Data on hospitalization for coccidioidomycosis in the state of New York from 1992 to 1997 were obtained from the New York Statewide Planning and Research Cooperative System, which compiles uniform inpatient data from all general acute-care hospitals (all public and private hospitals and hospital-based and free-standing ambulatory surgery facilities).

Patients were selected if their hospitalizations listed a primary or secondary discharge diagnosis of coccidioidomycosis (ICD9 114.0-114.5, 114.9) during 1992 to 1997. Information was collected from hospital discharge data abstracts and uniform billing forms for age at admission, sex, race, ethnicity, county of residence, insurance status, seasonality, disposition of patient, hospital length of stay, and total hospital cost.

Hospital medical records were requested from the 16 patients with positive C. immitis cultures available for multilocus genotyping (1994-1997). These records were reviewed for travel histories.

All suspect C. immitis isolates were transported and processed with full safety precautions in a BSL-3 facility. The isolates were subcultured on modified Sabouraud's agar and Mycosel agar, morphologic and microscopic features were examined, and cultures were processed for specific nucleic acid detection. The AccuPROBE C. immitis culture identification test (Gen-Probe Inc., San Diego, CA) was used to confirm the fungal identity. Before 1993, most culture identifications were confirmed by using a C. immitis exoantigen test kit (Scott Laboratories Inc., Fiskeville, RI).

The fungal DNA was extracted from mycelial cultures according to the method of Pan et al. (12). PCR-RFLP was done as described by Burt et al. (13), with minor modifications. We studied five of the 10 loci originally used for each C. immitis isolate because analysis showed these loci to be most useful for strain differentiation (13). The loci/restriction enzymes used included a1/BsrI, bg/DdeI, bl/DdeI, bq/HinFI, and e1/BsmI. Accordingly, California isolates were expected to show polymorphism at e1, and Arizona isolates could be either positive or negative at each locus, while isolates from Texas would be polymorphic only for the a1, bq, and e1 loci.

Hospital discharge data from 1992 to 1997 in New York showed 181 hospitalizations with coccidioidomycosis as the primary or secondary discharge diagnoses (Table 1). The yearly distribution of hospitalizations was 30 (1992), 28 (1993), 30 (1994), 33 (1995), and 32 (1996), with no discernible seasonal pattern. Coccidioidomycosis was the primary diagnosis in 75 hospitalizations and the secondary diagnosis in 106. The clinical diagnoses according to ICD9 Codes included 105 hospitalizations for pulmonary coccidioidomycosis (114.0, 114.5), one with extrapulmonary coccidioidomycosis (114.1), five with coccidioidal meningitis (114.2), 18 with other forms of coocidioidomycosis (114.3), 19 with chronic pulmonary coccidioidomycosis (114.4), and 33 with unspecified coccidioidomycosis (114.9). The mean length of stay was 16 days, with an average hospital cost of $22,516.00.

One hundred sixty-one patients were hospitalized, of whom five had two hospitalizations, two had three, one had four, and one had nine (Table 1). Thirty-two (20%) patients had HIV, 16 patients (10%) had cancer, and 18 patients (11%) died while hospitalized. Overall, 120 (75%) of these 161 patients were either >54 years of age, had cancer, or were HIV infected.

Mean and median ages for HIV-infected patients were 36 and 33 years, compared with 66 and 64 years for cancer patients and 59 and 63 years for others. Further comparison of HIV-infected patients with patients who had neither cancer nor HIV infection showed differences in percentages with a primary diagnosis of coccidioidomycosis (22% vs 50%), male sex (81% vs 50%), black race (31% vs 5%), and death during hospitalization (31% vs 6%).

Figure

Figure. Coccidioidomycosis in New York, 1989–1997. The figure on the left highlights coccidioidomycosis-endemic areas in the United States (adapted from Kirkland TN, Fierer [2]). The figure on the right depicts...

Forty-nine C. immitis cultures were identified at the state mycology laboratory from 1989 to 1997 (Figure).

Sixteen patient isolates were available for multilocus genotyping. The hospital records of these patients were reviewed to determine the likely source of infection. All had a history of travel to the Southwest, with 12 of 16 patients traveling to Arizona. However, date(s) of travel could not be correlated with onset of the disease. The typing of five alleles allowed unambiguous matches with Arizona genotypes for eight isolates from patients whose histories supported this conclusion (Table 2). It was also possible to type the respective C. immitis isolates to Arizona in five instances in which patients gave a history of travel to more than one disease-endemic area (four with travel to Arizona). The Arizona genotypes of three isolates (474-97, 639-97, and 779-97) did not match, as these patients had a history of travel to California, California and Mexico, and California, respectively.

Before 1994, the Centers for Disease Control and Prevention (CDC) definition for coccidioidomycosis relied solely on a physician's clinical diagnosis (5). The current Council of State and Territorial Epidemiologists/CDC surveillance case definition requires the presence of both clinically compatible symptoms and laboratory evidence of infection (5). The laboratory criteria include a positive culture, positive histopathologic results, molecular evidence of C. immitis, a positive serologic test, or a positive skin test. The retrospective nature of this study and our exclusive reliance on hospital discharge summary data precluded detailed evaluation of diagnostic criteria. Nevertheless, the average of 30 hospital discharges per year suggests that coccidioidomycosis may be more common in New York residents than previously recognized. The largest number of positive cases was reported from New York City, but the rate was not increased, and cases were reported throughout the state. Furthermore, while information on travel history was limited, all 16 patients from whom information was obtained had traveled to disease-endemic areas before becoming ill.

Immunocompromised persons, infants, and non-Caucasians are considered at increased risk for coccidioidomycosis (1). The mean age of patients in this series was 54 years, which is in agreement with >40-year age-group most commonly diagnosed in disease-endemic areas (5,6). As expected, a large proportion of patients had other diagnoses consistent with underlying immunosuppression, including a primary diagnosis of HIV in 20% and cancer in 10%. Overall, 75% were 55 years of age, had cancer, or were HIV infected. The availability of C. immitis cultures from only 10% of cases precluded any definitive investigations of geographic areas in which New Yorkers are at greater risk for infection.

The discovery of recombination in clinical isolates of C. immitis (14) has shown genetic evidence of two sexually distinct taxa in this pathogen, which was previously thought to reproduce asexually (15). A corollary of these investigations was the development of a multilocus genotyping scheme for determining the geographic origin of clinical isolates. Although the discharge data available for this study did not include travel information, we were able to delineate Arizona isolates in a subpopulation of our study for which cultures and medical record travel data were available. Thus, a patient's travel history was correlated with disease acquisition in a particular area by independently demonstrating the geographic pattern of the patient's isolate in 13 (81%) of 16 patients evaluated. However, the typing scheme could not determine the geographic origin of two C. immitis isolates, which came from patients thought to have traveled to both California and Mexico. This discrepancy may have resulted from the limited number of alleles used for typing, incomplete travel histories, shortcomings of the typing scheme for California isolates, or the possibility that the two isolates belonged to an outlier group of strains. Further testing is needed on a larger set of isolates from different geographic areas to evaluate this typing scheme.

As a result of this study, laboratories participating in the New York State Clinical Laboratory Evaluation Program (Mycology) were alerted about the hazards posed by C. immitis. A safe procedure for initial examination of suspect cultures was recommended. The laboratories were also provided with guidelines for safe transport of C. immitis to the state mycology laboratory for confirmation and characterization.

The diagnosis and clinical management of coccidioidomycosis in areas where the disease is not endemic pose unique challenges. The clinical symptoms of the disease mimic those of other infectious diseases, which may result in misdiagnosis and inappropriate treatment (1,16). Additionally, C. immitis cultures could be easily confused with a number of nonpathogenic fungi by unsuspecting laboratory personnel, who are at risk for infection. Increased awareness among health-care providers is likely to help in the timely diagnosis of coccidioidomycosis and the prevention of associated illness and death among patients in nondisease-endemic areas.