Improving Heart Failure Risk Stratification in the ED

描述（由申请人提供）：急诊科（ED）医生面临的一个关键挑战是如何最好地管理出现心力衰竭症状的患者。目前，大多数被评估为心力衰竭的患者都住进了医院，但并不是所有这些患者都需要这样的强化治疗，高达50%的患者是可以避免的。提高急诊医师有效、安全地管理低风险患者的能力对于避免不必要的住院治疗至关重要。我们建议开发一种决策工具，该工具来源于前瞻性收集的ED数据，用于预测住院或门诊患者死亡风险以及严重的院内或院外并发症。此外，拟议的项目将验证该决策工具在三种不同的ED环境中跨种族和社会经济不同的患者群体的有用性和普遍性。为了开发我们的决策工具，急诊医生在急诊室就诊的前两个小时内通常可以获得的100多个变量将被考虑纳入统计风险模型。与现有的使用住院病人数据的模型不同，这些措施代表了实际的临床实践，通常用于决定病人的处置。我们将在患者心力衰竭评估期间收集标准化数据。依赖图表审查或大型数据集分析可能导致数据丢失和不一致。我们将纳入所有评估为心力衰竭的患者，而不考虑最终诊断，从而避免基于确诊患者的模型固有的选择偏差。我们提出的一个基本创新是使用5天结果进行初级分析，30天结果进行次级分析的工具。这克服了30天结果模型的局限性，该模型高度依赖于不可预测的访问后患者和提供者行为。该项目的另一个新颖方面是结合急诊医学、心脏病学和生物统计学的专业知识，准确地将治疗后的结果分配给急性表现。结果将转化为一种算法，并在全世界传播。这是实现合理分配医院资源以降低心力衰竭护理成本这一广泛目标的第一步。在与结果和有效性研究人员的合作下，我们计划进行进一步的研究，以测试我们的风险模型的有效性。

DESCRIPTION (provided by applicant): A critical challenge facing emergency department (ED) physicians is how best to manage patients presenting with symptoms of heart failure. Currently, most patients being evaluated for heart failure are admitted to the hospital, yet not all of these patients warrant such intensive treatment, and up to 50% of these admissions could be avoided. Improving the ability of the emergency physician to effectively and safely manage low-risk patients is essential to avoid unnecessary hospitalizations. We propose developing a decision tool derived from prospectively gathered ED data that will predict risk for inpatient or outpatient death and serious in-hospital or out-of-hospital complications. Further, the proposed project will validate the usefulness and generalizability of this decision tool in three different ED environments across racially and socioeconomically diverse patient populations. To develop our decision tool, over 100 variables routinely available to the emergency physician within the first two hours of ED presentation will be considered for inclusion in a statistical risk model. Unlike exisitng models using inpatient data, these measures are representative of actual clinical practice and routinely used to decide a patient's disposition. We will collect standardized data during a patient's evaluation for heart failure. Relying on chart review or large dataset analyses can lead to missing and inconsistent data. We will include all patients evaluated for heart failure regardless of final diagnosis, thus avoiding selection bias inherent in models based on patients with a definitive diagnosis. A fundamental innovation we propose is a tool using 5-day outcomes for primary analyses, and 30-day outcomes for secondary analyses. This overcomes the limitation of 30-day outcome models that are highly dependent on unpredictable, post-visit patient and provider behavior. Another novel aspect of the proposed project is the combining of expertise in emergency medicine, cardiology, and biostatistics to accurately assign post-treatment outcomes to acute presentations. Results will be translated into an algorithm that will be disseminated worldwide. This is the first step toward achieving our broad objective of appropriate allocation of hospital resources to reduce costs of heart failure care. In collaboration with outcomes and effectiveness researchers, we plan to conduct further studies to test the efficacy of our risk model.