Computerized Adaptive Testing to Direct Delivery of Hospital-Based Rehabilitation

 描述(申请人提供)：超过60%的虚弱、老年和慢性病患者在住院期间丧失功能能力。其结果是不必要地增加了跌倒、近期再次住院和出院到机构而不是家庭环境的风险。我们的 目前的临床护理模式系统性地产生了这些通常可以避免的功能损失，因为它不必要且无效地依赖忙碌的临床医生来检测功能衰退并启动简单而有效的治疗计划。迫切需要通过以下方式重新定义这种情况：1)最大限度地减少对负担过重的临床医生的依赖，以发现、解决和监测患者的康复需求；以及2)立即启动资源适当的个性化护理计划。需要一种非负担的评估策略，能够准确地描述患者不断发展的功能能力和康复需求。然而，目前还没有任何措施提供这一功能。我们建议通过改进现有的急性护理后(AM-PAC)题库的项目反应理论(IRT)建模的活动测量来开发和评估所需的评估工具，以创建与疾病无关的多维计算机化自适应测试(MCAT)。除了提高效率和精确度外，MCAT还可以明确地解决当项目或IRT模型的心理测量学特性在临床相关的亚组之间存在差异时对测量有效性的威胁(如已报道的针对老年人、有强烈症状的人和少数族裔的情况)。这些可实现的努力的好处将是巨大的，因为这种多维方法的使用将从根本上扩大评估范围和效率，而不会以一种可以轻松整合到当前护理实践的方式相应增加临床医生的负担。更具体地说，我们的方法将是丰富目前的AM-PAC银行，并将它们管理到1500名住院患者。这些数据，结合现有的AM-PAC数据集，将被用来评估不同的项目功能、相关特征范围内的项目密度、多维IRT模型拟合以及分数分层的切点，以匹配物理治疗师(PT)分配的功能阶段。所得到的参数估计将用于创建固定长度的简表(SF)和MCAT，以指定将患者分配到链接到经验证的护理计划的功能阶段，以及检测有意义的变化。我们将对600名住院患者和300名代理人实施MCAT和SF，其中50%是口头的，50%是通过片剂的，他们将同时接受经验丰富的患者的评估。结果数据将被用来评估MCAT和SF相对于PT评估的黄金标准的有效性、与临床结果的关联、对代理人(即照顾者与患者)的稳健性、模式影响以及响应性。

 DESCRIPTION (provided by applicant): Over 60% of frail, elderly, and chronically ill patients lose functional capacity during hospitalization. The result is an unnecessarily increased risk of falls, near-term rehospitalizations, and discharges to institutional rather than home settings. Our current clinical care model systemically engenders these often avoidable losses of function due to its unnecessary and ineffective reliance on busy clinicians to detect functional decline and initiate simple, yet effective, treatment plans. There is a critical need to reconceptualize this situation by: 1) minimizing reliance on overburdened clinicians to detect, address, and monitor patients' rehabilitation needs; and 2) immediately initiating resource-appropriate individualized care plans. A non-burdensome assessment strategy capable of accurately profiling patients' evolving functional capabilities and rehabilitation needs is needed. However, no measure currently offers this capability. We propose to develop and evaluate the needed assessment tool by refining the extant item response theory (IRT)-modeled Activity Measure for Post Acute Care (AM-PAC) item banks to create a disease-independent multidimensional computerized adaptive test (MCAT). In addition to enhanced efficiency and precision, MCAT may definitively address threats to measurement validity that arise when an item's or an IRT model's psychometric properties differ across clinically relevant subgroups (as has been reported for the aged, the intensely symptomatic, and ethnic minorities). The benefits of these achievable efforts will be enormous given that the use of this multidimensional approach will radically enhance assessment scope and efficiency without a commensurate increase in clinician burden in a manner that can be easily integrated into current care practices. More specifically, our approach will be to enrich the current AM-PAC banks and administer them to 1500 hospitalized patients. These data, combined with existing AM-PAC data sets, will be used to evaluate differential item functioning, item density across relevant trait ranges, multidimensional IRT model fit, and cut points for score stratification to match physical therapist (PT)-assigned functional stages. The resultant parameter estimates will be used to create a fixed-length short form (SF) and MCAT specified to assign patients to functional stages that link to validated care plans, as well as to detect meaningful change. We will administer the MCAT and SF to 600 hospitalized patients and 300 proxies, 50% verbally and 50% via tablets, who will concurrently be evaluated by experienced PTs. The resultant data will be used to evaluate the MCAT's and SF's validity relative to the gold standard criterion of PT assessment, association with clinical outcomes, robustness to proxy (i.e., caregiver vs. patient) and mode effects, and responsiveness.