In healthcare priority setting, the concept of severity often plays an important role, to voice the aspect of need. It has been observed that severity is an under-theorized concept. The recent pandemic raised a number of practical but also theoretical issues; one such aspect is how to handle severity in heterogeneous patient groups, e.g. when we have to vaccinate a large group, but only a fraction of them is at the risk of suffering severe disease. The aim of this article is to explore, what we call such heterogeneous severity. In the article, we explore three approaches, the trump, the average and the additive approach and find that all of these alternatives are wanting given the rationale for taking severity into account. Instead, we examine a double threshold and a stepwise trump approach, and find that they both have a better match with the rationale for taking severity into account when having to consider budget constraints. However, neither of these will provide input to acceptable cost-effectiveness threshold, and therefore will have to be combined with a weighted additive approach. Comparing the double threshold and stepwise trump approach, we find the latter advantageous in avoiding problematic threshold effects.