RESUMO
We tested the hypothesis that a deuterium oxide (D2O) tracer could discriminate among patterns of clinically significant, imperfect compliance during drug trials. A model was developed to predict deuterium concentration during multiple dose regimens. After developing a regression equation to predict one of the model parameters for children, we selected healthy children (N = 20) at random to receive one of five 10-day D2O regimens. Five urine samples were obtained from each child during 15 days and analyzed for deuterium level by mass spectrometry. Each child's height, weight, age, and the first four urinary deuterium levels were used to estimate the amount and timing of deuterium administration. These estimates were compared with the five regimens to determine the closest match between estimate and regimen. The closest matching regimen was the regimen actually administered to 19 (95%) of the 20 children. Two of these children had D2O administration estimates that could be confused with another regimen. The correlation between the model's predicted levels and the measured levels of all urine samples was 0.96. We conclude that a D2O tracer shows excellent promise as a quantitative method of assessing compliance with liquid medications under specified conditions.