Potentially preventable hospitalizations in medicare patients with diabetes: A comparison of primary care provided by nurse practitioners versus physicians

Background: Few comparisons exist of the quality of primary care provided by nurse practitioners (NPs) versus physicians. Methods: Patients with a diagnosis of diabetes in 2007-2010 (n = 345,819) who received all primary care from NPs or from generalist physicians in a given year were selected from a national sample of Medicare beneficiaries. We compared the rate of potentially preventable hospitalizations among patients who received primary care from NPs versus generalist physicians. Various statistical methods-including multivariable analysis, inverse probability weighting of propensity score, nonpooling propensity score adjustment and matching, and instrumental variable (IV) analysis-were used to control for differences in patient characteristics between the 2 groups. Results: Patients who received all of their primary care from NPs or from physicians differed by age, sex, race/ethnicity, socioeconomic status, residential area, and number of provider visits in the previous year. Nonpooling propensity score matching substantially reduced the differences, but neither IV approach satisfactorily reduced the differences. In multivariable analyses, receipt of primary care from an NP was associated with a decreased risk of hospitalization for potentially preventable conditions (OR: 0.90; 95% CI, 0.87-0.93). Similar results were found using conditional logistic regression models with propensity methods. We found smaller reductions in our analyses of "other hospitalizations" (OR: 0.96; 95% CI, 0.95-0.98). Both IV analyses showed associations between NP care and lower potentially preventable hospitalizations, but only 1 result was statistically significant. Conclusions: Using potentially preventable hospitalizations as a quality indicator, primary care provided by NPs was at least comparable with that provided by generalist physicians.