The purpose of this paper is to describe electronic monitoring device (EMD) (e.g., MEMS caps) use among HIV-infected adults enrolled in a randomized clinical trial and to make explicit some of the benefits and caveats of using electronic monitoring device technology. This is a descriptive, exploratory study of EMD use among 128 HIV-infected adults treated with at least three antiretroviral agents. Thirty-six percent of the sample admitted that they did not use the EMD consistently. Forty-one percent of the subjects reported taking out more than one dose at a time and 26% reported opening the EMD but not taking the medication. Special subject-related issues accounted for only a small percentage of all reported problems with EMD use (e.g., transient housing, incarceration, substance abuse relapse and drug treatment). Results of this study suggest that EMDs may underestimate antiretroviral adherence among HIV-infected adults. Recommendations for improving EMD data quality are presented.

An adaptive approach to Poisson regression modelling is presented for analysing event data from electronic devices monitoring medication-taking. The emphasis is on applying this approach to data for individual subjects although it also applies to data for multiple subjects. This approach provides for visualization of adherence patterns as well as for objective comparison of actual device use with prescribed medication-taking. Example analyses are presented using data on openings of electronic pill bottle caps monitoring adherence of subjects with HIV undergoing highly active antiretroviral therapies. The modelling approach consists of partitioning the observation period, computing grouped event counts/rates for intervals in this partition, and modelling these event counts/rates in terms of elapsed time after entry into the study using Poisson regression. These models are based on adaptively selected sets of power transforms of elapsed time determined by rule-based heuristic search through arbitrary sets of parametric models, thereby effectively generating a smooth non-parametric regression fit to the data. Models are compared using k-fold likelihood cross-validation.