Methods: Preparatory for a penicillin allergy testing protocol, we conducted a retrospective chart review examining the influence of a penicillin allergy label and antibiotic choice on treatment of MSSA bacteremia in adults admitted between 2010 and 2015. Patients expiring before antibiotic susceptibilities resulted were excluded. Outcomes included rates of 30- and 90-day mortality, 90-day recurrence and subsequent allergic reaction. Two exposure variables were evaluated: 1) reported penicillin allergy and 2) antibiotic choice. We constructed multivariable logistic regression models to compare risk of outcomes for those exposed and unexposed while adjusting for confounding factors (age, infection site and co-morbidities).
Results: The study included 335 patients with MSSA bacteremia. Patients labeled with allergy had similar 30- and 90-day mortality rates to non-allergic patients (17 vs 16%, p=0.84 and 20 vs 22%, p=0.67, respectively) but a higher 90-day recurrence rate (2.2% vs 0.2%, p<0.01). Patients receiving vancomycin had substantially higher 30- (26%, p=0.002) and 90-day (28%, p=0.02) mortality rates compared to β-lactams (6-9%, see figure 1). β-lactam use was significantly associated with reduced 30- (OR 0.26, p=0.001) and 90-day (OR 0.423, p=0.02) mortality rates in our regression models (figures 2,3). 41% of patients labeled with penicillin allergy received cefazolin with no allergic reactions documented.
Conclusion: Treatment of MSSA bacteremia with vancomycin was associated with significantly higher mortality rates than β-lactams. Patients labeled with penicillin allergy were more likely to suffer recurrent bacteremia, but not higher mortality. Frequent use of cefazolin in supposedly penicillin allergic patients may explain the absence of a mortality difference by reported allergy. Beta-lactams should be strongly preferred over vancomycin for MSSA bacteremia, and can be safely given to selected patients despite a penicillin allergy label.
Figure 2: 30 day mortality (c-statistic 0.66):
Figure 3: 90 day mortality (c-statistic 0.70):
C. K. Cunningham, None
R. H. Drew, None
C. Sarubbi, None
R. W. Moehring, None