Factors associated with loss to follow-up after point-of-care diabetic retinopathy examination in a teleophthalmology program: A retrospective cohort study
Francisco J. Bonilla-Escobar, Sarah Atta, Maria R. Eibel, Joshua S. Wesalo, Jordan T. Chang, Daniel Sánchez-Cano, Andrew Eller, Evan L. WaxmanObjective
To identify patient-, clinical-, and system-level factors associated with loss to follow-up after point-of-care diabetic retinopathy examination (DRE).
Methods
We conducted a retrospective cohort study of adults ≥18 years undergoing DREs at 75 primary care clinics in Pennsylvania, USA (2017–2019), restricted to examinations with recommended follow-up. The unit of analysis was the DRE. Electronic medical records were used to obtain six-month follow‑up outcomes and patient-, clinical-, and system-level variables. Loss to follow-up was defined as failure to attend an eye care visit within 180 days of DRE. Examinations with duplicate results, those from patients who died within six months, or from sites where the program was discontinued were excluded. Multivariable mixed-effects logistic regression models with random intercepts for patient and clinic were used. Analyses were conducted in the full cohort and separately for gradable and indeterminate exams (ungradable/no image).
Results
Among 3,121 DREs from 2,959 patients, 32.3% completed follow-up. In the full cohort, higher odds of loss to follow-up were associated with increasing age (adjusted odds ratio [aOR]=1.02, 95% confidence interval [95%CI]=1.00–1.03, p = 0.019), not being married (aOR=1.36, 95%CI = 1.04–1.77, p = 0.03), findings suggestive of age-related macular degeneration (AMD) on DRE (aOR=1.80, 95%CI = 1.14–2.85, p = 0.012), and higher no-show rates (aOR=1.04, 95%CI = 1.02–1.05, p < 0.0001). History of eye disease (aOR=0.50, 95%CI = 0.37–0.69, p < 0.0001) and patient contact within 30 days (e.g., same/next day: aOR=0.37, 95%CI = 0.20–0.67, p = 0.001) were associated with lower odds of loss to follow-up; contact at 31-60 days was not significantly associated. Findings were consistent across full and gradable models; only no-show rate remained significant in the indeterminate model. Socioeconomic variables, comorbidities, and diabetic retinopathy severity were not independently associated with loss to follow-up after adjustment.
Conclusions
Adherence was associated with patient (sociodemographic: age, marital status), clinical (AMD on DRE, prior eye disease), and system-level factors (no-show rates, patient contact within 30 days), supporting targeted, multifactorial interventions.