OBJECTIVES: Cost-effectiveness studies of health technologies where data are extrapolated from short term clinical trials to estimate results over appropriate decision making time horizons typically assume that the decision context remains constant.  More recently, it has been argued that a ‘life cycle’ approach should be routinely applied to technology assessments, including incorporating anticipated changes in price.  Since the results from these different approaches can have important implications for reimbursement and conditional coverage decisions, it is critical that rationale and methods for incorporating future changes are more clearly defined.  The aims of this research are to: i) establish the key principles and assessments related to incorporating future changes when extrapolating cost-effectiveness results; ii) consider how additional information and analyses could be used to more appropriately inform a range of policy decisions.  METHODS: We reviewed existing literature investigating future changes in the decision context due to price, technological developments and the information available to decision makers to examine their relevance and implications for cost-effectiveness methods and approaches to extrapolation.  A case study evaluating the cost-effectiveness of biologics for the treatment of psoriatic arthritis is used to demonstrate the impact of modelling a changing context on both the magnitude and the uncertainty around cost-effectiveness results.  RESULTS: Our results show that incorporating anticipated changes in the decision context, particularly where decisions made are irreversible, can have a dramatic effect on the extrapolation of cost-effectiveness results both in terms of determining the optimal treatment decision and in terms of characterising the level and consequences of uncertainty around these results.  CONCLUSIONS: A changing decision context can have a significant impact on cost-effectiveness results and the conclusions drawn from these results.  This changing decision context so far as it can be anticipated should therefore be modelled to allow us to make the best possible decisions incorporating all available information.