Measuring intake of a prescribed medication A bottle count and a tracer technique compared

Abstract

The adequacy of bottle or pill counts for assessing patients' intake of medication has not been established. A bottle count was evaluated in this 2 year follow‐up of 105 patients with peptic ulcer. We added a trace of sodium bromide to a liquid antacid which enabled us to measure intake. There were 1,445 counts of empty bottles and 1,364 blood bromide level determinations. For 10 patients, the bottle counts did not appear to provide an accurate measure of medicine intake. These patients consistently turned in more empty bottles than iustified by their blood bromide levels. For the group as a whole, individual observations (covering time periods ranging from 1 to 3 months) showed only moderate correspondence between bottle counts and their matched bromide levels (r = 0.51). The correspondence was particularly poor when 7 or more bottles were "missing" (r = 0.22) or when the period of observation exceeded 35 days. Despite the inclusion of some erraneous observations, however, when a 12 or 24 month series of observations was pooled for each patient, the bottle count appeared to be moderately accurate (r = 0.80). A correlation of this magnitude stillleaves an errar of up to 36 per cent (i.e., 1 to 0.802 ), and the errar might be even larger if it were not possible to have the rigorous controls used in this study.