Characteristics of included studies.

Background
Most US children with acute otitis media [AOM] receive prompt antibiotic treatment, though guidelines encourage watchful waiting. Previous systematic reviews of antibiotics versus watchful waiting have focused on symptom resolution and RCTs, limiting the assessment of serious, rare complications. We sought to evaluate these complications by including observational studies.
Methods
RCTs and observational studies that compared antibiotics to placebo or watchful waiting for pediatric clinician diagnosed AOM were identified [PubMed/MEDLINE, Embase, Cochrane Database of Systematic Reviews, Central Register of Controlled Trials, and Web of Science] and reviewed for meta-analysis. Two reviewers independently extracted study characteristics, patient characteristics, and outcomes. We assessed publication bias, study bias with ROBINS-1 and RoB-2 and used random-effects models to assess treatment effects.
Results
24 studies were included. Antibiotics decreased the risk of acute mastoiditis [incidence 0.02%, RR 0.48, 95% CI 0.40–0.59; NNT 5,368]. This protective effect may be underestimated because of misclassification of non-suppurative conditions as AOM. Intracranial complications remained too rare to assess. Antibiotics markedly increased the risk of adverse effects [incidence 10.5%, RR 1.49, 1.27–1.73; NNH 23]. Studies used non-specific criteria for acute mastoiditis, potentially underestimating treatment effects.
Conclusions
Prompt antibiotic therapy reduces the risk for some AOM complications. The NNT to prevent serious, rare complications is high, while the NNH is relatively low. Large-scale population-based observational studies using real-world datasets with validated measures of severe complications are needed to improve understanding of risk factors for serious AOM complications, facilitate more selective antibiotic therapy, and optimize individual outcomes and public health.

背景
大多数美国急性中耳炎（acute otitis media, AOM）患儿会接受即时抗生素治疗，尽管临床指南推荐观察等待策略。既往针对抗生素与观察等待对比的系统综述多聚焦于症状缓解与随机对照试验（randomized controlled trial, RCT），限制了对严重、罕见并发症的评估。本研究纳入观察性研究，旨在评估此类并发症。

方法
本研究检索了PubMed/MEDLINE、Embase、Cochrane系统综述数据库、对照试验中央注册库及Web of Science，筛选对比抗生素与安慰剂或观察等待方案治疗儿科临床确诊AOM的随机对照试验及观察性研究，用于荟萃分析。由2名研究者独立提取研究特征、患者特征及结局指标。我们采用ROBINS-1与RoB-2工具评估发表偏倚及研究偏倚，并使用随机效应模型评估治疗效应。

结果
最终纳入24项研究。抗生素可降低急性乳突炎的发病风险（发生率0.02%，相对风险RR=0.48，95%置信区间CI：0.40~0.59；需治疗人数NNT=5368）。由于将非化脓性疾病误分类为AOM，该保护效应可能被低估。颅内并发症仍过于罕见，无法进行评估。抗生素显著增加不良反应风险（发生率10.5%，RR=1.49，95%CI：1.27~1.73；需伤害人数NNH=23）。现有研究对急性乳突炎采用非特异性诊断标准，可能低估了治疗效应。

结论
即时抗生素治疗可降低部分AOM并发症的发病风险。预防严重、罕见并发症的需治疗人数较高，而需伤害人数相对较低。未来需采用经过验证的严重并发症评估指标、基于真实世界数据的大规模人群观察性研究，以加深对AOM严重并发症危险因素的认识，助力更具选择性的抗生素治疗策略，并优化个体结局与公共卫生管理。