Trends and Predictors of 90-Day Sepsis-Associated Readmissions Among Adults Hospitalized With Sepsis.

Background Sepsis survivors remain vulnerable to recurrent hospitalization after discharge, yet factors associated with longer-term sepsis-associated readmission are incompletely understood. We evaluated temporal patterns and factors associated with 90-day sepsis-associated readmission following adult sepsis hospitalization. Methods We conducted a multicenter retrospective cohort study using de-identified electronic health record and administrative data from 50 hospitals between January 2018 and December 2025. Adults hospitalized with sepsis were eligible. Index hospitalizations in which the patient died during the hospitalization or was discharged to hospice were excluded from the readmission analysis. The primary outcome was recurrent hospitalization within 90 days in which sepsis, severe sepsis, or septic shock was coded as the primary diagnosis. Demographic, clinical, comorbidity, infection-related, and hospitalization-related variables were evaluated using multivariable logistic regression. Calendar year was initially modeled as a continuous predictor, with a sensitivity analysis treating year categorically to assess the assumption of a linear temporal relationship. Results Among 95,655 eligible index sepsis hospitalizations included in the readmission analysis, 6,095 (6.37%) were followed by a sepsis-associated readmission within 90 days. Index hospitalizations followed by readmission involved older patients, had greater comorbidity burden, longer index hospitalizations, and were more frequently discharged to non-home settings. Factors independently associated with increased readmission odds included greater Elixhauser Comorbidity Index, non-home discharge, pressure ulcer, severe sepsis, and abnormal lactate. Annual readmission rates demonstrated year-to-year variation, ranging from 5.52% in 2021 to 7.22% in 2023. When calendar year was modeled categorically, the overall year effect was significant (p < 0.001), with significantly higher adjusted odds of readmission in 2023 (OR 1.191, 95% CI 1.070-1.325) and 2024 (OR 1.136, 95% CI 1.019-1.266) compared with 2018. The categorical-year model provided significantly better fit than the continuous-year model (p < 0.001), indicating a nonuniform temporal pattern. In-hospital mortality among sepsis-associated readmissions was 32.3%. Conclusions Sepsis-associated 90-day readmission was associated with substantial mortality and multiple markers of post-sepsis vulnerability, including comorbidity burden, discharge disposition, and acute illness severity. These findings identify candidate factors associated with recurrent sepsis-related hospitalization and may inform future research on post-discharge risk assessment and intervention; however, prospective and external validation are required before these factors can be applied to individual patient-level risk prediction.