2017 HSR&D/QUERI National Conference
4010 — Intermediate Diabetes Outcomes by Primary Care Provider Type: Differences among Patients of Physicians, Nurse Practitioners, and Physician Assistants
Lead/Presenter: George Jackson, COIN - Durham
All Authors: Jackson GL (Durham HSR&D Center of Innovation)
Smith VA (Durham HSR&D Center of Innovation)
Edelman D (Durham HSR&D Center of Innovation)
Woolson SL (Durham HSR&D Center of Innovation)
Hendrix CC (Durham HSR&D Center of Innovation)
Everett CM (Duke Physician Assistant Division)
Berkowitz TS (Durham HSR&D Center of Innovation)
White BS (Durham HSR&D Center of Innovation)
Morgan PA (Duke Physician Assistant Division)
Examine differences in intermediate diabetes outcomes among patients of physician, nurse practitioner (NP), or physician assistant (PA) primary care providers (PCPs).
The study included pharmaceutically-treated VA diabetes patients receiving primary care in 2012 and 2013. A patient's PCP was defined as the provider most often visited in the primary care clinic in 2012 and 2013. We examined the relationship between the profession of the PCP and both continuous level and dichotomous control of hemoglobin A1c (HbA1c), systolic blood pressure (SBP), and low-density lipoprotein cholesterol (LDL-C) based on the mean of outcome measurements in 2013. To reduce the potential for selection bias among professions, we utilized inverse probability of PCP type weighting (propensity score analysis) by gender, age, race, ethnicity, marital status, homelessness, co-pay status, mental health diagnoses, health status (diagnostic cost group [DCG] score), travel distance, availability of specialized diabetes services at the VA, rurality, and U.S. region. Hierarchical linear mixed models with random intercepts to account for clustering by facility and PCP were used to analyze continuous outcomes. Logistic regression models fit with generalized estimating equations and an exchangeable correlation structure and empirical standard errors, to account for clustering within facility, were fit to dichotomous control outcomes.
The sample included 368,481 patients from 568 facilities. PCPs were physicians, NPs, and PAs for 75.0%, 18.1%, and 6.9% of patients respectively. No clinically-significant differences (based on a priori thresholds) were observed between outcomes of patients for physicians, NPs, and PAs. Compared to physicians: 1) HbA1c differences were -0.05% (95%CI = -0.07:-0.03) for NPs and 0.01% (95%CI = -0.02:0.03) for PAs; 2) SBP differences were -0.27mmHg (95%CI = -0.46:-0.08) for NPs and -0.29mmHg (95%CI = -0.54:-0.03) for PAs; and 3) LDL-C differences were 1.04mg/dl (95%CI = 0.60:1.48) for NPs and 1.81mg/dl (95%CI = 1.22:2.41) for PAs. There were no clinically-significant differences in the dichotomized control of the three outcomes. Results held when examining only patients on insulin or with DCG scores > 2.0.
We observed no clinically-significant variation among key diabetes outcomes based on the profession of the patient's PCP.
VA has recently expanded NP scope of practice. Our study suggests that similar chronic illness outcomes may be achieved by physician, NP, and PA PCPs.