One-Year Changes in Diabetes Eye Care Among Providers After Interventions to Improve Process of Care

Abstract

Purpose: : Prior studies show that significant opportunities exist to assist primary eye care providers (optometrists and ophthalmologists) in more fully utilizing best-evidence eyecare for patients with diabetes. Short-term results (3 to 8 months) presented last year demonstrated no significant changes in process quality of care as measured by open-ended vignettes across providers randomized to one of 3 groups - 1) CME course only; 2) CME plus a templated exam (chart) form and an office poster with reference photos and care recommendations (low technology) and 3) CME plus a real-time tablet computer with an integrated decision support system (high technology). Understanding the longer-term impact (12 to 16 months) of such interventions is essential to assessing their true efficacy.

Methods: : Providers completed an instrument that had been psychometrically balanced with the baseline and short-term surveys, based on the results of earlier studies. Vignettes have previously been shown to be valid and reliable proxies for chart abstractions and simulated patients. We analyzed the content from 82 provider assessments of what they would typically do for patients with different stages of diabetic retinopathy. Comparisons to baseline scores were made both within groups and among the 3 arms, using linear regression and adjusting for baseline scores.

Results: : The vignette-based content of care did not change for those providers who received only CME. While scores in the high technology arm improved between 3 and 7 points for new patient evaluations (on a 100 point scale), none of the changes were statistically or clinically significant. However, significant DECREASES of up to 12 points were noted in summary scores for new and follow-up patient evaluations among low technology providers.

Conclusions: : Use of low technology aids designed to improve process quality of care - a common technique across all of medicine - was associated with a DECREASE in process quality as measured by provider vignette responses at 1 year. In contrast, use of a tablet computer with individualized decision support systems was associated with a tendency to higher process quality. Whether these findings are borne out by ongoing chart abstraction has critical ramifications for understanding both the efficacy and the underlying mechanisms of process quality improvement as well as for methods of measurement.

Keywords: clinical (human) or epidemiologic studies: health care delivery/economics/manpower • diabetes • clinical (human) or epidemiologic studies: treatment/prevention assessment/controlled clinical trials