Practice Context Affects Efforts to Improve Diabetes Care for Primary Care Patients: A Pragmatic Cluster Randomized Trial

BackgroundEfforts to improve primary care diabetes management have assessed strategies across heterogeneous groups of patients and practices. However, there is substantial variability in how well practices implement interventions and achieve desired outcomes.ObjectiveTo examine practice contextual features that moderate intervention effectiveness.DesignSecondary analysis of data from a cluster randomized trial of three approaches for implementing the Chronic Care Model to improve diabetes care.ParticipantsForty small to mid-sized primary care practices participated, with 522 clinician and staff member surveys. Outcomes were assessed for 822 established patients with a diagnosis of type 2 diabetes who had at least one visit to the practice in the 18 months following enrollment.Main measuresThe primary outcome was a composite measure of diabetes process of care, ascertained by chart audit, regarding nine quality measures from the American Diabetes Association Physician Recognition Program: HgA1c, foot exam, blood pressure, dilated eye exam, cholesterol, nephropathy screen, flu shot, nutrition counseling, and self-management support. Data from practices included structural and demographic characteristics and Practice Culture Assessment survey subscales (Change Culture, Work Culture, Chaos).Key resultsAcross the three implementation approaches, demographic/structural characteristics (rural vs. urban + .70(p = .006), +2.44(p < .001), -.75(p = .004)); Medicaid: < 20 % vs. ≥ 20 % (-.20(p = .48), +.75 (p = .08), +.60(p = .02)); practice size: < 4 clinicians vs. ≥ 4 clinicians (+.56(p = .02), +1.96(p < .001), +.02(p = .91)); practice Change Culture (high vs. low: -.86(p = .048), +1.71(p = .005), +.34(p = .22)), Work Culture (high vs. low: -.67(p = .18), +2.41(p < .001), +.67(p = .005)) and variability in practice Change Culture (high vs. low: -.24(p = .006), -.20(p = .0771), -.44(p = .0019) and Work Culture (high vs. low: +.56(p = .3160), -1.0(p = .008), -.25 (p = .0216) were associated with trajectories of change in diabetes process of care, either directly or differentially by study arm.ConclusionsThis study supports the need for broader use of methodological approaches to better examine contextual effects on implementation and effectiveness of quality improvement interventions in primary care settings.