Combination therapy vs. monotherapy for Gram-negative bloodstream infection: matching by predicted prognosis.

The utility of empirical combination antimicrobial therapy for Gram-negative bloodstream infection (BSI) remains unclear. This retrospective, quasi-experimental matched cohort study examined the impact of empirical combination therapy on mortality in patients with Gram-negative BSI. Hospitalized adults with Gram-negative BSI from 1 January 2010 to 31 December 2013 at Palmetto Health Hospitals in Columbia, SC, USA were identified. Patients receiving combination therapy or beta-lactam monotherapy were matched 1:1 based on age, sex and Bloodstream Infection Mortality Risk Score (BSIMRS). Multivariate Cox proportional hazards regression with propensity score adjustment was used to examine overall 28-day mortality and within predefined BSIMRS categories (<5 and ≥5). A total of 380 patients receiving combination therapy or monotherapy for Gram-negative BSI were included in the study. Median age was 66 years and 204 (54%) were female. Overall, 28-day mortality in patients who received combination therapy and monotherapy was 13% and 15%, respectively (P = 0.51). After stratification by BSIMRS, mortality in both combination therapy and monotherapy groups was 1.1% in patients with BSIMRS <5 (P = 0.98) and 27% and 32%, respectively, in patients with BSIMRS ≥5 (P = 0.47). After adjusting for propensity to receive combination therapy, risk of mortality was not significantly different for combination therapy compared to monotherapy (hazard ratio [HR] 0.90, 95% confidence interval [CI] 0.51-1.60). This finding persisted for both subgroups of BSIMRS <5 (HR 0.96, 95% CI 0.04-24.28) and BSIMRS ≥5 (HR 0.83, 95% CI 0.46-1.48). There is no survival benefit from empirical combination therapy over monotherapy in patients with Gram-negative BSI, regardless of predicted prognosis at initial presentation.