Doctor and author Ben Goldacre points out how drug trials can be badly designed to exaggerate the beneficial effects of a drug - through unrepresentative patient selection, comparing it against something rubbish, trials that are too short/stop early, are too small, measure uninformative outcomes, bundle their outcomes together in odd ways, ignore drop-outs, change their main outcome after they have finished, and perform dodgy subgroup analysis. This last one is my favourite. In the well known ISIS-2 trial which looked at the benefits of giving aspirin in a heart attack, aspirin was found to improve outcomes. But the researchers decided to do a subgroup analysis just for fun. This showed that while aspirin is effective overall, it does not work in patients born under the star signs Libra and Gemini. As far as I know, we do not withhold aspirin in these patients ...
Nicola Cooper MBChB FAcadMEd FRCPE FRACP is a Consultant Physician and Honorary Clinical Associate Professor based in Derby, UK. She works with a team of educators and students at the Division of Medical Sciences and Graduate Entry Medicine teaching clinical reasoning.