DLW is in the news today as the UK Government's "Nudge Unit" - now operating independently as Behavioural Insights - has been looking at alleged under-reporting of calorie intake in official statistics.
One aspect of accuracy is how repeatable a measure is - if you take the same person and repeat a DLW test in the same circumstances do you get the same answer ? This has been done :-
My biggest query about DLW is whether it has been validated as a way to determine energy intake in a controlled situation where food eaten was independently verified. Most of the published work uses DLW to determine "calories out" and therefore points the finger of under-reporting at anyone reporting a smaller value for "calories in".
The US CALERIE project rides to my rescue with a 7 day calorimeter study snappily titled "Validation study of energy expenditure and intake during calorie restriction using doubly labeled water and changes in body composition". The methodology was to measure energy expenditure by DLW while subjects were contained in a metabolic chamber to provide another measure of energy expenditure. This came out fairly well :
"We found close agreement (R = 0.88) between EE measured in the metabolic chamber and EEDLW during CR. Using the measured respiratory quotient, we found that the mean (±SD) EEDLW was 1934 ± 377 kcal/d and EE measured in the metabolic chamber was 1906 ± 327 kcal/d, ie, a 1.3 ± 8.9% overestimation."So DLW reported slightly higher values than the chamber gas analysis but the difference was less than 30 kcal/day.
The next step was to combine the DLW data with changes in body composition and weight, this didn't turn out so well :
"EI calculated from EEDLW and from changes in ES was 8.7 ± 36.7% higher than the actual EI provided during the chamber stay (1596 ± 656 kcal/d)."Oops. DLW based calculations estimate a food intake 8.7% higher than that delivered to subjects living in a metabolic chamber. Providing 1430 kcal of food on average led to calculated energy intakes of ~1600 kcal on average using DLW for energy expenditure and weight / DEXA composition to determine the change of energy stores in the body.
So here we see a difference of 170 kcal/day under tightly controlled conditions. Did the subjects sneak in a candy bar each day ? Is the DEXA analysis suspect - it reported a fat-free mass loss of 0.7 kg from the 0.9 kg weight loss. Had all of the 0.9 kg been fat loss the change in energy stores would have risen from a loss of 337 kcal/day to over 1000 kcal/day so this is obviously a sensitive part of the analysis.
So far I am seeing DLW as a method that might be 100 kcal/day off when estimating energy expenditure, and when combined with body weight and composition changes the error may increase when trying to use this as a measure of energy intake.
A couple of other references for measured intake vs DLW - in female athletes the DLW + bodyweight change estimate was found to be "221 +/- 550 kcal/day in excess of those calculated from food records (2,193 +/- 466 kcal/day)". Note that 221 +/- 550 means a range from +770 to -330 which is a considerable range of uncertainty.
In another study Underweight adults also failed to perform to expectation, overfeeding by 720 kcal/day for 8 weeks failed to increase measured energy expenditure by DLW and 1/3 of the extra energy went unaccounted for during the ~2 kg weight gain.
So where does this leave us ? I don't believe doubly labelled water is sufficiently robust and proven by validation to allow it to be used to determine whether reported energy intake by other means is flawed.
There seems to be too much blind faith in the calorie hypothesis at work - the population got fatter therefore their calorie intake must have gone up even though careful measuring efforts don't show this. Does the science of DLW and assumptions about calories and weight gain really trump detailed reporting of food intake ?