Using meta-analytic procedures to synthesize changes in patient-centered outcomes after medication adherence interventions.
Strategies to locate studies included online searches of 13 databases and 19 research registries, hand searches of 57 journals, and author and ancestry searches of all eligible studies. Search terms included patient compliance, medication adherence, and related terms. Searches were conducted for all studies published since 1960. Eligible published or unpublished primary studies tested medication adherence interventions and reported medication knowledge, quality of life, physical function, and symptom outcomes. Primary study attributes and outcome data were reliably coded. Overall standardized mean differences (SMDs) were analyzed using random-effects models. Dichotomous and continuous moderator analyses and funnel plots were used to explore risks of bias.
Thorough searching located 141 eligible reports. The reports included 176 eligible comparisons between treatment and control subjects across 23,318 subjects. Synthesis across all comparisons yielded statistically significant SMDs for medication knowledge (d = 0.449), quality of life (d = 0.127), physical function (d = 0.142), and symptoms (d = 0.182). The overall SMDs for studies focusing on subsamples of patients with specific illnesses were more modest but also statistically significant. Of specific symptoms analyzed (depression, anxiety, pain, energy/vitality, cardiovascular, and respiratory), only anxiety failed to show a significant improvement after medication adherence interventions. Most SMDs were significantly heterogeneous, and risk of bias analyses suggested links between study quality and SMDs.
Modest but significant improvements in patient-centered outcomes were observed after medication adherence interventions.