Bacterial Genetic Determinants of Virulence Associated with In-Hospital Mortality among Outpatient Dialysis Patients with Bloodstream MRSA Infections

Background: Infection with MRSA is an important cause of healthcare-associated infections, leading to morbidity and mortality. S. aureus produces a number of toxins which could increase mortality. Prior studies have looked at the toxin profiles among S. aureus isolates causing infection, but none have examined the association between toxins and outcomes in a nationally representative sample. Objective: We hypothesized that one or more bacterial toxins are associated with mortality in patients with MRSA infections. We measured the toxin gene profiles in MRSA isolates taken from dialysis patients with infections complicated by bacteremia. Methods: Study subjects were selected from patients enrolled in the CDC Invasive MRSA Active Bacterial Core Surveillance Program from 2005 to 2006. Inclusion criteria were positive culture within 48 hours of admission, hospitalization for infection, initial isolation of MRSA from blood, outpatient dialysis within the last year, and the availability of a MRSA isolate for testing. The only exclusion criterion was the presence of a polymicrobial infection. The primary study outcome was mortality within 30 days of positive culture. Study variables included the pulsed-field gel electrophoresis type and the presence or absence of each of 33 genes for S. aureus toxin in the isolate, identified by PCR. Results: 267 patients met the study criteria. 37 patients (14%) died within 30 days of infection. Of the 33 toxin genes, 3 factors (see, etb, agrIV) did not appear in any isolate, 6 factors (sec, seh, tst, can, fnbB, agr III) occured in <5% of isolates, and 13 factors (sei, seg, sdrC, sdrD, bbp, clfB, hlg, efb, ebpS, clfA, icaA, V8, chp) appeared in >95% of isolates. None of the 11 remaining evaluable genes (sea, seb, sed, sej, eta, PVL, sdrE, fnbA, map-eap, agrI, agrII) were significantly associated with mortality, though sea approached significance (RR 1.8, 95% CI 0.9-3.8). 80% of sea-positive isolates were pulsed-field type USA500. Prior cerebrovascular accident (CVA) was associated with mortality. Age was associated with sea and mortality. Cirrhosis was also associated with sea and mortality. In a multivariate model of the association between sea, age, prior CVA and cirrhosis with mortality, sea was significantly positively associated with mortality (OR 3.0, 95% CI 1.0-9.1). Cirrhosis (OR 12.8, 95% CI 3.3-49.4), prior CVA (OR 2.8, 95% CI 1.3-6.3) and age (OR for each year increase 1.04, 95% CI 1.01-1.07) were also significantly associated with mortality. Conclusions: Host factors such as age, cirrhosis and prior CVA were strongly and significantly associated with mortality in this population. Among bacterial virulence genes, only sea was significantly associated with mortality. Our results are limited by a small number of the outcome of interest and lack of information on infection treatment.