Diabetes, a leading cause of morbidity and mortality, is projected to increase dramatically over the coming decades with the largest percent increase among those older than 75 years. Current diabetes guidelines recognize that treatment recommendations regarding tight control of glycosylated hemoglobin (A1c) and cardiovascular disease (CVD) risk factors may not be appropriate for older patients and those with comorbid conditions due to increased risk for adverse events. These increased risks have even more significance if the benefits of tight control are limited due to shortened life expectancy. If the long-term benefits are not relevant and the short-term risks are significant, there may be little reason to pursue tight control of A1c and CVD risk factors such as elevated low-density lipoprotein (LDL) and blood pressure (BP). To address this, guidelines recommend individualizing treatment for patients over age 65. Unfortunately, little evidence exists to support treatment decisions for older patients with diabetes generally and for those with comorbid conditions. If the outcomes of tight control were known for these patients, existing guidelines would be significantly strengthened. An evidence base is particularly important as treatment guidelines are increasingly used to develop publicly-reported quality metrics and guide pay-for-performance (P4P) efforts. Given these public and financial applications of guidelines, it is critical to determine whether the definitions used to identify patients with diabetes and assign them to a provider influence the conclusions of effectiveness studies. Unless new information is obtained to modify guidelines appropriately, financial incentives to adhere tightly to recommendations could have perverse effects that might actually diminish the quality of care.
Our specific aims are to: (1) examine the relationship between patients'control levels for A1c, LDL, and BP and short-term negative health outcomes (ER visits, hospitalizations, and death), and to determine whether this relationship depends on (2) the presence of comorbid congestive heart failure and chronic kidney disease, and (3) different definitions for assigning patients to providers. Our sample includes approximately 3,559 Medicare fee-for-service patients with diabetes who were cared for by a large midwestern multi-specialty physician group during 2003-2004. We link clinical results data (e.g., A1c, LDL, BP values) from the electronic health record to Medicare administrative data from 2003-2004. Our analytic approach, marginal structural modeling, takes advantage of the longitudinal nature of our data. Overall, the results from our investigation will have important implications for diabetes treatment guidelines, development of quality metrics, construction of pay-for-performance thresholds, and targeting of QI interventions. Narrative Prioritization of care is critical for patients with diabetes and comorbid conditions, because the short-term risks of tight adherence to diabetes treatment guidelines may outweigh the long-term benefits. We propose to examine the tight adherence to diabetes treatment guidelines on short-term negative health outcomes for Medicare fee-for-service patients with diabetes, including those who have two common and serious conditions chronic kidney disease and congestive heart failure. The results from our investigation will have important implications for diabetes treatment guidelines, development of quality metrics and pay-for-performance efforts, and targeting of quality improvement interventions.
