Correspondence
Oral Antibiotics for Febrile Patients with Neutropenia Due to Cancer Chemotherapy
N Engl J Med 2000; 342:55-58January 6, 2000
- Article
To the Editor:
In the July 29 issue, Kern et al.1 and Freifeld et al.2 reach similar conclusions about the efficacy of oral as compared with intravenous antibiotic therapy for febrile patients with neutropenia due to cancer chemotherapy who have a low risk of complications. It is a common mistake to assume that the absence of a statistical difference between two treatments means that they are equivalent. This criticism is of particular importance in a study involving a relatively small patient population, with a low failure rate in each group.
In the study by Kern et al., the end point was the rate of treatment success (i.e., recovery without a change in the regimen). Similarly, in the study by Freifeld et al., treatment was considered successful if the patient survived without a change in the regimen. The prognosis is usually good in low-risk patients with fever and neutropenia, particularly when the origin of the fever is unexplained.3 As Finberg and Talcott note in their accompanying editorial,4 the end point of these studies should have been the rate of treatment failure, not success. With treatment failure as the end point, neither study appears to have sufficient power to support a definite conclusion. In our opinion, the authors do not provide sufficient data on which to base a conclusion that oral antibiotics are as safe and effective as antibiotics delivered intravenously in low-risk patients with chemotherapy-induced neutropenia.
4 ReferencesEmmanuel Blot, M.D.
François Héron, M.D.
Rouen University Hospital, F-76031 Rouen CEDEX, France1
Kern WV ,Cometta A ,de Bock R ,Langenaeken J ,Paesmans M ,Gaya H . Oral versus intravenous empirical antimicrobial therapy for fever in patients with granulocytopenia who are receiving cancer chemotherapy. N Engl J Med 1999;341:312-318
Full Text | Web of Science | Medline2
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 1999;341:305-311
Full Text | Web of Science | Medline3
Talcott JA ,Siegel RD ,Finberg R ,Goldman L . Risk assessment in cancer patients with fever and neutropenia: a prospective, two-center validation of a prediction rule. J Clin Oncol 1992;10:316-322
Web of Science | Medline4
Finberg RW ,Talcott JA . Fever and neutropenia -- how to use a new treatment strategy. N Engl J Med 1999;341:362-363
Full Text | Web of Science | Medline
To the Editor:
Both Freifeld et al. and Kern et al. report that oral and intravenous antibiotics have equivalent efficacy in febrile patients receiving cancer chemotherapy. Even though both groups of investigators enrolled only patients with a low risk of complications, unexpectedly high rates of administration of granulocyte or granulocyte–macrophage colony-stimulating factor were reported. In the study by Freifeld et al., 200 of 232 episodes of fever and neutropenia were treated with these growth factors, and the rate of such treatment in the study by Kern et al. was also very high (>60 percent). The guidelines of the American Society of Clinical Oncology provide a rational basis for the administration of granulocyte or granulocyte–macrophage colony-stimulating factor in patients with fever and neutropenia.1 It has repeatedly been shown that treatment with growth factors shortens the hospital stay, shortens the duration of fever, and reduces the rate of use of antibiotic treatment but does not prolong survival.2,3 The guidelines suggest that growth factors be administered to high-risk patients with neutropenia, mainly those with severe sepsis or organ involvement such as renal failure or pulmonary infiltrates.
The reason for the absence of a survival advantage among patients with neutropenia who are treated with growth factors is unclear. The explanation may be that most low-risk patients recover after antibiotic treatment alone. The mortality rate among patients with severe sepsis or organ compromise is high, and most of these patients die before treatment with growth factors can improve the clinical course.
It is surprising that in both the United States and Europe, growth factors are administered in many cases in which there is no indication for this expensive treatment. I suggest that the American Society of Clinical Oncology guidelines be used in treatment protocols for patients with cancer who have chemotherapy-induced neutropenia and fever.
3 ReferencesMichael Lishner, M.D.
Meir Hospital, Kfar-Saba 44281, Israel1
1997 Update of recommendations for the use of hematopoietic colony-stimulating factors: evidence-based, clinical practice guidelines. J Clin Oncol 1997;15:3288-3288
Web of Science | Medline2
St Onge J ,Jacobson RJ . The role of hematopoietic growth factors in the treatment of neoplastic diseases. Semin Hematol 1992;29:Suppl2:53-63
Web of Science | Medline3
Anaissie EJ ,Vartivarian S ,Bodey GP , et al. Randomized comparison between antibiotics alone and antibiotics plus granulocyte-macrophage colony-stimulating factor (Escherichia-coli-derived) in cancer patients with fever and neutropenia. Am J Med 1996;100:17-23
CrossRef | Web of Science | Medline
To the Editor:
Neither clinical trials nor statistical models have led to accepted definitions of low risk, and there is no consensus on how low the risk of complications or death has to be to justify treating patients outside the hospital. There are potential hazards when low-risk patients are hospitalized, and regardless of the setting in which they are treated, a small number will die. We are not aware of any data showing that hospitalization can prevent these rare events, but we do know that admitting low-risk patients exposes them to potential iatrogenic complications and drug-resistant nosocomial infections.
In the study by Freifeld et al., the low response rates for documented infections (41 percent in the oral-therapy group and 33 percent in the intravenous-therapy group) and the high rates of use of hematopoietic growth factors raise questions about the patient population. Low-risk patients do not need treatment with growth factors, yet these agents were administered during 86 percent of episodes of fever and neutropenia in the study by Freifeld et al., suggesting that the patients were receiving dose-intensive chemotherapy and were not truly at low risk for complications. The expected duration of neutropenia was used in both studies to help identify low-risk patients. Unfortunately, clinicians are incorrect in predicting the duration of neutropenia in 10 to 15 percent of febrile episodes. As Kern et al. report, the response rate for patients with neutropenia of more than seven days' duration is much lower (49 percent in their study) than that for patients with neutropenia lasting for seven days or less (83 percent, P<0.001). Until better predictive models are available, only clinically stable outpatients with solid tumors and an expected duration of neutropenia of seven days or less should be considered to be at low risk.
The National Comprehensive Cancer Network has recently published guidelines1 for risk assessment that include suggestions for outpatient treatment of low-risk patients. On the basis of a decade of clinical investigations, we disagree with Finberg and Talcott's closing comment. We can with confidence identify and safely treat low-risk febrile patients with neutropenia outside the hospital.
1 ReferencesEdward B. Rubenstein, M.D.
Kenneth V.I. Rolston, M.D.
Young Jun Kim, M.P.H.
University of Texas M.D. Anderson Cancer Center, Houston, TX 770301
National Comprehensive Cancer Network
Web of Science | Medline
To the Editor:
Both Freifeld et al. and Kern et al. suggest that oral therapy may improve the quality of life in febrile patients with neutropenia, but this suggestion is not supported by the data from their studies. Since neither group reported an evaluation of the quality of life, we must depend on the data on toxicity. In the study by Kern et al., gastrointestinal symptoms occurred in 15 percent of the patients in the oral-therapy group but in only 2 percent of those in the intravenous-therapy group. In the study by Freifeld et al., nausea and vomiting or diarrhea occurred during 28 percent of episodes of fever and neutropenia treated with oral antibiotics but in only 6 percent of those treated with intravenous antibiotics (for all adverse effects, the proportions were 29 percent and 7 percent, respectively). Among the patients with documented infections — an important subgroup with a generally higher rate of complications than among patients with fever of unknown origin — there was intolerance of the active study drug in 22 percent of episodes in the oral-therapy group but in none of those in the intravenous-therapy group. Thus, the quality of life may well be reduced by oral antimicrobial treatment because of its gastrointestinal toxicity in patients with neutropenia during cancer chemotherapy.
Orhan Sezer, M.D.
Universitätsklinikum Charité, 10098 Berlin, GermanyAuthor/Editor Response
The authors reply:
To the Editor: The size of the sample in our double-blind, controlled equivalence trial provided an 80 percent probability that the upper 95 percent confidence limit for the true difference in the success rates for oral treatment and intravenous treatment was no more than 15 percent. Since the observed difference was 3 percent in favor of oral medications, the two regimens were declared equivalent. These statistical methods are standard and are not, as Drs. Blot and Héron imply, merely extrapolated assumptions made on the basis of data from a “relatively small patient population.” It is unclear how the evaluation of treatment failure as the end point of the study would have been beneficial, since it is simply the complement of the end point we chose — treatment success.
As Rubenstein et al. point out, there are no universally accepted definitions of low risk. However, there is a concurrence that patients considered to be at low risk should have a short anticipated duration of neutropenia, variably defined as less than or equal to 7 days or 10 days, and no coexisting medical conditions. Rubenstein et al. suggest that clinicians are frequently incorrect in predicting the duration of neutropenia but must nonetheless make such a prediction in defining low risk. We chose the more liberal definition — neutropenia for 10 days or less — in order to include patients with somewhat longer periods of neutropenia. The mean duration of neutropenia in our trial was 3.6 days (range, 1 to 14), indicating that clinicians can accurately predict the duration of neutropenia in most cases.
Since nearly all the study patients were also enrolled in trials of treatment protocols for primary cancer at the National Cancer Institute, we had no control over the administration of colony-stimulating factors. The profound influence of these agents on the duration of neutropenia, fever, and the hospital stay and on the overall incidence of documented infection must be recognized, as Dr. Lishner suggests.1 Concomitant treatment with colony-stimulating factors might convert some longer, higher-risk periods of neutropenia to those conforming to the definition of low risk. However, such treatment usually reduces the period of neutropenia by a few days, at most.2,3 Since the mean duration of neutropenia in our study was less than four days, it is unlikely that treatment with colony-stimulating factors had a dramatic effect on many of the episodes in our study.
Having shown that oral and intravenous regimens are medically equivalent, we agree with Dr. Sezer that it is now essential to determine whether oral antibiotics really do (or do not) improve the quality of life for low-risk patients with cancer. Although oral medication had some type of adverse gastrointestinal effect in 28 percent of episodes of fever and neutropenia, the rate of attrition due to these effects was only 9 percent. Many patients with cancer might welcome the opportunity to be free of intravenously delivered antibiotics and to be treated at home, even if it meant some gastrointestinal distress due to oral antibiotics.
3 ReferencesAlison G. Freifeld, M.D.
University of Nebraska Medical Center, Omaha, NE 68198-5400Seth M. Steinberg, Ph.D.
National Cancer Institute, Bethesda, MD 20892Philip A. Pizzo, M.D.
Children's Hospital, Boston, MA 021151
Johnston EM ,Crawford J . Hematopoietic growth factors in the reduction of chemotherapeutic toxicity. Semin Oncol 1998;25:552-561
Web of Science | Medline2
Crawford J ,Ozer H ,Stoller R , et al. Reduction by granulocyte colony-stimulating factor of fever and neutropenia induced by chemotherapy in patients with small-cell lung cancer. N Engl J Med 1991;325:164-170
Full Text | Web of Science | Medline3
Kawano Y ,Takaue Y ,Mimaya J , et al. Marginal benefit/disadvantage of granulocyte colony-stimulating factor therapy after autologous blood stem cell transplantation in children: results of a prospective randomized trial. Blood 1998;92:4040-4046
Web of Science | Medline
To the Editor:
Drs. Blot and Héron should keep in mind that our study was specifically designed to prove the equivalence of success rates for oral and intravenous therapy. The null hypothesis was that the absolute value of the difference between the success rates in the two treatment groups would be greater than 10 percent. A rejection of the null hypothesis was required to conclude that the treatment regimens were equivalent. Both per protocol and intention-to-treat analyses of success rates showed an equivalence at P values of less than 0.05. Thus, the likelihood that there was a difference in success rates between oral and intravenous therapy but that the study failed to detect it is small. If the end point had been treatment failure rather than success, it would not have changed the statistical methods, sample size, or results. Admittedly, the power of the study to detect a difference in secondary end points, such as survival, was limited.
The frequent use of hematopoietic growth factors, noted by Dr. Lishner, is in fact surprising. In many centers, patients are being treated according to research protocols that may include guidelines for the use of growth factors that differ from those issued for routine practice. Most patients who were given growth factors received them before the onset of fever. The use of growth factors was similar in the two study groups. Therefore, it did not alter the primary results of the study.
We appreciate the comments of Rubenstein et al. Although there is no statistical model that reliably predicts the absence of complications in an outpatient setting, outpatient therapy for fever and neutropenia has been an option for a number of years.1,2 Many factors related to the patient, his or her family, the patient–physician relationship, the institution, the available outpatient settings, and the geographic area need to be considered, and good clinical judgment is indispensable, making it difficult to develop a formal prediction rule that is universally applicable and acceptable. We emphasize that our results should not be interpreted as demonstrating that oral therapy in an outpatient setting is the preferred regimen for low-risk patients, but it now seems clear and adequately proved that the success of empirical antibiotic therapy does not depend on the means of drug administration.
Our study was not designed to measure the quality of life. However, unlike Dr. Sezer, we believe that oral therapy for fever and neutropenia will improve the quality of life for many patients, particularly if they can be treated as outpatients. At least two groups of investigators have reported that patients have favorable attitudes toward treatment at home.3,4 Oral drugs were associated with more frequent gastrointestinal adverse events than were intravenous drugs in our study, but the overall rates of adverse events (36 percent in the oral-therapy group and 31 percent in the intravenous-therapy group) or of adverse events possibly or probably related to the study drugs (16 percent and 15 percent, respectively) were similar. How the imbalance in the types of adverse events in the two groups affected the patients' quality of life is unknown. We wish to stress that in both our study and that reported by Freifeld et al., the rates of successful treatment were similar on an intention-to-treat basis (i.e., with modifications of treatment related to intolerance or any other adverse events included in the analysis).
4 ReferencesWinfried V. Kern, M.D.
Medizinische Universitätsklinik und Poliklinik, D-89070 Ulm, GermanyAlain Cometta, M.D.
Centre Hospitalier Universitaire Vaudois, CH-1011 Lausanne, Switzerland1
Uzun O ,Anaissie EJ . Outpatient therapy for febrile neutropenia: who, when and how? J Antimicrob Chemother 1999;43:317-320
CrossRef | Web of Science | Medline2
Davis DD ,Raebel MA . Ambulatory management of chemotherapy-induced fever and neutropenia in adult cancer patients. Ann Pharmacother 1998;32:1317-1323
CrossRef | Web of Science | Medline3
Talcott JA ,Whalen A ,Clark J ,Rieker PP ,Finberg R . Home antibiotic therapy for low-risk cancer patients with fever and neutropenia: a pilot study of 30 patients based on a validated prediction rule. J Clin Oncol 1994;12:107-114
Web of Science | Medline4
Shemesh E ,Yaniv I ,Drucker M , et al. Home intravenous antibiotic treatment for febrile episodes in immune-compromised pediatric patients. Med Pediatr Oncol 1998;30:95-100
CrossRef | Medline
To the Editor:
The comments of Blot and Héron raise the knotty question of when to conclude, on the basis of empirical evidence, that two treatments do not have important differences in efficacy. The calculations of statistical power for the best-designed studies contain an irreducibly subjective element, the clinically significant difference that the study was designed to exclude. Furthermore, the truth of the null hypothesis cannot be evaluated without reference to information outside a particular study (the Bayesian approach).1 In the current case, we concluded that the consistent results of the studies by Freifeld et al. and Kern et al., which were conducted in populations with different characteristics and both of which found fewer clinical failures and other adverse events with oral antibiotic regimens than with intravenous treatment, made clinically important differences in efficacy unlikely.
Although we have long appreciated the efforts of Rubenstein and colleagues at the M.D. Anderson Cancer Center to develop outpatient treatment strategies for patients with fever and neutropenia,2 we disagree with their comments. Thoughtful clinicians may differ in their interpretation of data. We believe that multivariable modeling, the approach we used to define low-risk patients with febrile neutropenia,3 provides the best opportunity to compare and validate the best predictors of risk. However, we are aware that other groups, including the group at the M.D. Anderson Cancer Center, have postulated their own criteria for low risk in pilot trials of outpatient therapy for fever and neutropenia, justifying their own criteria retrospectively on the basis of infrequent serious complications. Similarly, experienced physicians may estimate differently the net hazard or benefit when the increased risk of iatrogenic harm to inpatients is exchanged for the decreased opportunity to identify correctable new medical problems because of the diminished surveillance of outpatients. These conflicting opinions can be resolved only by a randomized clinical trial with sufficient statistical power.
The reference by Rubenstein and colleagues to the National Comprehensive Cancer Network guidelines illustrates the distinction between opinion and evidence that we were trying to make in our editorial. The network produces guidelines based on expert opinion, relying on group dynamics and the proliferation of acceptable options to resolve conflicts. Evidence-based guidelines systematically characterize the strength of existing evidence and endorse only approaches supported by strong evidence. Although Rubenstein and colleagues believe that low-risk patients with fever and neutropenia can be identified and safely treated outside the hospital, we remain convinced that evidence-based guidelines would not yet support their conclusion.
3 ReferencesRobert W. Finberg, M.D.
James A. Talcott, M.D.
Harvard Medical School, Boston, MA 021151
Goodman SN . Toward evidence-based medical statistics. 1. The P value fallacy. Ann Intern Med 1999;130:995-1004
Web of Science | Medline2
Rubenstein EB ,Rolston K ,Benjamin RS , et al. Outpatient treatment of febrile episodes in low-risk neutropenic patients with cancer. Cancer 1993;71:3640-3646
CrossRef | Web of Science | Medline3
Talcott JA ,Siegel RD ,Finberg R ,Goldman L . Risk assessment in cancer patients with fever and neutropenia: a prospective, two-center validation of a prediction rule. J Clin Oncol 1992;10:316-322
Web of Science | Medline
