Measuring the Productivity of Diabetes Treatment, 1997-2005
Posted: 14 Jun 2007
Date Written: June 2007
This research contributes to the literature on the value of health spending by estimating quality-adjusted price indices for treatment of patients with diabetes. We analyze rich data on patient outcomes and actual transaction prices for a sample of 735 patients identified in a large electronic diabetes management database (classified as type 1, type 2, or indeterminate type diabetes). To assure we capture all medical spending, we focus on employees or dependents of a large health care employer in the Midwestern United States (18 years or older) continuously enrolled in a self-insured health plan from 1997 to 2005 or to date of death.
Measures of patient outcomes include HbA1c, cardiovascular risks using the UK Prospective Diabetes Study equation, and mortality. To disentangle quality changes attributable to medical care from those associated with aging, we age-adjust the quality measures. Actual transactions prices represent direct medical spending for all medical care, including pharmaceuticals.
Because time since onset is a documented risk factor for complications, we structure the analysis so that we can document whether improved medical care has slowed the rate of progression of the disease. To minimize bias toward the sicker patients who receive more frequent tests, we define 3-year windows and use the last outcome measure for each window.
We find that patient outcomes have improved and spending has increased substantially over the nine year period. The percent of patients with HbA1c below 7, for example, increased across almost all diagnosis cohorts, so that by the 2003-2005 period, patients exhibiting this level of control ranged from 50% (of patients diagnosed before 1985) to 72% (of those diagnosed after 2000). Median annual spending increased 28% for those diagnosed before 1985, 76% for patients diagnosed between 1985 and 1996, and almost 4-fold for patients diagnosed during the 1997-99 period. Whether quality-adjusted prices have been increasing or not depends on the value of quality improvements. We estimate quality-adjusted price indices using techniques previously developed for treatment of heart attacks and depression (Cutler et al. 1998, Berndt et al. 2002), controlling for patient baseline severity, age and years since diagnosis.
Keywords: productivity, Diabetes, price indices, value for money, chronic disease management
JEL Classification: I10
Suggested Citation: Suggested Citation