OBJECTIVES: Network meta-analysis (NMA) is recognized as a statistical technique able to provide a broader overview of the effects of all available comparators in one single model, accounting for both direct and indirect evidence. However, limit research using NMA to evaluate non-pharmacological complex interventions to improve medication adherence exist, and inconsistencies in the assessment of adherence are still common in the literature. We aimed to evaluate the effect of different measures of adherence in the comparative effectiveness of complex interventions to enhance patients’ adherence to prescribed medications.

METHODS: A systematic review with NMA was performed. Electronic searches were conducted in PubMed (November 2017). Studies of interventions that aimed to improve medication adherence in short-periods (until 3 months) were included. The evaluated measures of adherence were: self-report, pill count and MEMS (medication event monitoring system). To standardize the results obtained with these measures, an overall composite measure and an objective composite measure were also calculated. NMAs for each measure of adherence were built and the surface under the cumulative ranking curve analyses (SUCRA) were conducted.

RESULTS: Ninety-one studies were included in the NMAs. Results obtained for all measures of adherence were similar between them and to both composite measures. For both composite measures, interventions containing economic + technical components were the best option (90% of probability in SUCRA analysis) with statistical superiority against almost all the other interventions and standard care (odds ratio with 95% credibility interval ranging from 0.09 to 0.25 [0.02, 0.98]).

CONCLUSIONS: NMA showed to be a reliable technique to compare different measures of medication adherence of complex interventions in short-periods follow-up. Different measures produced similar results. The use of composite measures may be a reliable alternative to establish a broader picture of adherence.