Prediction of Hypoglycemia By Quality of Life Measures: A Retrospective Analysis of Electronically Patient-Reported Data

Abstract

We aimed to examine the utility of electronically patient-reported data (e-PRD) in exploring the risk of diabetes-related hypoglycemia and to evaluate hypoglycemia prediction by the quality of life (QoL) measures. A retrospective analysis of e-PRD for patients with diabetes mellitus who completed the American Diabetes Association's Low Blood Sugar Questionnaire (Hypo-Q) and the Patient-Reported Outcomes Measurement Information System (PROMIS) for QoL assessment. Associations between Hypo-Q answers and PROMIS scores were described using Spearman correlations and 95% confidence intervals, or medians and quartiles. Records of 538 subjects were reviewed; 55% were female, 95% were non-Hispanic, with a mean age (± SD) of 53 ± 15 years. Patients with type 1 diabetes had a longer disease duration and more hypoglycemic episodes (P<.001) with higher PROMIS Physical and Mental T-scores (P<.001, both), when compared to patients with type 2 diabetes. The latter had a higher number of co-existing conditions. Having >5 episodes of either moderate or severe hypoglycemia in a year were reported by 18% and 5% of all patients, respectively. Mean PROMIS Physical and Mental health T-scores were 46 ± 10 and 47 ± 10, respectively. Patients with fewer moderate and severe hypoglycemic episodes had better Physical (P = .047 and P<.001) and Mental (P = .015 and P<.001) PROMIS T-scores with incremental decreases in the odds of hypoglycemia with each point increase in PROMIS T-scores. e-PRD of QoL measures and Hypo-Q were effective in exposing the risks for hypoglycemia and reproducing published findings with significant associations between QoL measures and hypoglycemia risks while providing new insights.