REVISITING THE LENGTH OF STAY "SWEET SPOT" AFTER HFREF HOSPITALIZATION USING DAYS ALIVE AND OUT OF HOSPITAL
Author(s)
Sibo Liu, MS, PhD Candidate1, Nicholas Graves, PhD1, Audry Lee, MRCP2, Chun Fan Lee, PhD3, Sameera Senanayake, MD, PhD1, Khung Keong Yeo, MBBS2, Derek Hausenloy, PhD4, Sanjeewa Kularatna, MD, PhD1.
1Health Services Research & Population Health (HSRPH), Duke-NUS Medical School, Singapore, Singapore, 2Department of Cardiology, National Heart Centre Singapore, Singapore, Singapore, 3Centre for Biomedical Data Science (CBDS), Duke-NUS Medical School, Singapore, Singapore, 4Cardiovascular & Metabolic Disorders (CVMD) Program, Duke-NUS Medical School, Singapore, Singapore.
1Health Services Research & Population Health (HSRPH), Duke-NUS Medical School, Singapore, Singapore, 2Department of Cardiology, National Heart Centre Singapore, Singapore, Singapore, 3Centre for Biomedical Data Science (CBDS), Duke-NUS Medical School, Singapore, Singapore, 4Cardiovascular & Metabolic Disorders (CVMD) Program, Duke-NUS Medical School, Singapore, Singapore.
OBJECTIVES: Hospital length of stay (LOS) for heart failure (HF) hospitalization is often interpreted as a potentially modifiable marker of care efficiency and resource use, but conventional event-based endpoints may create the impression of an intermediate LOS “sweet spot”. Days alive and out of hospital (DAOH) may help clarify this relationship by summarizing cumulative post-discharge health from a patient-centered perspective. We aimed to evaluate the association between LOS and 30-day post-discharge DAOH, and to compare this association with conventional time-to-first-event and cause-specific readmission outcomes.
METHODS: We conducted a nationwide cohort study using the Singapore Cardiovascular Longitudinal Outcomes Database. Patients discharged alive after an index hospitalization for HF with reduced ejection fraction (HFrEF) between 2011-2021 were included. The primary outcome was 30-day post-discharge DAOH. LOS associations were estimated using zero-one-inflated beta regression for DAOH and Cox models for conventional event-based endpoints. Models adjusted for established prognostic factors and tested for nonlinearity. Sex-based effect modification was assessed for DAOH.
RESULTS: Among 11,916 patients (mean age 68.6 ± 13.3 years; 31.2% women), median LOS was 4 (interquartile range [IQR], 3-7) days. Mean 30-day DAOH was 28.0±5.0 days and 8,979 (75.4%) had 100% DAOH. LOS was nonlinearly associated with 30-day DAOH (P-association<0.001; P-nonlinear <0.001), with adjusted DAOH remaining similar across approximately 1-4 days and declined progressively with longer stays. There was little evidence that this association differed by sex (P-interaction=0.912). Conventional endpoints showed heterogeneous patterns: all-cause mortality increased monotonically with longer LOS, all-cause readmission and the composite of readmission and/or death showed shallow U-shaped patterns, and cause-specific readmission associations varied by cause.
CONCLUSIONS: In this HFrEF cohort, DAOH did not support an intermediate LOS “sweet spot” for cumulative early post-discharge health. DAOH may complement conventional endpoints by reframing LOS evaluation around patient-centered time alive out of hospital, as well as supporting discharge planning, benchmarking, and policy evaluation.
METHODS: We conducted a nationwide cohort study using the Singapore Cardiovascular Longitudinal Outcomes Database. Patients discharged alive after an index hospitalization for HF with reduced ejection fraction (HFrEF) between 2011-2021 were included. The primary outcome was 30-day post-discharge DAOH. LOS associations were estimated using zero-one-inflated beta regression for DAOH and Cox models for conventional event-based endpoints. Models adjusted for established prognostic factors and tested for nonlinearity. Sex-based effect modification was assessed for DAOH.
RESULTS: Among 11,916 patients (mean age 68.6 ± 13.3 years; 31.2% women), median LOS was 4 (interquartile range [IQR], 3-7) days. Mean 30-day DAOH was 28.0±5.0 days and 8,979 (75.4%) had 100% DAOH. LOS was nonlinearly associated with 30-day DAOH (P-association<0.001; P-nonlinear <0.001), with adjusted DAOH remaining similar across approximately 1-4 days and declined progressively with longer stays. There was little evidence that this association differed by sex (P-interaction=0.912). Conventional endpoints showed heterogeneous patterns: all-cause mortality increased monotonically with longer LOS, all-cause readmission and the composite of readmission and/or death showed shallow U-shaped patterns, and cause-specific readmission associations varied by cause.
CONCLUSIONS: In this HFrEF cohort, DAOH did not support an intermediate LOS “sweet spot” for cumulative early post-discharge health. DAOH may complement conventional endpoints by reframing LOS evaluation around patient-centered time alive out of hospital, as well as supporting discharge planning, benchmarking, and policy evaluation.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
HSD104
Topic
Health Service Delivery & Process of Care
Disease
Cardiovascular Disorders (including MI, Stroke, Circulatory)