Frontiers of pain part 2: Within the pareto frontier
Frontiers of pain part 2: Finding the best of the best
When we want to figure out whether one intervention is more cost-effective than another, without knowing the true value of the intensity-weighting, we can sweep the weighting across its entire range and see how the two interventions compare for these values.
In my previous post (link), I discussed how the Cumulative Pain Framework (CPF) can be combined with Multi-Objective Optimisation (MOO) to help us narrow down a list of interventions to find the pareto frontier: the set of interventions which are more cost-effective than the others. However, that system does not help us in deciding between two interventions which are both in the pareto frontier, i.e. it can help us find the set of best interventions, but not the best of the best, which is what we’re looking for.
Here is the basic idea: we have two interventions, A and B, and we know how many hours of each pain-intensity they are expected to reduce and how much they are expected to cost, and we want to know which is more cost effective. We can’t calculate their relative cost-effectiveness directly, as this depends on the total expected pain reduction, which in turn depends on the pain-intensity-weighting. Instead, we will let the weighting range across all possible values and see the resulting range of relative cost-effectiveness. So we will end up with a statement something like: “A is between 80% as effective, and 200% more cost effective than B”. Along with some heuristics, we can use this to build up some confidence about which intervention is most cost-effective.
The maths of how this works is discussed in Appendix 1, but for now I will go straight into how this can be used in practice
Here are the steps to follow when given a list of interventions and the aim is to choose the one which is most cost-effective.
