A case was defined as culture-confirmed C. neoformans infection in an inpatient admitted to hospital A during 2013. We reviewed microbiology records at hospital A for 1992–2013 to identify all cases and establish a background rate for positive cryptococcal cultures at hospital A. We telephoned microbiology staff and ICPs at surrounding hospitals to inquire about any recent changes noted in rates of isolation of Cryptococcus spp.

Case-patients’ medical records were reviewed, and data were abstracted by using a standardized case report form that included demographic and clinical information. Case-patients’ family members (or case-patients, if alive) were interviewed by using a standardized questionnaire to identify potential exposures to Cryptococcus. Questions included whether patient had contact with pigeons or other birds and whether case-patients had engaged in any common activities in the community that could have resulted in a common acute exposure to C. neoformans before hospital admission. We interviewed laboratory managers and technicians at hospital A to assess specimen collection and processing methods and to identify any potential sources for contamination of specimens with C. neoformans.

To assess the possibility of acute, nosocomially acquired C. neoformans infection, we took several steps. Case-patient charts were reviewed for commonalities in physical location, procedures, and providers. ICU staff, hemodialysis staff, and ICPs at hospital A were interviewed to gain an understanding of how patients are cared for in the ICU, dialysis procedures, and infection control practices. We also asked hospital leadership about any known history of illegal activity or recent disciplinary action of ICU staff members. We asked facilities management staff about any bird habitats at the hospital and recent construction activity. Because C. neoformans thrives in bird guano, we also interviewed all ICU staff members about contact with birds and sampled the hands using Handi-wipes (8) and homes using Sponge-Sticks (3M Co., St. Paul, MN, USA) and vacuum filter socks (X-Cell 100 Dust Sampling Sock, Midwest Filtration Co, Cincinnati, OH, USA) (9) of ICU staff members who reported substantial bird exposure and had contact with case-patients. Environmental sampling was conducted in the ICU with Sponge-Stick swabs.

Finally, we conducted a retrospective cohort study to identify factors associated with cryptococcosis. Patients admitted to the hospital A ICU during April 1–December 31, 2013 (the period during which the cases occurred), were included. Patients were identified by querying the electronic medical record database. Data extracted from the medical record included length of stay in the ICU, receipt of glucocorticoids in the ICU, type of steroid if one was administered, and whether the patient was cared for by a specific respiratory therapist. For the purposes of the cohort analysis, only case-patients who were admitted to the ICU were included (n = 5). Poisson regression models were fit to the data to examine the relationship between the potential risk factors and cryptococcosis and to estimate the rate ratios between different exposure groups.

Clinical isolates were confirmed as C. neoformans by melanin production on L-DOPA media and by lack of growth after inoculation on canavanine-glycine-bromothymol blue media (10). Isolates were subtyped by using multilocus sequence typing (MLST). The URA5, IGS1, CAP59, LAC1, GPD1, PLB1, and SOD1 gene fragments were amplified as described (11) for all isolates. Allele numbers and sequence types were determined by using the online C. neoformans MLST database (12). Environmental samples were processed and plated onto birdseed benomyl agar, incubated at 35°C, and observed for growth at 4, 7, and 14 days.

We investigated 6 cases of cryptococcosis that occurred in 2013 in a community hospital ICU. For most of the case-patients, the disease appears to have developed while they were in the ICU after admission for other diagnoses. The patients experienced a fulminant clinical course after the diagnosis of cryptococcosis and died soon thereafter. There was no identifiable point source for the infections in the hospital or the community. Receipt of short-term steroids in the ICU was significantly associated with cryptococcosis in this cohort.

The cluster was characterized by several atypical clinical features. First, active cryptococcal disease is usually associated with HIV infection or organ transplant–associated immunosuppression. Five of 6 patients in this cluster were non–HIV-infected and nontransplant patients (NHNT); however, each did have other predisposing conditions, including chronic renal failure, chronic lung disease, hematologic malignancies, and other malignancies that might have put them at risk for cryptococcal disease (13). Second, C. neoformans BSI is extremely uncommon, especially among NHNT patients (14,15), yet 4 case-patients had blood cultures yielding C. neoformans; only 1 case-patient (the renal transplant recipient) had meningitis, a more typical manifestation of this disease. Next, acute respiratory failure and overwhelming sepsis, as experienced by 5 patients in this cluster, are atypical manifestations of cryptococcal disease (16); cryptococcosis is generally a subacute infection with insidious onset of nonspecific symptoms (5), especially among NHNT patients, who may have prolonged symptoms before diagnosis (17). In contrast, the 5 NHNT patients in this cluster had a relatively short duration of symptoms and experienced respiratory failure, septic shock, or both and died within days of their positive culture.

Because C. neoformans infections rarely result from acute fungal exposure, and because person-to-person transmission of cryptococcosis—if it exists at all—is exceedingly uncommon (6,7), focal clusters or outbreaks of cryptococcosis are not expected and, to our knowledge, have not previously been reported. Disease usually results from reactivation of latent infection in immunosuppressed hosts; reactivation in 1 host is an independent event that is not necessarily linked to reactivation in another host. The cases we investigated were clustered in space (ICU) and time (2013, with 3 cases occurring with 10 days of each other in late April through early May). The atypical patients and the unusual clinical manifestations involved in this cluster may be an indication that the source or mechanisms of infection and disease, though not identified during the investigation, were not typical for cryptococcosis. If this were a chance clustering of independent occurrences of reactivation of latent cryptococcal infection, we would have expected to see more patients who fit the typical risk profile and have more common manifestations of the disease, and all would not have occurred at a single hospital.

We searched for a point source in both the community and hospital settings. There has been no precedent for C. neoformans being found in the hospital environment or being transmitted from health care worker hands, but we investigated these possibilities because they have been implicated in outbreaks with other organisms. We conducted a thorough investigation to identify any hospital sources of Cryptococcus spp. but did not find a hospital source through environmental assessment. However, this outcome was limited by the fact that samples were taken ≈12 weeks after the first 3 cases occurred. The meaning of the 3 different MLST patterns among the 6 case-patient isolates is unclear. This finding may be consistent either with a single point source containing multiple strains of Cryptococcus spp. (as demonstrated previously) (17) or a different, non–point-source cause of infection.

Although nearly two thirds of patients who seek treatment with cryptococcal disease have advanced HIV infection or have received an organ transplant (3), long-term oral steroid use is also a known risk factor for cryptococcal disease. In a study of cryptococcal disease among NHNT patients, the median daily dose of prednisone or prednisone-equivalent that patients were receiving was 20 mg, and the median duration of immunosuppression before cryptococcal disease was 7 months (15). Although use of steroids in the treatment of sepsis is generally not favored, steroids may be used for a short term in cases of septic shock not responsive to other interventions. A single 10-day tapering course of hydrocortisone used to treat refractory septic shock, as was administered to some of these case-patients, is roughly equivalent to getting 20 mg of prednisone for 15–20 days. Although cases of cryptococcal disease developing in patients receiving low-dose and short-term oral corticosteroids have been reported (18), clusters of cryptococcal disease in an ICU setting after receipt of short-term steroids have not been previously described. A combination of multiple chronic underlying conditions, including diabetes, renal failure, and malignancy, which are also known but less frequently associated risk factors for cryptococcosis, and short-term steroid use in the ICU may have contributed to reactivation disease or dissemination of acute infection acquired from an unknown source in the community or the hospital. Further research should be conducted to better understand the relationship between short-term steroid use and the risk for opportunistic infections, including cryptococcal disease.

This investigation has several limitations. First, we considered that all patients with positive C. neoformans cultures had cryptococcosis. However, the manifestations were unusual, raising the possibility that some or all case-patients might not have had true cryptococcal disease. Unfortunately, serum CrAg testing results, pathologic specimens, and autopsy findings, all of which might have helped definitively determine if true cryptococcal disease was present, were not available for most case-patients.

Second, although we cannot definitively rule out the possibility of a pseudo-outbreak from laboratory contamination of specimens, C. neoformans, unlike environmental molds, is not typically a laboratory contaminant. Thus, we believed that laboratory contamination was a less likely explanation for the cluster. The findings of C. neoformans in different types of specimens also made laboratory contamination less likely.

Third, we did not investigate the possibility of a common contaminated intravenous medication as a source for these infection and therefore cannot rule out this explanation as a cause of the outbreak. Again, given the different sites of infection (both respiratory and bloodstream) among case-patients, a common-source contaminated intravenous fluid also appeared unlikely.

Finally, we were only able to look at a limited number of co-variates in the cohort study because of the reliance on an automated query of the electronic database. Individual chart review to determine severity of illness or underlying conditions was not possible for all 1,600 patients in the ICU. Therefore, we could not control for differences in reasons for admission to the ICU and severity of underlying illness.

Although we did not find a point source for infections in the hospital or community, we found that short-term steroids used in the ICU were associated with case status. To clarify whether this association between short-term steroid use and cryptococcosis is generalizable, similar studies examining rates and duration of ICU steroid use and cryptococcosis should be conducted at other hospitals. C. neoformans infection may need to be included in the differential diagnosis of a patient receiving short-term steroids in the ICU setting.

We recommended heightened vigilance for cryptococcal infection among ICU patients at hospital A, especially those receiving steroid treatment. We also asked that physicians carefully assess the need for steroid use in patients admitted to the ICU and weigh the risk for possible cryptococcal infection against the benefits of steroid use in each patient’s case.