Shayan Muhammad, Lew Daphne, Mancini Michael, Foraker Randi E, Doering Michelle, Mueller Kristen L
Cordell Institute for Policy in Medicine & Law, Washington University in St. Louis, United States.
Division of Biostatistics, Washington University in St. Louis, United States.
Prev Med. 2023 Mar;168:107443. doi: 10.1016/j.ypmed.2023.107443. Epub 2023 Feb 3.
To conduct a systematic review of methodologies, data sources, and best practices for identifying, calculating, and reporting recurrent firearm injury rates in the United States.
In accordance with PRISMA guidelines, we searched seven electronic databases on December 16, 2021, for peer-reviewed articles that calculated recurrent firearm injury in generalizable populations. Two reviewers independently assessed the risk of bias, screened the studies, extracted data, and a third resolved conflicts.
Of the 918 unique articles identified, 14 met our inclusion criteria and reported recurrent firearm injury rates from 1% to 9.5%. We observed heterogeneity in study methodologies, including data sources utilized, identification of subsequent injury, follow-up times, and the types of firearm injuries studied. Data sources ranged from single-site hospital medical records to comprehensive statewide records comprising medical, law enforcement, and social security death index data. Some studies applied machine learning to electronic health records to differentiate subsequent new firearm injuries from the index injury, while others classified all repeat firearm-related hospital admissions after variably defined cut-off times as a new injury. Some studies required a minimum follow-up observation period after the index injury while others did not. Four studies conducted survival analyses, albeit using different methodologies.
Variability in both the data sources and methods used to evaluate and report recurrent firearm injury limits individual study generalizability of individual and societal factors that influence recurrent firearm injury. Our systematic review highlights the need for development, dissemination, and implementation of standard practices for calculating and reporting recurrent firearm injury.
对美国识别、计算和报告复发性枪支伤害率的方法、数据来源和最佳实践进行系统综述。
根据PRISMA指南，我们于2021年12月16日在七个电子数据库中搜索了经同行评审的文章，这些文章计算了可推广人群中的复发性枪支伤害情况。两名评审员独立评估偏倚风险、筛选研究、提取数据，第三名评审员解决冲突。
在识别出的918篇独特文章中，14篇符合我们的纳入标准，报告的复发性枪支伤害率为1%至9.5%。我们观察到研究方法存在异质性，包括所使用的数据来源、后续伤害的识别、随访时间以及所研究的枪支伤害类型。数据来源从单家医院的医疗记录到包含医疗、执法和社会保障死亡指数数据的全州综合记录不等。一些研究将机器学习应用于电子健康记录，以区分后续新的枪支伤害与初次伤害，而另一些研究则将在不同定义的截止时间后所有与枪支相关的重复住院病例归类为新伤害。一些研究要求在初次伤害后有最短随访观察期，而另一些研究则没有。四项研究进行了生存分析，尽管使用了不同的方法。
用于评估和报告复发性枪支伤害的数据来源和方法的变异性限制了个体研究对影响复发性枪支伤害的个体和社会因素的可推广性。我们的系统综述强调了制定、传播和实施计算和报告复发性枪支伤害的标准做法的必要性。
