Identifying Potentially Preventable Death From Sepsis

Document Type

Conference Proceeding

Publication Date

4-20-2023

Publication Title

Am J Respir Crit Care Med

Abstract

Rationale: Sepsis is a major cause of hospital mortality, contributing to 1 out of every 2-3 US hospital deaths. However, it is unclear how often sepsis deaths are preventable, as patients hospitalized with sepsis are often elderly, frail, or have multimorbidity. We sought to identify a cohort of previously healthy patients hospitalized for sepsis to assess for potentially preventable sepsis deaths. Methods: This cohort study included adults hospitalized with community-onset sepsis at 13 hospitals in the Michigan Hospital Medicine Safety Consortium (HMS), a Collaborative Quality Initiative within Blue Cross Blue Shield of Michigan Value Partnership Program. Hospitalization data were abstracted into the HMS-Sepsis registry (11/2020-2/2022) by professional abstractors. All patients met CDC sepsis surveillance criteria. Using this registry, we serially excluded older, frail, multimorbid patients to enrich for a “healthy sepsis” sub-cohort while retaining ≥10% of hospitalizations. Results: Of 3,053 patients in the HMS-Sepsis registry, 840 (27.5%) died within 90 days. A majority (1,949, 63.8%) were ≥70 years old, had baseline cognitive or functional impairments, or had cancer. After excluding these patients (Figure 1), 90-day mortality decreased to 15.2%. After applying all 16 exclusions (Figure 1), the final healthy sepsis cohort included 369 (12.1%) patients, with 90-day mortality of 10.0%. All healthy sepsis patients had acute organ dysfunction on presentation, most commonly respiratory dysfunction (82.4% on ≥4L NC), lactate elevation (27.4%), altered mental status (12.7%), and shock (8.8%). Only 1.4% had treatment limitations on admission vs 19.0% in the comorbid sepsis cohort, p<0.001. Compared to healthy sepsis patients who survived, healthy sepsis patients who died were more often admitted directly to the ICU (81.1% vs 20.2%, p<0.001) and had worse organ dysfunction on presentation (median PaO2:FiO2 110 vs 217, p<0.001; median lactate 2.9 vs 2.2 mmol/L, p=0.03; mental status alteration 27.9% vs 11.1%, p=0.006; shock 21.6% vs 7.5%, p=0.01). Compared to comorbid sepsis deaths, healthy sepsis deaths occurred later (86.5% after day 7) and in-hospital (70.3%; median length of stay 15 days). Healthy sepsis mortality varied by hospital, with a pooled 90-day mortality of 3.5% in the highest-performing tercile of hospitals (encompassing 116/369 healthy sepsis hospitalizations) vs 16.5% in the lowest tercile. Conclusions: After excluding older, multimorbid patients, the resulting healthy sepsis cohort had substantially lower mortality, particularly at top-performing hospitals. Acute respiratory failure was the most common organ dysfunction, and deaths generally occurred beyond day 7 after long hospital courses, providing potential targets for preventing sepsis deaths.

Volume

207

Issue

1

Find It @ Sladen

Share

COinS