Monday, August 13, 2007

NHS Scale research testing

Ben Goldacre appears to be running a series on how pharmaceutical corporations contribute to the high cost of health care in the Guardian: Evil ways of the drug companies Spectacularly expensive cost of trial and error

In this post I will discuss the second article where he makes an interesting argument for whole scale medical testing in the NHS to reduce medical costs:

Imagine there are two drugs called Sixofone and Halfadozen. Nobody knows which one is better for treating sickitis. The drug companies periodically do trials, but funnily their own drug always seems to come out the best.

In our clinic, the doctor sees 200 patients in a month with sickitis. If she gives Sixofone or Halfadozen to her patients at random, on a whim, as a reflection of her own indecisiveness, then the regulation is the same as for any prescription: give reasonable verbal information about risks and benefits, thus obtaining informed consent, and so on.

But let's say the same doctor, in a spirit of inquiry, recognises that there is massive clinical uncertainty - nobody knows which of these drugs is better after all - and she wants to test her hunches, rather than simply act on them. Then she's in trouble.

She could run a speculative little trial, giving 100 patients Sixofone, and compare them with the others on Halfadozen. But she'd be struck off, unless she went through so much red tape that the idea becomes unmanageable.

After passing through the endless obstructive and bureaucratic hoops of the ethics committees, from each and every patient she would need to obtain elaborate written consent, and give huge amounts of information. Remember, nobody knows which is best, and if she wanted to hand over the same drugs, entirely at whim, outside of her "trial", there'd be no problem. This is a huge double standard, and it costs us knowledge.

But if we ever had a scientist in charge of health, instead of tinkering with payments to big pharma, they would do one simple thing: move hell and high water to collect and collate the best and cheapest evidence on healthcare. First you would give huge amounts of money to the Cochrane Collaboration, which collects and collates data independently on all healthcare interventions, and is quietly one of the most subversive organisations ever to be created, because it blows the lid on false commercial claims.

Doctors waste money by irrational prescribing decisions, so give them clean, clear, accessible information, and that will stop.

But more than that, we are in a unique position to generate data: we have inherited an extraordinary information resource, in this vast monolith of the NHS.

Nobody knows what the best treatment for stroke is, but if we randomised every single new stroke patient in the UK, over one week, into a rough trial, we'd have our answers in a couple of years.

Part of the target of his article is research ethics committees, who he claims are in effect complicit in a huge double standard, namely that doctors can prescribe according to their whims in practice, but not in research... Of course in practice this only applies when we have two reasonably equal treatments. I'm happy to agree there are sometimes biases built into the system of review, for example there is a huge bias towards the existing standard of treatment, regardless of how weak the evidence base is for that treatment. But I think this issue here is a non-starter. If we have two treatments for a condition which appear to be equally valuable, then almost certainly through out the country physicians will already be prescribing both of these. Why go through the unnecessary cost and hassle of a clinical trial, when we could get similar results, just by consulting already existing patients notes. Research ethics committee permission would still need to be obtained of course, but this is much less of a hassle than constructing a clinical trial.

There is of course also a presumption built into the article, namely that there is such a thing as a best treatment for a specific condition. But it is unclear that this is the case, it is likely instead that things are more complex than this, and what will suit one patient will not suit others. At best we can establish something less grand, namely the medicine most likely to be best.

No comments: