Estimation of adherence to medication treatment in presence of censoring

Abstract

The purpose of this study is to provide a theoretical framework for the analysis of medication adherence based on longitudinal data from electronic medication monitors and to suggest methods for unbiased estimation of the effect of time and covariates on adherence.Methods: After defining the statistical summaries involved in adherence analyses and the assumptions necessary for their estimation, we address the issue of bias encountered when adherence is estimated on censored data. We suggest 2 unbiased methods to estimate adherence: (i) indirect combining implementation and persistence; and (ii) based on weights, allowing estimation of the effect of time and covariates on adherence via generalized estimating equations models. Results:We applied the proposed methods to investigate the effect of sex on adherence in a sample of 43 oncology patients followed 1 year. Implementation was higher for men than for women at baseline (98.8 vs. 97.5%, odds ratio [OR] 2.08, 95% confidence interval [CI]: 1.00-4.35), whereas the relationship was reversed at 1 year (94.5 vs. 96.4%, OR 0.65, 95%CI: 0.28-1.52). Adherence declined faster in men, with yearend values of 46.3% for men and 92.2% for women (OR 0.07, 95%CI: 0.02-0.26).Conclusion: Estimation of adherence is a complex statistical issue with longitudinal and duration data, possibly censored, interleaving. This study provides a theoretical framework and suggests methods for unbiased estimation of adherence as a function of time and covariates. This allows the effect of an intervention to be estimated in clinical trials, and helps healthcare providers reframe adherence programmes to address covariates such as sex.

The differential impact of a 6-versus 12-month pharmacist-led interprofessional medication adherence program on medication adherence in patients with diabetic kidney disease: the randomized PANDIA-IRIS study

“…The probability of an optimal implementation could be determined by the ratio of treatments taken optimally to the total number of monitored treatments (i.e., on day x, if patients have an optimal implementation to 4/5 of their medications, the probability of an optimal implementation at day x would be 80%). Our research raises the point that the operational definitions of implementation should be evaluated further in polypharmacy patients, as well as the statistical methodology ( Pasquier et al, 2022 ).…”

“…Our definition of treatment implementation states that patients need to implement optimally all their medications monitored on day x in order to have an overall optimal medication implementation at day x. This definition has been commonly used in previous research studies ( Schneider et al, 2019 ; Bandiera et al, 2022b ; Pasquier et al, 2022 ). However, patients with DKD are polypharmacy, and patients included in the PANDIA-IRIS study often used more than one EM.…”

“…We used electronic monitoring for 24 months, which is considered the most robust methodology to objectively and longitudinally measure medication implementation over time providing an adherence history. The statistical analysis procedures used on repeated adherence electronic monitoring measures were previously developed and validated ( Schneider et al, 2019 ; Pasquier et al, 2022 ), and the analysis of the implementation to the different drug classes allowed determining the differential effect of the IMAP on implementation to antidiabetics, antihypertensive drugs, and statins.…”

Medication adherence research comes of age

“…Pasquier et al 11 also address the methodological challenges around adherence measurement. The authors note that the analysis of data from electronic monitoring systems, while perhaps seen as the Gold Standard because of the granular data on medication taking, can be challenging from a methodological perspective.…”

Using a Semiautomated Procedure (CleanADHdata.R Script) to Clean Electronic Adherence Monitoring Data: Tutorial

“…Digital health innovations and technologies can be used to contribute and facilitate interventions aiming at supporting medication adherence [ 14 ]. In parallel, the quality of adherence data analysis has increased due to guidelines on operational adherence definitions [ 15 ], advanced statistical analysis [ 16 , 17 ], and semiautomated procedures to analyze adherence in health care databases (eg, the Adhere-R package [ 18 ]). However, to date, the research community lacks guidance on the methodology to collect, manage, clean, and analyze adherence electronic monitoring database, whereas it is known that heterogeneous data management impacts adherence outcomes [ 19 ].…”