Malady: Febrile Neutropenia
Myth: Antibiotic therapy needs to be continued in patients with febrile neutropenia, especially those who are high-risk with haematological malignancy or undergoing transplant, until marrow recovery (absolute neutrophil count [ANC] increasing above 0.5).
Remedies:
- Culture and septic screen appropriately to identify clinically and microbiologically documented infections to simplify therapy and treatment duration.
- Extensive emerging evidence suggests both low- and high-risk patients can safely stop antibiotics when afebrile, clinically well and have completed an adequate (empirical or directed/targeted) treatment course irrespective of neutrophil count.
- Consider non-infectious and atypical infectious (especially invasive fungal disease) causes for those with persistent fevers and negative cultures despite broad-spectrum antibiotic therapy.
Dear reader, picture this. A forlorn haematologist, weary from a hard day of thinking, looking for a pick-me-up. Staring into their Mirror of Erised – what do they see that they desire the most? An appropriate request for intravenous immunoglobulin? A Howell-Jolly body on a blood smear? No, it is of course piperacillin holding hands with tazobactam.
Now, it is a tricky predicament to treat a severely unwell patient who cannot mount an immune response. Though simple and protocolised enough to start, antibiotics for febrile neutropenia meet very variable ends. It is an area of great practice variation, where the decision is not just down to logic and reason but instead heavily relies on the risk tolerance of the treating physician. There is surprisingly discordant and outdated advice on the topic but recent guidance suggests we are often conservative by following traditional endpoints for antibiotic therapy.
Defining the Problem
First off – definitions. Because nobody actually tells you this, the true definition of febrile neutropenia requires a single temperature ≥ 38.3C or a 1-hour sustained temperature ≥ 38C. A microbiologically documented infection (MDI) is one where we’ve grown a pathogen that corresponds to a site of infection. A clinically documented infection (CDI) is one where there is a clinical focus of infection (usually cellulitis and pneumonia) where it was not possible to grow a pathogen. And we of course have the pyrexias of unknown origin (PUO) – which generally are the patients where it is most difficult to make this judgement call.
The standard of care is to use anti-Pseudomonal beta-lactam therapies in the initial treatment of neutropenic fevers. This generally comes down to one of piperacillin-tazobactam, cefepime or a carbapenem in most institutions. Interestingly, historical data suggests that the causes of microbiologically documented infections have shifted from Gram-negative dominant (in the infancy of febrile neutropenic syndromes) to Gram-positive organisms thereafter (attributed to a combination of long-access lines, widespread antimicrobial use effective against Gram-negative organisms, current chemotherapy regimens) [1].
Risk Stratification
Febrile neutropenia carries high morbidity and mortality, that much is known. But to get further within the weeds, there are more shades of grey one can appreciate. The degree of neutropenia is a predictor of poorness outcomes. Neutropenia is a term liberally used, but the risk of serious infections increases at a threshold of <0.5×109 and even more with agranulocytosis [2]. Similarly, the duration of neutropenia is a factor. There are therapies, for instance haematopoietic stem cell transplantation, where the anticipated duration of neutropenia may far exceed that of a short burst of a chemotherapeutic agent. There are disease-associated factors (such as conditioning regimens for transplantation patients, AML patients being high-risk, solid organ tumour patients being lower risk), patients’ comorbidities, additional immunosuppression (such as graft-versus-host disease prophylaxis) to name a few other considerations.
As with all things in medicine, there are scoring systems to also guide stratification. A more commonly used tool in this setting is the MASCC Risk Index which was initially developed as tool to guide which patients can be managed in an outpatient setting or with oral therapies alone for their febrile neutropenia presentations – a noble cause indeed. Those with a higher MASCC score (>= 21) are considered lower risk, and vice versa. These calculations help to guide the crux of the issue – which patients can we safely de-escalate or stop antibiotic therapy for, to avoid the emergence of antimicrobial resistance and most importantly, not cause harm to the patients.
#1 Continuation – Awaiting Engraftment / Recovery of Neutropenia
A common practice is to continue therapy until recovery of neutropenia, past the 0.5×10^9 mark which is seen as evidence of marrow recovery and lower-risk of severe infection. This may also be spurred on by filgrastim/G-CSF agents.
As put forth in the 2010 IDSA guidelines, empirical therapy is generally recommended to continue in CDI, MDI and PUO until marrow recovery [3]. Closer to home, this is echoed in the 2011 Australian guidance [4] where patients with PUO who defervesce with 3-5 days of parenteral empirical therapy are recommended to continue antibiotics until marrow recovery. In those with prolonged durations of (expected) neutropenia, it defers decision making to the treating clinician, with a suggestion to consider cessation if skin and mucous membranes are intact, and no invasive procedures or further myelosuppressive therapies are planned [4]. A similar approach is outlined in the paediatric guidance from the Royal Children’s Hospital Melbourne. RCH makes a distinction to deem low-risk patients eligible for cessation of antimicrobials when afebrile with close follow-up, irrespective of counts [5].
The basis for this seems to have originated from this 1979 study [6] that demonstrated higher rates of adverse outcomes with fevers and infection after discontinuation of antibiotics when granulocytopenic. The antibiotic regimen studied here was a combination of Keflin (cefalotin), gentamicin and carbenicillin – if two of these don’t ring a bell, that proves my point of antiquity quite sufficiently. You can imagine however that this prolongs treatment duration for patients who may not have a true bacterial infection or have already completed an adequate and safe treatment course. Complicating this is the fact that beta-lactams may cause drug fevers, and along with vancomycin, they are known causes of myelosuppression.
#2 De-Escalation – Oral Step-Down
Another option proposed for febrile neutropenia with PUO is oral step-down therapy, which is generally the domain of oral fluoroquinolones (mainly levofloxacin and ciprofloxacin for their anti-Pseudomonal activities) and oral amoxicillin-clavulanic acid for its broad spectrum of activity including anaerobes. The aforementioned IDSA guidelines recommend consideration of an oral fluoroquinolone until marrow recovery in those with a CDI/MDI who have completed an appropriate therapeutic course and are asymptomatic [3]. The authors of two informative UpToDate articles on the management of febrile neutropenia follow an approach of early antibiotic cessation in PUO after 48 hours of apyrexia in low-risk patients, and err on the side of fluoroquinolone prophylactic step-down in high-risk neutropenic patients after 72 hours without fevers [7,8].
In fact many American centres recommend prophylactic antibiotic use for high-risk patients with prolonged durations of anticipated neutropenia (such as stem cell transplantation and CAR-T therapies) [9], but the strength of this recommendation has been questioned more recently [10]. A complication that will be increasingly faced with this approach is the rapid development of resistance – recent European data demonstrates rates of >50% fluoroquinolone resistance in Gram-negative bacteremias in haematology patients [11]. This has the potential to compromise a very valuable oral antibiotic option for Pseudomonal and other drug-resistant bacterial infections.
#3 Discontinuation – Stop When Persistently Afebrile Irrespective of Counts
More recent evidence has begun to emerge, on the background of a multitude of recent papers of varying quality in the hierarchy of evidence-based medicine (3 major meta-analyses, multiple RCTs and lots of observational studies). The general consensus suggests that bathing patients in the river Tazocin until their marrow is shocked into recovery is no longer necessary.
This is a matter that has seen profound change in expert opinion. The 2010 IDSA guidelines [3] did not suggest cessation of antibiotics prior to ANC recovery in high-risk patients as an option, and a recent paediatric guideline from 2023 identified this as a knowledge gap [12].
A landmark clinical trial, the aptly named How Long study, showed that discontinuation of empirical therapy was safe after 72 hours of apyrexia when clinically well, irrespective of neutrophil count amongst high-risk haematological febrile neutropenia patients [13]. On the back of this, and multiple other recent studies, the 10th European Conference on Infections in Leukaemia (ECIL-10) in late 2025 recommends antibiotic discontinuation regardless of neutrophil count (summarised in a neat table as below) [14]! The ECIL have been comparative trailblazers on this front, with the previous ECIL-4 guidelines published in 2013 also advocating for this [15].
Notably, the data is not airtight and suffers from the flaws of being heterogeneous. Many intermediate-risk patients such as those undergoing autologous stem-cell transplantation (as opposed to the higher-risk allogeneic transplantation which generally results in longer durations of neutropenia), and those with lymphoma (as opposed to leukaemia) comprised considerable parts of some cohorts. Some studies picked endpoints that may be less clinically informative, focusing on days without empirical antibiotic therapy as opposed to more directly patient-forward outcomes (e.g. mortality benefit). Notably, adoption rates of the ECIL-4 guidelines were low amongst European Bone Marrow Transplant centres [16], likely reflecting clinician hesitation. However, there is a reproducible trend of truth in the safe cessation of antibiotics in neutropenic patients who are well and afebrile prior to marrow recovery.
#4 Road Less Travelled – Stop Antibiotics Despite Fevers?
A brief note for the Palme d’Or. Can we stop antibiotics in neutropenic patients with PUO? What are we treating anyway, other than our anxieties? Across guidelines, escalation of antibiotics for fevers alone is dissuaded, but what about the converse?
We’re not quite there yet, is the answer, with one RCT showing a higher mortality rate when empiric carbapenem therapy was ceased at 72 hours in high-risk neutropenia patients who are still persistently febrile [17 – as part of a secondary sub-group analysis in their appendix, so caution is advised]. Realistically, it is also difficult for clinicians treating high-risk patients to not re-commence antibiotics when the patient spikes another fever two minutes after they cease.
Continuing antibiotics till marrow recovery, even in high-risk febrile neutropenic patients, may (and perhaps should) soon become a relic of the past, with some evidence even suggesting shorter term antibiotic therapy may confer mortality benefits [18]. It is also clear that it is simpler to treat those with a CDI or MDI, so emphasising collection of microbiological samples prior to antibiotics (when safe and without delaying commencement of therapy) would help that cause.
Giants’ Shoulders:
[2] Highly influential historical study that shows a dose-response relationship between lower granulocyte count and infection
[7 & 8] Excellent UpToDate articles summarising the approach to febrile neutropenia for low-risk and high-risk patients
[11] Recent European update of the microbiology of febrile neutropenia – meta analysis from the ECIL
[13] The How-Long Study
[14] ECIL-10 Guidelines on Antibiotic Therapy for Febrile Neutropenia. Highly recommend reading this, with more useful information in its appendix.
Also Cited Above:
[1] An overview of the microbiology of febrile neutropenia (slightly outdated)
[3] IDSA Guidelines 2011 for Febrile Neutropenia
[4] Australian Guidelines for Febrile Neutropenia
[5] RCH Guidelines for Paediatric Febrile Neutropenia
[6] 1979 study in the American Journal of Medicine evaluating continuation of KGC through granulocytopaenia
[9] Survey on practices across cancer centres in the USA
[10] Commentary article that questions the role of routine fluoroquinolone prophylaxis for patients with prolonged durations of neutropenia
[12] International Pediatric Fever and Neutropenia Guideline Panel 2023 Update
[15] Outdated ECIL-4 guidelines
[16] Survey of EBMT Centres regarding antibiotic practices
[17] Recent 2022 study evaluating very short course carbapenem therapy in high-risk neutropenic PUO
[18] Meta-analysis of early de-escalation of antibiotics for febrile neutropenia demonstrating mortality benefit
Discover more from Myths & Maladies
Subscribe to get the latest posts sent to your email.
