Diabetes self-management education and support (DSMES) is known to improve diabetes treatment outcomes, but this limited resource is not always targeted toward the highest risk patients in current models of care. Predictive modeling, which relies on large data sets to predict outcomes, is one tool which can be used to direct resources to patients with the greatest need for these services.

University of Pittsburgh Medical Center (UPMC) diabetes care providers along with members of the clinical analytics team created a predictive model, including patient age, HbA1c, and insulin use among other factors, to assess individual patient risk of future poor glycemic control. The model was applied to patients who received DSMES via diabetes care and education specialists (DCES) from December 2018 to December 2020 to evaluate the impact of DSMES according to risk strata. Patients were categorized into high, medium and low risk groups based on clinical data prior their initial DSMES encounter. HbA1c change was calculated using the HbA1c prior to DSMES and most recent HbA1c.

Of 1,247 patients eligible for retrospective assessment, 50% were categorized as low risk (n=621), 4% (n=55) medium, and 46% high risk (n=571). Mean HbA1c change after DSMES was -0.54% for low risk, -0.72% for medium risk, and -1.19% for the high risk group. Mean HbA1c after DSMES was 7.6% for low risk, 9.7% for medium risk and 9.6% for high risk patients. 57.5% of low, 59% of medium and 46.6% of high risk patients were seen by an endocrinologist.

In this analysis, high risk patients who received DSMES had larger improvement in HbA1c and were less likely to be seen by an endocrinologist. To optimize utilization of diabetes care regarding DCES, predictive modeling can be used to target resources to the highest risk patients.


M. Zupa: None. J. S. Krall: Research Support; Self; Becton, Dickinson and Company, Sanofi. K. Collins: None. J. Ng: Research Support; Self; Sanofi-Aventis. L. M. Siminerio: Advisory Panel; Self; Abbott Diabetes, Bayer U. S., Research Support; Self; Becton, Dickinson and Company.


National Institute of Diabetes and Digestive and Kidney Diseases (5T32DK007052-45)

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.