What Is a Valid Mapping Algorithm in Cost-Utility Analyses? A Response From a Missing Data Perspective

ObjectivesAlthough numerous mapping algorithms from a non–preference-based measure to a target health utility measure have been developed and applied in cost-utility analyses (CUAs), conditions for a mapping algorithm to work well in a CUA are still unclear. In this research, we formulate the mapping problem as a missing data problem and clarify these conditions.MethodsWe defined a valid mapping algorithm based on the purpose of mapping (ie, not for prediction but for CUA), and derived a sufficient set of conditions for a valid mapping algorithm. We also conducted a simulation study to investigate properties of a mapping algorithm under situations where the conditions are satisfied and violated.ResultsThe derived sufficient conditions indicate that the complete overlap of the source measure with the target health utility measure is important and that a covariate that is omitted from a mapping algorithm but has an effect on the target health utility measure not captured by the source measure may invalidate a mapping algorithm. The conditions cannot be verified from data in a CUA but can be supported using external data. A simulation study showed that when at least 1 of the 3 conditions was violated, a mapping algorithm provided biased health utility estimates in a CUA, and that prediction accuracy did not necessarily reflect performance of a mapping algorithm in a CUA.ConclusionThe derived conditions provide a fundamental basis for better practices in developing and selecting a mapping algorithm.     

心血管疾病患者健康相关生命质量及其影响因素的多水平模型分析

目的 探讨心血管疾病患者的健康相关生命质量及其影响因素,为改善患者生命质量提供参考。 方法 选取2015年中国健康与养老追踪调查(CHARLS)中符合要求的心血管疾病患者1 201例,收集患者及地区信息,基于欧洲五维健康量表(EQ-5D-3L)测量健康效用值反映其健康相关生命质量。采用K-W检验对患者和地区因素进行单因素分析。以患者水平为水平1,地区水平为水平2,建立两水平回归模型,分析患者和地区因素对心血管疾病患者健康相关生命质量的影响。 结果 健康效用值平均值为0.749±0.166。单因素分析结果显示,心血管疾病患者健康相关生命质量的影响因素包括性别、受教育水平、婚姻状况、是否患有高血压、是否患有心脏病、是否患有中风、睡眠时长、吸烟状况、饮酒状况、每周中等强度体力活动时间、社交活动频率、体质量指数(BMI)、所在地区(P<0.05)。多水平模型结果显示,对心血管疾病患者健康相关生命质量有显著影响的因素有性别、受教育水平、婚姻状况、是否患有高血压病、是否患有血脂异常、是否患有心脏病、是否患有中风、吸烟状况、每周中等强度体力活动时间、社交活动频率、BMI以及地区水平的地区因素(P<0.05)。 结论 提高心血管疾病患者的健康相关生命质量既要考虑受教育水平、吸烟状况、BMI等个体因素,还应考虑不同地区患者的异质性,从而制定不同地区的针对性策略。     

Cost–Utility Analysis of Pembrolizumab Versus Chemotherapy as First-Line Treatment for Metastatic Non-Small Cell Lung Cancer With Different PD-L1 Expression Levels

To evaluate the cost–utility of pembrolizumab versus chemotherapy as the first-line setting for metastatic nonsmall cell lung cancer (NSCLC) from the US health care system perspective, a Markov model was developed to compare the lifetime cost and effectiveness of pembrolizumab versus chemotherapy for untreated metastatic NSCLC, based on the clinical data derived from phase III randomized controlled trial (KEYNOTE- 042; ClinicalTrials.gov; NCT02220894). Weibull distribution was fitted to simulate the parametric survival functions. Drug costs were collected from official websites, and utility values were obtained from published literature. Total costs, quality-adjusted life years (QALYs), and incremental cost-effectiveness ratios (ICERs) were computed as primary output indicators. The impact of different PD-L1 expression levels on ICER was also evaluated. One-way and probabilistic sensitivity analyses were performed to assess the model uncertainty. Compared with chemotherapy, patients treated with pembrolizumab provided an additional 1.13, 1.01, and 0.59 QALYs in patients with PD-L1 expression levels of ≥50%, ≥20%, and ≥1%, with corresponding incremental cost of $53,784, $47,479, and $39,827, respectively. The resultant ICERs of pembrolizumab versus chemotherapy were $47,596, $47,184, and $68,061/QALY, in three expression levels of PD-L1, respectively, all of which did not exceed the WTP threshold of 180,000/QALY. Probability sensitivity analysis outcome supported that pembrolizumab exhibited evident advantage over chemotherapy to be cost-effective. One-way sensitivity analysis found that ICERs were most sensitive to utility value of pembrolizumab in progression survival state. All the adjustment of parameters did not qualitatively change the result. For treatment-naive, metastatic NSCLC patients with PD-L1+, pembrolizumab was estimated to be cost-effective compared with chemotherapy for all PD-L1 expression levels at a WTP threshold of $180,000/QALY in the context of the US health care system.     

Innovative Thinking on Endpoint Selection in Clinical Trials

ABSTRACT

In clinical trials, selection of appropriate study endpoints is critical for an accurate and reliable evaluation of safety and effectiveness of a test treatment under investigation. In practice, however, there are usually multiple endpoints available for measurement of disease status and/or therapeutic effect of the test treatment under study. For example, in cancer clinical trials, overall survival, response rate, and/or time to disease progression are usually considered as primary clinical endpoints for evaluation of safety and effectiveness of the test treatment under investigation. Once the study endpoints have been selected, sample size required for achieving a desired power is then determined. It, however, should be noted that different study endpoints may result in different sample sizes. In practice, it is usually not clear which study endpoint can best inform the disease status and measure the treatment effect. Moreover, different study endpoints may not translate one another although they may be highly correlated one another. In this article, we intend to develop an innovative endpoint namely therapeutic index based on a utility function to combine and utilize information collected from all study endpoints. Statistical properties and performances of the proposed therapeutic index are evaluated theoretically. A numerical example concerning a cancer clinical trial is given to illustrate the use of the proposed therapeutic index.     

Statistical inference for decision curve analysis,with applications to cataract diagnosis

Statistical learning methods are widely used in medical literature for the purpose of diagnosis or prediction. Conventional accuracy assessment via sensitivity, specificity, and ROC curves does not fully account for clinical utility of a specific model. Decision curve analysis (DCA) becomes a novel complement as it incorporates a clinical judgment of the relative value of benefits (treating a true positive case) and harms (treating a false positive case) associated with prediction models. The preference of a patient or a policy-maker is formulated statistically as the underlying threshold probability, above which the patient would choose to be treated. Net benefit is then calculated for possible threshold probability, which places benefits and harms on the same scale. We consider the inference problems for DCA in this paper. Interval estimation procedure and inference methodology are provided after we derive the relevant asymptotic properties. Our formulation can accommodate the classification problems with multiple categories. We carry out numerical studies to assess the performance of the proposed methods. An eye disease dataset is analyzed to illustrate our proposals.