Association of baseline poor attendance to scheduled clinic visits by primary care patients with diabetes and subsequent hospitalization

Patients with diabetes mellitus (DM) with poor attendance to scheduled clinic visits may be at increased risk of hospitalization and require proactive outreach to improve adherence to care. We examined this association in a primarily Hispanic cohort identified from a DM registry of a safety net health care system in San Antonio, TX. Of 14,784 patients in the DM registry with >2 primary care visits in the system and >2 hemoglobin A1c (HbA1c) tests from 1/2008 to 10/2012, we studied 8,854 patients with at least 2 clinic visits and at least one HbA1c >6.5% in a baseline interval (1/2008 to 10/2011) and at least one visit after 90 days in the follow-up year (10/2011 to 10/2012). Our outcome is any hospitalization in the follow-up year. Our primary predictor is % visit non-attendance calculated as: no show + canceled divided by all scheduled clinic visits (quartiles) in the baseline interval. Using logistic regression, we examined the association of % visit non-attendance with any follow-up hospitalization (yes/no) adjusting for: age, sex, race-ethnicity, months of follow-up after 10/2011 and baseline variables (insurance type, clinical comorbidities adapted from Elixhauser's measure, mean HbA1c [<8, 8-8.9, >9%], insulin vs. non-insulin therapy, N clinic visits). The study cohort had a mean age of 52 (SD=11), 60.4% women, 80.1% Hispanic, 6.7% Non-Hispanic black, 11.2% non-Hispanic white, and 2% other. Over half (51%) had county-funded financial assistance for care (CareLink), 32.1% Medicare, 10.6% Medicaid, and 6.3% private insurance. Median visit non-attendance was 31.5% (IQR=24.2%-39.4%) with an average of 22.4 scheduled clinic visits per patient (SD=15.2) in the baseline interval. 9.7% of subjects were hospitalized in the follow-up year. After adjustment, the odds of hospitalization increased with visit non-attendance. Compared with the 1st quartile (best group), the adjusted odds ratios (AOR) were: 2nd quartile AOR=1.36 (95% CI 1.08, 1.70), 3rd quartile AOR=1.51 (1.20, 1.89), and 4th quartile AOR=1.99 (1.59, 2.49). The AORs of hospitalization were significantly greater for: men, more comorbidities, non-Hispanic white vs. black, CareLink, and longer follow-up. In the only significant interaction, the odds of hospitalization was increased for patients with a baseline mean HbA1c <8% on insulin but not on oral medications. In this largely Hispanic cohort with DM, greater non-attendance to scheduled clinic visits in a 3-year baseline timeframe is significantly associated with hospitalization in a subsequent year. Strategies are needed to prevent hospitalizations by addressing barriers to attending scheduled care in this vulnerable Hispanic population.