Standard error and confidence interval for QALY weights

There are some problems with the standard errors of QALY weights proposed by Groot (2000, Journal of Health Economics 19). The standard errors show smaller values than those of Groot when we recalculate using his method. Moreover, we correct the derivation of his approximation and derive corrected values. Because mean and variance do not exist for a distribution of QALY weights, using standard errors for statistical inference may lead to problems even when an approximation is used. In this paper, we verify the statistical properties of Groot's standard errors by simulation. We find that the corrected standard errors hold the same properties as a normal distribution under specific conditions. In general, however, it would be appropriate to use our simulation method to obtain critical values or p-value.