Performer: University of Florida
Principal Investigator: Almut Winterstein, PhD
Project Duration: 9/30/15-9/29/17
Regulatory Science Challenge
Antihyperglycemic agents have been included as one of the three important drug classes in a national action plan to reduce preventable adverse drug events. While hypoglycemia has been recognized as one of the national patient safety priorities, hyperglycemia has also been associated with negative outcomes for hospitalized patients, including increases in infection and mortality rates. The research group at the University of Florida recently developed two prediction algorithms for hypo- and hyperglycemia in hospitalized patients, which have demonstrated excellent predictive performance in the two largest hospitals affiliated with the University of Florida (UF Shands and UF Jacksonville). The feasibility and effectiveness of the prediction algorithms in reducing hypo- and hyperglycemia rates need to be evaluated.
The University of Florida research team will conduct this study at UF Shands and UF Jacksonville. The two prediction algorithms will run fully automated in the electronic health records (EHR) to generate morning reports of patients ranked according to their risks for hypo- and hyperglycemia. The patient rankings will be integrated in the routine work queue of clinical pharmacists, who will be expected to follow up on all high-risk patients and to document their interventions for all patients in the 90th risk score percentile. The effectiveness of the prediction algorithms will be evaluated based on changes in hypo- and hyperglycemia rates.
- Implement the hypo- and hyperglycemia prediction algorithms in the EHR at UF Shands and UF Jacksonville, for automated retrieval of risk factors and generation of patient scores
- Present the risk scores to clinical pharmacists to guide interventions
- Evaluate the effectiveness of the algorithms in reducing hypo- and hyperglycemia rates