Purpose: To evaluate alternative approaches taken to estimate the population that could benefit from research and to demonstrate that explicitly modelling future change (technologies, prices and information) leads to more appropriate estimates of the value of information (VoI).

Methods: A review of the VoI literature identified a number of approaches to estimating the population including: considering only the current population; arbitrary time horizons (10, 20 years, infinite) or threshold population values. However, empirically based estimates of the time horizons for decisions in health exist and specifying different time horizons for different types of parameters (eg natural history and effectiveness) remains an option. We explore different approaches in the context of a stylised model of treatment for acute coronary syndrome and demonstrate the impact of uncertainty in time horizon on VoI. We also develop a general approach, which explicitly models future changes in technologies, prices, and information, and demonstrate the impacts on VoI estimates.

Results: The literature implicitly either ignores the value of research for future populations or relies on arbitrary assumptions about finite time horizons, with no analyses to date addressing different time horizons for different types of parameters. Empirically-based estimates of the time horizon may lead to underestimation of the value of information using common assumptions (e.g., &pound486.8m versus &pound158.6m at 10 years). Failing to incorporate uncertainty in the estimates leads to biased estimates of VoI (&pound486.8m deterministic versus &pound384.5m probabilistic). However, all these approaches implicitly use time horizon as a proxy for future changes in technologies, prices and information. We demonstrate that explicitly modelling future changes means that the VoI for the decision problem may increase or fall over time, but the VoI for the group of parameters that can be evaluated by current research tends to decline. We show that finite and infinite time horizons for the decision problem represent special cases (e.g., price shock or no changes respectively). In addition, we show how this type of analysis can be used to inform the timing of research decisions.

Conclusions: The value of information depends on future changes in technologies, prices, and evidence, and finite time horizons for decision problems may not represent an adequate proxy. However, the challenge of modelling future change also has implications for estimates of cost-effectiveness.