Candida albicans and Non-albicans Bloodstream Infections in Adult and Pediatric Patients: Comparison of Mortality and Costs

Introduction

Candida species are isolated in up to 15% of nosocomial bloodstream infections and are associated with a mortality rate of 5% to 71%.¹ Candida bloodstream infections are a particular concern for patients who are immunocompromised, patients in intensive care settings, patients with central lines, and patients receiving parenteral nutrition and/or prolonged broad-spectrum antibiotics. A recent study reported that the incidence of candidemia in adults increased by 50% from 2000 to 2005.² Candida albicans remains the leading cause of Candida bloodstream infections.³ However, the prevalence of non-albicans Candida infections continues to increase, comprising 35% to 65% of all Candida bloodstream infections.⁴ The most common non-albicans Candida infections are C tropicalis, C parapsilosis, C glabrata, and C krusei.⁴

Candida bloodstream infections are associated with significant hospital costs.^1,5 A 1997 analysis estimated the cost of candidemia in adults to be approximately $34,000 to $45,000.⁶ A study of candidemia in children found a mortality rate of 16%, mean length of stay of 45 days, and hospital charges of approximately $184,000.⁵ These analyses relied on data before 2001. Since that time, however, the incidence of candidemia has increased, the proportion of candidemias caused by non-albicans species has grown, and there have been changes in antifungal resistance patterns. Given the changing epidemiology of candidemia, we sought to use a contemporary cohort of patients to (1) compare the length of stay, inpatient costs, and mortality associated with candidemia among infants, children, and adults; and (2) compare these outcomes between patients with C albicans versus non-albicans bloodstream infections in each age group.

Materials and Methods

The population in this retrospective cohort study included 3 age cohorts: infants aged less than 24 weeks postnatal, children aged 24 weeks through 18 years, and adults older than 18 years. All patients hospitalized at Duke University Medical Center from February 1996 through July 2007 who had a blood culture positive for Candida were included in the analysis. Candida species identification was available for all patients. Patients with a C albicans and non-albicans bloodstream infection identified during the same hospitalization were excluded from the analysis (n = 36); patients with multiple non-albicans Candida infections were included. Duke University Medical Center is a tertiary and quaternary care hospital with more than 900 beds and approximately 41,000 admissions per year. Detailed cost-accounting data were available for patients hospitalized since December 2002; cost data before this period were not available. We evaluated inpatient costs, length of stay, and mortality for each age group from the date of the first blood culture positive for Candida through the date of discharge.

We used generalized linear models specified with gamma distributions and log links to compare costs, negative binomial models to compare length of stay, and chi-square tests to compare mortality. Reported P values and confidence intervals are 2-tailed. Statistical significance was defined as P ≤ 0.05. The institutional review board of the Duke University Health System approved this study.

Results

We identified 101 infants, 143 children, and 854 adults with Candida bloodstream infections. Complete data on length of stay were available for 99/101 (98%) infants, 138/143 (97%) children, and 846/854 (99%) adults. Data on inpatient costs were available for 37/101 (37%) infants, 63/143 (44%) children, 332/854 (39%) adults. Nineteen patients (1.8%) had more than one non-albicans Candida species identified during a hospitalization.

We first compared mean length of stay, inpatient costs, and mortality between age groups for patients with bloodstream infections attributable to any Candida species. Infants with a Candida bloodstream infection had significantly longer inpatient stays (51 days) than children (37 days; P = 0.01), and both infants and children were hospitalized longer than adults (21 days) (P < 0.001 for both comparisons). Although inpatient costs for infants and children were similar ($142,394 and $133,871, respectively; P = 0.77), both groups incurred significantly higher costs than adults ($56,725; P < 0.001 for both comparisons). However, a significantly higher proportion of adults with candidemia died while hospitalized (43%), compared with children (28.7%; P = 0.001). Inpatient mortality among infants (35.6%) was not statistically different from mortality for adults or children. After adjustment for mortality and length of stay, costs were not significantly higher among infants and children compared with adults (P = 0.12), indicating that the intensity of medical resource use on a daily basis was similar across age groups, and the higher costs for infants and children were attributable to longer hospital stays and lower mortality. Differences between age groups among those with C albicans bloodstream infections and non-albicans bloodstream infections were largely consistent with the results when all Candida species were combined (see Table).

Table.

Length of Stay, Inpatient Costs, and Mortality in Infants, Children, and Adults With Candida Bloodstream Infections

Infection	Length of Stay, days			Inpatient Costs, $			No. of Deaths/ No. of Patients (%)
	n	Mean (SD)	Median (IQR)	n	Mean (SD)	Median (IQR)
All Candida species
Infants	99	51.0 (46.5)*†	35.0 (13.0–86.0)	37	142,394 (98,472)†	117,505 (69,211–212,558)	36/101 (35.6)
Children	138	36.9 (43.2)*‡	22.0 (14.0–42.0)	63	133,871 (188,858)‡	54,699 (26,110–165,465)	41/143 (28.7)‡
Adults	846	20.7 (22.5)†‡	14.0 (7.0–26.0)	332	56,725 (69,751)†‡	31,743 (14,949–69,542)	367/854 (43.0)‡
Candida albicans
Infants	52	58.2 (51.8)*†	41.5 (17.0–95.0)	21	160,580 (112,143)*†	153,376 (69,211–223,738)	20/52 (38.5)*
Children	52	34.7 (35.5)*‡	22.5 (14.0–41.5)	27	90,201 (103,537)*‡	49,620 (24,863–144,038)	10/55 (18.2)*‡
Adults	365	18.8 (22.1)†‡	12.0 (6.0–23.0)	126	47,516 (67,217)†‡	24,855 (9,903–62,716)	153/369 (41.5)‡
Non-albicans Candida
Infants	47	43.0 (39.0)†	33.0 (13.0–63.0)	16	118,525 (73,713)†	99,197 (69,896–174,147)	16/49 (32.7)
Children	86	38.3 (47.4)‡	22.0 (10.0–42.0)	36	166,623 (229,444)‡	82,657 (28,667–216,091)	31/88 (35.2)
Adults	481	22.2 (22.8)†‡	15.0 (8.0–27.0)	206	62,357 (70,827)†‡	36,957 (18,640–85,255)	214/485 (44.1)

Abbreviation: IQR, interquartile range.

^*P ≤ 0.05 for comparisons between infants and children.

^†P ≤ 0.05 for comparisons between infants and adults.

^‡P ≤ 0.05 for comparisons between children and adults.

In the second set of analyses, we compared these same outcome measures for patients with albicans versus non-albicans Candida bloodstream infections within each age group. About half (51%) of all Candida bloodstream infections in infants were attributed to C albicans. Among infants, mortality was similar between those with C albicans and non-albicans bloodstream infections (38.5% versus 32.7%; P = 0.54). We did not observe a statistically significant difference in mean costs ($160,580 versus $118,525; P = 0.22) or length of stay (58.2 versus 43 days; P = 0.14) between infants with C albicans and infants with non-albicans Candida bloodstream infections. Approximately 80% of non-albicans bloodstream infections in infants were attributable to C parapsilosis.

Conversely, more than half of the Candida bloodstream infections in children (62%) were attributable to a non-albicans species. Although length of stay was similar in children with non-albicans and those with C albicans bloodstream infections (38.3 versus 34.7 days; P = 0.55), mean inpatient costs were significantly higher for children with non-albicans infections ($166,623 versus $90,201; P = 0.03). Furthermore, children with non-albicans bloodstream infections were approximately twice as likely to die than children with C albicans bloodstream infections (35.2% versus 18.2%; P = 0.03). C parapsilosis (42%) and C tropicalis (27%) were responsible for most non-albicans bloodstream infections in children, and mortality associated with these infections was 29.7% and 41.7%, respectively. C glabrata was the causative organism in 7 children, of whom 4 (57.1%) died.

In adults, 57% of Candida bloodstream infections were attributable to non-albicans species. Adults with non-albicans bloodstream infections incurred significantly higher inpatient costs ($62,357 versus $47,516; P = 0.02) and longer lengths of stay (22.2 versus 18.8 days; P = 0.01) than adults with a C albicans bloodstream infection. Inpatient mortality was similar between adults with non-albicans and C albicans bloodstream infections (44.1% versus 41.5%; P = 0.44). C glabrata was the most prevalent non-albicans species in adult patients, accounting for 49.3% of cases and a mortality rate of 45.8%.

Discussion

In our cohort, non-albicans Candida bloodstream infections were more common than C albicans bloodstream infections in adults and children, a trend consistent with previous studies showing an increasing presence of non-albicans candidemia.⁷ Conversely, C albicans was more common in the infant cohort. This finding is also consistent with previous analyses in neonates.⁸

In our study, mortality was higher in adults than in children. This finding is consistent with survival estimates reported by Pappas et al.³ Conversely, mortality in infants was more similar to the higher mortality among adulta. We found mortality for infants to be 36%, which is higher than previous estimates of 20% for premature neonates.⁹ As noted above, 80% of the non-albicans bloodstream infections in infants were attributable to C parapsilosis. A retrospective study performed in our neonatal intensive care unit during this period (January 1995 through July 1998) revealed higher mortality (8/19) for neonates with C parapsilosis as compared with previous reports.¹⁰ This difference may have been related to time of catheter removal after a positive blood culture and likely contributes to the higher mortality in the present study.

Cost data prior to December 2002 were not available. Outcomes among the 3 age groups were similar between those with and without cost data, with the exception of lower mortality among adult patients with cost data (35.5% versus 47.7%; P < 0.001)—which may reflect improvements in antifungal therapy since 2002—and longer inpatient stays among infants with cost data (65.1 days versus 42.6 days; P = 0.05).

These data demonstrate that Candida infections are associated with significant mortality and costs. Compared with adults, children had longer stays and higher costs, but these findings may be partly attributable to the higher mortality rate among adult patients, as costs were negatively associated with mortality. Nevertheless, even among those who survived, mean length of stay was longer (49.1 versus 24.2 days) and inpatient costs were higher ($122,103 versus $57,376) among infants and children compared to adults (P < 0.001 for both comparisons).

Our study is limited by a single institution’s experience, retrospective design, and lack of information on demographic characteristics, clinical status, susceptibility testing, antifungal prophylaxis, and cause of death. Because outcomes were evaluated from the date of the first positive culture through discharge, some proportion of these outcomes are not attributable to candidemia alone. For example, gestational age at birth and birth weight may have influenced outcomes, independent of candidemia. Additionally, the small numbers of infants and children with C glabrata, C tropicalis, and C parapsilosis limit our ability to evaluate the relative significance of these specific non-albicans species on inpatient outcomes. However, this study included a large number of consecutive patients with blood cultures positive for Candida. Future studies with richer clinical data and larger samples will be needed to more fully evaluate attributable mortality and cost associated with specific non-albicans Candida species among infants, children, and adults.