In search of a standard when analyzing medication adherence in patients with heart failure using claims data: a systematic review

Abstract

To determine and address medication non-adherence is important and may improve patient outcomes. Our aim was to perform a systematic review on the operationalization of adherence measures in patients with heart failure using claims data. We searched the MEDLINE, EMBASE, and Cochrane databases. A quality check of the full-text articles was performed using the Quality in Prognosis Studies tool. We included 28 studies. Eleven studies used the medication possession ratio, 15 the proportion of days covered, and six studies used own methods to calculate medication adherence. Almost all studies used an observation period of 1 year. Twenty-one studies considered angiotensin converting enzyme inhibitors or angiotensin receptor antagonists, 17 beta-blockers, 11 diuretics, eight mineralocorticoid receptor antagonists, six digitalis-glycosides, and six statins. Eight studies looked at single substance classes; 18 studies analyzed more than one drug class. Different assumptions for the dosage were used. In seven studies, switching within therapeutic classes was allowed. Three studies allowed stockpiling, nine studies censored hospital stays, and four studies censored death. Seventeen studies dichotomized the results, applying the cutoff ≥ 0.8 to define a patient as adherent. Eight studies analyzed adherence as a continuous value. The reporting quality of the methodology applied varied widely between studies. A gold standard is currently not available for the quantification of medication adherence in patients with chronic heart failure via claims data. To assess the methodology, the following parameters should be described: method of measurements, observation period, substances, dosing, switches, stockpiling, truncations, statistical analysis, and cutoff for adherence.

Pharmacy‐based interdisciplinary intervention for patients with chronic heart failure: results of the PHARM‐CHF randomized controlled trial

“…The effect of the intervention on the PDC as a continuous variable is quantitatively smaller than the effect on the proportion of patients with a PDC ≥ 80%. The 80% cut‐point for PDC is widespread practice and was used because it is associated with improved clinical outcomes for several medications/diseases including HF . Unfortunately, we were unable to calculate a PDC for 35 patients.…”

Adherence to heart failure management medications following cardiac resynchronization therapy

“…Although no gold standard exists for evaluating medication adherence, PDC, evaluated as both a continuous and binary outcome, is one of the most commonly accepted and used measures of adherence, as it more conservatively estimates adherence than other measures (i.e. medical possession ratio) 18,21 . The consistent increase in adherence observed following the index date suggests that either CRT or factors related to CRT, such as decreases in adverse symptom severity, contribute to significantly increasing adherence.…”

“…Second, adherence was measured according to the number of days of possible medication coverage, as reflected by days of supply in pharmacy claims, and thus may be overestimated; whether or not all medication supplied to the patient is actually used as directed is unknown. This adherence calculation did, however, use PDC, a widely accepted measurement and a more conservative estimate of adherence than alternatives (i.e., medical possession ratio), which should help mitigate this limitation 18,24 . Third, the study was limited to patients aged 18 to 64 years of age, as patients aged 65 years or older are eligible for prescription drug coverage through Medicare Part D plans.…”

“…Finally, the claims database that we used does not include information about contraindications or patients reasons for initiating and/or discontinuing certain therapies. Despite these limitations, the real-world approach used in this study of thousands of patients is appropriate for studying the impact of CRT implantation on patient medication adherence and has been applied and tested across multiple chronic medication classes (and regimens) to produce accurate assessments of medication adherence in patient populations 18 .…”

“…Pre-and post-index adherence was assessed for ACE-I, ARB, BB, and furosemide (the latter was used as a proxy for loopdiuretics as a whole, due to it being the most predominantly used agent in the class) 17 . Previous studies assessing adherence have demonstrated that evaluating both individual GDMT and medication regimens can improve the reliability of results in claims studies 18 . As a sensitivity analysis, adherence to GDMT-based HF medication regimens (ACE-I/ ARB/direct renin inhibitor [DRI] þ BB; ACE-I/ARB/DRI þ BB þ furosemide) was also assessed.…”