Epidemiology and Outcomes of Antibiotic De-escalation in Patients with Suspected Sepsis in US Hospitals

Abstract Background Little is known about the frequency, hospital-level variation, predictors, and clinical outcomes of antibiotic de-escalation in suspected sepsis. Methods We retrospectively analyzed all adults admitted to 236 US hospitals between 2017-2021 with suspected sepsis (defined by a blood culture draw, lactate measurement, and intravenous antibiotic administration) who were initially treated with ≥2 days of anti-MRSA and anti-pseudomonal antibiotics but had no resistant organisms requiring these agents identified through hospital day 4. De-escalation was defined as stopping anti-MRSA and anti-pseudomonal antibiotics or switching to narrower antibiotics by day 4. We created a propensity score for de-escalation using 82 hospital, demographic, and clinical variables, matched de-escalated to non-de-escalated patients, and then assessed associations between de-escalation and outcomes. Results Among 124,577 eligible patients, antibiotics were de-escalated in 36,806 (29.5%) including narrowing in 27,177 (21.8%) and cessation in 9,629 (7.7%). De-escalation rates varied widely between hospitals (median 29.4%, IQR 21.3-38.0%). Predictors of de-escalation included less severe disease on day 3-4, positive cultures for non-resistant organisms, and negative/absent MRSA nasal swabs. De-escalation was more common in medium, large, or teaching hospitals in the Northeast or Midwest. De-escalation was associated with lower adjusted risks for acute kidney injury (OR 0.80, 95% CI: 0.76-0.84), ICU admission after day 4 (OR 0.59, 95% CI: 0.52–0.66), and in-hospital mortality (OR 0.92, 95% CI: 0.86-0.996). Conclusions Antibiotic de-escalation in patients with suspected sepsis is infrequent, variable across hospitals, linked with clinical and microbiologic factors, and associated with lower risk for acute kidney injury, ICU admission, and in-hospital mortality.