Such statements like "measure x prevented/caused n deaths" should generally be considered highly dubious unless applied to a very narrow context (such as a rand

Such statements like "measure x prevented/caused n deaths" should generally be considered highly dubious unless applied to a very narrow context (such as a randomized double blind placebo control study, etc.). There is nothing that can really prevent death; everyone will die sooner or later. So speaking of "preventing/causing death" is an abbreviated way of speaking. It makes most sense when it is about "immediate death", but the longer it takes for a cause to develop its presumed effect, the less the effect can be solely attributed to this particular cause. It then makes no real sense to just count the numbers of deaths related to a particular cause, if it is only one of an actually unquantifiable number of contributory causes. But analysing the structure of the statement closer, it becomes even more dubious, because (after applying some small interpretative clarifications) it follows the basic structure: "measure x prevented n of d, because it has an effect e that is (indirectly!)[1] somehow correlated with d". To evaluate the extent to which such a claim could probably be valid would require a considerable amount of background information. Just presenting it as a statement without any substantial discussion is not enlighting at all, but just polarizes the audience: if the statement fits into someone's already established belief system it is readily accepted at truth, and if it contradicts someone's belief system it is instantly rejected. There is no chance to learn anything from it. [1] In the case of hypercapnia, CO2 can directly affect health, but I do not think that was the point of the statement.