EXPERIMENTAL RELIABILITY OF THE DOUBLY LABELED WATER TECHNIQUE

Experimental reliability of the doubly labeled water technique. Am. J. Physiol. 266 (Endocrinol. Metab. 29): E510-E515, 1994. - The experimental reliability of measuring CO2 production rates (rCO(2)) with the doubly labeled water (DLW) technique was assessed in five young healthy men (23 +/- 4 yr; 66.1 +/- 4.6 kg). To minimize the confounding effects of fluctuations in physical activity and eating patterns on variation in energy expenditure, the subjects lived under sedentary living conditions by confinement to their own room at a Clinical Research Center and were maintained on a fixed and known level of energy intake. rCO(2) was determined in duplicate over two identical g-day study periods after separate loading doses of deuterium and oxygen-18. Turnover rates were determined from multipoint sampling to reduce error from analytical uncertainty. Dilution spaces were determined by both the intercept and plateau methods. The average experimental variation for rCO(2) estimates was approximately +/- 8.5% and was not significantly different among three published calculation models that differ in their assumptions regarding the relationship between the dilution spaces of deuterium and oxygen-18. The experimental reliability of +/- 8.5% exceeds theoretical values generated from calculations based on propagation of error from analytical uncertainty. Between subjects, the experimental variation ranged from 1 to 21%, and the half-width of the 95% confidence interval for the precision of rCO(2) estimates was high (+/- 12 mol/day) relative to the mean reported value of similar to 16 mol/day. We conclude that 1) after minimizing the potential for variation in energy expenditure due to physical activity and dietary factors the experimental variability of the DLW technique in adult humans is +/- 8.5% and highly variable between subjects; 2) theoretical values suggest that the DLW technique has a better reliability than is measured empirically; and, 3) theoretical calculations underestimate the experimental variability of the DLW technique.