A June 2009 Health Affairs paper by Ari Hoffman and Steven Pearson includes this typology of marginal medicine:
(1) Inadequate evidence of comparative net benefit for any indication;  (2) use beyond boundaries of established net benefit. […] (3) higher cost when established benefit is comparable to other options; and (4) relatively high cost for incremental benefit compared to other options.
They provide examples of each of the four types:
- Inadequate evidence of net benefit: Proton beam therapy.
- Use beyond boundaries of established net benefit: Off-label drug use, “Vagus nerve stimulation for depression (approved for seizures) and chelation therapy for autism (approved for heavy metal toxicity).”
- Higher cost with no increase in benefit: Brand drugs when generics exist, “One example has been demonstrated by a study using decision-analysis modeling to compare four different strategies for diagnosing deep vein thrombosis. The four strategies involve differing combinations of tests; all four strategies were found to be equally effective. But the results showed that the approach beginning with clinical evaluation and blood tests worked just as well as all of the options that started off immediately with some form of expensive high-tech imaging.”
- High cost, low incremental benefit: Chemotherapy for advanced cancer, expensive antifungal agents for onychomycosis (toenail fungus), “The drug clopidogrel (Plavix), for example, helps prevent myocardial infarction and stroke, which are clearly of great clinical significance to individual patients; however, compared to patients taking aspirin, only for approximately one in every 200 patients taking clopidogrel will a myocardial infarction or stroke be prevented; all other patients taking the drugwould not have had a stroke or infarction anyway.It has been estimated that giving all eligible patients clopidogrel instead of aspirinwould cost an extra $34 billion per year.”
Related: Typology of health care technology.