Is the Use of Unanchored Matching-Adjusted Indirect Comparison Always Superior to Naïve Indirect Comparison on Survival Outcomes? A Simulation Study.

IF 3.1 4区医学 Q1 ECONOMICS

Applied Health Economics and Health Policy Pub Date : 2025-02-23 DOI:10.1007/s40258-025-00952-1

Ying Liu, Xiaoning He, Jia Liu, Jing Wu

{"title":"Is the Use of Unanchored Matching-Adjusted Indirect Comparison Always Superior to Naïve Indirect Comparison on Survival Outcomes? A Simulation Study.","authors":"Ying Liu, Xiaoning He, Jia Liu, Jing Wu","doi":"10.1007/s40258-025-00952-1","DOIUrl":null,"url":null,"abstract":"Objective: To compare the performance of matching-adjusted indirect comparison (MAIC) and naïve indirect comparison (NIC) under a wide range of data scenarios on survival outcome.Methods: A simulation study included 729 (36) single-arm trial data scenarios, which were created by performing a three-level full factorial arrangement of six situational variables, including individual patient data (IPD) sample size, aggregate data (AgD) sample size, covariate strength, covariate correlation, covariate overlap, and relative treatment effect. In each scenario, 1000 repetitions of simulated datasets were generated using the Monte Carlo approach. MAIC and NIC methods were used to estimate the relative treatment effect of each simulated dataset. The performance was evaluated in terms of bias, empirical standard error (ESE), mean squared error (MSE), and confidence interval coverage, respectively.Results: MAIC yielded relatively unbiased estimates of relative treatment effect compared with NIC in most scenarios, with better coverage and MSE but higher ESE. None of the situational variables had a significant impact on the bias and coverage of MAIC. However, increasing IPD sample size and covariate overlap significantly reduced the ESE and MSE of MAIC. In scenarios with low covariate overlap and high covariate strength, the bias of MAIC was larger and even greater than that of NIC.Conclusions: The performance of MAIC consistently demonstrates advantage over NIC across various scenarios. MAIC often provides more unbiased estimates and achieves confidence interval coverage close to nominal values compared with NIC. While MAIC may exhibit higher ESE in specific scenarios, this additional uncertainty can offer a more accurate reflection of variability, enhancing the robustness of the results. Researchers should thoroughly comprehend the influencing factors and interactions affecting the performance of these methods and judiciously apply research findings.","PeriodicalId":8065,"journal":{"name":"Applied Health Economics and Health Policy","volume":" ","pages":""},"PeriodicalIF":3.1000,"publicationDate":"2025-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Applied Health Economics and Health Policy","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s40258-025-00952-1","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ECONOMICS","Score":null,"Total":0}

引用次数: 0

Abstract

Objective: To compare the performance of matching-adjusted indirect comparison (MAIC) and naïve indirect comparison (NIC) under a wide range of data scenarios on survival outcome.

Methods: A simulation study included 729 (3⁶) single-arm trial data scenarios, which were created by performing a three-level full factorial arrangement of six situational variables, including individual patient data (IPD) sample size, aggregate data (AgD) sample size, covariate strength, covariate correlation, covariate overlap, and relative treatment effect. In each scenario, 1000 repetitions of simulated datasets were generated using the Monte Carlo approach. MAIC and NIC methods were used to estimate the relative treatment effect of each simulated dataset. The performance was evaluated in terms of bias, empirical standard error (ESE), mean squared error (MSE), and confidence interval coverage, respectively.

Results: MAIC yielded relatively unbiased estimates of relative treatment effect compared with NIC in most scenarios, with better coverage and MSE but higher ESE. None of the situational variables had a significant impact on the bias and coverage of MAIC. However, increasing IPD sample size and covariate overlap significantly reduced the ESE and MSE of MAIC. In scenarios with low covariate overlap and high covariate strength, the bias of MAIC was larger and even greater than that of NIC.

Conclusions: The performance of MAIC consistently demonstrates advantage over NIC across various scenarios. MAIC often provides more unbiased estimates and achieves confidence interval coverage close to nominal values compared with NIC. While MAIC may exhibit higher ESE in specific scenarios, this additional uncertainty can offer a more accurate reflection of variability, enhancing the robustness of the results. Researchers should thoroughly comprehend the influencing factors and interactions affecting the performance of these methods and judiciously apply research findings.

查看原文本刊更多论文

目的比较匹配调整间接比较法（MAIC）和天真间接比较法（NIC）在多种数据情景下对生存结果的影响：模拟研究包括 729 个（36 个）单臂试验数据情景，这些情景是通过对六个情景变量（包括单个患者数据（IPD）样本量、总体数据（AgD）样本量、协变量强度、协变量相关性、协变量重叠和相对治疗效果）进行三级全因子排列而创建的。在每种情况下，使用蒙特卡罗方法生成 1000 次重复的模拟数据集。使用 MAIC 和 NIC 方法估算每个模拟数据集的相对治疗效果。分别从偏差、经验标准误差（ESE）、均方误差（MSE）和置信区间覆盖率等方面对其性能进行了评估：在大多数情况下，与 NIC 相比，MAIC 对相对治疗效果的估计相对无偏，覆盖率和 MSE 更高，但 ESE 更高。没有一个情景变量对 MAIC 的偏差和覆盖率有显著影响。然而，增加 IPD 样本规模和共变量重叠度会显著降低 MAIC 的 ESE 和 MSE。在协变量重叠度低、协变量强度高的情况下，MAIC 的偏差更大，甚至超过了 NIC：结论：在各种情况下，MAIC 的性能始终优于 NIC。与 NIC 相比，MAIC 通常能提供更无偏的估计值，并实现接近名义值的置信区间覆盖率。虽然 MAIC 在特定情况下可能表现出更高的 ESE，但这种额外的不确定性可以更准确地反映变异性，从而增强结果的稳健性。研究人员应全面了解影响这些方法性能的影响因素和相互作用，并明智地应用研究成果。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Applied Health Economics and Health Policy Economics, Econometrics and Finance-Economics and Econometrics

CiteScore

6.10

自引率

2.80%

发文量

期刊介绍： Applied Health Economics and Health Policy provides timely publication of cutting-edge research and expert opinion from this increasingly important field, making it a vital resource for payers, providers and researchers alike. The journal includes high quality economic research and reviews of all aspects of healthcare from various perspectives and countries, designed to communicate the latest applied information in health economics and health policy. While emphasis is placed on information with practical applications, a strong basis of underlying scientific rigor is maintained.