ORIGINAL ARTICLE Association of Health Plans’ Healthcare Effectiveness Data and Information Set (HEDIS) Performance With Outcomes of Enrollees With Diabetes Jeffrey S. Harman, PhD,* Sarah Hudson Scholle, DrPH,† Judy H. Ng, PhD,† L. Gregory Pawlson, MD, MPH,† Russell E. Mardon, PhD,‡ Samuel C. (Chris) Haffer, PhD,§ Sarah Shih, MPH,¶ and Arlene S. Bierman, MD, MS Background: Few quality of care evaluations examine the relation- ship between clinical processes and patient outcomes. Objective: To determine the association between health plan per- formance on Healthcare Effectiveness Data and Information Set (HEDIS) clinical processes and intermediate outcome measures and Health Outcomes Survey (HOS) self-reported physical and mental health scores among Medicare plan enrollees with diabetes. Research Design: Secondary data analysis of 2002 HEDIS and 2001–2003 HOS data. Subjects: This study focused on Medicare plan enrollees with self-reported diabetes (N = 8184). Measures: Plan-level HEDIS diabetes care measures for 2002 and longitudinal, patient-level 2001–2003 HOS physical and mental health outcomes scores. Hierarchical linear models estimated the relationship between plan HEDIS performance on diabetes process of care and intermediate outcome measures and 2-year changes in enrollee HOS physical and mental health scores. Results: Each 10% point improvement in plan performance on HEDIS intermediate outcomes (ie, the proportion of well-controlled diabetes) was related to significant positive increase in the proba- bility of being healthy as measured by both enrollee physical health scores (7 percentage point increase, P  0.05) and mental health scores (11 percentage point increase, P  0.01). Similar increases in plan process of care measures were associated with increases in the probability of being healthy as measured by enrollee mental health scores (11 percentage point increase, P  0.001). Conclusions: This study represents one of the first attempts to link plan HEDIS performance to changes in enrollee health. The results suggest that improved quality of care, as measured by process and intermediate outcomes measures for diabetes, can result in better health among patients with diabetes. Further research should address whether this relationship exists in other quality measures, clinical conditions, and populations. Key Words: Medicare, aged, managed care programs, outcome and process assessment (health care), hierarchical linear models (Med Care 2010;48: 217–223) D iabetes is one of the most costly and prevalent chronic conditions in the Medicare population, prompting con- tinued efforts to evaluate quality of diabetes care. 1,2 Although measurement of quality in diabetes care is relatively ad- vanced, there are still limited data that directly link clinical process of care measures (eg, screening for hemoglobin A1c HbA1c in patients with diabetes) or intermediate outcomes (eg, control of HbA1c levels in patients with diabetes) with indicators of the end results of healthcare, such as functional outcomes or health status. There is a special need for research that relates clinical processes to patient outcomes in the Medicare population, where management of chronic conditions may have an espe- cially critical effect on quality of life. Although global health outcomes, such as mortality and functional status, are con- sidered to be the most critical indicators of health care quality, 3 these outcomes are often difficult to link to the actions of health providers. 4,5 Thus, in most quality evalua- tions, measures of process or intermediate outcomes are used. 6 However, to be considered valid measures of quality, measures of process or intermediate outcomes must be clearly linked to global health outcomes, which are more meaningful from the patient perspective. 7–9 A prior study of patient out- comes and health plan performance data found positive corre- lations between clinical measures and member health status, but did not consider longitudinal outcomes or health differ- ences in plan populations. 10 This study tests a direct link between clinical process measures and patient health status outcomes using data from From the *Department of Health Services Research, Management and Policy, College of Public Health and Health Professions, University of Florida, Gainesville, FL; †National Committee for Quality Assurance, Washington, DC; ‡Westat, Rockville, MD; §Centers for Medicare and Medicaid Services, Baltimore, MD; ¶The New York City Department of Health, New York, NY; and Li Ka Shing Knowledge Institute, Univer- sity of Toronto, Toronto, Ontario, Canada. Supported by the Centers for Medicare and Medicaid Services, US Depart- ment of Health and Human Services. The views expressed in this article are the authors’ and do not necessarily represent the views of the US Department of Health and Human Services, or Centers for Medicare & Medicaid Services. Reprints: Sarah Hudson Scholle, MPH, DrPH, National Committee for Quality Assurance, 1100 13th Street NW, Suite 1000, Washington, DC 20005. E-mail: scholle@ncqa.org. Copyright © 2010 by Lippincott Williams & Wilkins ISSN: 0025-7079/10/4803-0217 Medical Care • Volume 48, Number 3, March 2010 www.lww-medicalcare.com | 217