"All-or-none" (bundled) process and outcome indicators of diabetes care

To evaluate processes and outcomes of diabetes care using bundled indicators from a primary care registry of osteopathic training programs.

Retrospective cohort analysis.

This study examined care delivered to 7333 patients across 95 family practice and internal medicine residency programs (July 1, 2005, through September 15, 2008) to determine diabetes care performance using measures of processes of care and outcomes. Two summary (bundled) reports of care for each measure were constructed. The first used the frequency of indicated care delivered (indicator-level bundle), and the second used the frequency of patients' receiving all indicated care (patient-level bundle).

Use of the indicator-level bundle demonstrated that outcomes goals were achieved at a rate of 44.5%. Use of the patient-level bundle demonstrated that outcomes goals were achieved at a rate of only 16.2%, a significant difference (P <.001). Eight evidence-based processes of diabetes care were then examined using the 2 bundling methods. The indicator-level analysis mean rate for the bundled processes of care was 77.3%, whereas the patient-level analysis mean rate was only 33.5%. This was also significantly different (P <.001).

The method of bundling care measures can have a profound effect on the reporting of goals achieved. This can in turn influence the assessment of provider performance and opportunity gaps in diabetes care delivery. In this study, providers were more likely to achieve processes-of-care goals when diabetes care was bundled at the indicator level than at the patient level. Standardization of summary reporting of diabetes care should be developed to enhance consistent interpretation of performance.