以严重程度为锚定的卒中后吞咽困难临床试验患者报告结局框架

文献信息

中文摘要

背景： 卒中后吞咽困难常见，并与吸入性肺炎、营养不良、住院时间延长及生活质量下降相关。评估卒中后吞咽困难干预措施的试验在选取既以患者为中心又方法学稳健的终点方面持续面临挑战。
目的： 推导并论证一种用于卒中相关吞咽困难的患者报告结局（PRO）终点，该终点支持具有临床意义的解读、统计学效率以及与监管期望的一致性。
方法： 我们综合了监管指南、结局研究标准以及卒中特异性证据，开发了一种在卒中后90天时基于状态的二分类PRO应答者终点，该终点整合了吞咽特异性生活质量（SWAL-QOL）和整体健康相关生活质量（EuroQol-5D；EQ-5D）。应答者定义根据基线吞咽困难严重程度进行校准，使用在急性期后早期评估的吞咽相关美国国立卫生研究院卒中量表（NIHSS）子成分。蒙特卡洛模拟（10,000例模拟卒中后吞咽困难患者）在临床现实的严重程度分布和治疗效应下，比较了联合应答者终点与（i）连续SWAL-QOL和（ii）效用加权改良Rankin量表（UW-mRS）分析的性能。估计了个体随机化和阶梯楔形设计的样本量需求。
结果： 在异质性急性卒中人群中，以严重程度为锚定的应答者终点比连续或基于变化的分析更具统计学效率且更易解读。在模拟中，联合SWAL-QOL+EQ-5D应答率从42.7%（对照组）增至66.2%（治疗组），绝对差异为23.5%（比值比2.58，95% CI 2.38-2.81；需治疗人数≈4-5），在中度和重度吞咽困难中效应最大。相比之下，平均SWAL-QOL改善8.3分（接近或低于常引用的最小临床重要差异阈值），平均UW-mRS改善0.017，表明整体残疾测量指标的敏感性有限。
结论： 一种结合SWAL-QOL和EQ-5D的、经严重程度校准的二分类PRO应答者终点，为评估卒中后吞咽困难的治疗干预提供了一个以患者为中心、统计学高效且方法学稳健的框架。

英文摘要

BACKGROUND: Post-stroke dysphagia is common and is associated with aspiration pneumonia, malnutrition, longer hospitalization, and reduced quality of life. Trials evaluating interventions for post-stroke dysphagia face persistent challenges in selecting endpoints that are both patient-centered and methodologically robust.
OBJECTIVES: To derive and justify a patient-reported outcome (PRO) endpoint for stroke-related dysphagia that supports clinically meaningful interpretation, statistical efficiency, and alignment with regulatory expectations.
METHODS: We synthesized regulatory guidance, outcomes-research standards, and stroke-specific evidence to develop a dichotomous, status-based PRO responder endpoint at 90 days post-stroke that integrates swallowing-specific quality of life (SWAL-QOL) and global health-related quality of life (EuroQol-5D; EQ-5D). Responder definitions were calibrated to baseline dysphagia severity using swallowing-relevant National Institutes of Health Stroke Scale (NIHSS) subcomponents assessed in the early post-acute period. Monte Carlo simulations (10,000 simulated patients with post-stroke dysphagia) compared performance of the combined responder endpoint versus (i) continuous SWAL-QOL and (ii) utility-weighted modified Rankin Scale (UW-mRS) analyses under clinically realistic severity distributions and treatment effects. Sample size requirements were estimated for individually randomized and stepped-wedge designs.
RESULTS: Severity-anchored responder endpoints were more statistically efficient and more interpretable than continuous or change-based analyses in heterogeneous acute stroke populations. In simulation, combined SWAL-QOL + EQ-5D responder rates increased from 42.7% (control) to 66.2% (treatment), an absolute difference of 23.5% (odds ratio 2.58, 95% CI 2.38-2.81; number needed to treat ≈ 4-5), with the largest effects in moderate and severe dysphagia. In contrast, mean SWAL-QOL improved by 8.3 points (near or below commonly cited minimal clinically important difference thresholds), and mean UW-mRS improved by 0.017, indicating limited sensitivity of global disability measures.
CONCLUSION: A severity-calibrated, dichotomous PRO responder endpoint combining SWAL-QOL and EQ-5D provides a patient-centered, statistically efficient, and methodologically robust framework for evaluating therapeutic interventions in post-stroke dysphagia.