Phase-specific comparative effectiveness of local epistaxis interventions: a systematic review and meta-analysis

Local epistaxis studies often conflate failure of initial hemostatic control with bleeding after apparent control. We systematically reviewed randomized and nonrandomized comparative studies of local nonoperative epistaxis interventions. Outcomes were harmonized into acute failure (0–3 h) and post-hemostasis rebleeding windows (> 3 h to 7 days, > 7 days to < 6 months, and ≥ 6 months). Arm-based generalized linear mixed models (GLMMs) provided the primary time-harmonized framework; window-specific network and direct pairwise meta-analyses served as supportive analyses. Fifty-three studies provided 196 study-arm records across 15 intervention-group nodes. In the acute window, non-absorbable packing, topical hemostatic biomaterial, and tranexamic acid had the lowest predicted hemostatic-failure range. In post-hemostasis windows, radiofrequency coagulation, microwave ablation, laser therapy, and electrocautery showed the most stable low-risk rebleeding profile, whereas no treatment and packing-based strategies more often occupied higher-risk ranges. Anchor-time predictions suggested lower risk at 24 h, rising through 72 h and 7 days, remaining higher at 1 and 3 months, and declining by 6 months; radiofrequency coagulation was the clearest exception. GLMMs, window-specific network meta-analysis, and direct pairwise evidence broadly reproduced the main comparative directions, although network certainty was limited. Local epistaxis interventions appeared to follow phase-specific rather than fixed comparative patterns. Treatments linked to favorable immediate control were not necessarily those linked to the lowest later rebleeding risk. Because several later-window comparisons used sparse, mixed-design, or selected-population evidence, the findings should be treated as hypothesis-generating signals for phase-specific comparative trials, not as evidence for immediate practice change. Systematic review registration PROSPERO CRD420251230572.