JAMA 25 Jul 2007 Vol 298

Hypertrophic cardiomyopathy sometimes first presents as sudden death in apparently healthy young people, especially during sport. Over recent years, people (mean age 42) with an antemortem diagnosis of HCM have increasingly been fitted with implantable cardioverter-defibrillators, especially if they are considered at high risk because of a family history of sudden death, massive LV hypertrophy, ventricular tachycardia on monitoring, or unexplained syncope. But do these ICDs save lives, or are they just a shocking waste of money? This big registry study looks at outcomes in 506 patients with HCM and ICDs throughout the developed world. Within 4 years, 20% of patients had an appropriate ICD activation which may have saved their lives – more so in younger people. On the other hand, 27% of patients experienced an inappropriate activation and one died because the device failed to activate. So this technology is useful but far from perfect, as previous reports in older patients have shown.

Speaking of previous reports in older patients, we have read quite a lot about hip protectors over the years, but I hope we will hear less now that it is clear how useless they are. The study involved over a thousand nursing home residents who were supplied with underclothes containing a hip protector sewn into one side. In this way the other hip acted as the control, but the study was terminated at 20 months because there were more fractures on the “protected” side. Oops.

Now and again I take time out and actually read a paper properly, analysing the text and the figures with the help of such primitive statistical knowledge as may from time to time have accidentally lodged in my brain. But there are whole classes of paper which all but a tiny handful of doctors have to take on trust – those involving multiple logistic regression, genetic analysis using complex arrays, and most meta-analyses. Peter Gøtzsche (director of the Nordic Cochrane Centre) is not the kind of chap who takes anything on trust, and for his zeal in spotting inappropriate methodology he deserves the title of Peter Gotcha. Here he gets those who use standardized mean differences wrongly when reporting continuous variable data. OK, Peter, it’s a fair cop. I shall know better in the future.

After getting a bit off track with erythema migrans and cardiac tamponade, the Rational Clinical Examination returns to the high road of primary care with “Does This Child Have Appendicitis?”. This is an example of a red flag or “limited rule-out” diagnostic category – one you must try never to miss. Now we don’t understand the natural history of appendicitis all that well, and it is typical of childhood infective illness in that the clinical features may only become apparent over time; so that if you even think of the possibility you must be sure to see the child a second time, or let the surgeons do the worrying. This is a fairly useful trawl through the studies, though only one out of 42 was from a context that resembled primary care (an A&E department), and it is plagued with vague terms like “clinical gestalt” (why not just say “suspicion”). I will continue to rely on the history, examination and the Mars Bar test: plus, if in doubt, seeing the kid a few hours later.