Predictors of adverse outcome from candidal infection in a tertiary care hospital

Transcription

1 Journal of Infection (2004) 49, Predictors of adverse outcome from candidal infection in a tertiary care hospital R. Ben-Abraham a, N. Keller b, N. Teodorovitch c, A. Barzilai c, R. Harel d, Z. Barzilay d, G. Paret d, * a Department of Anesthesiology and Critical Care, Tel-Aviv Sourasky Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel b Department of Clinical Microbiology, Tel Aviv University, Tel Aviv, Israel c Department of Pediatric Infectious Diseases, Chaim Sheba Medical Center, Sackler Faculty of Medicine, Tel Hashomer 52621, Israel d Department of Pediatric Intensive Care, The Chaim Sheba Medical Center, Sackler Faculty of Medicine, Tel Hashomer 5262l, Israel Accepted 27 February 2004 Available online 18 August 2004 KEYWORDS Candida; Infection; Outcome Summary Objectives. To retrospectively delineate predictors of adverse outcome by looking at the demographic features, therapy and outcome of systemic candida infection in a large tertiary care university-affiliated medical center. Methods. We reviewed the clinical data on 186 inpatients with candidemia over a 6- year period. The major reason for their hospital admission was an underlying malignancy or an infection other than candidemia. Results. Candida albicans, tropicalis, parapsilosis, glabrata and krusei caused 54, 22, 13, 8 and 3% of the candidemia episodes, respectively. The overall mortality was 42% and it was highest in patients suffering from candidemia of the glabrata species (73%). Forty-eight (63%) of the 76 patients who received no anti-fungal treatment died compared to 38 (34%) of 110 patients who were treated ðp, 0:05Þ: Predictors of adverse outcome were intensive care unit stay, renal failure, thrombocytopenia and the need for mechanical ventilation or inotropic support. Conclusions. We identified four predictors of mortality from candidemia infection. Their validity should be further assessed and the specific candida strains and their susceptibility need to be methodically identified. Our data support immediate initiation of therapy at first identification of infection. Q 2004 The British Infection Society. Published by Elsevier Ltd. All rights reserved. Introduction Fungal infections are prominent among the causes of nosocomial infection. 1,2 Great vessel cannulation, *Corresponding author. Tel.: þ ; fax: þ address: gparet@post.tau.ac.il prolonged endotracheal intubation, the administration of wide spectrum antibiotic therapy and the use of total parenteral nutrition are recognized risk factors for the development of invasive fungal infections. 3,4 In addition, the use of modern regimens of chemotherapy and bone marrow or organ transplantation techniques have caused a substantial increase in immunocompromised patients who /$30.00 Q 2004 The British Infection Society. Published by Elsevier Ltd. All rights reserved. doi: /j.jinf

2 318 R. Ben-Abraham et al. are highly susceptible to disseminated fungal infections which have often proved to be lethal. 5,6 Candida species were reportedly responsible for up to 15% of all hospital-acquired bloodstream infections, 7 with Candida albicans constituting more than half of fungal isolates 2 and bearing a mortality rate of greater than 50%. 8 Although it is the one most frequently isolated, the proportion of infections caused by other species is increasing 9 as well as by strains resistant to amphotericin B-the golden standard of anti-fungal chemotherapeutic agent. 1,10 The purpose of this study is to search for predictors of mortality from candida infections by examining data on the demographic features, therapy and outcome of systemic candida infection in a large tertiary care university-affiliated hospital that were accumulated during a 6-year period. Based on these findings, we support the administration of anti-fungals at first identification of infection. Material and methods The Chaim Sheba Medical Center is a 1200-beds multidisciplinary, one of the six major tertiary medical centers in Israel. The medical records and case notes from January 1992 to January 1998 of its inpatients suffering from candidemia were reviewed by the authors. Blood culture identification method used the Bactec BD, CA, USA. The presence of the condition was defined by at least one blood culture being positive for each on of the candida species, which was identified from the microbiological test logs. 11 The retrieved information included data on demographics, basic disease/condition, cause of current admission, and the service in which the patient was hospitalized, duration of hospital stay, clinical course, the presence of infective complications and outcome. Candidemic patients were routinely screened for candida endophthalmitis and infective endocarditis. Candidemia was considered as being a direct cause of death when it occurred within 1 week from patient demise. 7 Emphasis was placed on the identification of data pertaining to the presence of risk factors for systemic fungal infections in the 30 days before the detection of a positive cultures, i.e. the placement of central catheters, hyperalimentation, immunosuppression, chemotherapy, neutropenia (lowest absolute neutrophile count, 1000/mm 3 ), thrombocytopenia (, / mm 3 ), renal failure (creatinine.2 mg/dl). Statistical evaluation of the data was done using the BMDP statistical software. 12 The analysis of variance (ANOVA) and Pearson chi-square tests were used to compare clinical data to patient outcome, and stepwise logistic regression was used to quantitatively identify the relationship between the clinical parameters and outcome. Statistical significance was assigned at a P value less than Data are presented as mean ^ standard deviation. Results Patient demographics Demographic, risk factors and clinical data on a cohort of 186 patients with a M/F ratio of 60%/40% and mean age of 48 years (range 5 days 91 years) with candidemia are presented in Table 1. Mortality The overall mortality of our series was 79 (42%), with a similar distribution being noted throughout the study period. The direct reasons for death included candidemia or its complications (meningitis, endocarditis, hepatitis) in 14 (17.5%), septic shock in 51 (65%) (of which 35 (70%) was caused by polymicrobial infection), cardiogenic shock in 5 (6.3%), complication of malignancy in 5 (6.3%), prematurity in 1 (1.3%) and other reasons 3 (3.6%). Risk factors Duration of neutropenia was 9 days ^ 8.4 (range 1 43). No differences were found between the prevalence of the following risk factors for fungemia (e.g., steroids, immunosuppression, broadspectrum antibiotics and hyperalimentation) throughout the studied years and as a function of the different medical services (i.e., medical, surgical and paediatric). Data on multivariate analysis are presented in Table 2. Stay in the intensive care unit (ICU), thrombocytopenia, creatinine, and the need for mechanical ventilation or inotropic support were found as predictors of mortality in candidemic patients. Data on patients distribution on the hospital wards as a function of the years and medical services are presented in Table 3. In the latter years of the study, fewer patients were hospitalized in the ICU and a larger percentage of the patients were cared for in the hemato-oncology departments.

3 Predictors of adverse outcome from candidal infection in a tertiary care hospital 319 Table 1 Demographic, risk factors and clinical data on a cohort of 186 patients with candidemia Variables Total no. Mortality OR 95% CI P value Population 186 (100%) 79 (42%) Age 48 year (5 days 91 years) Male/female 112/74 (60%/40%) 43/36 (42%/43%) Broad-spectrum anti-bacterial chemotherapy 164 patients (88.6%) 73 (44%) Chemotherapy 35 (19%) 11 (31%) Neutropenia 26 (14%) 10 (37%) Steroids 48 (26%) 23 (48%) Immunosuppression 3 (1.6%) 0 (0%) Hyperalimentation 60 (32%) 30 (50%) Prior surgery 76 (41%) 32 (42%) ICU stay 103 (70%) 52 (50%) Thrombocytopenia 37 (20%) 24 (65%) Renal failure 26 (14%) 18 (70%) Ventilatory support 102 (59%) 55 (54%) Inotropic support 58 (31%) 45 (78%) Central venous catheters 125 (67%) 57 (46%) Arterial lines 86 (46%) 47 (55%) Anti-fungal therapy 110 (59%) 31 (29%) Time to initiation of anti-fungal therapy 2.9 days (1 10 days) Treatment initiation 2 days (1 4 days) Treatment duration 13 days (1 45 days) Amphotericin B 96 (52%) 34 (35%) Fluconazole 14 (7.5%) 4 (29%) The P value refers to univariate statistical analysis of categorical variables. OR, odds ratio; CI, confidence interval. Treatment Of the 76 patients who received no anti-fungal treatment, 48 (63%) died compared to 38 (34%) of 110 patients who were treated ðp, 0:05Þ: No statistically significant correlation was found between time of blood sampling, treatment initiation and duration and mortality. A retrospective review of the medical charts revealed no reason for failing to administer antifungal therapy. The policy for giving anti-fungal therapy was found to vary significantly between wards: the majority of the patients with candidemia (73%) received no anti-fungal therapy in the internal medicine service, while amphotericin B was used more frequently in the ICU ðp, 0:05Þ; and fluconazole was used more frequently in the Table 2 Multivariate logistic regression predicting mortality in candidemic patients ðp, 0:05Þ Population and factor analysed OR 95% CI Intensive care unit Creatinine Thrombocytopenia Mechanical ventilation Inotropic support OR, odds ratio; CI, confidence interval. hemato-oncology service including its prophylactic administration ðp, 0:05Þ: Because of the failure of the initial anti-fungal therapy, additional drugs were given to 17 patients (amphotericin B to 3 patients, fluconazole to 10 and 5-flucytosine to 4). Intravascular catheters were removed in 100/125 (80%) patients. Data on cause of admission and the distribution of mortality are presented in Table 4. The relative distribution over time of candida in the hospital at large showed a trend toward a lower rate of candida isolation in the medical and surgical wards as well as in the ICUs, with the exception of the hematooncology department where an increased rate of candida isolation was observed over the years (Table 3). In addition, there was a steady rise in the frequency of bone marrow transplantation among patients with candidemia. Microbiology The candida species isolated from patients with candidemia and the related mortality are presented in Table 5. The frequency of candidemia due to nonalbicans candida species increased significantly throughout the study period (Fig. 1). Non-albicans candida species were identified in the blood cultures of 33 (41%) of the non-survivors. Whereas fungemia due to C. albicans was present in 61% of the patients in the first 2 years of the study period,

4 320 R. Ben-Abraham et al. Table 3 Clinical subspecialties of candidemic patients as a function of the years of the study Year of the study No (mortality) Internal medicine a Surgical Paediatric ICU b HO BMT (11%) 5 (24%) 5 (24%) 3 (14%) 8 (38%) 0 (0%) 0 (0%) (16%) 5 (17%) 4 (14%) 5 (17%) 13 (45%) 2 (7%) 0 (0%) (20%) 3 (8%) 13 (36%) 1 (3%) 15 (42%) 3 (8%) 0 (0%) (22%) 8 (20%) 7 (17%) 3 (7%) 19 (46%) 4 (10%) 1 (2%) (16%) 5 (16%) 11 (35%) 1 (3%) 7 (23%) 6 (19%) 1 (3%) (15%) 11 (40%) 1 (4%) 2 (7%) 6 (21%) 6 (21%) 2 (7%) Sum (Mortality) 20 (54%) 10 (24%) 2 (13%) 42 (62%) 4 (19%) 1 (25%) HO, hemato-oncology; ICU, intensive care unit; BMT, bone marrow transplant. a Including geriatric and neurologic. b Respiratory, cardiac, paediatric, neurosurgical and neonatal. its relative frequency dropped to 43% during the last 2 years and there was a parallel increase in the rate of infection with C. tropicalis (from 14 to 40%). This trend was well demonstrated among the hematooncology patients where 33% of the infections were caused by C. tropicalis. Multiple candida species were identified in the blood cultures of four patients (2.1%), all of whom survived after being treated with amphotericin B. The additional candidal species were C. tropicalis 2, and parapsilosis 2. Ninety patients (48%) had polymicrobial sepsis (positive blood cultures for candida and another microbial agent). These included Gram-positive cocci (Staphylococcus aureus, epidermidis and Streptococcus) in 49 cases (54.5%) and aerobic Gram-negative bacilli (Klebsiella pneumoniae, Pseudomonas aeruginosa and Escherichia coli) in 41 cases (45.5%). Discussion Our data support previous publications on the rising trend of candidal bloodstream infection among Table 4 Reasons for hospital admission, type of medical service used and distribution of overall crude mortality of 186 patients with candidemia Reason for admission No. (%) of patients Mortality Total no. 186 (100%) 79 (42%) Malignancy 78 (42%) 25 (32%) Infection a 25 (13%) 16 (64%) Cardiac surgery 18 (10%) 8 (44%) Other surgical 18 (10%) 11 (61%) Other internal 18 (10%) 9 (50%) Prematurity 14 (7%) 6 (43%) Burns 9 (5%) 4 (44%) Multiple trauma 6 (3%) 0 (0%) a Infection was defined as clinical evidence for active infective process (i.e., pneumonia, skin infection, sepsis). severely ill patients. 2,13 Indeed, hemato-oncology malignancies or infections were the leading causes of patients admission in our series. These patients were inevitably exposed to multiple risk factors for fungemia, such as steroids, immunosuppression, broad-spectrum antibiotics and hyperalimentation. Our crude mortality rate from candidemia (42%) is similar to that of other reports. 14,15 Sepsis from unknown sources was the most frequent cause of death (65%), whereas candidemia was found to be a primary cause of mortality in only 17.5% of the cases. These results support the fact that candidemia is sometimes only a marker of disease severity in critically ill patients. In fact, superinfection with candida was reported to occur in as many as 12% of blood samplings in critically ill patients. 16 The higher mortality among the untreated patients group as compared to the treated ones is similar to what had already been reported on the lethality of candidemia. 8,17 An expectant approach, i.e. removal of a central catheter, following the finding of positive blood culture for candida, without the immediate initiation of anti-fungal therapy was the common treatment policy in our institution. The anti-fungal therapy was initiated only after repeated positive blood cultures confirmed the presence of candida. This practice was recently changed according to the international consensus on practice guidelines for the treatment of candidiasis which recommended both immediate empirical start of anti-fungals and removal of central venous access. 9 Our current data point to a high mortality rate associated with positive blood cultures for candida, thus establishing a strong case to establish a policy of immediately initiating anti-fungal therapy when positive blood cultures of candidemia are identified for the first time. This is further supported by the facts that it is difficult both to predict which patient will develop serious consequences of the infection and to differentiate between transient or sustained

5 Predictors of adverse outcome from candidal infection in a tertiary care hospital 321 Table 5 Candida species isolated from patients with candidemia as a function of the study year ðn ¼ 186Þ Species isolated No. (%) of candidemic episodes Mortality Total no. of candidemic episodes 190 (100%) 79 (42%) Albicans 103 (54%) 46 (45%) a Tropicalis 41 (22%) 15 (37%) Parapsilosis 25 (13%) 5 (20%) Glabrata 15 (8%) 11 (73%) a Krusei 6 (3%) 2 (33%) a P, 0:05: candidemia. In addition, delays in microbiological diagnosis sometimes occur because none of the non-culture methods (serological assays) are adequately sensitive or specific. 18 The mean number of days (2.9) from the onset of candidemia until the initiation of therapy in our investigation was similar to the figures reported by Viudes et al. 17 However, while they found increased mortality among patients whose candidemia lasted longer than 2 days, we did not find that a delay of a few days in initiating treatment to be related to increased mortality. Analysis of our data revealed that fluconazole was the favored first-line anti-fungal agent outside the intensive care facilities. Its good bio-availability when given orally and ease of administration compared to amphotericin B are advantages for planning anti-fungal therapy in the set-up of the general ward. Amphotericin is still regarded as the gold standard anti-fungal agent despite its widely known serious allergic side effects and renal toxicity. Although there are data pointing to equal utility of fluconazole in non-high risk patients, 19 it must be kept in mind that some candida species, such as glabrata or krusei, may be intrinsically resistant to fluconazole. 20 Thus, careful identification of the specific strains and their susceptibility should be undertaken in the laboratory. In these cases, prompt institution of amphotericin B treatment is live saving, especially in high-risk patients for whom another fungal infection such as Aspergillus spp., which is not sensitive to fluconazole can be present. 6,21 The highest mortality was observed in patients suffering from the glabrata species of candidemia, as had been described by Nguyen et al. 10 The species distribution in our patients and the shift in time towards non-albicans species probably result from a combination of the effects of the nosocomial flora of our hospital and the increase in the number of immunocompromised hosts or the use of prophylactic anti-fungal agents (i.e., fluconazole) in the various hospital wards especially the hematooncology service. Similar trend toward fewer Figure 1 Distribution of candida strains throughout the study period. Y-axis, percentage of patients with candidemia, X-axis, years of the study.

6 322 R. Ben-Abraham et al. fungemic episodes due to C. albicans and more due to non-albicans species was recently reported by others. 22,23 For example, Malani et al. 22 in their survey of yeast causing fungemia over a 12-year period at a large tertiary medical center. In their report on a total of 966 unique episodes of fungaemia, the trend toward increased isolation of non-albicans species was attributed to increase use of fluconazole. 22 Clearly, further comparative research is needed to better define candida species flora in view of the associated high mortality and complication rate. 10 Candidal infection can cause disseminating disease. 24,25 The small number of patients who suffered from systemic spread of the infection is probably an underestimation since no post-mortem examination was done. It is reasonable to assume that some of the patients who probably had developed systemic complications of candida had died before these were clinically manifested thus, preventing an accurate diagnosis to be made antemortem. Indeed, the report by Prescott et al. stressed the importance of necropsy in the clinicopathological audit of deaths in this group of patients. 26 In addition, systemic complications following candidemia-like endophthalmitis or osteomyelitis resulting from hematogenous spread of the fungus can appear up to 2 years following the first identification of candidemia. 27 In our study, the failure to conduct long-term surveillance probably led to these cases being missed in our data bank. One-third of our studied population was hospitalized in one of the ICUs. ICU hospitalization was identified as a risk factor for mortality from candidemia, together with thrombocytopenia and impaired renal function, all pointing to patient disease complexity, which mandates invasive therapeutic interventions. 28,29 Indeed, mechanical ventilation and inotropic support were also identified as predictors of mortality. The increased awareness of the medical team and better patient monitoring in the ICU have undoubtedly contributed to higher detection rates of candidemia in this patient group. When ICU admission rates were analysed as a function of time, a trend towards a decrease rate of ICU hospitalization in favor of the hemato-oncology wards was revealed. We hypothesized that this trend was due to improved rationing and cost-containment considerations on the part of the medical team. 30 The establishment of clinical predictors for outcome that would be capable of differentiating between survivors and non-survivors among candidemic patients would facilitate the selection of proper therapies without additional morbidity potential. Several limitations of the current study need to be considered. The main limitation of our analysis is that, due to the retrospective nature of the study, relatively limited clinical information was collected during the course of hospitalization, preventing us from analysing our data according to the denominator of per 1000 hospital admission. Furthermore, we were unable to evaluate the role of specific treatment factors in patient outcome, such as changes in drug therapy. Although a fairly large population was examined, the non-randomization of innovative techniques makes it difficult to evaluate their role in patient survival. Conclusion Non-albicans candidemia carries high mortality, especially if not treated. The validity of the predictors of mortality identified from our data should be further assessed, focusing upon their ability to predict outcome from non-albicans candidemia in view of the growing frequency of infections caused by these species. This will help to establish a better plan of species-targeted therapy for minimizing systemic complications and multiorgan involvement and for improving prognosis. Acknowledgements We thank Esther Eshkol for editorial assistance. References 1. Blumberg HM, Jarvis WR, Soucie JM, Edwards JE, Patterson JE, Pfaller MA, Rangel-Frausto MS, Rinaldi MG, Saiman L, Wiblin RT, Wenzel RP, National Epidemiology of Mycoses Survey (NEMIS) Study Group. Risk factors for candidal bloodstream infections in surgical intensive care unit patients: the NEMIS prospective multicenter study. The National Epidemiology of Mycosis Survey. Clin Infect Dis 2001;33: De Marie S. New developments in the diagnosis and management of invasive fungal infections. Haematologica 2000;85: Nolla-Salas J, Sitges-Serra A, Leon-Gil C, Martinez-Gonzalez J, Leon-Regidor MA, Ibanez-Lucia P, Torres-Rodriguez JM. Candidemia in non-neutropenic critically ill patients: analysis of prognostic factors and assessment of systemic antifungal therapy. Study Group of Fungal Infection in the ICU. Intensive Care Med 1997;23: Alvarez-Lerma F, Nolla-Salas J, Leon C, Palomar M, Jorda R, Carrasco N, Bobillo F, EPCAN Study Group. Candiduria in

7 Predictors of adverse outcome from candidal infection in a tertiary care hospital 323 critically ill patients admitted to intensive care medical units. Intensive Care Med 2003;29: De Pauw BE, Meis JF. Progress in fighting systemic fungal infections in haematological neoplasia. Support Care Cancer 1998;6: Bohme A, Karthaus M. Systemic fungal infections in patients with hematologic malignancies: indications and limitations of the antifungal armamentarium. Chemotherapy 1999;45: Wey SB, Mori M, Pfaller MA, Woolson RF, Wenzel RP. Risk factors for hospital-acquired candidemia: a matched case control study. Arch Intern Med 1989;149: Ostrosky-Zeichner L. New approaches to the risk of Candida in the intensive care unit. Curr Opin Infect Dis 2003;16: Rex JH, Walsh TJ, Sobel JD, et al. Practice guidelines for the treatment of candidiasis. Clin Infect Dis 2000;30: Nguyen MH, Peacock Jr JE, Morris AJ, et al. The changing face of candidemia: emergence of non-candida albicans species and antifungal resistance. Am J Med 1996;100: Knoke M, Bernhardt H, Schulz K, Schroder G, Zimmermann K. Funguria and Candida-specific immunoglobulins in patients with systemic candidosis. Mycoses 2000;43: Dixon WJ, editor. BMDP statistical software. Berkeley, CA: University of California Press; Lewis RE, Klepser ME. The changing face of nosocomial candidemia: epidemiology, resistance and drug therapy. Am J Health Syst Pharm 1999;56: Gudlaugsson O, Gillespie S, Lee K, Vande Berg J, Hu J, Messer S, Herwaldt L, Pfaller M, Diekema D. Attributable mortality of nosocomial candidemia, revisited. Clin Infect Dis 2003; 37: Charles PE, Doise JM, Quenot JP, Aube H, Dalle F, Chavanet P, Milesi N, Aho LS, Portier H, Blettery B. Candidemia in critically ill patients: difference of outcome between medical and surgical patients. Intensive Care Med 2003;29: Telenti A, Steckelberg JM, Stockman L, Edson RS, Roberts GD. Quantitative blood cultures in candidemia. Mayo Clin Proc 1991;66: Viudes A, Peman J, Canton E, Ubeda P, Lopez-Ribot JL, Gobernado M. Candidemia at a tertiary-care hospital: epidemiology, treatment, clinical outcome and risk factors for death. Eur J Clin Microbiol Infect Dis 2002;21: Walsh TJ, Hathorn JW, Sobel JD, et al. Detection of circulating candida enolase by immunoassay in patients with cancer and invasive candidiasis. N Engl J Med 1991;324: Rex JH, Bennett JE, Sugar AM, et al. A randomized trial comparing fluconazole with amphotericin B for the treatment of candidemia in patients without neutropenia. N Engl J Med 1994;331: Wingard JR. Importance of candida species other than Candida albicans as pathogens in oncology patients. Clin Infect Dis 1995;20: Wingard JR, Merz WG, Rinaldi MG, Johnson TR, Karp JE, Saral R. Increase in Candida krusei among patients with bone marrow transplantation and neutropenia treated prophylactically with fluconazole. N Engl J Med 1991;325: Malany PN, Bradley SF, Little RS, Kauffman CA. Trends in species causing fungaemia in a tertiary care medical center over 12 years. Mycoses 2001;44: Diekema DJ, Messer SA, Brueggemann AB, Coffman SL, Doern GV, Herwaldt LA, Pfaller MA. J Clin Microbiol 2002;40: Maksymiuk AW, Thongprasert S, Hopfer R, Luna M, Fainstein V, Bodey GP. Systemic candidiasis in cancer patients. Am J Med 1984;77: Gregory RK, Powles RL, Treleaven JG, et al. Systemic candidiasis with candida vasculitis due to Candida kruzei in a patient with acute myeloid leukemia. Bone Marrow Transplant 1999;23: Prescott RJ, Harris M, Banerjee SS. Fungal infections of the small and large intestine. J Clin Pathol 1992;45: Bodey GP. Hematogenous and major organ candidiasis. In: Bodey GP, editor. Candidiasis: pathogenesis, diagnosis and treatment. 2nd ed. New York: Raven Press Ltd; p Anaissie EJ, Rex JH, Uzun O, Vartivarian S. Predictors of adverse outcome in cancer patients with candidemia. Am J Med 1998;104: Voss A, le Noble JL, Verduyn Lunel FM, Foudraine NA, Meis JF. Candidemia in intensive care unit patients: risk factors for mortality. Infection 1997;25: Bion JB. Cost containment: Europe, the United Kingdom. New Horiz 1994;2:341 4.