When we treat a disease it has a net marginal cost – we will call it A, and assume it is positive, i.e. it costs the health service more than the counter-factual. It also has a benefit, measured perhaps in Quality-Adjusted Life Years (QALYs) or Disability-Adjusted Life Years (DALYs). A cost per DALY can then be calculated. If we know the value of a QALY (£20m in England, for example), then we can calculate the expected benefit in monetary terms. In England, an intervention providing an incremental 2.5 QALYs would represent a monetary value of 2.5 x 20 = £50k – call that B. Then the Expected Net Benefit is B – A; if the intervention has net mean costs of £30k per patient, then the ENB is £20k.
What’s wrong with that? Well nothing, unless the treatment is life-saving. In that case, insofar as the new treatment is effective, some patients who would have died now survive to generate health costs in the future. So strictly speaking the downstream (discounted) cost, C, should be included, so that Expected Net Benefit is B – A – C. This is never done, perhaps because it makes the dismal science just too dismal. In the case of individual clinical treatments alternative uses of the money may involve a treatment that saves a similar proportion of life years so downstream costs can be ‘pruned’. If not, then the treatment that saves more lives gets penalised, which seems wrong even if it is logical within the paradigm. But clinical treatments are one thing, calculating the societal cost of a disease is another. Here it seems particularly wrong to ignore downstream contingent costs unless these are very remote (say over 50 years), in which case discounting takes care of them. It is common, for example, for public health academics to claim that public health interventions, such as reducing salt or sugar in the diet, would save £XX million per year. Yet these calculations exaggerate returns to society because the costs of looking after survivors are ignored in the calculations. Many public health interventions would still be a very good investment even if downstream costs were included. So there is no need to exaggerate the benefit by ignoring downstream costs contingent on earlier deaths avoided.