Richard Lehman’s journal review – 17 October 2011

Richard LehmanJAMA   12 Oct 2011  Vol 306
1549    It has been a bad week for vitamin supplements. Worst hit, as usual, has been vitamin E. The SELECT trial began collecting 35 000+ healthy men with normal feeling prostates back in 2001 and randomised them to get a selenium supplement, a vitamin E supplement, both, or placebo. Those given vitamin E had a “significantly increased” risk of prostate cancer –by a factor of 1.6 per thousand person years. Even selenium seems to do a tiny bit more harm than good in this respect. Once again a properly conducted large RCT puts to bed a mass of anecdote, observational “evidence,” bad little studies, and health supplement advertising.

1566   I’ve never quite understood how α-tocopherol managed to get the status of a vitamin with its own capital letter, whereas the truly vital amine folic acid is plonked somewhere with the other B vitamins and often forgotten about. Since the 1970s, we have become keenly aware of its role in early fetal neurodevelopment, especially the prevention of neural tube defects. This prospective observational study from Norway looks at nearly 40,000 children born to mothers who did or did not take folic acid supplements from 4 weeks before conception to 8 weeks after. Severe language delay at 3 years was twice as common in those who did not. The case for folic acid fortification of bread flour grows ever stronger.

1582    But that would be of little use to the one in a hundred mothers who has coeliac disease. I don’t know if anyone has studied neurodevelopmental outcomes in children born to mothers who have subsequently been diagnosed with this common and variable condition. In this clinical review and case discussion, it is claimed that its prevalence has been steadily rising over the last 50 years. This is anything but the case: its estimated prevalence used to be one in 2-10,000; then along came serological testing in the 1990s, and it shot up to 1 in 100 and has stayed there since. It is also described as a disease which shortens life and causes osteoporosis: but again, this only applies to classical grossly symptomatic CD. Taking in the whole spectrum of antibody positive disease, there is little evidence of harm to asymptomatic individuals. But it is still worth testing for in everyone with recurrent anaemia and/or tiredness-all-the-time: some people’s lives are changed by treatment. Others just find it a bore and revert to eating gluten.

1593   Most pharmaceutical and medical device companies conduct human experiments (clinical trials) and then fail to publish the results of some of them at all, or publish them selectively. Through widespread ignorance, and occasional collusion, we have put up with this ethically outrageous situation for decades. “Evidence-based medicine” is often missing half the evidence; patients who gave consent to risk harm so that others might benefit are cheated. There have been occasional calls for open access to all trial data from industry but legislators and regulatory bodies have paid little attention, and until now nobody has come up with a working model for individual patient data to be analysed in a way that guarantees integrity and independence. But in the last 4 months, Harlan Krumholz at Yale has not only come up with the model but has also persuaded the largest medical device company, Medtronic, to support and fund two totally independent analyses of its data relating to a controversial bone product. I can’t believe my luck in being involved with this benchmark project, which is outlined here by its originator and a close colleague. If any company present knoweth of any just cause or impediment why its human trial data should not be disclosed in this way, let it speak now, or forever hold its peace and deliver them up.

NEJM  13 Oct 2011  Vol 365
1376   Barrett’s oesophagus is a classic example of a condition which has been talked up out of all proportion to its significance. It provides a comfortable living for an army of endoscopists and needless anxiety to countless thousands of individuals who live in daily fear of developing a nasty and usually lethal kind of cancer. In fact just 5,000 people a year die of oesophageal cancer in the USA, and the great majority of these have never had Barrett’s oesophagus. It is true that this is six times the figure in 1975, but it is still a tiny risk. Multiply it by 11 and it is still small, and this is the figure arrived at for the added risk from having a diagnosis of Barrett’s in this Danish whole-population database study. It agrees reasonably well with a previous Northern Ireland whole-population study. There is absolutely no evidence that regular upper GI endoscopy makes any difference. The all-cause mortality of people with and without Barrett’s is identical. As I’ve said before, I do wish he had never invented his oesophagus.

1406    Influenza is a diagnosis we seldom make with confidence in children between six months and six years of age. It is usually a mild febrile illness which rarely needs specific treatment. This Finnish trial shows that by giving two shots of an influenza vaccine with oil-and-water adjuvant, you can effectively protect children from a selected range of influenza viruses, as shown by PCR testing. Following each dose, there is a 60+% chance of a mild febrile illness and a local reaction. So you would have to be jolly worried about your child getting flu to want them to have this vaccine.

1417    In this article about adult primary care following childhood acute lymphoblastic leukaemia, there are two striking figures. One displays survival improvement from 1975 (40ish %) to 2003 (90+%): it’s a great teaching slide to illustrate incremental progress in therapeutics. The red line wiggles slowly upwards as treatment regimens are gradually refined. The other picture is of identical twins, one of who had treatment for ALL as a child. She is unrecognisable as a sister, let alone a twin: several inches shorter, with a moon face and obesity, as lasting legacies of prolonged steroid treatment and an irradiated pituitary. If you have such a survivor in your practice, it’s worth having a look at this piece: though if you carried out every bit of “primary care” that it advises, you might have to cancel the rest of your appointments for a couple of days.

Lancet  15 Oct 2011  Vol 378
1388   Here is a fine example of a genre of Outcomes Research first suggested by Francis Clifton in 1732: the surgical case-series. Two neurosurgeons from Queen Square report on the outcomes of 649 consecutive operations performed for adult epilepsy since 1990. They managed to cure half their patients completely, and most of the rest appear to have benefited. Nice when you can flourish personal results like these in The Lancet, while seeking generalizable lessons on case-selection and surgical technique in adult epilepsy.

1396   As I try to put together a book on Outcomes Research in one of the world’s leading centres of outcomes research, I am naturally keen to learn a bit about the subject. So far I have picked up some of the basics. You need data that are reliable. You need outcomes measures that are well defined and well documented. To link the two you need statisticians who understand the material. If you tried to use the software on your own you would go mad and produce garbage. You might anyway. Often you have to work with data that happen to be there, rather than the data you would like to have. These data are usually complex and so you have to think very hard about what variables matter, how they inter-relate, and what adjustments need to be made. Now this study of preoperative anaemia related to postoperative mortality and morbidity is fairly straightforward. The data come from American College of Surgeons’ National Surgical Quality Improvement Program database, a prospective validated outcomes registry from 211 hospitals worldwide in 2008. Anaemia is defined by haematocrit, which is confusing to UK readers, but is perfectly logical. The post-op outcomes seem reasonably well ascertained. So we can be pretty sure about the bottom line: if you come to non-cardiac surgery with a low haematocrit, you tend to have worse 30 day outcomes. Even mild anaemia is important. But the key clinical questions that arise are not answered in this study: we don’t have any information about the causes or duration of anaemia in this heterogeneous population, nor any information about any therapy given pre-operatively. Like all outcomes research, it suggests that a lot more outcomes research is needed.

BMJ  15 Oct 2011  Vol 343
The BMJ shows admirable impartiality in publishing this study comparing the speeds with which various medical updating sites actually update their material. The BMJ’s own product, Clinical Evidence, is the slowest by a good margin. The paper is written in a translated-from-the-Italian style which is hard to follow; it acknowledges that its methods were essentially bibliometric and did not examine content in detail. So it is quite possible that Dynamed, which dashes to cite everything the moment it appears, may suffer by tending towards the opposite extreme. Given the quality of much of what we call evidence, sometimes slow adoption is a good thing.

Looking at the contributors to diabetes and lipid guidelines in the USA and Canada, the authors of this study find that half of them have conflicts of interest. This is amazing. Where did they find the half without conflicts of interest? This murky topic draws a superb editorial from Edwin Gale and commentary from Fiona Godlee. Something must be done about this: and slowly, I think, there are signs that it is going to be done. But even without conflicts of interest, guidelines will always err towards over-investigation and overtreatment. Try not to use them.

Arch Intern Med  10 Oct 2011  Vol 171
1625    In an already famous contribution to the Less Is More series, a miscellaneous group of authors analyse data from the Iowa Women’s Health Study. In 1986, two-thirds of the cohort (mean age 62) were taking vitamin and/or mineral supplements; by 2004 (mean age 82) this had increased to 85%. There are striking differences between those who did and didn’t in these ill-balanced groups: for example, those not taking supplements did twice as much exercise. So in interpreting the mortality differences cited in this study, bear in mind how much depends on the accuracy of its multivariable adjusted proportional hazards regression models. Still, I am quite content to believe that all this stuff does no properly nourished person any good, and unnecessary iron supplementation almost certainly does harm.

1655   “Two major factors associated with skeletal fracture in older persons are intrinsic bone strength and risk of falling.” Hard to disagree with that. Now there are lots of risk-of-falling scores which we can look up but seldom use. This one requires just a stopwatch and a chair and it is called TUG, which stands for Timed Up and Go. Place your postmenopausal lady of advanced years in a chair and command her to get up and walk 3m and sit down again. This should take less than 10.2 seconds. Measure her hip bone density. Put the TUG score and the BMD together and you have a measure of non-vertebral fracture risk.

Fungus of the Week: Craterellus cornucopioides

Walking up Mount Tom in Rhode Island with my son Tom a week ago, I stopped to tie my bootlace, and he drew my attention to a little cluster of grey tubes emerging from the ground nearby. They were all but hidden in the leaf litter, and only by careful searching did we find a few more of these fabled trompettes des morts scattered in the vicinity. I could have stayed there all day, grubbing about like a hedgehog, but I sensed that this was not the young man’s idea of a good time. We strode to the summit and admired the vistas of New England fall colors. The next day we did Mount Tom in neighbouring Massachussets: a formidable forest climb, but no more trompettes to be found. The next day it was the turn of Mount Tom in Connecticut: fabulous views, but again no trompettes.

Alas, there are only two ways to find this fungus: by happy accident, or by knowing where it grows. To some extent this applies to all fungi, but Craterellus is so well-camouflaged that you have almost to be on all fours to spot it. So far as I know, dogs and pigs have never been trained for this purpose, though these little grey funnels are the nearest thing in taste to an above-ground truffle.

If you have the good fortune to come across some, bake a sea-bass or grill a Dover sole in their honour. Tear the trompettes into small pieces and cook them briefly in melted butter with pieces of shallot. If flavour means more to you than appearance, now add some white wine and reduce briskly: then add thick cream and boil briefly. Pour this sauce on the white fish and sprinkle with finely chopped parsley and a little chive. If appearance is important, you could do the wine and cream reduction separately and add the black trompettes after pouring this white sauce over the fish. But you will pay for your black-on-white showmanship by losing some of the magical intensity of flavour they can impart to the cream: a poor exchange.