Primary Care Health Care Use for Patients With Type 2 Diabetes During the COVID-19 Pandemic

Abstract

The coronavirus disease 2019 (COVID19) pandemic may have delayed care for chronic disease (1). We sought to examine factors associated with total and virtual health care use for primary care visits for patients with type 2 diabetes (T2D) during the pandemic. Using our electronic medical record, we identified patients in the Cleveland Clinic Health System with a diagnosis of T2D with a prepandemic primary care visit between 1 August 2019 and 14 March 2020 (period 0), with follow-up data collected during two pandemic periods: period 1, from 15 March 2020 to 30 June 2020, when in-person visits were rapidly converted to virtual (telephone or video); and period 2, from 1 July 2020 to 15 November 2020, when in-person visits resumed. We obtained demographic characteristics including age, sex, race, insurance type, median income estimated by zip code based on the American Community Survey 2014–2018 5-year estimates (2), and baseline glycosylated hemoglobin (HbA1c). Patient characteristics were summarized with appropriate descriptive statistics. We assessed factors associated with completing any follow-up visits using logistic regression and then assessed factors associated with completing virtual visits using a mixed-effects logistic regression model with a random intercept at the patient level. All statistical analyses were performed in R (cran.r-project.org), and statistical significance was established at two-sided P values <0.05. There were 76,015 patients with T2D who completed a primary care visit in baseline period 0. The median age was 66.2 years (SD 13.2), 50.7% were women, 21.7% were Black, 71.0% were White, and 7.4% were other race, with insurance distribution 43.2% for private, 46.5% Medicare, 9.5% Medicaid, and 0.8% other insurance, median income $59,000 (SD $22,000 [3% missing]), and baseline HbA1c categories of #7% (53 mmol/mol) (59.6% of patients), >7–8% (>53–64 mmol/mol) (19.5%), >8–9% (>64–75 mmol/mol) (9.3%), and >9% (>75 mmol/mol) (11.3%), with 0.3% missing. The total number of primary care visits for this patient population was 191,928 in period 0 (3.5% virtual), 56,524 in period 1 (67.4% virtual), and 80,483 in period 2 (22.2% virtual). In the model adjusted for age, sex, race, median income (per $10,000), insurance type, and HbA1c category (#7% [53 mmol/mol], >7–8% [53–64 mmol/mol], >8–9% [>64–75 mmol/mol], and >9% [>75 mmol/mol]), we found higher odds of any visit completion with increasing age per 1 year (odds ratio 1.008, 95% CI 1.007–1.010, period 1; 1.009, 1.008–1.010, period 2) and for Black patients (1.288, 1.238–1.339, period 1; 1.087, 1.043–1.133, period 2) as well as for those with Medicare and Medicaid insurance. Men had lower odds of a visit for both periods (0.871, 0.846–0.897, period 1; 0.848, 0.823–0.875, period 2). There were slightly lower odds of any visit completion with increasing median income per $10,000 (0.964, 0.957–0.971, period 1; 0.981, 0.973–0.988, period 2). Compared with patients with a baseline HbA1c #7% (53 mmol/mol), patients in all HbA1c categories >7% (53 mmol/mol) had higher odds of a visit in period 1, while patients with a baseline HbA1c >9% (75 mmol/mol) had lower odds of a visit in period 2 (0.825, 0.785–0.867). The model predicting subsequent completion of the subset of virtual visits (Table 1) shows lower odds of completing follow-up virtual visits with increasing age and for men. Black patients had