NEJM 14 Jun 2007 Vol 356

This meta-analysis of rosiglitazone and cardiovascular events appeared on the NEJM website some weeks ago, and few readers with an interest in diabetes can have failed to read something or other about it, or even the paper itself. I think most people will find it extraordinarily difficult to analyze. The additional risk falls just within the limits of statistical significance and the comparison table with other treatments is deeply confusing. The Chief Medical Officer of GlaxoSmithKline has taken the opportunity of an early rebuttal in the correspondence pages of The Lancet (p.1996), where he is also referred to as the King of Prussia. The King of GSK cites an unpublished observational study he has sponsored which apparently shows no real-life difference in outcomes between diabetics whose regimes did or didn’t include rosiglitazone. Among the BMJ letters (p.1236), a GP from Wales raises once again the question of why diabetologists are so fixated on glycaemic control, which is of little prognostic value, and won’t simply come out and say that metformin improves outcomes and other drugs don’t.

Now here’s an interesting question: if you have more medical conditions, does the quality of your care get better or worse? It seems that in American cities, the answer is – surprisingly – better. But I can’t be bothered to speculate as to whether the methods of this study are robust or whether you can extrapolate such findings to anywhere else.