Design: Systematic review and meta-analysis of randomised trials
Population: 9 trials, 602 patients with brucellosis
Primary outcome: Treatment failure, relapse, adverse events and time to defervescence
Effect: Failure RR 2.11 (1.26–3.51); relapse RR 1.88 (0.94–3.74); defervescence 1.20 days longer

A systematic review and meta-analysis in PLOS Neglected Tropical Diseases, published on 23 October, pooled nine randomised trials with 602 patients comparing fluoroquinolone-based regimens with standard doxycycline-based combinations, such as doxycycline with rifampicin or streptomycin.

Fluoroquinolone regimens had about twice the risk of treatment failure, with no heterogeneity across trials, and took about 1.2 days longer to clear fever. Relapse was numerically higher but not significantly so, and adverse events did not differ significantly. Certainty was moderate for failure and low for relapse and harms.

Brucellosis is common in India among people who handle livestock or drink unpasteurised milk, and fluoroquinolones are sometimes reached for because they are familiar. This analysis supports sticking with doxycycline-based combinations as first-line treatment for adults, in line with long-standing practice.

Fluoroquinolones may still have a role when standard drugs cannot be used, but they should be a second choice.

Use a doxycycline-based combination as first-line treatment for adult brucellosis.
Reserve fluoroquinolone-based regimens for when standard drugs are contraindicated or unavailable.
Ask about livestock contact and raw milk in prolonged fever of unclear cause.
Follow patients for relapse after treatment, whichever regimen is used.

Why it matters

It gives a pooled answer to a common improvisation in brucellosis treatment.

The statistics, in plain English

A risk ratio of 2.11 (95% CI 1.26 to 3.51) for failure means fluoroquinolone regimens failed about twice as often, and the whole interval sits above 1. Relapse had an RR of 1.88 (0.94 to 3.74), which crosses 1, so the trial data cannot confirm a difference there.