Prevalence, Risk Factors, and Outcomes Associated With Delayed Second Doses of Antibiotics in Sepsis

INTRODUCTION: Best practice guidelines and quality measures emphasize timely administration of initial antibiotics in patients with sepsis but there are limited data on the prevalence, risk factors, and clinical impact of delays in subsequent doses.

METHODS : We retrospectively identified all adult patients admitted to a large academic hospital who triggered an electronic sepsis alert in the emergency department (ED), received ≥2 doses of vancomycin or an antipseudomonal beta-lactam, and were discharged with an ICD-10 sepsis code between January 2018-December 2019. We calculated the prevalence of delays in second doses of vancomycin or antipseudomonal beta-lactams, defined as ≥25% of the recommended dose interval (accounting for initial renal function), and conducted multivariable regression analyses to assess for risk factors for delays (including demographics, comorbidities, SOFA score, lactate, creatinine clearance, source of infection, time in the ED, ICU admission, ED vs non-ED location at time of 2nd dose, need for stress dose steroids) and the association between delays and short-term mortality (in-hospital death or discharge to hospice).

RESULTS: The cohort included 449 patients with sepsis, of whom 123 (27.4%) had delays in second doses of vancomycin or antipseudomonal beta-lactams. Short-term mortality occurred in 38 patients (30.9%) in the delayed group and 89 (27.3%) in the non-delayed group (p=0.45). On multivariable analysis, only location in a non-ED unit at the time second doses were due was significantly associated with delays (OR 2.75, 95% CI 1.20-6.32). In the mortality model, significant risk factors included malignancy (OR 2.11, 95% CI 1.26-3.53), respiratory infection (OR 1.91, 95% CI 1.15-3.17), and elevated SOFA score (OR 1.16, 95% CI 1.08-1.25), but not delayed second antibiotic doses (OR 1.19, 95% CI 0.69-2.05).

CONCLUSIONS: Over a quarter of patients treated for sepsis in the ED experienced delays in second doses of antibiotics; these delays were more likely when the patient had transferred out of the ED before the second dose was due, suggesting that transitions of care may have contributed. Delayed second doses were not significantly associated with mortality, although our study may be underpowered to detect small differences in this outcome.