IS TELEHEALTH EFFECTIVE FOR DIABETES CARE FOR THE UNDERSERVED, MINORITY AND COMPLEX PATIENTS

ABSTRACT

BACKGROUND: The rapid conversion to telehealth services as an alternative to in-person ambulatory care in response to COVID-19 required abrupt adaptations by patients with diabetes and their providers, that may have resulted in poorer outcomes for subgroups of the population. METHODS: We conducted a longitudinal population study from a diabetes registry with clinical and administrative data maintained for all patients with diabetes seen at an academic medical center. From this registry, we identified all patients seen at least once in the year before and after 03/20/20 at any of the 16 ambulatory care clinics at this site (n=9760) who also had ≧ 1 HbA1c value in both periods (n=4710), and those with ≧ 2 visits and ≧ 2 HbA1c values in the same periods (n=1553). We compared patient characteristics (age, gender, race/ethnicity, Charlson comorbidity score), clinic site [Federally Qualified Health Centers (FQHCs) vs. other ambulatory care sites], total number of ambulatory visits and number of telehealth visits, mean HbA1c mean value and % in poor control (HbA1c ≧ 9%) for both groups of patients. We used odds ratios for bivariate comparisons and logistic regression for multivariable analyses. RESULTS: The mean age of patients with ≧ 1 visit in the pre-post periods was 62.5 [SD 14.0], 47% were female, 40% were Hispanic, 28% had a Charlson score greater than the median, 37% were seen at an FQHC, and 18.9% had poor glycemic control (HbA1c ≧9%). Characteristics for patients with ≧ 2 visits were comparable. Poor control was more likely among those seen at FQHC sites (OR=3.17, p<.0002), those ≧65 years (OR=3.53, p<.0001), those with substantial comorbidity (Charlson ≧ median value, OR=1.40, p=.0011), Hispanic patients (OR=3.08, p<.0001) and those seen by telehealth (OR=1.59, p<.0001). Results for patients with ≧ 2 visits and corresponding HbA1c values were comparable. Parameter estimates from the logistic regression model predicting HbA1c ≧ 9% were all statistically significant and in the expected direction for the variables considered. CONCLUSIONS: Telemedicine is currently being considered for continuation as an accepted, efficient and safe mode of healthcare delivery. However, it may not be effective for specific subgroups of patients with chronic diseases such as diabetes in which patient partnership and the provider patient relationship are key to optimizing outcomes. Further, advances in the delivery of telehealth care, including easily accessed high quality technologies are needed to ensure that remote healthcare delivery does not further increase disparities in health outcomes, particularly for the poor, underserved, minorities, elderly and those with complex diseases.