How Long

Published

Aguilar-Guisado M, et al. "Optimisation of empirical antimicrobial therapy in patients with haematological malignancies and febrile neutropenia (How Long study): an open-label, randomised, controlled phase 4 trial". Lancet Haematology. 2017. 4(12):e573-e583.
PubMed • Full text

Clinical Question

Among patients admitted to hematology wards with high-risk neutropenic fever without microbial diagnosis, does stopping empiric antibiotics once the patient has been clinically stable & afebrile for 72 hours increase the number of antibiotic-free days compared to continuing antibiotics until neutrophil recovery?

Bottom Line

Among patients with high-risk neutropenic fever without microbial diagnosis who are admitted to hematology wards, cessation of empiric antibiotics at 72 hours of clinical stability & resolution of fever led to fewer days of total empiric antibiotic use without increase in adverse events when compared to antibiotic cessation driven by end of neutrophilia.

Major Points

Management of febrile neutropenia with antibiotics has historically involved use of empiric antimicrobial therapy (EAT) until resolution of fever and neutropenia. This practice was not supported by high-quality evidence. Among patients with febrile neutropenia, neutropenia may persist for weeks or months, and continuing broad-spectrum antibiotics for such a long duration may be associated with more harm than benefit.

Published in 2017, the How Long trial sought to compare shorter or longer durations of empiric antimicrobial therapy among patients on inpatient hematology wards with febrile neutropenia. The shorter duration experimental arm stopped antibiotics 72 hours after clinical improvement (resolution of fever, clinical stability, normalization of vital signs), while the longer control arm stopped antibiotics at resolution of neutropenia. Use of antibiotics followed local practices and clinical guidance of treating physicians. The shorter intervention arm had more EAT-free days than the longer control arm (13 vs. 16 days). There was no apparent increase in mortality or infections during the follow-up period; rather, there might have been fewer serious adverse events with the shorter course of antibiotics. However, the study was not powered to detect statistically significant changes in patient-important outcomes such as mortality. The population only included 4 deaths in 157 patients, which is lower than other observational studies of patients with febrile neutropenia. This study provides some support for shorter courses of empiric antibiotics in neutropenic fever.

Guidelines

As of March 2021, no guidelines have been published that reflect the results of this trial.

Design

• Open-label, multicenter, randomised controlled trial
• N=157 patients
  • Experimental group (stop at 72 hours of clinical stability) (n=78)
  • Control group (continue until neutrophil recovery) (n=79)
• Setting: 6 hospitals in Spain
• Enrollment: 2012-2016
• Follow-up: 28 days for the primary outcome
• Analysis: intention-to-treat
• Primary outcome: # of empirical antimicrobial therapy-free days

Population

Inclusion Criteria

• Aged ≥18 years on inpatient hematology service
• ≥1 of the following:
  • Acute leukemia
  • Lymphoproliferative disease
  • Multiple myeloma
  • Myelodysplastic syndrome
  • Bone marrow aplasia
  • Autologous or allogeneic hematopoietic stem cell transplantation
• High-risk febrile neutropenia, defined by:
  • Expected duration of neutropenia ≤0.5x10⁹/L of ≥7 days

Exclusion Criteria

• Fever with other clinical explanation, such as:
  • Known infection with microbiological diagnosis
  • Fever from non-infectious etiology:
    • Transfusion-related
    • Drug fever
    • GVHD
    • Other underlying disease
• Epilepsy
• Fluoroquinolone-related tendinopathy
• Pregnancy or breastfeeding
• HIV infection
• Renal dysfunction with creatinine clearance less than 30 mL/min
• Patients receiving CYP3A5-acting medication

Baseline Characteristics

From the experimental group:

• Demographics: 52y, 54% female
• Hematologic disease
  • Acute leukemia: 51% (39% in control)
    • Treatment: Induction/reinduction: 31%, other chemotherapy 10%, autologous HSCT 4%, allogeneic HSCT 6%
  • Lymphoma: 29% (37% in control)
    • Treatment: Chemotherapy 6%, autologous HSCT 22%, allogeneic HSCT 2%
  • Chronic lymphocytic leukemia: 3%
    • Treatment: Chemotherapy 3%
  • Multiple myeloma: 9% (18% in control)
    • Treatment: Chemotherapy 0%, autologous HSCT 8%, allogeneic HSCT 1%
  • Myelodysplastic syndrome: 3%
    • Treatment: Allogeneic HSCT 3%
  • Other diagnosis: 5%
    • Treatment: Chemotherapy 1%, autologous HSCT 4%
  • By treatment:
    • Chemotherapy or immunosuppression: 50% (39% in control)
    • Autologous HSCT: 37% (54% in control)
    • Allogeneic HSCT: 12% (6% in control)

Interventions

• Patients were randomized to a group:
  • Experimental - Antibiotics stopped at 72 hours from resolution of fever, signs and symptoms of infection, and normalization of vital signs
  • Control - Antibiotics stopped when the neutrophil count reached ≥0.5×10⁹ cells/L
• Antibiotics were selected based upon local protocols, but options included:
  • Cefepime + ceftazidime 2g TID
  • Meropenem 1 g TID
  • Imipenem 500 mg QID
  • Pip-tazo 4 g TID
  • Amikacin 20 mg/kg daily
  • Ciprofloxacin 400 mg BID
  • Levofloxacin 500 mg BID
  • Vancomycin per usual dosing and targeting
  • Teicoplanin 6 mg/kg BID x3 doses then daily
  • Aztreonam 2 g TID
• Persistent fever workup followed predefined guidelines
• Modification to antibiotic treatment was driven by primary clinical management

Outcomes

Presented as experimental (stop at 72h of clinical stability) vs. control (stop at neutrophil recovery)

Primary Outcome

# of empirical antimicrobial therapy-free days

The authors present this intention-to-treat analysis. They also present per-protocol and modified per-protocol population analyses. The overall associations were similar for each endpoint, as is presented in Table 3 on pg e578 of the publication.

16.1 vs. 13.6 days (absolute difference -2.4 days; 95% CI -4.6 to -0.3; P=0.026)

Secondary Outcomes

Safety events

Crude mortality: 1 death vs. 3 deaths (P=0.62)

Days of fever: 5.7 days vs. 6.3 days (P=0.53)

Infections by end of follow-up

Bacterial infections: 39% vs. 46%

Viral infections: 17% vs. 11%

Fungal infections: 22% vs. 37%

Additional Outcomes

Median neutropenia

14 vs. 11 days (P=0.13)

Neutropenia at the end of empiric antimicrobial therapy

53% vs. 10% (P<0.0001)

Adverse Events

Any, as rate

159 vs. 138/1000 patient-days (P=0.057)

Serious: 5 vs. 13/1000 patient-days (P=0.0087)

Serious adverse events, as proportion

Statistical comparison not provided, see additional details in Table 5 on pg e580.

Any: 14% vs. 34%

Non-infectious adverse events: 5% vs 14%

Infectious adverse events: 9% vs. 20%

Criticisms

• A single center recruited 83% of the patients, possibly limiting its external validity.
• Not blinded.
• Unclear if EAT-free days is a patient-centered, meaningful outcome.
• Not powered to assess mortality differences.

Funding

Publicly funded by the Spanish Ministry of Economy, Industry and Competitiveness, with support from other Spanish and European public institutions.