Many randomized trials have used overall survival as the primary endpoint for establishing non-inferiority of one treatment compared to another. However, if a treatment is non-inferior to another treatment in terms of overall survival, clinicians may be interested in further exploring which treatment results in better health utility scores for patients. Examining health utility in a secondary analysis is feasible, however, since health utility is not the primary endpoint, it is usually not considered in the sample size calculation, hence the power to detect a difference of health utility is not guaranteed. Furthermore, often the premise of non-inferiority trials is to test the assumption that an intervention provides superior quality of life or toxicity profile without compromising the survival when compared to the existing standard. Based on this consideration, it may be beneficial to consider both survival and utility when designing a trial. There have been methods that can combine survival and quality of life into a single measure, but they either have strong restrictions or lack theoretical frameworks. In this manuscript, we propose a method called HUS (Health Utility adjusted Survival), which can combine survival outcome and longitudinal utility measures for treatment comparison. We propose an innovative statistical framework as well as procedures to conduct power analysis and sample size calculation. By comprehensive simulation studies involving summary statistics from the PET-NECK trial,¹ we demonstrate that our new approach can achieve superior power performance using relatively small sample sizes, and our composite endpoint can be considered as an alternative to overall survival in future clinical trial design and analysis where both survival and health utility are of interest.

中文翻译：

---

健康效用调整生存期：临床试验设计的复合终点

许多随机试验都使用总生存期作为主要终点，以确定一种治疗方法相对于另一种治疗方法的非劣效性。然而，如果一种治疗在总生存率方面不劣于另一种治疗，临床医生可能有兴趣进一步探索哪种治疗可以为患者带来更好的健康效用评分。在二次分析中检查健康效用是可行的，但是，由于健康效用不是主要终点，因此在样本量计算中通常不考虑它，因此无法保证检测健康效用差异的能力。此外，非劣效性试验的前提通常是检验这样的假设：与现有标准相比，干预措施可以提供优越的生活质量或毒性特征，而不影响生存。基于此考虑，在设计试验时同时考虑生存和效用可能是有益的。已经有一些方法可以将生存和生活质量结合到一个单一的衡量标准中，但它们要么有很强的限制，要么缺乏理论框架。在这篇手稿中，我们提出了一种称为 HUS（健康效用调整生存）的方法，它可以结合生存结果和纵向效用测量来进行治疗比较。我们提出了一个创新的统计框架以及进行功效分析和样本量计算的程序。通过涉及 PET-NECK 试验汇总统计数据的综合模拟研究，¹我们证明我们的新方法可以使用相对较小的样本量实现卓越的功效性能，并且我们的复合终点可以被视为未来临床试验设计中总体生存率的替代方案以及对生存和健康效用都感兴趣的分析。