GEO-sirs - geospatial and social influences on readmissions for sepsis

The National Institute of Nursing Research awarded Vanderbilt University $125,478 on September 30, 2026, under the Nursing Research program (CFDA 93.361) to conduct GEO-SIRS, a study integrating electronic health record data, patient-reported non-medical barrier screening, and geospatially derived contextual indicators to characterize post-discharge outcomes among sepsis survivors in a multi-site academic health system. The research addresses unplanned 30-day hospital readmissions and 30-day all-cause mortality in sepsis patients by evaluating the quality and completeness of non-medical barrier information (including food insecurity, housing instability, transportation barriers, utility insecurity, and interpersonal safety) captured in EHRs and standardizing these data using ICD-10-CM Z codes. The study will apply rigorous statistical and geospatial methods to measure neighborhood-level contextual exposures including deprivation indices, rurality, and proximity to essential services. Performance occurs in Nashville, Tennessee, with a project completion date of July 31, 2028.

Abstract

Sepsis is a major cause of hospitalization and death in the united states, and unplanned 30-day readmissions after sepsis are common, costly, and clinically consequential. Current approaches to readmission risk assessment rely primarily on clinical information recorded during hospitalization and often omit measurable non-medical barriers that may affect recovery and care transitions after discharge. These barriers include food insecurity, housing instability, transportation barriers, utility insecurity, and interpersonal safety. Each barrier can be measured using standardized screening items and, when documented in structured form, ICD-10-CM z codes. In parallel, post-discharge risk may also vary with predefined, address-linked contextual exposures that can be measured at the neighborhood level (e.g., deprivation indices, rurality, proximity to essential services). GEO-sirs will integrate electronic health record (EHR) data, patient-reported non-medical barrier screening, and geospatially derived contextual indicators to improve characterization of measurable post-discharge outcomes among sepsis survivors in a multi-site academic health system. The primary outcome is unplanned 30-day hospital readmission, defined using EHR encounter data; a secondary outcome is 30-day all-cause mortality, captured using EHR-linked mortality indicators. Analyses will rely on prespecified, clearly defined variables directly relevant to these outcomes, including clinical severity and demographic characteristics available in the EHR, to support valid inference and reproducible research. Aim 1: we will evaluate the quality and completeness of non-medical barrier information captured in electronic health records and standardize these data using ICD-10-CM z codes to improve consistency, comparability, and future analytic reproducibility. Aim 2: we will use rigorous statistical and geospatial methods to identify geographic heterogeneity in sepsis readmission risk and compare patient-reported and geospatially derived measures of non-medical barriers and contextual exposures to determine where each source adds value and where important information may be missing. By emphasizing measurable exposures, prespecified clinically meaningful outcomes, and reproducible methods, GEO-sirs will generate a stronger evidence base for future risk stratification and targeted care transition strategies after sepsis.