Machine vs HRM Exercise Calories: the whole story

For those who don’t want to read the whole thing this is the summary:

I posted this about 3 wks ago, but it did not seem to get much traffic, so I am trying again--especially since I still see questions on this topic every day.

If you don't want to read the whole thing, below is a summary:

Machines and HRMs calculate calories differently.
HRMs do not measure calories directly.
For some exercises, machines can actually be the more accurate number.
HRM accuracy depends greatly on the accuracy of the user setup information.
HRM are only accurate (such as they are) under very specific exercise conditions, i.e. steady-state cardio exercise. Just because an HRM displays a number doesn’t mean the number has any value.
HRM accuracy can also be affected by a number of other factors.

Estimating calories burned during exercise is of primary interest to most people who are trying to lose weight. There are three main sources of exercise calorie information:


Calorie readouts on machines.
Calorie estimating feature on Heart Rate Monitors (HRMs)
Data tables (such as MFP)

I have addressed #3 in another blog, so I am going to focus on #1 and #2.

The “conventional wisdom” is that “machines cannot be believed” and “HRMs are always the most accurate”. Neither of these statements is completely true. It IS true that cardio machines and HRMs estimate calories in fundamentally different ways.

Before we look at that in more detail, let’s make sure everyone is on the same page.

During aerobic exercise, the rate at which calories are expended is a product of two factors:

INTENSITY x BODY WEIGHT

That’s it. (Factors like age, height, and gender are necessary for heart rate monitor estimates, but those factors are only needed for HRMs, as will be explained later).

So: more intense workloads will burn more calories than less intense workloads and, at any given workload, heavier people will burn more calories than lighter people.

Next: the energy cost for any given exercise workload is relatively fixed. For example the energy cost of walking at 3.0 mph and 5% elevation on a treadmill (w/out holding on) is approximately 5.4 METs (a MET is a measure of aerobic intensity). That 5.4 MET intensity is the same for everyone—regardless of age, gender, or fitness level.

So, every given speed, elevation, watt level, etc, has a relatively fixed energy cost. If we can measure the workload, and we have formulae that can accurately calculate the energy cost for a given workload, it is straightforward arithmetic to determine the calories expended.

Machines
Most modern commercial cardiovascular equipment uses computer chips to control the operation of the equipment. The digital control means that exercise workloads can be accurately measured.

For simple movements, such as treadmill walking and running, or stationary cycling, there are longstanding, validated equations for estimating energy expenditure.

Since these machines can accurately measure workload intensity, and since they are programmed with accurate equations for estimating calorie expenditure, all they need is body weight to calculate exercise calories. So, for stationary cycling and treadmill walking (w/out holding on), as long as you enter body weight, the machine readings should be accurate—or at least as accurate as any indirect estimate can be. (It’s a little different for running—not only do you have differences between treadmill running and outdoor running, research I have seen suggests that, starting with speeds above 6.5 mph, the actual energy estimate equation itself starts to overestimate calories—so treadmill running calories are likely 10%-25% high).

Elliptical cross trainers are a different story. While the machines can consistently measure a work output, there is no consistent movement design for cross trainers. Each manufacturer’s equipment is different. Therefore, there is no one validated equation that can be applied to all cross trainers. Each manufacturer can program the machine any way they want. The only way to have an accurate calorie estimate for a cross trainer is for the manufacturer to develop and validate their own custom algorithm for each machine—that is expensive and time-consuming. As I have mentioned before, the only company I know that has even tried to do this is Life Fitness, and they have only done it on a select few models. So, in the case of elliptical, it is best to assume that NONE of them are particularly accurate.

It is important to repeat that cardio machines estimate calories differently than HRMs—they measure the actual workload being performed and use that for the estimates. The accuracy of the machine readings depends on the following factors:


How accurately the machine can measure actual exercise workload
Does the machine allow user to input weight?
How valid are the equations programmed into the machine?

Based on these factors, treadmill walking and stationary cycle calorie estimates should be pretty accurate, running off by up to 25% (my estimate), and elliptical cross trainers – who knows? (Stairmasters could be fairly accurate, but their equations are based on not holding on at all and I know very few people who do that).

Heart Rate Monitors
Heart Rate Monitors estimate calories by a completely different method than machines. It is important to emphasize up front that HRMs do not “measure” calories—they don’t have any special sensors or anything. In fact, the ONLY thing that HRMs measure is heart rate. That’s it—the calorie function is just a very indirect estimate.

How do HRMs estimate calories? They use algorithms based on the relationship between heart rate and oxygen uptake (VO2). It’s increased oxygen uptake that is actually responsible for increased calorie burn.

During steady-state cardiovascular exercise, there is a relatively fixed relationship between heart rate and VO2. An increase in workload requires an increase in VO2 which leads to an increase in heart rate. (A decrease in workload has the opposite effect). We know some approximate relationships between HR levels and VO2 level—e.g. 70% of HRmax is equal to 57% of VO2 max, 85% of HRmax is equal to 70% VO2max, etc.

Put very simply—if we know a person’s HRmax, HRrest, and VO2max, the individual “scale” for that person can be established. If the VO2max is 40, and we know the heart rate is 85% of HRmax, and we know that 85% HRmax = 70% VO2max, then we can calculate that the exercise workload is 28 (70% x 40), and that, along with body weight, allows us to calculate calories.

HRM manufacturers create algorithms that attempt to capture and refine this relationship, as well as make the algorithm applicable to the widest range of individuals. They validate the algorithms by taking a group of subjects and comparing the VO2/calorie expenditure predicted by the algorithm to actual VO2 measurements taken with a metabolic cart. It is in the generation of these algorithms that factors such as age, gender, and height, for example, are necessary factors to improve accuracy (don’t ask me why—I am NOT a mathematician or statistician).

Just because an HRM requires more setup info (height, age, gender, etc) does NOT mean it is "more accurate" or that it is "more customized for your body". The extra setup info is actually necessary to compensate for the inherent weakness in an HRM -- the fact that it cannot measure your actual workload.

It must be pointed out that this method is very much based on RELATIVE intensities rather than fixed workloads. In other words, an exercise heart rate of 150 beats/min means absolutely nothing unless we know the context – the persons HRmax, HRrest and VO2max.

It also must be emphasized that the HRM algorithms are ONLY valid under conditions in which there is the consistent relationship between heart rate and VO2 as described above. If heart rate increases without an increase in VO2, then the calorie numbers are bogus. REMEMBER: HRMs do not measure calories—they estimate calorie burn under very specific circumstances.

The accuracy of HRM calorie estimates depends on these factors:

Accuracy/quality of the algorithms programmed into the device (these are usually proprietary for each manufacturer).
Accuracy of the user setup information
How closely the exercise activity matches the research conditions under which the calorie counting algorithms were developed (i.e. steady-state cardiovascular exercise).

As you can see, there is no single answer as to which method is “most accurate”.

For simple activities, such as treadmill walking, walking/running on level ground outdoors, and stationary cycling (on a quality commercial bike), the calorie displays on the machines can be quite accurate because they measure actual workload.

For activities such as cross trainers, stairclimbers and more unstructured activities like group exercise classes, walking outdoors on hilly terrain, then HRMs are likely to be more accurate.

However, if HRMs are not set up properly with accurate user data, then the accuracy of the numbers can be WAY off. Under the best of circumstances, HRMs are only about 80% accurate. Conditions such as illness, lack of sleep, hot weather, and cardiovascular drift can significantly lower HRM calorie accuracy even further.

For activities/conditions that are different that the conditions under which HRM calorie equations were developed, HRM calorie numbers are not accurate at all. These include: activities of daily living or being at rest, lifting weights, thermal stress (e.g. hot yoga).

UPDATE: The moral of the story is, as always, not to look at ANY calorie estimate as a precise number. Most of you know this already, but it always bears repeating.