0145 Worsening ESS Thresholds Predict Five-year Mortality Risk: An NLP-informed Model

Abstract Introduction The Epworth Sleepiness Scale (ESS) is a validated tool for capturing self-reported excessive daytime sleepiness. We examined thresholds for association between ESS changes and risk of 5-year all-cause mortality. Methods ESS values were mined from progress note-derived free form texts using a natural language processing (NLP) pipeline with reported accuracy of 91%. Veterans in the VHA diagnosed with a sleep disorder or who had participated in a sleep study from January 13, 2000, to August 13, 2018, and with two exams at least 12 weeks apart were identified. Patients were categorized as having normal (0-10) or abnormal (11-24) ESS and characterized by changes in ESS over time: normal-normal, normal-abnormal, abnormal-abnormal, and abnormal-normal. Cox proportional hazards models and elbow plots were conducted to evaluate magnitude of and thresholds for risk of 5-year all-cause mortality adjustment for age and sex. Results At baseline, qualifying Veterans were mostly white (74.2%) and male (92.2%) with a mean age of 55.8 ± 13.3 (N = 21,605). One quarter experienced all-cause mortality within five years of their second exam (25.9%). Veterans with abnormal ESS at the second exam had a 6% greater risk of five-year all-cause mortality (aHR: 1.06; 95%CI:1.00,1.11) than those with normal ESS, after adjusting for age and sex. Mortality risk was 13% higher among those converting from normal-abnormal (aHR: 1.13; 95%CI: 1.08,1.32), compared to normal-normal, particularly among those 60 years (aHR: 1.20; 95%CI:1.08,1.32), after adjustment. Regardless of baseline ESS, participants whose score increased over 3-points had significantly higher risk of five-year mortality compared to lesser changes (aHR: 1.08; 95% CI: 1.01, 1.16), which was intensified among those 60 years of age and over (aHR: 1.14, 95% CI: 1.05, 1.23), after adjustment. Conclusion These results suggest that abnormal ESS may serve as a clinical marker for increased risk of 5-year all-cause mortality. Providers should pay particular attention to ESS increases over 3-points, particularly among those 60 and older. Support (if any) NHLBI K25 HL152006 J.R.; ZOLL Respicardia (31002) A.S., AHBPCE, FY2024-002 J.R.; and IQuESt F&A (CIN 13-413) at the Michael E. DeBakey VA Medical Center, and Baylor College of Medicine in Houston, TX.