Xie Shitong, Hong Tianqi, Geng Jialu, Luo Chang, Fang Haoran, Wu Jing
School of Pharmaceutical Science and Technology, Tianjin University, Tianjin, China.
Center for Social Science Survey and Data, Tianjin University, Tianjin, China.
Appl Health Econ Health Policy. 2025 Aug 18. doi: 10.1007/s40258-025-00992-7.
Despite an increasing number of mapping studies being conducted in China, there is an absence of a systematic reviews, which makes it difficult to inform the applications and further assess the methodological consistency, accuracy, and applicability of existing mapping studies. The objective of this review is to consolidate existing evidence, identify methodological gaps, and provide recommendations for improving mapping studies conducted among the Chinese population.
A systematic literature search was conducted in 14 databases from inception to May 31, 2025 to identify studies that developed mapping algorithms to estimate health utility values, specifically among Chinese populations. A data template was applied to extract dataset information, source and target measures, mapping types (direct vs indirect), models used, goodness-of-fit indicators, validation methods, and the optimal mapping algorithms selected. Potential challenges for future related studies were further discussed.
A total of 33 studies was included. Most studies (87.9%) focused on mapping disease-specific non-preference-based measures (PBMs) to generic PBMs. The studies covered a broad range of disease areas, including oncology (36.4%), musculoskeletal disorders (15.2%), metabolic diseases (15.2%), cardiovascular diseases (9.1%), and neurological conditions (6.1%). All studies used direct mapping, with the ordinary least squares model (n = 37) being used most frequently, followed by Tobit model (n = 32) and Beta model (n = 22). Eleven studies explored indirect mapping, with the Ordered Logit and Ordered Probit models being the most employed techniques. Thirty-two studies conducted internal validation, with the N-fold cross-validation being the most used method-no study conducted external validation. The sample size ranged from 133 to 3320, with a median sample size of 553. Conducted conceptual analysis was performed in 81.8% of the studies to assess the degree of overlap between the source measure and target measure; 72.7% of the studies reported the utility/score distributions, and 15.2% of studies further reported the response distributions.
This systematic review provides insights into methodologies employed in mapping studies in China and identifies key areas for improvement. Addressing issues related to sample size, conceptual overlap, model selection, and validation methods will enhance the quality and applicability of mapping algorithms, ultimately supporting more robust cost-utility analyses in the Chinese healthcare system.
尽管中国开展的映射研究数量不断增加，但缺乏系统评价，这使得难以了解现有映射研究的应用情况，并进一步评估其方法的一致性、准确性和适用性。本综述的目的是整合现有证据，识别方法学上的差距，并为改进针对中国人群开展的映射研究提供建议。
从数据库建库至2025年5月31日，在14个数据库中进行了系统的文献检索，以识别开发映射算法来估计健康效用值的研究，特别是针对中国人群的研究。应用数据模板提取数据集信息、源指标和目标指标、映射类型（直接映射与间接映射）、使用的模型、拟合优度指标、验证方法以及选择的最佳映射算法。进一步讨论了未来相关研究可能面临的挑战。
共纳入33项研究。大多数研究（87.9%）专注于将特定疾病的非基于偏好的指标（PBMs）映射为通用PBMs。这些研究涵盖了广泛的疾病领域，包括肿瘤学（36.4%）、肌肉骨骼疾病（15.2%）、代谢疾病（15.2%）、心血管疾病（9.1%）和神经系统疾病（6.1%）。所有研究均采用直接映射，其中普通最小二乘法模型（n = 37）使用最为频繁，其次是 Tobit 模型（n = 32）和 Beta 模型（n = 22）。11项研究探索了间接映射，有序 Logit 模型和有序 Probit 模型是最常用的技术。32项研究进行了内部验证，N 折交叉验证是最常用的方法，没有研究进行外部验证。样本量从133到3320不等，中位数样本量为553。81.8%的研究进行了概念分析，以评估源指标和目标指标之间的重叠程度；72.7%的研究报告了效用/分数分布，15.2%的研究进一步报告了反应分布。
本系统综述深入了解了中国映射研究中采用的方法，并确定了关键的改进领域。解决与样本量、概念重叠、模型选择和验证方法相关的问题将提高映射算法的质量和适用性，最终支持中国医疗保健系统中更可靠的成本效用分析。
