Inaccurate calorie count for walking

[Lena_Amsterdam]

Hi! Does anyone know if Fitbit has any plans to fix the wildly inaccurate calorie counts for walking? 

I have no problem with running, which is pretty spot-on. Same for cycling. But walking is overestimated by a factor of 3. I have seen quite a number of people complaining about it in different posts for the last year or two, but I cannot seem to find any relevant response from Fitbit as to whether this will be fixed sometime soon. 

Thank you!

[GershonSurge]

@Lena_Amsterdam,

I agree that according to the research, the average person burns fewer calories than I do when walking. However, we don't know the extremes. No way to find that out either. 

What I have found is the Surge overestimates calories by 10-15%. The Zip underestimates by the same amount. The Charge 2 is about right. I computed this when I was losing weight and meticulously tracking the calories I ate.

I'm at the point where I step on the Aria scale every day or to and adjust how much I eat based on my weight. I don't track calories any more.

[Baltoscott]

---

@Dominique wrote:

FWIW, I currently wear three Fitbits: Surge, Charge 2 and Alta HR. This morning, I did a 5.5 km walk in 54 minutes one way, and (slightly later on) 5.5 km in 53 minutes on the way back (average speed: 6.1 km = 3.8 mph). I recorded both with my Surge. These same walks were autodetected by my Charge 2 and Alta HR.

Here were the calories for each tracker:

Surge: 486 + 446 (avg: 466)

Charge 2: 443 + 415 (avg: 429)

Alta HR: 482 + 479 (avg: 480)

According to the Compendium of Physical activities, "walking at 4.0 mph on level, firm surface" is 5.0 METS. I actually walked on uneven, non-flat terrain, which was probably more demanding. My calculated BMR is 1438, so in 55 minutes I would expend 5.0 x 55 x 1438 / 1440 = 275 calories. Based on that, all three Fitbits would overestimate calories burned, but not by a factor of 3.

---

I think the overestimate is related to walking activities that take heart rate into account.  I had a couple auto detected activities yesterday that showed calorie burns in the range reported by @Lena_Amsterdam, which I agree seems high.  One of the autodetected activities was about 15 minutes, on pavement in DC, at my normal "exercise" walking pace of about 3.5 mph, so I used it as a base for comparison.  I am a 56 yr/old male currently weighing 174 lbs.

This morning, I added two fake 15 minute manual walking activities for a time frame shortly after I woke up when I was preparing or eating breakfast. Low activity. For one, (starting at 6 AM in the picture), I did not enter a distance, but instead said I walked at 3.5 mph -- a drop down list is available to choose from.  For the second I put in 0.7 miles, which would be 2.8 mph.  For that one, fitbit calculated a step count based on my stride length divided into the distance I reported.  Both activities overwrote my true activity level up to that point upping my total calorie burn slightly.  

As expected, the 2.8 mph activity generated fewer calories (56) than the 3.5 mph activity (64) over the 15 minute test period.  Curiously, in both cases, the reported calorie burn was significantly less than the 148 calories attributed to my 15 minute autodetected walk from the day before.  

I don't know whether it true or not that the calorie burn reported in activities includes BMR.  Mine is currently around 17 calories for 15 minutes based on this calculator.  Even if you subtract that amount from the autodetected calorie burn it still come in at 131, more than twice either estimate from fitbit's standard non-tracker influenced estimate for walking.  

I doubt that I am really burning significantly more calories when walking than the average guy my age and size, so I agree that the calorie burn assigned to HR walking activities is high for me (but maybe not @GershonSurge).  

But in the end I don't lose much sleep over it.  I have successfully lost 25+ lbs since early January at the rate of almost 2 lbs/week.  During that time period I have tracked all the calories I've consumed in MyFitnessPal, and let fitbit record the calories I've burned based primarily on whatever the tracker records 24/7 other than daily bike rides (which I import from Garmin via Strava).  The caloric deficit that fitbit reports equates to around a projected weekly loss of at least 50% more than my actual results -- usually 3-4 lbs / week.  Based on this experiment I deduce that at least part of the reason for the difference is because in some circumstances fitbit is overestimating my calorie burn.  But I am also aware that the caloric values assigned to the foods I eat are based on averages, and may in some cases be dramatically wrong (and of course my estimates of consumption will be off as well, especially for meals I don't make and weigh myself), so there surely are errors on that side of the equation as well.  In the end, at least with respect to weight loss/gain/monitoring, I just start with the numbers that fitbit reports and then adjust based on actual results.  That works fine.

[Vinny0158]

I have a versa 2 watch connected to my phone and I walk every night approximately 5 miles but I have not ever seen it calibrate my calories what’s wrong with this?

[Dominique]

---

@Vinny0158 wrote:

I have a versa 2 watch connected to my phone and I walk every night approximately 5 miles but I have not ever seen it calibrate my calories what’s wrong with this?

---

I must admit I don’t understand the "calorie calibration" part. Once you’ve created your account and connected your tracker to it, Fitbit "knows" your personal data (age, gender, height, weight) and its algorithm starts estimating calories based on detected activity, heart rate etc.

What do calories look like during your nightly walks? Do they look wrong to you? Too high, too low?

[kolleenf]

Has there been a fix found for the overestimation of calories burned while walking? I’ve been using the Charge 3 for a few days now and have found that - like with Lena - my calories burned with steps / walking is about 3x higher than what would be normal for me. I am 5 ft, 113lbs, and according to Fitbit, my ~10,000 steps today burned 1,600+ calories, and this is above my BMR. That’s an insane overestimation!

[Baltoscott]

@kolleenf - I may not have understand your post, but the daily calorie burn Fitbit reports includes your BMR.  For you, BMR is probably around 1100 kcal or a bit more.  10k steps over the course of the day is a rough proxy for “lightly active” so 1600 calories for the day seems in the ballpark (calculator.net)

 For what it is worth, I find Fitbits to be helpful in analyzing exercise trends as my weight changes (if weight and exercise went up last month when I though I was eating the same, I was probably wrong) and tracking sleep and heart rate changes.  I don’t pay much attention to calorie burn for exercise. 

[Heybales]

---

@kolleenf wrote:

Has there been a fix found for the overestimation of calories burned while walking? I’ve been using the Charge 3 for a few days now and have found that - like with Lena - my calories burned with steps / walking is about 3x higher than what would be normal for me. I am 5 ft, 113lbs, and according to Fitbit, my ~10,000 steps today burned 1,600+ calories, and this is above my BMR. That’s an insane overestimation!

---

It's an overestimation when HR-based calorie burn is calculated for a low intensity exercise like walking.

Any calculation for HR-based is going to be inflated when down at the bottom of the aerobic exercise range - same as it's inflated when at the top of the range approaching anaerobic.

And no they haven't corrected.

Formula-based would be so much more accurate - studies have been around for decades based on research studies that show how accurate they are.

Oh - it's the distance the steps takes you, not the steps by themselves.

That's why it's always good to confirm Fitbit sees a correct distance when walking about a mile at average daily pace - about 2 mph.

That way it can dynamically adjust up or down based on going faster or slower.

And since 10K steps is pretty major, an incorrect stride length setting can lead to bad distance and calorie burn calculation for all your normal daily activity calories too.

Never mind the exercise level walking with HR-based calories.

What kind of distance did that 10K steps give you?

[kolleenf]

The distance on the Fitbit app reads as 6.59km, which I believe is also an overestimation. We are still in government-mandated quarantine where I’m from so that distance is really just me walking around at home, and up and down the stairs.

I had intended to use the Charge 3 primarily as a fitness tracker to help me determine net calories / as tool for macro counting. I understand no fitness tracker is 100% accurate with counting calories, but I would at least like to get better estimates.

I haven’t customized my stride length. Will this get me closer to a solution? 

[Heybales]

---

@kolleenf wrote:

The distance on the Fitbit app reads as 6.59km, which I believe is also an overestimation. We are still in government-mandated quarantine where I’m from so that distance is really just me walking around at home, and up and down the stairs.

I had intended to use the Charge 3 primarily as a fitness tracker to help me determine net calories / as tool for macro counting. I understand no fitness tracker is 100% accurate with counting calories, but I would at least like to get better estimates.

I haven’t customized my stride length. Will this get me closer to a solution? 

---

Sadly it depends.

Where do the majority of your daily steps (and therefore distance and calories) come from - daily activity or exercise?

If it's from getting out and exercise level pace walking or jogging - then the device is slipping into HR-based calorie burn - and if at the bottom of the aerobic range - it'll be inflated.

Only way to get around that for walks anyway is for Fitbit to get past the initial 2 weeks learning, see what your resting HR is and levels reached in small daily activity - and stop slipping into HR-based calorie burn when you aren't doing a high intensity walk.

If majority of steps is from daily activity, not a straight workout - then the stride length is more important. But you need a decent distance to confirm and have figures to work with it - like walking across the living room not useful.

And need to go at pace that is in middle of range from grocery store shuffle to exercise level - whatever that midpoint is.

Because during those times it's all about distance and mass and time for calorie burn.

Sounds like you may be limited right now though.

Forgot to add - ways to improve the HR-based calorie burn.

RestingHR is tracked and used in formula - no way to correct that.

But HRmax is estimated from 220-age. And that is used in formula too.

And women have better chance of being more than 10 bpm outside that calculated figure than being within it.

So for instance if your HRmax was higher than calculated - then right now any HR appears to burn more than it really does. 

Have you ever done a workout where you hit a really high HR from max effort within about 5-10 min of starting the workout?

Like some interval workout where the rests were just too short to recovery, and next set just kept pushing and you saw a HR higher than you'd ever seen?

[kolleenf]

Thanks for the detailed explanation. Yours seems to be the answer that addresses my concern the clearest and most straightforwardly.

My daily calories from exercise come from interval / circuit / weight training, and so I can only account for those activities being responsible for majority of my active calories. However, that’s only from about an hour of my day. The rest of my calories recorded burned are from the usual daily activity.

My last 3 workouts I wore the Charge 3 and another HR-measuring device, and the max HR read the same for both devices.

I finally had the chance to measure my stride lengths today for both walking and running, and I’ve just customized them on the Fitbit app. I’ll be observing how this changes things the next few days.

[Heybales]

That will probably help the majority of the day then.

Most people find it the case that unless they just do hours of exercise (long bike rides or runs) and that wipes them out so they are also very sedentary rest of the day - exercise is not the major component of daily burn.

Now, a fun caveat to confirming HR during exercise is right, or the HR-based calorie burn.

There is no direct link between HR and calorie burn, lose correlation with several possible caveats to it.

While a certain HR bpm may indeed correspond to a certain level of effort and therefore calorie burn, when your body gets too hot HR goes up merely to move more blood used for cooling, not because you are working harder. When the body is already stressed or tired it can be higher than normal also. If you have sweat a lot and harder to move the blood around then a higher HR also.

So there are all kinds of reasons a HR will go higher than needed, and for HR-based calorie burn calculations - that means inflated calorie burn.

The other kicker those calculations are only a good estimate for aerobic steady-state workouts, same HR for 3-5 min at a time. That's the range I referenced when I said inflated down at the bottom and the top of the range.

The problem is many types of workouts are not steady-state, and the increase in HR has nothing to do with increased oxygen requirements for increased fat burn that the calculations are for.

Like you could get done with a set of squats and HR is racing, and then you sit on the bench to recover for next set. Your HR doesn't immediately drop to normal HR needed to sit on a bench, it slowly drifts down - that entire time being inflated for the level of effort and calorie burn actually occurring.

Your workouts are opposite of steady-state with HR moving all over the place, and likely going into anaerobic if being done right.

For those, database entries are actually more accurate for calorie burn, for Weights and Circuit training. Still use the watch to create an Activity Record so you know the times and can observe HR and other stats - but create your own Workout Record using same Start/duration times and let Fitbit estimate the calories.

That Workout will replace the calorie burn in the daily stats, so you can still keep the Activity for rest of the stats.

For interval running/walking for instance - again making your own Workout Record afterwards with the distance Fitbit said and the time it said, and let it compute calorie burn based on that info will be more accurate.

I'm sure you aren't doing those workout solely for the purpose of calorie burn anyway - but rather the good response it has on transforming your body and making you more fit.

[kolleenf]

You’re correct. Calories burned are really just a tool for me, and not really the end goal here.

Still, it’s interesting to see how Fitbit has been interpreting workout data, and I’m going to try your suggestion to use Activity Record but create my own Workout Record using the same duration times — just to see how much of a variance it makes. 

Thanks again! 

[Heybales]

I need to correct myself on part of that.

A few of the models (can't recall now) - and after upgrades maybe more - already log Weights correctly not by HR but rather by database rate of burn which is more accurate.

If you know your BMR / 1440 min in day x 3.5 METS = per min rate of burn.

So say BMR 1500 / 1440 x 3.5 x 30 min = 109 calories total.

So if you manually log that workout and get almost exactly the same as Fitbit showed anyway - you got the right model!

Circuit training is 8.0 - 8.5 METS multiplier - I never confirmed if that workout was switched over to database rate of burn.

[ca4man]

It really depends on what model you're using and it varies by the release year.  I can't find any scientific studies that conclude calories are off by a factor of 3.  I did find an article referencing an 18-month-old study that concluded that the Charge 2 overestimated calories burned while walking by about 50%, but was fairly accurate for running (4% underestimate).   

https://www.huffingtonpost.co.uk/entry/fitbits-overestimate-calories-burned-by-50-percent-study-clai...

[Heybales]

---

@ca4man wrote:

It really depends on what model you're using and it varies by the release year.  I can't find any scientific studies that conclude calories are off by a factor of 3.  I did find an article referencing an 18-month-old study that concluded that the Charge 2 overestimated calories burned while walking by about 50%, but was fairly accurate for running (4% underestimate).   

https://www.huffingtonpost.co.uk/entry/fitbits-overestimate-calories-burned-by-50-percent-study-clai...

---

That study was worthless and ran by people that don't understand how the devices work.

Yes - walking is on the lowest end of the aerobic exercise zone - and this means the HR-based calorie burn will show the worst inflation compared to the middle of the aerobic zone.

Right below and going into the anaerobic zone is the same issue.

Using distance-based calorie burn would be a whole lot more accurate - which the other devices in the study likely used.

But in this study which was a workout session - they took devices and threw them on people and ran the test.

These newer devices require 1-2 weeks of seeing workout data, seeing resting HR, ect, to get a better estimate of HR-based calorie burn for that person.

You'll still be inflated at the bottom & top of the aerobic range though. But not that bad.

[geoff1998]

I agree Fitbit is wildly inaccurate measuring calories burned during walking exercise

running is comparable to Apple, as is elliptical 

walking is silly

apple device reading me as burning around 225 to 250 calories per hour walking at 5.00 km’s per hour vs 400 to 420 for fitbit

i wear both every day ( on alternate wrists), and do my comparisons are contemporaneous 

it’s always been an anomaly with Fitbit ( and yes I’m sure it’s the Fitbit that’s wrong)