Limits of confidence in tracer compounds as a means of measuring patient compliance with medication.

Chemical tracers are thought to be a desirable means of measuring patient compliance with medication. However, no compound has emerged as a standard. To explain this observation, the authors explored the mathematics of simulated tracers with half-lives 5 to 140 days over a month of daily dosing. To model assay variability, they added random "noise" to calculated tracer serum levels. They then fitted those altered levels to the dosing pattern most likely to have produced them, given known kinetics of the tracers. The large number of possible dosing patterns over a month (268,435,456) magnified uncertainty in interpreting noise-altered levels. At best, with an assay variability of 5%, compliance could be estimated only within 4 to 8 doses per month. The tracer with a half-life of 30 days performed best. Pairing tracers did not improve uncertainty. The authors found that even a model neglecting many real-world variables suggests a limited role for tracers in measuring compliance.