When I said last week that the true scientist rejoices when her/his hypothesis is refuted, I wasn’t trying to restate Popperian orthodoxy but making the point that those who wish to see knowledge expand shouldn’t care whether it does so by proving them right or wrong – the main thing is that they’ve helped in the process. My remark was somewhat informed by the fact that I’ve done work which proved the opposite of what I was hoping to prove. Moreover, I was dabbling in cardiology, where hypotheses are often burdened with tremendous investments of personal conviction. A great chap for sorting these out is Harlan Krumholz at Yale – at the rate of about one major paper a week. If you don’t believe me, look on Medline. Here he and his team did a literature search and found 85 reported genetic risk factors for acute coronary syndrome and then checked them out in 811 patients who had acute coronary events in Kansas City. Only one of them showed even a nominal statistical correlation (p=.03).

Atrial fibrillation is a common hazard in the first 3-4 days after cardiac surgery, but in a nice simple randomised trial, a group of investigators from Finland have discovered a nice simple way to cut the risk by almost half – 100mg of intravenous hydrocortisone. Inject a little steroid when you’ve finished finicky surgery and you’ll stop Finns fibrillating.

Robert Southey, least loved or read of the Lake Poets, spent the years 1810-1819 sitting in a Lakeland cottage writing an immense History of Brazil, a place he never set eyes on. Time prevents me from searching his volumes for any reference to medicine, and indeed I can’t think of any major Brazilian contribution to medical science, though it did provide us with the raw materials for curare, first investigated in the 1930s. But here is a Brazilian proof-of-concept study which may herald a cure for type 1 diabetes. Or not; it is a small case-series of newly diagnosed patients treated with a risky regime of immunosuppression in order to receive back their own bone marrow stem cells. In the following months, most of them no longer needed insulin and showed a drop in HbA1c and a rise in C-peptide, indicating islet-cell recovery. For a long list of caveats, see the accompanying editorial.

Epidemic avian influenza is bad for egg production. When seeking to protect the human population from avian strains of pandemic influenza, it is best to find ways of producing lots of vaccine quickly – and ways that do not depend on eggs. Insect cells containing baculovirus seem to be the answer, if the promise of this small study is confirmed. Say goodbye to H5N1 sleepless nights? Not just yet.