Choice of Metrics and Other Considerations for Benefit–Risk Analysis in Subgroups