We describe a framework for the analysis of heterogeneous treatment effects (HTE). Our framework recognizes three types of HTE analyses: confirmatory, exploratory and descriptive. HTE commonly pertains to the variation in treatment effect on the efficacy endpoints. We denote the variation in treatment effect on adverse events as heterogeneous safety effects (HSE). We discuss the potential usefulness of descriptive HTE and HSE analyses for the efficacy and safety outcomes, respectively. We discuss the methodological challenges of studying HTE/HSE in non-experimental settings, with a particular emphasis on post-marketing surveillance (PMS). Finally, we propose some new ideas for future work in HSE analysis in PMS studies such as considering trial data in conjunction with PMS and the use of consistency of subgroup effects as a validation device for pharmacovigilance.