New perspective on parametric confidence intervals for the cost-effectiveness ratio