Markov Mixed Effects Modeling Using Electronic Adherence Monitoring Records Identifies Influential Covariates to HIV Preexposure Prophylaxis

Abstract

Adherence is a major factor in the effectiveness of preexposure prophylaxis (PrEP) for HIV prevention. Modeling patterns of adherence helps to identify influential covariates of different types of adherence as well as to enable clinical trial simulation so that appropriate interventions can be developed. We developed a Markov mixed-effects model to understand the covariates influencing adherence patterns to daily oral PrEP. Electronic adherence records (date and time of medication bottle cap opening) from the Partners PrEP ancillary adherence study with a total of 1147 subjects were used. This study included once-daily dosing regimens of placebo, oral tenofovir disoproxil fumarate (TDF), and TDF in combination with emtricitabine (FTC), administered to HIV-uninfected members of serodiscordant couples. One-coin and first- to third-order Markov models were fit to the data using NONMEM 7.2. Model selection criteria included objective function value (OFV), Akaike information criterion (AIC), visual predictive checks, and posterior predictive checks. Covariates were included based on forward addition (α = 0.05) and backward elimination (α = 0.001). Markov models better described the data than 1-coin models. A third-order Markov model gave the lowest OFV and AIC, but the simpler first-order model was used for covariate model building because no additional benefit on prediction of target measures was observed for higher-order models. Female sex and older age had a positive impact on adherence, whereas Sundays, sexual abstinence, and sex with a partner other than the study partner had a negative impact on adherence. Our findings suggest adherence interventions should consider the role of these factors.

Assessment of Demographic and Socio-Behavioral Factors on Adherence to HIV Pre-Exposure Prophylaxis Using a Markov Modeling Approach

“…The current report agrees with our prior report (Madrasi et al, 2017) on Markov modeling of MEMS based adherence data on the approach and covariate effects. There are two important differences between these two analyses.…”

“…The dependence of future state on the current state (persistence, execution) in the electronic adherence measurements was evident by the better fit of the Markov model compared with the logistic model. A similar observation was made in previous reports on the analysis of electronic adherence data (Girard et al, 1998; Madrasi et al, 2017). Some of the issues associated with ignoring the Markov element in the data when present include increased type I error rates on covariate inclusion, overestimation of information content in the data, and unrealistic simluation of individual time course of the outcomes (Silber et al, 2009).…”

“…For instance, participants older than 25 years and/or female sex had lower odds of transitioning from the dose-taking state to the dose-missing state and thus had higher persistence. Younger age has been reported to negatively influence PrEP adherence in other studies (Amico and Stirratt, 2014; Haberer et al, 2017a; Madrasi et al, 2017). This finding might be related to the level of maturity associated with age.…”

“…For generating this transition for the first day in an individual, the previous state was considered as 1, which signifies a “dose taken” state. The first dosing state was assumed to be 1 based on the high overall adherence; moreover, prior modeling showed limited impact of the initial state assumption on outcomes (Madrasi et al, 2017). Transitions [01] and [11] signify transition from a “dose-missing” state to a “dose-taken state” and staying in a “dose-taking” state, respectively.…”

“…Adherence can be quantified and modeled by deriving probability parameters for transition between the discrete states that explain an individual’s dose-taking pattern over a period. A Markov mixed-effects modeling approach has been implemented for analyzing adherence, describing drug holiday patterns, and identifying influential covariates of PrEP adherence from the MEMS-documented adherence data sets (Girard et al, 1998; Fellows et al, 2015; Madrasi et al, 2017). The Markov modeling approach has also been used to analyze data in diverse studies, such as those dealing with adverse effects of drugs, seizures counts, and patterns of sleep stages (Karlsson et al, 2000; Zingmark et al, 2005; Ito et al, 2008; Henin et al, 2009; Bisaso et al, 2015).…”