VOLUME

23

NUMBER

30

OCTOBER

20

2005

JOURNAL OF CLINICAL ONCOLOGY

O R I G I N A L

R E P O R T

Feasibility of Withholding Antibiotics in Selected Febrile Neutropenic Cancer Patients

Claudi Oude Nijhuis, Willem A. Kamps, Simon M.G. Daenen, Jourik A. Gietema, Winette T.A. van der Graaf, Harrie J.M. Groen, Edo Vellenga, Els M. ten Vergert, Karin M. Vermeulen, Hillie G. de Vries-Hospers, and Eveline S.J.M. de Bont
From the Division of Pediatric Oncology/Hematology, Beatrix Childrens Hospital, Division of Hematology, Division of Medical Oncology, Division of Pulmonary Diseases, Department of Internal Medicine, Ofce for Medical Technology Assessment, Department of Medical Microbiology, University Hospital Groningen, the Netherlands. Submitted November 12, 2004; accepted July 12, 2005. Supported by a Grant from the Association of University Hospitals in the Netherlands. Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 31-June 3, 2003. Authors disclosures of potential conicts of interest are found at the end of this article. Address reprint requests to E.S.J.M. de Bont, MD, PhD, Division of Pediatric Oncology, Beatrix Childrens Hospital, University Hospital Groningen, P.O. Box 30.001, 9700 RB Groningen, the Netherlands; e-mail: e.s.j.m.de.bont@bkk.umcg.nl. 2005 by American Society of Clinical Oncology 0732-183X/05/2330-7437/$20.00 DOI: 10.1200/JCO.2004.00.5264

Purpose To investigate the feasibility of withholding antibiotics and early discharge for patients with chemotherapy-induced neutropenia and fever at low risk of bacterial infection by a new risk assessment model. Patients and Methods Outpatients with febrile neutropenia were allocated to one of three groups by a risk assessment model combining objective clinical parameters and plasma interleukin 8 level. Patients with signs of a bacterial infection and/or abnormal vital signs indicating sepsis were considered high risk. Based on their interleukin-8 level, remaining patients were allocated to low or medium risk for bacterial infection. Medium-risk and high-risk patients received standard antibiotic therapy, whereas low-risk patients did not receive antibiotics and were discharged from hospital after 12 hours of a febrile observation. End points were the feasibility of the treatment protocol. Results Of 196 assessable episodes, 76 (39%) were classied as high risk, 84 (43%) as medium risk, and 36 (18%) as low risk. There were no treatment failures in the low-risk group (95% CI, 0% to 10%). Therefore, sensitivity of our risk assessment model was 100% (95% CI, 90% to 100%), the specicity, positive, and negative predictive values were 21%, 13%, and 100%, respectively. Median duration of hospitalization was 3 days in the low-risk group versus 7 days in the medium- and high-risk groups (P .0001). The incremental costs of the experimental treatment protocol amounted to a saving of 471 (US $572) for every potentially low-risk patient. Conclusion This risk assessment model appears to identify febrile neutropenic patients at low risk for bacterial infection. Antibiotics can be withheld in well-dened neutropenic patients with fever. J Clin Oncol 23:7437-7444. 2005 by American Society of Clinical Oncology

INTRODUCTION

Bacterial infections are common and potentially serious complications of cancer treatment. The most important risk factor for infection is an absolute neutrophil count (ANC) less than 0.5 109/L and its dura1-3 tion. In addition, disruptions of the physical defense barriers of skin and mucosae, secondary to chemotherapy, leads to increased expo-

sure to potentially pathogenic microorganisms. In situ venous access catheters are further risk factors for bacteremia. Fever during chemotherapy-induced neutropenia can be the rst sign of bacterial infection and, therefore, requires prompt and careful attention. In the early 1970s, the introduction of empirical use of intravenous broad-spectrum antibiotics for febrile neutropenic patients reduced infection-related
7437

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Oude Nijhuis et al

mortality signicantly.4 Until recently, routine hospitalization for administration of intravenous broad-spectrum antibiotics is still standard care and the paradigm in most hospitals. Patients are discharged when they are afebrile for at least 24 hours and the ANC has recovered.5 Although this standard treatment resulted in a dramatic decline in morbidity and mortality from infection, there are reports indicating that febrile neutropenic patients are a heterogeneous patient group with respect to their risk for bacterial infection and medical complications.6-8 Nearly half of febrile neutropenic patients who receive standard antibiotic treatment have no microbiologically or clinically documented bacterial infection.9 Therefore, new treatment strategies for febrile neutropenic patients with a low risk for bacterial infection are under investigation. Emerging resistance and side effects to antibiotics, compliance of the patient, and costs of hospitalization make this important. Furthermore, hospital admissions expose patients to nosocomial pathogens and reduce the quality of life of patients and their families. There are a number of studies that show that adult patients with a low risk of infection can be treated safely with oral rather than intravenous antibiotics.10-12 From this perspective as well as the observation that some outpatients with febrile neutropenia can have a favorable outcome without antibiotics justied a new approach: to withhold antibiotics in patients with a low risk of bacterial infection. This is only feasible if a strict, but simple, model with high sensitivity for detecting the absence of a bacterial infection is identied and applied. Previous studies have shown that plasma interleukin-6 (IL-6) and -8 (IL-8) levels were significantly increased in microbiologically documented infections compared to levels in unexplained fevers in neutropenic cancer patients.13-15 Moreover, it was demonstrated that IL-8 as well as IL-6 plasma levels measured at the onset of febrile neutropenia can dene a patient group at low risk for bacteremia.13 Because age had less inuence on IL-8 plasma levels than on IL-6 plasma levels, IL-8 plasma levels were considered more appropriate to dene a subgroup of patients with a low risk of infection. In this prospective study, we evaluated the feasibility of withholding antibiotics and early hospital discharge in febrile neutropenic patients with a low risk of bacterial infection. A new risk assessment model consisting of objective clinical parameters and the plasma IL-8 level was used to allocate risk groups. In addition, cost analysis was performed.
PATIENTS AND METHODS Patients Outpatient cancer patients presenting with fever and chemotherapy-induced neutropenia were eligible. Fever was dened as a single body temperature higher than 38.5C, or two or
7438

more recordings of a body temperature higher than 38.0C during a 6-hour period. Neutropenia was dened as ANC less than 0.5 109/L or leukocytes less than 1.0 109/L. Patients who had received antibiotics, or who had undergone a stem-cell transplant in the previous month were excluded. The use of selective gut decontamination or Pneumocystis jiroveci pneumonia prophylaxis was not an exclusion criterion. Study Design This single-center prospective intervention study was performed at the Departments of Pediatric Oncology, Internal Hematology, Medical Oncology, and Pulmonary Diseases of the University Hospital Groningen from April 1999 until August 2002 in children and until October 2002 in adult patients. The institutional review board approved the study protocol and all patients and/or their parents, dependent on age, gave written informed consent for participation. All patients underwent a physical examination at presentation. In addition to routine blood and radiologic evaluations, diagnostic blood cultures (two from a peripheral blood vessel, 30 minutes apart, and one from the venous access port if present) were performed and a plasma sample was taken to measure the plasma IL-8 concentration at entry and at 24 hours after study entry. All patients were followed until day 8 from admission or until discharge from hospital, whichever was the longest. Risk Assessment A risk assessment model constructed on physical examination and IL-8 level was applied. Patients with signs of a local bacterial infection and/or abnormal vital signs suggesting sepsis were classied as high risk. Abnormal vital signs indicating sepsis were dened as: systolic blood pressure less than 90 mmHg in adults or less than 2 SD for age in children, or both heart rate higher than 100/min and respiratory rate higher than 20/min in adults or both heart and respiratory rate higher than 2 SD for age in children, according to systemic inammatory response syndrome (SIRS) criteria.16,17 Patients with plasma IL-8 level below the cutoff value were classied as low risk. Patients with an IL-8 above the cutoff value were classied as medium risk. Patients with a second IL-8 within 24 hours below the cutoff value and unchanged physical ndings remained in the low-risk group. Patients with a second IL-8 level above the cutoff moved to the medium-risk group, and those who developed abnormal vital signs and/or clinical signs suggesting bacterial infection, to the high-risk group (Fig 1). Withholding broad-spectrum antibiotics from low-risk patients is a completely new concept for which safety is not known. So at the start of the study it was decided that a safety analysis was performed after the rst 75 included episodes of outpatient cancer patients with safe limits for IL-8 (40 ng/L) as well as for vital signs. Outcome was considered unsafe if one of the following situations occurred1: increase in temperature ( 38.5C) after a 24-hour afebrile period until day 5 after inclusion2; hemodynamic instability meeting the criteria for abnormal vital signs attributable to infection in patients classied as low risk; and3 isolation of a bacterial pathogen from the blood cultures in patients classied as low risk. An unfavorable outcome determined the start of antimicrobial treatment and readmission to the hospital and was classied as a failure. Death of patient during the febrile neutropenic episode attributable to infection who was classied as a low-risk episode on physical examination and IL-8 plasma level was an
JOURNAL OF CLINICAL ONCOLOGY

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Withholding Antibiotics in Neutropenia

Fig 1. Risk Assessment Model After physical examination (P.E.), patients with signs of a local bacterial infection and/or abnormal vital signs were classied into the high-risk group. The cutoff value for the interleukin (IL)-8 test was 60 ng/L. Patients with an IL-8 greater than 60 ng/L were classied as medium risk, patients with an IL-8 60 ng/L were classied as low risk.

absolute reason to stop the inclusion in the low-risk group. No failures or death were achieved at 75 febrile episodes. After 75 episodes without failures, we were reassured that the IL-8 cutoff value could be raised from 40 ng/L up to 60 ng/L and pediatric criteria of heart frequency and respiratory rate could be adjusted to wider acceptable ranges.18 Plasma IL-8 concentrations were determined using a chemiluminescence immunoassay according to the manufacturers instructions (Immulite, Diagnostic Products Corporation, Los Angeles, CA) as a routine procedure operated 24 hours a day. The result of this IL-8 determination was available within 30 minutes. The cutoff value was based on a receiver operating curve analysis in a prior study set on 40 ng/L.13 After 75 episodes, it was raised to 60 ng/L as described above. Treatment All patients were admitted to the hospital. Low-risk patients did not receive intravenous antibiotics, except for those with severe mucositis who received oral amoxicillin-clavulanic acid. Low-risk patients were discharged from hospital when they were afebrile for 12 hours irrespective of their ANC. Following discharge, low-risk patients were contacted daily by the research physician until day 8 of the study protocol. High-risk and medium-risk patients received broadspectrum intravenous antibiotics. Adults received cefuroxim (1,500 mg, three times daily) and tobramycin (3 mg/kg, once daily), and children ceftazidime (50 mg/kg, three times daily to a maximum of 6 g/d). Antibiotic treatment was stopped and patients were discharged when the blood culture was negative, patients were afebrile for at least 24 hours, and the ANC was greater than 0.5 109/L. No antipyretics were given during the 8 day study period. Tramadol was given when analgesia was necessary. End Points The primary end point of this study was feasibility of the treatment protocol. This was evaluated by the number of failures in the low-risk group, which were dened as either positive blood cultures at the time of admission, persistent fever, or recurrent fever in combination with prolonged neutropenia. Persistent fever was dened as a body temperature higher than 38.5C for a minimum of 12 hours during the period of 48 to 72 hours after admission. Recurrent fever was dened as a new fever during the
www.jco.org

rst 5 days of the study period, after having been afebrile for a minimum of 24 hours. The secondary end point of the study was the diagnostic value of our risk assessment model to identify a patient group at low risk of bacterial infection. This was evaluated by assessing the sensitivity, specicity, and predictive values of the risk assessment model for the presence of bacteremia. Other secondary end points were duration of fever, neutropenia, intravenous antibiotic therapy, hospitalization (related to the febrile episode), and costs in the three risk groups. Cost Analysis To assess cost effectiveness of the treatment protocol for febrile neutropenia, a cost-minimization analysis was performed. The economic evaluation was performed from a societal perspective. Data on costs were registered on the case report form and in cost diaries. To value costs, Dutch standard prices were used.19 For IL-8 measurements, no standard prices were available and costs were estimated based on true resources used and time invested. Costs for the experimentally treated low-risk patients were assessed prospectively. Specication of cost categories is given in Table 1. Costs for the standard treatment (hospitalization and

Table 1. Mean Total Costs per Patient Low-Risk Group Experimental Protocol () 39 15 1,051 14 20 35 4 5 35 1,218 Standard Treatment () 39 2,618 317 2,974

Category of Costs IL-8 measurements Tests/cultures/x-ray Hospitalization Medication GP consults Informal care at home Formal care at home Out of pocket costs Travel costs Total

Abbreviation: IL, interleukin; GP, general practitioner.

7439

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Oude Nijhuis et al

intravenous antibiotics) were estimated on the basis of mediumrisk patients without bacteremia. Incremental costs were calculated according to the following formula:

C incr(E-S)

(p 2(E)

2(p 1(E)

p 3(E))) Cn_IL8 p 2(E) (mcp 2(E) mcp 1(E))

mcp1 mean costs (excluding IL-8 determination) of medium-risk patients without bacteremia (Table 1 for cost categories) mcp2 mean costs (excluding IL-8 determination) of lowrisk patients (Table 1 for cost categories) Statistical Analysis Statistical analyses were performed using Statistical Package for Social Sciences (SPSS) for Windows, version 10.0 (SPSS Inc. Chicago, IL). The protocol was considered feasible if the failure rate in the lowrisk group was less than 10%. It was estimated that 35 low-risk episodes (using an alpha level of 5% and a power of 80%) and a total of 175 episodes were required (assuming 20% of all episodes to be low risk). Patient characteristics and end points of the risk groups were compared using 2 analysis (sex and malignancy) or Kruskal-Wallis tests (age, leukocytes, ANC, and end points in Table 2). When signicant differences between the risk groups

Cincr incremental costs per patient E experimental protocol S standard treatment p1 proportion of patients who were classied as mediumrisk after IL-8 determination p2 proportion of patients who were discharged from the hospital earlier (low-risk group) p3 proportion of patients who were classied as low-risk on the basis of the rst IL-8 level, but changed to the medium-risk group because the second IL-8 was above the cutoff value Cn_IL8 costs of determination of plasma IL-8 level

Table 2. Characteristics of the Patients With Episodes of Febrile Neutropenia Overall Characteristic Episodes Patients Sex Female Male Age, years Median Range % of patients 16 Leukocytes at admission, 109/L Median Range ANC at admission, 109/L Median Range Type of cancer Hematologic malignancy ALL AML Myeloma Lymphoma Solid tumor Brain tumor Hepatoblastoma Rhabdomyosarcoma Osteosarcoma Ewings sarcoma Other soft tissue sarcoma Neuroblastoma gastrointestinal carcinoma Germ cell malignancy Lung cancer Breast cancer Gynecological cancer Miscellaneous malignancies No. 196 128 57 71 25 0-77 42 0.5 0.2-1.1 0.07 0.02-0.14 53 41 19 7 8 19 75 59 3 2 6 7 6 10 2 3 9 8 11 5 3 1.1 0.2-5.2 0.08 0.00-0.46 8 33 2 1 1 4 16 67 1 1 2 1 1 5 3 2 % 45 55 Low Risk No. 36 24 10 14 19 0-65 44 0.4 0.0-4.5 0.05 0.00-0.41 21 44 9 3 3 6 27 56 1 1 3 3 2 3 2 2 2 2 2 2 2 % 42 58 Medium Risk No. 84 48 24 24 25 0-72 38 0.4 0.0-2.7 0.10 0.00-0.48 24 43 8 3 4 9 32 57 1 1 3 3 7 1 2 6 6 1 1 % 50 50 High Risk No. 76 56 23 33 32 0-77 % 41 59 P .627 .240 .0001 .259 .67
2

NOTE. Sex, age, and type of cancer analyzed in patients (n 128); leukocytes and ANC in episodes (n 196). Sex and type of cancer analyzed by leukocytes, and ANC by Kruskal-Wallis and Mann-Whitney. Abbreviations: ANC, absolute neutrophil count; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia. Signicant difference between low-risk and high-risk as well as medium-risk groups.

; age,

7440

JOURNAL OF CLINICAL ONCOLOGY

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Withholding Antibiotics in Neutropenia

were found by the Kruskal-Wallis test, Mann-Whitney analysis using Bonferroni correction was used for assessing differences in characteristics and end points between low-risk and highrisk, and low-risk and medium-risk patients. It was possible to enroll patients several times. Data obtained from patients who were enrolled more than once were tested for statistical dependence. First, risk group classication of subsequent compared to initial episodes in the same patients was tested .267). Further, by 2 analysis; no dependence was found (P patient characteristics and end points of the study were compared between initial and subsequent episodes using Mann-Whitney analysis. No signicant differences between these two groups of episodes were found. Data obtained in subsequent episodes seemed, therefore, not dependent on data from initial episodes in patients. In addition, the main end points of the study were also analyzed in initial episodes only and mentioned in the results accordingly. A two-sided P value of less than .05 was considered statistically signicant. RESULTS

On day 2, the plasma IL-8 level rose above the cutoff value in six of the 42 low-risk patients (14%), with no change in physical signs suggesting bacterial infection. These six patients changed from the low-risk to the medium-risk group, and received intravenous antibiotic therapy. Of all febrile neutropenic episodes, 84 (43%) were assigned to the medium-risk group, and 36 (18%) to the low-risk group. All of the initially potentially low-risk episodes (n 42) had a favorable clinical course, an absence of major medical complications despite the absence of any antibiotic treatment. Feasibility of the Treatment Protocol Of the 36 febrile neutropenic episodes in patients assessed as low risk and therefore not treated with intravenous antibiotics, only one patient received oral amoxicillinclavulanate acid because of severe mucositis. There were no failures (0%; 95% CI, 0% to 10%; initial episodes only, n 24; 95% CI, 0% to 15%, ie, none of the low-risk patients had a positive blood culture on admission, nor persistent or recurrent fever during the study period, nor discontinuation of the assigned strategy for any reason before neutropenia resolved). Diagnostic Value of the Risk Assessment Model Bacteremia was detected in none of the patients allocated to the low-risk group by the risk assessment model. Therefore, the sensitivity of the risk assessment model was 100% (95% CI, 90% to 100%). Specicity, positive predictive value, and negative predictive value were 21%, 13%, and 100%, respectively (Table 3). End Points per Risk Group Table 4 shows the outcome of the febrile neutropenic episodes according to risk group. A total of 324 blood cultures in 196 episodes were performed. Bacteremia was found in 21 (11%) of the 196 febrile neutropenic episodes, none in the low-risk, nine (11%) in the medium-risk, and 12 (16%) in the high-risk groups. In nine patients of the medium-risk group, the following microbiological isolates in the blood stream were found: Escherichia coli (n 1), Klebsiella pneumoniae (n 1), Pseudomonas aeruginosa (n 1), Acinetobacter (n 1), Coagulase-negative Staphylococcus (n 1), Streptococcus mitis (n 3), and multiorganism (n 1). The following microbiological isolates were found in the 12 high-risk patients: Escherichia coli (n 3), Klebsiella

Characteristics of the Patients A total of 137 patients with 207 febrile neutropenic episodes were enrolled onto the study. Forty of the 137 patients were enrolled more than once: 25 patients had two episodes, and 15 patients had three or more episodes. Eleven (5%) of the 207 febrile neutropenic episodes could not be evaluated due to protocol violations, such as erroneous administration of antibiotics to potentially low-risk patients (n 6); withholding antibiotic therapy on admission or on day 2 to medium- or high-risk patients (n 4); and omitting blood cultures on admission (n 1). The characteristics of the 128 patients with in total 196 episodes are shown in Table 2. No signicant differences between the risk groups were found, except for leukocytes, which were higher at admission in the low-risk group than in the other two risk groups, but this was not reected in the neutrophil count. Risk Assessment Of the total 196 febrile neutropenic episodes, 76 episodes (39%) were assessed as high risk for bacterial infection. In 45 of the 76 episodes, a local bacterial infection was suspected: skin infection in 19 cases; pneumonia in six cases; urinary tract infection in ve cases; sinusitis and otitis both in four cases; dental infection in three cases; and other infection in four cases. In seven of these cases patients also had abnormal vital signs indicating sepsis. Thirty-one of the 76 episodes were classied as high-risk group because of abnormal vital signs indicating sepsis. Of the remaining 120 potentially low-risk episodes, 78 (65%) had plasma IL-8 levels above the cutoff value and were classied as medium risk. The other 42 episodes (35%) were included in the low-risk group. All patients were admitted to the hospital and treated according to the protocol for their risk group.
www.jco.org

Table 3. Diagnostic Value of the Risk Assessment Model NonLow-Risk Bacteremia No bacteremia Total 21 139 160 Low-Risk 0 36 36 Total 21 175 196

7441

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Oude Nijhuis et al

Table 4. Study End Points According to Risk Group Low Risk (n End Point Bacteremia No. of patients % IL-8 level at admission, ng/L Duration of fever, days Duration of neutropenia, days Duration of intravenous antibiotic treatment, days Hospitalization related to the febrile episode, days Median 36) Range Medium Risk (n Median 84) High Risk (n Median 76) Range P

Range

0 0 23 1 3 0 3 7-45 1-4 1-16 1-9 120 2 6 6 7

9 10.7 29-4,927 1-12 1-50 2-17 3-22 92 3 4 6 7

12 15.8 13-9,384 1-21 1-42 0-31 2-34

.008 .012 .0001

Signicant difference between low-risk and high-risk as well as medium-risk groups (Mann-Whitney). Signicant difference between low-risk and medium-risk groups (Mann-Whitney).

pneumoniae (n 1), Enterobacter (n 1), coagulasenegative Staphylococcus (n 1), Streptococcus mitis (n 4), and multiorganism (n 2). Median plasma IL-8 levels at admission were signicantly lower in the low-risk group (23; range, 7 to 45 ng/L) compared with the medium-risk group (120; range, 29 to 4,927 ng/L) and the high-risk group (92; range, 13 to 9,384 ng/L). No difference for IL-8 admission levels was found between medium-risk and high-risk group. No intravenous antibiotics were given to patients in the low-risk group. Intravenous antibiotic therapy was given for a median duration of 6 days in the medium-risk group5-8 and 6 days in the high-risk group.4-9 The median duration of hospitalization in the low-risk group was 3 days2-4 compared with 7 days5-8 in the medium-risk group and 7 days6-10 in the highrisk group (P .0001 and P .0001, respectively). Median duration of hospitalization for initial febrile neutropenic episodes only (n 128) was 3 days2,3 in the low-risk group versus 7 days 6-9 in the medium-risk group, and 8 days6-9 in the high-risk group (P .0001 and P .0001, respectively). Cost Analysis Mean total costs per patient were 1,218 (US $1,480) in the low-risk group compared with 2,974 (US $3,614) in the medium and high-risk groups who received standard treatment. This was mainly explained by shorter hospitalization and lower costs for medication in the low-risk group (Table 1). Since the low-risk group cannot be ascertained in advance, the incremental costs were determined using all patients, except the high-risk patients, in the formula. The incremental costs of the experimental treatment protocol amounted to a savings of 471 (US $572) for every potentially low-risk patient when further risk assessment is performed and risk-based treatment according to our protocol given.
DISCUSSION

Until now, risk stratication of neutropenic febrile patients has provided a scoring system which has led to risk-adjusted
7442

management. For low-risk patients the main advantage has been the use of oral instead of intravenous antibiotics.20 The present study shows that withholding antibiotics and early hospital discharge in adult and pediatric cancer patients with febrile neutropenia at low risk for bacterial infection is also feasible. The results were based on outpatients who were selected by a new risk assessment model using objective clinical parameters and the plasma IL-8 level. The rule that every febrile neutropenic patient deserves hospitalization and intravenous broad-spectrum antibiotic therapy is already controversial. There are subpopulations that can be discharged earlier,21-23 or that may have oral broad-spectrum antibiotics without major problems,10-12 or that can be treated with oral antibiotics in the outpatient setting.24,25 This study moves a step further by showing that some patients actually do not need antibiotics at all and can be discharged from the hospital early without antimicrobial therapy. Febrile neutropenia has been reported to decrease the quality of life in cancer patients.26,27 This seems to be due to decreased physical well being and less functional capacity due to hospitalization. Therefore, dening a group of low-risk patients that need not to be admitted or can be discharged early would be an important step forward. The present study even withholds antibiotics from lowrisk patients. Previous studies have reported on the development of risk assessment models for febrile neutropenia in cancer patients. Talcott et al7 identied subgroups of adult patients presenting with febrile neutropenia at higher risk of medical complications than a remaining group of low-risk patients: hospitalized patients, patients with comorbidity, and patients with uncontrolled cancer. In a validation study, 34% of the high-risk patients were found to suffer from serious medical complications compared to only 5% of the low-risk patients.28 However, 11 of the 104 low-risk patients had bacteremia, which is a reason for caution when Talcotts scoring system is applied.
JOURNAL OF CLINICAL ONCOLOGY

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Withholding Antibiotics in Neutropenia

More recently, the Multinational Association for Supportive Care in Cancer (MASCC) developed a new risk index, which identied adult febrile neutropenic patients at low risk for medical complications.29 Favorable characteristics in this risk model were as follows: low burden of illness; absence of hypotension; absence of chronic pulmonary obstructive disease; solid tumor or no previous fungal infection in the case of hematologic malignancy; absence of dehydration; outpatient status; and age younger than 60 years. However, the burden of illness appears to be a subjective characteristic. We applied the MASCC risk index on the 90 episodes in adult patients in our study. According to MASCC, 60 episodes (67%) would be at low risk for serious medical complications. In four (6.6%) of the 60 low-risk episodes according to the MASCC risk index, a microbiologically documented bacterial infection was dened following microbiological isolates in the blood stream (Escherichia coli [n 2] and Streptococcus mitis [n 2]). Comparing to the current study, 16 of the 60 episodes at low risk according to the MASCC risk index were assigned to the low-risk group. None of the 16 low risk episodes appeared to have a positive microbiological isolate. However, the MASCC risk index was designed to select patients at low risk for medical complications who might be treated with oral antibiotics in the hospital. Our risk assessment model was designed to select patients at low risk for bacterial infection who will not receive antibiotic therapy and are discharged from the hospital early. Therefore, more stringent criteria were necessary. In children, variable admission parameters were identied that may predict low risk for invasive bacterial infection in chemotherapy treated neutropenic patients.23,30-32 Initial monocyte count of 0.1 109/L has been found as a low risk parameter.23,30 More recently, a risk prediction model with ve clinical and laboratory admission parameters (C-reactive protein 90 mg/L, hypotension, identication of relapse of leukemia as the cancer type, platelet count 50 109/L and chemotherapy within 7 days of the hospital visit) was used and was shown to be helpful in discriminating between children with high or low risk for invasive bacterial infection.12,31 In the present study, based on our analysis of our own results,13 we tried to take a step further; actually showing that with objective criteria a well-dened subpopulation of neutropenic patients could be discriminated that has a favorable outcome even without any antimicrobial therapy. Here, we show that in neutropenic children as well as in neutropenic adults a subpopulation exists that does not need antibiotic therapy at all.
REFERENCES
1. Bodey GP, Buckley M, Sathe YS, et al: Quantitative relationships between circulating

In our study, it was possible to enroll patients several times. To prove statistical independence of data obtained from subsequent episodes in one patient is difcult. However, we checked risk group classication and end points of the study and did not nd any sign of statistical dependence with this relatively small number. Therefore, we were condent to analyze all episodes in addition to initial episodes. The percentage of bacteremia in the total study population was 11%, which is low compared to the incidence of bacteremia of 23% in four large reported European Organisation for Research and Treatment of Cancer studies.9 However, the European Organisation for Research and Treatment of Cancer studies described both hospitalized patients and outpatients with febrile neutropenia, whereas our study only enrolled outpatients who are generally considered at lower risk for bacteremia. Recent studies including only outpatients have reported a similar incidence of bacteremia to ours.10,31,33 Mean cost for withholding antibiotics and early discharge in low-risk patients is 41% of the cost of a hospitalbased medium-risk strategy. An ambulatory strategy might help to reduce health care costs. Medium-risk patients in our study may be comparable to low-risk patients selected by the MASCC index. According to the guidelines, these patients may be treated with oral antibiotics in a hospital setting.5 Moreover, in children, studies with oral cexime show similar outcome compared with continued intravenous antibiotics after 48 hours.23,34 Therefore, medium-risk episodes according to our criteria may be suitable for a strategy with continued oral antibiotic treatment in the future. In conclusion, we show that there is a subpopulation of febrile neutropenic patients who actually do not need antibiotic therapy for favorable outcome. We identied this subpopulation at low risk for bacterial infection based on objective clinical parameters and the plasma IL-8 level. This risk assessment strategy can be used independent of age for children as well as adults. Further studies will help us to identify these patients more accurately and test safety in a multicenter setting. Before recommending implementation of such a risk assessment strategy in different settings, it is essential to perform an identical study designed to test the decision rule and to have an experienced medical team that is able to follow these patients at home closely.

Authors Disclosures of Potential Conicts of Interest The authors indicated no potential conicts of interest.
3. Vogel CL, Wojtukiewicz MZ, Carroll RR, et al: First and subsequent cycle use of peglgrastim prevents febrile neutropenia in patients with breast cancer: A multicenter, double-blind, placebo-controlled phase III study. J Clin Oncol 23:1178-1184, 2005

leukocytes and infection in patients with acute leukemia. Ann Intern Med 64:328-340, 1966 2. Feld R, Bodey GP: Infections in patients with malignant lymphoma treated with combined chemotherapy. Cancer 39:1018-1025, 1977

www.jco.org

7443

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.

Oude Nijhuis et al

4. Schimpff S, Satterlee W, Young VM, et al: Empiric therapy with carbenicillin and gentamicin for febrile patients with cancer and granulocytopenia. N Engl J Med 284:1061-1065, 1971 5. 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 6. Pizzo PA: Management of fever in patients with cancer and treatment-induced neutropenia. N Engl J Med 328:1323-1332, 1993 7. Talcott JA, Finberg R, Mayer RJ, et al: The medical course of cancer patients with fever and neutropenia. Clinical identication of a low-risk subgroup at presentation. Arch Intern Med 148: 2561-2566, 1988 8. Elting LS, Rubenstein EB, Rolston KV, et al: Outcomes of bacteremia in patients with cancer and neutropenia: Observations from two decades of epidemiological and clinical trials. Clin Infect Dis 25:247-259, 1997 9. Hann I, Viscoli C, Paesmans M, et al: A comparison of outcome from febrile neutropenic episodes in children compared with adults: Results from four EORTC studies. International Antimicrobial Therapy Cooperative Group (IATCG) of the European Organisation for Research and Treatment of Cancer (EORTC). Br J Haematol 99:580-588, 1997 10. Freifeld A, Marchigiani D, Walsh T, et al: A double-blind comparison of empirical oral and intravenous antibiotic therapy for low-risk febrile patients with neutropenia during cancer chemotherapy. N Engl J Med 341:305-311, 1999 11. Kern WV, Cometta A, de Bock R, et al: Oral versus intravenous empirical antimicrobial therapy for fever in patients with granulocytopenia who are receiving cancer chemotherapy. International Antimicrobial Therapy Cooperative Group of the European Organization for Research and Treatment of Cancer. N Engl J Med 341:312-318, 1999 12. Santolaya ME, Alvarez AM, Aviles CL, et al: Early hospital discharge followed by outpatient management versus continued hospitalization of children with cancer, fever, and neutropenia at low risk for invasive bacterial infection. J Clin Oncol 22:3784-3789, 2004 13. de Bont ES, Vellenga E, Swaanenburg JC, et al: Plasma IL-8 and IL-6 levels can be used to dene a group with low risk of septicaemia among cancer patients with fever and neutropenia. Br J Haematol 107:375-380, 1999 14. Kern WV, Heiss M, Steinbach G: Prediction of gram-negative bacteremia in patients with

cancer and febrile neutropenia by means of interleukin-8 levels in serum: Targeting empirical monotherapy versus combination therapy. Clin Infect Dis 32:832-835, 2001 15. Engel A, Mack E, Kern P, et al: An analysis of interleukin-8, interleukin-6 and C-reactive protein serum concentrations to predict fever, gramnegative bacteremia and complicated infection in neutropenic cancer patients. Infection 26:213221, 1998 16. Bone RC, Balk RA, Cerra FB, et al: Denitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Crit Care Med Chest 101:1644-1655, 1992 17. Report of the Second Task Force on Blood Pressure Control in Children1987. Task Force on Blood Pressure Control in Children. National Heart, Lung, and Blood Institute, Bethesda, Maryland. Pediatrics 79:1-25, 1987 18. The cardiovascular system, in Behrman R, Vaughan V (eds): Nelson Textbook of Pediatrics (ed 13). Philadelphia, PA, W.B. Saunders Company, 1987, pp. 943 19. Oostenbrink JB, Koopmanschap MH, Rutten FFH: Guidelines for cost-studies, methods and unit prices for economic evaluations in health care (in Dutch). Dutch Health Insurance Board, Diemen, 2000 20. Klastersky J: Management of fever in neutropenic patients with different risks of complications. Clin Infect Dis 39:S32-S37, 2004 21. Aquino VM, Tkaczewski I, Buchanan GR: Early discharge of low-risk febrile neutropenic children and adolescents with cancer. Clin Infect Dis 25:74-78, 1997 22. Mullen CA, Buchanan GR: Early hospital discharge of children with cancer treated for fever and neutropenia: Identication and management of the low-risk patient. J Clin Oncol 8:1998-2004, 1990 23. Klaassen RJ, Allen U, Doyle JJ: Randomized placebo-controlled trial of oral antibiotics in pediatric oncology patients at low-risk with fever and neutropenia. J Pediatr Hematol Oncol 22: 405-411, 2000 24. Hidalgo M, Hornedo J, Lumbreras C, et al: Outpatient therapy with oral ooxacin for patients with low risk neutropenia and fever: A prospective, randomized clinical trial. Cancer 85: 213-219, 1999 25. Innes HE, Smith DB, OReilly SM, et al: Oral antibiotics with early hospital discharge compared with in-patient intravenous antibiotics

for low-risk febrile neutropenia in patients with cancer: A prospective randomised controlled single centre study. Br J Cancer 89:43-49, 2003 26. Macquart-Moulin G, Viens P, Palangie T, et al: High-dose sequential chemotherapy with recombinant granulocyte colony-stimulating factor and repeated stem-cell support for inammatory breast cancer patients: Does impact on quality of life jeopardize feasibility and acceptability of treatment? J Clin Oncol 18:754-764, 2000 27. Remick SC, Sedransk N, Haase RF, et al: Oral combination chemotherapy in conjunction with lgrastim (G-CSF) in the treatment of AIDSrelated non-Hodgkins lymphoma: Evaluation of the role of G-CSF; quality-of-life analysis and long-term follow-up. Am J Hematol 66:178-188, 2001 28. Talcott JA, Siegel RD, Finberg R, et al: Risk assessment in cancer patients with fever and neutropenia: A prospective, two-center validation of a prediction rule. J Clin Oncol 10:316-322, 1992 29. 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 30. 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 31. 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 32. 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 33. Vellenga E, Uyl-de Groot CA, de Wit R, Keizer HJ, et al: Randomized placebo-controlled trial of granulocyte-macrophage colony-stimulating factor in patients with chemotherapy-related febrile neutropenia. J Clin Oncol 14:619-627, 1996 34. Shenep JL, Flynn PM, Baker DK, et al: Oral cexime is similar to continued intravenous antibiotics in the empirical treatment of febrile neutropenic children with cancer. Clin Infect Dis 32:36-43, 2001

7444

JOURNAL OF CLINICAL ONCOLOGY

Downloaded from jco.ascopubs.org on September 6, 2009 . For personal use only. No other uses without permission. Copyright 2005 by the American Society of Clinical Oncology. All rights reserved.