Background: Medication non-adherence is common among renal dialysis patients. High degrees of non-adherence in randomized controlled trials (RCTs) can lead to failure to detect a true treatment effect. Cardio-protective pharmacological interventions have shown no consistent benefit in RCTs involving dialysis patients. Whether non-adherence contributes to this lack of efficacy is unknown. We aimed to investigate how medication adherence and drug discontinuation were assessed, reported and addressed in RCTs, evaluating cardiovascular or mortality outcomes in dialysis patients.
Methods: Electronic database searches were performed in MEDLINE, EMBASE & Cochrane CENTRAL for RCTs published between 2005-2015, evaluating self-administered medications, in adult dialysis patients, which reported clinical cardiovascular or mortality endpoints, as primary or secondary outcomes. Study characteristics, outcomes, methods of measuring and reporting adherence, and data on study drug discontinuation were analyzed.
Results: Of the 642 RCTs in dialysis patients, 22 trials (12 placebo controlled), which included 19,322 patients, were eligible. The trialed pharmacological interventions included anti-hypertensives, phosphate binders, lipid-lowering therapy, cardio-vascular medications, homocysteine lowering therapy, fish oil and calcimimetics. Medication adherence was reported in five trials with a mean of 81% (range: 65-92%) in the intervention arm and 84.5% (range: 82-87%) in the control arm. All the trials that reported adherence yielded negative study outcomes for the intervention. Study-drug discontinuation was reported in 21 trials (mean 33.2%; 95% CI, 22.0 to 44.5, in intervention and 28.8%; 95% CI, 16.8 to 40.8, in control). Trials with more than 20% study drug discontinuation, more often yielded negative study outcomes (p = 0.018). Non-adherence was included as a contributor to drug discontinuation in some studies, but the causes of discontinuation were not reported consistently between studies, and non-adherence was listed under different categories, thereby potentiating the misclassification of adherence.
Conclusions: Reporting of medication adherence and study-drug discontinuation in RCTs investigating cardiovascular or mortality endpoints in dialysis patients are inconsistent, making it difficult to compare studies and evaluate their impact on outcomes. Recommendations for consistent reporting of non-adherence and causes of drug discontinuation in RCTs will therefore help to assess their impact on clinical outcomes.