Adjunctive corticosteroids in non-HIV immunocompromised patients with Pneumocystis jirovecii pneumonia: a systematic review and meta-analysis

Background:

Pneumocystis jirovecii pneumonia (PJP) in non-HIV immunocompromised adults carries 30%–60% mortality. Adjunctive corticosteroids are standard in HIV-PJP but remain contested in non-HIV PJP.

Objectives:

To determine whether adjunctive corticosteroids reduce mortality and invasive mechanical ventilation (IMV) in non-HIV PJP, and whether effect estimates differ between randomized and observational study designs.

Design:

Systematic review and meta-analysis.

Data sources and methods:

Six databases and two trial registers were searched from inception to February 7, 2026. Randomized and observational studies comparing adjunctive corticosteroids versus no corticosteroids or placebo in non-HIV adults with PJP were eligible. Random-effects meta-analyses (restricted maximum likelihood) of two co-primary outcomes included subgroup, sensitivity, E -value, and Grading of Recommendations Assessment analyses.

Results:

Twenty-eight studies were included (one randomized controlled trial (RCT), 27 observational; ∼ 5300 patients): 22 contributed corticosteroid-versus-no-corticosteroid comparisons and six contributed dose-comparison data only. Evidence was discordant by design (interaction p = 0 . 006 ): the RCT showed lower IMV (hazard ratio (HR) 0.36, 95% CI 0.14–0.90) and lower 90-day mortality (HR 0.59, 0.37–0.93); regression-adjusted observational studies suggested harm (odds ratio (OR) 1.56, 1.10–2.22); propensity-based estimates were null-compatible. Pre-specified mortality pools were non-significant short-term (OR 1.22, 0.88–1.70) and intermediate-term (OR 1.39, 0.85–2.29). Secondary infections ( k = 4 ): OR 1.04 (0.63–1.72); RCT 0.59 (0.32–1.06). E -values for RCT effects (3.44 IMV; 2.24 mortality) exceeded the 1.81 E -value of the adjusted observational mortality pool. Certainty was very low for pooled mortality and moderate for RCT outcomes.

Conclusion:

Reliable inference rests on the single RCT rather than the pooled observational estimate, which is discordant by design and most consistent with confounding by indication. In the RCT, corticosteroids did not significantly reduce 28-day mortality but lowered IMV and 90-day mortality and sped respiratory recovery, without clear excess secondary infections (low certainty). Corticosteroids may be considered for severe respiratory failure using the PIC (ClinicalTrials.gov NCT02944045) protocol; adequately powered confirmatory trials are needed.

Trial registration : PROSPERO CRD420261304396.