Characteristics of Cases With Polymicrobial Bloodstream Infections Involving Candida in Multisite Surveillance, 2017

Abstract

Background: Candidemia is associated with high morbidity and mortality. Although risk factors for candidemia and other bloodstream infections (BSIs) overlap, little is known about patient characteristics and the outcomes of polymicrobial infections. We used data from the CDC Emerging Infections Program (EIP) candidemia surveillance to describe polymicrobial candidemia infections and to assess clinical differences compared with Candida-only BSIs. Methods: During January 2017–December 2017 active, population-based candidemia surveillance was conducted in 45 counties in 9 states covering ~6% of the US population through the CDC EIP. A case was defined as a blood culture with Candida spp in a surveillance-area resident; a blood culture >30 days from the initial culture was considered a second case. Demographic and clinical characteristics were abstracted from medical records by trained EIP staff. We examined characteristics of polymicrobial cases, in which Candida and ≥1 non-Candida organism were isolated from a blood specimen on the same day, and compared these to Candida-only cases using logistic regression or t tests using SAS v 9.4 software. Results: Of the 1,221 candidemia cases identified during 2017, 215 (10.2%) were polymicrobial. Among polymicrobial cases, 50 (23%) involved ≥3 organisms. The most common non-Candida organisms were Staphylococcus epidermidis (n = 30, 14%), Enterococcus faecalis (n = 26, 12%), Enterococcus faecium (n = 17, 8%), and Staphylococcus aureus, Klebsiella pneumoniae, and Stenotrophomonas maltophilia (n = 15 each, 7%). Patients with polymicrobial cases were significantly younger than those with Candida-only cases (54.3 vs 60.7 years; P < .0004). Healthcare exposures commonly associated with candidemia like total parenteral nutrition (relative risk [RR], 0.82; 95% CI, 0.60–1.13) and surgery (RR, 0.99; 95% CI, 0.77–1.29) were similar between the 2 groups. Polymicrobial cases had shorter median time from admission to positive culture (1 vs 4 days, P < .001), were more commonly associated with injection drug use (RR, 1.95; 95% CI, 1.46–2.61), and were more likely to be community onset-healthcare associated (RR, 1.91; 95% CI, 1.50–2.44). Polymicrobial cases were associated with shorter hospitalization (14 vs 17 days; P = .031), less ICU care (RR, 0.7; 95% CI, 0.51–0.83), and lower mortality (RR, 0.7; 95% CI, 0.50–0.92). Conclusions: One in 10 candidemia cases were polymicrobial, with nearly one-quarter of those involving ≥3 organisms. Lower mortality among polymicrobial cases is surprising but may reflect the younger age and lower severity of infection of this population. Greater injection drug use, central venous catheter use, and long-term care exposures among polymicrobial cases suggest that injection or catheter practices play a role in these infections and may guide prevention opportunities.

Funding: None

Disclosures: None