• Adherence in Exercise Meta-Analyses: Assessment and Effect on Study Outcomes

      Slack, Marion; Bae, Jeffrey; Kobleski, Robert; College of Pharmacy, The University of Arizona (The University of Arizona., 2005)
      Objective: The purpose of this study was to explore whether current meta-analyses on exercise interventions assess adherence and/or compliance of the studies included in the meta-analyses and to determine if subject adherence had any effect on outcomes of the analyses. Methods: Data was collected through a search of the MEDLINE database using the key words exercise, adherence, compliance, clinical trials, and meta-analysis. Data on study title, author, number of studies screened, number in meta-analysis, range of sample sizes, total number of subjects, primary intervention, primary outcome, how study quality was assessed, how adherence was assessed, whether adherence was used as a control variable, and did adherence affect the outcome was recorded on a paper and pencil data extraction form. Data was analyzed by constructing a table describing the meta-analyses and calculating the number and percent of analyses that included adherence. The table allowed for the evaluation of the strength and methodology of each piece of literature with respect to acknowledging adherence as a significant variable in the strength and legitimacy of each analysis. Results: Nineteen meta-analyses met our search criteria and were evaluated. Five of the nineteen meta-analyses (26 percent) described a method for assessing adherence. It was found that none of these used adherence as a control variable. Four of the nineteen meta-analyses did not assess the quality of the studies contained within the analysis. One of these meta-analyses suggested that adherence may have confounded outcomes, but did not provide any data to address their concerns. Conclusions: In meta-analyses, adherence is unlikely to be addressed. Current meta-analyses frequently lack methods for assessing adherence, and do not use adherence as a control variable. Whether adherence to exercise regimens affects outcomes cannot be determined from current meta-analyses.