586 P&T® • October 2006 • Vol. 31 No. 10 Disclosure. Dr. Gagne has indicated that he is an Outcomes Research Fellow for Ortho-McNeil Janssen Scientific Affairs, LLC. This article is supported by a grant from Pfizer, Inc. This study was presented at the 46th Annual Interscience Con- ference on Antimicrobial Agents and Chemotherapy in San Francisco on September 29, 2006. Costs Associated with Candidemia in a Hospital Setting Joshua J. Gagne, PharmD, Ross E. Breitbart, MS, Vittorio Maio, PharmD, MS, MSPH, David L. Horn, MD, Christine W. Hartmann, MSS, PhD, Robert Swanson, PhD, and Neil I. Goldfarb cause of all hospital-acquired bloodstream infections. 1–3 The negative health consequences associated with candidemia are very high. 4–7 Previous studies report mortality rates attributable to candidemia ranging from 19% to 49% in adults. 4,6,8,9 In addition, length of stay (LOS) in the hospital increases significantly among patients with candidemia. One study estimated a mean LOS increase of 21.1 days for children and 10.1 days for adults. 4 Although many studies have investigated the epidemiology and clinical consequences of candidemia, 10–14 the information about its effect on inpatient costs is limited. One study of hospitalized adults found a mean increase in hospital charges attributable to candidemia of almost $40,000. 4 Another study estimated average major costs associated with candidemia at approximately $35,000 for Medicare patients and $45,000 for privately insured patients. 6 These studies relied on claims data to identify study popula- tions. Because of the acknowledged limitations of administrative data, 15 the methods employed in these studies might have led to an inaccurate sample selection and, consequently, to a possi- ble underestimation of the actual burden of candidemia. The objective of this study was to determine the cost of candidemia in patients at a large urban academic tertiary-care hospital. METHODS Study Design, Data Sources, and Population We conducted a retrospective cohort study at Thomas Jeffer- son University Hospital in Philadelphia, Pennsylvania. Patients infected with candidemia during a hospitalization between Jan- uary 1, 2004, and December 31, 2005, were identified through blood cultures that tested positive for Candida species. A posi- tive blood culture is considered the standard method of diagno- sis for candidemia. 4 The hospital’s microbiology laboratory information system was used to obtain all microbiological data. The microbiology laboratory used the germ tube assay, as well as morphologic ob- servation of chlamydospore formation on cornmeal agar plates, to identify Candida albicans. Other species were identified by the Rapid Yeast Identification System (Remel Diagnostics). All additional data were retrieved via the hospital’s inpatient cost-accounting system, Eclipsys (Eclypsys Corporation), includ- ing all demographic characteristics, International Classification of Disease (ICD-9) diagnosis codes, patient LOS, discharge dis- positions, payers, total and department costs, and Diagnosis- ABSTRACT Objective: Candida species are the leading cause of fungal infections in hospitalized patients and are the fourth most com- mon cause of all nosocomial bloodstream infections. Mortality and length of stay (LOS) attributable to candidemia are extremely high; however, less is known about its impact on inpatient care costs. The goal of this study was to determine the inpatient cost of candidemia at a large urban academic tertiary- care hospital. Methods: All patients with a confirmed positive blood culture for Candida at Thomas Jefferson University Hospital, between January 1, 2004, and December 31, 2005, were identified via the microbiology laboratory information system. Data from the hos- pital’s true-cost accounting system were used to calculate costs, LOS, and mortality rates. We determined the added burden of candidemia by comparing the observed costs and LOS for candi- demia cases with expected Diagnosis-Related Group (DRG)– adjusted costs and LOS and non-candidemia cases. Differences in costs and LOS were also compared across Candida species. Results: From the 68,526 total hospitalizations during the study period, we identified 287 encounters positive for candidemia, yield- ing a frequency of 419 cases of candidemia per 100,000 admis- sions. The mean age of patients with candidemia was 58 years; 52% were female, and 60% were Caucasian. The most commonly identified species were C. albicans (n = 127; 44% of cases) and C. glabrata (n = 75; 26% of patients). Candidemia was associated with additional costs and LOS of $68,311 (95% CI, $57,513–$79,108) and 23.1 days (95% confidence interval [CI], 19.3–26.8 days), respec- tively. No statistically significant differences in total costs, LOS, or mortality were found between species. Conclusion: Inpatient candidemia cases are extremely costly, regardless of the involved species. Most of the costs were attrib- utable to the nursing department, with pharmacy costs consti- tuting about 20% of total costs. BACKGROUND Candida species account for about 75% of all fungal infections in hospitalized patients, and they are the fourth most common Dr. Gagne is an Outcomes Research Fellow, Dr. Maio is a Research Assistant Professor, Dr. Hartmann is Assistant Professor, and Mr. Goldfarb is Director of Research, all in the Department of Health Policy at Jefferson Medical College, Thomas Jefferson Uni- versity, in Philadelphia, Pennsylvania. At the time of this writing, Mr. Breitbart was a Research Intern in the Department of Health Policy. Dr. Horn is Director of Clinical Research in the Division of Infectious Diseases and Environmental Medicine at Thomas Jefferson University. Dr. Swanson is a Senior Director/Team Leader at Pfizer in New York City, New York. continued on page 589