Adjusting for partially-observed utilities and costs in trial-based cost-effectiveness analysis: a comparison of different methods and their performance