In the Wikipedia page of the Metabolic equivalent of task, one MET is defined as
1 MET = 1.162 W/Kg.
In the same page, it's claimed that the activity level of bicycling, on flat, 10–12 mph (16–19 km/h), light effort is 6.0 MET, that is, almost 7 W/Kg, if we plug the definition above.
Now, 7 W/Kg is the FTP of Alberto Contador[1] from his Tour de France-winning days, which nobody would call light effort.
My question is: What is it here that gets lost in the translation from MET to W/Kg?
Our bodies burn oxygen to produce energy. That energy powers things like your brain, your heart, your lungs, and locomotion - or cycling! Oxygen consumption is like power. It can be expressed in milliliters of O2 per minute or we can normalize that for body weight. Or, we can express that in metabolic equivalents of task (METs), where 1 MET is defined as 3.5 mL O2/kg min, equivalent to 1.162 W/kg generated by the whole body.
Cycling power measures work done to the bicycle. However, your body has to burn energy to turn your legs. The human body is only about 22% efficient at producing cycling power. As with all machines, the balance is wasted as heat. The math below should demonstrate that if your body is working at 6.0 METs, that should correspond to 1.5-ish W/kg delivered to the bicycle - but humans vary in their efficiency, so that's a range. That's probably easy-ish for many trained cyclists. Someone in the general population who maybe cycles a couple miles to the store may find that to be a moderate effort, and/or they might sustain it for less time than a trained athlete.
METs aren't really a consumer-facing thing. They're used by researchers to measure the intensity of various physical activities. When you read the government recommendations to get a few hours of at least moderate physical activity, and here's a list of moderate intensity exercises/activities, that's one use of METs. Sports where you can stick a strain gauge on equipment are conducive to direct measurement of power. But for things like yoga, housework, yardwork, swimming, etc, where do you stick the strain gauge?
In the lab, we can measure oxygen burned by the body. This corresponds to the power generated by the aerobic energy system, because that system uses oxygen to produce energy (by processing glucose into ATP, consuming oxygen and producing CO2).1 This measurement involves strapping a mask to your face and measuring the amount of oxygen burned. This (free to access) paper has a description of the method of indirect calorimetry.
If we stuck a portable oxygen sensor to your face, we could have it transmit your oxygen consumption (absolute or normalized for weight) and have it transmit to your head unit or smart watch. You could have the head unit express that in absolute VO2, VO2 per kg or lb weight, or METs. You could then measure your threshold and pace intervals in terms of that reading, just like you could do with cycling power. Of course, the mask is probably more obtrusive than your power meter, so I imagine people will generally prefer the latter.
By convention, 1 MET is defined as 3.5 mL O2/kg min, equivalent to 1.162 W/kg (produced by the body, not delivered to the crank). The math is shown on the Wikipedia article. That oxygen consumption rate was based on one 70 kg, 40 year old, healthy man. Our actual resting metabolic rates are probably a bit less than this and there is person to person variation.2
Finally, METs can be used to compare different exercises by intensity. For example, moderate walking, yoga, and light housework were estimated to burn about 3.0-3.5 METs - that is, you burn 3.0-3.5x your resting metabolic rate. The American College of Sports Medicine's exercise guidelines have some guidelines on aerobic activity, and they want you to be doing things that are at least 3-6 METs a few times a week. The Wikipedia page in the question shows MET estimates for a number of physical activities. That covers walking, housework, tennis, and lifting weights. For comparison, moderate to hard swimming is listed as 8-11 METs, football (soccer in the US) as 10.3, and running about an 8:47 mile pace at 11.2.
Gross efficiency or gross mechanical efficiency is the power delivered to the crank divided by the power generated by the body to turn your legs and run essential functions. Hopker and colleagues (2019) (should be ungated, link is to Pubmed) compared efficiency between trained competitive male and female cyclists. At lactate threshold (roughly equivalent to cycling at functional threshold power), women had an average gross efficiency (GE) of 23.2%, standard deviation 3.5 percentage points. Men had an average GE of 21.2%, standard deviation 1.7 percentage points. The authors reported that the distribution of GE was roughly normal. So, a bit over 95% of persons like those in the sample are within 2 standard deviations of the mean.
6.0 METs is, by definition, 6.972 W/kg of energy consumed by the body. We can estimate that 95% of men have GE ranging from 17.8% to 24.6%.3 So, if a group of trained male cyclists cycled with their O2 meters at 6.0 METs, most of them would have their power meters reporting 1.24 to 1.72 W/kg. Yes, we normally show raw watts, but you could make them report W/kg, and Zwift does this already.
Since going from W/kg in cycling power to METs is an arithmetic exercise (you could just assume 22% efficiency for simplicity), those who are curious can work out how many METs they're burning at higher power. For example, a lot of fitter cyclists can hold over 3 W/kg for some time. Pro women can have thresholds in the 5 W/kg range and higher. Pro men can be higher. The top of the men's field may be pushing 7 W/kg or so. If you do the arithmetic, that's a lot of METs.
The American College of Sports Medicine and the American Heart Association recommend that (2007) people try to get 150 mins per week of moderate (~3-6 METs, e.g. brisk walking) aerobic exercise, or 60 minutes per week of vigorous (~6+ METs) aerobic exercise. Those guidelines are aimed at healthy adults in the general population age 18-65.
What if you're a consumer of about average fitness who's looking to get the recommended amount of physical activity? I wouldn't worry too much about exactly how many METs you're working at. First of all, we are all at different fitness levels, so we can sustain different amounts of METs. It's about what you can do now. Ideally you increase your duration and/or intensity as you train, but do this at a pace you can sustain. Of course, the harder you go, the less exercise time you need to see the benefits.
In cycling terms, I'd guess that a lot of adults age 18-65 with some experience and without major health conditions should be able to do at least about 1.5 W/kg for a while, maybe 15-30 minutes - remember that this is about 6.0 METs, which is the threshold for vigorous exercise as defined above. Don't forget you can coast, maybe take a breather. You can compare your perceived effort to taking a moderate pace walk as well. Some gym equipment can display MET consumption. Remember that this is estimated, and we don't now how accurate the estimates are! Your own actual MET consumption may differ from the equipment's estimate. You can obviously use that to guide your effort but you have to pair that with perceived exertion.
Since smart watches are so common, you also can check your heart rate. For example, I am probably at 60-80 bpm during the day when mostly sitting, and probably 90-100 when taking a walk around the neighborhood (moderate exercise). At a cycling pace that's easy for me but is probably about 6 METs, I usually get to about 120 bpm. My numbers are only for illustration. Everyone's resting, maximum, and typical heart rates during exercise will vary, sometimes by a lot, so you have to base your judgment on your own heart rate response.
For those who are actually interested in METs, there's a publicly available compendium of MET values here. Here's the page for cycling. Below is a table from the 1993 version, although the page has updated values as of 2011.
| Description | METs |
|---|---|
| Bicycling, <10mph, general, leisure, to work/for pleasure | 4.0 |
| Bicycling, 10-12 mph, leisure, slow, light effort | 6.0 |
| Bicycling, 12-14 mph, leisure, moderate effort | 8.0 |
| Bicycling, 14-16 mph, racing or leisure, fast, vigorous effort | 10.0 |
| Bicycling, 16-19 mph, racing or not drafting, or >19 mph drafting, very fast, racing general | 12.0 |
| Bicycling, >20 mph, racing, not drafting | 16.0 |
| Bicycling, BMX/MTB | 8.5 |
If you go through the site and read the 1993 and 2000 articles in peer-reviewed journals where they explain their purpose and process, not all of the MET values are directly measured. Some are estimated using the authors' best judgment from similar activities. Plus don't forget that we would expect different people and different bikes to have some variation in MET costs to travel a certain speed - on a high-performance road bike, I would be surprised if it took even 4 METs to travel at 12 mph on flat ground.
Anyway, METs have been used in epidemiological studies. For example, we might have participants catalog their physical activity with a diary. Then we might calculate the number of MET-hours they accumulated during a week. Then we might see if that predicts health outcomes. The exercise recommendations earlier probably stem from some study like that. For this purpose, you can't achieve high precision. Thus, those approximations above are just what epidemiologists have to live with in this setting.
Footnotes
1: I think this omits the contribution of the anaerobic energy system. That system processes glucose without burning oxygen, and can deliver energy rapidly. This accounts for a negligible amount of energy when you're exercising at a moderate intensity. If you're doing a 10s-2 minute max effort, then a lot of energy is coming from anaerobic systems. If you were able to measure VO2 consumption to pace your intervals, it would probably lag your power output, as heart rate does. And if you were doing a 1 minute effort and trying to ramp up your power, you might see your VO2 consumption remaining flat.
2: Humans have a range of resting metabolic rates, i.e. not everyone burns 1.0 METs. This study found that the standard MET definition overestimated actual resting O2 expenditure in their sample by an average of 35%, with variations in fat-free mass accounting for a lot of the variation in resting VO2. But this doesn't affect the exercise prescription, it just means you burn <1 MET at rest.
3: Previously I subtracted 1 MET from the reported 6 METs to account for base metabolic rate. I now think this wasn't correct; in the lab, you would be measuring the total energy consumed during an activity, but this should include your base metabolic rate. From our power meter numbers, we'd estimate the total energy burned by the body (Strava appears to assume 22% gross mechanical efficiency, at least for men). I think these two quantities are comparable.
If I'm wrong, then the energy estimates from the METs numbers are a bit lower than what I reported above.