Friday, 25 January 2008

The proof of the pudding

The Lancet today reports that a "collaborative reanalysis" of epidemiological studies "shows beyond doubt that ovarian cancer can be prevented by the long term use of different generations of oral contraceptives." (Full text free with your Athens password!) In simple terms this means that if you take an oral contraceptive pill you are less likely to get ovarian cancer.

The point that interests me here is the way in which this conclusion was reached - not by doing any trials or studies, but by reading existing studies. The authors looked at epidemiological studies that included over 100 women. They defined what was meant by "ovarian cancer" and "oral contraceptive use". They looked at the figures they had and used them to come to wider conclusions than could be reached from any one study alone. Their findings are expressed using confidence intervals and there is an awful lot of statistics in the paper.

In health research there are levels of evidence. These range from (at the bottom) "expert opinion" to "systematic reviews". A systematic review is the gold standard of evidence. Like the Lancet paper it gathers together a number of studies - Randomised Controlled Trials - selects those that fall within defined limits, and using a "meta analysis" of the data attempts to draw wider conclusions than can be drawn from one trial alone.

If this is confusing, don't despair. There is plenty of help available, starting with a veritable mountain of books. One of the best for a straightforward accessible read, covering all the types of evidence and how to look for them and assess them, is Trisha Greenhalgh's How to Read a Paper, now in its third edition.

There are very specific books: Chalmer's Systematic Reviews, Systematic Reviews in Healthcare by Glasziou, and Systematic Reviews to Support Evidence-Based Medicine from the Royal Society of Medicine Press.

There are general books on understanding research: Studying a Study and Testing a Test, Reading Research, Bandolier's Little Book of Making Sense of the Medical Evidence.

There are endless titles on reading and understanding statistics, and evidence-based practice. Your local NHS library is, of course, the best place to start looking.

Online, CASP has tools to help you appraise a systematic review and randomised controlled trials. The Centre for Evidence-Based Medicine has tools to help you understand the maths and the jargon.

If you want to find systematic reviews the place to look is the Cochrane Library.

As to whether the lowered risk of ovarian cancer outweighs the increased risk of breast, cervical or thyroid this post, at least, I'm not even going there.

(c) creative commons attributed

No comments: