Projektname
Meta-Analysis of Interventions to Improve Medication Adherence

Forschungsleitung
Vicki Conn

Forschungsteam
Sabina De Geest; Adam Hafdahl; Cindy Russell; David Mehr; Diane Johnson; Ginny Pepper; Jacqueline Dunbar-Jacob; Todd Ruppar

Zusammenarbeit mit
National Institute of Health, University of Missouri, USA

Laufzeit
2010 bis 2013

Projektbeschreibung
Inadequate medication adherence (MA) contributes to increased morbidity, mortality, patient and provider frustration, and health care costs. Alarming persistent nonadherence is consistently documented in diverse samples. Scientific evidence about the efficacy of interventions is essential to develop MA interventions that improve health outcomes and reduce costs. Research testing many MA interventions has yielded results that are sometimes conflicting and often unclear. These primary studies have not been quantitatively synthesized, which seriously impedes ­progress in both practice and research. This project’s purpose is to integrate scientific knowledge about interventions designed to increase MA. The project addresses these specific aims: 1) Determine the strength of ­
the research base about interventions to increase MA. 2) Specify and quantify the effect of interventions on MA. 3) Distinguish factors (e.g., intervention characteristics such as dose reduction or packaging, participant ­attributes such as diagnoses or ethnicity) that moderate the effect of interventions to increase MA.
This research team has used the proposed methods in several syntheses, including a recent preliminary meta-analysis of MA interventions tested in randomized trials with older adults. An extensive and rigorous literature search will avoid the bias that typical limited searches can cause. Strategies include searches by computer ­
and journal searches by hand, searches of ancestry lists and registries databases, reviews of graduate projects, examinations of conference/association abstracts, and contacts with senior authors on retrieved studies and principal investigators of NIH-funded studies. Independent data extractors will reliably code for intervention, methodological, and participant attributes. Analysis plans include: d-index to standardize the magnitude ­
of effect, sample size weighted calculations, random-­effects models, homogeneity (Q) assessment, publication bias analysis (I2), and to facilitate interpretation – Common Language Effect Size and conversion to original metric. Moderator analyses using meta-analysis analogues of regression and ANOVA will reveal which intervention characteristics (e.g., MA feedback, prompts) are associated with larger increases in MA. Minority and gender differences in intervention effectiveness will be examined as well. Provisional multivariate moderator analyses using meta-regression with selected subsets
of moderators will be conducted. The major impact of the project will be derived from the moderator analyses which will determine which intervention characteristics are linked with the best MA outcomes. Findings will improve public health by synthesizing diverse results in order to aid the design of interventions that help people increase their MA and achieve therapeutic goals. Study findings will have a major immediate impact on science by identifying unanswered research questions as well as areas in which the science has been settled.