Trial Troubles
New on ripe-tomato.org
Randomised trials are the best way to test treatments – to protect us from duds, and direct us to those that work. But they can mislead. Randomisation is necessary, but not sufficient. Measure enough endpoints, peek at the data, do enough tests, and enthusiasts can always squeeze out a “positive” result.
To protect ourselves we should agree before we start on the basic design, how many participants, the primary endpoint and statistical tests – it’s called trial registration. All good journals insist on it – or claim to – but sometimes lower their standards for exciting papers. Ripe-tomato.org is checking up.
We will report on unregistered and late registered trials, and on altered primary endpoints, sample sizes or statistical tests. We will note trial funding, and compare the sins of pharma with the public and charitable sectors. Here’s our first, second, and third offenders,
Jim Thornton
More trial troubles
Ripe-tomato.org 