Here's a quick question for those who advocate a government role in cost effectiveness research: Is it cost effective to treat aspiration pneumonia? The answer, of course, is yes. Or no. Well, yes and no.
If a forty year old trauma patient aspirates immediately after urgent surgery, then yes, it is relatively inexpensive and efficacious to treat with a short course of antibiotics. On the other hand, if the patient is a 74 year old stroke patient with recurrent aspiration, end stage renal disease, and prostate cancer, treating his most recent pulmonary event will prolong the course of dialysis, potentially involve future stays in the ICU on a ventilator, gastrosomy tube, etc., all for the expected benefit of the patient being kept alive so he can die of something more horrible. It is easier to see that treatment might not be cost effective if we consider the patient to have widely metastatic prostate cancer.
If we take this last patient and start removing, one by one, the co-morbidities that make treating aspiration pneumonia such an expensive proposition, at what point do arrive at the objective cut-off that treating him is cost-effective? It is the fact that identifying such a cut-off seems reasonable that is problematic, because it implies that whatever body decides cost-effectiveness will eventually be forced to decide on case-by-case bases. This is the achilles heel of all high minded healthcare system reformers, and all of those overly academic types that worship at the altar of "evidence based medicine." The practice of medicine is founded on the doctor-patient relationship, not the doctor-population relationship or the doctor-health board relationship, or the doctor-utopian social justice fantasist relationship.
It should be axiomatic that a doctor not give a patient medicine to treat the anxiety of a family member. The principle at work is that the provision or withholding of a therapeutic intervention should not be for the benefit of third parties. Similarly, we should not conclude that an intervention is not cost-effective for a particular patient, merely because some egghead has decided that it is not cost effective for a heterogeneous population.
Population studies and statistical models and various and sundry other such are useful for identifying general principles that a competent physician may incorporate into the treatment decision that he makes for his specific patient. They are not reasonable bases upon which to dictate those same decisions from a remote authority for whom that patient is just a data point.