Dasbach Erik J, Elbasha Elamin H
Merck & Co. Inc., Kenilworth, NJ, USA.
Merck Center for Observational and Real-World Evidence, Merck Research Laboratories, Merck & Co., Inc., UG1C-60, PO Box 1000, North Wales, PA, 19454-1099, USA.
Pharmacoeconomics. 2017 Jul;35(7):673-683. doi: 10.1007/s40273-017-0508-2.
Decision-analytic models for cost-effectiveness analysis are developed in a variety of software packages where the accuracy of the computer code is seldom verified. Although modeling guidelines recommend using state-of-the-art quality assurance and control methods for software engineering to verify models, the fields of pharmacoeconomics and health technology assessment (HTA) have yet to establish and adopt guidance on how to verify health and economic models. The objective of this paper is to introduce to our field the variety of methods the software engineering field uses to verify that software performs as expected. We identify how many of these methods can be incorporated in the development process of decision-analytic models in order to reduce errors and increase transparency. Given the breadth of methods used in software engineering, we recommend a more in-depth initiative to be undertaken (e.g., by an ISPOR-SMDM Task Force) to define the best practices for model verification in our field and to accelerate adoption. Establishing a general guidance for verifying models will benefit the pharmacoeconomics and HTA communities by increasing accuracy of computer programming, transparency, accessibility, sharing, understandability, and trust of models.
用于成本效益分析的决策分析模型是在各种软件包中开发的，而计算机代码的准确性很少得到验证。尽管建模指南建议使用软件工程的最新质量保证和控制方法来验证模型，但药物经济学和卫生技术评估（HTA）领域尚未建立并采用关于如何验证健康和经济模型的指南。本文的目的是向我们的领域介绍软件工程领域用于验证软件是否按预期运行的各种方法。我们确定这些方法中有多少可以纳入决策分析模型的开发过程中，以减少错误并提高透明度。鉴于软件工程中使用的方法范围广泛，我们建议开展更深入的举措（例如，由ISPOR-SMDM特别工作组）来定义我们领域中模型验证的最佳实践，并加速采用。建立模型验证的一般指南将通过提高计算机编程的准确性、透明度、可及性、共享性、可理解性和模型的可信度，使药物经济学和HTA社区受益。
