Spatio-temporal and healthcare trends of non-endemic, invasive fungal infections in the United States, National Hospital Discharge Survey--1996 to 2006.

Non-endemic, invasive fungal infections (IFI) remain a major cause of morbidity and mortality but their healthcare epidemiologic patterns require further elucidation. The 1996-2006 records in the National Hospital Discharge Survey (NHDS) of a hospitalized sub-cohort of HIV, hematologic malignancy, and transplant patients were analyzed. The objective was to determine independent predictors of non-endemic IFI, apart from other known predisposing host factors. Population-weighted, univariate analyses identified potential variables to include in multivariate models. Risk ratios for IFI using logistic regression and calculated incidence rate ratios (IRR) for IFI-associated mortality using a discrete, proportional hazards model were estimated. A total of 372 IFI hospital discharges, with a case-fatality proportion of 11.7% were identified. There was a significant trend toward increasing IFI hospitalizations (86.2%) in smaller hospitals (< 500 beds). Most IFIs occurred during the spring (37.6%, P = 0.01) and in the Midwest and South (41%) sections of the US, and lasted more than 7 days (61.7%, P < 0.0001). However, multivariable analysis revealed that the risk for IFI hospitalization was greatest during the autumn in the Midwest (RR=6.25 [1.57-24.9], P = 0.009) and in the Northeast (RR=8.14 [2.03-32.6], P = 0.003). Transfer from another healthcare facility conferred over a 3-fold increase risk (RR = 3.38 [2.30-4.97]) whereas a clinician referral reduced the risk by 36% (RR=0.64 [0.44-0.88]). The IFI-related mortality rate was least for the young, regardless of area and season (IRR(0-14years) = 0.155 [0.044-0.550]). Maintaining a steady rate over the past decade, non-endemic IFI hospitalizations exhibit a significant differential distribution in time and space. Prevention efforts that incorporate these trends may lessen IFI healthcare burden.