It is in theory useful. Wouldn't bringing some clarity to a complex problem solving field help?
In practice it becomes very complicated. Physicians end up faced with all sorts of variable recommendations. So, if I have a IIb level of recommendation -- should I give that medication or not? Actually -- no one knows. It's a guess.
I'm not convinced that it really solves any problems. And the consequence of the EBM model is that the system has to deal with best practice initiatives that lag the research, and the decisions about what actions to mandate are made by a very few people and the system then enforces those actions in a very arrogant and condescending way, as if doctors who practice differently are backwards and ill informed hacks.
For example -- sepsis has gotten a lot of attention in the medical system over the last 15 years. Back then, one group of docs from one hospital were doing research about massive infusions of fluids through IVs to treat the low blood pressures that came with bad infections. They thought it helped, and eventually their research got enough attention that it was included in the new model of centralized medical protocols being spread throughout the system.
This new model is based on the idea of evidence based medicine and big data, with the idea being that by crunching large amounts of data or summarizing the aggregate of many different studies, you can figure out the single best approach to a problem, with the idea being that everyone should then do it that way. CMS (the medicare system) started to roll out specific treatments for specific medical conditions, including guidelines about how quickly those treatments had to be started. They tracked this data, and if you didn't treat enough people the way they said to, you would be penalized financially.
So the current state of affairs is that anyone with a specific combination of physical exam and/or lab findings to suggest sepsis gets large amounts of IV fluids. But this often causes patients to go into congestive heart failure. And meanwhile, no one has been able to replicate those initial studies that suggested large amounts of fluids were a good idea, and the current state of research suggests that it doesn't help at all. Plus, the criteria for sepsis decided upon by CMS wasn't the criteria for sepsis used in the initial Detroit studies.
The consequence of all of this is that the system spends a lot of time and money doing something that may, in all likelihood, cause more harm than good.
As far as I'm concerned, the system is grappling with the fact that there is no ideal way to treat a problem in a complex, self adjusting system. There are a few "fixes" that are no brainers -- I don't see anyone volunteering to be part of a double blind placebo controlled experiment on the efficacy of parachutes, for example.
The system didn't need big data to help it figure out that insulin was a good idea for diabetics. Or that IV fluids helps if you are dehydrated. Or that reducing joint dislocations is a good idea. But once you have picked that low hanging fruit, what remains isn't unclear because we aren't crunching large enough data sets, but because the very nature of it is too variable to be precise.
EBM is an attempt to provide the illusion of clarity for something that is by nature complex and unclear, and then to force the conclusion of the politicized health officials onto the rest of the system as if it is clear. I'm not sure that is a good thing.