Individualizing Steroid Use in Pneumonia

项目摘要/摘要 肺炎是世界范围内导致死亡的主要传染病。虽然炎症是病理的标志 肺炎的特点是，使用皮质类固醇来钝化严重的炎症反应仍然存在 未定义。几十年来对不同人群的研究未能显示出一致的益处，导致 关于使用皮质类固醇的指导方针相互矛盾。尽管皮质类固醇已被证明对 新冠肺炎肺炎、糖皮质激素在其他感染性肺炎中的应用尚不清楚，具有较高的 临床实践中的可变性。实践中的这种可变性提供了一个机会，可以看到皮质类固醇是如何 肺炎的给药可以根据患者的具体特征进行优化和定制。的目标是 这项研究是为了使先进的统计和机器学习方法适用于已有的稳健方法 来自电子健康记录的观察数据，以识别临床恶化的预测因素并发展 社区获得性肺炎患者使用类固醇的个体化治疗规则。这将是 通过三个具体目标完成：1)开发和验证机器学习预测工具 住院疾病进展，2)开发和测试类固醇使用的个体化治疗规则(ITR)， 3)进行一项单中心可行性临床试验，比较ITR和生物标记物引导的皮质类固醇 社区获得性肺炎患者常规护理的用法和剂量。我们假设：1)a 结合人口统计学、生理参数、临床和实验室数据将准确地 预测住院期间疾病进展的风险并确定受益的类固醇反应型患者 辅助性皮质类固醇治疗大于潜在危害，以及2)ITR和生物标记物引导 皮质类固醇的使用和剂量将是可行的。这个职业发展奖将提供重要的 未来更大规模临床试验的初步数据侧重于优化皮质类固醇在肺炎中的使用，而 培训初级调查员成为独立研究员所需的基本技能。

PROJECT SUMMARY/ABSTRACT Pneumonia is the leading infectious cause of death worldwide. While inflammation is a hallmark pathologic feature of pneumonia, the use of corticosteroids to blunt the profound inflammatory response remains undefined. Decades of studies in heterogeneous populations failed to show a consistent benefit, leading to conflicting guidelines on the use of corticosteroids. Although corticosteroids have been shown to be effective in COVID-19 pneumonia, the use of corticosteroids in other infectious pneumonia remains unclear with high variability in clinical practice. This variability in practice presents an opportunity to see how corticosteroid administration in pneumonia can be optimized and tailored to patient specific characteristics. The objective of this study is to adapt advanced statistical and machine learning methods to already available robust observational data from the electronic health record to identify predictors of clinical deterioration and to develop an individualized treatment rule for steroid use in patients with community acquired pneumonia. This will be accomplished through three specific aims: 1) To develop and validate a machine learning prediction tool for in-hospital disease progression, 2) To develop and test an individualized treatment rule (ITR) for steroid use, and 3) To conduct a single center feasibility clinical trial comparing ITR and biomarker guided corticosteroid use and dosing to usual care in patients with community acquired pneumonia. We hypothesize that: 1) a combination of demographics, physiological parameters, clinical and laboratory data will be accurate in predicting risk of in-hospital disease progression and identifying steroid-responsive patients in whom benefit from adjunct corticosteroid treatment outweighs potential harm, and 2) ITR and biomarker-guided corticosteroid use, and dosing will be feasible. This career development award will provide important preliminary data for future larger clinical trials focused on optimizing corticosteroid use in pneumonia while training a junior investigator in the essential skills needed to become an independent researcher.