Predictors of recall of over-the-counter and prescription non-steroidal anti-inflammatory drug exposure

Abstract

Purpose: Because of the difficulty in establishing a gold standard, data on accuracy of recall of over-the-counter (OTC) medication use are sparse.

Methods: We studied a cohort of 1889 persons living in the Philadelphia area to assess recall of non-aspirin non-steroidal anti-inflammatory drug (NANSAID) use during the preceding 8 weeks. Our analyses were based on the assumption that among the group of subjects, on average, the reported usage of NANSAIDs should not vary over the previous 8 weeks. To model the effect of time on reported usage while allowing for the inherent correlation of responses within subjects over time, we employed alternating logistic regression.

Results: We documented a significant decline in reported use of OTC NANSAIDs but not prescription NANSAIDs during the 8-week study period (p = 0.3 for frequent prescription NANSAIDs, p = 0.2 for infrequent prescription NANSAIDs, p < 0.001 for frequent OTC NANSAIDs, and p < 0.001 for infrequent OTC NANSAIDs). Reported rates of frequent and infrequent OTC NANSAID consumption declined from 6.3 to 4.6% and from 17.1 to 12.8% between the most recent week and eight weeks prior, respectively.

Conclusions: Interviews focusing on medications used on an as needed basis should be performed as close as possible to the index date. Likewise, data on frequent use of OTC NANSAIDs may be more reliable than that on infrequent use, particularly when subjects are asked to recall more than a few weeks back in time.