Understanding Facilitators and Barriers in the Hospital Discharge Processes of Newly Prescribed Insulin: A Mixed-Methods Study

Patients, newly prescribed insulin, being discharged from the hospital are at high risk of adverse outcomes. An electronic enterprise data warehouse (EDW) algorithm was created and validated to identify these inpatients electronically. Qualitative interviews were also conducted to assess barriers in the discharge process. The EDW algorithm to identify inpatients (09/01/18-08/31/19), newly prescribed insulin at discharge, was created by identifying screening indicators (e.g., admission/discharge medication lists, discharge summary). Iterative adjustments to the algorithm were made after chart review and included review of medication reconciliation (med rec), admission/discharge orders, and insulin orders (types/delivery). The EDW list was compared to the list of patients who received insulin teaching from the Certified Diabetes Care and Education Specialist (CDCES), during the same period. Providers (N=8, 3 endocrine attending MDs, 2 fellow MDs, 3 resident MDs) were interviewed in key informant interviews (N=3) and focus groups (N=2); transcripts were independently coded by 2 coders, utilizing a constant comparative method to generate key themes. The EDW list (N=554) was audited by EHR review (n = 42, 8%); 83% (35/42) were correctly identified as newly discharged on insulin. Of the 7 incorrectly identified, 4 likely had incomplete med rec. The EDW algorithm was unable to correctly identify patients with inaccurate/incomplete med rec, patients transferring from outside hospitals or those without e-Rx at discharge (vouchers, call-in). The CDCES list (N=257) was audited (n=25, 10%), and of patients not meeting criteria (n=15), some had prior insulin prescribed (n=5), and most ended up not discharged on insulin after CDCES insulin teaching (n=9). Comparison of the EDW and CDCES lists had 177 patients (32% of EDW list) in common, with 377 on the EDW list with no CDCES consultation. An audit (n=21/377, 5%) of these EDW patients, who did not have CDCES or endocrinology consultation, revealed patients across service lines, with minimal formal documentation of insulin training/education. Key identified themes from interviews identified barriers including lack of availability of a CDCES after-hours and on weekends, low health literacy/numeracy, and lack of time during stay. In training MDs noted variability in discharge prescribing by supervising MDs and the need to assess “chart lore,” given cut and paste documentation in EHR. This study suggests that an EDW algorithm can be used to identify patients newly being discharged on insulin, for whom teaching by a CDCES is recommended. The data suggest the need for more targeted and increased CDCES capacity as only a portion of those eligible for insulin teaching were seen while others were seen but then not discharged on insulin. Additional resources for insulin teaching are needed and standardized training and documentation need to be developed.