How should missing cost data be handled in economic evaluations?

• A. Use imputation or sensitivity analyses to assess impact of missing values.
• B. Exclude observations with missing data.
• C. Substitute missing costs with the mean of observed values.
• D. Ignore missing data if the amount is small.

Answer: (A)

Handling missing cost data in economic evaluations requires approaches that preserve information and quantify uncertainty. Imputation methods estimate the missing costs based on observed data patterns, and using multiple imputation adds and then combines several plausible values to reflect uncertainty about what the true costs would be. Pairing imputation with sensitivity analyses tests how results change under different reasonable assumptions about the missing values, showing whether conclusions are robust or sensitive to the gaps in data. This combination helps maintain sample size, reduce bias, and transparently assess the impact of missing data on conclusions.

Excluding observations with missing data risks biasing results if the missingness is related to the costs or outcomes, and it reduces precision. Substituting missing costs with the mean ignores uncertainty and can distort the distribution of costs and the estimated effects. Ignoring missing data altogether, even when the amount seems small, can still lead to biased or unstable conclusions if the missingness is not completely random.