My FitBit versus DirectLife Experiment

My Weightology Weekly subscribers know that I've written a lot about NEAT and the use of accelerometers to help track NEAT.  In particular, I've written a number of articles looking at the accuracy of some accelerometers for estimating energy expenditure, including the DirectLife, the FitBit, and the Sensewear Armband.  I recently obtained a Fitbit, and have had a DirectLife for a while now.  I was curious how the Fitbit and Directlife would compare in regards to their energy expenditure estimates.  Since I'm not a subscriber to the DirectLife service, I can only see my energy expenditure for the current and previous day, so I can't look at a large sample and compare it to the Fitbit.  However, it's still interesting to compare the estimates for a single day.  I thought yesterday might be a good day to compare the two devices, since I had a good mixture of physical activity and sedentary behavior, with a weight training session in the morning and an afternoon walk with my wife.  Throughout the rest of the day I was at the computer either working on our monthly finances or writing.

The total daily energy expenditure estimate from the Fitbit was 2,887 calories.  The DirectLife doesn't give you a total daily energy expenditure estimate; rather, it only tells you energy expenditure through physical activity.  The DirectLife gave me an estimate of 899 calories.  To get an estimate of my total daily energy expenditure, I would need to add my resting metabolic rate (RMR, the energy you expend at rest to maintain basic life functions) to that number, along with the calories expended through the thermic effect of feeding (the heat produced from the digestion of food).  It's been a long time since I've had my RMR formally tested; however, I remember at the time it was measured at about 90% of what the Mifflin equation predicted for me.  If I plug my numbers into the Mifflin equation now, it produces an RMR estimate of 1695 calories.  If I take 90% of that number (assuming that my true RMR is still around 90% of Mifflin), then that puts my RMR at 1526 calories.  To get the thermic effect of feeding, I need to take 10% of my average daily calorie intake, which lately has been around 2,250 calories per day.  That brings us to 1526 + 225 = 1751 calories.  Finally, if I add the DirectLife estimate of 899 calories, that gives a final value to 2,650 calories.

So, my energy expenditure estimate for the Fitbit was 2,887 and for the DirectLife was 2,650 calories.  Now, to be fair, I had reported in my Fitbit account that I had done a weight training session, so the software automatically tacked on about 200 calories to my estimate.  I cannot report such activities to the Directlife without having the subscription.  If I hadn't reported the weight training session in the Fitbit, it would've estimated me at around 2,687....very close to the Directlife estimate of 2,650.

How close are these estimates to my true energy expenditure?  I have no way of knowing without being a subject in an experiment that uses doubly labeled water.  But it's interesting to compare the estimates of both devices nonetheless.