The conditional reimbursement policy for expensive medicines in the Netherlands requires real-world data collection on cost-effectiveness within a four years period (T=4) after the initial decision to reimburse a drug (T=0). This introduces new sources of uncertainty, which are less important in an RCT than in real life. This may affect the priorities for further (real-world) research as determined in a VOI analysis. OBJECTIVES: Identifying and modeling types of uncertainty that are usually not parameterized at T=0 but may become relevant at T=4. Include them in the VOI analysis. METHODS: We use a hypothetical model with four states and parameters related to transition and exacerbation probabilities, costs and utilities. Three additional uncertainties were parameterized: persistence, compliance and broadening of indication. Persistence refers to the duration of the treatment and it is determined by the probability of dropping out of the treatment. Compliance is characterized by the fraction of the treatment benefit obtained due to not taking the medication as it was indicated. The impact of indication broadening is modeled as the percentage of the RCT treatment effect realized in the outcome study. These extra parameters were included in the VOI analysis. RESULTS: Priorities change when new uncertainties are introduced in the model. Initially, the EVPPI was highest for transition probabilities followed by utilities; and it was very low for exacerbation probabilities and costs. After new uncertainties are included, compliance and broadening of indication (which is applied only to the new treatment) become as relevant as utilities. Persistence however has little impact in the model. CONCLUSIONS: VOI analysis at T=0 should anticipate and parameterize new types of uncertainty that may emerge during a four year outcomes study. This would help to focus the real-world outcomes study on those parameters that reduce uncertainty in the decision to continue the reimbursement most.