A comprehensive prognostic model for older adults discharged to skilled nursing facilities.

项目摘要/摘要 2019年，估计有20%的住院老年人出院到熟练护理设施(SNF)， 许多人随后经历了潜在的不良后果，包括再次住院，进入长期- 长期护理(LTC)，而不是返回社区，并在6个月内死亡。在这一关键时期 住院后的过渡期，临床医生、患者和家属往往有不一致的期望 关于SNF逗留的轨迹，这可能会导致对护理的不满和对护理的分歧 计划。对于管理这些患者的SNF临床医生来说，围绕急性和慢性的管理做出决策 病情、治疗偏好和提前护理计划因缺乏准确的预后而受阻。 信息。为SNF入院后的各种结果提供个性化的风险评估可能会有所帮助 框定这些重要的讨论，促进共同决策。鉴于没有广泛使用的 对于特别出院的老年人的预后工具，本研究的目标是开发一种 易于使用和节俭的预测模型，联合建模多个结果。20%的样本 社区居住的65岁及以上的医疗保险受益人从医院出院到SNF遗嘱 用于调查两个具体目标：(1)开发和内部验证第一天的预后模型 在SNF入院的第一天使用，提供包括住院再治疗在内的多种结果的风险估计 入院、出院而不再入院，延长SNF停留时间&>100天(建议过渡到 LTC)和6个月死亡率，以及(2)开发更新的模型，该模型使用 来自最小数据集(MDS)评估的详细信息，这些信息可能无法随时获得或广泛使用 由临床医生于入院第1天收集。这些目标的结果将是两个易于使用的节俭 模型可以提供对SNF入院后结果的准确和校准的估计。 意义和创新：拟议研究项目的结果将通过以下方式直接影响临床实践 允许SNF临床医生将特定的患者特征输入到网络计算器中，以获得 多项结果，可以围绕临床管理与患者和家属进行对话 决策和未来规划。这个项目是创新的，因为它将是第一个SNF结果预测 使用最少的变量同时预测多个临床相关结果的模型 易于在临床实践中实施。这项工作的未来方向将涉及更多的调查 高级建模技术，如动态预测和多状态建模，并最终探索 提供这种预后信息如何提高对护理和其他以患者为中心的服务的满意度 一项随机临床试验的结果。

PROJECT SUMMARY/ABSTRACT An estimated 20% of hospitalized older adults were discharged to a skilled nursing facility (SNF) in 2019 with many subsequently experiencing potentially adverse outcomes, including hospital re-admission, entering long- term care (LTC) rather than returning to the community, and death within 6 months. During this critical transition period following a hospitalization, clinicians, patients, and families often have discordant expectations about the trajectory of the SNF stay which can lead to dissatisfaction with care and disagreements over care plans. For SNF clinicians managing these patients, decision making around management of acute and chronic conditions, treatment preferences, and advance care planning is hindered by a lack of accurate prognostic information. Providing individualized risk estimates for a variety of outcomes following SNF admission can help frame these important discussions and facilitate shared decision making. Given that there are no widely used prognostic tools for older adults specifically discharged to a SNF, the objective of this study is to develop an easy-to-use and parsimonious prediction model that jointly models multiple outcomes. A 20% sample of community-dwelling Medicare beneficiaries aged 65 years and older discharged from a hospital to a SNF will be used to investigate two specific aims: (1) develop and internally validate a Day 1 prognostic model to be used on day 1 of SNF admission that provides risk estimates of multiple outcomes including hospital re- admission, discharge home without readmission, prolonged SNF stay >100 days (suggesting transition to LTC), and 6-month mortality and (2) develop an Updated model which provides refined estimates using detailed information from the Minimum Data Set (MDS) assessment that may not be readily available or widely collected by clinicians on day 1 of SNF admission. The result of these aims will be 2 easy-to-use parsimonious models that can provide accurate and well calibrated estimates of outcomes following a SNF admission. Significance and Innovation: Results from the proposed research project will directly inform clinical practice by allowing SNF clinicians to input specific patient characteristics into a web calculator to obtain risk estimates for multiple outcomes that can frame conversations with patients and families around clinical management decisions and future planning. This project is innovative because it will be the first SNF outcome prediction model to predict multiple clinically relevant outcomes simultaneously using a minimal number of variables that can be easily implemented in clinical practice. Future directions of this work will involve investigating more advanced modeling techniques, such as dynamic predictions and multi-state modeling, and ultimately explore how providing this prognostic information can improve satisfaction with care and other patient-centered outcomes in a randomized clinical trial.