meaning of metabolic tasks

• Type 2 Diabetes
• Heart Disease
• Digestive Health
• Multiple Sclerosis
• COVID-19 Vaccines
• Occupational Therapy
• Healthy Aging
• Health Insurance
• Public Health
• Patient Rights
• Caregivers & Loved Ones
• End of Life Concerns
• Health News
• Thyroid Test Analyzer
• Doctor Discussion Guides
• Hemoglobin A1c Test Analyzer
• Lipid Test Analyzer
• Complete Blood Count (CBC) Analyzer
• What to Buy
• Editorial Process
• Meet Our Medical Expert Board

What Is Metabolism?

• How it's Regulated
• Factors Impacting It
• Can You Increase It?

You’ve probably heard the term "metabolism" thrown around. Some people may blame their body size on their metabolism , describing it as fast or slow. Or maybe you’ve tried to boost your metabolism through a particular diet or exercise regimen.

But what exactly is your metabolism, and is there anything you can do to change it? This article will define metabolism, its types, how it’s regulated, and the factors that impact it.

Westend61 / Getty Images

"Metabolism" is a term that refers to all chemical processes or changes in your body at the cellular level. At any given moment, thousands of complex chemical processes are happening in your cells to keep you healthy and thriving. Such processes help with breathing, circulating blood, controlling your body temperature, and ensuring your brain and nerves function.

You have to consume food to create energy for these chemical processes. However, your body cannot use the food directly. The energy in the food must be converted into a form your cells can use for normal functioning, including growth, development, reproduction, repair, and elimination (getting rid of waste from the body).

Types of Metabolism

There are two types of metabolism: anabolism and catabolism . Anabolism processes require energy, while catabolism processes create or release energy. Both co-occur in the body.

Anabolism is any chemical process involved with synthesis, or building, of complex molecules from simpler molecules. Your body is constantly repairing and building new structures necessary for life.

Sometimes anabolism is visible, such as in building muscle for sports, a healing wound, a growth spurt, or pregnancy . Other times, it happens without our noticing, such as when producing new blood cells , repairing DNA (deoxyribonucleic acid; the hereditary material in humans and other organisms), mineralizing bone, or synthesizing hormones like insulin , estrogen , or testosterone .

Catabolism is any chemical process involved with the degradation, or the breakdown, of complex molecules into simpler molecules. The process is highly regulated by enzymes in the body, which control chemical reactions.

When you eat food, your body needs to digest it and catabolize or break it down into a form usable by the body. Another example of a catabolic process is when your body breaks down muscle and fat to release stored energy during strenuous exercise or starvation.

What Regulates Metabolism?

Metabolism is regulated by hormones , chemicals acting as messengers in the body.

While many hormones are involved in various chemical processes in the body, the thyroid hormone is largely responsible for regulating your metabolism.

Hormones are made in the endocrine glands , including the pituitary, pineal, thymus, thyroid, adrenal glands, and pancreas, and send chemical messages to all parts of the body, including organs, organ systems, and tissues like bone, skin, and muscle.

Hyperthyroidism , or excess thyroid hormone, increases your basal metabolic rate (resting energy expenditure) and encourages weight loss, the breakdown of fats and protein for energy, and reduced cholesterol levels . Hypothyroidism , or reduced thyroid hormone, has the opposite effect, lowering overall metabolism.

Factors Impacting Metabolism

Your metabolism, or the rate at which your body's chemical reactions use energy, can be affected by many factors, such as:

• Age: A large-scale study showed that metabolism has these four distinct phases:
• From birth to 1 year old, metabolism is very high (approximately 50% above normal adult metabolism).
• From age 1 to 20, metabolism will slowly decline.
• From age 20 to 60, metabolism will remain steady.
• From age 60 onward, metabolism will decline.
• Sex : Overall, women's metabolic rate is about 5% to 10% lower than men's. This is generally explained by differences in body composition since fat has less metabolic activity than muscle tissue. Women tend to carry approximately 10% more fat than men of a similar age and body mass index ( BMI ).
• Diet: Resting metabolic rate is regulated by the number of calories consumed in your diet relative to energy used. Research suggests that when excess calories are consumed, resting metabolic rate increases, whereas fasting and low-calorie consumption (under 1,000 calories per day) causes basal metabolic rate to drop.
• Exercise: The impact of exercise on metabolic rate is controversial, despite being widely studied. However, research suggests that exercise may regulate metabolism because bed rest in sedentary individuals leads to lower metabolic rates. Additionally, in highly trained runners who suddenly stop running, their resting metabolic rate can decrease by 7% to 10%.
• Sleep : A study in which healthy participants were sleep deprived (getting just four hours per night) showed that their resting metabolic rate decreased compared with participants who had a full night of rest (10 hours per night). However, when the sleep-deprived participants returned to regular sleep, their metabolic rate increased, returning to their normal rate.
• Hyper- or hypothyroidism : Since thyroid hormone is largely responsible for regulating your resting metabolic rate, excess or reduced thyroid hormone can impact your metabolism.
• Injury or disease : When a person is undergoing illness or injury, such as cancer, sepsis, trauma, or burns, their body's usage of energy increases by 20% to 25%. Metabolism also increases during the healing process after an operation by 15% to 30%. This is why proper nutrition is critical to healing and success after operation and treatment.

Can You Increase Your Metabolism?

Several factors contribute to your metabolism or energy usage. Unfortunately, much of your metabolism is out of your control. That's because your basal metabolic rate, or resting energy expenditure, makes up about 70% of your metabolism. This includes all normal cellular functions occurring without conscious participation, such as breathing, pumping blood, and a functioning brain and nervous system.

Another 10% of energy is used to digest your food. Physical activity accounts for the remaining 20%. This includes not just voluntary exercise but maintaining your posture and fidgeting.

While you may not be able to change your metabolism drastically, some things can help:

• Build more muscle: While the direct effect of physical activity itself may not increase your daily energy expenditure by a lot, the impact of building more lean muscle mass may, as it is more metabolically active tissue.
• Get enough rest: Sleep deprivation can lower your metabolism, so getting enough sleep may give you a boost if you are regularly sleep deprived.
• Eat enough food: Similarly to getting enough rest, you may also want to examine if you get enough calories each day. While most people may choose calorie-restricted diets to lose weight, this is counterproductive, as fasting or extremely low-calorie diets actually decrease your basal metabolism.

Metabolism is the sum of all chemical reactions in the body required to sustain life. These chemical processes involve energy and the breakdown and buildup of molecules and are regulated by hormones.

Your metabolism can be impacted by age, sex, diet, exercise, sleep, and injury or disease. While some lifestyle factors can affect your metabolism, no magic bullet or pill will drastically increase your metabolism.

However, a balanced diet with sufficient calories, regular exercise, and enough sleep can help regulate your metabolism.

National Cancer Institute. Metabolism .

Metallo CM, Heiden MGV. Understanding metabolic regulation and its influence on cell physiology .  Mol Cell . 2013;49(3):388-398. doi:10.1016/j.molcel.2013.01.018

Sánchez López de Nava A, Raja A. Physiology, metabolism . In:  StatPearls . StatPearls Publishing; 2022.

Stárka L, Dušková M. What is a hormone?   Physiol Res . 2019.  doi:10.33549/physiolres.934509

Mullur R, Liu YY, Brent GA. Thyroid hormone regulation of metabolism .  Physiol Rev . 2014;94(2):355-382. doi:10.1152/physrev.00030.2013

Pontzer H, Yamada Y, Sagayama H, et al. Daily energy expenditure through the human life course .  Science . 2021;373(6556):808-812. doi:10.1126/science.abe5017

Yoo J, Fu Q. Impact of sex and age on metabolism, sympathetic activity, and hypertension .  FASEB j . 2020;34(9):11337-11346. doi:10.1096/fj.202001006RR

Molé PA. Impact of energy intake and exercise on resting metabolic rate .  Sports Med . 1990;10(2):72-87. doi:10.2165/00007256-199010020-00002

Spaeth AM, Dinges DF, Goel N. Resting metabolic rate varies by race and by sleep duration .  Obesity (Silver Spring . 2015;23(12):2349-2356. doi:10.1002/oby.21198

Şimşek T, Şimşek HU, Cantürk NZ. Response to trauma and metabolic changes: posttraumatic metabolism .  Ulus Cerrahi Derg . 2014;30(3):153-159. doi:10.5152/UCD.2014.2653

Science Direct. Energy metabolism - an overview .

By Rebecca Valdez, MS, RDN Rebecca Valdez is a registered dietitian nutritionist and nutrition communications consultant, passionate about food justice, equity, and sustainability.

By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts.

• Weight Management
• Nutrition Facts
• Nutrition Basics
• Meal Delivery Services
• Apparel & Accessories
• Fitness Gear
• Nutrition Products
• Gift Guides
• Recipe Nutrition Calculator
• Weight Loss Calorie Goal
• BMI Calculator
• Body Fat Percentage Calculator
• Calories Burned by Activity
• Daily Calories Burned
• Pace Calculator
• Editorial Process
• Meet Our Review Board

Using Metabolic Equivalent for Task (MET) for Exercises

Sean is a fact-checker and researcher with experience in sociology, field research, and data analytics.

The metabolic equivalent for task (MET) is a unit that estimates the amount of energy used by the body during physical activity, as compared to resting metabolism . The unit is standardized so it can apply to people of varying body weight and compare different activities.

What Is a MET?

MET can be expressed in terms of oxygen use or kilocalories (what you commonly think of as calories). By using MET, you can compare the exertion required for different activities.

At rest or sitting idly, the average person expends 1 MET, which equals:

• 1 kilocalorie per kilogram of body weight times minutes of activity
• 3.5 milliliters of oxygen per kilogram of body weight times minutes of activity

At 2 MET you are using twice the calories per minute than you do at rest. The number of calories burned each minute depends on your body weight. A person who weighs more will burn more calories per minute.

MET Levels for Different Activities

In studies comparing different activities, the use of oxygen is measured since the body uses oxygen to expend calories . The Compendium of Physical Activities lists MET for hundreds of activities. The harder your body works during any given activity, the more oxygen is consumed and the higher the MET level.

• Under 3 MET: Light-intensity activities
• 3 to 6 MET: Moderate-intensity aerobic physical activities
• Over 6 MET: Vigorous-intensity aerobic physical activities

Moderate-Intensity

Moderate-intensity physical activity  is a level of body effort that is active but not strenuous. Characteristics of moderate-intensity physical activity include:

• Causes an increase in breathing and/or heart rate
• Results in 3 to 6 metabolic equivalents (MET) of effort

Your activity level is probably moderate if you are actively moving, potentially lightly sweating, and breathing harder than usual but can still carry on a normal conversation. Examples of moderate physical activities include things like walking outside or on a treadmill at a speed of about 3 mph, shooting a basketball, biking at a speed of about 10 mph or slower, doing water aerobics, ballroom dancing, or playing doubles tennis.

Vigorous-Intensity

Vigorous-intensity physical activity  burns more than 6 MET. During vigorous activity, you will sweat more, breath harder, and use more oxygen. At most, you will be able to utter only a couple of words between breaths.

Examples of vigorous physical activity include jogging and running (either outdoors or on a treadmill), playing tennis, swimming laps , playing basketball or soccer, or doing calisthenics  like push-ups and jumping jacks. Any of these activities can be done with varying levels of effort.

The key for vigorous-intensity physical activity is that the activity must be performed with intense effort. You will definitely know you are exercising. Vigorous intensity physical activity may be performed less frequently than moderate-intensity physical activity, as it is more demanding on the body.

A Word From Verywell

To get benefits for your health, you should get a variety of aerobic physical activity each week. The minimum suggested is either 150 minutes at moderate intensity or 75 minutes at a vigorous intensity, or a combination of the two spread out through the week.   These activities need to be performed for at least 10 minutes at a time. More is better, so it is good to find activities you enjoy to add to your healthy lifestyle.

Ainsworth BE, Haskell WL, Herrmann SD, et al. 2011 Compendium of Physical Activities: A Second Update of Codes and MET Values .  Med Sci Sports Exerc . 2011;43(8):1575-1581. doi:10.1249/MSS.0b013e31821ece12

Physical Activity Guidelines Committee. 2018 Physical Activity Guidelines Advisory Committee Scientific Report . U.S. Department of Health and Human Services; 2018.

U.S. Department of Health and Human Services. Physical Activity Guidelines for Americans . 2nd edition. 2019.

By Elizabeth Quinn, MS Elizabeth Quinn is an exercise physiologist, sports medicine writer, and fitness consultant for corporate wellness and rehabilitation clinics.

By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts.

Staying Active

Although many people view exercise as a way to lose weight, it plays a key role in the wellbeing of the body beyond weight loss. Research strongly supports its benefits across a range of physical and mental health conditions for people of all ages. However, busy lifestyles and an environment that encourages being sedentary for many hours of the day (driving door-to-door, sitting at an office desk, relaxing for the evening in front of a television) have led to exercise ranking low as a priority for many people.

Types of Exercise

All types of exercises offer health benefits. Performing different types of exercises can expand the range of benefits even further. But it is important to remember that some exercise is better than none, and that most everyone can participate in some form of exercise safely.

Aerobic/Cardiovascular physical activity. These are activities that are intense enough and performed long enough to maintain or improve one’s heart and lung fitness. Examples: walking , jogging, dancing , bicycling , basketball, soccer, swimming

Muscle-strengthening activity. This may be referred to as resistance training. These activities maintain or increase muscle strength, endurance, and power. Examples: weight machines, free weights, resistance elastic bands, Pilates, daily activities of living (lifting children, carrying groceries or laundry, climbing stairs)

Flexibility training. This may be referred to as stretching. It lengthens or flexes a skeletal muscle to the point of tension, and holds for several seconds to increase elasticity and range of motion around a joint. Improving flexibility can enhance the overall physical performance of other types of exercise. Examples: dynamic stretches performed with movement ( yoga , tai chi), static stretches without movement (holding a pose for several seconds or longer), passive stretching (using an external force like a strap or wall to hold an elongated pose), and active stretching (holding a pose without an external force)

Balance training. These activities are intended to throw off one’s balance to improve body control and stability. They can help to prevent falls and other injuries. Examples: standing on one foot, walking heel-to-toe in a perfectly straight line, standing on a balance or wobble board

Measures of Exercise Intensity

Although just moving more and sitting less offers health benefits, how much energy you use while exercising can increase those health benefits further. This is referred to as energy intensity.

The Borg Scale of Perceived Exertion measures your exercise intensity by rating how you feel. It is based on observations like higher heart rate, heavier and faster breathing, increased sweating, and muscles feeling tired. It does not use actual measurements of these occurrences but a personal self-check.

The scale uses numbers from 6 to 20. The lowest rating is “no feeling of exertion,” at number 6, and the highest rating is “very, very hard,” at number 20. Moderate activities register 11 to 14 (“fairly light” to “somewhat hard”) while vigorous activities usually rate 15 or higher (“hard” to “very, very hard”). Dr. Gunnar Borg, who created the scale, set it to run from 6 to 20 as a simple way to estimate heart rate—multiplying the Borg score by 10 gives an approximate heart rate for a particular level of activity. [1]

Exercise workouts may vary in intensity throughout the session. You can use the Borg Scale to change the intensity, by speeding up or slowing down movements or applying more or less resistance (such as increasing the incline on a treadmill or turning the resistance control knob on a stationary bicycle).

Target Heart Rate

Calculating your heart rate and target heart rate can be used to measure exercise intensity. First determine your maximum heart rate : subtract your age from 220 (example: the maximum heart rate for a 40-year-old person would be 220 – 40 = 180 beats per minute). The target heart rate for moderate-intensity exercise is between 65-75% of your maximum heart rate (or 77-93% of maximum heart rate for vigorous exercise). So for the 40-year-old person with a maximum heart rate of 180, the target heart rate falls somewhere between 117-135 beats per minute for moderate exercise, or 139-167 for vigorous exercise.

Then measure your actual heart rate in either of these two ways:

• Midway through the exercise, stop to check your pulse. Place the tips of your index and middle fingers at the wrist and press lightly on the artery in line with the thumb. Count the heartbeats for 30 seconds and multiply by 2.
• Wear a heart rate monitor. Some pedometers have a built-in heart rate monitor that displays your current heartbeats per minute.

MET stands for the metabolic equivalent of task. One MET is the amount of energy used while sitting quietly. Physical activities may be rated using METs to indicate their intensity. For example, reading may use about 1.3 METs while running may use 8-9 METs. METs can also be translated into light, moderate, and vigorous intensities of exercise.

• Sedentary —Uses 1.5 or fewer METs. Examples are sitting, reclining, or lying down.
• Light intensity —Uses from 1.6-3.0 METs. Examples are walking at a leisurely pace or standing in line at the store.
• Moderate intensity —Uses from 3.0-6.0 METs. Examples are walking briskly , vacuuming, or raking leaves.
• Vigorous intensity —Uses from 6.0+ METs. Examples are walking very quickly, running, taking an aerobics class, or shoveling snow.

Exercise experts measure activity in metabolic equivalents, or METs. One MET is defined as the energy it takes to sit quietly. For the average adult, this is about one calorie per every 2.2 pounds of body weight per hour; someone who weighs 160 pounds would burn approximately 70 calories an hour while sitting or sleeping.

Moderate-intensity activities are those that get you moving fast enough or strenuously enough to burn off three to six times as much energy per minute as you do when you are sitting quietly, or exercises that clock in at 3 to 6 METs. Vigorous-intensity activities burn more than 6 METs.

One limitation to this way of measuring exercise intensity is that it does not consider the fact that some people have a higher level of fitness than others. Thus, walking at 3 to 4 miles-per-hour is considered to require 4 METs and to be a moderate-intensity activity, regardless of who is doing the activity—a young marathon runner or a 90-year-old grandmother. As you might imagine, a brisk walk would likely be an easy activity for the marathon runner, but a very hard activity for the grandmother.

This table gives examples of light-, moderate-, and vigorous-intensity activity for healthy adults:

Physical Activity Through the Life Course

In the U.S., the Department of Health and Human Services provides specific guidelines for physical activity for different life stages and conditions: [2]

• Children ages 3 through 5 . Try to be physically active throughout the day. Adult caregivers should encourage children this age to engage in active playing for at least 3 hours daily.
• Children and adolescents ages 6 through 17 . At least 1 hour daily of moderate-to-vigorous activity with both aerobic and strength movements.
• Adults . Move more frequently throughout the day and sit less. Engage in at least 150 to 300 minutes weekly (spaced throughout the week) of moderate-intensity aerobic exercise and at least 2 days weekly of muscle-strengthening exercises. Greater health benefits may be seen with more than 300 minutes weekly of exercise.
• Older adults . Follow similar activity guidelines as those for adults but also include a focus on balance training. Although discussing the start of a new exercise regimen with one’s doctor is a good practice for all ages, it is especially important with this age group because of the higher likelihood of having health conditions or physical limitations that may require modified exercises. See additional physical activity considerations for older adults . 
• Women who are pregnant or postpartum . Aim for 150 minutes weekly (spaced throughout the week) of moderate-intensity aerobic exercise. If vigorous exercise was performed regularly prior to pregnancy, one may continue this throughout pregnancy after discussing with their doctor. See additional physical activity considerations for pregnancy . 
• Adults with physical disabilities and chronic conditions . Follow similar activity guidelines as those for adults if able to exercise, but discuss with one’s doctor about the types and amounts of activity that would be appropriate for specific conditions. Any exercises within one’s ability is encouraged, to avoid being completely sedentary. See additional physical activity considerations for chronic conditions . 

The frequency, duration, and intensity of exercise are helpful terms to consider when deciding on an exercise regimen.

• Frequency: How often will you do the activity—once a day, three times a week, twice a month?
• Duration: How long is the exercise session—20 minutes, 1 hour, 30 minutes split into two sessions in one day?
• Intensity: How much energy is needed—light versus vigorous activity, 3 METs versus 6 METs?

Wearable trackers for physical activity

Generally these trackers are pretty accurate when measuring steps taken. But other measures, such as how many calories are burned, can overestimate or underestimate the actual amount. Studies looking at the accuracy of devices in tracking calories used while exercising tend to be small in size. In one study, 14 participants wearing different popular brand devices walked and ran. The estimated calorie usage displayed on the devices was compared with measurements from indirect calorimetry (a reliably accurate technique to measure calorie output). The results were mixed. Some of the devices were accurate for calorie expenditure with running but not walking and visa versa. Some of the devices overestimated the amount of calories used during exercise. [5] Other studies found similar discrepancies. [6]

Tracking devices can be useful for personal motivation and accountability, but the data should be interpreted with caution as there are variable readings among devices. The accuracy of the data may also vary within the same device when performing different intensities of exercise. [6] They are best used with other methods to gauge fitness levels, such as monitoring the frequency, duration, and perceived exertion of your exercise routine. It’s also important to have motivation to exercise because you enjoy how you feel during and after the exercise, not just to reach a certain number on a tracker.

Exercise Safety

Safety should be a major priority when exercising. Any physical activity carries the risk of injury, whether you are just starting an exercise regimen or are a seasoned fitness buff. But don’t let this stop you from moving because the health benefits of being active far outweigh any risks. Using caution and patience can reduce the risk of injuries.

Common workout injuries include:

• Strained or pulled muscles
• Ankle sprain
• Knee strain
• Inflamed tendons or ligaments
• Rotator cuff (shoulder) injury
• Overuse injuries caused by repetitive movements using mainly one part of the body

In very rare cases, vigorous physical activity may lead to a heart attack or sudden death. Active people have a lower risk of serious or fatal heart problems than inactive people.

Common missteps:

Not talking with your doctor first. If you are new to exercise or have medical conditions, let your doctor know what type of exercise you’ll be starting. They can review the format to ensure it is safe with a specific health condition.

Doing too much too soon. This is very common as people may be highly motivated when starting a new exercise program. However, forcing your body to move with too much intensity can be jarring to the heart, muscles, and joints that may lack strength from inactivity. This often leads to injury. Even if starting slowly feels too easy, plan to progress exercise gradually. Start with light to moderate-intensity movements for a shorter amount of time, and continue this for a few weeks. As you develop strength and stamina, you can add minutes and higher-intensity movements every few weeks.

Leaving out the warm-up and cool-down. A warm-up before exercising includes light movements that initiate the flow of blood and loosening of muscles and joints. An example would be 5-10 minutes of marching in place, doing arm circles, and neck rolls. After exercising, the cool-down is important to slow down the body and heart rate steadily, as a sudden stop in movements can interfere with blood flow to the brain and cause lightheadedness or dizziness. A cool-down could be simply slowing the pace of whatever exercise being performed for 10 minutes (if jogging, change to a walk; if on a stationary bicycle, release any tension on the resistance knob and peddle slower). The cool-down period may also include stretches that are most effective when the muscles are warmed from exercising; stretches help to lengthen muscles that will protect against injuries. A cool-down with stretching can also lessen muscle soreness the following day.

How to get started safely

• If you have a chronic health condition or are pregnant, let your doctor know of your desire to start exercising.
• If you are sedentary, start with activities that are lower impact and require a light-to-moderate effort, such as walking , gardening, stationary bicycling, or swimming. Progress gradually—it’s especially important to “start low and go slow.”
• Choose the right equipment. If you’re cycling, wear a bike helmet. If you’re going out for a walk, pull on a well-fitting pair of sneakers instead of a pair of flip flops. Generally, exercise shoes should be replaced very 4-6 months as the cushioning wears out.
• Find a safe place to work out. Seek out streets that have sidewalks or bicycle lanes, or visit a local park. Play basketball on well-maintained courts.
• Pay attention to the weather. In the middle of a heat wave, exercise in the morning or evening when it’s cooler out, exercise indoors, or hit the swimming pool instead of the tennis court. Be aware of signs of overheating like dizziness, nausea, headache, cramping, and a racing heart rate that doesn’t slow down even when stopping the exercise.
• Stay hydrated with water. The amount will vary depending on the temperature (more is needed in very hot and humid conditions) and level of exercise. For moderate workouts of one hour or less, bring about 24 ounces of water to drink during and after exercising.
• Choose healthy “fuel.” A diet with adequate amounts of healthy protein and carbohydrates is sufficient to fuel the body for low to moderate amounts of physical activity, such as an hour of jogging or bicycling.
• Be wary of supplement claims. Fitness gurus and advertisements touting workout supplements as crucial for peak performance, fat loss, and explosive muscle growth might have you believing you can’t effectively exercise without them. Although some supplements have been researched for use in regular high-intensity, strenuous physical activity (such as marathon training or power lifting), it’s important to note they are not regulated for safety. Be sure to consult with a doctor before incorporating them into your exercise routine and discuss if there are any potential contraindications if you have existing medical conditions. Learn more about the research on common workout supplements .
• Listen to your body. If you feel very fatigued, pain, or lightheaded while exercising, slow down the workout or end it early.

What’s the difference between muscle pain and muscle soreness?

On the other hand, sharp pain in a specific area that starts during the workout or soon after could be a sign of a more serious injury. See a doctor if the pain does not lessen in a day or two or you notice swelling or bruising around the painful area.

10 Tips to Keep Moving

• Plan exercise into your day .  Intention is an important first step. Set aside a specific time in your schedule to exercise and write it in your planner.
• Accountability helps . If your motivation is lagging, connect with a friend or family member with a similar goal to move more. A workout partner can help keep you on track and motivate you to get out the door.
• Try counting steps . Step-counters or pedometers are an easy, inexpensive way to remind yourself to move. Working up to 10,000 steps per day can be a good general goal. If that seems too intimidating, measure your steps on an average day and increase by 1000 steps every two weeks.
• Keep it brisk .  When you walk, make it brisk, since this may help control weight better than walking at a leisurely pace. What is brisk enough? Walk as though you are meeting someone for lunch and you are a little late.
• Turn off the TV, computer, and smart phone .  Chances are that if you turn off these devices for an hour or two, you will automatically move more and curb your “sit time.” Fill the time by doing household chores, running errands, playing with the kids, or taking a stroll around your neighborhood.
• Turn sit time into fit time . Try to combine movement with a sedentary activity that you already do. For example, perform squats, marching in place, jumping jacks, push-ups, or sit-ups while watching TV or throughout each commercial. Fidgeting, or its scientific term non-exercise activity thermogenesis (NEAT), also uses extra energy. Studies show that people who are lean incorporate more NEAT movements throughout the day compared with people who are overweight. [7] This might mean pacing around while talking on the phone, tapping your feet when sitting, drumming your fingers on a desk or your leg, or wiggling your toes. For inspiration on how to move “creactively” wherever you are, check out Activating a Move-Friendly World .
• Move at the office . If you work long shifts or care for a busy family after hours, fitting in a workout can be daunting. So focus on moving at the office even if you have a sedentary desk job. Make climbing stairs and avoiding elevators the norm, park as far from the front office door as possible, set a reminder to get up and walk for 5 minutes each hour (that could add up to 40 minutes in a day!), or follow a short desk exercise video online.
• Split the workout . If you are new to exercise and find a 30-minute session challenging, split it into two 15-minute sessions. The fitness benefit may actually be greater if you can exercise with higher energy and intensity in two shorter bouts, than if you tried to exercise for 30 minutes but slowed down from fatigue towards the end.
• Sign up for a class or a specific event . Check out the fitness class schedule at your local gym, yoga studio, or community center. Some offer virtual classes with a live instructor but which you can do at home. Or sign up for a specific event like a road race or walk-for-charity a few months out; this can help drive you to train regularly the weeks leading up to the event. You may find that having a target date or the structure of a weekly class keeps you consistent.
• Reward yourself .  Set short-term goals—then acknowledge and reward yourself when achieving them. Positive affirmations are key to building confidence as you commit to ongoing fitness goals. Treat yourself to new exercise shoes, clothing, or workout gear; a new book; or a massage.

The influence of music in exercise

• Physical Activity Considerations for Special Populations
• More Research on Physical Activity 
• Walking for Exercise
• HIIT (High Intensity Interval Training)
• Yoga for Exercise
• Zumba Fitness
• Borg G.A. Psychophysical bases of perceived exertion.  Medicine and Science in Sports and Exercise . 1982; 14:377-381.
• The U.S. Department of Health and Human Services. Executive Summary: Physical Activity Guidelines for Americans, 2 nd https://health.gov/sites/default/files/2019-10/PAG_ExecutiveSummary.pdf Accessed 10/19/20.
• Brickwood KJ, Watson G, O’Brien J, Williams AD. Consumer-based wearable activity trackers increase physical activity participation: systematic review and meta-analysis. JMIR mHealth and uHealth . 2019;7(4):e11819.
• Cadmus-Bertram LA, Marcus BH, Patterson RE, Parker BA, Morey BL. Randomized trial of a Fitbit-based physical activity intervention for women. American journal of preventive medicine . 2015 Sep 1;49(3):414-8.
• Price K, Bird SR, Lythgo N, Raj IS, Wong JY, Lynch C. Validation of the Fitbit One, Garmin Vivofit and Jawbone UP activity tracker in estimation of energy expenditure during treadmill walking and running. Journal of medical engineering & technology . 2017 Apr 3;41(3):208-15.
• Evenson KR, Goto MM, Furberg RD. Systematic review of the validity and reliability of consumer-wearable activity trackers. International Journal of Behavioral Nutrition and Physical Activity . 2015 Dec 1;12(1):159.
• Villablanca PA, Alegria JR, Mookadam F, Holmes Jr DR, Wright RS, Levine JA. Nonexercise activity thermogenesis in obesity management. InMayo Clinic Proceedings 2015 Apr 1 (Vol. 90, No. 4, pp. 509-519). Elsevier.
• Karageorghis CI, Priest DL. Music in the exercise domain: a review and synthesis (Part I). International review of sport and exercise psychology . 2012 Mar 1;5(1):44-66.
• Ballmann CG. The influence of music preference on exercise responses and performance: A review. Journal of Functional Morphology and Kinesiology . 2021 Apr 8;6(2):33.

Last reviewed December 2022

Terms of Use

The contents of this website are for educational purposes and are not intended to offer personal medical advice. You should seek the advice of your physician or other qualified health provider with any questions you may have regarding a medical condition. Never disregard professional medical advice or delay in seeking it because of something you have read on this website. The Nutrition Source does not recommend or endorse any products.

• More from M-W
• To save this word, you'll need to log in. Log In

Definition of metabolic

Examples of metabolic in a sentence.

These examples are programmatically compiled from various online sources to illustrate current usage of the word 'metabolic.' Any opinions expressed in the examples do not represent those of Merriam-Webster or its editors. Send us feedback about these examples.

Word History

borrowed from German metabolisch, borrowed from Greek metabolikós "changeable, subject to change," from metabolḗ "change, transition" (from metabol-, stem in noun derivation of metabállein "to put into a different position, turn about, change, alter," from meta- meta- + bállein "to reach by throwing, let fly, strike, put, place") + -ikos -ic entry 1 — more at devil entry 1

Note: The term was introduced by the German physiologist Theodor Schwann (1810-82) in Die Mikroskopischen Untersuchungen über die Uebereinstimmung in der Struktur und dem Wachsthum der Thiere und Pflanzen (Berlin, 1839), p. 229: "Die Frage über die Grundkraft der Organismen reducirt sich also auf die Frage über die Grundkräfte der einzelnen Zellen. Wir müssen nun die allgemeinen Erscheinungen der Zellenbildung betrachten, um zu finden, welche Kräfte man zur Erklärung derselben in den Zellen voraussetzen muss. Diese Erscheinungen lassen sich unter zwei natürlichen Gruppen bringen: Erstens Erscheinungen, die sich auf die Zusammenfügung der Moleküle zu einer Zelle beziehn; man kann sie die plastischen Erscheinungen der Zellen nennen; zweitens Erscheinungen, die sich auf chemische Veränderungen, sowohl der Bestandtheile der Zelle selbst, als des umgebenden Cytoblastems beziehn; diese kann man metabolische Erscheinungen nennen (τὸ μεταβολικὸν [sic] was Umwandlung hervorzubringen oder zu erleiden geneigt ist)." — "The question, then, as to the fundamental powers of organized bodies resolves itself into that of the fundamental powers of the individual cells. We must now consider the general phenomena attending the formation of cells, in order to discover what powers may be presumed to exist in the cells to explain them. These phenomena may be arranged in two natural groups: first, those which relate to the combination of the molecules to form a cell, and which may be denominated the plastic phenomena of the cells; secondly, those which result from chemical changes either in the component particles of the cell itself, or in the surrounding cytoblastema [fluid held to be the formative substance from which cells arise], and which may be called metabolic phenomena ( tò metabolikòn, implying that which is liable to occasion or to suffer change)." ( Microscopical Researches into the Accordance in the Structure and Growth of Animals and Plants, translator Henry Smith, London, 1847).

1841, in the meaning defined above

Phrases Containing metabolic

• metabolic syndrome
• basal metabolic rate

Dictionary Entries Near metabolic

metabolic heat

Cite this Entry

“Metabolic.” Merriam-Webster.com Dictionary , Merriam-Webster, https://www.merriam-webster.com/dictionary/metabolic. Accessed 27 Nov. 2023.

Kids Definition

Kids definition of metabolic, medical definition, medical definition of metabolic, more from merriam-webster on metabolic.

Nglish: Translation of metabolic for Spanish Speakers

Britannica English: Translation of metabolic for Arabic Speakers

Subscribe to America's largest dictionary and get thousands more definitions and advanced search—ad free!

Can you solve 4 words at once?

Word of the day.

See Definitions and Examples »

Get Word of the Day daily email!

Games & Quizzes

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 04 June 2020

Analysis of human metabolism by reducing the complexity of the genome-scale models using redHUMAN

• Maria Masid   ORCID: orcid.org/0000-0001-6470-5083 1 ,
• Meric Ataman   ORCID: orcid.org/0000-0002-7942-9226 2 &
• Vassily Hatzimanikatis   ORCID: orcid.org/0000-0001-6432-4694 1  

Nature Communications volume  11 , Article number:  2821 ( 2020 ) Cite this article

10k Accesses

12 Citations

34 Altmetric

Metrics details

• Cancer metabolism
• Computational models
• Data integration
• Metabolic engineering
• Metabolic pathways

An Author Correction to this article was published on 23 July 2020

This article has been updated

Altered metabolism is associated with many human diseases. Human genome-scale metabolic models (GEMs) were reconstructed within systems biology to study the biochemistry occurring in human cells. However, the complexity of these networks hinders a consistent and concise physiological representation. We present here redHUMAN, a workflow for reconstructing reduced models that focus on parts of the metabolism relevant to a specific physiology using the recently established methods redGEM and lumpGEM. The reductions include the thermodynamic properties of compounds and reactions guaranteeing the consistency of predictions with the bioenergetics of the cell. We introduce a method (redGEMX) to incorporate the pathways used by cells to adapt to the medium. We provide the thermodynamic curation of the human GEMs Recon2 and Recon3D and we apply the redHUMAN workflow to derive leukemia-specific reduced models. The reduced models are powerful platforms for studying metabolic differences between phenotypes, such as diseased and healthy cells.

Introduction

An altered metabolism is a hallmark of several human diseases, such as cancer, diabetes, obesity, Alzheimer’s, and cardiovascular disorders 1 , 2 . Understanding the metabolic mechanisms that underlie this reprogramming guides the discovery of new drug targets and the design of new therapies. To this effect, tremendous efforts are now being made to use the large amounts of now-available multi-omics experimental data to gain insight into the metabolic alterations occurring in different phenotypes. Unfortunately, current mathematical models can be too complex for this analysis, rendering them too cumbersome to employ for many systems biology studies.

In the field of systems biology, genome-scale metabolic models (GEMs) integrate available omics data with genome sequences to provide an improved mechanistic understanding of the intracellular metabolism of an organism. GEMs have been reconstructed for a large diversity of organisms spanning from bacteria to mammals 3 , 4 , 5 and are valuable tools for studying metabolism 6 , 7 . The mathematical representation of GEMs through the stoichiometric matrix 7 is amenable to methods such as flux balance analysis (FBA) 8 and thermodynamic-based flux balance analysis (TFA) 9 , 10 , 11 , 12 , 13 , which ensure that the modeled metabolic reactions retain feasible concentrations and their directionalities obey the rules of thermodynamics, to predict reaction rates and metabolite concentrations when optimizing for a cellular function, such as growth, energy maintenance, or a specific metabolic task. Additionally, GEMs can be used for gene essentiality 14 , drug off-target analysis 15 , metabolic engineering 16 , 17 , 18 , and the derivation of kinetic models 19 , 20 , 21 , 22 .

The first human GEM was reconstructed in 2007 23 , 24 . Since then, the scientific community has been working to develop high-quality human GEMs, including HMR 2.0 25 , Recon 2 26 , Recon 2.2 27 , and Recon 3D 28 . The human GEMs used for the analysis in this article are Recon 2 and Recon 3D. Recon 2 is composed of 7440 reactions with 4821 of them associated to 2140 genes, and 2499 unique metabolites across seven compartments: cytosol, mitochondria, peroxisome, Golgi apparatus, endoplasmic reticulum, nucleus, and lysosome. Recon 3D is the latest consensus human GEM. It is an improved more comprehensive version of the previous GEMs consisting of 10,600 reactions, with 5938 of them associated with 2248 genes, and 2797 unique metabolites compartmentalized as Recon 2 with an additional compartment for the mitochondria intermembrane space.

Human GEMs reconstruct the metabolic reactions occurring in several human cell types. However, a given cell type only leverages a portion of these reactions. This motivates the development of methods to generate context-specific metabolic models that can be used to study the differences in metabolism for different cell types 29 , for healthy and diseased cells 30 , 31 , and for cells growing under diverse extracellular conditions. Some examples of such methods are (1) GIMME 32 , mCADRE 33 , and tINIT 34 to reconstruct tissue-specific models based on omics data and a set of tasks or a specific objective function; (2) redGEM–lumpGEM 35 , 36 to reconstruct models around a specific set of subsystems of interest for the study; and (3) iMM 37 , 38 to characterize the extracellular medium and the metabolites that are essential for growth under each condition. Context-specific metabolic models have been extensively used to understand the differences in metabolism between cancer cells and their healthy counterparts 39 , 40 , 41 , 42 , 43 , 44 , 45 .

In this article, we present redHUMAN, a workflow to reconstruct thermodynamic-curated reductions of the human GEMs Recon 2 and Recon 3D. We integrate the thermodynamic properties of the metabolites and reactions into the GEMs and use redGEM–lumpGEM to reconstruct reduced models around specific subsystems. Furthermore, we introduce redGEMX, a method to identify the pathways required to connect the extracellular compounds to a core network. redGEMX guarantees that the reduced models have all the feasible pathways that consume and produce the components of the extracellular environment of the cell. Finally, we use metabolic data for leukemia as an example of how to integrate experimental data to derive disease- and tissue-specific metabolic models.

Overall workflow

In order to generate reduced models from human GEMs, we developed redHUMAN, a six-step workflow that can be applied to any GEM or desired model system. The overall workflow is briefly described here and shown in Fig.  1 , and the details of each step in its application to the human GEMs Recon 2 and Recon 3D to generate thermodynamic-curated reductions are provided in the subsequent sections. For the workflow, the thermodynamic information for compounds and reactions, which is assembled from earlier studies or estimated using established group contribution methods, is first integrated into the GEM. Second, the subsystems, or families of pathways with a specific functional role for a biological process, are selected based on the objectives of the specific study. These pathways are explicitly represented and constitute the core of the reduced model. For example, when studying cancer metabolism, this can include reported subsystems that are deregulated in cancer cells in addition to the standard central carbon pathways. Third, these subsystems are expanded using reacti\ons from the GEM to create a connected core network. In this step, we include every reaction that connects core metabolites and that is not a member of the formal definition of the selected subsystems in the core model. In steps four and five, we include the shortest pathways to connect the extracellular metabolites from the defined medium as well as the shortest pathways to generate the biomass components from the core network. These steps guarantee that the model has all pathways that are essential for survival and growth of the cells based on the availability of nutrients. In the sixth step, experimental data for a specific physiological state is integrated in the model, and the final model is verified through checks that ensure the consistency of the reduced model with the original GEM.

(1) Thermodynamic curation: the Gibbs free energy of compounds and reactions are estimated and used to define the reaction directionality. (2) Subsystem selection: the subsystems relevant for the study are selected. (3) Network expansion: the initial subsystems are connected using reactions from the GEM to generate a core network. (4) Extracellular medium connection: the pathways that connect the extracellular medium components to the core network are identified. (5) Biosynthetic reaction generation: the pathways required to produce the biomass building blocks are classified. (6) Data integration and consistency checks: experimental values are integrated and the model is verified through consistency checks.

Thermodynamic curation of the human GEMs (Step 1)

We first determine the directionality of the chemical reactions of the network, which is directly associated with their corresponding Gibbs free energy. The Gibbs free energy of a reaction can be estimated from the thermodynamic properties of its reactants and products. Therefore, we curated the GEMs Recon 2 and Recon 3D (see “Methods”) and integrated the thermodynamic properties for 52.4% of the 2499 unique metabolites from Recon 2 and 67.5% of the 2797 unique metabolites from Recon 3D (Fig.  2a and Supplementary Data  1 ). Three main reasons prevented the estimation of the thermodynamic properties of the metabolites: (1) an unknown molecular structure (SMILE), (2) an incomplete elemental description (for example, an R in the structure), and (3) groups in the structure for which an estimated free energy does not exist (for example, >N − group). We observed that as the number of metabolites increases from Recon 2 to Recon 3D, the percentage of thermodynamic coverage increases as well. This is due to the improved annotation of the metabolite structures in Recon 3D. Using the thermodynamic properties of the compounds as constraints (see “Methods”), we estimated the Gibbs free energy for 51.3% of the 7440 reactions present in Recon 2 and 61.6% of the 10,600 reactions in Recon 3D. These constraints ensured that the reactions in the computed flux distributions operated in thermodynamically feasible directions.

a Thermodynamics for the unique compounds in Recon 2 (orange) and Recon 3D (blue). The percentage is relative to the total number of unique compounds. b Size of the core network when the expansion is performed for different degrees. c Number of reactions that pairwise connect the subsystems for Recon 2 (values below the diagonal) and Recon 3D (values above the diagonal) for degree D  = 1.

Subsystem selection to build the core (Step 2)

A proper metabolic model contains the pathways that are essential for the survival of the cell as well as the pathways that are informative of a specific metabolic behavior. In this work, we were interested in the metabolism of cancer cells. Thus, we selected as core subsystems: (a) the central carbon pathways that provide the energy, redox potential, and biomass precursors, and (b) the subsystems that have been reported to be altered in cancer cells 46 , 47 , 48 , 49 . Consequently, the core subsystems for our models were glycolysis, pentose phosphate pathway, citric acid cycle, oxidative phosphorylation, glutamate metabolism, serine metabolism, urea cycle, and reactive oxygen species detoxification. We have estimated the thermodynamic properties for the metabolites and the reactions in these initial subsystems. In the case of Recon 2, we provide an estimate for the Gibbs free energy of formation for 236 metabolites (94.4% of the total in the initial subsystems) and the Gibbs free energy of reaction for 143 reactions (83.1% of the reactions in the initial subsystems). In the case of Recon 3D, we provide estimated values of the thermodynamic properties for 288 metabolites (97.6%) and for 183 reactions (91.0%).

Network expansion (Step 3)

Subsequently, to reconstruct the core network we pairwise connected the chosen subsystems using redGEM (see “Methods”). The algorithm first performed an intra-expansion of the initial subsystems. In this process, each initial subsystem was expanded to include additional reactions from the GEM whose reactants and products belong to that subsystem. These reactions can be assigned to different subsystems in the GEM which are not any of the initial subsystems and the core network would miss these additional reactions if we had considered the formal definition of the initial subsystems. The initial core subsystems of Recon 2 contained a total of 180 reactions. After the intra-expansion, 135 reactions from 21 subsystems were added. Examples of these added reactions included three from pyruvate metabolism that interconvert acetyl-CoA, acetate, malate, and pyruvate, which are all metabolites that participate in the citric acid cycle subsystem. For Recon 3D, 171 reactions from 24 subsystems were added to the 211 reactions from the initial core subsystems.

Next, the algorithm performed a directed graph search to find the reactions from the GEM that connected the subsystems for different degrees D (Fig.  2b and Supplementary Table  1 ), wherein D represents the distance (in number of reactions) between pairs of metabolites from the subsystems. Our final models included the connections for degree D  = 1, that is, all the reactions that in one step connect two metabolites (excluding cofactors) belonging to any of the initial subsystems. A degree D  = 1 was enough to pairwise connect all the initial subsystems (Fig.  2c ). This resulted in a Recon 2 core network of 356 metabolites and 617 reactions and a Recon 3D core network of 440 metabolites and 796 reactions.

Extracellular medium connection (Step 4)

Cells adapt their metabolism to the available nutrients in their extracellular environment. Consequently, a correct definition of the medium in the metabolic model is fundamental for an adequate representation of the intracellular metabolism. Given the complexity of the extracellular medium, it is particularly important to identify and classify the essentiality of the medium components and the pathways used for their metabolism. To this end, we curated the representation of the interactions of the cell with its environment into the human GEMs. First, we did not allow the exchange of intracellular metabolites lacking associated transport reactions or transport molecules containing P, CoA, or ACP (acyl carrier protein). Secondly, we allowed the synthesis of generic fatty acids from palmitate, with reactions from Recon 2 and Recon 3D (Supplementary Note  1 ). We next characterized the in silico minimal medium composition required for growth in the human GEMs by applying iMM (see “Methods”), which identifies the minimal set of metabolites that need to be uptaken to simulate growth. The results showed that Recon 2 required a medium with glucose, the nine essential amino acids, and some inorganics (PO 4 , NH 4 , SO 4 , O 2 ), and Recon 3D simulated growth in a medium with glucose, the nine essential amino acids, the same inorganics as Recon 2, and one of the two essential fatty acids (alpha-linolenic acid and linoleic acid). The presence of the two essential fatty acids in the iMM of Recon 3D is a consequence of the improvement of the lipid metabolism 28 , where the essential fatty acids participate in the synthesis of phospholipids. This demonstrates how the algorithms and workflow can be used to compare and validate updated model reconstructions for the same organisms or between different organisms.

Seeking to identify the pathways that human cells use to uptake and secrete extracellular metabolites, we next developed the method redGEMX (see “Methods”). This algorithm finds the pathways from the GEM that are needed to connect the extracellular metabolites to the core network defined by redGEM. In this work, we considered a complex medium composition of 34 metabolites (Fig.  3a ), and redGEMX found the corresponding GEM reactions that connected 26 of these extracellular metabolites (we excluded the inorganics and the fatty acids) to the core network.

a Extracellular medium composition defined in the models. b Graph of the subnetwork from Recon 2 for the uptake of l -histidine and the medium components required for its metabolism. Green represents the metabolites from the subnetwork, and orange represents the metabolites of the core network where the subnetwork is connected. In blue, the medium metabolite under study ( l -histidine) and in pink, the extracellular metabolites co-utilized to metabolize l -histidine. The pathway starts with the transport of l -histidine from the extracellular space to the cytosol, where it is sequentially transformed into urocanate (urcan_c), 4-imidazolone-5-propanoate (4izp_c), N-formimidoyl- l -glutamate (forglu_c), l -glutamate (glu_L_c), 5-formiminotetrahydrofolate (5forthf_c), 5-10-methenyltetrahydrofolate (methf_c), and 10-formyltetrahydrofolate (10thf_c). 4-Aminobutanoate (4abut_c) is converted to l -glutamate through a reaction from the subsystem glutamate metabolism, and finally, l -glutamate is connected to the TCA cycle.

An example of one of these connected metabolites is the essential amino acid l -histidine which affects many aspects of human physiology, including cognition functions and allergic reactions. The classical pathway to metabolize l -histidine consists of four steps that sequentially convert it into urocanate, 4-imidazole-5-propanoate, N-formimidoyl- l -glutamate, and ultimately, l -glutamate 50 . Interestingly, the resulting redGEMX subnetwork for l -histidine uses this classical pathway to connect it to the Recon 2 core metabolites l -glutamate and 4-aminobutanoate, both from the subsystem glutamate metabolism. The subnetwork is composed of 22 reactions, and it contains not only the classical pathway but also all the additional reactions required to balance the cofactors and by-products (Fig.  3b ). These additional reactions are essential for an active main pathway, as they include the utilization of NH 4 , the sources of water and tetrahydrofolate, and the conversion of the by-product 5-formiminotetrahydrofolate to 10-formyltetrahydrofolate, which regenerates tetrahydrofolate. Cellular metabolism has evolved to give flexibility to the cells to survive and function under different conditions. This flexibility is captured in the metabolic networks with the existence of alternative pathways. For this reason, using redGEMX we found three alternative pathways of minimum size (22 reactions) to connect l -histidine to the core network of Recon 2. The alternatives emerge from the existence of different transport reactions for the extracellular metabolites. In the case of Recon 3D, l -histidine is connected to the core network using 20 reactions, and there exist two pathways of minimum size. The subnetworks connect l -histidine to the Recon 3D core metabolites l -glutamate, 5-10-methylenetetrahydrofolate, 2-oxoglutarate, and pyruvate using the classical pathway to metabolize l -histidine. The different topology of the Recon 2 and Recon 3D networks manifests in differences in the pathways used to metabolize and synthesize the compounds, thus, it is important to characterize which are the pathways used in the models. Following this approach, we added the reactions that compose all the alternative subnetworks of minimum size to the core networks to connect the 26 extracellular metabolites (Supplementary Table  2 and Supplementary Data  2 ).

The subnetworks generated with redGEMX provide a new perspective on the current understanding of metabolic pathways, as they not only contain the main pathway but they also include other reactions necessary to supply and consume all the cofactors and by-products. Moreover, the alternatives can be used to hypothesize which pathways cells use when growing under different conditions, such as when different nutrients are present in the environment or under different intracellular regulations when different enzymes are operational. If metabolomics data are available, the subnetworks generated with redGEMX can be classified based on pathway favorability as it has been recently done in refs. 9 , 51 , 52 .

Biosynthetic reactions generation (Step 5)

Cellular metabolic functions, such as growth, structure maintenance, and reproduction, require the synthesis of several metabolites. In metabolic models, this is represented using the biomass reaction 53 , whose reactants, named biomass building blocks or BBBs, are the metabolites that the cell needs to survive and perform its functions. Therefore, the last step necessary for reconstructing the reduced models is the integration of the pathways necessary to synthesize the 37 BBBs that compose the defined biomass in Recon 2 and Recon 3D. Among them, 19 are uptaken directly from the extracellular medium or produced within the core network. To find the minimum number of reactions in the GEM that we need to add to the core network for the synthesis of the remaining 18 BBBs, we used lumpGEM (see “Methods”). Similarly to redGEMX, lumpGEM generates subnetworks that account for the synthesis, degradation, and balancing of all the by-products and cofactors required by the main pathway. The alternative subnetworks generated with lumpGEM can assess the flexibility of the cells to use alternative pathways to produce the BBBs, which can lead to survival in different conditions and drug resistance. Using lumpGEM, we calculated all the alternative subnetworks (set of reactions) of minimum size to capture the flexibility of the network for the biosynthesis of the BBBs (Fig.  4a , Supplementary Table  3 , and Supplementary Data  3 ). The reactions that compose each of these subnetworks were summed up together to form an overall reaction that represented the subnetwork. These lumped reactions were then added to the core network.

a Size of lumped reactions for Recon 2 and Recon 3D, and the corresponding number of alternatives to synthesize the BBBs that cannot be produced by the core nor uptaken from the extracellular medium. b , c Subnetwork for the synthesis of phosphatidylserine. Orange represents the metabolites from the core network. Blue represents the metabolites from the subnetwork for phosphatidylserine synthesis. Pink represents the extracellular metabolites. Phosphatidylserine synthesis starts from the core metabolites glycerol 3-phosphate (glyc3p_c), from glycolysis, and acetyl CoA (accoa_c), from TCA. In the first reaction, acetyl CoA is transformed into malonyl CoA (maloca_c). The next reaction (KAS8) represents the synthesis of palmitate (hdca_c) in the elongation cycle 74 . A CoA molecule is attached to palmitate to form palmitoyl CoA (pmtcoa_c), from which the two generic fatty acids are derived. These two generic fatty acids are attached to glycerol 3-phosphate to form lysophosphatidic acid (alpa_hs_c) and phosphatidic acid (pa_hs_c). Finally, serine (ser_L_c) is attached to phosphatidic acid to form phosphatidylserine (ps_hs_c). b Subnetwork from Recon 2 and corresponding lumped reaction. c The four alternative subnetworks of minimum size from Recon 3D. Phosphatidic acid can be produced with two generic fatty acids or with one generic fatty acid and the essential fatty acid linoleic acid (lnlc_e) (light blue reactions). Phosphatidylserine can be directly produced from phosphatidic acid by attaching serine (green reaction) or through the formation of phosphatidylcholine (red reaction) and then changing choline (chol_c) for serine (orange reaction).

The subnetworks generated with lumpGEM have the same size and number of alternatives in both Recon models for most of the BBBs, indicating that both models have the same level of flexibility for synthesizing the BBBs, with the exception of l -cysteine, dTTP and the purine nucleotides (ATP, GTP and their deoxy equivalents), cholesterol, and the phospholipids and sphingolipids. The core network of Recon 2 contains a reaction that produces l -cysteine, however, the core network of Recon 3D requires two reactions to produce it. The subnetworks that produce dTTP have the same size in both models, but a different number of alternatives. The subnetworks to produce the purine nucleotides have one more reaction and more alternatives in Recon 3D. Cholesterol is another BBB whose subnetworks agree in size for both models, but Recon 3D has more alternatives than Recon 2. The explosion of alternatives in Recon 3D is due to the parallel description of the synthesis of cholesterol in three compartments, namely cytosol, peroxisome, and endoplasmic reticulum. The differences in the lumped reactions for the phospholipids and sphingolipids between both models are due to the introduction of the essential fatty acid in their synthesis in Recon 3D.

As an example of the subnetworks that produce the BBBs, we show the synthesis of the phospholipid phosphatidylserine (Fig.  4b, c ). The standard KEGG pathway 54 for the synthesis of phosphatidylserine comprises four steps, wherein glycerol 3-phosphate is converted to lysophosphatidic acid, phosphatidic acid, CDP-diacylglycerol, and phosphatidylserine. In Recon 2, the subnetwork generated with lumpGEM for the synthesis of phosphatidylserine was composed of eight reactions. It included the KEGG pathway with the exception of the CDP-diacylglycerol intermediate, which was not connected to phosphatidylserine in the GEMs. Instead, phosphatidylserine was produced directly from phosphatidic acid by attaching serine. Additionally, the subnetwork contained the reactions required to generate from acetyl-CoA the fatty acids that would attach to glycerol 3-phosphate and to lysophosphatidic acid, which are important to consider for the final synthesis of phosphatidylserine. All the reactions involved in the synthesis of phosphatidylserine were lumped together in one reaction.

For Recon 3D, the phosphatidylserine synthesis subnetwork was generated with the same eight reactions, but in this case, four alternative subnetworks existed (Fig.  4c and Supplementary Table  4 ), indicating that Recon 3D has a higher flexibility in producing this BBB. The alternatives emerged from the presence of two reactions in Recon 3D that could be substituted by two other reactions in the subnetwork. One of these reactions arose from the participation of the essential fatty acid linoleate in phospholipid generation, resulting in an alternative form of synthesizing one of the tails of phosphatidic acid. Specifically, the reaction ARTPLM2, which converts palmitoyl CoA into a generic fatty acid, is not required, and instead, the essential fatty acid linoleate is transported from the extracellular medium, transformed into linoleyl-coA and attached to the lysophosphatidic acid to form phosphatidic acid. Because the core network of Recon 3D included a reaction that transforms phosphatidylcholine in phosphatidylserine, the other substitution occurred in the last step, where serine was replaced by choline and phosphatidylcholine was synthesized. The lumped reactions can be classified based on the thermodynamic favorability of their subnetworks, if metabolomics data are available, as in refs. 9 , 51 , 52 .

The analysis performed with lumpGEM allows to characterize and classify the metabolic pathways and their alternatives, leading to an in-depth understanding of the flexibility of metabolism. In the context of GEMs, such detailed analysis of the subnetworks is often a difficult task due to their large size and interconnectivity.

By applying the redHUMAN workflow, we reconstructed four reduced metabolic models for human metabolism (Table  1 ). Two of them have Recon 2 as the parent GEM, and the other two are generated from the Recon 3D GEM. For both GEMs, we generated one model with the minimum set of pathways required to simulate growth, that is, one lumped reaction per BBB with subnetworks of minimum size, and another model with higher flexibility containing all the alternative pathways of minimum size required to simulate growth. The reduced models have a thermodynamic coverage of more than 92% of the compounds and more than 61% of the reactions.

Data integration and metabolic tasks (Step 6)

Once the reduced models were generated, we investigated the metabolic tasks captured by the reduced models and we identified how the models should be curated to recover the tasks that they could not perform. First, we sequentially tested in the generated reduced models the thermodynamically feasibility of 57 metabolic tasks defined by Agren et al. 34 . The four models captured 45 of the 57 tasks, including rephosphorylation of nucleoside triphosphates, uptake of essential amino acids, de novo synthesis of nucleotides, key intermediates and cholesterol, oxidative phosphorylation, oxidative decarboxylation, and growth (Fig.  5a ).

a The 57 metabolic tasks tested in the generated reduced models. R2, R3: Recon 2, Recon 3D reduced model with one lumped reaction per BBB. R2s, R3s: Recon 2, Recon 3D reduced model with Smin. Classification of metabolic tasks in those captured by the models (green) and those not captured by the models (red). MT1: rephosphorylation of nucleoside triphosphates, MT2: de novo synthesis of nucleotides, MT3: uptake of essential amino acids, MT4: de novo synthesis of key intermediates, MT5: de novo synthesis of other compounds, MT6: protein turnover, MT7: electron transport chain and TCA, MT8: beta oxidation of fatty acids, MT9: de novo synthesis of phospholipids, MT10: vitamins and co-factors, MT11: growth. b Gene essentiality of the reduced models and their corresponding GEM. R2s has 829 genes associated to reactions, 37 of which are essential both in the reduced model and in Recon 2 and 12 are essential only in the reduced model. R3s has 828 genes associated to reactions, from which 23 are essential in both the reduced model and Recon 3D. The reduced model presents an additional 44 essential genes. c Thermo-flux variability analysis (TVA) for reactions in the reduced models. Orange represents fluxes in the reduced Recon 2 model and blue represents fluxes in the reduced Recon 3D models. The black lines correspond to the fluxes in the GEM.

The tasks not captured by the models encompassed the synthesis of protein from amino acids, beta oxidation of fatty acids, inositol uptake, and vitamin and co-factor metabolism. We classified the causes behind their limitation into two categories: (1) the model reconstruction, specifically the definition of the biomass, or (2) the reduction properties, that is, the subsystems included in the reduction and the representation of parts of the network as lumped reactions. To recover these tasks such that they are captured by the model, the following actions should be performed: the synthesis of proteins from amino acids and vitamin and co-factor metabolism can be recovered by modifying the biomass to account for their synthesis and utilization; the inclusion of lipid metabolism subsystems can recover the beta oxidation of fatty acids; and finally, the utilization of inositol can be recovered by adding the explicit reactions that compose the subnetworks, as it was found to be hidden in the lumped reactions of phosphatidyl-inositol. This demonstrates that redHUMAN allows to build reduced models consistent not only with the GEM but also with the metabolic tasks, and these models are suitable for targeted modifications and expansions.

We next demonstrated how generic reduced models were used to integrate data to study disease physiology. We first integrated experimental data from the NCI60 cell lines in the reduced models to define the physiology of leukemia cells. In particular, we considered the exometabolomics of the cell lines HL-60, K-562, MOLT-4, CCRF-CEM, RPMI-8226, and SR, which correspond to leukemia 40 , 55 . Additionally, we limited the maximal growth to the doubling time reported for leukemia cells, which is 0.035 h −1 , and we constrained according to literature values the maximum uptake rate of oxygen to 2 mmol·gDW −1 ·h −1   40 and the ATP maintenance to 1.07 mmol·gDW −1 ·h −1   56 (Supplementary Tables  5 and 6 ). We tested that all the models achieved the maximum growth when maximizing for the biomass reaction using TFA.

Next, to analyze the impact that the deletion of each gene had on the network, we performed in silico gene knockout by artificially removing a gene and measuring how the network was affected. The genes whose knockout prevented the synthesis of biomass could then be investigated as potential targets for limiting cell proliferation. The consistency of the workflow used to generate the reduced models ensures that they capture the essentiality from the GEM, that is, the genes that are part of the reduced models and are essential in the GEM they are also essential in the reduced model (Fig.  5b and Supplementary Tables  7 and 8 ). Furthermore, the reduced models allow the assignment of functionality to the essential genes using the lumped reactions. For example, the gene GART is associated with the enzymes phosphoribosylglycinamide formyltransferase, phosphoribosylglycinamide synthetase, and phosphoribosylaminoimidazole synthetase, which are all part of the subnetworks for the synthesis of the nucleotides ATP, GTP, dATP, and dGTP. Silencing this gene prevents the synthesis of these BBBs, and consequently, the models cannot synthesize biomass.

Finally, because the model reduction affects the flexibility of the network with respect to the GEM, we performed thermodynamic flux variability analysis (TVA) on the common reactions between the GEM and the reduced model. The top 20 reactions whose rate ranges changed the most in absolute value included reactions from glycolysis, the pentose phosphate pathway, folate metabolism, and nucleotide interconversion among others (Fig.  5c ). For reactions such as phosphoglycerate kinase (PGK), transaldolase (TALA), and methenyltetrahydrofolate cyclohydrolase (MTHFC), the ranges of reaction rates in the reduced model decreased with respect to the corresponding reaction rates in the GEM. Some reactions, such as nucleoside-diphosphate kinase (NDPK9), were bidirectional in the GEM and became unidirectional in the reduced models. On the other hand, there were also reactions such as fumarase, (FUM) lactate dehydrogenase (LDHL), or ribose-5-phosphate isomerase (RPI) whose flux ranges fully agreed between the reduced model and the GEM. Interestingly, if we look at the percentage of rate flexibility change, the reactions from the initial subsystems did not experience a large relative change in their rates, with the exception of the reactions whose participants are precursors for the lumped reactions of the BBBs as their reaction rates are now constrained closer to the physiological state. A final calibration of the models is done using the transcriptomics data from the NCI data repository ( https://www.ncbi.nlm.nih.gov/sites/GDSbrowser?acc=GDS4296 ) for the corresponding leukemia cell lines. We have identified that, in the four models presented in this study, over 99% of the enzymes with gene associations (more than 75% of the total enzymes) are expressed in the NCI60 leukemia cell lines (Supplementary Table  9 ). This suggests that the pathways selected for initializing and expanding the metabolic core network are highly relevant for the specific physiology, which are also consistent with the important pathways identified in the experimental and medical studies 46 , 48 , 57 .

Physiology analysis

redHUMAN helps to navigate large human genome-scale metabolic models to explore and classify the metabolic pathways that cells use to function and survive under specific conditions. The thermodynamic curation performed in the genome-scale models guarantees that the reactions obey the laws of thermodynamics, discarding possible pathways that would not be compatible with the bioenergetics of the cell. As an example of how thermodynamics reduces the space of solutions to the thermodynamically feasible pathways, we analyzed the flux variability with and without thermodynamic constraints in the Recon 3D reduced model that has all the alternative lumped reactions of minimum size (Smin). The reactions l -glutamate 5-semialdehyde dehydratase (from arginine metabolism) and l -glutamate 5-semialdehyde:NAD+ oxidoreductase (from urea cycle) are bidirectional when flux variability is performed without thermodynamics and become unidirectional when their thermodynamic information is taken into account. Therefore, integrating thermodynamic information reduces the space of reaction directionality and the physiological solution space, and it eliminates thermodynamic infeasible reactions, excluding some pathways.

The leukemia-specific models generated in this study are powerful tools to analyze how the metabolic pathways are altered with respect to other cancer cells or normal cells. In particular, we can analyze how leukemia cells utilize the nutrients available in the microenvironment to biosynthesize the precursors required for growth and cellular functionality. As an example, we identified the minimal number of reactions that are required for the synthesis of phosphatidyl-serine in the reduced Recon 3D model with all the alternative lumped reactions of minimum size. We found that at least 76 reactions should be active for the production of phosphatidyl-serine including the interactions with the extracellular medium, i.e., for some alternatives the uptake of glucose, histidine, linoleic acid, oxygen, and phosphate, and the secretion of succinate, ammonia, carbon dioxide, and water. The main pathways active within the subnetwork of 76 reactions are glycolysis, the citric acid cycle, serine metabolism, and the electron transport chain. This type of analysis will enlighten our knowledge on how cells adapt their metabolism to the microenvironment allowing researchers to hypothesize how and why the cancer cells change their expression profile to adapt and survive.

For a better understanding of the altered metabolisms that accompany many human diseases, we have herein presented a workflow to generate reduced models for common human GEMs that can reduce the complexity of these systems to the relevant processes to be studied, making detailed in silico analyses of metabolic changes possible.

During the last years, there has been an increased generation of metabolomics data that better study what is happening in the physiology of cell metabolism compared to other omics data. This has created a need to expand the classical constraint-based modeling methods to include metabolomics information. Our thermodynamic formulation and application of TFA 12 , 51 , 58 , 59 in redHUMAN allows to integrate endo- and exo-metabolomics in the models, constraining the concentration of the metabolites according to physiological data. The size of the model is directly related to the percentage of metabolites that need to be measured. Therefore, the continuous expansion in the size of genome-scale models increases the demand of larger sets of metabolomics, and such data are not always available. In addition, there is a community effort to expand constraint-based models to include information on enzyme abundancy relating the metabolic fluxes with enzymatic data and allowing to integrate transcriptomics and proteomics data into the models. These data are currently limited, but they can be continuously updated and integrated as they become available 60 , 61 .

Moreover, most of the existing methods to build context-specific models are data-driven, that is, the reduced models are extracted from a GEM by considering only the enzymes associated to highly expressed data, or literature-based pathways. Then, they include additional reactions that are required to simulate growth and cellular functions 33 , 34 , 62 . The main difficulty with these methods is the large amount of data required to fully characterize the initial set of reactions or core reactions. The lack of data could lead to unconnected parts and the impossibility to include reactions that could be important for the specific physiology, affecting the final model and the predictions.

redHUMAN reconstructs reduced models considering only the pathways of interest and their stoichiometric connectivity. The reduced models are built unbiased from the data, guaranteeing thermodynamic feasibility and consistency with the GEM and the metabolic tasks. The reduced models can then be used to construct context-specific models by integrating omics data, accommodating to also integrate partial data without sacrificing reactions from the network. Overall, the reduced size of the new models and their conceptual organization overcomes some of the main challenges in building genome-scale context-specific models as for example, the barrier of data network coverage. The reduced models generated with redHUMAN are powerful representations of the specific parts of the network, and they have promising applications as they are suitable to use with existing methods including MBA 62 , tINIT 34 , mCADRE 33 , uFBA 63 , GECKO 64 , ETFL 65 , TEX-FBA 66 , and IOMA 67 .

Based on our results, we propose the following approach to using these models as tools to explain and compare phenotypes. First, generate a reduced model around a desired set of subsystems and for a defined extracellular medium, and check that the model captures the metabolic tasks. Subsequently, build physiology-specific models by integrating experimental data into the reduced models. Then, test the consistency of the reduced network with respect to its parent GEM. Finally, integrate different sets of omics data, including expression, to compare different physiologies, such as diseased vs healthy or within several types of cancers. This approach will help to better investigate the alterations in metabolism that occur as diseases develop and progress. Moreover, the same procedure can be used to analyze systematically and consistently metabolic models for the same organism and to compare metabolic models of different organisms, enhancing our understanding of their similarities and differences.

Throughout this paper, we have considered a specific set of subsystems, a specific medium, and the biomass definition from the GEMs. In the future, the reduced models could be further expanded to include other pathways, a more complex medium, or more biomass components. To introduce new subsystems or pathways into the core network, redGEM should be run to find the pairwise connections between the added pathways and the rest of the core. For an expansion of the medium, redGEMX would find the connections necessary for using the new extracellular metabolites. In a similar manner, a further curation of the biomass reaction could increase the number of BBBs, requiring lumpGEM to be run to find the biosynthesis pathways for those compounds. If a higher consistency was required between the GEM and the corresponding reduction, we could find the reactions missing in the reduced model to satisfy that condition. Moreover, we have selected a set of metabolic tasks to test the generated reduced model based on the definition within the original GEM. However, these sets of tasks can be expanded or redefined according to the needs of the specific studies, which can be based on expert knowledge or experimental data, as done in ref. 68 .

Furthermore, in this study, we have used metabolomics, proteomics, and growth data from the NCI60 cell lines to define a generic physiology for leukemia cells. The core networks of the reduced models are structurally the same across growth conditions and depend only on the structure of the corresponding GEMs. Therefore, these generic models are robust to variations in growth or data for the same physiology, and thus data for individual leukemia cell lines can be used without changing the workflow. However, if there are important differences in the data, for example across different physiological conditions, the authors suggest running the lumpGEM workflow with data integration and generate alternative subnetworks and lumped reactions, which in turn will capture the different flux profiles for each physiological state.

Overall, our analysis demonstrates how redHUMAN facilitates the characterization of differences in metabolic pathways across models and phenotypes.

Thermodynamic curation of the genome-scale models (GEMs)

The thermodynamic curation of the human GEMs Recon 2 and Recon 3D aims to include thermodynamic information, i.e., the Gibbs free energy of formation for the compounds and the corresponding error for the estimation, into the model. The workflow to obtain this information is as follows.

We first used MetaNetX ( http://www.metanetx.org ) 69 to annotate the compounds of the GEMs with identifiers from SEED 70 , KEGG 54 , CHEBI 71 , and HMDB 72 . We then used Marvin (version 18.1, 2018, ChemAxon http://www.chemaxon.com ) to transform the compound structures (canonical SMILES) into their major protonation states at pH 7 and to generate MDL Molfiles. We used the MDL Molfiles and the Group Contribution Method to estimate the standard Gibbs free energy of the formation of the compounds as well as the error of the estimation 59 .

Since the model for Recon 3D already incorporates the structure for 82% of the metabolites in the form of SMILES, we used those SMILES and followed the previous workflow from the point of obtaining the major forms at pH 7 using Marvin.

Furthermore, we have integrated in the models the thermodynamic properties for the compartments of human cells, including, pH, ionic strength, membrane potentials, and generic compartment concentration ranges from 10 pM to 0.1 M (Supplementary Table  10 ).

Thermodynamics-based flux analysis (TFA)

TFA estimates the feasible flux and concentration space according to the laws of thermodynamics 11 , 12 , 13 . TFA is formulated as a mixed-integer linear programming (MILP) problem that incorporates the thermodynamic constraints to the original FBA problem. The Gibbs free energy of the elemental and charge balanced reactions is calculated as a function of the standard transformed Gibbs free energy of formation (depending on pH and ionic strength) and the concentrations of the products and reactants.

Considering a network with m metabolites and n reactions, the Gibbs free energy, \(\Delta _r{\mathrm{G}}_i^\prime ,\) for reaction i is:

where \(i = 1, \ldots ,n,j = 1, \ldots ,m.\) n i,j is the stoichiometric coefficient of compound j in reaction i ; \(\Delta _f{\mathrm{G}}_j^{\prime o}\) is the standard Gibbs free energy of formation of compound j ; x j is the concentration of the compound j , R is the ideal gas constant, \(R = 8.31 \cdot 10^{ - 3}\frac{{\mathrm{KJ}}}{{\mathrm{K}}\;{\mathrm{mol}}}\) , and T is the temperature. In this case, T  = 298 K.

The value of the Gibbs free energy determines the directionality of the corresponding reaction and the thermodynamically feasible pathways. With this formulation, we included the concentrations of the metabolites as variables in the mathematical formulation. TFA allows the integration of metabolomics data into the model.

Characterizing the extracellular in silico minimal media (iMM)

iMM is formulated as a MILP problem that introduces new variables and constraints to the TFA problem to find the minimum set of extracellular metabolites necessary to simulate growth or a specific metabolic task with the GEM 37 , 38 . iMM identifies the minimum number of boundary reactions (uptakes and secretions) that need to be active. The method defines new binary variables in the TFA problem that represent the state of each boundary reaction, active or inactive. New constraints link the new binary variables to the corresponding reaction rates such that if the reaction is inactive, then it should not carry flux. The objective of the problem is to maximize the number of inactive reactions.

Assuming a network with m metabolites and n reactions, the mathematical formulation of the iMM problem is the following:

where n b is the total number of boundary reactions in the model, z k are new binary variables for all the boundary reactions, S is the stoichiometric matrix, v are the net fluxes for all the reactions and \(v_i^F,v_i^R\) are the corresponding net-forward and net-reverse fluxes, so that, \(v_{i} = v_{i}^{F} - v_{i}^{R},\;\;{\mathrm{for}}\;{\mathrm{all}}\;\; i = 1, \ldots ,n\) . \({\boldsymbol{v}}_L\) and \({\boldsymbol{v}}_U\) are the lower and upper bound, respectively, for all the reactions in the network. \(\Delta _r{\mathbf{G}}^\prime\) is the Gibb’s free energy of the reactions defined in TFA. \({\mathbf{b}}_{}^F\) and \({\mathbf{b}}_{}^R\) are the binary variables for the forward or reverse fluxes of all the reactions (coupled to TFA). M is a big constant (bigger than all upper bounds) and C is an arbitrary large number. In this case, if \(z_k = 0\) , then reaction k is active.

redGEM, redGEMX, and lumpGEM

The redGEM, redGEMX, and lumpGEM algorithms seek to generate systematic reductions of the GEMs starting from chosen subsystems (or lists of reactions and metabolites, such as the synthesis pathway of a target metabolite), based on the studied physiology and the specific parts of the metabolism that are of interest.

redGEM is a published algorithm 35 that extracts the reactions that pairwise-connect the initial subsystems from the GEM, generating a connected network named the core network.

The inputs for redGEM are (i) the GEM, (ii) the starting subsystems or an initial set of reactions, (iii) the extracellular medium metabolites, (iv) a list with the GEM cofactor pairs, and (v) the desired degree of connectivity. The algorithm then performs an expansion (by graph search) of the starting subsystems by finding the reactions that pairwise-connect the subsystems up to the selected degree (see ref. 35 for further details). For example, for a degree equal to 2, it will connect the metabolites from the starting subsystem that are one and two reactions away in the GEM.

redGEMX is a formulated algorithm that finds the pathways in the GEM that connect the extracellular medium to the core network generated with redGEM (Fig.  6 ). These pathways are added to the core network.

a Classification of the reactions from the GEM into core (green) and non-core reactions (orange), and classification of the extracellular metabolites from the GEM into those that are part of the medium that we want to connect (blue), those that are present in the core (pink), and the others (gray). The algorithm will block the non-core reactions that involve only extracellular metabolites as well as the boundary and transport reactions of the metabolites that are not part of the medium (gray). b The algorithm finds the minimal set of reactions that are required to connect each of the medium metabolites (blue) to the core network, uses the core network to balance the reactions, and secretes metabolites from the medium (blue or pink).

The redGEMX method involves five steps:

Classify the extracellular metabolites of the GEM into 3 classes:

Those that are part of the medium that we want to connect,

Those that are already present in the inter-connected subsystems network,

Those that do not belong to (a) nor (b).

Classify the reactions from the GEM into 2 classes:

Those that belong to the inter-connected subsystems network (core-reactions),

those that do not belong to the inter-connected subsystems network (non-core reactions).

Block the flux through the reactions in the GEM that involve only extracellular metabolites.

Block the flux through the boundary reactions of other metabolites in the GEM (1c). Steps (3) and (4) guarantee that the subnetwork reaches the core network.

Force the uptake of a medium metabolite (1a, one-by-one) and minimize the number of non-core reactions (2b) required to connect this extracellular metabolite to any core metabolite participating in a core reaction (2a). Note that the subnetwork will contain any reaction required to balance the by-products secreted by the subnetwork and/or the core network.

The redGEMX is a MILP problem that is formulated as follows:

Consider the TFA problem of the model that we want to reduce.

Create binary variables z i for each non-core reaction (2b). Non-core reactions are denoted as R nc .

Generate a constraint that controls the flux for each non-core reaction:

where b F and b R are the binary variables for the forward and reverse fluxes of all the reactions (coupled to the TFA constraints); when z i  = 1, the corresponding reaction is inactive.

Build the following MILP problem for each extracellular medium metabolite (1a)

subject to:

where v eM,j is the flux of the j th extracellular medium metabolite (1a), and c is a small number.

lumpGEM is a published algorithm 36 that generates elementally balanced lumped reactions for the synthesis of the biomass building blocks (BBBs). Using a MILP formulation, lumpGEM identifies the smallest subnetwork (minimum number of reactions from the GEM) required to produce each BBB from metabolites that belong to the core network using reactions from the GEM that are not part of the core. With this formulation, we can identify all the alternative subnetworks (of minimal size or larger) for the synthesis of each BBB (one by one). lumpGEM generates, for each BBB, an overall lumped reaction by adding all the reactions that constitute each subnetwork (see ref. 36 for further details). Note here, different subnetworks can give rise to the same overall lumped reaction. This implies that although we produce all the alternative subnetworks with their associated lumped reactions, only the unique lumped reactions will be added to the final reduction.

The simulations of this article have been done with Matlab 2017b and CPLEX 12.7.1. Escher 73 has been used to draw the subnetworks in the figures.

Reporting summary

Further information on research design is available in the  Nature Research Reporting Summary linked to this article.

Data availability

The models generated in this work and the data integrated in the models to define the physiology of leukemia cells are available under the APACHE 2.0 license at https://github.com/EPFL-LCSB/redhuman .

Code availability

The scripts to generate the results for this paper are available under the APACHE 2.0 license at https://github.com/EPFL-LCSB/redhuman . The code for TFA is available at https://github.com/EPFL-LCSB/mattfa . The code to reduce the human GEMs (redGEM), to connect the extracellular medium to the core (redGEMX), and to generate the biosynthetic lumped reactions (lumpGEM) are available at https://github.com/EPFL-LCSB/redgem .

Change history

23 july 2020.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144 , 646–674 (2011).

CAS   PubMed   Google Scholar  

Mardinoglul, A. & Nielsen, J. New paradigms for metabolic modeling of human cells. Curr. Opin. Biotechnol. 34 , 91–97 (2015).

Google Scholar  

Orth, J. D. et al. A comprehensive genome-scale reconstruction of Escherichia coli metabolism. Mol. Syst. Biol. 7 , 535 (2011).

PubMed   PubMed Central   Google Scholar  

Kerkhoven, E. J., Pomraning, K. R., Baker, S. E. & Nielsen, J. Regulation of amino-acid metabolism controls flux to lipid accumulation in Yarrowia lipolytica . npj Syst. Biol. Appl. 2 , 16005 (2016).

PubMed   PubMed Central   CAS   Google Scholar  

Heavner, B. D., Smallbone, K., Price, N. D. & Walker, L. P. Version 6 of the consensus yeast metabolic network refines biochemical coverage and improves model performance. Database https://doi.org/10.1093/database/bat059 (2013).

Mardinoglu, A., Gatto, F. & Nielsen, J. Genome-scale modeling of human metabolism a systems biology approach. Biotechnol. J. 8 , 985 (2013).

PubMed   CAS   Google Scholar  

O’Brien, E. J., Monk, J. M. & Palsson, B. O. Using genome-scale models to predict biological capabilities. Cell 161 , 971–987 (2015).

Orth, J. D., Thiele, I. & Palsson, B. O. What is flux balance analysis? Nat. Biotechnol. 28 , 245–248 (2010).

Henry, C. S., Broadbelt, L. J. & Hatzimanikatis, V. Thermodynamics-based metabolic flux analysis. Biophys. J. 92 , 1792–1805 (2007).

ADS   PubMed   CAS   Google Scholar  

Soh, K. C. & Hatzimanikatis, V. Network thermodynamics in the post-genomic era. Curr. Opin. Microbiol. 13 , 350–357 (2010).

Ataman, M. & Hatzimanikatis, V. Heading in the right direction: thermodynamics-based network analysis and pathway engineering. Curr. Opin. Biotechnol. 36 , 176–182 (2015).

Soh, K. C. & Hatzimanikatis, V. Constraining the flux space using thermodynamics and integration of metabolomics data. In Methods in Molecular Biology 49–63 (2014).

Salvy, P. et al. pyTFA and matTFA: a Python package and a Matlab toolbox for thermodynamics-based flux analysis. Bioinformatics 35 , 167–169 (2018).

PubMed Central   Google Scholar  

Zhang, C. et al. ESS: a tool for genome-scale quantification of essentiality score for reaction/genes in constraint-based modeling. Front. Physiol. 9 , 1355 (2018).

ADS   PubMed   PubMed Central   Google Scholar  

Lewis, N. E., Nagarajan, H. & Palsson, B. O. Constraining the metabolic genotype–phenotype relationship using a phylogeny of in silico methods. Nat. Rev. Microbiol. 10 , 291–305 (2012).

Nielsen, J. Systems biology of metabolism. Annu. Rev. Biochem. 86 , 245–275 (2017).

Bordbar, A. & Palsson, B. O. Using the reconstructed genome-scale human metabolic network to study physiology and pathology. J. Intern. Med. 271 , 131–141 (2012).

Zhang, C. & Hua, Q. Applicatons of genome-scale metabolic models in biotechnology and systems medicine. Front. Physiol. 6 , 413 (2016).

Chakrabarti, A., Miskovic, L., Soh, K. C. & Hatzimanikatis, V. Towards kinetic modeling of genome-scale metabolic networks without sacrificing stoichiometric, thermodynamic and physiological constraints. Biotechnol. J. 8 , 1043–1057 (2013).

Miskovic, L., Tokic, M., Fengos, G. & Hatzimanikatis, V. Rites of passage: requirements and standards for building kinetic models of metabolic phenotypes. Curr. Opin. Biotechnol. 36 , 146–153 (2015).

Srinivasan, S., Cluett, W. R. & Mahadevan, R. Constructing kinetic models of metabolism at genome-scales: a review. Biotechnol. J. 10 , 1345–1359 (2015).

Stanford, N. J. et al. Systematic construction of kinetic models from genome-scale metabolic networks. PLoS ONE 8 , e79195 (2013).

Duarte, N. C. et al. Global reconstruction of the human metabolic network based on genomic and bibliomic data. Proc. Natl Acad. Sci. USA 104 , 1777–1782 (2007).

ADS   PubMed   CAS   PubMed Central   Google Scholar  

Thiele, I. & Palsson, B. O. A protocol for generating a high-quality genome-scale metabolic reconstruction. Nat. Protoc. 5 , 93–121 (2010).

Mardinoglu, A. et al. Genome-scale metabolic modelling of hepatocytes reveals serine deficiency in patients with non-alcoholic fatty liver disease. Nat. Commun. 5 , 3083 (2014).

ADS   PubMed   Google Scholar  

Thiele, I. et al. A community-driven global reconstruction of human metabolism. Nat. Biotechnol. 31 , 419 (2013).

Swainston, N. et al. Recon 2.2: from reconstruction to model of human metabolism. Metabolomics 12 , 109 (2016).

Brunk, E. et al. Recon3D enables a three-dimensional view of gene variation in human metabolism. Nat. Biotechnol. 36 , 272 (2018).

Ebrahim, A. et al. Multi-omic data integration enables discovery of hidden biological regularities. Nat. Commun. 7 , 13091 (2016).

ADS   PubMed   PubMed Central   CAS   Google Scholar  

Aurich, M. K. et al. Prediction of intracellular metabolic states from extracellular metabolomic data. Metabolomics 11 , 603–619 (2015).

Schmidt, B. J. et al. GIM(3)E: condition-specific models of cellular metabolism developed from metabolomics and expression data. Bioinformatics 29 , 2900–2908 (2013).

Becker, S. A. & Palsson, B. O. Context-specific metabolic networks are consistent with experiments. PLoS Comput. Biol. 4 , e1000082 (2008).

ADS   MathSciNet   PubMed   PubMed Central   Google Scholar  

Wang, Y., Eddy, J. A. & Price, N. D. Reconstruction of genome-scale metabolic models for 126 human tissues using mCADRE. BMC Syst. Biol. 6 , 153 (2012).

Agren, R. et al. Identification of anticancer drugs for hepatocellular carcinoma through personalized genome-scale metabolic modeling. Mol. Syst. Biol. 10 , 721 (2014).

Ataman, M., Gardiol, D. F. H., Fengos, G. & Hatzimanikatis, V. redGEM: systematic reduction and analysis of genome-scale metabolic reconstructions for development of consistent core metabolic models. PLoS Comput. Biol. 13 , e1005444 (2017).

Ataman, M. & Hatzimanikatis, V. lumpGEM: systematic generation of subnetworks and elementally balanced lumped reactions for the biosynthesis of target metabolites. PLoS Comput. Biol. 13 , e1005513 (2017).

Tymoshenko, S. et al. Metabolic needs and capabilities of Toxoplasma gondii through combined computational and experimental analysis. PLoS Comput. Biol. 11 , e1004261 (2015).

Chiappino-Pepe, A., Tymoshenko, S., Ataman, M., Soldati-Favre, D. & Hatzimanikatis, V. Bioenergetics-based modeling of Plasmodium falciparum metabolism reveals its essential genes, nutritional requirements, and thermodynamic bottlenecks. PLoS Comput. Biol. 13 , e1005397 (2017).

Folger, O. et al. Predicting selective drug targets in cancer through metabolic networks. Mol. Syst. Biol. 7 , 501 (2011).

Zielinski, D. C. et al. Systems biology analysis of drivers underlying hallmarks of cancer cell metabolism. Sci. Rep. 7 , 41241 (2017).

Shlomi, T., Benyamini, T., Gottlieb, E., Sharan, R. & Ruppin, E. Genome-scale metabolic modeling elucidates the role of proliferative adaptation in causing the Warburg effect. PLoS Comput. Biol. 7 , e1002018 (2011).

Warburg, O. On the metabolism of cancer cells. Naturwissenschaften 12 , 1131–1137 (1924).

ADS   CAS   Google Scholar  

Fenninger, L. D. & Mider, G. B. Energy and nitrogen metabolism in cancer. Adv. Cancer Res. 2 , 229–251 (1954).

Cairns, R. A., Harris, I. S. & Mak, T. W. Regulation of cancer cell metabolism. Nat. Rev. Cancer 11 , 85–95 (2011).

Dang, C. V. Links between metabolism and cancer. Genes Dev. 26 , 877–890 (2012).

Di Filippo, M. et al. Zooming-in on cancer metabolic rewiring with tissue specific constraint-based models. Comput. Biol. Chem. 62 , 60–69 (2016).

PubMed   Google Scholar  

Hosios, A. M. et al. Amino acids rather than glucose account for the majority of cell mass in proliferating mammalian cells. Dev. Cell 36 , 540–549 (2016).

Hay, N. Reprogramming glucose metabolism in cancer: can it be exploited for cancer therapy? Nat. Rev. Cancer 16 , 635–649 (2016).

Wise, D. R. & Thompson, C. B. Glutamine addiction: a new therapeutic target in cancer. Trends Biochem. Sci. 35 , 427–433 (2010).

Kanarek, N. et al. Histidine catabolism is a major determinant of methotrexate sensitivity. Nature 559 , 632–636 (2018).

Henry, C. S., Broadbelt, L. J. & Hatzimanikatis, V. Discovery and analysis of novel metabolic pathways for the biosynthesis of industrial chemicals: 3-hydroxypropanoate. Biotechnol. Bioeng. 106 , 462–473 (2010).

Du, B., Zielinski, D. C., Monk, J. M. & Palsson, B. O. Thermodynamic favorability and pathway yield as evolutionary tradeoffs in biosynthetic pathway choice. Proc. Natl Acad. Sci. USA 115 , 11339–11344 (2018).

PubMed   CAS   PubMed Central   Google Scholar  

Feist, A. M. & Palsson, B. O. The biomass objective function. Curr. Opin. Microbiol. 13 , 344–349 (2010).

Kanehisa, M., Furumichi, M., Tanabe, M., Sato, Y. & Morishima, K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. 45 , D353–D361 (2017).

Jain, M. et al. Metabolite profiling identifies a key role for glycine in rapid cancer cell proliferation. Science 336 , 1040–1044 (2012).

Kilburn, D. G., Lilly, M. D. & Webb, F. C. The energetics of mammalian cell growth. J. Cell Sci. 4 , 645–654 (1969).

DeBerardinis, R. J. & Chandel, N. S. Fundamentals of cancer metabolism. Sci. Adv. 2 , e1600200 (2016).

Hatzimanikatis, V. et al. Exploring the diversity of complex metabolic networks. Bioinformatics 21 , 1603–1609 (2005).

Jankowski, M. D., Henry, C. S., Broadbelt, L. J. & Hatzimanikatis, V. Group contribution method for thermodynamic analysis of complex metabolic networks. Biophys. J. 95 , 1487–1499 (2008).

Wilhelm, M. et al. Mass-spectrometry-based draft of the human proteome. Nature 509 , 582–587 (2014).

Uhlen, M. et al. Proteomics. Tissue-based map of the human proteome. Science 347 , 1260419 (2015).

Jerby, L., Shlomi, T. & Ruppin, E. Computational reconstruction of tissue-specific metabolic models: application to human liver metabolism. Mol. Syst. Biol. 6 , 401 (2010).

Bordbar, A. et al. Elucidating dynamic metabolic physiology through network integration of quantitative time-course metabolomics. Sci. Rep. 7 , 46249 (2017).

Sanchez, B. J. et al. Improving the phenotype predictions of a yeast genome-scale metabolic model by incorporating enzymatic constraints. Mol. Syst. Biol. 13 , 935 (2017).

Salvy, P. & Hatzimanikatis, V. The ETFL formulation allows multi-omics integration in thermodynamics-compliant metabolism and expression models. Nat. Commun.   11 , 30 (2020).

Pandey, V., Hernandez Gardiol, D., Chiappino-Pepe, A. & Hatzimanikatis, V. TEX-FBA: a constraint-based method for integrating gene expression, thermodynamics, and metabolomics data into genome-scale metabolic models. bioRxiv 101101/536235 (2019).

Yizhak, K., Benyamini, T., Liebermeister, W., Ruppin, E. & Shlomi, T. Integrating quantitative proteomics and metabolomics with a genome-scale metabolic network model. Bioinformatics 26 , i255–i260 (2010).

Richelle, A., Chiang, A. W. T., Kuo, C. C. & Lewis, N. E. Increasing consensus of context-specific metabolic models by integrating data-inferred cell functions. PLoS Comput. Biol. 15 , e1006867 (2019).

Moretti, S. et al. MetaNetX/MNXref—reconciliation of metabolites and biochemical reactions to bring together genome-scale metabolic networks. Nucleic Acids Res. 44 , D523–D526 (2016).

Overbeek, R. et al. The subsystems approach to genome annotation and its use in the project to annotate 1000 genomes. Nucleic Acids Res. 33 , 5691–5702 (2005).

Hastings, J. et al. ChEBI in 2016: Improved services and an expanding collection of metabolites. Nucleic Acids Res. 44 , D1214–D1219 (2016).

Wishart, D. S. et al. HMDB 4.0: the human metabolome database for 2018. Nucleic Acids Res. 46 , D608–D617 (2018).

King, Z. A. et al. Escher: a web application for building, sharing, and embedding data-rich visualizations of biological pathways. PLoS Comput. Biol. 11 , e1004321 (2015).

Harayama, T. & Riezman, H. Understanding the diversity of membrane lipid composition. Nat. Rev. Mol. Cell Biol. 19 , 281–296 (2018).

Download references

Acknowledgements

We would like to thank Dr. G. Fengos, Dr. J. P. Vieira and Dr. D. R. Weilandt for constructive discussions and feedback; and Dr. K. Butler for valuable comments proofreading this paper. This project has received financial support from the European Unionʼs Horizon 2020 research and innovation programme under the Marie Skłodowska Curie Grant Agreement No. 675585 SyMBioSys, and under Grant Agreement No. 686070, and from the RTD project MicroScapeX (grant 2013/158) funded by SystemsX.ch, the Swiss Initiative for Systems Biology evaluated by the Swiss National Science Foundation.

Author information

Authors and affiliations.

Laboratory of Computational Systems Biotechnology, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland

Maria Masid & Vassily Hatzimanikatis

Computational and Systems Biology, Biozentrum, University of Basel, Basel, Switzerland

Meric Ataman

You can also search for this author in PubMed   Google Scholar

Contributions

M.M., M.A., and V.H. conceptualized the paper, defined the methodology, and performed the investigation. M.M. and M.A. designed the code and performed the simulations. M.M. and V.H. wrote the manuscript and all authors edited and reviewed the manuscript. V.H. acquired the funding and the resources, and supervised the project.

Corresponding author

Correspondence to Vassily Hatzimanikatis .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information, supplementary data 1, supplementary data 2, supplementary data 3, reporting summary, description of additional supplementary files, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and Permissions

About this article

Cite this article.

Masid, M., Ataman, M. & Hatzimanikatis, V. Analysis of human metabolism by reducing the complexity of the genome-scale models using redHUMAN. Nat Commun 11 , 2821 (2020). https://doi.org/10.1038/s41467-020-16549-2

Download citation

Received : 23 September 2019

Accepted : 07 May 2020

Published : 04 June 2020

DOI : https://doi.org/10.1038/s41467-020-16549-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Dynamic partitioning of branched-chain amino acids-derived nitrogen supports renal cancer progression.

• Marco Sciacovelli
• Aurelien Dugourd
• Christian Frezza

Nature Communications (2022)

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: Cancer newsletter — what matters in cancer research, free to your inbox weekly.

Metabolic Equivalent Task (MET)

Measuring physical activity using mets.

Physical activity is measured in metabolic equivalents task units (METs).  One MET is defined as the energy it takes a 70 kg man to sit quietly. In other words, one MET is defined as the oxygen consumption of a 70-kg man at rest.

Another rendition of this definition is, “A MET is defined as the resting metabolic rate, that is, the amount of oxygen consumed at rest, sitting quietly in a chair, approximately 3.5 ml 0 2 /kg/min (1.2 kcal/min for a 70-kg person).”

For the average adult (70kg person), this is about one calorie ( i.e. 1kcal ) for every one kilogram (2.2 pounds) of body weight per hour. That is someone who weighs 70 kg (154 lbs) would burn about 70 calories (70 kcal) in one hour while sitting or sleeping.

The U.S. Dept of Health & Human Services says, “1 MET is the rate of energy expenditure while sitting at rest. It is taken by convention to be an oxygen uptake of 3.5 milliliters per kilogram of body weight per minute. Physical activities frequently are classified by their intensity using the MET as a reference” (HHS 2008, 53).

• Light-intensity activities are defined as < 3 METs.
• Moderate-intensity activities are defined as ≥ 3.0 METs to < 6 METs. Walking at 3.0 miles per hour requires 3.3 METs of energy expenditure and is therefore considered a moderate-intensity activity.
• Vigorous-intensity activities are defined as ≥ 6.0 METs. Running at 10 minutes per mile (6.0 mph) is a 10 MET activity and is therefore classified as vigorous-intensity (HHS 2008, 55).

Moderate-intensity activities are activities that get a person moving fast enough or strenuously enough to burn off three to six times as much energy per minute as a 70 kg person would burn when sitting quietly. In other words, these exercises require 3 to 6 METs. Vigorous-intensity activities burn more than 6 METs.

Limitations to using METs to measure physical activity/exercise intensity.

• All METs are not created equal. That is, all METs do not bring the same health benefits. “Physical activity at work is shown to not provide the same health gain as recreational PA,” (Holtermann 2018, 1).
• Measuring with METs does not consider personal characteristics. “It does not consider the fact that some people have a higher level of fitness than others. Thus, walking at 3 to 4 miles per hour is considered to require 4 METs and to be a moderate-intensity activity, regardless of who is doing the activity—a young marathon runner or a 90-year-old grandmother. As you might imagine, a brisk walk would likely be an easy activity for the marathon runner, but a very hard activity for the grandmother” (Harvard, n.d.).

The table below shows different activity levels and corresponding METs.

MET = metabolic equivalent;

Click to access METs-Metabolicequivalents.pdf

www.hsph.harvard.edu/nutritionsource/mets-activity-table/, Last Accessed 5/23/2019

U.S. Department of Health and Human Services.  2008 Physical Activity Guidelines for Americans. https://health.gov/paguidelines/pdf/paguide.pdf. Last Accessed 5/23/2019

• Well Child Check
• Well Adult Exam
• DERMATOLOGY
• Forms & Charts
• History Templates
• PE Templates
• Quick Physical Exam
• A good Assessment & Plan
• DIFFERENTIAL DIAGNOSIS
• Clinic Quick Reference
• Hospital Quick Reference
• InfoGraphics
• Book Chapter in Time of Care
• Coding Guide
• Office Management

• Pop culture
• Writing tips
• Daily Crossword
• Word Puzzle
• Word Finder
• Word of the Day
• Synonym of the Day
• Word of the Year
• Language stories
• All featured
• Gender and sexuality
• All pop culture
• Grammar Coach ™
• Writing hub
• Grammar essentials
• Commonly confused
• All writing tips

of, relating to, or affected by metabolism .

undergoing metamorphosis.

Origin of metabolic

Other words from metabolic.

• met·a·bol·i·cal·ly, adverb
• hy·per·met·a·bol·ic, adjective

Words Nearby metabolic

• metabolic heat
• metabolic pathway
• metabolic syndrome

Dictionary.com Unabridged Based on the Random House Unabridged Dictionary, © Random House, Inc. 2023

How to use metabolic in a sentence

Our bodies release certain metabolites—products of our metabolic activities.

Finding the genetic basis for the animal’s metabolic feats could clarify the mitochondrial genome’s function, helping to find treatments for human metabolic diseases.

Basal metabolic rate is the minimum amount of calories required for basic functions at rest.

In birds, metabolic rates played a massive role for birds, while isolating mechanisms like body hair or fat reserves had a prominent role for mammals.

Adipose tissue is therefore important for good metabolic health.

What if they were to measure body composition or hormone levels or metabolic rate?

A “set point” would determine weight by regulating metabolic activity, hunger and even the desire to do exercise.

After all, here we are, in the middle of a global obesity crisis; and there they are running a metabolic lab on television.

They are much more likely to have metabolic syndrome—a condition that puts you at high risk for diabetes and heart disease.

Eventually, he developed metabolic syndrome—a condition that presages diabetes and heart disease.

It stimulates metabolic activity of tissue cells and secures more complete oxidation of energy-yielding elements.

Not as our bodies do the stars need continual metabolic replacing food.

(b) The medium became toxic from the accumulation of metabolic products.

A still more important difference is that the metabolic force is peculiar to the cell.

Now, an indispensable condition of the ripening of the ovum in the female organism is that the metabolic process shall be normal.

Metabolic equivalents (METS) in exercise testing, exercise prescription, and evaluation of functional capacity

Affiliation.

• 1 Department of Kinanthropology, School of Human Kinetics, University of Ottawa, Canada.
• PMID: 2204507
• DOI: 10.1002/clc.4960130809

One metabolic equivalent (MET) is defined as the amount of oxygen consumed while sitting at rest and is equal to 3.5 ml O2 per kg body weight x min. The MET concept represents a simple, practical, and easily understood procedure for expressing the energy cost of physical activities as a multiple of the resting metabolic rate. The energy cost of an activity can be determined by dividing the relative oxygen cost of the activity (ml O2/kg/min) x by 3.5. This article summarizes and presents energy expenditure values for numerous household and recreational activities in both METS and watts units. Also, the intensity levels (in METS) for selected exercise protocols are compared stage by stage. In spite of its limitations, the MET concept provides a convenient method to describe the functional capacity or exercise tolerance of an individual as determined from progressive exercise testing and to define a repertoire of physical activities in which a person may participate safely, without exceeding a prescribed intensity level.

Publication types

• Activities of Daily Living*
• Energy Metabolism*
• Exercise Test
• Exercise Therapy*
• Oxygen Consumption*

• Reference Manager
• Simple TEXT file

People also looked at

Mini review article, metabolic equivalents of task are confounded by adiposity, which disturbs objective measurement of physical activity.

• 1 Department of Clinical Physiology and Nuclear Medicine, Kuopio University Hospital, Kuopio, Finland
• 2 Institute of Biomedicine/Physiology, School of Medicine, University of Eastern Finland, Kuopio, Finland

Physical activity refers any bodily movements produced by skeletal muscles that expends energy. Hence the amount and the intensity of physical activity can be assessed by energy expenditure. Metabolic equivalents of task (MET) are multiplies of the resting metabolism reflecting metabolic rate during exercise. The standard MET is defined as 3.5 ml/min/kg. However, the expression of energy expenditure by body weight to normalize the size differences between subjects causes analytical hazards: scaling by body weight does not have a physiological, mathematical, or physical rationale. This review demonstrates by examples that false methodology may cause paradoxical observations if physical activity would be assessed by body weight scaled values such as standard METs. While standard METs are confounded by adiposity, lean mass proportional measures of energy expenditure would enable a more truthful choice to assess physical activity. While physical activity as a behavior and cardiorespiratory fitness or adiposity as a state represents major determinants of public health, specific measurements of health determinants must be understood to enable a truthful evaluation of the interactions and their independent role as a health predictor.

Introduction

This review discusses critically the metabolic equivalent of task (MET) ( Weir, 1949 ; Jetté et al., 1990 ), and its applicability to measure energy expenditure or to assess amount and intensity of physical activity. Physical activity is a major public health determinant ( Lim et al., 2012 ) and physical activity introduces public health promoting potential. Physical activity research consists also behavioral aspect, which has been taken into account in elaborate review by Hills et al. (2014) , but from physiological point of view specific and truthful measurement is needed to assess physical activity related health interactions. While physiologists are familiar with physiological relevance and potential problems of body size differences related scaling ( Tanner, 1949 ; Weir, 1949 ; Hoppeler and Weibel, 2005 ; Lorenzo and Babb, 2012 ; Tompuri et al., 2014a ), there are physiological procedures which have been adapted to be used at the population level in health sciences. For example scaling by body weight or METs has often been used when aiming to assess physical activity ( Strath et al., 2001 , 2013 ; Brage et al., 2004 , 2005 ; Corder et al., 2005 ; Crouter et al., 2008 ; US Department, 2008 ; Warren et al., 2010 ; World Health Organization, 2010 ). This paper enlightens the background of the MET as well as rational body size-related scaling, and demonstrates how methodological confounding may affect physical activity assessment, which in turn may bias further analytical conclusions.

Physiological Basis for Measurement of Physical Activity

The definition of the physical activity as “ any bodily movements produced by skeletal muscles that result in energy expenditure ” includes a rationale why energy expenditure can be used to assess physical activity ( Caspersen et al., 1985 ). Physical activity determines the most variable portion of the energy expenditure; moreover, the rate of the energy expenditure is directly linked to the intensity of the physical activity ( Strath et al., 2013 ). While oxygen uptake has equivalency with energy expenditure during rest and physical activity ( Weir, 1949 ; Dennis and Noakes, 1998 ), muscle blood flow is closely related to the oxygen demand of the exercising muscles ( Andersen and Saltin, 1984 ). Furthermore, the heart rate reflects the level of the oxygen supply ( Stringer et al., 1997 ) and energy metabolism regardless of the type of dynamic exercise ( Strath et al., 2000 ; Achten and Jeukendrup, 2003 ). Therefore, ambulatory heart rate recording can be used to assess long-term energy expenditure. As compared with other methods to assess physical activity, such as questionnaires or movement sensors, an objective measurement of the energy expenditure during dynamic exercise by ambulatory heart rate recording has value in defining the intensity of physical activity especially at greater intensities ( Warren et al., 2010 ).

The energy cost of a physical activity can be expressed by the METs that reflect the metabolic rate ( Weir, 1949 ; Jetté et al., 1990 ). As a physiological reference for a man who weighs 70 kg, one MET has been defined as 40 kcal/square meter of the body surface area * h . However, its derivative 1 kcal/kg * h , which refers an oxygen uptake of 250 ml/min, corresponds to the standard MET expression that is scaled by body weight i.e., 3.5 ml/min/kg ( Jetté et al., 1990 ; Byrne et al., 2005 ). This value reflects the resting metabolism during quiet sitting. During physical activity, multiples of the resting metabolisms refer the metabolic rate and aims to standardize the energy cost greater than resting metabolism ( Warren et al., 2010 ). Therefore, standard MET vs. time integral at daily level should reflect the amount and intensity of physical activity.

Conversely, the standard expression of one MET has been criticized because resting metabolism varies according to the physiological state ( Saris et al., 2003 ; Byrne et al., 2005 ; Harrell et al., 2005 ; Kozey et al., 2010 ; Wilms et al., 2014 ). The size of the individual is a major determinant of the basal metabolism ( Kleiber, 1947 ; Ravussin et al., 1986 ) as well as maximal workload and oxygen uptake ( Jensen et al., 2001 ; Wasserman et al., 2005 ). Therefore, to enable comparison between individuals the absolute values should be scaled by body size ( Jensen et al., 2001 ; Wasserman et al., 2005 ; Strath et al., 2013 ). Already in 1883, metabolism was discovered to be proportional to the surface of the body ( Rubner, 1883 ), and in 1949, while determining the ratio between oxygen uptake and energy expenditure, Weir proposed that the metabolic rate should be expressed by the body's surface area instead of body weight ( Weir, 1949 ). On other hand, exercise testing has originally been used among endurance athletes who are lean subjects, and in competitive sports power produced per body weight as an indicator of functional capacity matters more than measurement of the cardiorespiratory capacity in the physiological context ( Lee et al., 2002 ). However, scaling by body weight is widely used at the population level in health sciences and epidemiology.

Rationale of the Body Size Differences Related Scaling

Simple measures, such as height or weight, can easily be used to assess body size, but the problem is that these measures may not be able to distinguish the relevant physiological differences between subjects. The metabolic size matters more as compared to dimensional differences. While the body weight includes fat mass, energy metabolism is related to lean mass. Therefore, scaling by body weight can cause a statistical problem due to “mathematical coupling” with adiposity ( Firebaugh and Gibbs, 1985 ). Body surface area ( DuBois and DuBois, 1916 ), as a fractal and indirect indicator ( Heaf, 2007 ) of body size, cannot identify metabolically relevant lean mass content in the way modern body composition measurements do ( Fosbøl and Zerahn, 2015 ). Physiological ( Goran et al., 2000 ; Tompuri et al., 2014a ), mathematical ( Firebaugh and Gibbs, 1985 ), or physical rationales for scaling oxygen uptake or energy expenditure by body weight cannot be found.

Basal metabolism is strongly related to lean tissue ( Ravussin et al., 1986 ), and physical activity-related energy expenditure is produced by skeletal muscles ( Caspersen et al., 1985 ; Hoppeler and Weibel, 2000 ). The skeletal muscle mass per se is a major determinant for increased metabolism during exercise ( Cooper et al., 1984 ; Tipton and Franklin, 2006 ) and the absolute maximal oxygen uptake ( Turley and Wilmore, 1997 ; LeMura et al., 2001 ). Therefore, scaling by lean mass is a physiologically rational method to perform body size related normalization ( Osman et al., 2000 ; American Thoracic Society, American College of Chest Physicians, 2003 ; Krachler et al., 2014 ). Correspondingly, while fat tissue is energy metabolically inactive during exercise ( Andersen and Saltin, 1984 ; Goran et al., 2000 ), fat mass represents most inter-individually variable compartment of the body ( Fomon et al., 1982 ; Bazzocchi et al., 2013 ). Therefore, scaling by body weight has been criticized ( Tanner, 1949 ; Lorenzo and Babb, 2012 ). Hence, while energy expenditure has equivalency with oxygen uptake, body size-related normalization of the energy expenditure should be done by lean mass as in case of the oxygen uptake.

Physically fat mass represents a load that must be carried during physical activity. According to Newton's second law of motion, mass, such as extra mass by adiposity, increases the force. Therefore, adiposity increases total work of exercise during locomotion-related physical activity ( Cureton and Sparling, 1980 ). Respectively, body fat excess increases basal metabolism because adiposity also increases the amount of lean mass ( Wasserman et al., 2005 ; Heymsfield et al., 2014 ). However, body fat per se does not affect the slope of the increase in oxygen uptake with an increase in the external workload or the maximal aerobic capacity ( Goran et al., 2000 ; Wasserman et al., 2005 ) (Figure 1 ). Interestingly, the observation that energy expenditure during physical activity increases as body weight increases has been assumed to justify the use of body weight in the scaling of energy expenditure ( Strath et al., 2013 ). However, this assumption lacks physiological and physical rationale. While exercise, such as movement from one place to another, is one task causing energy cost, the subject must simultaneously carry his excess fat mass, which is an additional task. Both tasks are physical activities, which are produced by skeletal muscles. Therefore, scaling to normalize body size differences should be performed by metabolic size, i.e., by lean mass.

Figure 1. Additional work by extra load (marked by red color and dashed line) due to adiposity or due to backpack limits maximal performance and increases sub-maximal performance related energy cost during locomotion related exercise . Maximal oxygen uptake referring maximal energy expenditure is similar between subjects as well as lean mass. Increase of oxygen uptake by exercise does not differ between lean (solid line), adipose and lean with backpack subjects ( Wasserman et al., 2005 ). Dimensions of the curves are normative and directional.

Examples of Methodological Hazards Encountered when Using Standard MET

Unfortunately the current review is not only theoretical speculation, but has practical relevance. Standard METs have been used in many original methodological publications of physical activity ( Strath et al., 2001 ; Brage et al., 2004 , 2005 ; Corder et al., 2005 ; Crouter et al., 2008 ), and these publications have been cited in reviews ( Warren et al., 2010 ; Strath et al., 2013 ). Furthermore, body weight scaled METs have been used in recommendations for physical activity measurement in North America ( US Department, 2008 ) and in global recommendations on physical activity by the World Health Organization ( World Health Organization, 2010 ).

While measuring physical activity, the potential problems due to body weight scaled METs may be clarified through simple examples using imaginary subjects: (a) a lean subject, (b) an obese subject, and (c) a lean subject with a backpack representing an extra load to be carried (Figure 1 ). To make this example easier to understand, these subjects have identical lean mass, absolute maximal oxygen uptake and maximal heart rate. These examples demonstrate how extra weight carried during similar sub-maximal or maximal performance effects on alternative indicators of physical activity, such as (1) absolute energy expenditure, (2) lean mass proportional energy expenditure, (3) relative intensity of physical activity and (4) body weight-scaled energy expenditure, i.e., standard METs.

Absolute Energy Expenditure

A more obese subject has a greater absolute energy cost as compared with a leaner subject when they are moving at a similar sub-maximal speed because excess body fat is an extra load that must be carried (Figure 1 ). This agrees with Newton's second law of motion. While the ratio of work and oxygen cost due to locomotion in adipose is similar to leaner subject, the additional energy cost is caused by the greater work because of extra load due to the additional fat mass. The comparison is similar between lean subjects with and without the extra load.

In maximal exercise, the subjects would have similar maximal absolute energy expenditure because they have similar maximal absolute oxygen uptake [ml/min]. However, a lean subject would be able to achieve a greater level of external performance than the obese subject because the lean subject uses less of the physiological reserve to carry his body weight, as observed by Cureton and Sparling (1980) (Figure 1 ).

Lean Mass Proportional Energy Expenditure

When using scaling energy expenditure by lean mass, the same exercise would introduce a similar lean mass proportional energy cost due to exercise performance ( Wasserman et al., 2005 ) (Figure 1 ). However, additional work due to the extra load increases the lean mass proportional oxygen uptake and energy expenditure as compared to the situation without extra load.

Maximal levels of the lean mass proportional energy expenditures would be similar, but subjects with a backpack or greater fat mass would be unable to perform similar locomotion related maximal performance as the lean subject without an extra load ( Cureton and Sparling, 1980 ).

Relative Intensity of Physical Activity

The relative intensity of physical activity refers to the percentage of the maximal oxygen uptake or percentage of maximal heart rate. As compared to a lean subject with a similar absolute maximal oxygen consumption, an obese subject would experience a greater relative intensity of exercise at any given sub-maximal performance level because excess fat mass must be carried during locomotion ( Cureton and Sparling, 1980 ). Correspondingly, if the lean subject carries a backpack, he would achieve a greater intensity at any sub-maximal performance level as compared to the situation without any extra load.

Maximal energy expenditure, maximal oxygen uptake or maximal heart rate can be achieved with or without extra load. However, the extra load diminishes the maximal external performance, such as running speed ( Cureton and Sparling, 1980 ) (Figure 1 ), which also agrees with Newton's second law of motion.

Energy Expenditure by Body Weight Scaled METs

If the energy expenditure were expressed proportionally to body weight, i.e., using standard METs, the obese subject would have a lower level of physical activity than a lean subject with the same absolute energy expenditure. If the subjects walk at the same pace, the absolute or lean mass proportional energy cost as well as intensity relative to maximal oxygen intake or heart rate of the adipose subject would be greater. This false observation results because body weight includes confounding by adiposity when comparing subjects. The extra load caused by the fat mass does not in and of itself impair aerobic capacity or maximal energy expenditure ( Goran et al., 2000 ; Tompuri et al., 2014a ), even though the extra load impairs physiological performance ( Cureton and Sparling, 1980 ) and increases energy cost as compared to lean subjects without the extra load.

Summary of Practical and Analytical Relevance

It is paradoxical that a subject who has a greater absolute energy expenditure and a greater relative intensity during similar performance may be classified as less physically active if using standard METs. Moreover, it is interesting that the original definition of the MET ( Weir, 1949 ) was aware of the potential problems caused by body size normalization by body weight. Therefore, the use of by body weight-scaled standard METs should be avoided when assessing physical activity-related energy expenditure.

When measuring physical activity, it is important to measure essential dimensions instead of irrelevant confounders ( Warren et al., 2010 ). Similarly, it is important to recognize the metabolically relevant size when performing body size-related normalization ( Goran et al., 2000 ; Hoppeler and Weibel, 2000 ; Krachler et al., 2014 ). However, although scaling by body weight represents physiologically historic burden, reporting by METs represents a kind of state of practice in objective measurement of physical activity, maybe because the MET is a simple measure ( Jetté et al., 1990 ; Hills et al., 2014 ). For example as a well-known method METs and scaling by body weight are often used in studies considering survival rates ( Holick et al., 2008 ) or applied physiology ( Turzyniecka et al., 2010 ). However, because of scaling confounded by adiposity there may be undetected interactions between relevant measures. The body lean mass vs. fat mass ratio declines with aging ( Bazzocchi et al., 2013 ) and would introduce inevitably decline in physical activity if using standard METs. Similarly, cardiovascular decline ( Santulli et al., 2013 ) and impairment in insulin sensitivity ( Santulli et al., 2012 ) are prominent features with aging and are also affected by adiposity, which should be taken into account analytically, especially if analyzing with METs.

Adiposity is a major public health risk ( Lim et al., 2012 ), and confounding by adiposity refers to a situation whereby body weight-scaled measures reflects adiposity instead of cardiorespiratory fitness and physical activity per se . In general, as the prevalence of adiposity has increased at the population level ( Vuorela et al., 2011 ), potential confounding by adiposity has become even more important. This causes an increased risk of false conclusions, if relevant residual confounding has not been detected when assessing interactions e.g., between physical activity, adiposity and cardiorespiratory fitness ( Wong et al., 1999 ; Tompuri et al., 2014a , b ; Wilms et al., 2014 ). For example, when using standard METs, confounding by adiposity may result in the biased conclusion that greater intensity and a greater amount of physical activity would be healthier ad infinitum , because more fit and leaner, i.e., healthier subjects, would be classified as engaging in increasingly intensive physical activity as compared to adipose and unfit subjects even if physical activity levels based on absolute energy expenditure were similar.

It is important to realize that statistical adjustment for adiposity may not completely eliminate problems related to residual confounding ( Wong et al., 1999 ) introduced by standard METs. It has been observed that the standard MET as compared to individually measured resting metabolism disproportionally impacts subgroups of the population and causes analytical errors when assessing physical activity ( Kozey et al., 2010 ). Theoretically individually measured resting metabolism instead of the standard MET would potentially diminish analytical problems ( Byrne et al., 2005 ; Wilms et al., 2014 ), because similar scaling error would affect both resting metabolism and energy expenditure by exercise. On other hand, resting metabolic rate is a quite artificial measure ( Hoppeler and Weibel, 2005 ), and also measurement errors will be multiplied along energy expenditure multiplies, which may cause analytical problems ( Wong et al., 1999 ).

To understand physical activity as a behavior and cardiorespiratory fitness or adiposity as a state representing major determinants of public health, the specific measurements of these determinants must be understood to enable a truthful evaluation of their interactions and their independent role as a predictor of health outcomes.

A major advantage in the determination of energy expenditure is that different methods of measurement, such as movement counts by accelerometer and recordings of heart rate, can be combined, which may improve the accuracy of physical activity assessment over a broad range of intensity levels ( Warren et al., 2010 ). Energy expenditure can be assessed also without standard METs. Whereas, oxygen uptake is equivalent to energy expenditure ( Weir, 1949 ), interpretation of a maximal exercise performance with body composition measures ( Tompuri et al., 2014a ) depends on whether oxygen consumption is scaled by body weight or lean mass. Thus, we can conclude that scaling energy expenditure by lean mass would allow to avoid confounding by adiposity when comparing energy expenditure between subjects.

Conflict of Interest Statement

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

This work has been financially supported by a grant from the Research Committee of the Kuopio University Hospital Catchment Area for the State Research Funding (project 5031357, Kuopio, Finland). The language of this text was edited by David E. Laaksonen, MD, PhD, MPH.

Achten, J., and Jeukendrup, A. E. (2003). Heart rate monitoring: applications and limitations. Sports Med. 33, 517–538. doi: 10.2165/00007256-200333070-00004

PubMed Abstract | CrossRef Full Text | Google Scholar

American Thoracic Society, American College of Chest Physicians. (2003). ATS/ACCP Statement on cardiopulmonary exercise testing. Am. J. Respir. Crit. Care Med. 167, 211–277. doi: 10.1164/rccm.167.2.211

Andersen, P., and Saltin, B. (1984). Maximal perfusion of skeletal muscle in man. J. Physiol. 366, 233–249. doi: 10.1113/jphysiol.1985.sp015794

Bazzocchi, A., Diano, D., Ponti, F., Andreone, A., Sassi, C., Albisinni, U., et al. (2013). Health and ageing: a cross-sectional study of body composition. Clin. Nutr. 32, 569–578. doi: 10.1016/j.clnu.2012.10.004

Brage, S., Brage, N., Franks, P. W., Ekelund, U., and Wareham, N. J. (2005). Reliability and validity of the combined heart rate and movement sensor Actiheart. Eur. J. Clin. Nutr. 59, 561–570. doi: 10.1038/sj.ejcn.1602118

Brage, S., Brage, N., Franks, P. W., Ekelund, U., Wong, M. Y., Andersen, L. B., et al. (2004). Branched equation modeling of simultaneous accelometry and heart arte monitoring improves estimate of directly measured physical activity energy expenditure. J. Appl. Physiol . 96, 343–351. doi: 10.1152/japplphysiol.00703.2003

Byrne, N. M., Hills, A. P., Hunter, G. R., Weinsier, R. L., and Schutz, Y. (2005). Metabolic equivalent: one size does not fit all. J. Appl. Physiol . 99, 1112–1119. doi: 10.1152/japplphysiol.00023.2004

Caspersen, C. J., Powell, K. E., and Christenson, G. M. (1985). Physical activity, exercise, and physical fitness: definitions and distinctions for health-related research. Public Health Rep . 100, 126–131.

PubMed Abstract | Google Scholar

Cooper, D. M., Weiler–Ravell, D., Whipp, B. J., and Wasserman, K. (1984). Aerobic parameters of exercise as a function of body size during growth in children. J. Appl. Physiol. 56, 628–634.

Corder, K., Brage, S., Wareham, N. J., and Ekelund, U. (2005). Comparison of PAEE from combined and separate heart rate and movement models in children. Med. Sci. Sports Exerc . 37, 1761–1767. doi: 10.1249/01.mss.0000176466.78408.cc

Crouter, S. E., Churilla, J. R., and Bassett, D. R. Jr. (2008). Accuracy of the Actiheart for the assessment of energy expenditure in adults. Eur. J. Clin. Nutr . 62, 704–711. doi: 10.1038/sj.ejcn.1602766

Cureton, K. J., and Sparling, P. B. (1980). Distance running performance and metabolic responses to running in men and women with excess weight experimentally equated. Med. Sci. Sports Exerc. 12, 288–294. doi: 10.1249/00005768-198024000-00011

Dennis, S. C., and Noakes, T. D. (1998). Physiological and metabolic responses to increasing work rates: relevance for exercise prescription. J. Sports Sci . 16, S77–S84. doi: 10.1080/026404198366704

DuBois, D., and DuBois, E. F. (1916). A formula to estimate the approximate surface area if height and weight be known. Arch. Intern. Med . 17, 863–871. doi: 10.1001/archinte.1916.00080130010002

PubMed Abstract | CrossRef Full Text

Firebaugh, G., and Gibbs, J. P. (1985). User's guide to ratio variables. Am. Soc. Rev. 50, 713–722. doi: 10.2307/2095384

CrossRef Full Text | Google Scholar

Fomon, S. J., Haschke, F., Ziegler, E. E., and Nelson, S. E. (1982). Body composition of reference children from birth to age 10 years. Am. J. Clin. Nutr . 35, 1169–1175.

Fosbøl, M. Q., and Zerahn, B. (2015). Contemporary methods of body composition measurement. Clin. Physiol. Funct. Imaging 35, 81–97. doi: 10.1111/cpf.12152

Goran, M., Fields, D. A., Hunter, G. R., Herd, S. L., and Weinsier, R. L. (2000). Total body fat does not influence maximal aerobic capacity. Int. J. Obes. Relat. Metab. Disord. 24, 841–848. doi: 10.1038/sj.ijo.0801241

Harrell, J. S., McMurray, R. G., Baggett, C. D., Pennell, M. L., Pearce, P. F., and Bangdiwala, S. I. (2005). Energy costs of physical activities in children and adolescents. Med. Sci. Sports Exerc. 37, 329–336. doi: 10.1249/01.MSS.0000153115.33762.3F

Heaf, J. G. (2007). Review article - The origin of the 1.73m 2 body surface area normalization: problems and implications. Clin. Physiol. Funct. Imaging 27, 135–137. doi: 10.1111/j.1475-097X.2006.00718.x

Heymsfield, S. B., Gonzalez, M. C. C., Shen, W., Redman, L., and Thomas, D. (2014). Weight loss composition is one-fourth fat-free mass: a critical review and critique of this widely cited rule. Obes. Rev . 15, 310–321. doi: 10.1111/obr.12143

Hills, A. P., Mokhtar, N., and Byrne, N. M. (2014). Assessment of physical activity and energy expenditure: an overview of objective measures. Front. Nutr . 1:5. doi: 10.3389/fnut.2014.00005

Holick, C. N., Newcomb, P. A., Trentham-Dietz, A., Titus-Ernstoff, L., Bersch, A. J., Stampfer, M. J., et al. (2008). Physical activity and survival after diagnosis of invasive breast cancer. Cancer Epidemiol. Biomarkers Prev . 17, 379–386. doi: 10.1158/1055-9965.EPI-07-0771

Hoppeler, H., and Weibel, E. R. (2000). Structural and functional limits for oxygen supply to muscle. Acta Physiol. Scand . 168, 445–456. doi: 10.1046/j.1365-201x.2000.00696.x

Hoppeler, H., and Weibel, E. R. (2005). Scaling functions to body size: theories and facts. J. Exp. Biol. 2008, 1573–1574. doi: 10.1242/jeb.01630

Jensen, K., Johansen, L., and Secher, N. H. (2001). Influence of body mass on maximal oxygen uptake: effect of sample size. Eur. J. Appl. Physiol . 84, 201–205. doi: 10.1007/s004210170005

Jetté, M., Sidney, K., and Blümchen, G. (1990). Metabolic equivalents (METS) in exercice testing, exercise prescription, and evaluation of functional capacity. Clin. Cardiol. 13, 555–565. doi: 10.1002/clc.4960130809

Kleiber, M. (1947). Body size and metabolic rate. Physiol. Rev. 27, 511–541.

Kozey, S., Lyden, K., Staudenmayer, J., and Freedson, P. (2010). Errors in MET estimates of physical activities using 3.5 ml·kg −1 ·min −1 as the baseline oxygen consumption. J. Phys. Act. Health 7, 508–516. Available online at: http://www.umass.edu/physicalactivity/newsite/publications/Sarah%20Keadle/papers/1.pdf

Krachler, B., Savonen, K., Komulainen, P., Hassinen, M., Lakka, T. A., and Rauramaa, R. (2014). Cardiopulmonary fitness is a function of lean mass, not total body weight: the DR's EXTRA study. Eur. J. Prev. Cardiol . doi: 10.1177/2047487314557962. [Epub ahead of print].

Lee, H., Martin, D. T., Anson, J. M., Grundy, D., and Hahn, A. G. (2002). Physiological characteristics of successful mountain bikers and professional road cyclists. J. Sports Sci . 20, 1001–1008. doi: 10.1080/026404102321011760

LeMura, L. M., von Duvillard, S. P., Cohen, S. L., Root, C. J., Chelland, S. A., Andreacci, J., et al. (2001). Treadmill and cycle ergometry testing in 5– to 6–year–old children. Eur. J. Appl. Physiol. 85, 472–478. doi: 10.1007/s004210100461

Lim, S. S., Vos, T., Flaxman, A. D., Danaei, G., Shibuya, K., Adair-Rohani, H., et al. (2012). A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380, 2224–2260. doi: 10.1016/S0140-6736(12)61766-8

Lorenzo, S., and Babb, T. G. (2012). Quantification of cardiorespiratory fitness in healthy nonobese and obese men and women. Chest 141, 1031–1039. doi: 10.1378/chest.11-1147

Osman, A. F., Mehra, M. R., Lavie, C. J., Nunez, E., and Milani, R. V. (2000). The incremental prognostic importance of body fat adjusted peak oxygen consumption in chronic heart failure. J. Am. Coll. Cardiol. 36, 2126–2131. doi: 10.1016/S0735-1097(00)00985-2

Ravussin, E., Lillioja, S., Anderson, T. E., Christin, L., and Bogardus, C. (1986). Determinants of 24-hour energy expenditure in man. Methods and results using a respiratory chamber. J. Clin. Invest . 78, 1568–1578. doi: 10.1172/JCI112749

Rubner, M. (1883). Ueber den einfluss der körpergrösse auf stoff-und kraftwechsel. Z. Biol . 19, 535–562.

Google Scholar

Santulli, G., Ciccarelli, M., Trimarco, B., and Iaccarino, G. (2013). Physical activity ameliorates cardiovascular health in elderly subjects: the functional role of the B adenergic system. Front. Physiol. 4:209. doi: 10.3389/fphys.2013.00209

Santulli, G., Lombardi, A., Sorriento, D., Anastasio, A., Del Giudice, C., Formisano, P., et al. (2012). Age-related impairment in insulin release: the essential role of beta(2)-adenergic receptor. Diabetes 61, 692–701. doi: 10.2337/db11-1027

Saris, W. H., Blair, S. N., van Baak, M. A., Eaton, S. B., Davies, P. S., Di Pietro, L., et al. (2003). How much physical activity is enough to prevent unhealthy weight gain? Outcome of the IASO 1st Stock Conference and consensus statement. Obes. Rev. 4, 101–114. doi: 10.1046/j.1467-789X.2003.00101.x

Strath, S. J., Bassett, D. R. Jr., Swartz, A. M., and Thompson, D. L. (2001). Simultaneous heart rate-motion sensor technique to estimate energy expenditure. Med. Sci. Sports Exerc . 33, 2118–2123. doi: 10.1097/00005768-200112000-00022

Strath, S. J., Kaminsky, L. A., Ainsworth, B. E., Ekelund, U., Freedson, P. S., Gary, R. A., et al. (2013). Guide to the assessment of physical activity: clinical and research applications: a scientific statement from the american heart association. Circulation 128, 2259–2279. doi: 10.1161/01.cir.0000435708.67487.da

Strath, S. J., Swartz, A. M., Bassett, D. R. Jr., O'Brien, W. L., King, G. A., and Ainsworth, B. E. (2000). Evaluation of heart rate as a method for assessing moderate intensity physical activity. Med. Sci. Sports Exerc . 32, S465–S470.

Stringer, W. W., Hansen, J. E., and Wasserman, K. (1997). Cardiac output estimated noninvasely from oxygen uptake during exercise. J. Appl. Physiol . 82, 908–912.

Tanner, J. M. (1949). Fallacy of per–weight and per–surface area standards, and their relation to spurious correlation. J. Appl. Physiol . 2, 1–15.

Tipton, C. M., and Franklin, B. A. (2006). “The language of exercise,” in ACSM's Advanced Exercise Physiology , eds C. M. Tipton, M. N. Sawka, C. A. Tate, and R. L. Terjung (Philadelphia, PA: Lippincott Williams & Wilkins), 3–10.

Tompuri, T., Lintu, N., Savonen, K., Laitinen, T., Laaksonen, D., Jääskeläinen, J., et al. (2014a). Measures of cardiorespiratory fitness in relation to measures of body size and composition among children. Clin. Physiol. Funct. Imaging . doi: 10.1111/cpf.12185. [Epub ahead of print].

Tompuri, T., Vaisto, J., Lintu, N., Haapala, E. A., Lindi, V., and Lakka, T. A. (2014b). Physical activity and watching TV, but not cardiorespiratory fitness, are related to metabolic risk among children: the Physical Activity and Nutrition in Children (PANIC) study. JPAH 11, S191. Abstract retrieved from Abstract in 2014 Global Summit on the Physical Activity of Children database.

Turley, K. R., and Wilmore, J. H. (1997). Cardiovascular responses to treadmill and cycle ergometer exercise in children and adults. J. Appl. Physiol . 83:948–957.

Turzyniecka, M., Wild, S. H., Krentz, A. J., Chipperfield, A. J., Clough, G. F., and Byrne, C. D. (2010). Diastolic function is strongly and independently associated with cardiorespiratory fitness in central obesity. J. Appl. Physiol . 108, 1568–1574. doi: 10.1152/japplphysiol.00023.2010

US Department. (2008). 2008 Physical Activity Guidelines for Americans . Available online at: http://www.health.gov/paguidelines/guidelines/ (Accessed March 10, 2015).

Vuorela, N., Saha, M. T., and Salo, M. K. (2011). Change in prevalence of overweight and obesity in Finnish children - comparison between 1974 and 2001. Acta Paediatr . 100, 109–115. doi: 10.1111/j.1651-2227.2010.01980.x

Warren, J. M., Ekelund, U., Besson, H., Mezzani, A., Geladas, N., Vanhees, L., et al. (2010). Assessment of physical activity - a review of methodologies with reference to epidemiological research: a report of the exercise physiology section of the European Association of Cardiovascular Prevention and Rehabilitation. Eur. J. Cardiovasc. Prev. Rehabil. 17, 127–139. doi: 10.1097/HJR.0b013e32832ed875

Wasserman, K., Hansen, J. E., Sue, D. Y., Stringer, W. W., and Whipp, B. J. (2005). Principles of Exercise Testing and Interpretation Including Pathophysiology and Clinical Applications, 4th Edn . Philadelphia, PA: Lippincott Williams & Wilkins.

Weir, J. B. (1949). New methods for calculating metabolic rate with special reference to protein metabolism. J. Physiol . 109, 1–9. doi: 10.1113/jphysiol.1949.sp004363

Wilms, B., Ernst, B., Thurnheel, M., Weisser, B., and Schultes, B. (2014). Correction factors for the calculation of metabolic equivalents (MET) in overweight to extremely obese subjects. Int. J. Obes. (Lond). 38, 1383–1387. doi: 10.1038/ijo.2014.22

Wong, M. Y., Day, N. E., and Wareham, N. J. (1999). Measurement error in epidemiology: the design of validation studies II: bivariate situation. Stat. Med. 18, 2831–2845.

World Health Organization. (2010). Global Recommendations on Physical Activity for Health . Geneva: WHO Press.

Keywords: adiposity, energy expenditure, intensity, lean mass, metabolic equivalent, MET, PAEE, physical activity

Citation: Tompuri TT (2015) Metabolic equivalents of task are confounded by adiposity, which disturbs objective measurement of physical activity. Front. Physiol . 6:226. doi: 10.3389/fphys.2015.00226

Received: 13 March 2015; Accepted: 27 July 2015; Published: 11 August 2015.

Reviewed by:

Copyright © 2015 Tompuri. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Tuomo T. Tompuri, Department of Clinical Physiology and Nuclear Medicine, Kuopio University Hospital, PO Box 100, FIN-70211 Kuopio, Finland, [email protected]; [email protected]

Metabolic Equivalent of Tasks

Metabolic equivalent of tasks (METS), which is sometimes shortened to metabolic equivalents, is a standard measure in physiology used to denote the physical intensity involved while exercising at different intensities for various physical activities. METS are calculated as the ratio of the work metabolic rate (rate of energy consumption during a specific physical activity) to the resting metabolic rate (rate of energy consumption while sleeping). This plural form is sometimes stated in its singular form: metabolic equivalent of task (MET).

METS are measures of the amount of oxygen used by the body during various types of activities or exercises. The term MET is used to express an individual's metabolic rate while performing some type of task based on that person's resting metabolic rate. The resting MET is measured as 1. That is, 1 MET is the amount of energy, as determined from the amount of oxygen inhaled, used by the body at rest. Activities considered to be resting include sitting quietly on a chair or reading a book while lying on a bed.

When more strenuous (nonresting) activities are pursued, such as walking at about 3 mi. (4.8 km) per hour, bicycling on flat terrain, and mowing the grass with a push mower, the body expends more energy and, thus, the body inhales more oxygen. Then, when such activities occur, MET values are higher. The METS for various nonresting tasks range from 1.5 to 18.0.

Demographics

Anyone can use METS to determine the intensity of a workout or for everyday activities around the house. When METS are used, calories per hour can be determined for a weight-loss program or to gauge intensity levels for a particular fitness level desired.

Description

The various values for METS were first standardized within the Survey of Activity, Fitness, and Exercise (SAFE Study—1987 to 1989 ). Called the Compendium of Physical Activities, they were developed by Dr. William Haskell, from Stanford University (California). In 2017, Dr. Haskell was a professor (research) of medicine, emeritus, at the Stanford Prevention Research Center.

Originally, Haskell created a five-digit code that identified the category (heading) of physical activity (the first two digits) and the type of activity (the last three digits). For example, “01010” referred to “01” for “bicycling” and “010” stood for “bicycling at less than 10 miles per hour, for leisure, to work, or for pleasure.” The first version of the Compendium was published in 1993 and the last update to it was published in 2011 (as of December 2016).

The MET is a unit of standard metabolic equivalent used to estimate an individual's metabolic rate while performing some task based on that person's resting metabolic rate (RMR). Two well known measures of RMR, which is also denoted as 1 MET, are the consumption of (1) 1 kilocalorie (kcal) per kilogram (kg) of body weight per hour (h) and (2) 3.5 milliliters (ml) of oxygen (O2) per kilogram (kg) of body weight per minute (min). MET values range from 0.9 MET when sleeping to 18.0 MET when running at 10.9 mi. (17.5 km) per hour.

One MET is also defined as 58.2 watts per meter squared (W/m 2 ) or equivalently as 18.4 British thermal units per hour times feet squared (BTU/h × ft 2 ). These terms equal the rate of energy produced per unit surface area (either in square meters or square feet) of an average adult at rest in the sitting position.

The value of 1 MET is the base metabolic equivalent of task, or the metabolic rate of a person who is at rest. The amount of energy expended for any activity is measured in METS, which are multiples of a person's base metabolic rate (1 MET).

A value of 3 MET to 6 MET is considered to be moderate physical activities. Moderate (low-intensity) physical activity is generally determined to be any activity that causes the heart rate or breathing to increase. Such activities usually burn 3.5 to 7.0 calories (cal) per minute (kcal/min).

Any MET values from 7 to 18 are considered vigorous physical activities. Examples of vigorous (high-intensity) activities are playing a strenuous game of tennis, doing demanding calisthenics, snow skiing down a hill, water skiing on a lake, and swimming competitive laps in a pool. When such levels are attained, breathing and heart rates are rapid, and the body is forced to expend quite a bit of energy to perform such activities. Vigorous physical activities expend 8.0 or more calories per minute; this value may vary, however, depending on fitness level, weight, age, and other such factors.

For instance, a man who is horseback riding has a MET of about 3.5, meaning that his metabolic rate is 3.5 times than his resting metabolic rate. A woman who is backpacking has a 7 MET, or a metabolic rate that is seven times higher than her resting rate. METS for other activities include:

• active skindiving, 16.0 MET
• rigorous pedaling a stationary bicycle, 12.5 MET
• running at a pace of 8 mi. (12.9 km) per mile, 12.5 MET
• running at 12 mi. (19.4 km) per hour, 8.5 MET
• playing racquetball, 8.0 MET
• jogging at 6 mi. (9.7 km) per hour, 8.0 MET
• playing a game of basketball, 7.0 MET
• aerobic dancing, 6.0 MET
• performing gymnastics, 5.5 MET
• walking at 4 mi. (6.5 km) per hour on level, firm surface, 5.0 MET
• bicycling at 24 mi. (38.7 km) per hour, 5.0 MET
• doing calisthenics, 4.5 MET
• playing volleyball, 4.0 MET
• walking at 2 mi. (3.2 km) per hour on level, firm surface, 2.0 MET
• resting, 1.0 MET
• sleeping, 0.9 MET

According to the Physical Activity Guidelines for Americans (2008, the latest edition as of 2017), which is produced by the U.S. Office of Disease Prevention and Health Promotion, adults should engage in from 500 to 1,000 MET minutes per week.

Preparation

An Exercise Metabolic Test is one way to determine factors relating to MET. During the test, a person will exercise on a treadmill or a bicycle that is attached to an ergometer (a device for measuring work performed). The participant is hooked to a breathing device and gas analyzer. The operator of the equipment has the participant increase their level of exercise intensity, which causes the consumption of oxygen to increase. When the oxygen level does not rise further, but the exercise intensity level continues to increase, then the gas analyzer will determine the VO 2 max, or the maximum (max) volume (V) of oxygen (O 2 ).

The test shows how an individual responds to physical activity at different levels of intensity. The equipment also produces some of the following statistics: calories burned per minute while exercising; percent of calories from fat, carbohydrates, and protein; volume of carbon dioxide (CO 2 ) exhaled (VCO 2 ); and heart rate. The test is used for people who have breathing concerns such as shortness of breath, desire to lose (or gain) weight, or want to know their peak performance level.

There are no known medical risks for using metabolic equivalent of tasks. However, MET values are based on a large sample of people representing an even much larger population. Therefore, these values are average values and may vary depending on differences among individuals, such as fitness level, age, speed, intensity, conditions while exercising, and other such factors.

By knowing one's MET value for a particular exercise, it is possible to estimate the number of calories that will be burned while participating in physical activities. Many websites provide tools to calculate the number of calories burned while exercising. One of them, which provides a step–by–step procedure for this measurement, is “How to Calculate Calories Burned” from Ethika Clinic in Kolkata, India. It is found at: https://ethikaclinic.wordpress.com/a-healthy-life/calculatecalories-burnt/ .

• Where can I learn more about METS?
• How will METS help me become more healthy and fit?
• Why should I be concerned about METS while I am exercising?

Graham, Dan J., et al., eds. Emerging Technologies to Promote and Evaluate Physical Activity . Lausanne: Frontiers, 2014.

Heyward, Vivian, H., and Ann L. Gibson, eds. Advanced Fitness Assessment and Exercise Prescription . Champaign: Human Kinetics, 2014.

Malâ, Lucia, Tomâš, Malÿ, and František, Zahálka. Fitness Assessment . Prague: Charles University of Prague: Karolinum Press, 2014.

McArdle, William D., Frank I. Katch, and Victor L. Katch. Essentials of Exercise Physiology . Philadelphia: Wolters Kluwer, 2016.

Pescatello, Linda S., et al., eds. ACSM's Guidelines for Exercise Testing and Prescription . Baltimore: Wolters Kluwer Health/Lippincott Williams & Wilkins, 2014.

Swain, David P., and Clinton A. Brawner, eds. ACSM's Resource Manual for Guidelines for Exercise Testing and Prescription . Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins, 2014.

The Cooper Institute. MET Minutes: A Simple Common Value to Track Exercise Progress . CooperInstitute.org . (2014). http://www.cooperinstitute.org/2012/04/met-minutes-a-simple-common-value-to-trackexercise-progress/ (accessed December 28, 2016).

Maximal Oxygen Consumption Test (VO 2 max) . TopEnd Sports.com . http://www.topendsports.com/testing/tests/VO2max.htm (accessed December 22, 2016).

National Center for Biotechnology Information, US National Library of Medicine. Compendium of Physical Activities: Classification of Energy Costs of Human Physical Activities . NCBI.nlm.nih.gov . https://www.ncbi.nlm.nih.gov/pubmed/8292105 (accessed December 28, 2016).

Office of Disease Prevention and Health Promotion. Physical Activity Guidelines . Health.gov . (December 28, 2016). https://health.gov/paguidelines/ (accessed December 28, 2016).

Prevention Research Center, Arnold School of Public Health. The Compendium of Physical Activities . University of South Carolina. http://prevention.sph.sc.edu/tools/compendium.htm (accessed December 22, 2016).

Stanford Medicine. William Haskell . med.stanford.edu . https://med.stanford.edu/profiles/william-haskell (accessed December 28, 2016).

American College of Sports Medicine, 401 W. Michigan St., Indianapolis, IN, 46202-3233, (317) 634-9200, Fax: (317) 634-7817, http://www.acsm.org .

American Council on Exercise, 4851 Paramount Dr., San Diego, CA, 92123, (888) 825-3636, http://www.fitness.gov .

American Heart Association, 7272 Greenville Ave., Dallas, TX, 75231, 90245, (800) 242-8721, http://www.americanheart.org .

The Cooper Institute, 12330 Preston Rd., Dallas, TX, 75230, (972) 341-3200, Fax: (972) 341-3227, (800) 635-7050, http://www.cooperinstitute.org .

President's Council on Fitness, Sports & Nutrition, 1101 Wootton Pkwy., Ste. 560, Rockville, MD, 20852, (240) 276-9567, Fax: (240) 276-9860, [email protected], http://www.fitness.gov .

Shape America (Society of Health and Physical Educators), 1900 Association Dr., Reston, VA, 20191-1598, (703) 476-9527, (800) 213-7193, http://www.shapeamerica.org/ .

William A. Atkins, BB, BS, MBA

• Open access
• Published: 24 November 2023

Accelerometer-measured absolute versus relative physical activity intensity: cross-sectional associations with cardiometabolic health in midlife

• Jonatan Fridolfsson 1 ,
• Daniel Arvidsson 1 ,
• Elin Ekblom-Bak 2 ,
• Örjan Ekblom 2 ,
• Göran Bergström 3 , 4 &
• Mats Börjesson 5 , 6  

BMC Public Health volume  23 , Article number:  2322 ( 2023 ) Cite this article

287 Accesses

8 Altmetric

Metrics details

Observational studies investigating the association between accelerometer-measured physical activity and health all use absolute measures of physical activity intensity. However, intervention studies suggest that the physical activity intensity required to improve health is relative to individual fitness. The aim of this study was to investigate the associations between accelerometer-measured absolute and relative physical activity intensity and cardiometabolic health, and what implications these associations may have on the interpretation of health-associated physical activity.

A sample of the cross-sectional Swedish CArdioPulmonary bioImage Study (SCAPIS) consisting of 4,234 men and women aged 55–64 years was studied. Physical activity intensity was measured by accelerometry and expressed as absolute (e.g., metabolic equivalents of task) or relative (percentage of maximal oxygen consumption). Fitness was estimated by the submaximal Ekblom-Bak test. A composite (‘metabolic syndrome’) score combined measures of waist circumference, systolic blood pressure, high-density lipoprotein, triglycerides, and glycated hemoglobin. Associations of absolute and relative physical activity intensity with the health indicators (i.e., fitness and metabolic syndrome score) were studied by partial least squares regression. Analyses were stratified by fitness level.

Both absolute and relative physical activity intensity associated with the health indicators. However, the strongest associations for absolute intensity varied depending on fitness levels, whereas the associations for relative intensity were more synchronized across fitness groups. The dose–response relationship between moderate-to-vigorous intensity and the health indicators was stronger for relative than for absolute intensity. The absolute and relative moderate-to-vigorous intensity cut-offs intersected at the 5th fitness percentile, indicating that the absolute intensity cut-off is too low for 95% of individuals in this sample. While 99% of individuals fulfilled the general physical activity recommendations based on absolute intensity measures, only 21% fulfilled the recommendations based on relative intensity measures. In relation to a “sufficient” fitness level, 9% fulfilled the recommendations.

Conclusions

Accelerometer-measured relative physical activity intensity represents the intensity related to health benefits regardless of fitness level. Traditional absolute moderate intensity accelerometer cut-offs are too low for most individuals and should be adapted to the fitness level in the sample studied. Absolute and relative physical activity intensity cannot be used interchangeably.

Peer Review reports

The health benefits of physical activity (PA) are related to the intensity and volume of activity [ 1 ]. PA intensity can be expressed in absolute or relative terms [ 2 ]. Absolute PA intensity measures include energy expenditure, locomotion speed and mechanic work, all of which represent the same absolute intensity regardless of who is performing the activity. Relative PA intensity refers to absolute PA intensity in relation to individual maximal PA capacity and differs depending on individual cardiorespiratory fitness. Cardiorespiratory fitness refers to the maximal oxygen consumption of an individual. Examples of relative intensity measures are proportion of maximal oxygen consumption, proportion of maximal heart rate, and self-perceived exertion.

PA intensity is today typically measured by accelerometers. The acceleration measured is closely related to mechanical workload and is an absolute measure of PA intensity [ 3 ]. The absolute acceleration output is often translated to absolute energy expenditure, based on calibration studies using indirect calorimetry as reference [ 4 ]. This continuous estimate of energy expenditure is then used to determine the time spent at different absolute intensity levels. Although PA intensity is measured in absolute terms, it is often expressed in relative terms (as light, moderate, and vigorous). In absolute terms, sedentary is defined as time spent at an absolute energy expenditure below 1.5 metabolic equivalents of task (METs), where 1 MET represents an oxygen consumption of 3.5 mL/min/kg. Absolute light, moderate, vigorous and very-vigorous intensity is defined as time spent above 1.5, 3, 6 and 9 METs respectively [ 2 ]. The resulting time spent at these intensity levels is used in observational studies to investigate associations with health and fulfilment of recommendations on PA [ 1 ].

Relative PA intensity can be determined by the oxygen consumption of an activity in relation to individual maximal oxygen consumption and is defined as moderate intensity above 46%, vigorous above 64% and very-vigorous above 91% [ 5 ]. For the moderate intensity cut-off, the absolute 3 METs is equivalent to the relative 46% only in individuals with a maximal oxygen consumption of 22.8 mL/min/kg [ 5 ]. From a relative perspective, 3 METs would be too low for individuals with higher fitness level than 22.8 mL/min/kg, and too high for individuals with lower fitness. This mixture of absolute and relative PA intensity in the use of accelerometer data to determine PA levels causes confusion and misunderstanding in the interpretation of the results, and in the evaluation of the importance of PA to health.

The current consensus in PA research is that the main health benefits of PA come from time spent at moderate-to-vigorous intensity [ 2 ]. Recommendations state that adults should undertake 150–300 min per week of moderate intensity PA, or 75–150 min per week of vigorous intensity, or an equivalent combination of both [ 1 ]. The PA guidelines suggest that absolute or relative intensity can be used interchangeably for most individuals with the exception of older adults with low fitness level, for whom relative intensity is more appropriate [ 1 , 2 ]. However, fitness levels vary considerably between individuals and also decrease with age [ 6 ]. This suggest absolute and relative intensity should not be used interchangeably. In addition, PA recommendations are mainly based on self-reported PA, which lack detail about intensity [ 4 ].

To understand how the different intensity measures should be used, their association with health outcomes must be studied in more detail by objective measures. The association between PA and cardiometabolic health is of particular interest due to the importance of PA to reduce the risk of cardiovascular disease. [ 7 ] The association between PA and cardiometabolic health is considered to be both moderated and mediated by fitness [ 2 ]. The moderation by fitness refers to the association between PA and cardiometabolic health being different depending on fitness level. The mediation by fitness refers to the association between higher PA and better cardiometabolic health being caused by an increase in fitness level.

Furthermore, there is conflicting evidence from previous studies as to whether absolute light, moderate or vigorous intensity is required for health benefits. In older individuals, associations with cardiometabolic health can be found at absolute light intensity, whereas in younger individuals absolute vigorous intensity may be required for any significant associations with cardiometabolic health [ 8 , 9 , 10 ]. The age differences between these samples could be considered a proxy for differences in fitness level. Relative intensity was not considered in these highly diverse study samples, which might be a contributing factor to the lack of consensus.

Previous studies have demonstrated that accelerometer-measured absolute PA intensity can be translated to relative intensity [ 11 , 12 , 13 , 14 , 15 ]. This is done by applying available calibration equations to estimate oxygen consumption from accelerometry data, and then relating the estimated oxygen consumption to the measured fitness level. However, the health benefits of accelerometer-measured relative PA intensity have not been investigated. The purpose of this study was therefore to investigate the associations of accelerometer-measured absolute and relative PA intensity with cardiometabolic health, and what implications these associations may have on the interpretation of health-associated PA.

Study sample

A sample of the Swedish multicenter observational study SCAPIS (Swedish CArdioPulmonary bioImage Study) [ 16 ] was analyzed. SCAPIS includes 30 154 randomly selected men and women aged 50–64 years, with objectively measured PA and extensive measurements of markers of cardiovascular health [ 16 , 17 ]. Cardiorespiratory fitness was estimated in a subsample from the study center in Gothenburg. All participants with estimated fitness, valid PA measurements, and measurements of cardiometabolic health indicators were included in this study ( n  = 4 176). The data collection was carried out in 2013–2018. SCAPIS has been approved by the ethics committee at Umeå University (no. 2021–228-31 M) and the current study has received specific approval by the Regional ethical board in Gothenburg (no. 638–16). Written, informed consent was retrieved from all participants.

Physical activity and fitness

Participants wore a triaxial accelerometer (ActiGraph model GT3X + , wGT3X + or wGT3X-BT, ActiGraph, Pensacola, Florida, USA) in an elastic belt over their right hip for seven continuous days. They were instructed to take the accelerometer off during sleep and water-based activities. Raw accelerometer data were extracted and processed using the 10 Hz frequency extended method (FEM) [ 18 ]. This method enables more detailed and accurate measurement of PA intensity compared to the most commonly used processing method using ActiGraph counts [ 3 , 18 , 19 ]. Triaxial accelerometer data were combined to a vector magnitude and reduced to 3 s epochs. Non-wear time was defined as 60 min of zero accelerometer output with allowance of up to 2 min of interruptions below the sedentary threshold [ 20 ]. A valid day was defined as at least 10 h of wear time and a valid measurement as at least 4 valid days [ 21 ]. The variables retrieved from the processed accelerometer data represent time spent at different intensity levels.

In this study, absolute PA intensity was expressed as metabolic equivalents of task (METs), oxygen consumption and locomotion speed [ 18 , 22 ]. The definition of one MET is the oxygen consumption during rest, generally considered to be 3.5 mL/min/kg [ 2 ]. On an absolute scale, PA intensity is generally defined as sedentary (< 1.5 METs), light (≥ 1.5- < 3 METs), moderate (≥ 3- < 6 METs), vigorous (≥ 6- < 9 METs) and very-vigorous (≥ 9 METs) [ 2 , 21 ], which correspond to an oxygen consumption of < 5.25, ≥ 5.25- < 10.5, ≥ 10.5- < 21.0, ≥ 21.0- < 31.5 and ≥ 31.5 mL/min/kg respectively. This study expressed relative PA intensity as proportion of estimated maximal oxygen consumption during activity, standardized to bodyweight. On a relative scale, PA intensity is defined as light (< 46%), moderate (≥ 46- < 64%), vigorous (≥ 64- < 91%) and very-vigorous (≥ 91%) of maximal oxygen consumption according to the American College of Sports Medicine [ 5 ].

Maximal oxygen consumption (referred to as fitness) was estimated by the Ekblom-Bak submaximal cycle ergometer test [ 23 ]. The testing procedure include cycling at two submaximal workloads, the first at light intensity and the second at estimated moderate intensity, while the heart rate response is measured. The ratio of the difference in heart rate and the difference in workload is calculated and compared to published reference values. The Ekblom-Bak test has high validity as reference to direct measurement with cross validated R 2 adj = 0.90 and standard error of estimate: 0.30 L/min for all ages, and R 2 adj = 0.84 and standard error of estimate: 0.33 L/min for 50–64 years old [ 23 ]. Exclusion criteria for the fitness test were ongoing infections, known unstable cardiovascular disease, indication of cardiac disease from electrocardiography patterns, medication with beta-blockers, weight above 125 kg or resting heart rate above 100 beats per minute. In addition, participants who refrained from performing the fitness test was not tested.

Processed accelerometer output was used to estimate the time spent at different absolute and relative PA intensity levels [ 3 , 18 ]. A detailed PA intensity spectrum consisting of 22 intensity variables was generated to represent an absolute PA intensity pattern [ 19 , 24 ]. The bin edges dividing the PA intensity spectrum variables were 0, 40, 80, 160, 240 mg, and then increasing in 80 mg intervals to 1600 mg and above. The same intensity spectrum was used to represent relative PA intensity by first translating the spectrum cut-offs from accelerometer output to oxygen consumption, and then dividing by estimated maximal oxygen consumption. Translation was done by a linear regression model based on published reference values [ 18 ]. The regression coefficients were Y = 0.02683X + 5.108, where Y represents oxygen consumption in mL/min/kg and X represents processed accelerometer output in mg. Traditional crude cut-points representing sedentary, light-, moderate-, vigorous- and very-vigorous PA were used for reference [ 5 , 18 ].

Metabolic syndrome score

To provide an overall measure of individual cardiometabolic health, a metabolic syndrome score was used [ 25 , 26 ]. The metabolic syndrome refers to the clustering of several cardiometabolic risk factors, including central obesity, dyslipidemia, hyperglycemia and elevated blood pressure, and can be used to predict the risk of future cardiovascular disease [ 27 ]. Central obesity was represented by waist circumference, dyslipidemia by high-density lipoprotein (HDL) and triglycerides, hyperglycemia by glycated hemoglobin (HbA1c), and hypertension by systolic blood pressure (SBP) [ 27 ]. Waist circumference was measured by measuring tape according to standardized methods [ 28 ]. A venous blood sample was collected after an overnight fast and was used for measuring HDL, triglycerides and HbA1c levels. SBP was measured twice in each arm by an automated device (Omron M10-IT, Omron Health care Co, Kyoto, Japan) and the mean of the measurements was used. Examinations were performed on two or three occasions within two weeks. All variables were measured within two weeks (at the examinations before or after the week of accelerometer use).

All measured risk factors have a positive association with cardiovascular disease (e.g. higher values are associated with higher risk of disease), except HDL which has a negative association with cardiovascular disease. Therefore, HDL was multiplied by -1 to have a positive association with the other risk factors. For dyslipidemia to not be more influential in the composite score due to more variables, the mean of sex standardized Z-scores of HDL and triglycerides was first calculated [ 26 ]. Subsequently, the mean of sex standardized Z-scores of waist circumference, HbA1c, SBP and the combined value of HDL and triglycerides was calculated as the metabolic syndrome composite score.

Statistical analyses

To investigate differences in PA intensity depending on fitness, analyses were stratified by fitness level. Fitness was standardized by sex and used for dividing the sample into tertiles. Differences between fitness groups in risk factors and absolute and relative crude PA levels were investigated by ANOVA with post hoc t-tests. Detailed absolute and relative PA intensity patterns were visualized by standardizing the PA volume in the intensity spectrum variables in the whole study sample to a mean of zero and a standard deviation of one. Standardization was performed by subtracting the mean and dividing by the standard deviation for each intensity variable. Subsequently, the mean of each standardized variable was presented for each fitness group [ 24 ].

The proportion of individuals fulfilling the PA recommendations of 150 min of moderate-to-vigorous PA [ 1 ] was calculated using both absolute and relative PA measures. Previous research has suggested a maximal oxygen consumption capacity of 31.5 and 35 mL/min/kg for women and men, respectively, to be a sufficiently high fitness level to achieve most potential health benefits [ 29 ]. The proportion of individuals fulfilling the PA recommendations with a moderate-to-vigorous intensity cut-off in relation to a sufficiently high fitness level was calculated. The 46% moderate-to-vigorous intensity cut-off for an individual with this fitness level is equivalent to an oxygen consumption of 14.5 and 16.1 mL/min/kg for men and women, or 4.1 and 4.6 METs, respectively.

Since the variables representing the PA intensity spectrum are highly colinear, partial least squares regression (PLS) was used to investigate the association with fitness and metabolic syndrome score [ 30 , 31 ]. Separate PLS models were used to investigate the associations of absolute and relative PA intensity in the whole study sample and in the 3 fitness groups. If the variables were skewed, they were square root transformed. The number of latent variables (PLS components) was selected based on Monte Carlo resampling with 10 3 repetitions and a cut-off of a quarter of a standard deviation, to ensure that the PLS model was significantly better than a model with fewer components [ 32 ].

Selectivity ratio plots were used to represent the contribution of each PA intensity spectrum variable to the association with the outcome [ 33 ]. The selectivity ratio represents the explained variance in the PA spectrum variables from the latent variables. To get an estimate of the explained variance in the health variable, the selectivity ratio is multiplied by the overall explained variance of the PLS model. However, the explained variance in the outcome by a single intensity level should not be interpreted independently, since the results from the PLS regression represent a pattern of the PA variables combined rather than independent associations with the separate variables.

The 95% confidence interval of the selectivity ratio was calculated by bootstrapping and the statistical uncertainty of the PLS model was assessed by permutation tests, both with 10 4 repetitions [ 31 ]. Furthermore, the PA intensity level where less than one third of the individuals had any movement was indicated as dashed lines in the figures. In addition, to facilitate interpretation of the results from the more advanced analyses, the dose–response association between absolute and relative moderate-to-vigorous PA and the metabolic syndrome score and fitness was calculated using linear regression. All data processing and statistical analyses were performed in MATLAB 2022a (MathWorks, Natick, MA, USA).

Characteristics and group differences

The number of participants at the Gothenburg study site was 6 266, and the number of participants with valid fitness tests were 4 513. Of the participants with valid fitness tests, 4234 had valid accelerometer measured PA. The number of participants with measurements of fitness, PA, and metabolic syndrome score, and thus used in all further analyses, was 4 176. The 4 176 participants with valid data had on average more favorable cardiometabolic health indicators compared to the excluded participants from the Gothenburg study site as well as compared to the entire SCAPIS sample. The group differences are presented in the Supplementary Table 1 in the Additional file 1.

The median age of the study sample was 57 years, and the range was 50.1–65.5 years. The sample was stratified into tertiles according to their fitness levels (low, medium, and high). Fitness tertile limits were 34.0 mL/min/kg and 39.3 mL/min/kg for men and 28.1 mL/min/kg and 33.4 mL/min/kg for women. The low fitness group overall had the least favorable cardiometabolic health indicators, and the high fitness group had the most favorable cardiometabolic health indicators (Table 1 ). In addition, the mean age was 58.4, 57.2, and 56.0 years for the low, medium, and high fitness group, respectively. The means of the standardized PA intensity spectrum variables are presented for each fitness group in Fig.  1 . The PA patterns show that the low fitness group was more sedentary and less active at all absolute intensity levels, whereas the intensity pattern of the high fitness group was the opposite. Relative intensity PA generally displayed reversed group differences with the low fitness group being most active; however, in the vigorous intensity range, the high fitness group was still more active than the other groups.

Fitness-group means of standardized PA intensity spectrum represented as absolute (left) and relative mean group fitness (right). Values are calculated in relation to total sample mean and standard deviation where 0 represents the mean and 1 the standard deviation for each of the PA intensity spectrum variables. Dashed lines indicate where less than one third of the individuals had any movement. Shaded areas represent 95% confidence intervals. Since no established relative cut off for sedentary is available, the absolute cut off at 1.5 METs was used as reference for relative sedentary time based on the average fitness in the sample. SD standard deviation, SED sedentary, LPA light PA, MPA moderate PA, VPA vigorous PA, VVPA very-vigorous PA

Association between physical activity intensity and cardiometabolic health

In terms of absolute PA intensity, the results of the PLS analysis of the whole study sample show that there was a significant association between PA intensity and the metabolic syndrome score and fitness from the mid-moderate PA range to the lower part of the very-vigorous PA range (Fig.  2 , left panels). However, the results of the fitness stratified PLS analyses show that significant associations only in the moderate PA intensity range in the low fitness group. In the medium fitness group, the strongest associations were in the moderate PA range for metabolic syndrome score and in the vigorous PA range for fitness. In the high fitness group, the strongest associations were around the vigorous to very-vigorous cut-off for both health indicators.

The absolute (left) and relative (right) PA intensity patterns associated with metabolic syndrome score (top) and fitness (bottom). The selectivity ratio represents the influence of each PA intensity level in the association with the outcome. The thick lines represent the main statistically significant part based on 95% confidence intervals, and the dashed lines indicate where less than one third of the individuals had any movement. Since no established relative cut off for sedentary is available, the absolute cut off at 1.5 METs was used as reference for relative sedentary time based on the average fitness in the sample. SED sedentary, LPA light PA, MPA moderate PA, VPA vigorous PA, VVPA very-vigorous PA

In terms of relative PA intensity, the results of the PLS analyses show that the intensities related to metabolic syndrome score and fitness were more synchronized between fitness groups (Fig.  2 , right panels). The lowest PA intensity that was significantly associated with the metabolic syndrome score and fitness was close to the relative moderate cut-off at 46% of maximal oxygen consumption for all fitness groups. Furthermore, the strongest association was apparent in the vigorous PA range corresponding to between 64 and 91% of maximal oxygen consumption. The association weakened in the very-vigorous PA range, presumably due to excessive zero score.

All PLS models were statistically significant with p  < 0.01. More details regarding the PLS models are found in Supplementary Table 2  in Additional file 1.

With coarse measures of PA, representing time spent at moderate-to-vigorous intensity, the linear regression coefficients from relative moderate-to-vigorous intensity were significantly larger than their absolute intensity counterparts for both fitness and cardiometabolic health indicators (Fig.  3 ). PA associated more strongly with fitness than with the metabolic syndrome score, especially relative intensity in the high fitness group.

Dose–response relationship from time spent at absolute or relative moderate-to-vigorous physical activity (MVPA) as independent variable and metabolic syndrome (left) and fitness (right) as dependent variable in each of the three fitness groups. Coefficients from regression models are expressed as number of standard deviations difference in the health variable from one minute increase in time spent at MVPA per day. Error bars indicate 95% confidence intervals

Fulfilment of physical activity recommendations

In the whole study sample, the mean length of time spent at absolute moderate-to-vigorous intensity was 500 (95% confidence interval 495–506) minutes per week, and 99% (99–100%) fulfilled the recommendations of at least 150 min of moderate-to-vigorous PA intensity per week. When relating PA intensity to individual fitness, the mean moderate-to-vigorous PA level was 91 (87–94) minutes per week and 21% (20–22%) fulfilled the PA recommendations. Furthermore, when relating moderate PA intensity to a sufficiently high fitness of 31.5 and 35 mL/min/kg for women and men, respectively, [ 29 ] the mean moderate-to-vigorous PA level was 56 (54–58) minutes per week and only 9% (8–10%) fulfilled the PA recommendations.

Figure  4 visualizes the translation between absolute and relative intensity. The intersections between the dashed and solid lines represent the fitness level where absolute intensity corresponds to relative intensity. For moderate intensity, this intersection is at the 5 th percentile, suggesting that the absolute intensity cut-off is too low for 95% of the sample. The vigorous and very-vigorous intersections are at the 45 th and 55 th percentiles, respectively, which are more representative of the average fitness level in the sample.

Translation between absolute and relative PA intensity levels. Individual fitness level on the x-axis and different measures of absolute intensity on the y-axis. Background colors with solid borders represent relative intensity. Dashed lines denote absolute intensity. The black dotted line represents the distribution of fitness in the study sample. The intersects between the dashed and solid lines represent the fitness level where the absolute and relative PA intensity is congruent. In individuals with a higher fitness than this, time spent at different intensities will be overestimated, which is emphasized by the colored arrows. For example, a relatively unfit individual with a maximal oxygen consumption of 23 mL/min/kg (x-axis) will have a relative MPA accelerometer cut point at 201 mg (y-axis) which corresponds to a locomotion speed of 3.2 km/h (slow walking) and a MET of 3.0 (at the intersection of the green dashed and solid lines). For comparison, an individual with a higher maximal oxygen consumption of 35 mL/min/kg will have a relative MPA accelerometer cut point at 388 mg which corresponds to a locomotion speed of 4.8 km/h (brisk walking) and a MET of 4.6. If the cut point of the unfit individual (MET of 3.0) is applied to the individual with higher fitness, the time spent in MPA is overestimated. LPA, light PA, MPA moderate PA, VPA vigorous PA, VVPA very-vigorous PA

Health-related physical activity intensity

In this study, we investigated associations between accelerometer-measured absolute and relative PA intensities and cardiometabolic health in a subgroup of participants from SCAPIS. A key finding was that relative PA intensity, determined from accelerometer data and submaximal fitness test in combination, identified the level that was most strongly associated with cardiometabolic health across all fitness levels, whereas the level of absolute PA intensity associated with cardiometabolic health shifted depending on fitness level. Furthermore, when stratifying for fitness level, the PA intensity patterns of the associations found in this study seem to align with the relative cut-offs of moderate and vigorous intensity that have been suggested based on previous intervention and physiological studies [ 2 , 5 ]. In relative terms, moderate PA intensity was required to observe significant associations with cardiometabolic health, and associations between PA intensity and cardiometabolic health were stronger for vigorous intensity.

Our results also showed that expressing PA intensity associated with cardiometabolic health in absolute terms was misleading for most individuals in this sample of adults aged 50–64 years. In particular, the absolute moderate intensity cut-off was too low for 95% of the individuals. This implies that relative moderate-to-vigorous PA was considerably more intense than absolute moderate-to-vigorous PA for most individuals and subsequently explains the significantly stronger dose–response relationship with cardiometabolic health and less time spent at this intensity.

In relative terms, the low fitness group was most active overall, which is in line with previous research [ 13 , 14 ]. However, the detailed analyses in this study showed that the high fitness group was most active at relative vigorous intensity. This might partly explain why this group had a higher fitness level than the low fitness group, despite spending much less time at relative moderate intensity. However, we cannot exclude the possibility that a higher, possibly genetically predisposed, fitness allows for more time spent at higher absolute intensities and that fitness and cardiometabolic risk (e.g. obesity) have shared genetic factors [ 34 ].

The differences in associations between absolute and relative PA and cardiometabolic health may explain some of the controversies apparent in previous research regarding the intensity of PA required to provide health benefits. Some previous studies have suggested that accelerometer-measured absolute light intensity PA is associated with cardiometabolic health [ 8 , 9 ]. These studies typically include older individuals who are less healthy and have lower fitness than in our study sample. On the contrary, studies including healthy young individuals have suggested that absolute moderate intensity PA is not sufficient for associations with health, but that absolute vigorous intensity is required [ 10 ]. Thus, for older individuals with low aerobic fitness, absolute light intensity PA could be classified as relative moderate intensity PA, whereas for younger individuals with high aerobic fitness, relative moderate intensity PA may require absolute vigorous intensity PA. Similarly, this could also explain why previous studies in individuals with low fitness suggest an inverse association between sedentary time and health [ 35 ]. Time spent sedentary is negatively correlated with time spent at light intensity [ 36 ]. This means that an association between sedentary time and health could hypothetically be due to an association between light intensity PA and health, which in turn could be considered relative moderate intensity in individuals with low fitness.

Recommendations on physical activity

The results of this study suggest that absolute and relative intensity cannot be used interchangeably for most individuals when analyzing accelerometer data, in contrast to the current recommendations on PA [ 1 , 2 ]. Instead, more emphasis should be put on the relative intensity of a given activity, for example using self-perceived exertion of PA rather than on specific activities at absolute intensities (e.g., “a substantial increase in breathing rate” instead of “brisk walking”).

If an individual is sufficiently physically active to increase fitness, the absolute intensity required to improve fitness further will also increase. Given that the health benefits of increased fitness might level off at a sufficiently high fitness level of 31.5 and 35 mL/min/kg for women and men, respectively, [ 29 ] general PA recommendations should be developed with the long-term goal to achieve these fitness levels in a middle-age population. Absolute moderate intensity would then be considered to be above 4.1 and 4.6 METs for men and women, respectively, [ 29 ] substantially higher than the current recommendations of 3 METs. The mean fitness level of 33.9 mL/min/kg in the whole sample in our study is close to the suggested sufficient fitness level and implies that these cut-points would also represent health beneficial PA intensity in this sample. These cut-points are also similar to previous research regarding accelerometer-measured moderate intensity [ 11 , 15 ]. The message in the current recommendations that every move counts is, however, still important. To improve their health and fitness, individuals with low fitness should start with PA intensity that is light on the absolute scale but corresponds to moderate on the relative scale, and then progress to more intense PA. These findings may further support individualized exercise prescription and monitoring, where relative rather than absolute intensity levels are of importance for maintenance and adherence.

Limitations

The cross-sectional study design does not allow causality to be determined. The results do not directly show which PA intensity level is most beneficial for interventions, but rather which intensity level is associated with health outcomes within each fitness group. Additionally, the results should not be interpreted as showing that individuals with low fitness would not gain health benefits from very high intensity exercise. The weaker association at absolute vigorous intensity is rather explained by very few low fit individuals performing this kind of activity.

The age span of the whole study sample was limited and the participants had a more favorable health status than in other large-scale studies [ 37 , 38 ]. The traditional absolute moderate intensity cut-off at 3 METs might be more appropriate for another population with less favorable health and lower fitness level. In that case, the absolute vigorous cut-off would instead be too high according to the results of this study. In addition, the studied sample displayed slightly more favorable cardiometabolic health indicators compared to the entire SCAPIS study. This is presumably due to the exclusion criteria of the fitness test, which likely exclude individuals with low fitness to a larger degree. However, these differences in cardiometabolic health indicators are generally smaller than the differences between fitness tertiles. The entire SCAPIS study has been shown to have only minimal selection bias in relation to the Swedish population [ 39 ].

The current methods used for processing raw accelerometer data have been shown to capture PA intensity more accurately, [ 3 , 18 , 19 ] but typically result in more time spent at moderate-to-vigorous intensity compared to traditional processing and cut-points [ 19 , 40 ]. This explains the very high fulfilment of absolute PA recommendations when using absolute intensity with crude moderate-to-vigorous cut-offs by accelerometry in this study and implies that these levels cannot be directly compared to other studies. However, this emphasizes the benefit of using a more detailed intensity spectrum instead of crude cut-points. Furthermore, this provides additional evidence that absolute moderate intensity is too low to represent health-beneficial PA in relation to the guidelines.

The stratification approach naturally misses variation available in the dataset, which weakens the overall association in the PLS models. Because of the normal distribution of fitness in the sample, there is substantially less variation in fitness in the medium fitness group. Furthermore, this study solely focuses on the effect of different PA intensities and assumes that the recommendations of 150 min per week of moderate-to-vigorous intensity PA is applicable to objectively measured PA. Future studies should incorporate the results of this study regarding PA intensity and investigate the effects of different PA volume.

Implications

Although the results of this study suggest that health-beneficial PA intensity is relative to individual fitness, the results from measurement of relative PA intensity should be interpreted with caution. In cross-sectional studies, a consideration of individual relative intensity PA is comparable to including fitness as a covariate and controlling the association between PA and health outcomes for fitness level. This distorts potential associations because it usually suggest low fit individuals are the most active [ 12 , 13 , 14 ]. In intervention studies, however, the effect of PA is more likely to be better represented by relative intensity than by absolute intensity [ 2 , 5 ]. In these settings, accelerometer-measured individual relative intensity can be directly applied to study the effect of an intervention on PA level or potential effects of PA level on other outcomes.

Instead of considering individual relative intensity in cross-sectional studies, PA intensity relative to the sample mean could be considered. This implies that the cut-points used for analyzing accelerometer data are adjusted to represent health-beneficial PA intensity in the sample studied. This can be done based on the relative cut-points of 46%, 64% and 91% of fitness level and is visualized in Fig.  4 . In the present study sample, with an average fitness level of 33.9 ml/min/kg, the cut-points used would be 4.5, 6.2 and 8.8 METs for moderate, vigorous, and very-vigorous PA intensity respectively (e.g., 0.46 × 33.9 = 15.6 ml/min/kg; 15.6 / 3.5 = 4.5 METs). This is equivalent to 391, 618 and 959 mg based on the regression coefficients presented in the methods section (e.g., (0.46 × 33.9 – 5.108) / 0.02683 = 391 mg). As another example, in patients with heart failure and a fitness level of 14.1 ml/min/kg, [ 41 ] the corresponding cut-points would be 1.9, 2.6 and 3.7 METs for moderate, vigorous and very-vigorous PA. In this example, even absolute light intensity could be considered relatively vigorous. Clearly, there is a large discrepancy between these cut-points, which emphasizes the need for adjusting them to be relevant for the sample studied. The same adjustments can be used on ActiGraph counts by applying the regression equations for the widely used cut-points from e.g. Freedson et al. [ 42 ]. However, the use of relative intensity cut-points limits the comparability of PA levels between studies.

For precise interpretation and prescription of PA intensity, maximal or submaximal fitness tests would ideally be considered in both research and clinical practice. Alternatively, when fitness tests are not available, fitness level may be estimated from age, sex and self-perceived physical capacity [ 43 ]. To aid in clinical practice, we provide a tool for translation between relative and absolute intensity (Fig.  4 ), to be considered when interpreting research on physical activity or when prescribing PA as treatment.

Our study suggests that accelerometer-measured relative moderate-to-vigorous PA intensity represents the PA intensity associated with health regardless of fitness level. Absolute moderate PA intensity represents health-associated PA only in a small proportion of individuals. Relative intensity cut-offs representing moderate and vigorous PA intensity align with the associations to health indicators observed in this study when applied to accelerometer data. Traditional absolute moderate intensity PA accelerometer cut-offs are too low for most individuals and should be adopted to individual fitness or the fitness level in the sample studied. Absolute and relative PA intensity cannot be used interchangeably in PA recommendations, and more emphasis should be put on relative intensity, for example by using self-perceived exertion, when communicating the PA intensity required to benefit health.

Availability of data and materials

The data that support the findings of this study are available from the SCAPIS study organization ( www.scapis.org ) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the SCAPIS study organization.

Abbreviations

Glycated hemoglobin

High density lipoprotein

Metabolic equivalents of task

Physical activity

Partial least squares regression

Systolic blood pressure

Swedish cardiopulmonary bioimage study

Bull FC, Al-Ansari SS, Biddle S, Borodulin K, Buman MP, Cardon G, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. 2020;54(24):1451–62.

Article   PubMed   Google Scholar  

2018 Physical Activity Guidelines Advisory Committee. 2018 Physical Activity Guidelines Advisory Committee Scientific Report. Washington, DC: U.S. Department of Health and Human Services; 2018 p. 779. Available from: https://health.gov/paguidelines/second-edition/report/pdf/PAG_Advisory_Committee_Report.pdf

Fridolfsson J, Börjesson M, Arvidsson D. A Biomechanical Re-Examination of Physical Activity Measurement with Accelerometers. Sensors. 2018;18(10):3399.

Article   PubMed   PubMed Central   Google Scholar  

Arvidsson D, Fridolfsson J, Börjesson M. Measurement of physical activity in clinical practice using accelerometers. J Intern Med. 2019;286(2):137–53.

Article   CAS   PubMed   Google Scholar  

Garber C, Blissmer B, Deschenes M, Franklin B, Lamonte M, Lee IM, et al. Quantity and Quality of Exercise for Developing and Maintaining Cardiorespiratory, Musculoskeletal, and Neuromotor Fitness in Apparently Healthy Adults: Guidance for Prescribing Exercise. Med Sci Sports Exerc. 2011;43(7):1334–59.

Ekblom-Bak E, Ekblom Ö, Andersson G, Wallin P, Söderling J, Hemmingsson E, et al. Decline in cardiorespiratory fitness in the Swedish working force between 1995 and 2017. Scand J Med Sci Sports. 2019;29(2):232–9.

Lear SA, Hu W, Rangarajan S, Gasevic D, Leong D, Iqbal R, et al. The effect of physical activity on mortality and cardiovascular disease in 130 000 people from 17 high-income, middle-income, and low-income countries: the PURE study. The Lancet. 2017;390(10113):2643–54.

Article   Google Scholar  

LaMonte MJ, Lewis CE, Buchner DM, Evenson KR, Rillamas-Sun E, Di C, et al. Both Light Intensity and Moderate-to-Vigorous Physical Activity Measured by Accelerometry Are Favorably Associated With Cardiometabolic Risk Factors in Older Women: The Objective Physical Activity and Cardiovascular Health (OPACH) Study. J Am Heart Assoc. 2017;6(10):e007064.

Loprinzi PD, Lee H, Cardinal BJ. Evidence to Support Including Lifestyle Light-Intensity Recommendations in Physical Activity Guidelines for Older Adults. Am J Health Promot. 2015;29(5):277–84.

Owens S, Galloway R, Gutin B. The Case for Vigorous Physical Activity in Youth. Am J Lifestyle Med. 2016;11(2):96–115.

Gil-Rey E, Maldonado-Martín S, Gorostiaga EM. Individualized Accelerometer Activity Cut-Points for the Measurement of Relative Physical Activity Intensity Levels. Res Q Exerc Sport. 2019;90(3):327–35.

Gil-Rey E, Maldonado-Martín S, Palacios-Samper N, Gorostiaga EM. Objectively measured absolute and relative physical activity intensity levels in postmenopausal women. Eur J Sport Sci. 2019;19(4):539–48.

Vähä-Ypyä H, Sievänen H, Husu P, Tokola K, Vasankari T. Intensity Paradox—Low-Fit People Are Physically Most Active in Terms of Their Fitness. Sensors. 2021;21(6):2063.

Vähä-Ypyä H, Sievänen H, Husu P, Tokola K, Mänttäri A, Heinonen OJ, et al. How adherence to the updated physical activity guidelines should be assessed with accelerometer? Eur J Public Health. 2022;32(Supplement_1):i50-5.

Siddique J, Aaby D, Montag SE, Sidney S, Sternfeld B, Welch WA, et al. Individualized Relative-Intensity Physical Activity Accelerometer Cut Points. Med Sci Sports Exerc. 2020;52(2):398–407.

Bergström G, Berglund G, Blomberg A, Brandberg J, Engström G, Engvall J, et al. The Swedish CArdioPulmonary BioImage Study: objectives and design. J Intern Med. 2015;278(6):645–59.

Ekblom-Bak E, Börjesson M, Bergman F, Bergström G, Dahlin-Almevall A, Drake I, et al. Accelerometer derived physical activity patterns in 27.890 middle-aged adults: The SCAPIS cohort study. Scand J Med Sci Sports. 2022;32(5):866–80.

Fridolfsson J, Börjesson M, Buck C, Ekblom Ö, Ekblom-Bak E, Hunsberger M, et al. Effects of Frequency Filtering on Intensity and Noise in Accelerometer-Based Physical Activity Measurements. Sensors. 2019;19(9):2186.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Fridolfsson J, Börjesson M, Ekblom-Bak E, Ekblom Ö, Arvidsson D. Stronger Association between High Intensity Physical Activity and Cardiometabolic Health with Improved Assessment of the Full Intensity Range Using Accelerometry. Sensors. 2020;20(4):1118.

Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, Mcdowell M. Physical Activity in the United States Measured by Accelerometer. Med Sci Sports Exerc. 2008;40(1):181–8.

Migueles JH, Cadenas-Sanchez C, Ekelund U, Nyström CD, Mora-Gonzalez J, Löf M, et al. Accelerometer Data Collection and Processing Criteria to Assess Physical Activity and Other Outcomes: A Systematic Review and Practical Considerations. Sports Med. 2017;47(9):1821–45.

Arvidsson D, Fridolfsson J, Buck C, Ekblom Ö, Ekblom-Bak E, Lissner L, et al. Reexamination of Accelerometer Calibration with Energy Expenditure as Criterion: VO2net Instead of MET for Age-Equivalent Physical Activity Intensity. Sensors. 2019;19(15):3377.

Björkman F, Ekblom-Bak E, Ekblom Ö, Ekblom B. Validity of the revised Ekblom Bak cycle ergometer test in adults. Eur J Appl Physiol. 2016;116(9):1627–38.

Fridolfsson J, Buck C, Hunsberger M, Baran J, Lauria F, Molnar D, et al. High-intensity activity is more strongly associated with metabolic health in children compared to sedentary time: a cross-sectional study of the I.Family cohort. Int J Behav Nutr Phys Act. 2021;18(1):90.

Gurka MJ, Filipp SL, Pearson TA, DeBoer MD. Assessing Baseline and Temporal Changes in Cardiometabolic Risk Using Metabolic Syndrome Severity and Common Risk Scores. J Am Heart Assoc Cardiovasc Cerebrovasc Dis. 2018;7(16):e009754.

Ahrens W, Moreno LA, Mårild S, Molnár D, Siani A, De Henauw S, et al. Metabolic syndrome in young children: definitions and results of the IDEFICS study. Int J Obes. 2005;2014(38 Suppl 2):S4-14.

Google Scholar  

Eckel RH, Grundy SM, Zimmet PZ. The metabolic syndrome. The Lancet. 2005;365(9468):1415–28.

Article   CAS   Google Scholar  

World Health Organization. Waist circumference and waist-hip ratio: report of a WHO expert consultation. Geneva. 2008;8–11:2011.

Blair SN, Kohl HW III, Paffenbarger RS Jr, Clark DG, Cooper KH, Gibbons LW. Physical Fitness and All-Cause Mortality: A Prospective Study of Healthy Men and Women. JAMA. 1989;262(17):2395–401.

Aadland E, Kvalheim OM, Anderssen SA, Resaland GK, Andersen LB. Multicollinear physical activity accelerometry data and associations to cardiometabolic health: challenges, pitfalls, and potential solutions. Int J Behav Nutr Phys Act. 2019;16(1):74.

Fridolfsson J, Arvidsson D, Andersen LB, Thorsson O, Wollmer P, Rosengren B, et al. Physical activity spectrum discriminant analysis—A method to compare detailed patterns between groups. Scand J Med Sci Sports. 2021;31(12):2333–42.

Kvalheim OM, Arneberg R, Grung B, Rajalahti T. Determination of optimum number of components in partial least squares regression from distributions of the root-mean-squared error obtained by Monte Carlo resampling. J Chemom. 2018;32(4):e2993.

Kvalheim OM. Interpretation of partial least squares regression models by means of target projection and selectivity ratio plots. J Chemom. 2010;24(7–8):496–504.

Schnurr TM, Gjesing AP, Sandholt CH, Jonsson A, Mahendran Y, Have CT, et al. Genetic Correlation between Body Fat Percentage and Cardiorespiratory Fitness Suggests Common Genetic Etiology. PLoS ONE. 2016;11(11):e0166738.

Ekelund U, Tarp J, Steene-Johannessen J, Hansen BH, Jefferis B, Fagerland MW, et al. Dose-response associations between accelerometry measured physical activity and sedentary time and all cause mortality: systematic review and harmonised meta-analysis. BMJ. 2019;366:l4570.

van der Ploeg HP, Hillsdon M. Is sedentary behaviour just physical inactivity by another name? Int J Behav Nutr Phys Act. 2017;23(14):142.

Millett ERC, Peters SAE, Woodward M. Sex differences in risk factors for myocardial infarction: cohort study of UK Biobank participants. BMJ. 2018;7:k4247.

Muntner P, Miles MA, Jaeger BC, Hannon L III, Hardy ST, Ostchega Y, et al. Blood Pressure Control Among US Adults, 2009 to 2012 Through 2017 to 2020. Hypertension. 2022;79(9):1971–80.

Bergström G, Persson M, Adiels M, Björnson E, Bonander C, Ahlström H, et al. Prevalence of Subclinical Coronary Artery Atherosclerosis in the General Population. Circulation. 2021;144(12):916–29.

Ekblom-Bak E, Olsson G, Ekblom Ö, Ekblom B, Bergström G, Börjesson M. The Daily Movement Pattern and Fulfilment of Physical Activity Recommendations in Swedish Middle-Aged Adults: The SCAPIS Pilot Study. PLoS ONE. 2015;10(5):e0126336.

Canada JM, Trankle CR, Buckley LF, Carbone S, Abouzaki NA, Kadariya D, et al. Severely Impaired Cardiorespiratory Fitness in Patients With Recently Decompensated Systolic Heart Failure. Am J Cardiol. 2017;120(10):1854–7.

Freedson PS, Melanson E, Sirard J. Calibration of the Computer Science and Applications Inc accelerometer. Med Sci Sports Exerc. 1998;30(5):777–81.

Jackson AS, Blair SN, Mahar MT, Wier LT, Ross RM, Stuteville JE. Prediction of functional aerobic capacity without exercise testing. Med Sci Sports Exerc. 1990;22(6):863–70.

Download references

Acknowledgements

We thank Rosie Perkins (Institute of Medicine, University of Gothenburg) for editing the manuscript.

Open access funding provided by University of Gothenburg. The main funding body of the Swedish CArdioPulmonary bioImage Study (SCAPIS) is the Swedish Heart–Lung Foundation. The study is also funded by the Knut and Alice Wallenberg Foundation, the Swedish Research Council, VINNOVA (Sweden’s Innovation agency), the University of Gothenburg and Sahlgrenska University Hospital, Karolinska Institutet and Stockholm County Council, Linköping University and University Hospital, Lund University and Skåne University Hospital, Umeå University and University Hospital, Uppsala University and University Hospital. ÖE and EEB were funded by Skandia Risk&Hälsa and MB by Heart-and Lung foundation (grant 20210270).

Author information

Authors and affiliations.

Center for Health and Performance, Department of Food and Nutrition and Sport Science, Faculty of Education, University of Gothenburg, Box 300, 405 30, Gothenburg, Sweden

Jonatan Fridolfsson & Daniel Arvidsson

Department of Physical Activity and Health, Swedish School of Sport and Health Sciences, Stockholm, Sweden

Elin Ekblom-Bak & Örjan Ekblom

Department of Molecular and Clinical Medicine, Institute of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

Göran Bergström

Department of Clinical Physiology, Region Västra Götaland, Gothenburg, Sweden

Center for Lifestyle Intervention, Department of Molecular and Clinical Medicine, Institute of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

Mats Börjesson

Sahlgrenska University Hospital, Region Västra Götaland, Gothenburg, Sweden

You can also search for this author in PubMed   Google Scholar

Contributions

JF, DA and MB conceptualized the study. EEB, ÖE, GB and MB were responsible for the data collection. JF performed the data processing, analysis, visualization and wrote the initial manuscript draft. DA, EEB, ÖE, GB and MB made substantial manuscript revisions.

Corresponding author

Correspondence to Jonatan Fridolfsson .

Ethics declarations

Ethics approval and consent to participate.

SCAPIS has been approved by the ethics committee at Umeå University (no. 2021–228-31 M) and the current study has received specific approval by the Regional ethical board in Gothenburg (no. 638–16). Written, informed consent was retrieved from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: table 1. .

Characteristics of the study sample, individuals from the Gothenburg site excluded due to missing measurement of fitness and PA, and the entire SCAPIS sample. Mean (standard deviation).  Table 2. PLS model details. Number of PLS components was chosen based on cross validation.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Cite this article.

Fridolfsson, J., Arvidsson, D., Ekblom-Bak, E. et al. Accelerometer-measured absolute versus relative physical activity intensity: cross-sectional associations with cardiometabolic health in midlife. BMC Public Health 23 , 2322 (2023). https://doi.org/10.1186/s12889-023-17281-4

Download citation

Received : 02 October 2023

Accepted : 21 November 2023

Published : 24 November 2023

DOI : https://doi.org/10.1186/s12889-023-17281-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Cardiometabolic risk factors
• Cardiovascular disease
• Cardiovascular fitness
• Recommendations

BMC Public Health

ISSN: 1471-2458

• Submission enquiries: [email protected]
• General enquiries: [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• Advanced Search
• Journal List
• Bone Joint Res
• v.11(5); 2022 May

The metabolic equivalent of task score

Thomas c. edwards.

1 MSk Lab, Imperial College London, Sir Michael Uren Biomedical Engineering Research Hub, London, UK

Brogan Guest

Kartik logishetty, alexander d. liddle, justin p. cobb.

This study investigates the use of the metabolic equivalent of task (MET) score in a young hip arthroplasty population, and its ability to capture additional benefit beyond the ceiling effect of conventional patient-reported outcome measures.

From our electronic database of 751 hip arthroplasty procedures, 221 patients were included. Patients were excluded if they had revision surgery, an alternative hip procedure, or incomplete data either preoperatively or at one-year follow-up. Included patients had a mean age of 59.4 years (SD 11.3) and 54.3% were male, incorporating 117 primary total hip and 104 hip resurfacing arthroplasty operations. Oxford Hip Score (OHS), EuroQol five-dimension questionnaire (EQ-5D), and the MET were recorded preoperatively and at one-year follow-up. The distribution was examined reporting the presence of ceiling and floor effects. Validity was assessed correlating the MET with the other scores using Spearman’s rank correlation coefficient and determining responsiveness. A subgroup of 93 patients scoring 48/48 on the OHS were analyzed by age, sex, BMI, and preoperative MET using the other metrics to determine if differences could be established despite scoring identically on the OHS.

Postoperatively the OHS and EQ-5D demonstrate considerable negatively skewed distributions with ceiling effects of 41.6% and 53.8%, respectively. The MET was normally distributed postoperatively with no relevant ceiling effect. Weak-to-moderate significant correlations were found between the MET and the other two metrics. In the 48/48 subgroup, no differences were found comparing groups with the EQ-5D, however significantly higher mean MET scores were demonstrated for patients aged < 60 years (12.7 (SD 4.7) vs 10.6 (SD 2.4), p = 0.008), male patients (12.5 (SD 4.5) vs 10.8 (SD 2.8), p = 0.024), and those with preoperative MET scores > 6 (12.6 (SD 4.2) vs 11.0 (SD 3.3), p = 0.040).

The MET is normally distributed in patients following hip arthroplasty, recording levels of activity which are undetectable using the OHS.

Cite this article: Bone Joint Res  2022;11(5):317–326.

Article focus

• Substantial ceiling effects of some conventional patient-reported outcome measure (PROM) scores limit their ability to discriminate between high-functioning postoperative hip arthroplasty patients.
• This article introduces the concept of using metabolic equivalent of task (MET) values in a young hip arthroplasty cohort to determine whether this approach can capture additional benefit beyond the ceiling effect of conventional PROMs.

Key messages

• Postoperative ceiling effects of the Oxford Hip Score (OHS) and EuroQol five-dimension questionnaire (EQ-5D) miss clinically substantial gains in higher-level health-related quality of life.
• MET is a simple, versatile measure of physical activity, with no postoperative ceiling effect.
• Clinicians, researchers, and health economists who wish to capture the full benefit from hip arthroplasty surgery should consider using MET in addition to conventional PROMs.

Strengths and limitations

• This multicentre, multiple-surgeon study addresses an important concept, which will inform future hip arthroplasty researchers designing studies assessing new technologies or techniques.
• The large proportion of non-responders and included cohort being substantially younger than the national average for hip arthroplasty limits its generalizability to the wider population.

Introduction

Primary hip arthroplasty is an effective intervention for improving pain and restoring function. 1 Patient-reported outcome measures (PROMs) such as the Oxford Hip Score (OHS), 2 , 3 which is routinely collected before and after hip arthroplasty in the UK, reliably and predictably report considerable, cost-effective improvements in pain and function. 4 , 5 However, as a consequence of the efficacy of hip arthroplasty, the distribution of postoperative scores is highly negatively skewed. In two national registries, the modal score on the postoperative OHS was 100% (48/48 points) with up to 20% of patients recording this score. 6 , 7

The population of patients presenting for hip arthroplasty has evolved and their expectations are different from those undergoing surgery when the OHS was introduced. 8 In a study by Scott et al, 9 40% of hip arthroplasty patients considered returning to sporting activity a ‘very important’ preoperative expectation, but such activities are not captured by the OHS. By relying solely on this skewed metric, potential health gain in return to sporting activity from innovative techniques will remain undetected, while other patient groups may be inappropriately told that their clinical results are as good as they can get, despite dissatisfaction with the level of activity that they have achieved. 7 , 10 , 11 In the UK, the pre- to postoperative change in OHS is used by government-backed initiatives including ‘getting it right first time’ (GIRFT) and the NHS best practice tariff (BPT), 12 , 13 to measure success at the institutional level. With the skewed OHS as the metric, the only way to improve the health gain obtained is by refusing care until the preoperative scores are low enough. This approach may have a role in healthcare rationing but should not impede the scientific desire to measure higher-level function. As degree of health gain is so closely related to preoperative score, and as preoperative score may vary, there is a need for a metric which measures outcome equally independently of preoperative health state.

Alternative scores have been developed with the aim of being more discriminative in high-functioning patients. Several constructs have been suggested, such as joint perception as measured by the Forgotten Joint Score (FJS) and physical activity scores. The FJS assesses patients’ awareness of their joint arthroplasty performing different tasks, with the optimal outcome being a ‘forgotten’ artificial joint. In a study comparing the outcomes of robotic and manual THA, the authors found no clinically relevant difference using the OHS, however the robotic group did substantially better using the FJS. 14 The authors support the idea that the ceiling effect of the OHS limits its use for comparing high-functioning postoperative patients, with their results indicating that the FJS may be more discriminative. While this sounds encouraging, other authors have reported ceiling effects of 20% to 30% using the FJS in a postoperative hip arthroplasty cohort, and so problems may still exist with skewed distributions using this score. 15 , 16

Physical activity metrics may be another solution, with a number of valid and reliable metrics such as the University of California, Los Angeles (UCLA) activity scale available. 17 This score appears to have no ceiling effect and is simple to use, however it only includes a small number of activities and does not account for the individual activity intensity. 17 One potential solution to this problem is the use of metabolic equivalent of task (MET) values, which numerically quantify the energy expenditure of over 800 activities comparing them to energy expenditure at rest. 18 This sophisticated, personalized approach to quantifying activity energy expenditure has been validated as a surrogate for general cardiovascular fitness, correlating well with both objective activity measures, such as pedometers, as well as the development of cardiovascular disease and mortality. 19 , 20 Exercise at an intensity that raises the heart rate is now well established as being an effective health maintenance intervention. MET values have been used to confirm this beneficial effect: in a twin study, ‘conditioning exercise’ offers substantial protection against risk of death when compared with sedentary or occasional exercising. 20 Although not yet commonly used following arthroplasty surgery, with a simplification to measure activity intensity without the performance time or frequency, the MET may be a robust way of comparing activity in postoperative hip arthroplasty patients, demonstrating activity levels that have real relevance to health and life expectancy.

This study therefore aims to answer two important questions: 1) does the MET have a postoperative ceiling effect that may limit its ability to discriminate between high-performing postoperative patients; and 2) can the MET demonstrate continued improvement and health gains beyond the maximal OHS, establishing differences between postoperative patients who score 48/48?

Study design

This study was a retrospective analysis of anonymous data, collected prospectively from consenting primary hip arthroplasty patients as part of an ongoing, longitudinal study of gait analysis in lower limb arthroplasty (REC reference: 14/SC/1243). Patients from this study were eligible for inclusion if they underwent a primary hip arthroplasty under one of 13 surgeons at 12 sites between 2014 and 2018. Patients were excluded if they had revision surgery, an alternative hip procedure, or incomplete data either preoperatively or at one-year follow-up. Demographic data and the patient-reported answers to three PROMs questionnaires (EuroQol five-dimension questionnaire (EQ-5D), OHS, and the MET) were recorded preoperatively and at one year postoperatively.

Demographic details

A total of 751 patients were initially identified on our electronic database; 73 were excluded having had an alternative or revision procedure, and a further 457 patients were excluded due to lack of preoperative or one-year PROM scores. Overall 221 patients, including 117 THAs (53%) and 104 HRAs (47%), with a mean age of 59 years (SD 11), were analyzed in this study. Demographic data are detailed in Table I . The 221 responding patients with full datasets were a mean four years younger than the 457 non-responders (59 years (SD 11) vs 63 years (SD 12), p < 0.001, independent-samples t -test).

Demographics and patient-reported outcome measures.

EQ-5D, EuroQol five-dimension questionnaire; HRA, hip resurfacing arthroplasty; IQR, interquartile range; MET, metabolic equivalent of task; OHS, Oxford Hip Score; SD, standard deviation; THA, total hip arthroplasty.

Using a similar methodology to Amstutz and Le Duff, 21 the MET asks patients to choose three physical activities that are important to them, and that are affected by their joint problem. These initially selected activities remain the same at all follow-up timepoints. Patients then rate the intensity at which they currently perform the activity on a visual scale from 0 to 100. METs are numerical values assigned to demonstrate the energy expenditure used performing different tasks. One MET is equivalent to energy expenditure during rest and is approximately equal to 3.5 ml O 2 kg -1 min -1 in adults. 19 Using Arizona State Universities compendium of activities, 18 the MET values associated with each activity are recorded. An example is running, which has a range of values between 4.5 (jogging on a mini-tramp) and 23 METs (running a 4.3 minute mile). Based on this reference range, the patient’s self-reported intensity score is then used to work out a value for the METs they are currently doing. This is done by subtracting the lower value of the MET reference range from the higher value, multiplying this by the percentage intensity expressed as a decimal, and then adding back on the lower reference value. Using the above as an example, if a patient rated their intensity as 50% in running (range 4.5 to 23 METs) their MET score would be worked out as ((23 to 4.5)*0.5) + 4.5 = 13.75 METs. The MET is the maximum value scored from the three chosen activities. In the full MET, the frequency and duration of physical activity are recorded; we omitted these aspects from the score in favour of intensity, to avoid measuring cardiorespiratory fitness and also to avoid under-representing performance through measuring high-intensity but infrequently performed activities, such as skiing. 21 , 22

Distribution of scores

Data were analyzed to demonstrate the distribution, presence of ceiling or floor effects, concurrent validity of the MET in terms of its responsiveness, and correlations between the MET and the two conventional PROMs. Other authors have suggested when validating physical activity metrics that a weak-to-moderate correlation would be expected between the activity metric and conventional PROMs. 8

Health gains

Recent literature has established that as preoperative OHS increases, the improvement in score decreases. 23 To investigate whether health gains in the MET and EQ-5D are also limited by the level of preoperative joint symptoms, the relationship between preoperative OHS and patient improvement at one year using the three metrics was plotted. Fractional polynomial regression plots were used to demonstrate the likely increase in each metric for a given preoperative OHS score.

48/48 sub-cohort analysis

A subgroup analysis was performed on a cohort of patients with the maximum postoperative OHS. Previous studies have highlighted that postoperative physical activity (as measured on the UCLA activity score) in hip arthroplasty patients may be higher in younger patients, male patients, those with higher preoperative activity levels, and those with lower BMI. 24 , 25 Based on this, the 48/48 scoring patient cohort was divided into categories (age < or > 60 years, male or female, preoperative MET < or > 6, and BMI < or > 25 kg/m 2 ) and compared using the MET and EQ-5D at one year postoperatively. Previous literature has classified activities as light (< 3), moderate (3 to 6), or vigorous (> 6) according to their MET values. 26 , 27 Therefore, in this analysis, the threshold for high preoperative activity was set at 6 METs.

Statistical analysis

Statistical analysis was performed using Stata/IC 10.1 (StataCorp, USA). Data were first tested for normality visually using histograms and normal Q-Q plots. To quantify the shape and symmetry of the distribution about the mean, kurtosis and skewness values were calculated. A standard normal distribution is generally considered to have a kurtosis value of 3 and a skew value of 0. 28 For independent data, parametric variables were compared using the independent-samples t -test, and non-parametric data were compared using the Mann-Whitney U test. Paired data were compared using the paired t -test. Ceiling and floor effects were calculated as the percentage of patients scoring the maximum or minimum scores, respectively. As previously indicated in the literature, ceiling or floor effects of > 15% were considered relevant. 7 Construct validity of the MET was assessed by examining the responsiveness of the score to change using the standardized response mean (SRM; calculated by dividing the mean change in score over the one-year time period by the standard deviation (SD) of that change), and concurrent validity was assessed by measuring correlations between scores as calculated using Spearman’s rank correlation coefficient. R s values of 0.3 to 0.5 were considered as weak-to-moderate correlations. 8 Statistical significance was set at p < 0.05.

Distribution

Preoperatively, the distribution of the OHS was normal (skew -0.14, kurtosis 2.57) and that of the EQ-5D was bimodal with a skewness value of -0.99 and a kurtosis of 4.03 ( Figure 1 and Figure 2 ). The MET demonstrates a slight positive skew of 0.59 representing a floor effect, with the commonest score being zero, on account of hip pain and near-normal kurtosis of 3.18 ( Figure 3 ).

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of Oxford Hip Scores (OHS) preoperatively and at one-year follow-up. Solid vertical lines represent mean values, dashed vertical lines represent the median.

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of EuroQol-5D (EQ-5D) index scores preoperatively and at one-year follow-up. Solid vertical lines represent mean values, dashed vertical lines represent the median.

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of metabolic equivalent of task (MET) scores preoperatively and at one-year follow-up. Solid vertical lines represent mean values, dashed vertical lines represent the median.

Postoperatively, both the OHS and EQ-5D scores demonstrate a substantial negative skew ( Figure 1 and Figure 2 ). The kurtosis value of the OHS was 13.07 with a skewness value of -3.12, while the EQ-5D demonstrated a kurtosis value of 6.23 and a skewness value of -1.96. The MET on the other hand exhibited a normal distribution postoperatively, centred around a mean of 10.7 (SD 3.8) with very little skew (0.40) and near normal kurtosis (4.46; Figure 3 and Table II ).

Distribution, ceiling and floor effects, and responsiveness of the three metrics before and after surgery.

Ceiling and floor effects calculated as the percentage of patients scoring the maximum and minimum possible scores, respectively. Skewness and kurtosis values numerically represent the distribution of scores. A normal distribution has a skew of 0 and a kurtosis of 3. Standardized response mean calculated as the mean difference in scores divided by the standard deviation of that difference.

EQ-5D, EuroQol five-dimension questionnaire; IQR, interquartile range; MET, metabolic equivalent of task; OHS, Oxford Hip Score; SD, standard deviation; SRM, standardized response mean.

Ceiling and floor effects

No floor effects were seen for the OHS or EQ-5D, but substantial ceiling effects of 41.6% and 53.8% were seen at one year follow-up in the OHS and EQ-5D, respectively. Preoperatively the MET had a moderate floor effect of 25.3%, while no relevant ceiling effect was noted postoperatively ( Table II ).

Spearman’s rank correlation coefficient was weak-to-moderate, but there were statistically significant correlations between the MET and EQ-5D both preoperatively (r s = 0.46, p < 0.001) and to a lesser extent at one-year follow-up (r s = 0.32 p < 0.001, Spearman’s rank correlation). Similarly, weak-to-moderate correlations were demonstrated between the MET and the OHS both preoperatively (r s = 0.46, p < 0.001, Spearman’s rank correlation) and at one-year follow-up (r s = 0.30, p < 0.001, Spearman’s rank correlation).

Improvement in score and responsiveness

All three metrics demonstrated excellent responsiveness with effect sizes as determined by the SRMs of > 1 ( Table III ). The fractional polynomial predictive plots in Figure 4 demonstrate a strong negative relationship between preoperative OHS and improvement in score for both EQ-5D ( Figure 4a ) and OHS ( Figure 4b ). For the MET score, this relationship is far less clear, with an initial decrease in improvement seen in patients who have lower preoperative OHS, while in patients with higher preoperative OHS the MET is progressively more responsive ( Figure 4c ).

Fractional polynomial regression plots demonstrating predicted improvement in score at one-year follow-up for a given preoperative Oxford Hip Score (OHS) for a) OHS, b) EuroQol five-dimension questionnaire (EQ-5D), and c) metabolic equivalent of task (MET). Fractional polynomial fit line with 95% confidence intervals (CIs) demonstrated in grey.

Outcome scores by category for 48/48 Oxford Hip Score sub-cohort.

Except for p-values marked with the † symbol (Mann-Whitney U test), all p-values in this table were calculated using the independent-samples t -test.

EQ-5D, EuroQol five-dimension questionnaire; IQR, interquartile range; MET, metabolic equivalent of task; SD, standard deviation.

A total of 92 postoperative patients scored 48/48 on the OHS following surgery. The histograms in Figure 5 demonstrate the distribution of EQ-5D in this group, which has a strong negative skew, with the majority of patients scoring the maximal score of 1 ( Figure 5a ). The MET on the other hand exhibits a near normal distribution of scores, despite all patients scoring the same on the OHS ( Figure 5b ). When subdivided into groups, patients aged under 60 years scored significantly higher on the MET than patients over 60 years of age (mean 12.7 (SD 4.7) vs 10.6 (SD 2.4), p = 0.008, independent-samples t -test), as did male patients (mean 12.5 (SD 4.5) vs 10.8 (SD 2.8), p = 0.024, independent-samples t -test) and patients with higher activity levels on their preoperative MET scores (mean 12.6 (SD 4.2) vs 11.0 (SD 3.3), p = 0.040, independent-samples t -test) ( Figure 6a ). No significant differences were found comparing patients by BMI using the MET or comparing any of the groups using the EQ-5D ( Figure 6b , Table III ).

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of a) metabolic equivalent of task score (MET) and b) EuroQol five-dimension questionnaire (EQ-5D), at one-year follow-up for the subgroup of patients who all scored 48/48 on the Oxford Hip Score (n = 92).

Column scatter for the subgroup of 92 patients scoring 48/48 on the Oxford Hip Score, compared by age group, BMI, preoperative metabolic equivalent of task (MET) score, and sex using: a) MET scores; and b) EuroQol five-dimension questionnaire (EQ-5D) scores. The solid horizontal line represents the median and the whiskers represent the interquartile range. Statistically significant p-values have been indicated.

This retrospective study set out to determine whether the MET score could capture differences in function that were not detectable by the OHS or EQ-5D in an active hip arthroplasty population. This question was answered: the MET score does deliver a symmetrical metric with a normal distribution in a postoperative population, capturing differences in activity levels that were not detectable using the OHS. This question is relevant for health economists, policy makers, and those designing clinical trials. Health benefits that can be captured simply, without the need for expensive equipment or licences, should help drive commissioning choices. By demonstrating that patients can improve past the OHS maximum score, we have revealed an opportunity that was otherwise denied: using this metric, surgeons who are currently penalized for failing to deliver adequate health gains may now be able to justify their offering of arthroplasty in younger and more active patients. By restricting health gains to the OHS, commissioners may unfairly restrict access to arthroplasty surgery, or unfairly penalize hospitals for not achieving satisfactory results if these decisions are based solely on health gains as measured by the OHS.

Although there is more work to be done in this area, the aspects of validity measured in the present study support its use as a metric for the outcome of hip arthroplasty surgery. The MET demonstrated evidence of concurrent validity with weak-to-moderate correlations found with both the OHS and EQ-5D. Naal et al 17 used a similar approach, establishing weak-to-moderate correlations with three different physical activity scores and the OHS. One potential limitation was that the present study did not validate the MET against another validated physical activity metric or objective physical activity measures such as a pedometer or exercise log. However, the authors note that the face validity of using MET values has already been well established by other similar MET-based scores. 18 , 19 Although not validated specifically for use in arthroplasty, the International Physical Activity Questionnaire (IPAQ) score is a MET-based score, shown to be valid and reliable for use in the general population measuring activity levels. 19 It differs from the MET, being a better measure of cardiorespiratory fitness, whereas the MET is personalized to patients’ sporting aspirations. The major advantage of the MET is that no matter what activities patients choose, the scores are comparable and relevant to their joint disease. Furthermore, the numeric MET values assigned by the University of Arizona are objective, being based upon oxygen consumption. 18 , 19 Therefore, the authors considered concurrent validity with other hip-specific and generic PROMs, alongside responsiveness, encouraging validation data for using the score in this cohort, however further work in this area would be an interesting avenue for future research.

Responsiveness is considered another aspect of construct validity. 29 The greater the responsiveness, the more accurate a metric is in detecting change when it has occurred. The MET had a SRM of 1.17, which indicates a large effect size or an excellent response to change over time. 8 The calculated SRMs for EQ-5D and OHS in this cohort were found to be similar to previously published literature, further validating our findings. 30

Unlike the OHS and EQ-5D, the postoperative MET had a normal distribution and exhibited no ceiling effect. Substantial postoperative ceiling effects were found for the OHS (41.6%) and EQ-5D (53.8%). In general, ceiling effects or floor effects are considered problematic when 15% or more of the cohort score the best or worst scores. 7 , 10 By having large numbers of patients scoring the best or worst scores, the metric is rendered insensitive to detecting differences at the extremes of the scale. 7 , 10 Other studies have demonstrated strong ceiling effects in the OHS of 19.9%, 6 and even more pronounced ceiling effects for the EQ-5D of 39.8%. 30 While the pattern of these findings support our results, our population demonstrated a much higher percentage ceiling effect for both metrics. This may be related to the studied population which included a younger, more active cohort than that used in other studies. While other scores have been developed with the aim of reducing the impact of ceiling effect, unfortunately problematic ceiling effects may still exist. In a recent study, the FJS was reported to have a ceiling effect of 31.9%, similar to those reported for more conventional PROMs. 15 In addition, the FJS has reported a substantial floor effect of 22.4%, suggesting that there may be problems discriminating at both ends of the score. 31

While the MET showed no postoperative ceiling effect, it did show a preoperative floor effect, similarly to the FJS. This is not surprising given that the formulation of the question specifies the selection of tasks that have been negatively affected by the respondent’s hip pain. A similar preoperative floor effect has been observed in validation studies looking at other physical activity-based outcome measures such as the Tegner score. 17 When using MET solely as an assessment of postoperative outcome rather than of preoperative disease state, this floor effect is unimportant. If it were to be used for the former, the question may have to be re-formulated.

Both the OHS and EQ-5D demonstrated very little predicted improvement towards the upper end of the preoperative OHS scale. The MET on the other hand shows continued predicted improvements, with a 6 MET improvement predicted for patients who score 48/48 on the OHS. A large registry study by Price et al 23 demonstrated a similar effect using the OHS, with the likelihood of seeing a meaningful clinical improvement decreasing with higher preoperative scores. The authors conclude that at a preoperative score of 40 or above, there was a 0% chance of meaningful improvement, suggesting this as a threshold for referral. 23 The present study suggests that even though these higher-scoring preoperative patients do not show improvement using the OHS, they do show considerable improvement using the MET. Setting a referral threshold at 40 may restrict access to high-functioning patients who may want to return to a preferred sporting activity.

While it is certainly important to use conventional PROMs to record health gains, the assumption that no further benefit can be achieved past the maximal score may mean that these overall health gains are under-represented. In doing so, one may unfairly restrict access to our highly effective surgical interventions for higher-functioning patients who are unable to perform their desired sporting activity. Without an additional activity metric, the considerable improvement in quality of life delivered by returning them to their preferred sporting activity may be reported as a failure, since the improvement in function captured by change in OHS may be smaller than average.

The subgroup analysis further emphasizes the point that the patients who score 48/48 are not necessarily performing at a similar level to one another. Despite identical OHS scores, patients > 60 years old had a mean MET of 10.6 METs compared to the 12.7 METs scored by the under 60s. A similar effect was noted for the male sex and those with higher preoperative MET scores. To put those scores into perspective, an activity such as Nordic walking at a fast pace scores 9.5 METs. 18 A fast run at 9 mph scores 12.8 METs, 18 so a difference of 2 to 3 METs translates into the difference between patients performing a fast walk or a fast run. Clinically this would likely be a noticeable benefit. Other studies have shown the effect of age, sex, and preoperative activity levels on postoperative physical activity. Williams et al, 24 in a study of 736 primary joint arthroplasty operations, found male sex, younger age, preoperative UCLA scores, and lower BMI to be overall predictors for achieving higher postoperative activity levels. The authors report that males are nearly five times more likely to achieve a UCLA activity score > 7 post-hip arthroplasty when compared to females (odds ratio 4.84, 95% confidence interval 2.93 to 7.99). 24 These findings have been corroborated by a number of other studies, concurring with the findings of the present study. 25 , 32 , 33

There are a number of limitations to this study. First, a large proportion of patients (61%) did not have preoperative or one-year postoperative scores, and the included patients were younger than those with missing data. It is possible that this younger cohort who completed the online questionnaire were more physically active and motivated than those who did not respond. Furthermore, our studied population was considerably younger than the national average for hip arthroplasty. While the authors believe this young population to be ideal for investigating the MET, it is worth noting that our findings may not be generalizable to the wider population of hip arthroplasty patients. Second, the MET does not factor in frequency of the activity, only intensity, so it cannot be used as a metric of fitness. Additionally, a high MET value may not correlate with impact on the hip joint, nor on the number of hip cycles. For instance, canoeing with vigorous effort scores a MET of 12.5. 18 This scores similarly to running at 9 mph (12.8 METs), 18 however running has greater impact on the hip joint and may not be attempted following hip arthroplasty in an effort to protect the longevity of the implant. Although our score did not take this into account, patients were asked to pick activities that were of importance to them and that their joint trouble affected, thus directing them to choose activities specific to the hip. Finally, as data in this study were retrospectively analyzed, there remains a risk of selection bias.

In conclusion, this study demonstrates that a simple, patient-centred activity metric (MET) can pick up important health gains in return to higher-level sporting activity, which are missed by the OHS in a younger, active population. The MET showed evidence of construct validity, good responsiveness to change, and no postoperative ceiling effect, with health gains not limited by preoperative OHS. A patient-centred physical activity metric may have a useful role in addition to conventional function-based PROMs scores where the functional outcome of hip arthroplasty is relevant.

Author contributions

T. C. Edwards: Conceptualization, Data curation, Formal analysis, Methodology, Writing – original draft.

B. Guest: Data curation, Formal analysis, Writing – review & editing.

A. Garner: Data curation, Formal analysis, Writing – review & editing.

K. Logishetty: Data curation, Formal analysis, Writing – review & editing.

A. D. Liddle: Conceptualization, Methodology, Formal analysis, Supervision, Writing – review & editing.

J. P. Cobb: Conceptualization, Methodology, Formal analysis, Supervision, Writing – review & editing.

Funding statement

The authors disclose receipt of the following financial or material support for the research, authorship, and/or publication of this article: an institutional research support grant from the Sir Michael Uren Foundation (as reported by J. P. Cobb). Infrastructure support was provided by the National Institute for Health Research (NIHR) Imperial Biomedical Research Centre (BRC).

Acknowledgements

The authors would like to acknowledge the support of the editors and reviewers of this manuscript, for their invaluable contributions.

Ethical review statement

Ethical approval was granted for data collected and used as part of this study (REC Reference: 14/SC/1243, IRAS ID: 136430).

Open access funding

The authors report that they received open access funding for their manuscript from the Imperial College London open access fund.

Follow T. C. Edwards @edwards_tomc

Follow A. Garner @DrAmyGarner

Follow K. Logishetty @klogishetty

Follow J. P. Cobb @orthorobodoc

© 2022 Author(s) et al. This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives (CC BY-NC-ND 4.0) licence, which permits the copying and redistribution of the work only, and provided the original author and source are credited. See https://creativecommons.org/licenses/by-nc-nd/4.0/

share this!

November 27, 2023

This article has been reviewed according to Science X's editorial process and policies . Editors have highlighted the following attributes while ensuring the content's credibility:

fact-checked

peer-reviewed publication

trusted source

written by researcher(s)

Stones inside fish ears mark time like tree rings. How they're helping us learn about climate change

by Clive Trueman, The Conversation

Warming waters

Looking ahead.

Journal information: Nature Communications

Provided by The Conversation

Explore further

Feedback to editors

Extreme rainfall increases exponentially with global warming: Study

10 minutes ago

A method for the early prediction of El Niño events with high hazard potential

15 minutes ago

May the 'Star Wars' vocabulary be with us

35 minutes ago

Algorithmic recommendation technology or human curation? Study of online news outlet suggests both

Alien haze, cooked in a lab, clears view to distant water worlds

Silk lines help pirate spiders trick, capture eight-legged prey

New study analyzes how people choose friendships at school

Researchers develop photoactivatable nanomedicine for the treatment of age-related macular degeneration

Pocket-sized DNA sequencers track malaria drug resistance in Ghana in near real-time

New fluorescence-based methods for fast and accessible light intensity measurements

Relevant physicsforums posts, weighing scale that can measure body fat via impedances.

20 hours ago

Are cones in eyes sized to the light's wavelengths they perceive?

Nov 24, 2023

Can a COVID vaccine cause a cancer (e.g., lymphoma) to be worse?

Remove cancer with centrifugal forces, some interesting info on vaccinations -- who guidelines on the influenza vaccine.

Nov 23, 2023

Sars-Cov2, RSV declining; Human Metapneumovirus increased

More from Biology and Medical

Related Stories

Farming tuna on land heralded as a win for sustainability, but there are serious concerns about animal welfare

Nov 1, 2023

Bluefin tuna tagged for the first time in UK waters with acoustic 'residency' tags

Jan 18, 2022

Bluefin tuna are back around the UK and a new study explains why

Jan 7, 2019

Evidence bolsters classification of a major spawning ground for Atlantic bluefin tuna off the northeast U.S.

Mar 4, 2022

Bluefin tuna reveal global ocean patterns of mercury pollution

Sep 13, 2021

How short-term heat waves impact apex marine predators

Sep 6, 2023

Recommended for you

The way a sperm tail moves can be explained by mathematics worked out by Alan Turing

Nov 26, 2023

Black dots on white fronts of African penguins may help them tell one another apart

Poisonous invasive plant exhibits twice as many genes as expected

Plastic waste in the water might be stopping, or interrupting, some shrimp-like creatures from reproducing

New human gene cluster sequence discovered

Let us know if there is a problem with our content.

Use this form if you have come across a typo, inaccuracy or would like to send an edit request for the content on this page. For general inquiries, please use our contact form . For general feedback, use the public comments section below (please adhere to guidelines ).

Please select the most appropriate category to facilitate processing of your request

Thank you for taking time to provide your feedback to the editors.

Your feedback is important to us. However, we do not guarantee individual replies due to the high volume of messages.

E-mail the story

Your email address is used only to let the recipient know who sent the email. Neither your address nor the recipient's address will be used for any other purpose. The information you enter will appear in your e-mail message and is not retained by Phys.org in any form.

Newsletter sign up

Get weekly and/or daily updates delivered to your inbox. You can unsubscribe at any time and we'll never share your details to third parties.

More information Privacy policy

Donate and enjoy an ad-free experience

We keep our content available to everyone. Consider supporting Science X's mission by getting a premium account.

E-mail newsletter

• Newsletters

Google DeepMind wants to define what counts as artificial general intelligence

AGI is one of the most disputed concepts in tech. These researchers want to fix that.

• Will Douglas Heaven archive page

AGI, or artificial general intelligence , is one of the hottest topics in tech today. It’s also one of the most controversial. A big part of the problem is that few people agree on what the term even means. Now a team of Google DeepMind researchers has put out a paper that cuts through the cross talk with not just one new definition for AGI but a whole taxonomy of them .

In broad terms, AGI typically means artificial intelligence that matches (or outmatches) humans on a range of tasks. But specifics about what counts as human-like, what tasks, and how many all tend to get waved away: AGI is AI, but better.

To come up with the new definition, the Google DeepMind team started with prominent existing definitions of AGI and drew out what they believe to be their essential common features. 

The team also outlines five ascending levels of AGI: emerging (which in their view includes cutting-edge chatbots like ChatGPT and Bard ), competent, expert, virtuoso, and superhuman (performing a wide range of tasks better than all humans, including tasks humans cannot do at all, such as decoding other people’s thoughts, predicting future events, and talking to animals). They note that no level beyond emerging AGI has been achieved.

“This provides some much-needed clarity on the topic,” says Julian Togelius, an AI researcher at New York University, who was not involved in the work. “Too many people sling around the term AGI without having thought much about what they mean.”

The researchers posted their paper online last week with zero fanfare. In an exclusive conversation with two team members—Shane Legg, one of DeepMind’s co-founders, now billed as the company’s chief AGI scientist, and Meredith Ringel Morris, Google DeepMind’s principal scientist for human and AI interaction—I got the lowdown on why they came up with these definitions and what they wanted to achieve.

A sharper definition

“I see so many discussions where people seem to be using the term to mean different things, and that leads to all sorts of confusion,” says Legg, who came up with the term in the first place around 20 years ago. “Now that AGI is becoming such an important topic—you know, even the UK prime minister is talking about it—we need to sharpen up what we mean.”

It wasn’t always this way. Talk of AGI was once derided in serious conversation as vague at best and magical thinking at worst. But buoyed by the hype around generative models, buzz about AGI is now everywhere.

When Legg suggested the term to his former colleague and fellow researcher Ben Goertzel for the title of Goertzel’s 2007 book about future developments in AI , the hand-waviness was kind of the point. “I didn’t have an especially clear definition. I didn’t really feel it was necessary,” says Legg. “I was actually thinking of it more as a field of study, rather than an artifact.”

His aim at the time was to distinguish existing AI that could do one task very well, like IBM’s chess-playing program Deep Blue, from hypothetical AI that he and many others imagined would one day do many tasks very well. Human intelligence is not like Deep Blue, says Legg: “It is a very broad thing.”

But over the years, people started to think of AGI as a potential property that actual computer programs might have. Today it’s normal for top AI companies like Google DeepMind and OpenAI to make bold public statements about their mission to build such programs.

“If you start having those conversations, you need to be a lot more specific about what you mean,” says Legg.

For example, the DeepMind researchers state that an AGI must be both general-purpose and high-achieving, not just one or the other. “Separating breadth and depth in this way is very useful,” says Togelius. “It shows why the very accomplished AI systems we’ve seen in the past don’t qualify as AGI.”

They also state that an AGI must not only be able to do a range of tasks, it must also be able to learn how to do those tasks, assess its performance, and ask for assistance when needed. And they state that what an AGI can do matters more than how it does it.  

It’s not that the way an AGI works doesn’t matter, says Morris. The problem is that we don’t know enough yet about the way cutting-edge models, such as large language models, work under the hood to make this a focus of the definition.

“As we gain more insights into these underlying processes, it may be important to revisit our definition of AGI,” says Morris. “We need to focus on what we can measure today in a scientifically agreed-upon way.”

Measuring up

Measuring the performance of today’s models is already controversial , with researchers debating what it really means for a large language model to pass dozens of high school tests and more. Is it a sign of intelligence? Or a kind of rote learning?

Assessing the performance of future models that are even more capable will be more difficult still. The researchers suggest that if AGI is ever developed, its capabilities should be evaluated on an ongoing basis, rather than through a handful of one-off tests.

The team also points out that AGI does not imply autonomy. “There’s often an implicit assumption that people would want a system to operate completely autonomously,” says Morris. But that’s not always the case. In theory, it’s possible to build super-smart machines that are fully controlled by humans.

One question the researchers don’t address in their discussion of what AGI is, is why we should build it. Some computer scientists, such as Timnit Gebru , founder of the Distributed AI Research Institute, have argued that the whole endeavor is weird. In a talk in April on what she sees as the false (even dangerous) promise of utopia through AGI , Gebru noted that the hypothetical technology “sounds like an unscoped system with the apparent goal of trying to do everything for everyone under any environment.” 

Most engineering projects have well-scoped goals. The mission to build AGI does not. Even Google DeepMind’s definitions allow for AGI that is indefinitely broad and indefinitely smart. “Don’t attempt to build a god,” Gebru said.

Artificial intelligence

This new data poisoning tool lets artists fight back against generative ai.

The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models. 

• Melissa Heikkilä archive page

Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist

An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.

Driving companywide efficiencies with AI

Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.

• MIT Technology Review Insights archive page

Generative AI deployment: Strategies for smooth scaling

Our global poll examines key decision points for putting AI to use in the enterprise.

Stay connected

Get the latest updates from mit technology review.

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at [email protected] with a list of newsletters you’d like to receive.

• Information for advertisers
• Information for librarians
• Carbon balance
• Reprints and permissions
• Core trainee prize
• Core trainee prize winners
• ABC Fellows

Find your institution

Your academic institution may have purchased access to content. Find your institution below and we'll take you to your institution's login page

Account login

Login here to access an Administrator account for an institution, a personal subscription or to manage your alerts.

Registration successful

Receive monthly Table of Contents alerts from Bone & Joint Research

Comprehensive article alerts can be set up and managed through your account settings

Visit Bone & Joint Research at:

Bone & Joint Research

• Journal overview
• Editorial board
• For authors
• For advertisers
• Publication ethics

The metabolic equivalent of task score

A useful metric for comparing high-functioning hip arthroplasty patients.

• Thomas C. Edwards
• Brogan Guest
• Kartik Logishetty
• Alexander D. Liddle
• Justin P. Cobb

This study investigates the use of the metabolic equivalent of task (MET) score in a young hip arthroplasty population, and its ability to capture additional benefit beyond the ceiling effect of conventional patient-reported outcome measures.

From our electronic database of 751 hip arthroplasty procedures, 221 patients were included. Patients were excluded if they had revision surgery, an alternative hip procedure, or incomplete data either preoperatively or at one-year follow-up. Included patients had a mean age of 59.4 years (SD 11.3) and 54.3% were male, incorporating 117 primary total hip and 104 hip resurfacing arthroplasty operations. Oxford Hip Score (OHS), EuroQol five-dimension questionnaire (EQ-5D), and the MET were recorded preoperatively and at one-year follow-up. The distribution was examined reporting the presence of ceiling and floor effects. Validity was assessed correlating the MET with the other scores using Spearman’s rank correlation coefficient and determining responsiveness. A subgroup of 93 patients scoring 48/48 on the OHS were analyzed by age, sex, BMI, and preoperative MET using the other metrics to determine if differences could be established despite scoring identically on the OHS.

Postoperatively the OHS and EQ-5D demonstrate considerable negatively skewed distributions with ceiling effects of 41.6% and 53.8%, respectively. The MET was normally distributed postoperatively with no relevant ceiling effect. Weak-to-moderate significant correlations were found between the MET and the other two metrics. In the 48/48 subgroup, no differences were found comparing groups with the EQ-5D, however significantly higher mean MET scores were demonstrated for patients aged < 60 years (12.7 (SD 4.7) vs 10.6 (SD 2.4), p = 0.008), male patients (12.5 (SD 4.5) vs 10.8 (SD 2.8), p = 0.024), and those with preoperative MET scores > 6 (12.6 (SD 4.2) vs 11.0 (SD 3.3), p = 0.040).

The MET is normally distributed in patients following hip arthroplasty, recording levels of activity which are undetectable using the OHS.

Cite this article: Bone Joint Res  2022;11(5):317–326.

Article focus

Substantial ceiling effects of some conventional patient-reported outcome measure (PROM) scores limit their ability to discriminate between high-functioning postoperative hip arthroplasty patients.

This article introduces the concept of using metabolic equivalent of task (MET) values in a young hip arthroplasty cohort to determine whether this approach can capture additional benefit beyond the ceiling effect of conventional PROMs.

Key messages

Postoperative ceiling effects of the Oxford Hip Score (OHS) and EuroQol five-dimension questionnaire (EQ-5D) miss clinically substantial gains in higher-level health-related quality of life.

MET is a simple, versatile measure of physical activity, with no postoperative ceiling effect.

Clinicians, researchers, and health economists who wish to capture the full benefit from hip arthroplasty surgery should consider using MET in addition to conventional PROMs.

Strengths and limitations

This multicentre, multiple-surgeon study addresses an important concept, which will inform future hip arthroplasty researchers designing studies assessing new technologies or techniques.

The large proportion of non-responders and included cohort being substantially younger than the national average for hip arthroplasty limits its generalizability to the wider population.

Introduction

Primary hip arthroplasty is an effective intervention for improving pain and restoring function. 1 Patient-reported outcome measures (PROMs) such as the Oxford Hip Score (OHS), 2 , 3 which is routinely collected before and after hip arthroplasty in the UK, reliably and predictably report considerable, cost-effective improvements in pain and function. 4 , 5 However, as a consequence of the efficacy of hip arthroplasty, the distribution of postoperative scores is highly negatively skewed. In two national registries, the modal score on the postoperative OHS was 100% (48/48 points) with up to 20% of patients recording this score. 6 , 7

The population of patients presenting for hip arthroplasty has evolved and their expectations are different from those undergoing surgery when the OHS was introduced. 8 In a study by Scott et al, 9 40% of hip arthroplasty patients considered returning to sporting activity a ‘very important’ preoperative expectation, but such activities are not captured by the OHS. By relying solely on this skewed metric, potential health gain in return to sporting activity from innovative techniques will remain undetected, while other patient groups may be inappropriately told that their clinical results are as good as they can get, despite dissatisfaction with the level of activity that they have achieved. 7 , 10 , 11 In the UK, the pre- to postoperative change in OHS is used by government-backed initiatives including ‘getting it right first time’ (GIRFT) and the NHS best practice tariff (BPT), 12 , 13 to measure success at the institutional level. With the skewed OHS as the metric, the only way to improve the health gain obtained is by refusing care until the preoperative scores are low enough. This approach may have a role in healthcare rationing but should not impede the scientific desire to measure higher-level function. As degree of health gain is so closely related to preoperative score, and as preoperative score may vary, there is a need for a metric which measures outcome equally independently of preoperative health state.

Alternative scores have been developed with the aim of being more discriminative in high-functioning patients. Several constructs have been suggested, such as joint perception as measured by the Forgotten Joint Score (FJS) and physical activity scores. The FJS assesses patients’ awareness of their joint arthroplasty performing different tasks, with the optimal outcome being a ‘forgotten’ artificial joint. In a study comparing the outcomes of robotic and manual THA, the authors found no clinically relevant difference using the OHS, however the robotic group did substantially better using the FJS. 14 The authors support the idea that the ceiling effect of the OHS limits its use for comparing high-functioning postoperative patients, with their results indicating that the FJS may be more discriminative. While this sounds encouraging, other authors have reported ceiling effects of 20% to 30% using the FJS in a postoperative hip arthroplasty cohort, and so problems may still exist with skewed distributions using this score. 15 , 16

Physical activity metrics may be another solution, with a number of valid and reliable metrics such as the University of California, Los Angeles (UCLA) activity scale available. 17 This score appears to have no ceiling effect and is simple to use, however it only includes a small number of activities and does not account for the individual activity intensity. 17 One potential solution to this problem is the use of metabolic equivalent of task (MET) values, which numerically quantify the energy expenditure of over 800 activities comparing them to energy expenditure at rest. 18 This sophisticated, personalized approach to quantifying activity energy expenditure has been validated as a surrogate for general cardiovascular fitness, correlating well with both objective activity measures, such as pedometers, as well as the development of cardiovascular disease and mortality. 19 , 20 Exercise at an intensity that raises the heart rate is now well established as being an effective health maintenance intervention. MET values have been used to confirm this beneficial effect: in a twin study, ‘conditioning exercise’ offers substantial protection against risk of death when compared with sedentary or occasional exercising. 20 Although not yet commonly used following arthroplasty surgery, with a simplification to measure activity intensity without the performance time or frequency, the MET may be a robust way of comparing activity in postoperative hip arthroplasty patients, demonstrating activity levels that have real relevance to health and life expectancy.

This study therefore aims to answer two important questions: 1) does the MET have a postoperative ceiling effect that may limit its ability to discriminate between high-performing postoperative patients; and 2) can the MET demonstrate continued improvement and health gains beyond the maximal OHS, establishing differences between postoperative patients who score 48/48?

Study design

This study was a retrospective analysis of anonymous data, collected prospectively from consenting primary hip arthroplasty patients as part of an ongoing, longitudinal study of gait analysis in lower limb arthroplasty (REC reference: 14/SC/1243). Patients from this study were eligible for inclusion if they underwent a primary hip arthroplasty under one of 13 surgeons at 12 sites between 2014 and 2018. Patients were excluded if they had revision surgery, an alternative hip procedure, or incomplete data either preoperatively or at one-year follow-up. Demographic data and the patient-reported answers to three PROMs questionnaires (EuroQol five-dimension questionnaire (EQ-5D), OHS, and the MET) were recorded preoperatively and at one year postoperatively.

Demographic details

A total of 751 patients were initially identified on our electronic database; 73 were excluded having had an alternative or revision procedure, and a further 457 patients were excluded due to lack of preoperative or one-year PROM scores. Overall 221 patients, including 117 THAs (53%) and 104 HRAs (47%), with a mean age of 59 years (SD 11), were analyzed in this study. Demographic data are detailed in Table I . The 221 responding patients with full datasets were a mean four years younger than the 457 non-responders (59 years (SD 11) vs 63 years (SD 12), p < 0.001, independent-samples t -test).

Demographics and patient-reported outcome measures.

Missing data, n = 163.

Non-parametric variable, data presented as median (interquartile range), otherwise data reported as mean (standard deviation, range) as indicated.

EQ-5D, EuroQol five-dimension questionnaire; HRA, hip resurfacing arthroplasty; IQR, interquartile range; MET, metabolic equivalent of task; OHS, Oxford Hip Score; SD, standard deviation; THA, total hip arthroplasty.

Using a similar methodology to Amstutz and Le Duff, 21 the MET asks patients to choose three physical activities that are important to them, and that are affected by their joint problem. These initially selected activities remain the same at all follow-up timepoints. Patients then rate the intensity at which they currently perform the activity on a visual scale from 0 to 100. METs are numerical values assigned to demonstrate the energy expenditure used performing different tasks. One MET is equivalent to energy expenditure during rest and is approximately equal to 3.5 ml O 2 kg -1 min -1 in adults. 19 Using Arizona State Universities compendium of activities, 18 the MET values associated with each activity are recorded. An example is running, which has a range of values between 4.5 (jogging on a mini-tramp) and 23 METs (running a 4.3 minute mile). Based on this reference range, the patient’s self-reported intensity score is then used to work out a value for the METs they are currently doing. This is done by subtracting the lower value of the MET reference range from the higher value, multiplying this by the percentage intensity expressed as a decimal, and then adding back on the lower reference value. Using the above as an example, if a patient rated their intensity as 50% in running (range 4.5 to 23 METs) their MET score would be worked out as ((23 to 4.5)*0.5) + 4.5 = 13.75 METs. The MET is the maximum value scored from the three chosen activities. In the full MET, the frequency and duration of physical activity are recorded; we omitted these aspects from the score in favour of intensity, to avoid measuring cardiorespiratory fitness and also to avoid under-representing performance through measuring high-intensity but infrequently performed activities, such as skiing. 21 , 22

Distribution of scores

Data were analyzed to demonstrate the distribution, presence of ceiling or floor effects, concurrent validity of the MET in terms of its responsiveness, and correlations between the MET and the two conventional PROMs. Other authors have suggested when validating physical activity metrics that a weak-to-moderate correlation would be expected between the activity metric and conventional PROMs. 8

Health gains

Recent literature has established that as preoperative OHS increases, the improvement in score decreases. 23 To investigate whether health gains in the MET and EQ-5D are also limited by the level of preoperative joint symptoms, the relationship between preoperative OHS and patient improvement at one year using the three metrics was plotted. Fractional polynomial regression plots were used to demonstrate the likely increase in each metric for a given preoperative OHS score.

48/48 sub-cohort analysis

A subgroup analysis was performed on a cohort of patients with the maximum postoperative OHS. Previous studies have highlighted that postoperative physical activity (as measured on the UCLA activity score) in hip arthroplasty patients may be higher in younger patients, male patients, those with higher preoperative activity levels, and those with lower BMI. 24 , 25 Based on this, the 48/48 scoring patient cohort was divided into categories (age < or > 60 years, male or female, preoperative MET < or > 6, and BMI < or > 25 kg/m 2 ) and compared using the MET and EQ-5D at one year postoperatively. Previous literature has classified activities as light (< 3), moderate (3 to 6), or vigorous (> 6) according to their MET values. 26 , 27 Therefore, in this analysis, the threshold for high preoperative activity was set at 6 METs.

Statistical analysis

Statistical analysis was performed using Stata/IC 10.1 (StataCorp, USA). Data were first tested for normality visually using histograms and normal Q-Q plots. To quantify the shape and symmetry of the distribution about the mean, kurtosis and skewness values were calculated. A standard normal distribution is generally considered to have a kurtosis value of 3 and a skew value of 0. 28 For independent data, parametric variables were compared using the independent-samples t -test, and non-parametric data were compared using the Mann-Whitney U test. Paired data were compared using the paired t -test. Ceiling and floor effects were calculated as the percentage of patients scoring the maximum or minimum scores, respectively. As previously indicated in the literature, ceiling or floor effects of > 15% were considered relevant. 7 Construct validity of the MET was assessed by examining the responsiveness of the score to change using the standardized response mean (SRM; calculated by dividing the mean change in score over the one-year time period by the standard deviation (SD) of that change), and concurrent validity was assessed by measuring correlations between scores as calculated using Spearman’s rank correlation coefficient. R s values of 0.3 to 0.5 were considered as weak-to-moderate correlations. 8 Statistical significance was set at p < 0.05.

Distribution

Preoperatively, the distribution of the OHS was normal (skew -0.14, kurtosis 2.57) and that of the EQ-5D was bimodal with a skewness value of -0.99 and a kurtosis of 4.03 ( Figure 1 and Figure 2 ). The MET demonstrates a slight positive skew of 0.59 representing a floor effect, with the commonest score being zero, on account of hip pain and near-normal kurtosis of 3.18 ( Figure 3 ).

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of Oxford Hip Scores (OHS) preoperatively and at one-year follow-up. Solid vertical lines represent mean values, dashed vertical lines represent the median.

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of EuroQol-5D (EQ-5D) index scores preoperatively and at one-year follow-up. Solid vertical lines represent mean values, dashed vertical lines represent the median.

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of metabolic equivalent of task (MET) scores preoperatively and at one-year follow-up. Solid vertical lines represent mean values, dashed vertical lines represent the median.

Postoperatively, both the OHS and EQ-5D scores demonstrate a substantial negative skew ( Figure 1 and Figure 2 ). The kurtosis value of the OHS was 13.07 with a skewness value of -3.12, while the EQ-5D demonstrated a kurtosis value of 6.23 and a skewness value of -1.96. The MET on the other hand exhibited a normal distribution postoperatively, centred around a mean of 10.7 (SD 3.8) with very little skew (0.40) and near normal kurtosis (4.46; Figure 3 and Table II ).

Distribution, ceiling and floor effects, and responsiveness of the three metrics before and after surgery.

Ceiling and floor effects calculated as the percentage of patients scoring the maximum and minimum possible scores, respectively. Skewness and kurtosis values numerically represent the distribution of scores. A normal distribution has a skew of 0 and a kurtosis of 3. Standardized response mean calculated as the mean difference in scores divided by the standard deviation of that difference.

Indicates number of included patients.

Paired t -test comparing preoperative and one-year postoperative scores.

EQ-5D, EuroQol five-dimension questionnaire; IQR, interquartile range; MET, metabolic equivalent of task; OHS, Oxford Hip Score; SD, standard deviation; SRM, standardized response mean.

Ceiling and floor effects

No floor effects were seen for the OHS or EQ-5D, but substantial ceiling effects of 41.6% and 53.8% were seen at one year follow-up in the OHS and EQ-5D, respectively. Preoperatively the MET had a moderate floor effect of 25.3%, while no relevant ceiling effect was noted postoperatively ( Table II ).

Spearman’s rank correlation coefficient was weak-to-moderate, but there were statistically significant correlations between the MET and EQ-5D both preoperatively (r s = 0.46, p < 0.001) and to a lesser extent at one-year follow-up (r s = 0.32 p < 0.001, Spearman’s rank correlation). Similarly, weak-to-moderate correlations were demonstrated between the MET and the OHS both preoperatively (r s = 0.46, p < 0.001, Spearman’s rank correlation) and at one-year follow-up (r s = 0.30, p < 0.001, Spearman’s rank correlation).

Improvement in score and responsiveness

All three metrics demonstrated excellent responsiveness with effect sizes as determined by the SRMs of > 1 ( Table III ). The fractional polynomial predictive plots in Figure 4 demonstrate a strong negative relationship between preoperative OHS and improvement in score for both EQ-5D ( Figure 4a ) and OHS ( Figure 4b ). For the MET score, this relationship is far less clear, with an initial decrease in improvement seen in patients who have lower preoperative OHS, while in patients with higher preoperative OHS the MET is progressively more responsive ( Figure 4c ).

Fractional polynomial regression plots demonstrating predicted improvement in score at one-year follow-up for a given preoperative Oxford Hip Score (OHS) for a) OHS, b) EuroQol five-dimension questionnaire (EQ-5D), and c) metabolic equivalent of task (MET). Fractional polynomial fit line with 95% confidence intervals (CIs) demonstrated in grey.

Outcome scores by category for 48/48 Oxford Hip Score sub-cohort.

Except for p-values marked with the † symbol (Mann-Whitney U test), all p-values in this table were calculated using the independent-samples t -test.

Number of patients.

Mann-Whitney U test.

EQ-5D, EuroQol five-dimension questionnaire; IQR, interquartile range; MET, metabolic equivalent of task; SD, standard deviation.

A total of 92 postoperative patients scored 48/48 on the OHS following surgery. The histograms in Figure 5 demonstrate the distribution of EQ-5D in this group, which has a strong negative skew, with the majority of patients scoring the maximal score of 1 ( Figure 5a ). The MET on the other hand exhibits a near normal distribution of scores, despite all patients scoring the same on the OHS ( Figure 5b ). When subdivided into groups, patients aged under 60 years scored significantly higher on the MET than patients over 60 years of age (mean 12.7 (SD 4.7) vs 10.6 (SD 2.4), p = 0.008, independent-samples t -test), as did male patients (mean 12.5 (SD 4.5) vs 10.8 (SD 2.8), p = 0.024, independent-samples t -test) and patients with higher activity levels on their preoperative MET scores (mean 12.6 (SD 4.2) vs 11.0 (SD 3.3), p = 0.040, independent-samples t -test) ( Figure 6a ). No significant differences were found comparing patients by BMI using the MET or comparing any of the groups using the EQ-5D ( Figure 6b , Table III ).

Histograms with kernel (Epanechnikov) density plots demonstrating distribution of a) metabolic equivalent of task score (MET) and b) EuroQol five-dimension questionnaire (EQ-5D), at one-year follow-up for the subgroup of patients who all scored 48/48 on the Oxford Hip Score (n = 92).

Column scatter for the subgroup of 92 patients scoring 48/48 on the Oxford Hip Score, compared by age group, BMI, preoperative metabolic equivalent of task (MET) score, and sex using: a) MET scores; and b) EuroQol five-dimension questionnaire (EQ-5D) scores. The solid horizontal line represents the median and the whiskers represent the interquartile range. Statistically significant p-values have been indicated.

This retrospective study set out to determine whether the MET score could capture differences in function that were not detectable by the OHS or EQ-5D in an active hip arthroplasty population. This question was answered: the MET score does deliver a symmetrical metric with a normal distribution in a postoperative population, capturing differences in activity levels that were not detectable using the OHS. This question is relevant for health economists, policy makers, and those designing clinical trials. Health benefits that can be captured simply, without the need for expensive equipment or licences, should help drive commissioning choices. By demonstrating that patients can improve past the OHS maximum score, we have revealed an opportunity that was otherwise denied: using this metric, surgeons who are currently penalized for failing to deliver adequate health gains may now be able to justify their offering of arthroplasty in younger and more active patients. By restricting health gains to the OHS, commissioners may unfairly restrict access to arthroplasty surgery, or unfairly penalize hospitals for not achieving satisfactory results if these decisions are based solely on health gains as measured by the OHS.

Although there is more work to be done in this area, the aspects of validity measured in the present study support its use as a metric for the outcome of hip arthroplasty surgery. The MET demonstrated evidence of concurrent validity with weak-to-moderate correlations found with both the OHS and EQ-5D. Naal et al 17 used a similar approach, establishing weak-to-moderate correlations with three different physical activity scores and the OHS. One potential limitation was that the present study did not validate the MET against another validated physical activity metric or objective physical activity measures such as a pedometer or exercise log. However, the authors note that the face validity of using MET values has already been well established by other similar MET-based scores. 18 , 19 Although not validated specifically for use in arthroplasty, the International Physical Activity Questionnaire (IPAQ) score is a MET-based score, shown to be valid and reliable for use in the general population measuring activity levels. 19 It differs from the MET, being a better measure of cardiorespiratory fitness, whereas the MET is personalized to patients’ sporting aspirations. The major advantage of the MET is that no matter what activities patients choose, the scores are comparable and relevant to their joint disease. Furthermore, the numeric MET values assigned by the University of Arizona are objective, being based upon oxygen consumption. 18 , 19 Therefore, the authors considered concurrent validity with other hip-specific and generic PROMs, alongside responsiveness, encouraging validation data for using the score in this cohort, however further work in this area would be an interesting avenue for future research.

Responsiveness is considered another aspect of construct validity. 29 The greater the responsiveness, the more accurate a metric is in detecting change when it has occurred. The MET had a SRM of 1.17, which indicates a large effect size or an excellent response to change over time. 8 The calculated SRMs for EQ-5D and OHS in this cohort were found to be similar to previously published literature, further validating our findings. 30

Unlike the OHS and EQ-5D, the postoperative MET had a normal distribution and exhibited no ceiling effect. Substantial postoperative ceiling effects were found for the OHS (41.6%) and EQ-5D (53.8%). In general, ceiling effects or floor effects are considered problematic when 15% or more of the cohort score the best or worst scores. 7 , 10 By having large numbers of patients scoring the best or worst scores, the metric is rendered insensitive to detecting differences at the extremes of the scale. 7 , 10 Other studies have demonstrated strong ceiling effects in the OHS of 19.9%, 6 and even more pronounced ceiling effects for the EQ-5D of 39.8%. 30 While the pattern of these findings support our results, our population demonstrated a much higher percentage ceiling effect for both metrics. This may be related to the studied population which included a younger, more active cohort than that used in other studies. While other scores have been developed with the aim of reducing the impact of ceiling effect, unfortunately problematic ceiling effects may still exist. In a recent study, the FJS was reported to have a ceiling effect of 31.9%, similar to those reported for more conventional PROMs. 15 In addition, the FJS has reported a substantial floor effect of 22.4%, suggesting that there may be problems discriminating at both ends of the score. 31

While the MET showed no postoperative ceiling effect, it did show a preoperative floor effect, similarly to the FJS. This is not surprising given that the formulation of the question specifies the selection of tasks that have been negatively affected by the respondent’s hip pain. A similar preoperative floor effect has been observed in validation studies looking at other physical activity-based outcome measures such as the Tegner score. 17 When using MET solely as an assessment of postoperative outcome rather than of preoperative disease state, this floor effect is unimportant. If it were to be used for the former, the question may have to be re-formulated.

Both the OHS and EQ-5D demonstrated very little predicted improvement towards the upper end of the preoperative OHS scale. The MET on the other hand shows continued predicted improvements, with a 6 MET improvement predicted for patients who score 48/48 on the OHS. A large registry study by Price et al 23 demonstrated a similar effect using the OHS, with the likelihood of seeing a meaningful clinical improvement decreasing with higher preoperative scores. The authors conclude that at a preoperative score of 40 or above, there was a 0% chance of meaningful improvement, suggesting this as a threshold for referral. 23 The present study suggests that even though these higher-scoring preoperative patients do not show improvement using the OHS, they do show considerable improvement using the MET. Setting a referral threshold at 40 may restrict access to high-functioning patients who may want to return to a preferred sporting activity.

While it is certainly important to use conventional PROMs to record health gains, the assumption that no further benefit can be achieved past the maximal score may mean that these overall health gains are under-represented. In doing so, one may unfairly restrict access to our highly effective surgical interventions for higher-functioning patients who are unable to perform their desired sporting activity. Without an additional activity metric, the considerable improvement in quality of life delivered by returning them to their preferred sporting activity may be reported as a failure, since the improvement in function captured by change in OHS may be smaller than average.

The subgroup analysis further emphasizes the point that the patients who score 48/48 are not necessarily performing at a similar level to one another. Despite identical OHS scores, patients > 60 years old had a mean MET of 10.6 METs compared to the 12.7 METs scored by the under 60s. A similar effect was noted for the male sex and those with higher preoperative MET scores. To put those scores into perspective, an activity such as Nordic walking at a fast pace scores 9.5 METs. 18 A fast run at 9 mph scores 12.8 METs, 18 so a difference of 2 to 3 METs translates into the difference between patients performing a fast walk or a fast run. Clinically this would likely be a noticeable benefit. Other studies have shown the effect of age, sex, and preoperative activity levels on postoperative physical activity. Williams et al, 24 in a study of 736 primary joint arthroplasty operations, found male sex, younger age, preoperative UCLA scores, and lower BMI to be overall predictors for achieving higher postoperative activity levels. The authors report that males are nearly five times more likely to achieve a UCLA activity score > 7 post-hip arthroplasty when compared to females (odds ratio 4.84, 95% confidence interval 2.93 to 7.99). 24 These findings have been corroborated by a number of other studies, concurring with the findings of the present study. 25 , 32 , 33

There are a number of limitations to this study. First, a large proportion of patients (61%) did not have preoperative or one-year postoperative scores, and the included patients were younger than those with missing data. It is possible that this younger cohort who completed the online questionnaire were more physically active and motivated than those who did not respond. Furthermore, our studied population was considerably younger than the national average for hip arthroplasty. While the authors believe this young population to be ideal for investigating the MET, it is worth noting that our findings may not be generalizable to the wider population of hip arthroplasty patients. Second, the MET does not factor in frequency of the activity, only intensity, so it cannot be used as a metric of fitness. Additionally, a high MET value may not correlate with impact on the hip joint, nor on the number of hip cycles. For instance, canoeing with vigorous effort scores a MET of 12.5. 18 This scores similarly to running at 9 mph (12.8 METs), 18 however running has greater impact on the hip joint and may not be attempted following hip arthroplasty in an effort to protect the longevity of the implant. Although our score did not take this into account, patients were asked to pick activities that were of importance to them and that their joint trouble affected, thus directing them to choose activities specific to the hip. Finally, as data in this study were retrospectively analyzed, there remains a risk of selection bias.

In conclusion, this study demonstrates that a simple, patient-centred activity metric (MET) can pick up important health gains in return to higher-level sporting activity, which are missed by the OHS in a younger, active population. The MET showed evidence of construct validity, good responsiveness to change, and no postoperative ceiling effect, with health gains not limited by preoperative OHS. A patient-centred physical activity metric may have a useful role in addition to conventional function-based PROMs scores where the functional outcome of hip arthroplasty is relevant.

1. Mei XY , Gong YJ , Safir O , Gross A , Kuzyk P . Long-term outcomes of total hip arthroplasty in patients younger than 55 years: a systematic review of the contemporary literature . Can J Surg . 2019 ; 62 ( 4 ): 249 – 258 . Crossref Google Scholar

2. Dawson J , Fitzpatrick R , Carr A , Murray D . Questionnaire on the perceptions of patients about total hip replacement . J Bone Joint Surg Br . 1996 ; 78-B ( 2 ): 185 – 190 . Google Scholar

3. Murray DW , Fitzpatrick R , Rogers K , et al. The use of the Oxford hip and knee scores . J Bone Joint Surg Br . 2007 ; 89-B ( 8 ): 1010 – 1014 . Crossref Google Scholar

4. Harris K , Dawson J , Gibbons E , et al. Systematic review of measurement properties of patient-reported outcome measures used in patients undergoing hip and knee arthroplasty . Patient Relat Outcome Meas . 2016 ; 7 : 101 – 108 . Crossref Google Scholar

5. Dakin H , Eibich P , Beard D , Gray A , Price A . The use of patient-reported outcome measures to guide referral for hip and knee arthroplasty . Bone Joint J . 2020 ; 102-B ( 7 ): 950 – 958 . Crossref Google Scholar

6. Paulsen A , Odgaard A , Overgaard S . Translation, cross-cultural adaptation and validation of the Danish version of the Oxford hip score: Assessed against generic and disease-specific questionnaires . Bone Joint Res . 2012 ; 1 ( 9 ): 225 – 233 . Crossref Google Scholar

7. Lim CR , Harris K , Dawson J , Beard DJ , Fitzpatrick R , Price AJ . Floor and ceiling effects in the OHS: an analysis of the NHS PROMs data set . BMJ Open . 2015 ; 5 ( 7 ): e007765 . Crossref Google Scholar

8. Dawson J , Beard DJ , McKibbin H , Harris K , Jenkinson C , Price AJ . Development of a patient-reported outcome measure of activity and participation (the OKS-APQ) to supplement the Oxford knee score . Bone Joint J . 2014 ; 96-B ( 3 ): 332 – 338 . Crossref Google Scholar

9. Scott CEH , Bugler KE , Clement ND , MacDonald D , Howie CR , Biant LC . Patient expectations of arthroplasty of the hip and knee . J Bone Joint Surg Br . 2012 ; 94-B ( 7 ): 974 – 981 . Crossref Google Scholar

10. Gulledge CM , Lizzio VA , Smith DG , Guo E , Makhni EC . What are the floor and ceiling effects of patient-reported outcomes measurement information system computer adaptive test domains in orthopaedic patients? A systematic review . Arthroscopy . 2020 ; 36 ( 3 ): 901 – 912 . Crossref Google Scholar

11. Edwards TC , Logishetty K , Cobb JP . Letter to the Editor on “Patient-reported outcomes following total hip arthroplasty: a multicenter comparison based on surgical approaches.” J Arthroplasty . 2020 ; 35 ( 9 ): 2686 – 2687 . Crossref Google Scholar

12. Briggs T . A national review of adult elective orthopaedic services in England: Getting It Right First Time . British Orthopaedic Association . 2015 . https://gettingitrightfirsttime.co.uk/wp-content/uploads/2018/07/GIRFT-National-Report-Mar15-Web.pdf ( date last accessed 5 April 2022 ). Google Scholar

13. No authors listed . 2022/23 National Tariff Payment System – a consultation notice . 2022 . https://www.england.nhs.uk/wp-content/uploads/2021/12/22-23-National-Tariff-Payment-System-a-consultation-notice-v2.pdf ( date last accessed 5 May 2022 ). Google Scholar

14. Clement ND , Gaston P , Bell A , et al. Robotic arm-assisted versus manual total hip arthroplasty . Bone Joint Res . 2021 ; 10 ( 1 ): 22 – 30 . Crossref Google Scholar

15. Puliero B , Blakeney WG , Beaulieu Y , Vendittoli PA . Joint perception after total hip arthroplasty and the forgotten joint . J Arthroplasty . 2019 ; 34 ( 1 ): 65 – 70 . Crossref Google Scholar

16. Larsson A , Rolfson O , Kärrholm J . Evaluation of Forgotten Joint Score in total hip arthroplasty with Oxford Hip Score as reference standard . Acta Orthop . 2019 ; 90 ( 3 ): 253 – 257 . Crossref Google Scholar

17. Naal FD , Impellizzeri FM , Leunig M . Which is the best activity rating scale for patients undergoing total joint arthroplasty? Clin Orthop Relat Res . 2009 ; 467 ( 4 ): 958 – 965 . Crossref Google Scholar

18. Ainsworth BE , Haskell WL , Herrmann SD , et al. 2011 Compendium of Physical Activities: a second update of codes and MET values . Med Sci Sports Exerc . 2011 ; 43 ( 8 ): 1575 – 1581 . Crossref Google Scholar

19. Hagströmer M , Oja P , Sjöström M . The International Physical Activity Questionnaire (IPAQ): a study of concurrent and construct validity . Public Health Nutr . 2006 ; 9 ( 6 ): 755 – 762 . Crossref Google Scholar

20. Kujala UM , Kaprio J , Sarna S , Koskenvuo M . Relationship of leisure-time physical activity and mortality: the Finnish twin cohort . JAMA . 1998 ; 279 ( 6 ): 440 – 444 . Crossref Google Scholar

21. Amstutz HC , Le Duff MJ . Effects of physical activity on long-term survivorship after metal-on-metal hip resurfacing arthroplasty: is it safe to return to sports? Bone Joint J . 2019 ; 101-B ( 10 ): 1186 – 1191 . Crossref Google Scholar

22. Le Duff MJ , Amstutz HC . Sporting activity after hip resurfacing: changes over time . Orthop Clin North Am . 2011 ; 42 ( 2 ): 161 – 167 . Crossref Google Scholar

23. Price AJ , Kang S , Cook JA , et al. The use of patient-reported outcome measures to guide referral for hip and knee arthroplasty . Bone Joint J . 2020 ; 102-B ( 7 ): 941 – 949 . Crossref Google Scholar

24. Williams DH , Greidanus NV , Masri BA , Duncan CP , Garbuz DS . Predictors of participation in sports after hip and knee arthroplasty . Clin Orthop Relat Res . 2012 ; 470 ( 2 ): 555 – 561 . Crossref Google Scholar

25. Lübbeke A , Zimmermann-Sloutskis D , Stern R , et al. Physical activity before and after primary total hip arthroplasty: a registry-based study . Arthritis Care Res (Hoboken) . 2014 ; 66 ( 2 ): 277 – 284 . Crossref Google Scholar

26. Vähä-Ypyä H , Vasankari T , Husu P , et al. Validation of cut-points for evaluating the intensity of physical activity with accelerometry-based mean amplitude deviation (MAD . PLoS One . 2015 ; 10 ( 8 ): e0134813 . Crossref Google Scholar

27. Jetté M , Sidney K , Blümchen G . Metabolic equivalents (METS) in exercise testing, exercise prescription, and evaluation of functional capacity . Clin Cardiol . 1990 ; 13 ( 8 ): 555 – 565 . Crossref Google Scholar

28. Hoffman JIE . Chapter 6 - Normal Distribution In Hoffman JIE . ed . Biostatistics for Medical and Biomedical Practitioners . Cambridge, Massachusetts : Academic Press , 2015 : 101 – 119 . Google Scholar

29. Revicki D , Hays RD , Cella D , Sloan J . Recommended methods for determining responsiveness and minimally important differences for patient-reported outcomes . J Clin Epidemiol . 2008 ; 61 ( 2 ): 102 – 109 . Crossref Google Scholar

30. Ostendorf M , van Stel HF , Buskens E , et al. Patient-reported outcome in total hip replacement. A comparison of five instruments of health status . J Bone Joint Surg Br . 2004 ; 86-B ( 6 ): 801 – 808 . Crossref Google Scholar

31. Hamilton DF , Loth FL , Giesinger JM , et al. Validation of the English language Forgotten Joint Score-12 as an outcome measure for total hip and knee arthroplasty in a British population . Bone Joint J . 2017 ; 99-B ( 2 ): 218 – 224 . Crossref Google Scholar

32. Dahm DL , Barnes SA , Harrington JR , Sayeed SA , Berry DJ . Patient-reported activity level after total knee arthroplasty . J Arthroplasty . 2008 ; 23 ( 3 ): 401 – 407 . Crossref Google Scholar

33. Wylde V , Blom A , Dieppe P , Hewlett S , Learmonth I . Return to sport after joint replacement . J Bone Joint Surg Br . 2008 ; 90-B ( 7 ): 920 – 923 . Crossref Google Scholar

Author contributions

T. C. Edwards: Conceptualization, Data curation, Formal analysis, Methodology, Writing – original draft.

B. Guest: Data curation, Formal analysis, Writing – review & editing.

A. Garner: Data curation, Formal analysis, Writing – review & editing.

K. Logishetty: Data curation, Formal analysis, Writing – review & editing.

A. D. Liddle: Conceptualization, Methodology, Formal analysis, Supervision, Writing – review & editing.

J. P. Cobb: Conceptualization, Methodology, Formal analysis, Supervision, Writing – review & editing.

Funding statement

The authors disclose receipt of the following financial or material support for the research, authorship, and/or publication of this article: an institutional research support grant from the Sir Michael Uren Foundation (as reported by J. P. Cobb). Infrastructure support was provided by the National Institute for Health Research (NIHR) Imperial Biomedical Research Centre (BRC).

Acknowledgements

The authors would like to acknowledge the support of the editors and reviewers of this manuscript, for their invaluable contributions.

Ethical review statement

Ethical approval was granted for data collected and used as part of this study (REC Reference: 14/SC/1243, IRAS ID: 136430).

Open access funding

The authors report that they received open access funding for their manuscript from the Imperial College London open access fund.

Follow T. C. Edwards @edwards_tomc

Follow A. Garner @DrAmyGarner

Follow K. Logishetty @klogishetty

Follow J. P. Cobb @orthorobodoc

© 2022 Author(s) et al. This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives (CC BY-NC-ND 4.0) licence, which permits the copying and redistribution of the work only, and provided the original author and source are credited. See https://creativecommons.org/licenses/by-nc-nd/4.0/

Information

11 No.5 | Pages 317 - 326

23 May 2022

https://doi.org/10.1302/2046-3758.115.BJR-2021-0445.R1

Clinical Research Fellow

MSk Lab, Imperial College London, Sir Michael Uren Biomedical Engineering Research Hub, London, UK

[email protected]

Physician Associate

NIHR Clinical Lecturer

Senior Clinical Lecturer

Chair in Orthopaedics

Downloaded 1030 times

Mei XY , Gong YJ , Safir O , Gross A , Kuzyk P . Long-term outcomes of total hip arthroplasty in patients younger than 55 years: a systematic review of the contemporary literature . Can J Surg . 2019 ; 62 ( 4 ): 249 – 258 .

Dawson J , Fitzpatrick R , Carr A , Murray D . Questionnaire on the perceptions of patients about total hip replacement . J Bone Joint Surg Br . 1996 ; 78-B ( 2 ): 185 – 190 .

Murray DW , Fitzpatrick R , Rogers K , et al. The use of the Oxford hip and knee scores . J Bone Joint Surg Br . 2007 ; 89-B ( 8 ): 1010 – 1014 .

Harris K , Dawson J , Gibbons E , et al. Systematic review of measurement properties of patient-reported outcome measures used in patients undergoing hip and knee arthroplasty . Patient Relat Outcome Meas . 2016 ; 7 : 101 – 108 .

Dakin H , Eibich P , Beard D , Gray A , Price A . The use of patient-reported outcome measures to guide referral for hip and knee arthroplasty . Bone Joint J . 2020 ; 102-B ( 7 ): 950 – 958 .

Paulsen A , Odgaard A , Overgaard S . Translation, cross-cultural adaptation and validation of the Danish version of the Oxford hip score: Assessed against generic and disease-specific questionnaires . Bone Joint Res . 2012 ; 1 ( 9 ): 225 – 233 .

Lim CR , Harris K , Dawson J , Beard DJ , Fitzpatrick R , Price AJ . Floor and ceiling effects in the OHS: an analysis of the NHS PROMs data set . BMJ Open . 2015 ; 5 ( 7 ): e007765 .

Dawson J , Beard DJ , McKibbin H , Harris K , Jenkinson C , Price AJ . Development of a patient-reported outcome measure of activity and participation (the OKS-APQ) to supplement the Oxford knee score . Bone Joint J . 2014 ; 96-B ( 3 ): 332 – 338 .

Scott CEH , Bugler KE , Clement ND , MacDonald D , Howie CR , Biant LC . Patient expectations of arthroplasty of the hip and knee . J Bone Joint Surg Br . 2012 ; 94-B ( 7 ): 974 – 981 .

Gulledge CM , Lizzio VA , Smith DG , Guo E , Makhni EC . What are the floor and ceiling effects of patient-reported outcomes measurement information system computer adaptive test domains in orthopaedic patients? A systematic review . Arthroscopy . 2020 ; 36 ( 3 ): 901 – 912 .

Edwards TC , Logishetty K , Cobb JP . Letter to the Editor on “Patient-reported outcomes following total hip arthroplasty: a multicenter comparison based on surgical approaches.” J Arthroplasty . 2020 ; 35 ( 9 ): 2686 – 2687 .

Briggs T . A national review of adult elective orthopaedic services in England: Getting It Right First Time . British Orthopaedic Association . 2015 . https://gettingitrightfirsttime.co.uk/wp-content/uploads/2018/07/GIRFT-National-Report-Mar15-Web.pdf ( date last accessed 5 April 2022 ).

No authors listed . 2022/23 National Tariff Payment System – a consultation notice . 2022 . https://www.england.nhs.uk/wp-content/uploads/2021/12/22-23-National-Tariff-Payment-System-a-consultation-notice-v2.pdf ( date last accessed 5 May 2022 ).

Clement ND , Gaston P , Bell A , et al. Robotic arm-assisted versus manual total hip arthroplasty . Bone Joint Res . 2021 ; 10 ( 1 ): 22 – 30 .

Puliero B , Blakeney WG , Beaulieu Y , Vendittoli PA . Joint perception after total hip arthroplasty and the forgotten joint . J Arthroplasty . 2019 ; 34 ( 1 ): 65 – 70 .

Larsson A , Rolfson O , Kärrholm J . Evaluation of Forgotten Joint Score in total hip arthroplasty with Oxford Hip Score as reference standard . Acta Orthop . 2019 ; 90 ( 3 ): 253 – 257 .

Naal FD , Impellizzeri FM , Leunig M . Which is the best activity rating scale for patients undergoing total joint arthroplasty? Clin Orthop Relat Res . 2009 ; 467 ( 4 ): 958 – 965 .

Ainsworth BE , Haskell WL , Herrmann SD , et al. 2011 Compendium of Physical Activities: a second update of codes and MET values . Med Sci Sports Exerc . 2011 ; 43 ( 8 ): 1575 – 1581 .

Hagströmer M , Oja P , Sjöström M . The International Physical Activity Questionnaire (IPAQ): a study of concurrent and construct validity . Public Health Nutr . 2006 ; 9 ( 6 ): 755 – 762 .

Kujala UM , Kaprio J , Sarna S , Koskenvuo M . Relationship of leisure-time physical activity and mortality: the Finnish twin cohort . JAMA . 1998 ; 279 ( 6 ): 440 – 444 .

Amstutz HC , Le Duff MJ . Effects of physical activity on long-term survivorship after metal-on-metal hip resurfacing arthroplasty: is it safe to return to sports? Bone Joint J . 2019 ; 101-B ( 10 ): 1186 – 1191 .

Le Duff MJ , Amstutz HC . Sporting activity after hip resurfacing: changes over time . Orthop Clin North Am . 2011 ; 42 ( 2 ): 161 – 167 .

Price AJ , Kang S , Cook JA , et al. The use of patient-reported outcome measures to guide referral for hip and knee arthroplasty . Bone Joint J . 2020 ; 102-B ( 7 ): 941 – 949 .

Williams DH , Greidanus NV , Masri BA , Duncan CP , Garbuz DS . Predictors of participation in sports after hip and knee arthroplasty . Clin Orthop Relat Res . 2012 ; 470 ( 2 ): 555 – 561 .

Lübbeke A , Zimmermann-Sloutskis D , Stern R , et al. Physical activity before and after primary total hip arthroplasty: a registry-based study . Arthritis Care Res (Hoboken) . 2014 ; 66 ( 2 ): 277 – 284 .

Vähä-Ypyä H , Vasankari T , Husu P , et al. Validation of cut-points for evaluating the intensity of physical activity with accelerometry-based mean amplitude deviation (MAD . PLoS One . 2015 ; 10 ( 8 ): e0134813 .

Jetté M , Sidney K , Blümchen G . Metabolic equivalents (METS) in exercise testing, exercise prescription, and evaluation of functional capacity . Clin Cardiol . 1990 ; 13 ( 8 ): 555 – 565 .

Hoffman JIE . Chapter 6 - Normal Distribution In Hoffman JIE . ed . Biostatistics for Medical and Biomedical Practitioners . Cambridge, Massachusetts : Academic Press , 2015 : 101 – 119 .

Revicki D , Hays RD , Cella D , Sloan J . Recommended methods for determining responsiveness and minimally important differences for patient-reported outcomes . J Clin Epidemiol . 2008 ; 61 ( 2 ): 102 – 109 .

Ostendorf M , van Stel HF , Buskens E , et al. Patient-reported outcome in total hip replacement. A comparison of five instruments of health status . J Bone Joint Surg Br . 2004 ; 86-B ( 6 ): 801 – 808 .

Hamilton DF , Loth FL , Giesinger JM , et al. Validation of the English language Forgotten Joint Score-12 as an outcome measure for total hip and knee arthroplasty in a British population . Bone Joint J . 2017 ; 99-B ( 2 ): 218 – 224 .

Dahm DL , Barnes SA , Harrington JR , Sayeed SA , Berry DJ . Patient-reported activity level after total knee arthroplasty . J Arthroplasty . 2008 ; 23 ( 3 ): 401 – 407 .

Wylde V , Blom A , Dieppe P , Hewlett S , Learmonth I . Return to sport after joint replacement . J Bone Joint Surg Br . 2008 ; 90-B ( 7 ): 920 – 923 .