Length of Stay and Mortality Associated With Febrile Neutropenia Among Children With Cancer
http://www.100md.com
《临床肿瘤学》
the University of Rochester Medical Center, Rochester, NY
ABSTRACT
PURPOSE: The aim of this study was to evaluate risk factors for longer length of stay (los) and mortality among hospitalized children with cancer who have febrile neutropenia.
METHODS: This study involved analysis of longitudinal data from the University HealthSystem Consortium database from 1995 to 2002. All patients who were 21 years or younger, with diagnostic codes for both neoplastic disease and febrile neutropenia at discharge, were included.
RESULTS: A total of 12,446 patients were identified for the study. The los was 5 days or less for 6,799 patients, and greater than 5 days for 5,647 patients. The mortality rate was 3%. On bivariate analysis, race, age, cancer type, and associated complications (bacteremia/sepsis, hypotension, pneumonia, and fungal infections) were significantly associated with longer length of stay and death. On multivariate analysis, age group, race, cancer type (acute myeloid leukemia, multiple cancers v acute lymphoblastic leukemia), and the complication variables were significantly associated with increased risk of longer los and death. Certain types of cancer (Hodgkin's disease, osteosarcoma/Ewing’s sarcoma, rhabdomyosarcoma, compared with acute lymphoblastic leukemia) and year of discharge after 1995 were significantly associated with a reduced risk of longer length of stay and/or mortality.
CONCLUSION: Race, age group, year of discharge, associated complications, and cancer type were significantly associated with risk of longer los and mortality. These factors may potentially help in identifying high-risk patients who might benefit from targeted antibiotic therapy or prophylactic hematopoietic growth factor support.
INTRODUCTION
Febrile neutropenia is a common complication of therapy among children with cancer. In recent years, several studies have evaluated risk factors for bacteremia or poor outcomes among patients with cancer and helped to establish the current guidelines for the treatment of febrile neutropenia by the Infectious Disease Society of America (IDSA), as well as the recommendations for the use of hematopoietic colony-stimulating factors by the American Society of Clinical Oncology (ASCO).1-4 However, not all guidelines are generalizable to children with cancer because of inadequate data in this population. Studies on the pediatric population are rendered difficult by the limited numbers of pediatric patients with cancer treated at a single institution. It is also problematic to study mortality as an outcome because the mortality associated with febrile neutropenia at any institution is usually small. Therefore, other surrogate markers for severity of disease are used, such as bacteremia or invasive bacterial infection, which are defined differently in various studies and therefore difficult to compare.5-8
The overall goal of this study was to better understand factors associated with outcomes of febrile neutropenia and its consequences in children. This study was a retrospective analysis of clinical data on pediatric patients from a national hospitalization database.
METHODS
Study Design
This study was designed as an analysis of prospectively collected data included in a hospitalization database maintained by the University HealthSystem Consortium (UHC). The major outcomes of interest were length of stay and mortality. The major independent variables of interest were age, sex, race, year of discharge, histopathologic diagnosis, and associated complications.
Source of Data
The UHC database contains information from discharge summaries of all patients with cancer and febrile neutropenia who were admitted between 1995 and 2002 to 115 member academic institutions in the United States. Because children with cancer are usually treated at tertiary care academic centers, the subjects from this database were considered to constitute a representative population of children with febrile neutropenia. All patients included in the analysis had diagnosis codes for both neoplastic disease (International Classification of Diseases, 9th revision, clinical modification [ICD-9-CM] code 140-239) and febrile neutropenia, and were 21 years or younger. Patients with a diagnostic code for bone marrow transplantation were excluded from the study. For those patients who had more than one admission for febrile neutropenia, only one hospitalization was selected randomly for the purpose of this analysis.
Statistical Analysis
The data set included information on age, sex, race, year of discharge, diagnostic codes at discharge (up to 16 diagnostic codes), length of hospitalization, and discharge status. The primary outcomes of this analysis were length of stay and mortality in the hospital, which served as markers for severity of illness. Based on the ICD-9-CM codes for diagnosis, subjects were classified as having acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), Hodgkin's disease (HD), non-Hodgkin's lymphoma (NHL), osteosarcoma and Ewing's sarcoma, rhabdomyosarcoma and other soft tissue sarcomas, neuroblastoma and renal tumors, hepatic tumors, and ophthalmic and ovarian/testicular tumors. Subjects who had a diagnostic code for cancer but who did not fit in any of the above cancer categories were classified as having "other" cancers. Subjects with diagnostic codes for more than one of these cancer categories were classified as having multiple cancers. Using the ICD-9-CM codes as listed in Table 1, additional variables were created to identify subjects with bacteremia/sepsis, other bacterial infections, hypotension, pneumonia, and fungal infections. Subjects who died during the hospitalization were identified using the discharge status.
The initial step in the analysis involved obtaining descriptive statistics on all dependent and independent variables. The bivariate association between each outcome variable (length of stay and mortality) and each of the independent variables was then evaluated. Length of stay was used as a dichotomous variable ( 5 days and > 5 days) for the purpose of the initial bivariate analyses. The cutoff point of 5 days for the purpose of the analysis was determined before the data analysis and was based on the clinical expectation that a patient who was admitted for less than 5 days most likely represented one who had an uncomplicated course in the hospital.
For the multivariate analysis, a three-level categorical outcome variable was defined to include patients who were discharged alive with a length of stay less than or equal to 5 days, patients who were discharged alive with a length of stay longer than 5 days, and patients who died. Variables found to have a significant association with the outcome variable in the bivariate analysis and potential confounders were included in the multivariate analysis. A step-wise approach was used to arrive at the final model, using nominal logistic regression with a generalized logit model. SAS software version 8 (SAS Institute Inc, Cary, NC) was used to run all the statistical analyses.
RESULTS
Predictor Variables
A total of 12,446 patients were identified for the study using the criteria described in the Methods section. The sample consisted of 7,045 (57%) male and 5,401 (43%) female patients. Of the 12,446 patients in the study, 8,349 (67%) subjects in the study were white, 1,262 (10%) were African American, 1,259 (10%) were Hispanic, 342 (3%) were Asian, and 1,234 (10%) were categorized as representing other racial/ethnic groups. In the overall study population, 502 (4%) of the subjects were infants, 7,563 (61%) were children between the ages of 1 and 12 years, and 4,377 (35%) subjects were older than 12 years of age. The median age of subjects in the study was 8 years (range, 0 to 21 years). The number of subjects entered each year from 1995 to 2002 ranged from 1,224 to 1,853 representing 10% to 15% of the study sample (data not shown). The distribution of cancer types is shown in Figure 1. ALL was the most common type of cancer and occurred in 34% of subjects. There were 1,988 (16%) patients with a diagnostic code for bacteremia or sepsis, 251 (2%) subjects with hypotension, 678 (5%) subjects with pneumonia, 1,150 (9%) subjects with other bacterial infections, and 1,056 (8%) with fungal infections. There were 2,598 subjects with one complication, and 1,157 subjects with multiple complications.
Outcome Variables
Length of stay. The median length of stay for all patients in the study was 5 days (range, 1 to 359 days). The study sample consisted of 6,799 (55%) patients with a length of stay of 5 days or less and 5,647 (45%) patients with length of stay longer than 5 days (as presented in Table 2).
The mean and median length of stay for the patients admitted for 5 days or less were the same (3 days), with a range of 1 to 5 days. The mean and median length of stay for the patients admitted for longer than 5 days were 19 days and 12 days, respectively, with a range of 6 to 359 days. As seen in Table 2, the racial/ethnic composition of subjects hospitalized for longer than 5 days was significantly different from those with a shorter length of stay. Specifically, subjects in each of the nonwhite categories were more likely to require a longer hospital stay than white patients. Age group also emerged as a significant risk factor for longer length of hospital stay. Adolescents were found to have a greater likelihood of being hospitalized for longer than 5 days as compared with infants and children. Subjects with a diagnosis of AML were found to be at greater risk of being hospitalized for longer than 5 days as compared with subjects with other cancer diagnoses. Subjects with an associated diagnosis of bacteremia/sepsis, hypotension, pneumonia, and other bacterial or fungal infections were also significantly more likely to have a longer length of stay as compared with subjects without these associated diagnoses (Table 2).
Mortality. Three hundred fifty-eight subjects in the study died, leading to a mortality rate of 3% (see Fig 2). Subjects who were white were less likely to die compared with subjects in each of the nonwhite categories (Table 3). Infants and adolescents were found to be at a greater risk of death than children. A diagnosis of AML or multiple cancers was associated with an increased risk of death as compared with other cancer types, as seen in Table 3. A diagnosis of bacteremia/sepsis significantly increased the risk of death: 9% of patients with bacteremia died, as compared with 2% mortality rate among patients without bacteremia. The incidence of bacteremia among patients who died was 50% as compared with 15% among survivors. Similarly, hypotension, pneumonia, and fungal infections were significantly associated with an increased risk of death (Table 3).
Multivariate Analysis
For the multivariate analysis, a three-level categoric outcome variable was used as described in the methods section. The group of patients who were discharged alive with a length of stay of 5 days was used as the reference group. Odds ratios (ORs) were obtained for each of the two remaining categories of outcome (patients who were discharged alive with length of stay longer than 5 days and patients who died) as compared with the patients who were discharged alive with a length of stay of 5 days or less. Using nominal logistic regression with a generalized logit model, year of discharge, race, age group, the associated complication diagnoses, and cancer type were significantly associated with risk of longer length of stay and/or death (Table 4).
With each consecutive year after 1995, the risk of longer length of stay decreased by 3% (OR = 0.97), and the risk of death decreased by 7% (OR = 0.93), after accounting for all the other variables in the model. The risk of both longer length of stay and death was consistently high in each of the other nonwhite categories as compared with white patients. Infants were found to have a significantly increased risk of death (OR = 1.75) as compared to children between the ages of 1 and 12 years. Adolescents were at a significantly higher risk of both longer length of stay (OR = 1.44) and death (OR = 3.05) as compared with children aged 1 to 12 years. The associated complications of bacteremia/sepsis, hypotension, pneumonia, and fungal infection were all significantly associated with longer length of stay and death. Of note, patients with bacteremia/sepsis had an almost 10-fold increase in the risk of death as compared to those without a diagnosis of bacteremia/sepsis. Cancer type was also found to be an independent predictor of both longer length of stay and death. As compared with ALL, patients with AML and multiple cancers had a significantly greater risk of both longer length of stay (OR = 2.86 and 1.20, respectively) and death (OR = 2.88 and 1.48, respectively). Patients with Hodgkin's disease, osteosarcoma/Ewing sarcoma and rhabdomyosarcoma were at a significantly lower risk of both longer length of stay and death as compared with subjects with ALL. The risk of longer length of stay was also significantly lower for subjects with neuroblastomas/kidney tumors and ovarian/testicular/ophthalmic tumors compared with ALL.
A secondary analysis was performed using mortality as a dichotomous variable (dead v alive) in a multivariate model to determine the risk of death as compared with being alive, for each predictor variable. Year of discharge, race, age group, cancer type, presence of sepsis, hypotension, pneumonia, other bacterial and fungal infections were again all found to be significantly associated with the risk of death (data not shown), as seen with the 3-category outcome. Each successive year after 1995 was associated with a significantly lower risk of death. All the nonwhite race/ethnicity groups were found to be at an increased risk of death. With regard to age group, adolescents were also found to have a significantly increased risk of death. However, infancy was not significantly associated with an increased risk of death (OR = 1.66, P = .0653). Among the cancer types, Hodgkin's disease, osteosarcoma and Ewing sarcoma, and rhabdomyosarcoma were associated with a significantly reduced risk of death. A diagnosis of AML was found to have an increased risk of death (OR = 1.34), but this was not statistically significant (P = .0667).
DISCUSSION
In recent years, there has been increasing interest in stratifying patients with febrile neutropenia into risk categories to explore various therapeutic strategies.2,3,9 The current study uses a multicenter database to identify risk factors for longer length of stay and death.
Death is the most devastating of all potential outcomes of febrile neutropenia. Yet, most studies of febrile neutropenia use bacteremia or bacterial infection as the main outcome variable because infection is the expected complication in a neutropenic patient.5-8 However, the availability of potent broad-spectrum antibiotics has made most bacterial infections curable today. Despite the widespread approach of treating all febrile neutropenic patients with empiric broad-spectrum antibiotics for potential bacteremia, febrile neutropenia in the child with cancer remains associated with a significant, albeit small, mortality rate. This suggests that certain host or disease-related factors may play a role in influencing mortality resulting from febrile neutropenia. Thus, mortality seems to be the more appropriate outcome for evaluation in the modern era. In the current study, the large sample in the database presented a remarkable opportunity to evaluate mortality associated with febrile neutropenia. It must be noted, however, that although all the patients in the study had febrile neutropenia during their course in the hospital, the cause of death may have been refractory cancer or other diagnoses with or without infection.
In this study, length of stay was also evaluated as an outcome of interest for several reasons. Cancer care currently accounts for approximately 10% of all health care expenditure in the United States.10,11 A substantial proportion of the health care costs incurred by patients with cancer can be attributed to the treatment of febrile neutropenia. A recent study from the Hospital for Sick Children, Toronto, Canada reported fever with neutropenia as one of the most common admitting diagnoses among pediatric oncology patients, second only to admissions for chemotherapy.8 It is commonly known that costs of hospitalization increase with longer length of stay. In this study, length of stay also served as a surrogate for severity of illness. In addition, evaluation of length of stay is important because longer periods of hospitalization impact the quality of life of affected patients.
The mortality rate in this study was approximately 3%. It must be noted that this figure is derived from patients who were admitted to the hospital with febrile neutropenia, as well as patients who may have been admitted for another medical condition and subsequently developed febrile neutropenia while in the hospital. The mortality rate in some recent pediatric studies of febrile neutropenia has ranged from 0.7% to 3.9%.5,7,8,12
In the current study, race, age group, year of diagnosis, cancer type, and associated complications were significantly associated with longer length of stay (greater than 5 days) and death.
As compared with white patients, all nonwhite groups were found to have a higher risk of both longer length of stay and death. To the best of our knowledge, this association has not been reported previously among patients with febrile neutropenia. There are studies that have evaluated the relationship between racial/ethnic groups and survival among children diagnosed with ALL, with conflicting results.13 Analyses of data from the two large children's cooperative groups have found differences in survival between patients of different races, with inferior survival among African American and Hispanic children.14,15 However, successive studies from a single large institution have not found any differences in survival across racial groups.16,17 With regard to febrile neutropenia, race may be a surrogate for social class or a confounder for an underlying association between socioeconomic status and poor outcomes. Additional information regarding socioeconomic status was not available in the current study to evaluate this relationship further. Alternatively, it is possible that there is a true biologic susceptibility conferred by race and, given the large sample size, the current study had adequate power to uncover the association between race and poor outcomes with febrile neutropenia. Therefore, we suggest that race should be taken into account while considering alternative treatment strategies in children with febrile neutropenia.
Age group emerged as another significant independent risk factor for both longer length of stay and death. In studies of adults with febrile neutropenia, older age has emerged as a significant risk factor for poor outcomes.18,19 Although infancy has been cited as a high-risk pediatric age group, most studies have very few infants and young children to evaluate this relationship.12,20-22 In the current study, the risk of death was significantly higher in infants as compared with children aged 1 to 12 years. Adolescence was also found to be a high-risk age group, with an increased risk of both longer length of stay and death. After controlling for race, year of discharge, cancer type, and complications, adolescents were almost 1.5 times as likely to be hospitalized longer and more than three times as likely to die as children aged 1 to 12 years. As in the case of race, it is not clear whether this reflects an underlying biologic susceptibility during this period. It is possible that adolescents are more reluctant to seek care for what they may perceive as an innocuous episode of fever, present late in the course of a febrile neutropenic episode and therefore develop more severe complications. It is also possible that there are differences in treatment received by adolescents, with some patients receiving care from pediatric oncologists and others being cared for by adult oncology teams.
Year of discharge was found to have an independent association with both longer length of hospital stay and death. Using the initial year (1995) as the comparison group, the risk of longer length of stay decreased by approximately 3% with every passing year. This is not surprising, given the trend toward early discharge in almost every sphere of medicine in recent years. However, it is noteworthy that there has been a significant decrease in the risk of death over the same period, by approximately 7% per year. Perhaps, this may be attributable to better supportive care. The role of granulocyte colony-stimulating factor in this decline in mortality would be interesting to evaluate. Unfortunately, this information was not available for this study. Of note, a previous study of trends in infection morbidity in a pediatric oncology ward throughout a 10-year period did not find a decrease in the frequency of febrile episodes or fatality of infections after recombinant hematopoietic growth factors became available.23
For the evaluation of risk by cancer type, ALL was used as the reference category because it was the most frequently encountered cancer type in the data set and because most clinicians are familiar with the treatment and treatment-related complications of ALL. It should be noted that an odds ratio of less than 1 for any of the cancer types in this analysis indicates a lower risk of the outcome as compared to ALL; it does not denote a protective effect as there were no noncancer controls. In comparison to ALL, a diagnosis of AML was associated with a three-fold increased risk of longer length of stay as well as death. A diagnosis of Hodgkin's disease, osteosarcoma/Ewing sarcoma or rhabdomyosarcoma was associated with a significant reduction in risk of both longer length of stay and death. There was a fairly large "other cancer type" category, which comprised 12% of the study population and consisted of numerous diagnoses that were too small individually to be considered as distinct categories in the analysis. This "other" group, which included mostly miscellaneous solid tumors, did not have a significant difference in outcomes as compared to ALL. This is consistent with studies in adult populations that have identified the diagnosis of solid tumors as a low risk factor.2 Similarly, pediatric studies have identified leukemia or relapse of leukemia as high risk factors for bacterial infection.5,24
Finally, as expected, the complication diagnoses were significantly associated with both longer length of stay and death. A diagnosis of sepsis or bacteremia conferred a 10-fold increase in the risk of death. Thus, despite the widespread use and availability of powerful antibiotics, bacteremia/sepsis remains the most important independent prognostic marker for mortality. Pneumonia was associated with an eight-fold increase in the risk of death. Fungal infections were associated with a five-fold increase in the risk of death. Several complication variables were associated with a longer length of stay and death. These findings are consistent with previous reports that comorbid conditions are high risk factors in adult populations.2,18,25 Hypotension was reported as an independent high risk factor for invasive bacterial infection in a prospective pediatric study,5 while several other prospective pediatric studies of low-risk patients exclude subjects with comorbidities.8,20
This study involved a retrospective analysis of an administrative data set, and therefore has several limitations. The selection of the variables, coding used, and the methods used for recording the data were predetermined. Therefore, it was not possible to evaluate additional variables that may have been of interest, such as the absolute neutrophil or monocyte count at the time of admission and discharge, use of growth factors, and type of chemotherapy used. The analysis was also limited by the ICD-9-CM diagnostic codes listed at discharge. There is no information in the data set about whether patients were at diagnosis, in remission, or in a palliative care mode at the time of the hospitalization. Reasons for admission were also not available. The subjects in the database include patients who were admitted with febrile neutropenia, as well as patients who were hospitalized for another reason and developed febrile neutropenia during the course of the hospitalization. Further studies are necessary to specifically evaluate patients admitted with febrile neutropenia. The cause of death is not available. Thus, patients in this database may have died from causes other than febrile neutropenia. This could theoretically bias the study. Approximately 12% of the study population were found to have ICD-9-CM codes for more than one cancer category and were classified as having multiple cancers. Some of these patients may truly have more than one cancer, some may have treatment-related second malignant neoplasms. In several cases, this may represent a coding error or misinterpretation of the diagnosis. It was not possible to evaluate this issue further in the data set. These patients were analyzed in a separate category so as to avoid a potential misclassification bias that might result from their inclusion in only one of the major histopathologic categories.
On the other hand, use of a pre-existing data set offered the advantages of speed, economy and large sample size. Current protocols for the treatment of febrile neutropenia are based on findings of randomized trials. However, subjects enrolled on clinical trials are usually different from the general population. This data set includes all patients admitted to the hospital unselected by any criteria other than febrile neutropenia and cancer. Therefore, the findings may be more generalizable to the entire population.
In conclusion, the results of this study indicate that nonwhite race, infancy and adolescence, bacteremia/sepsis, other bacterial infections, hypotension, pneumonia, fungal infections, and a diagnosis of AML or multiple cancers should be considered high risk factors in children with febrile neutropenia. The identification of risk factors for poor outcomes may help in devising protocols for providing targeted therapy based on risk criteria and thus help to reduce the cost of cancer care as well as reduce the adverse effects of antibiotic therapy. The findings of this study may also help in devising updated guidelines for hematopoeitic growth factor support in neutropenic children with cancer. Prospective studies of febrile neutropenia among children enrolled in large cooperative trials may be beneficial in evaluating these risk factors further.
Authors' Disclosures of Potential Conflicts of Interest
Although all authors completed the disclosure declaration, the following author or immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
NOTES
Presented at the 40th Annual Meeting of the American Society of Clinical Oncology, June 5-8, 2004.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Pizzo PA: Management of fever in patients with cancer and treatment-induced neutropenia. N Engl J Med 328:1323-1332, 1993
Hughes WT, Armstrong D, Bodey GP, et al: 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer. Clin Infect Dis 34:730-751, 2002
Ozer H, Armitage JO, Bennett CL, et al: 2000 update of recommendations for the use of hematopoietic colony-stimulating factors: Evidence-based, clinical practice guidelines—American Society of Clinical Oncology Growth Factors Expert Panel. J Clin Oncol 18:3558-3585, 2000
Pizzo PA: Fever in immunocompromised patients. N Engl J Med 341:893-900, 1999
Santolaya ME, Alvarez AM, Becker A, et al: Prospective, multicenter evaluation of risk factors associated with invasive bacterial infection in children with cancer, neutropenia, and fever. J Clin Oncol 19:3415-3421, 2001
Santolaya ME, Alvarez AM, Aviles CL, et al: Prospective evaluation of a model of prediction of invasive bacterial infection risk among children with cancer, fever, and neutropenia. Clin Infect Dis 35:678-683, 2002
Rackoff WR, Gonin R, Robinson C, et al: Predicting the risk of bacteremia in childen with fever and neutropenia. J Clin Oncol 14:919-924, 1996
Klaassen RJ, Goodman TR, Pham B, et al: "Low-risk" prediction rule for pediatric oncology patients presenting with fever and neutropenia. J Clin Oncol 18:1012-1019, 2000
Orudjev E, Lange BJ: Evolving concepts of management of febrile neutropenia in children with cancer. Med Pediatr Oncol 39:77-85, 2002
Parker SL, Tong T, Bolden S, et al: Cancer statistics, 1996. CA Cancer J Clin 46:5-27, 1996
Vincenzino JV: Health care costs: Market forces and reform. Oncology (Huntingt) 9:367-368, 371, 374, 1995
Baorto EP, Aquino VM, Mullen CA, et al: Clinical parameters associated with low bacteremia risk in 1100 pediatric oncology patients with fever and neutropenia. Cancer 92:909-913, 2001
Bhatia S: Influence of race and socioeconomic status on outcome of children treated for childhood acute lymphoblastic leukemia. Curr Opin Pediatr 16:9-14, 2004
Bhatia S, Sather HN, Heerema NA, et al: Racial and ethnic differences in survival of children with acute lymphoblastic leukemia. Blood 100:1957-1964, 2002
Pollock BH, DeBaun MR, Camitta BM, et al: Racial differences in the survival of childhood B-precursor acute lymphoblastic leukemia: A Pediatric Oncology Group Study. J Clin Oncol 18:813-823, 2000
Pui CH, Boyett JM, Hancock ML, et al: Outcome of treatment for childhood cancer in black as compared with white children: The St Jude Children's Research Hospital experience, 1962 through 1992. JAMA 273:633-637, 1995
Pui CH, Sandlund JT, Pei D, et al: Results of therapy for acute lymphoblastic leukemia in black and white children. JAMA 290:2001-2007, 2003
Klastersky J, Paesmans M, Rubenstein EB, et al: The Multinational Association for Supportive Care in Cancer risk index: A multinational scoring system for identifying low-risk febrile neutropenic cancer patients. J Clin Oncol 18:3038-3051, 2000
Morrison VA, Picozzi V, Scott S, et al: The impact of age on delivered dose intensity and hospitalizations for febrile neutropenia in patients with intermediate-grade non-Hodgkin's lymphoma receiving initial CHOP chemotherapy: A risk factor analysis. Clin Lymphoma 2:47-56, 2001
Mustafa MM, Aquino VM, Pappo A, et al: A pilot study of outpatient management of febrile neutropenic children with cancer at low risk of bacteremia. J Pediatr 128:847-849, 1996
Aquino VM, Herrera L, Sandler ES, et al: Feasibility of oral ciprofloxacin for the outpatient management of febrile neutropenia in selected children with cancer. Cancer 88:1710-1714, 2000
Jones GR, Konsler GK, Dunaway RP, et al: Infection risk factors in febrile, neutropenic children and adolescents. Pediatr Hematol Oncol 13:217-229, 1996
Wehl G, Allerberger F, Heitger A, et al: Trends in infection morbidity in a pediatric oncology ward, 1986-1995. Med Pediatr Oncol 32:336-343, 1999
Lucas KG, Brown AE, Armstrong D, et al: The identification of febrile, neutropenic children with neoplastic disease at low risk for bacteremia and complications of sepsis. Cancer 77:791-798, 1996
Talcott JA, Finberg R, Mayer RJ, et al: The medical course of cancer patients with fever and neutropenia: Clinical identification of a low-risk subgroup at presentation. Arch Intern Med 148:2561-2568, 1988(Swati K. Basu, Isabel D. )
ABSTRACT
PURPOSE: The aim of this study was to evaluate risk factors for longer length of stay (los) and mortality among hospitalized children with cancer who have febrile neutropenia.
METHODS: This study involved analysis of longitudinal data from the University HealthSystem Consortium database from 1995 to 2002. All patients who were 21 years or younger, with diagnostic codes for both neoplastic disease and febrile neutropenia at discharge, were included.
RESULTS: A total of 12,446 patients were identified for the study. The los was 5 days or less for 6,799 patients, and greater than 5 days for 5,647 patients. The mortality rate was 3%. On bivariate analysis, race, age, cancer type, and associated complications (bacteremia/sepsis, hypotension, pneumonia, and fungal infections) were significantly associated with longer length of stay and death. On multivariate analysis, age group, race, cancer type (acute myeloid leukemia, multiple cancers v acute lymphoblastic leukemia), and the complication variables were significantly associated with increased risk of longer los and death. Certain types of cancer (Hodgkin's disease, osteosarcoma/Ewing’s sarcoma, rhabdomyosarcoma, compared with acute lymphoblastic leukemia) and year of discharge after 1995 were significantly associated with a reduced risk of longer length of stay and/or mortality.
CONCLUSION: Race, age group, year of discharge, associated complications, and cancer type were significantly associated with risk of longer los and mortality. These factors may potentially help in identifying high-risk patients who might benefit from targeted antibiotic therapy or prophylactic hematopoietic growth factor support.
INTRODUCTION
Febrile neutropenia is a common complication of therapy among children with cancer. In recent years, several studies have evaluated risk factors for bacteremia or poor outcomes among patients with cancer and helped to establish the current guidelines for the treatment of febrile neutropenia by the Infectious Disease Society of America (IDSA), as well as the recommendations for the use of hematopoietic colony-stimulating factors by the American Society of Clinical Oncology (ASCO).1-4 However, not all guidelines are generalizable to children with cancer because of inadequate data in this population. Studies on the pediatric population are rendered difficult by the limited numbers of pediatric patients with cancer treated at a single institution. It is also problematic to study mortality as an outcome because the mortality associated with febrile neutropenia at any institution is usually small. Therefore, other surrogate markers for severity of disease are used, such as bacteremia or invasive bacterial infection, which are defined differently in various studies and therefore difficult to compare.5-8
The overall goal of this study was to better understand factors associated with outcomes of febrile neutropenia and its consequences in children. This study was a retrospective analysis of clinical data on pediatric patients from a national hospitalization database.
METHODS
Study Design
This study was designed as an analysis of prospectively collected data included in a hospitalization database maintained by the University HealthSystem Consortium (UHC). The major outcomes of interest were length of stay and mortality. The major independent variables of interest were age, sex, race, year of discharge, histopathologic diagnosis, and associated complications.
Source of Data
The UHC database contains information from discharge summaries of all patients with cancer and febrile neutropenia who were admitted between 1995 and 2002 to 115 member academic institutions in the United States. Because children with cancer are usually treated at tertiary care academic centers, the subjects from this database were considered to constitute a representative population of children with febrile neutropenia. All patients included in the analysis had diagnosis codes for both neoplastic disease (International Classification of Diseases, 9th revision, clinical modification [ICD-9-CM] code 140-239) and febrile neutropenia, and were 21 years or younger. Patients with a diagnostic code for bone marrow transplantation were excluded from the study. For those patients who had more than one admission for febrile neutropenia, only one hospitalization was selected randomly for the purpose of this analysis.
Statistical Analysis
The data set included information on age, sex, race, year of discharge, diagnostic codes at discharge (up to 16 diagnostic codes), length of hospitalization, and discharge status. The primary outcomes of this analysis were length of stay and mortality in the hospital, which served as markers for severity of illness. Based on the ICD-9-CM codes for diagnosis, subjects were classified as having acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), Hodgkin's disease (HD), non-Hodgkin's lymphoma (NHL), osteosarcoma and Ewing's sarcoma, rhabdomyosarcoma and other soft tissue sarcomas, neuroblastoma and renal tumors, hepatic tumors, and ophthalmic and ovarian/testicular tumors. Subjects who had a diagnostic code for cancer but who did not fit in any of the above cancer categories were classified as having "other" cancers. Subjects with diagnostic codes for more than one of these cancer categories were classified as having multiple cancers. Using the ICD-9-CM codes as listed in Table 1, additional variables were created to identify subjects with bacteremia/sepsis, other bacterial infections, hypotension, pneumonia, and fungal infections. Subjects who died during the hospitalization were identified using the discharge status.
The initial step in the analysis involved obtaining descriptive statistics on all dependent and independent variables. The bivariate association between each outcome variable (length of stay and mortality) and each of the independent variables was then evaluated. Length of stay was used as a dichotomous variable ( 5 days and > 5 days) for the purpose of the initial bivariate analyses. The cutoff point of 5 days for the purpose of the analysis was determined before the data analysis and was based on the clinical expectation that a patient who was admitted for less than 5 days most likely represented one who had an uncomplicated course in the hospital.
For the multivariate analysis, a three-level categorical outcome variable was defined to include patients who were discharged alive with a length of stay less than or equal to 5 days, patients who were discharged alive with a length of stay longer than 5 days, and patients who died. Variables found to have a significant association with the outcome variable in the bivariate analysis and potential confounders were included in the multivariate analysis. A step-wise approach was used to arrive at the final model, using nominal logistic regression with a generalized logit model. SAS software version 8 (SAS Institute Inc, Cary, NC) was used to run all the statistical analyses.
RESULTS
Predictor Variables
A total of 12,446 patients were identified for the study using the criteria described in the Methods section. The sample consisted of 7,045 (57%) male and 5,401 (43%) female patients. Of the 12,446 patients in the study, 8,349 (67%) subjects in the study were white, 1,262 (10%) were African American, 1,259 (10%) were Hispanic, 342 (3%) were Asian, and 1,234 (10%) were categorized as representing other racial/ethnic groups. In the overall study population, 502 (4%) of the subjects were infants, 7,563 (61%) were children between the ages of 1 and 12 years, and 4,377 (35%) subjects were older than 12 years of age. The median age of subjects in the study was 8 years (range, 0 to 21 years). The number of subjects entered each year from 1995 to 2002 ranged from 1,224 to 1,853 representing 10% to 15% of the study sample (data not shown). The distribution of cancer types is shown in Figure 1. ALL was the most common type of cancer and occurred in 34% of subjects. There were 1,988 (16%) patients with a diagnostic code for bacteremia or sepsis, 251 (2%) subjects with hypotension, 678 (5%) subjects with pneumonia, 1,150 (9%) subjects with other bacterial infections, and 1,056 (8%) with fungal infections. There were 2,598 subjects with one complication, and 1,157 subjects with multiple complications.
Outcome Variables
Length of stay. The median length of stay for all patients in the study was 5 days (range, 1 to 359 days). The study sample consisted of 6,799 (55%) patients with a length of stay of 5 days or less and 5,647 (45%) patients with length of stay longer than 5 days (as presented in Table 2).
The mean and median length of stay for the patients admitted for 5 days or less were the same (3 days), with a range of 1 to 5 days. The mean and median length of stay for the patients admitted for longer than 5 days were 19 days and 12 days, respectively, with a range of 6 to 359 days. As seen in Table 2, the racial/ethnic composition of subjects hospitalized for longer than 5 days was significantly different from those with a shorter length of stay. Specifically, subjects in each of the nonwhite categories were more likely to require a longer hospital stay than white patients. Age group also emerged as a significant risk factor for longer length of hospital stay. Adolescents were found to have a greater likelihood of being hospitalized for longer than 5 days as compared with infants and children. Subjects with a diagnosis of AML were found to be at greater risk of being hospitalized for longer than 5 days as compared with subjects with other cancer diagnoses. Subjects with an associated diagnosis of bacteremia/sepsis, hypotension, pneumonia, and other bacterial or fungal infections were also significantly more likely to have a longer length of stay as compared with subjects without these associated diagnoses (Table 2).
Mortality. Three hundred fifty-eight subjects in the study died, leading to a mortality rate of 3% (see Fig 2). Subjects who were white were less likely to die compared with subjects in each of the nonwhite categories (Table 3). Infants and adolescents were found to be at a greater risk of death than children. A diagnosis of AML or multiple cancers was associated with an increased risk of death as compared with other cancer types, as seen in Table 3. A diagnosis of bacteremia/sepsis significantly increased the risk of death: 9% of patients with bacteremia died, as compared with 2% mortality rate among patients without bacteremia. The incidence of bacteremia among patients who died was 50% as compared with 15% among survivors. Similarly, hypotension, pneumonia, and fungal infections were significantly associated with an increased risk of death (Table 3).
Multivariate Analysis
For the multivariate analysis, a three-level categoric outcome variable was used as described in the methods section. The group of patients who were discharged alive with a length of stay of 5 days was used as the reference group. Odds ratios (ORs) were obtained for each of the two remaining categories of outcome (patients who were discharged alive with length of stay longer than 5 days and patients who died) as compared with the patients who were discharged alive with a length of stay of 5 days or less. Using nominal logistic regression with a generalized logit model, year of discharge, race, age group, the associated complication diagnoses, and cancer type were significantly associated with risk of longer length of stay and/or death (Table 4).
With each consecutive year after 1995, the risk of longer length of stay decreased by 3% (OR = 0.97), and the risk of death decreased by 7% (OR = 0.93), after accounting for all the other variables in the model. The risk of both longer length of stay and death was consistently high in each of the other nonwhite categories as compared with white patients. Infants were found to have a significantly increased risk of death (OR = 1.75) as compared to children between the ages of 1 and 12 years. Adolescents were at a significantly higher risk of both longer length of stay (OR = 1.44) and death (OR = 3.05) as compared with children aged 1 to 12 years. The associated complications of bacteremia/sepsis, hypotension, pneumonia, and fungal infection were all significantly associated with longer length of stay and death. Of note, patients with bacteremia/sepsis had an almost 10-fold increase in the risk of death as compared to those without a diagnosis of bacteremia/sepsis. Cancer type was also found to be an independent predictor of both longer length of stay and death. As compared with ALL, patients with AML and multiple cancers had a significantly greater risk of both longer length of stay (OR = 2.86 and 1.20, respectively) and death (OR = 2.88 and 1.48, respectively). Patients with Hodgkin's disease, osteosarcoma/Ewing sarcoma and rhabdomyosarcoma were at a significantly lower risk of both longer length of stay and death as compared with subjects with ALL. The risk of longer length of stay was also significantly lower for subjects with neuroblastomas/kidney tumors and ovarian/testicular/ophthalmic tumors compared with ALL.
A secondary analysis was performed using mortality as a dichotomous variable (dead v alive) in a multivariate model to determine the risk of death as compared with being alive, for each predictor variable. Year of discharge, race, age group, cancer type, presence of sepsis, hypotension, pneumonia, other bacterial and fungal infections were again all found to be significantly associated with the risk of death (data not shown), as seen with the 3-category outcome. Each successive year after 1995 was associated with a significantly lower risk of death. All the nonwhite race/ethnicity groups were found to be at an increased risk of death. With regard to age group, adolescents were also found to have a significantly increased risk of death. However, infancy was not significantly associated with an increased risk of death (OR = 1.66, P = .0653). Among the cancer types, Hodgkin's disease, osteosarcoma and Ewing sarcoma, and rhabdomyosarcoma were associated with a significantly reduced risk of death. A diagnosis of AML was found to have an increased risk of death (OR = 1.34), but this was not statistically significant (P = .0667).
DISCUSSION
In recent years, there has been increasing interest in stratifying patients with febrile neutropenia into risk categories to explore various therapeutic strategies.2,3,9 The current study uses a multicenter database to identify risk factors for longer length of stay and death.
Death is the most devastating of all potential outcomes of febrile neutropenia. Yet, most studies of febrile neutropenia use bacteremia or bacterial infection as the main outcome variable because infection is the expected complication in a neutropenic patient.5-8 However, the availability of potent broad-spectrum antibiotics has made most bacterial infections curable today. Despite the widespread approach of treating all febrile neutropenic patients with empiric broad-spectrum antibiotics for potential bacteremia, febrile neutropenia in the child with cancer remains associated with a significant, albeit small, mortality rate. This suggests that certain host or disease-related factors may play a role in influencing mortality resulting from febrile neutropenia. Thus, mortality seems to be the more appropriate outcome for evaluation in the modern era. In the current study, the large sample in the database presented a remarkable opportunity to evaluate mortality associated with febrile neutropenia. It must be noted, however, that although all the patients in the study had febrile neutropenia during their course in the hospital, the cause of death may have been refractory cancer or other diagnoses with or without infection.
In this study, length of stay was also evaluated as an outcome of interest for several reasons. Cancer care currently accounts for approximately 10% of all health care expenditure in the United States.10,11 A substantial proportion of the health care costs incurred by patients with cancer can be attributed to the treatment of febrile neutropenia. A recent study from the Hospital for Sick Children, Toronto, Canada reported fever with neutropenia as one of the most common admitting diagnoses among pediatric oncology patients, second only to admissions for chemotherapy.8 It is commonly known that costs of hospitalization increase with longer length of stay. In this study, length of stay also served as a surrogate for severity of illness. In addition, evaluation of length of stay is important because longer periods of hospitalization impact the quality of life of affected patients.
The mortality rate in this study was approximately 3%. It must be noted that this figure is derived from patients who were admitted to the hospital with febrile neutropenia, as well as patients who may have been admitted for another medical condition and subsequently developed febrile neutropenia while in the hospital. The mortality rate in some recent pediatric studies of febrile neutropenia has ranged from 0.7% to 3.9%.5,7,8,12
In the current study, race, age group, year of diagnosis, cancer type, and associated complications were significantly associated with longer length of stay (greater than 5 days) and death.
As compared with white patients, all nonwhite groups were found to have a higher risk of both longer length of stay and death. To the best of our knowledge, this association has not been reported previously among patients with febrile neutropenia. There are studies that have evaluated the relationship between racial/ethnic groups and survival among children diagnosed with ALL, with conflicting results.13 Analyses of data from the two large children's cooperative groups have found differences in survival between patients of different races, with inferior survival among African American and Hispanic children.14,15 However, successive studies from a single large institution have not found any differences in survival across racial groups.16,17 With regard to febrile neutropenia, race may be a surrogate for social class or a confounder for an underlying association between socioeconomic status and poor outcomes. Additional information regarding socioeconomic status was not available in the current study to evaluate this relationship further. Alternatively, it is possible that there is a true biologic susceptibility conferred by race and, given the large sample size, the current study had adequate power to uncover the association between race and poor outcomes with febrile neutropenia. Therefore, we suggest that race should be taken into account while considering alternative treatment strategies in children with febrile neutropenia.
Age group emerged as another significant independent risk factor for both longer length of stay and death. In studies of adults with febrile neutropenia, older age has emerged as a significant risk factor for poor outcomes.18,19 Although infancy has been cited as a high-risk pediatric age group, most studies have very few infants and young children to evaluate this relationship.12,20-22 In the current study, the risk of death was significantly higher in infants as compared with children aged 1 to 12 years. Adolescence was also found to be a high-risk age group, with an increased risk of both longer length of stay and death. After controlling for race, year of discharge, cancer type, and complications, adolescents were almost 1.5 times as likely to be hospitalized longer and more than three times as likely to die as children aged 1 to 12 years. As in the case of race, it is not clear whether this reflects an underlying biologic susceptibility during this period. It is possible that adolescents are more reluctant to seek care for what they may perceive as an innocuous episode of fever, present late in the course of a febrile neutropenic episode and therefore develop more severe complications. It is also possible that there are differences in treatment received by adolescents, with some patients receiving care from pediatric oncologists and others being cared for by adult oncology teams.
Year of discharge was found to have an independent association with both longer length of hospital stay and death. Using the initial year (1995) as the comparison group, the risk of longer length of stay decreased by approximately 3% with every passing year. This is not surprising, given the trend toward early discharge in almost every sphere of medicine in recent years. However, it is noteworthy that there has been a significant decrease in the risk of death over the same period, by approximately 7% per year. Perhaps, this may be attributable to better supportive care. The role of granulocyte colony-stimulating factor in this decline in mortality would be interesting to evaluate. Unfortunately, this information was not available for this study. Of note, a previous study of trends in infection morbidity in a pediatric oncology ward throughout a 10-year period did not find a decrease in the frequency of febrile episodes or fatality of infections after recombinant hematopoietic growth factors became available.23
For the evaluation of risk by cancer type, ALL was used as the reference category because it was the most frequently encountered cancer type in the data set and because most clinicians are familiar with the treatment and treatment-related complications of ALL. It should be noted that an odds ratio of less than 1 for any of the cancer types in this analysis indicates a lower risk of the outcome as compared to ALL; it does not denote a protective effect as there were no noncancer controls. In comparison to ALL, a diagnosis of AML was associated with a three-fold increased risk of longer length of stay as well as death. A diagnosis of Hodgkin's disease, osteosarcoma/Ewing sarcoma or rhabdomyosarcoma was associated with a significant reduction in risk of both longer length of stay and death. There was a fairly large "other cancer type" category, which comprised 12% of the study population and consisted of numerous diagnoses that were too small individually to be considered as distinct categories in the analysis. This "other" group, which included mostly miscellaneous solid tumors, did not have a significant difference in outcomes as compared to ALL. This is consistent with studies in adult populations that have identified the diagnosis of solid tumors as a low risk factor.2 Similarly, pediatric studies have identified leukemia or relapse of leukemia as high risk factors for bacterial infection.5,24
Finally, as expected, the complication diagnoses were significantly associated with both longer length of stay and death. A diagnosis of sepsis or bacteremia conferred a 10-fold increase in the risk of death. Thus, despite the widespread use and availability of powerful antibiotics, bacteremia/sepsis remains the most important independent prognostic marker for mortality. Pneumonia was associated with an eight-fold increase in the risk of death. Fungal infections were associated with a five-fold increase in the risk of death. Several complication variables were associated with a longer length of stay and death. These findings are consistent with previous reports that comorbid conditions are high risk factors in adult populations.2,18,25 Hypotension was reported as an independent high risk factor for invasive bacterial infection in a prospective pediatric study,5 while several other prospective pediatric studies of low-risk patients exclude subjects with comorbidities.8,20
This study involved a retrospective analysis of an administrative data set, and therefore has several limitations. The selection of the variables, coding used, and the methods used for recording the data were predetermined. Therefore, it was not possible to evaluate additional variables that may have been of interest, such as the absolute neutrophil or monocyte count at the time of admission and discharge, use of growth factors, and type of chemotherapy used. The analysis was also limited by the ICD-9-CM diagnostic codes listed at discharge. There is no information in the data set about whether patients were at diagnosis, in remission, or in a palliative care mode at the time of the hospitalization. Reasons for admission were also not available. The subjects in the database include patients who were admitted with febrile neutropenia, as well as patients who were hospitalized for another reason and developed febrile neutropenia during the course of the hospitalization. Further studies are necessary to specifically evaluate patients admitted with febrile neutropenia. The cause of death is not available. Thus, patients in this database may have died from causes other than febrile neutropenia. This could theoretically bias the study. Approximately 12% of the study population were found to have ICD-9-CM codes for more than one cancer category and were classified as having multiple cancers. Some of these patients may truly have more than one cancer, some may have treatment-related second malignant neoplasms. In several cases, this may represent a coding error or misinterpretation of the diagnosis. It was not possible to evaluate this issue further in the data set. These patients were analyzed in a separate category so as to avoid a potential misclassification bias that might result from their inclusion in only one of the major histopathologic categories.
On the other hand, use of a pre-existing data set offered the advantages of speed, economy and large sample size. Current protocols for the treatment of febrile neutropenia are based on findings of randomized trials. However, subjects enrolled on clinical trials are usually different from the general population. This data set includes all patients admitted to the hospital unselected by any criteria other than febrile neutropenia and cancer. Therefore, the findings may be more generalizable to the entire population.
In conclusion, the results of this study indicate that nonwhite race, infancy and adolescence, bacteremia/sepsis, other bacterial infections, hypotension, pneumonia, fungal infections, and a diagnosis of AML or multiple cancers should be considered high risk factors in children with febrile neutropenia. The identification of risk factors for poor outcomes may help in devising protocols for providing targeted therapy based on risk criteria and thus help to reduce the cost of cancer care as well as reduce the adverse effects of antibiotic therapy. The findings of this study may also help in devising updated guidelines for hematopoeitic growth factor support in neutropenic children with cancer. Prospective studies of febrile neutropenia among children enrolled in large cooperative trials may be beneficial in evaluating these risk factors further.
Authors' Disclosures of Potential Conflicts of Interest
Although all authors completed the disclosure declaration, the following author or immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
NOTES
Presented at the 40th Annual Meeting of the American Society of Clinical Oncology, June 5-8, 2004.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Pizzo PA: Management of fever in patients with cancer and treatment-induced neutropenia. N Engl J Med 328:1323-1332, 1993
Hughes WT, Armstrong D, Bodey GP, et al: 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer. Clin Infect Dis 34:730-751, 2002
Ozer H, Armitage JO, Bennett CL, et al: 2000 update of recommendations for the use of hematopoietic colony-stimulating factors: Evidence-based, clinical practice guidelines—American Society of Clinical Oncology Growth Factors Expert Panel. J Clin Oncol 18:3558-3585, 2000
Pizzo PA: Fever in immunocompromised patients. N Engl J Med 341:893-900, 1999
Santolaya ME, Alvarez AM, Becker A, et al: Prospective, multicenter evaluation of risk factors associated with invasive bacterial infection in children with cancer, neutropenia, and fever. J Clin Oncol 19:3415-3421, 2001
Santolaya ME, Alvarez AM, Aviles CL, et al: Prospective evaluation of a model of prediction of invasive bacterial infection risk among children with cancer, fever, and neutropenia. Clin Infect Dis 35:678-683, 2002
Rackoff WR, Gonin R, Robinson C, et al: Predicting the risk of bacteremia in childen with fever and neutropenia. J Clin Oncol 14:919-924, 1996
Klaassen RJ, Goodman TR, Pham B, et al: "Low-risk" prediction rule for pediatric oncology patients presenting with fever and neutropenia. J Clin Oncol 18:1012-1019, 2000
Orudjev E, Lange BJ: Evolving concepts of management of febrile neutropenia in children with cancer. Med Pediatr Oncol 39:77-85, 2002
Parker SL, Tong T, Bolden S, et al: Cancer statistics, 1996. CA Cancer J Clin 46:5-27, 1996
Vincenzino JV: Health care costs: Market forces and reform. Oncology (Huntingt) 9:367-368, 371, 374, 1995
Baorto EP, Aquino VM, Mullen CA, et al: Clinical parameters associated with low bacteremia risk in 1100 pediatric oncology patients with fever and neutropenia. Cancer 92:909-913, 2001
Bhatia S: Influence of race and socioeconomic status on outcome of children treated for childhood acute lymphoblastic leukemia. Curr Opin Pediatr 16:9-14, 2004
Bhatia S, Sather HN, Heerema NA, et al: Racial and ethnic differences in survival of children with acute lymphoblastic leukemia. Blood 100:1957-1964, 2002
Pollock BH, DeBaun MR, Camitta BM, et al: Racial differences in the survival of childhood B-precursor acute lymphoblastic leukemia: A Pediatric Oncology Group Study. J Clin Oncol 18:813-823, 2000
Pui CH, Boyett JM, Hancock ML, et al: Outcome of treatment for childhood cancer in black as compared with white children: The St Jude Children's Research Hospital experience, 1962 through 1992. JAMA 273:633-637, 1995
Pui CH, Sandlund JT, Pei D, et al: Results of therapy for acute lymphoblastic leukemia in black and white children. JAMA 290:2001-2007, 2003
Klastersky J, Paesmans M, Rubenstein EB, et al: The Multinational Association for Supportive Care in Cancer risk index: A multinational scoring system for identifying low-risk febrile neutropenic cancer patients. J Clin Oncol 18:3038-3051, 2000
Morrison VA, Picozzi V, Scott S, et al: The impact of age on delivered dose intensity and hospitalizations for febrile neutropenia in patients with intermediate-grade non-Hodgkin's lymphoma receiving initial CHOP chemotherapy: A risk factor analysis. Clin Lymphoma 2:47-56, 2001
Mustafa MM, Aquino VM, Pappo A, et al: A pilot study of outpatient management of febrile neutropenic children with cancer at low risk of bacteremia. J Pediatr 128:847-849, 1996
Aquino VM, Herrera L, Sandler ES, et al: Feasibility of oral ciprofloxacin for the outpatient management of febrile neutropenia in selected children with cancer. Cancer 88:1710-1714, 2000
Jones GR, Konsler GK, Dunaway RP, et al: Infection risk factors in febrile, neutropenic children and adolescents. Pediatr Hematol Oncol 13:217-229, 1996
Wehl G, Allerberger F, Heitger A, et al: Trends in infection morbidity in a pediatric oncology ward, 1986-1995. Med Pediatr Oncol 32:336-343, 1999
Lucas KG, Brown AE, Armstrong D, et al: The identification of febrile, neutropenic children with neoplastic disease at low risk for bacteremia and complications of sepsis. Cancer 77:791-798, 1996
Talcott JA, Finberg R, Mayer RJ, et al: The medical course of cancer patients with fever and neutropenia: Clinical identification of a low-risk subgroup at presentation. Arch Intern Med 148:2561-2568, 1988(Swati K. Basu, Isabel D. )