Informatics approaches to assessing patient frailty in surgical care

摘要 手术并发症是常见的，昂贵的，致命的。老年患者的不良手术风险高 结果，特别是当他们表现出脆弱。虚弱是一种生理储备减少和丧失的状态。 适应压力的能力。在过去的十年里，虽然脆弱越来越被认为是一种 作为手术效果差的重要风险因素，将标准化虚弱度量整合到临床护理中， 没有实现。一个关键的障碍是，现有的脆弱性评估不是标准化的，客观的，或广泛的 这限制了它们在外科决策中的常规应用。长期目标是改善 对于老年人的外科护理，我们将评估两个“电子脆弱”指标，可以自动从 已作为常规临床护理的一部分收集的电子或数字数据。这些电子脆弱指标 包括：第一，电子健康记录（EHR）数据粒度患者简档（基于索赔数据的风险评分 或生理和实验室值），第二，从术前计算的肌肉损失评估， 断层扫描（CT）扫描（低骨骼肌质量，称为肌肉减少症，和脂肪浸润到肌肉中 表明身体功能降低，称为肌肉脂肪变性）。在目标1中，我们将计算这两个e-脆弱性 超过41，000名腹部手术患者的不同人群中的指标;描述重叠 通过两个e-虚弱指标指定为虚弱的患者之间的差异;并评估其与30天 再入院和其他不良手术结局（30天和1年死亡率、并发症、非家庭 出院，住院时间>7天）。在目标2中，我们将比较电子脆弱性指标的性能， 标准风险分层工具预测30天再入院和其他不良手术结局 （急性和慢性疾病严重程度指标）已经嵌入到今天的EHR使用交叉验证和 超过14，000例近期腹部手术的独立验证数据集。在目标3中，我们将研究 e-虚弱指标是否会改变患者从实现术后目标中获得的受益-包括 早期和持续的动员-在一个最大的增强术后恢复（ERAS）计划， 归国我们将研究电子脆弱指标作为生物年龄的显着指标，用于预测发病率， mortality.总之，电子脆弱指标显示出在外科领域识别高风险患者的巨大希望， 但它们需要整合到临床工作流程中，以实现可扩展性和可持续性。这项建议会 计算从EHR数据自动导出的标准化电子脆弱性指标，并提供新信息 关于这些电子脆弱指标对改善老年人外科护理的潜在价值。本研究 还将为未来的前瞻性干预奠定基础，将电子脆弱指标纳入临床护理， 改善考虑手术患者的风险分层和咨询，并加强围手术期护理， 虚弱的外科病人

ABSTRACT Surgical complications are common, costly, and deadly. Older patients are at high risk of adverse surgical outcomes, especially when they exhibit frailty. Frailty is a state of decreased physiologic reserve and loss of capacity to adapt to stressors. Over the past decade, while frailty has been increasingly recognized as an important risk factor for poor surgical outcomes, integration of a standardized frailty metric into clinical care has not been achieved. A key barrier is that existing frailty assessments are not standardized, objective, or widely available, limiting their routine application in surgical decision-making. With the long-term goal of improving surgical care for older adults, we will evaluate two “e-frailty” metrics that can be automatically derived from electronic or digital data that are already collected as part of routine clinical care. These e-frailty metrics include, first, granular patient profiles of electronic health record (EHR) data (risk scores based on claims data or on physiologic and laboratory values), and second, muscle loss assessed from pre-surgical computed tomography (CT) scans (low skeletal muscle mass, known as sarcopenia, and fatty infiltration into muscle indicative of reduced physical function, known as myosteatosis). In Aim 1, we will calculate these two e-frailty metrics among a diverse population of over 41,000 abdominal surgical patients; characterize the overlap between patients designated as frail by the two e-frailty metrics; and evaluate their associations with 30-day readmission and other adverse surgical outcomes (30-day and 1-year mortality, complications, non-home discharge, and length of stay >7 days). In Aim 2, we will compare the performance of e-frailty metrics for predicting 30-day readmission and other adverse surgical outcomes to that of standard risk stratification tools (acute and chronic illness severity metrics) already embedded in EHRs today using cross-validation and an independent validation dataset of over 14,000 more recent abdominal surgeries. In Aim 3, we will examine whether e-frailty metrics modify the benefits that patients derive from achieving postoperative targets -including early and sustained mobilization- in one of the largest Enhanced Recovery After Surgery (ERAS) programs in the nation. We will examine e-frailty metrics as salient indicators of biologic age for predicting morbidity and mortality. In sum, e-frailty metrics show great promise for identifying high-risk patients in the surgical domain, but they need to be integrated within clinical workflows to be scalable and sustainable. This proposal will compute standardized e-frailty metrics automatically derived from EHR data and provide new information regarding the potential value of these e-frailty metrics for improving surgical care for older adults. This study will also lay the groundwork for future prospective interventions integrating e-frailty metrics into clinical care to improve risk stratification and counseling of patients considering surgery and enhance perioperative care for frail surgical patients.