In September 2016, a team consisting of clinicians, epidemiologists, and microbiologists from INS, CDC, and the Ministries of Health of Barranquilla, Bogotá, and Cartagena conducted an outbreak investigation in 1 acute-care hospital serving adult and pediatric patients in Barranquilla, 2 adult hospitals in Bogotá, and 1 pediatric hospital in Cartagena. These facilities all had confirmed cases of C. auris candidemia after January 2015.

Figure

Figure. Epidemic curve for cases of Candida auris candidemia in Colombia, by hospital, January 2015–September 2016.

We identified 40 cases of C. auris candidemia occurring during January 2015–September 2016. In 3 hospitals, cases clustered in May–July 2016, whereas in the fourth hospital, cases occurred throughout the 21-month period (Figure). The median patient age was 23 years (interquartile range [IQR] 4 months–56 years); 12 patients (30%) were <1 year of age. Most patients (24; 60%) were male. Thirty-five (88%) patients had a documented concurrent condition before hospitalization; the most commonly reported conditions were hemodialysis for renal failure (23%); diabetes (18%); and immunocompromising conditions (16%), including cancer and transplant.

More than half (58%) of patients died while hospitalized; the overall 30-day mortality was 43%. The median length of admission was 46 days (IQR 34–69 days). For the 38 patients with known intensive care unit (ICU) admission, median length of ICU stay was 36 days (IQR 25–58 days). All case-patients were exposed to an invasive procedure. Central venous catheter (CVC) placement was present in all 40 patients (100%); other common procedures were intubation (35; 97%), surgical procedures (28; 70%), and hemodialysis (15; 38%) (Table 1).

Median time from admission to having a blood culture positive for C. auris was 22 days (IQR 18–31 days). The median time from placement of CVC to positive blood culture was 12 days (IQR 5–21 days). The most common treatments received for C. auris candidemia were fluconazole (16; 43%) and caspofungin (13; 34%). Nineteen patients (45%) received >1 antifungal drug; the median number of antimicrobial drugs administered to patients was 3.5 (range 1–6) (Table 1). The most frequent laboratory abnormalities were mild leukocytosis (median leukocyte count 12.7 × 10⁹ cells/L) and anemia (median hemoglobin 9.8 g/dL).

The median age of the 17 patients who died within 30 days of positive culture was 36 years (IQR 6 months–65 years). Neither age group (p = 0.18), sex (p = 0.9), nor preexisting concurrent condition (p = 0.30) were associated with 30-day mortality. Diabetes was associated with 30-day mortality (odds ratio [OR] 12, 95% CI 1.28–112.42; p = 0.03). The median length of hospital stay was longer for surviving case-patients (63 days, IQR 45–95 days) than for those who died (36 days, IQR 29–45 days; p<0.01). The median length of ICU admission was 32 days (IQR 27–44 days) for case-patients who died and 51 days (IQR 14–76 days) for those who survived (p = 0.3). No procedures were found to be associated with 30-day mortality. The median time from admission to positive blood culture was 22 days (IQR 21–37 days) for the patients who died and 21 days (IQR 12–28 days) in those who survived (p = 0.23). The median time from positive blood culture to death in those who died was 7.5 days (IQR 6–17 days). We were unable to ascertain whether death was attributable to C. auris fungemia.

Among the 12 cases in infants, median age was 34 days (IQR 17–107 days) and 9 (75%) were male; 6 (50%) died in the hospital, and 5 (42%) died within 30 days of C. auris culture. Five (50%) were preterm (<37 weeks’ gestation), and 6 (50%) had congenital heart disease (Table 2). Four (80%) of 5 premature infants survived.

We performed antifungal susceptibility testing on 34 available clinical isolates from 34 unique case patients at 3 hospitals. Six (18%) isolates were resistant to fluconazole, 10 (29%) to amphotericin B, and none to the echinocandin class (anidulafungin, caspofungin). Ten (50%) patients who died had been treated with fluconazole, 9 (43%) received caspofungin, 1 received amphotericin B, and 1 received voriconazole; 2 patients who died did not receive an antifungal drug. Of the 6 case-patients with isolates resistant to fluconazole, 4 (67%) died by 30 days. Of the 10 case-patients with isolates resistant to amphotericin B, 4 (40%) died by 30 days. Resistance to fluconazole was not associated with outcome (p = 0.21), nor was resistance to amphotericin B (p = 0.85) (Table 3). Fluconazole-resistant isolates were seen at all 3 hospitals and amphotericin B-resistant isolates at 2 hospitals, in Barranquilla and Cartagena (northern region of Colombia).

Cases of C. auris bloodstream infection in Colombia were associated with nearly 60% all-cause, in-hospital deaths and a 43% 30-day mortality, which is higher than that reported in the United States. Cases occurred primarily in patients in ICUs who had central venous catheters and other invasive devices, continuing to support the findings that these cases occur in patients with long stays involving multiple procedures (13). In contrast to the United States, where most cases have occurred in older adults, nearly one third of cases in Colombia were in infants, and the median patient age was 23 years (15). Although cases are occurring in vulnerable extremes of age populations, our findings suggest that no age group is unaffected and that any given hospital could be affected by an outbreak.

Clustering of patients in time and space, the findings of skin colonization, and the highly clonal nature of the isolates all strongly support person-to-person transmission occurring in the healthcare setting. Although Candida is often considered a commensal organism that most commonly colonizes the gastrointestinal tract, C. auris behaves more similarly to C. parapsilosis in its propensity to colonize the skin, which provides an opportunity for person-to-person spread. Outbreaks of C. parapsilosis have occurred in both adult and neonatal ICUs (16,17). In our investigation, 2 NICUs were involved; in at least 1 NICU, healthcare workers shared time between patients, providing an opportunity for transmission. The propensity of C. auris to colonize the skin may also explain the strong association with intubation, catheters, and feeding tubes in infants. The ability of C. auris to form biofilms may further enhance its ability to migrate into the bloodstream when provided a conduit through the skin (18). In this investigation, central lines were in place for a mean of 12 days at the time of positive culture. The tendency for skin colonization, biofilm production, and ability to cause invasive infection further emphasizes the need for diligent central line care.

The high death rate associated with C. auris infections, coupled with potential for nosocomial transmission, including among high-risk neonates, underscores the importance of infection control to prevent its spread. Current recommendations focus on early notification, hand hygiene, disinfection with a US Environmental Protection Agency–registered hospital-grade disinfectant effective against Clostridium difficile spores, and use of standard and contact precautions (19).

The characteristics of patients with C. auris candidemia were similar, but not identical, to those reported for candidemia caused by other species. Patients with C. auris bloodstream infection shared exposures such as ICU stay, central venous catheters, and surgical procedures (7). Whereas immunosuppressive conditions, including solid organ tumors, hematologic malignancy, and transplants are known risk factors for candidemia, they were uncommon in patients with C. auris candidemia (18). The reason for this may be that whereas other Candida species invade when there is a shift in the physiology of the host, C. auris is not a commensal, but rather a new exposure. Thirty-day mortality rates for C. auris candidemia in Colombia appear to be slightly higher (43%) than those seen for candidemia from other species, although the small sample size in our investigation limits the comparison (20). Half the infants in our cohort died, although age was not significantly associated with death. Furthermore, we compared characteristics between case-patients who survived and those who died and found few to be associated with outcome. Preterm birth was more common in those who survived, and only diabetes was associated with 30-day mortality. The high death rate is likely multifactorial but is most clearly related to long hospital stays, multiple invasive procedures, and lack of clinician awareness of resistance profiles.

In September 2016, a national laboratory alert was released in Colombia to increase reporting and surveillance catchment, as well as to enhance awareness of C. auris and the other species with which it was most commonly confused (21). According to the national guidelines for surveillance of healthcare-associated infections, aerobic and anaerobic blood cultures should be performed on any patient with signs of bloodstream infection (11). However, the capacity to perform blood cultures may not be available in all facilities; thus, it is possible that cases may be missed. The alert coincided with the initiation of our investigation and thus did not contribute heavily to our case counts, but it did increase reporting nationally.

Directed treatment is key to successful outcomes in fungemia. Resistance is another noteworthy feature of C. auris. Among C. auris isolates collected from 4 world regions, nearly all (93%) were resistant to fluconazole, and 41% were resistant to >2 classes of antifungal drugs (13). In our investigation, fluconazole was the most common antifungal drug administered, and treatment failure may have played a role in death rates. Of the isolates in this investigation, only about one fifth were resistant to fluconazole, but nearly one third were resistant to amphotericin B. The low proportion resistant to fluconazole was surprising; medical practices surrounding administration likely play a role. Forty-three percent of patients received fluconazole around the time of blood culture yielding C. auris, and of these, 57% died. The mortality rate was even higher (83%) among the patients with isolates resistant to fluconazole. Most C. auris isolates have been susceptible to echinocandin drugs, which remain the recommended first-line treatment for C. auris candidemia (9,13,22). However, the often high cost of these drugs can be an obstacle to their use in resource-limited settings.

Because of the retrospective nature of this investigation, data on all variables were not available for each case-patient. Data on previous exposure to antifungal drugs and laboratory values were the most commonly missing. These limitations, along with the small sample size, may have prevented the detection of associations between risk factors and outcome.

The findings of our investigation highlight the necessity of adherence to infection control recommendations, especially aspects of careful central line care and maintenance, hand hygiene, proper disinfection of medical equipment, and use of standard and contact precautions (https://www.cdc.gov/fungal/diseases/candidiasis/recommendations.html). C. auris remains an emerging pathogen with the potential for high levels of resistance to a limited body of antifungal drugs. Its propensity to colonize skin provides a means for person-to-person transmission and elevates the concern for healthcare-associated outbreaks. Further understanding of its comportment and attention to infection control and prevention recommendations are paramount to prevent further spread.