We used the California Patient Discharge Data Set, an administrative database developed by the Office of Statewide Health Planning and Development, to review hospitalization data for 2000–2011 (19). The database contains inpatient (defined as a person formally admitted to the hospital with the expectation of remaining overnight or longer) information reported by nonfederal hospitals in California (20). These data include diagnosis and procedure codes, demographic characteristics, hospital admission and discharge dates, source of admission, and hospital charges. Records for coccidioidomycosis-associated hospitalizations were identified as records with an admission date during January 1, 2000–December 31, 2011, and 1 of the following primary or secondary International Classification of Diseases, 9th revision (ICD-9) codes for coccidioidomycosis: 114.0 primary pulmonary, 114.1 primary extrapulmonary, 114.2 meningitis, 114.3 other forms of progressive, 114.4 chronic pulmonary, 114.5 pulmonary unspecified, and 114.9 unspecified.

Hospitalization records were matched by patients’ social security numbers, dates of birth, sex, race/ethnicity, and county of residence by using probabilistic record linkage methods as described (21,22). Matching of records by patient enabled the classification of all hospitalization records into initial and subsequent hospitalizations for each patient. Records missing the social security number or other data elements were included in the record linkage and subsequent analyses.

Numbers of coccidioidomycosis-associated hospitalizations were calculated for each year during 2000–2011, by initial and subsequent hospitalization, and by type of coccidioidomycosis. Hospitalization rates were calculated per 100,000 population by using population estimates from the California Department of Finance that were based on the 2010 census (23,24).

Subanalyses were performed for patient initial hospitalizations by sex, age group, race/ethnicity, and county and region of patient residence. Two regions of California were defined for this study: areas where coccidioidomycosis is endemic (hereafter referred to as the endemic region) and areas where coccidioidomycosis is less endemic (hereafter referred to as the less endemic region). The endemic region was defined as the 6 counties that had annual case rates consistently higher than those for the state during 2001–2011: Fresno, Kings, Kern, Madera, San Luis Obispo, and Tulare Counties (6,12–14). The less endemic region was defined as all other California counties. Descriptive statistics were also calculated for hospital admissions from prisons or jails.

We reviewed diagnosis codes for the initial hospitalization records to identify patients with the following concurrent conditions: immunocompromised state as defined by the Agency for Healthcare Research and Quality (25), HIV and AIDS (ICD-9 codes 042, V08), diabetes (ICD-9 code 250), and pregnant state (ICD-9 codes V22, V23, 630–679). The estimated percentage of the California population with these conditions was then provided as a reference. The estimated percentage of the California population with HIV infection or AIDS was determined by using data from the California Department of Public Health HIV/AIDS Surveillance Statistical Reports for 2004–2011, and the estimated percentages of the California population with diabetes and with pregnancy were determined by averaging data from the California Health Interview Survey (CHIS) for 2003, 2005, 2007, and 2009 (26,27). For diabetes, both the percentage in our study and the estimated population average were age-adjusted to the 2000 standard US population. We were unable to determine the prevalence of non-HIV immunocompromised state in the population.

Length of stay was calculated for all initial and subsequent hospitalizations and was defined as the number of days from admission to discharge for each hospitalization. Total length of stay per person was defined as the sum of the lengths of stay for an individual patient’s initial and subsequent hospitalizations during the study period.

Total hospital charge per patient was calculated by adding the total charges for all hospitalizations during 2000–2011. Total hospital charge per day was determined by dividing the total charge for each hospitalization by the length of stay for that record. The sum of charges for all hospitalizations was stratified by the 10 expected source of payment categories defined by the Office of Statewide Health Planning and Development (19). Hospital charges for Medi-Cal, Medicare, county indigent programs, and other government were combined to estimate the total charge to government payers. The annual sums of charges were adjusted for inflation and standardized to 2011 by using the US Department of Labor Consumer Price Index and were used to assess changes in total charges over time (28). All charge figures were calculated in US dollars. When a total charge was indicated as invalid, unknown, or no charge, the records were excluded from analyses. Approximately 8% of hospitalization records had invalid, unknown, or no charge indicated.

For all, initial, and subsequent hospitalizations and for all primary pulmonary, other forms of progressive, and meningeal coccidioidomycosis hospitalizations, z tests were used to compare rates for 2011 with those for 2000. In addition, bivariate relative risks were calculated for the effects of year (2000–2011), sex (female, male), age group (0–19, 20–39, 40–59, and >60 years), and race/ethnicity (White, African American, Hispanic, Other) on patient initial hospitalization. Factors with significant bivariate relative risks were then included in a multivariate negative binomial regression model, which was used to test for statistical significance of the trend in initial hospitalization statewide during 2000–2011. Negative binomial regression models controlling for sex, age group, and race/ethnicity were also used to test for statistical significance of the trends in initial hospitalization in the endemic and less endemic regions. Statistical significance was defined as a p value of <0.05. SAS 9.2 software (SAS Institute Inc., Cary, NC, USA) was used for analyses.

During 2000–2011, there were 25,217 coccidioidomycosis-associated hospitalizations in California; hospitalizations increased from 1,074 in 2000 to 3,197 in 2011 (Table 1). The hospitalization rate per 100,000 population in 2011 was 8.6, a significant increase from the 2000 rate of 3.2/100,000 population (p<0.0001). Of the 25,217 hospitalizations, 15,747 (62%) were initial hospitalizations and 9,470 (38%) were subsequent hospitalizations. Of the initial hospitalizations, 9,568 (61%) were for a primary diagnosis of coccidioidomycosis. The initial hospitalization rate per 100,000 population in 2011 was 5.0, a 2-fold increase from the 2000 rate of 2.3/100,000 population (p<0.0001), and the rate of subsequent hospitalizations in 2011 was 3.6/100,000 population, a >4-fold increase from the 2000 rate of 0.8/100,000 population (p<0.0001).

For all hospitalizations, the most common types of coccidioidomycosis diagnoses were for primary pulmonary (48%), other forms of progressive (18%), and coccidioidal meningitis (13%) (Table 1). The rate of any hospitalization for primary pulmonary coccidioidomycosis was ≈4-fold higher in 2011 than in 2000 (4.3 vs. 1.1 hospitalizations/100,000 population), and the rate of hospitalization for other forms of progressive disease and meningitis was ≈2-fold higher in 2011 than in 2000 (p<0.0001 for all).

Of 15,747 patients initially hospitalized for coccidioidomycosis, 3,824 (24%) were readmitted to a hospital during the study period: 2,004 (52%) were readmitted once, 746 (20%) were readmitted twice, and 1,074 (28%) were readmitted >3 times. The median time from initial hospitalization discharge to first readmission was 47 days (range 0–4,252 days). Of those readmitted, 3,006 (79%) were readmitted within 1 year after the initial hospitalization discharge.

A total of 1,220 (8%) patients hospitalized for coccidioidomycosis died during an initial or subsequent hospitalization. For these patients, there was a diagnosis during the final hospitalization of primary pulmonary coccidioidomycosis for 628 (51%), other forms of progressive disease for 309 (25%), and coccidioidal meningitis for 196 (16%).

During the study period, hospitalized patients were residents of 56 of the 58 California counties. Approximately 50% of patients initially hospitalized for coccidioidomycosis resided in 1 of the 6 endemic region counties. The rate of initial hospitalizations in the endemic region increased nearly 3-fold from 2000 (12.0 initial hospitalizations/100,000 population) through 2011 (34.6 initial hospitalizations/100,000); in the less endemic region, the rate increased 1.6-fold from 2000 (1.6 initial hospitalizations/100,000 population) through 2011 (2.5 initial hospitalizations/100,000 population) (Table 2). In the endemic region, the highest annual hospitalization rates were consistently observed in Kern and Kings Counties.

The annual number of hospitalizations increased steadily during 2000–2006, followed by a slight decrease through 2008 and then a steady increase through 2011 (Figure 1). Using multivariate analysis, we adjusted for sex, age group, and race/ethnicity and determined that the increasing trend in initial hospitalizations during 2000–2011 was significant (p<0.0001). Over this period, there was also an overall increase in the annual rates of initial hospitalizations in the endemic and less endemic regions; however, rates in the endemic region were 7–14 times higher than those in the less endemic region (Figure 2). When controlled for sex, age group, and race/ethnicity, trends in initial hospitalizations in the 2 regions increased significantly (p<0.0001) during 2000–2011.

During 2000–2011, the overall initial hospitalization rate increased 2-fold; when the analysis was stratified by sex, age group, and race/ethnicity, a similar increase was observed within most strata (Table 3). Most (69%) initial hospitalizations were for men; the annual initial hospitalization rate for men (6.7/100,000 population in 2011) was consistently more than twice that for women (3.2/100,000 population in 2011). In most years, the annual initial hospitalization rates increased steadily by advancing age group and then decreased slightly for patients >80 years of age; the highest 2011 initial hospitalization rate (8.5/100,000 population) was for persons 70–79 years of age. For race/ethnicity, the annual initial hospitalization rates were highest for African Americans (8.9/100,000 population in 2011) and lowest for Asian/Pacific Islanders (2.0/100,000 population in 2011).

Multivariate negative binomial regression analysis showed that male sex, older age group, and African American and Hispanic race/ethnicities were significantly associated with an increased risk for initial hospitalization (p<0.0001 for all) (Table 4). This increased risk was 2.48 times higher for men than women. The risk for initial hospitalization for African Americans and Hispanics was 2.09 and 1.31 times higher, respectively, than that for Whites. The risks for initial hospitalization for persons 20–39, 40–59, and >60 years of age were 4.22, 7.73, and 9.50 times higher, respectively, than that for persons 0–19 years of age (Table 4).

A total of 1,374 (8.7%) patients initially hospitalized with coccidioidomycosis were admitted from a prison or jail. Of those, 1,103 (80%) were admitted from prisons or jails in the endemic region of California. Thirty-eight percent of hospitalized persons from Kings County were initially admitted from prison or jail, compared with 25% of persons from Fresno County and 10% from Kern County. The median age of persons admitted from prison or jail was 42 years (range 18–88), and 99.5% of these patients were men. Of the 1,374 patients initially admitted to the hospital from prison or jail, 342 (25%) were White, 379 (28%) were African American, and 419 (30%) were Hispanic. The number of initial hospital admissions for persons in prison or jail increased from 28 in 2000 to 201 in 2011. When initial hospitalizations for persons in prison or jail were excluded from the multivariate analysis, the risk for initial hospitalization for men, persons in older age groups, and persons of African American or Hispanic race/ethnicity remained significantly higher than the risk for the reference group (p<0.05 for all) (data not shown).

Records for 33% (5,176) of the coccidioidomycosis-related initial hospitalizations had a diagnosis code indicating a concurrent condition that could increase the risk for infection or severe disease. Of those, 34% (1,760) had diagnosis codes for immunocompromising conditions. Approximately 2.8% (439) of initial hospitalizations had a diagnosis code of HIV infection or AIDS, compared with an estimated 0.3% of the California population. Approximately 25% (3,786) of adults initially hospitalized for coccidioidomycosis had a diagnosis of diabetes. When our data and the CHIS data were age-adjusted to the 2000 US standard population, 22% had a diagnosis of diabetes, compared with an age-adjusted 7.7% of the general California population. Twelve percent (185) of women 18–45 years of age had a diagnosis code indicating pregnancy, compared with the state average of 4% from CHIS.

During 2000–2011, the median length of stay per hospitalization was 6 days; this length of stay did not vary substantially over time. For ≈52% of those hospitalized, the total per person length of stay was >1 week. The median charge per day was ≈$6,800 US, and the median total hospital charge per patient was US $55,062 (range ≈$1,000 to >$6 million).

During 2000–2011, the total charges for all coccidioidomycosis-associated hospitalizations in California was US $2.2 billion, and the average annual total was US $186 million (Table 5). After we adjusted for inflation, the annual total charges increased from US $73 million in 2000 to US $308 million in 2011. For the expected source of payment category, private coverage had the highest total hospital charges during the study period (total US $713 million; average annual total US $59 million). Government payers (defined as Medi-Cal, Medicare, other government, and county indigent payers) were the expected source of payment for 62% of charges. For all government payers combined, the total (US $1.4 billion) and average annual total (US $115 million) of hospital charges were each nearly twice those for private coverage (Table 5).