{"id":1170,"date":"2015-10-09T20:49:49","date_gmt":"2015-10-09T19:49:49","guid":{"rendered":"https:\/\/blogs.bmj.com\/adc\/?p=1170"},"modified":"2015-10-05T14:34:40","modified_gmt":"2015-10-05T13:34:40","slug":"getting-from-the-diamond-to-the-drug-chart","status":"publish","type":"post","link":"https:\/\/blogs.bmj.com\/adc\/2015\/10\/09\/getting-from-the-diamond-to-the-drug-chart\/","title":{"rendered":"Getting from the diamond to the drug chart"},"content":{"rendered":"

\"\"“Come out of the dark and into the light – use systematic reviews in your clinical practice.” (OK, so that’s probably NOT a direct quote.)<\/p>\n

It’s been battered on about for a while in this blog that systematic reviews<\/a> give you – probably – the best idea of if a treatment, diagnostic test or prognostic factor\u00a0is real\/effective\/truthful.<\/p>\n

But it’s also true that the same, wonderful, complex and sometimes frankly unreadable pieces of research<\/a> don’t tell you how to take that Truth and get the pharmacy to dispense it<\/a>, or the physio team to do it, or the lab to tell you the answer.<\/p>\n

The answer might be here<\/a>, beautifully written by\u00a0Paul Glasziou<\/a>, an enormously wise practical theorist on evidence-based practice, and colleagues.<\/p>\n

<\/p>\n

They suggest that there are three conceptual approaches to taking the results of a well conducted systematic review and turning it into action. The first would be to pick a method used in one of the included trials, which has all the practicality of being what someone has actually done, but loses some of the groupiness of the synthetic component of an SR. It’s quick tho’, and eminently practical for a street user.<\/p>\n

The second is to break apart the components of an intervention, and see which bits are common, and are probably the key to why ‘it works’, then use that reconstituted hybrid\u00a0in practice. That assumes that the common elements are the effective ones – which may be wrong – but it’s not a daft idea.<\/p>\n

The third would be to use a proper theoretical understanding to propose how the intervention works, and then use the different studies and their elements to test if that’s right (using meta-regression<\/a> and that sort of thing). This takes bucket loads of data, good quality descriptions of the interventions and the outcomes, a uniformity of bias and a tonne of time. But might be the best way of doing it.<\/p>\n

To get a deeper understanding, DO go and read the full paper<\/a>. Have a think, and work out why you might pick different ways at different times. Comment, tweet or Facebook us in response. We’re here to help.<\/p>\n