Impact of Primary Care Usual Provider Type and Provider Interdependence on Outcomes for Patients with Diabetes

Research Objective Quality of diabetes care delivered to patients with different types of usual providers of care [i.e., physician, physician assistant (PA) or nurse practitioner (NP)] is similar. However, primary care (PC) providers often provide care to each other's patients (i.e., “share” common patients). The impact of patient sharing, or interdependence, is on quality of diabetes care is unknown. As a result, some providers and organizations hesitate to formalize patient sharing by creating multi‐provider teams due to concerns about the impact of impact of such teams on quality of care. We sought to both 1) evaluate the association of usual provider type (physician or PA/NP) provider and 2) interdependence on outcomes for patients with diabetes. Study Design This patient‐level cohort study used electronic health record data from 24 health system‐affiliated PC practices in central North Carolina. Patients' usual PC provider was the provider most frequently seen during 2016 and 2017. Patient‐level independent variables included demographic, medical complexity, and healthcare utilization (separate variables for PC, specialty, emergency department, and hospital). Provider panel‐level variables [usual provider of care type (physician or PA/NP), panel size, and provider interdependence (# shared patients / # supplemental providers then categorized into quartiles)]. We examined the association of diabetes quality (at least two hemoglobin A1c (HbA1c) tests, at least one low‐density lipoprotein (LDL) cholesterol test, mean HbA1c and LDL values) during 2017 with all variables simultaneously using logistic or linear regression with clustering by practice. Population Studied Adults with diabetes (N = 10,498) on 131 panels (physician = 111; PA/NP = 20). Principal Findings Ninety percent of patients had physicians as usual providers (N = 9462). Patient demographics, complexity and utilization were similar for patients of different usual provider types except for mean age (physician = 64.6; PA/NP = 59.7) and insurance type (% Medicaid/uninsured: physician: 5.5; PA/NP: 11.2). Most patients had at least two HbA1c tests (72%) and one LDL test (65%). Average HbA1c (7.5 mmHg) and LDL (109 mg/dL) was also similar by usual provider type. Average panel size was 80 diabetes patients (physician = 85; PA/NP = 52) Panels had a mean interdependence of 6.1 patients/supplemental provider (physician = 6.2; PA/NP = 5.8). There were no statistically significant differences in HbA1c or LDL testing by usual provider type or interdependence. Similarly, there was no statistically significant difference in HbA1c for mean HbA1c values by usual provider type or interdependence. However, increases in interdependence quartile resulted in increases in mean LDL values (β = 13.8, p = 0.016). Conclusions The quality of diabetes care does not differ based on provider type, either PA/NPs or physicians. Increases in provider interdependence (greater numbers of patients per provider) resulted in higher mean LDL values, but only the interdependence values in the highest quartile would result in a mean LDL in the borderline high range. Implications for Policy or Practice Organizations using or considering interdependent, multiple‐provider teams will likely not see a reduction in diabetes‐specific quality metrics. However, additional patient and provider outcomes should be evaluated, such as delivery of preventive services, outcomes important to patients with multiple chronic conditions, and provider satisfaction and burn‐out. Primary Funding Source National Institutes of Health.