Adjusting Estimates of the Expected Value of Information for Implementation: Theoretical Framework and Practical...

Area of Science:

• Health economics
• Decision analysis
• Health services research

Background:

• Value of information (VoI) calculations, including expected value of sample information (EVSI), typically assume perfect and instant implementation of new treatment recommendations.
• This assumption is often unrealistic in healthcare, where implementation is typically improved rather than perfect.

Purpose of the Study:

• To present a novel method for calculating the expected value of further research that incorporates the reality of improved implementation.
• Introduce a framework to quantify the impact of varying implementation levels on research value.

Main Methods:

• Extended an existing conceptual framework to include additional states of information and implementation.
• Developed a method to calculate implementation-adjusted EVSI (IA-EVSI) that accounts for different degrees of implementation.
• Illustrated calculations using a stylized case study in non-small cell lung cancer.

Main Results:

• In a case study, the expected value of sample information (EVSI) was £25 million, while the implementation-adjusted EVSI (IA-EVSI) was £8 million, indicating a significant overestimation with the traditional method.
• The IA-EVSI was influenced by the time horizon and the rate of implementation change; a higher rate led to a greater IA-EVSI.
• Decisions based on perfect implementation assumptions overestimated research value by approximately £17 million in the case study.

Conclusions:

• Traditional VoI measures rely on unrealistic assumptions about implementation, leading to potentially inflated estimates of research value.
• The proposed framework provides a more realistic assessment of the expected value of research by accounting for improved implementation.
• This approach offers a more accurate basis for healthcare decision-making regarding research investments.