Information sources and eligibility criteria

The information specialist (SB) searched the following electronic databases: Medline, Premedline, Embase, Cochrane Library, Cinahl, BNI, Psychinfo, Web of Science (SCI & SSCI), ISI proceedings and Biomed Central. The search was restricted to published randomised trials and systematic reviews of randomised trials. The final search was done on 7^th of November 2011.

Selection of studies

The information specialist (SB) conducted the first screen of the literature search results. One reviewer (KF) then selected potentially eligible studies by comparing their title and abstract to the inclusion criteria set out by the PICO question. Full text articles were obtained for studies identified as potentially relevant. These were read and checked against the inclusion criteria.

Results of the literature searches

Figure 15.1Study flow diagram

The literature searches identified 139 potentially relevant studies of which four were included as evidence. The structure of this question is such that it can only be properly answered by randomised studies comparing neutropenic patients with persistent fever, despite having been treated with an empiric antibiotic, to either stay on the empiric therapy or have some sort of treatment modification i.e. a different drug to replace or add to the empiric antibiotic. However, the overwhelming majority of papers identified in the literature search described studies in which patients had stopped empiric antibiotics before being randomised to one or two second line drugs. These studies would not answer this question.

The evidence base is very poor, consisting of four randomised studies, two of which are more than twenty years out of date. Patients (N=461 patients in total) with low granulocyte counts and persistent fever were randomised to either remain on the empiric antibiotic (alone or with an added placebo) or to primary treatment with the addition of another agent. The point at which these studies were initiated i.e. the number of days of persistent fever, varied between two and seven days. The length of stay and the incidence of over-treatment were not specifically addressed and nor was the patients' quality of life. None of the studies dealt adequately with the methods of randomisation, allocation or blinding and, although some authors stated that appropriate statistics had been used for data analysis, the details were sometimes scant or absent and very few outcomes had more than a P (probability) value reported. For these reasons, all four studies have been classified by GRADE as being of ‘low’ or ‘very low’ quality. The variability of data and study design precluded pooling.

Evidence summary

Generally, none of the studies demonstrated a significant difference between patients kept on empiric antibiotics and those given an additional drug or drugs (Table 15.1). The general consensus seemed to be that patients seemed to respond to the initial antibiotic treatment eventually and that glycopeptides in particular could potentially be of more harm than benefit if the infectious agent did not warrant such treatment. Bearing in mind the age of these studies, these points may no longer be of relevance.

Table 15.1

GRADE evidence profile for optimal time to change the primary empiric treatment in unresponsive fever.

Pizzo et al (1981) reported on fifty patients who, having received empiric antibiotics for fever and granulocytopenia of unknown infectious aetiology, had failed to respond to treatment after seven days. These patients were randomised to stop empiric antibiotics (group 1), continue with empiric antibiotics (group 2) or continue empiric antibiotics with the addition of amphotericin B (group 3). Six patients in group 1 experienced shock compared with no patients in the other two groups (P<0.001). The incidence of infectious complications was significantly higher (N=9) in group 1 compared with group 3 (N=2) (P=0.013) but not between group 1 and group 2 (N=6) i.e. for patients stopping versus continuing antibiotics. Although no statistical analyses were presented, the low patients numbers, low event rates and wide ranges of data suggest that there was no significant difference between time to the resolution of granulocytopenia or to defervescence between the three groups. The number of non-infectious complications did not differ significantly. The numbers of fatalities appeared to be similar between groups: N=5 in groups 1 and 2 versus N=3 in group 3. There were more patients with fungal infections in group 2 (N=5) compared with the other two groups but this might be a random effect but no evidence was offered to suggest causality.

A study by the EORTC antimicrobial therapy co-operative group (1989) compiled results from two consecutive trials on the use of empiric antibodies in patients with fever and granulocytopenia. One hundred and thirty-two patients who were unresponsive to treatment after four days, were randomised to continue antibiotics with (group 1) or without (group 2) amphotericin B. Clinical response was assessed five days after randomisation and considered a failure if the patient remained febrile. Under this criterion, 47/68 (69%) of patients in group 1 versus 34/64 (53%) of patients in group 2 experienced treatment success (P=0.09). More patients with a clinically documented infection at day 4, had a positive clinical response with combined treatment than with antibiotics alone (P=0.03). Similarly, patients that had not received prior anti-fungal prophylaxis had a better response to the combined treatment regime than antibiotics only (P=0.04) but other sub-group comparisons were not statistically significant. Fewer patients in group 1 had died by day 30 (11 versus 14 (P=0.039) but most deaths were described as being due to ‘other causes’ rather than being specifically treatment related. More patients in group 2 developed fungal disease than in group 1 but the difference was not significant. All sub-group analyses were of very low patient number.

Cometta et al. (2003) reported the results of a prospective double blinded trial in which one hundred and sixty-five patients who had persistent fever after 48-60 hours, were randomised to receive an empiric broad spectrum antibiotic plus vancomycin (group 1) or with saccharose solution (group 2). The main outcome of interest was the rate of fever resolution at three days post randomisation, which was not significantly different between study arms: 82/86 (95%) in group 1 versus 73/79 (92%) in group 2. More than half of the patients in both groups had their therapy modified, either by adding a glycopeptide to vancomycin or by stopping the placebo and giving amphotericin B. There was no significant difference in the time to fever resolution between groups, regardless of whether the treatment regime had been modified or not. Fewer (N=4) patients died in group 1 compared with group 2 (N=8) but the differences are unlikely to have been statistically significant. More patients in group 1 (N=9) experienced treatment related side effects compared with group 2 (N=3). The study had very low patient numbers and event rates and was underpowered to have detected a clinically meaningful difference between comparators for the main outcome. The study was closed for this reason.

Erjavec et al. (2000) conducted a randomised double blinded placebo-controlled study of one hundred and fourteen patients with febrile neutropenia who had persistent fever after three to four days of treatment with an empiric antibiotic. Group 1 continued with imipenum and added teicoplanin whilst group 2 had imipenum with a placebo. The primary outcome was the rate of treatment response after 72 hours. There was no significant difference between study arms: 25/56 (45%) in group 1 versus 27/58 (47%). The number of deaths throughout the study was 6 in group 1 and 4 in group 2. Many of the patients had received anti-bacterial prophylaxis and some had also been given G-CSF. The numbers of patients and event rates were low.

Evidence Statements

Information sources and eligibility criteria

The full search strategy is available in appendix X. We restricted the search to published randomised trials and systematic reviews of such trials. The search was done on the 23^rd of November 2010 and updated on 2^nd November 2011.

Selection of studies and data synthesis

The information specialist (SA) performed an initial screening of the literature search results. One reviewer (MSH) then selected possibly eligible studies by comparing their title and abstract to the inclusion criteria in the PICO question.

It was anticipated that evidence would come from trials comparing different times for switching to intravenous to oral antibiotics. However, in the absence of such studies, subgroup analyses was done (according to time-of-switch) in trials which compared switching from intravenous to oral antibiotics with continued intravenous antibiotics.

Results of the literature searces

Figure 16.1Study flow diagram

90 studies were identified in the literature searches. Of these, 83 were excluded because they were narrative reviews (N = 6), not in PICO (N = 59), not RCT (N = 16), reporting data already included (N = 1) or a comment (N = 1).

One Cochrane review (Vidal et al., 2004) was indentified for inclusion. The review included 6 RCTs (Flaherty et al., 1989; Giamarellou et al., 2000; Mullen et al., 1999; Paganini et al., 2000, 2003; Shenep et al., 2001). These 6 trials were also checked individually for outcomes not reported in the Cochrane review.

Detailed information about the populations, interventions, outcomes and overall risk of bias in the included trials is given in the Evidence and Grade Tables below.

Evidence Statements

Duration of fever / Treatment failure

Duration of fever was not reported in the systematic review (Vidal, et al., 2004). Three of the included trials reported this outcome but none of these reported a statistically significant difference in the duration of fever between treatment groups.

Vidal, et al., (2004) reported treatment failure as a composite outcome comprising one or more of the following: death; persistence, recurrence or worsening of clinical signs or symptoms of presenting infection; any addition to or modification of the assigned intervention Low quality evidence suggested no significant difference in the rate of treatment failure in the IV-to-oral group compared to the IV only group, RR 1.07 (0.9 to 1.27).

Table 16.1

GRADE evidence profile, Switching from intravenous to oral antibiotic therapy.

Results of the literature searches

Figure 17.1Study flow diagram

Two Randomised Controlled Trials (RCTs) evaluating duration of inpatient care in the management of suspected bacterial infection (Innes et al 2003 and Santolaya et al 2004) were identified. One non-randomised comparative feasibility study was identified (Lau et al 1994). Eight prospective observational studies (Cherif et al 2006; Girmenia et al 2007; Klastersky 2006; Nijhuis et al 2005; Dommett 2009; Lehrnbecher 2002 and Bash 1994) and seven retrospective observational studies (Tordecilla et al 1994; Aquino 1997; Mullen et al 1990; Griffin et al 1992; Wacker et al 1997; Hodgson-Viden et al 2005 and Tomiak et al 1994) evaluated optimal inpatient duration. We identified one survey of the management of febrile neutropenia. One systematic review published 11 years ago was also identified (Castagnola et al 2000).

Evidence statements

The evidence is summarized in Tables 17.1 and 17.2.

Table 17.1

GRADE evidence profile for early discharge with continued inpatient care.

Table 17.2

Early discharge criteria and rates.

REFERENCES

• Innes HE, Smith DB, O'Reilly SM, Clark PI, Kelly V, Marshall E. 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. British Journal of Cancer. 2003;89:43–49. [PMC free article: PMC2394220] [PubMed: 12838298]
• Santolaya ME, Alvarez AM, Aviles CL, Becker A, Cofre J, Cumsille MA, 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. Journal of Clinical Oncology. 2004;22:3784–3789. [PubMed: 15365075]
• Lau RC, Doyle JJ, Freedman MH, King SM, Richardson SE. Early discharge of pediatric febrile neutropenic cancer patients by substitution of oral for intravenous antibiotics. Pediatric Hematology & Oncology. 1994;11:417–421. [PubMed: 7947014]
• Bash RO, Katz JA, Cash JV, Buchanan GR. Safety and cost effectiveness of early hospital discharge of lower risk children with cancer admitted for fever and neutropenia. Cancer. 1994;74:189–196. [PubMed: 8004575]
• Cherif H, Johansson E, Bjorkholm M, Kalin M. The feasibility of early hospital discharge with oral antimicrobial therapy in low risk patients with febrile neutropenia following chemotherapy for hematologic malignancies. Haematologica. 2006;91:215–222. [PubMed: 16461306]
• Girmenia C, Russo E, Carmosino I, Breccia M, Dragoni F, Latagliata R, et al. Early hospital discharge with oral antimicrobial therapy in patients with hematologic malignancies and low-risk febrile neutropenia. Annals of Hematology. 2007;86:263–270. [PubMed: 17225113]
• Klastersky J, Paesmans M, Georgala A, Muanza F, Plehiers B, Dubreucq L, et al. Outpatient oral antibiotics for febrile neutropenic cancer patients using a score predictive for complications. Journal of Clinical Oncology. 2006;24(25):4129–4134. [PubMed: 16943529]
• Tomiak AT, Yau JC, Huan SD, Cripps MC, Goel R, Perrault DJ, et al. Duration of intravenous antibiotics for patients with neutropenic fever. Annals of Oncology. 1994;5:441–445. [PubMed: 8075051]
• Lehrnbecher T, Stanescu A, Kuhl J. Short courses of intravenous empirical antibiotic treatment in selected febrile neutropenic children with cancer. Infection. 2002;30(1):17–21. [PubMed: 11876510]
• Dommett R, Geary J, Freeman S, Hartley J, Sharland M, Davidson A, et al. Successful introduction and audit of a step-down oral antibiotic strategy for low risk paediatric febrile neutropaenia in a UK, multicentre, shared care setting. European Journal of Cancer. 2009;45:2843–2849. [PubMed: 19616427]
• Tordocilla CJ, Campbell BM, Joannon SP, Rodriguez RN. Neutropenia and fever. Revista Chilena de Pediatria. 1994;65(5):260–263.
• Aquino VM, Buchanan GR, Tkaczewski I, Mustafa MM. Safety of early hospital discharge of selected febrile children and adolescents with cancer with prolonged neutropenia. Medical and Pediatric Oncology. 1997;28(3):191–195. [PubMed: 9024515]
• Mullen CA, Buchanan GR. Early hospital discharge of children with cancer treated for fever and neutropenia: identification and management of the low-risk patient. Journal of Clinical Oncology. 1990;8:1998–2004. [PubMed: 2230891]
• Wacker P, Halperin DS, Wyss M, Humbert J. Early hospital discharge of children with fever and neutropenia: a prospective study. Journal of Pediatric Hematology/Oncology. 1997;19:208–211. [PubMed: 9201142]
• Hodgson-Viden H, Grundy PE, Robinson JL. Early discontinuation of intravenous antimicrobial therapy in pediatric oncology patients with febrile neutropenia. BMC Pediatrics. 2005;5:10. [PMC free article: PMC1156908] [PubMed: 15904510]
• Griffin TC, Buchanan GR. Hematologic predictors of bone marrow recovery in neutropenic patients hospitalized for fever: implications for discontinuation of antibiotics and early discharge from the hospital. Journal of Paediatrics. 1992;121:28–33. [PubMed: 1625089]
• Nijhuis CO, Kamps WA, Daenen SM, Gietema JA, van der Graaf WT, Groen HJ, et al. Feasibility of withholding antibiotics in selected febrile neutropenic cancer patients. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 2005;23(30) [PubMed: 16234511]
• Santos-Machado TM, De Aquino MZ, Almeida MTA, Bakhit S, Cristofani LM, Maluf PT, et al. Short-term intravenous antibiotic therapy and early discharge of febrile neutropenic patients. International Journal of Pediatric Hematology/Oncology. 99 A.D.;6(1):33–38.
• Innes H, Lim SL, Hall A, Chan SY, Bhalla N, Marshall E. Management of febrile neutropenia in solid tumours and lymphomas using the Multinational Association for Supportive Care in Cancer (MASCC) risk index: feasibility and safety in routine clinical practice. Supportive Care in Cancer. 2008;16:485–491. [PubMed: 17899215]
• Castagnola E, Paola D, Giacchino R, Viscoli C. Clinical and laboratory features predicting a favorable outcome and allowing early discharge in cancer patients with low-risk febrile neutropenia: a literature review. Journal of Hematotherapy & Stem Cell Research. 2000;9:645–649. [PubMed: 11091488]

Information sources and eligibility criteria

The search strategy will be available in the full guideline.

We restricted the search to published randomised trials and systematic reviews of such trials. The search was done on the 9^th of May 2011 and updated on 7^th November 2011.

Selection of studies and data synthesis

The information specialist (SB) performed an initial screening of the literature search results. One reviewer (MSH) then selected possibly eligible studies by comparing their title and abstract to the inclusion criteria in the PICO question.

It was anticipated that results from studies which compare early stopping with continuing empiric antibiotics until afebrile with a recovered neutrophil count would be pooled with the potential of doing subgroup analyses to compare the different stopping criteria for empiric antibiotics (e.g., neutrophil count) used by the included randomised trials.

Results of the literature searches

Figure 18.1Study flow diagram

Description of included studies

59 studies were identified in the literature searches. Of these, 55 were excluded because they were narrative reviews (N = 9), not in PICO (N = 36), not RCT (N = 9), or a protocol (N = 1).

Four RCTs were indentified for inclusion (Bjornsson, 1977; Klaassen, 2000; Pizzo, 1979; Santolaya, 1997). Two of these studies were conducted in children (Klaassen, 2000; Santolaya, 1997), one was conducted in adults (Bjornsson, 1977) and one study was conducted in a mixed population of children and adults (Pizzo, 1979). These four studies examined a variety of different antibiotic regimens. Detailed information about the populations, interventions, outcomes and overall risk of bias in the included trials is given in the Evidence and GRADE profile (Table 18.1) below.

Table 18.1

GRADE evidence profile for duration of empiric antibiotic therapy.

Evidence statements