Design: Systematic review and meta-analysis of proportions, stratified by design
Population: 15 studies, 5,440 adults with DLBCL on first-line R-CHOP or similar
Primary outcome: Incidence of febrile neutropenia
Effect: Observational cohorts 19.0% (95% CI 17.1% to 21.1%); trials 11.7% (5.7% to 22.7%)

Guidelines recommend primary G-CSF prophylaxis when a regimen's febrile neutropenia risk is about 20% or more, or 10% to 20% with patient risk factors. Estimates for R-CHOP have largely come from trials. This meta-analysis pooled 15 studies of adults with diffuse large B-cell lymphoma on first-line R-CHOP or similar regimens (5,440 patients, 885 febrile neutropenia events), analysed separately by design.

Across ten observational cohorts (4,020 patients), febrile neutropenia occurred in 19.0% (95% CI 17.1% to 21.1%), with modest heterogeneity and a prediction interval of 15.3% to 23.3%. Five trials gave 11.7% (5.7% to 22.7%), with high heterogeneity. The difference between designs narrowly missed significance and depended heavily on one trial.

In routine practice, then, R-CHOP sits at the 20% threshold rather than below it. Patients with DLBCL are typically older and often have additional risk factors, so for most of them, the case for primary G-CSF prophylaxis is stronger than trial figures suggest.

Consider primary G-CSF prophylaxis for most adults with DLBCL starting R-CHOP.
Assess risk factors such as age over 65, poor performance status, comorbidity and low albumin.
Do not rely on trial-reported febrile neutropenia rates when estimating routine-care risk.
Ensure patients know how to act on fever from the first cycle.

Why it matters

Trial figures may have led clinicians to underestimate how often R-CHOP causes febrile neutropenia in practice.

The statistics, in plain English

A prediction interval of 15.3% to 23.3% describes where the rate in a new routine-care setting is likely to fall, and it sits close to the 20% threshold. The trial estimate is lower but much less consistent between trials.