Abstract
Background and Aims: Despite the proliferation of care guidelines, widespread practice variation is common. Longitudinal electronic health records (EHR) provide an opportunity to observe how care variations are related to care outcomes. This study uses blood pressure management to illustrate the uses of the Disease Management Index (DMI) to relate care quality to cardiovascular outcomes.
Methods: The Blood Pressure Disease Management Index (BP-DMI) incorporates the level of control of BP and the time out of control into a single index scored from 0 to 100. The practice level DMI for a given year is the average DMI for all diagnosed hypertensive patients in the practice. We observed practice level BP-DMI scores for 199 to 374 primary care physician (PCP) practices for the years 1998–2007 (Oregon) and 2005–2007 (Hawaii) and related those scores to annual incidence rates of myocardial infarction (MI), stroke, and congestive heart failure (CHF) hospital days.
Results: Systolic BP (SBP) control was much worse than diastolic BP (DBP) control, particularly before 2005, but improved steadily from that point on. Practice variations in BP control were substantial. Propensity scores were used to adjust for the substantial differences in patient risk across practices. On average, clinicians treating elevated DBP had patients who were younger and healthier than those treating elevated SBP. Propensity adjusted incidence rates of MI, stroke, and CHF hospital days were strongly associated with both systolic (S) and diastolic (D) BP-DMI scores (p <.0001 to <.02). The estimated effect of increasing the SBP and DBP DMI scores from the 50th percentile to the 60th percentile of clinical practices for MI and stroke incidence, and CHF hospitalization was:
SBP: −7.2%, −4.6%, −8.4%, respectively
DBP: −17.6%, −20.1%, −20.0%, respectively.
Conclusions: Practice level BP-DMI scores over a ten-year period are strongly related to the incidence of cardiovascular events including hospital days for CHF. The DMI is an effective tool for relating practice patterns to morbidity risk and health care utilization.
- Received May 27, 2010.
- Accepted May 27, 2010.
