JAMA  5 Oct 2011  Vol 306
1461    American medicine is a mass of quirks and contradictions: like medicine anywhere else, but magnified by huge financial forces pulling in different directions. There is a big financial incentive for cardiologists to do percutaneous interventions in elderly patients with stable coronary disease, for example, though the evidence suggests that few of them need it. I don’t know how Medicare handles that dilemma, but it willingly pays for these patients to spend a night in hospital following PCI. And I guess because the hospitals and the cardiologists get paid, nobody complains. But the fact is that most of these patients could perfectly well go home the same day, as confirmed by the small subgroup of patients who actually did in this Medicare patient study. Yet the incentives are all weighted towards them staying in unnecessarily, including the fear of litigation when something goes wrong, as always will happen. So American PCI patients are stuck with a night in hospital, whether they need it or not, and ours are stuck with going home the same day, whether they are fit to or not. I would love to see the day when health services really begin to learn from each other, but until that happens there is only one safe rule: never copy the USA.

1483   Now I shall immediately appear to break that rule by praising this Commentary piece which explains the idea of time-limited trials near the end of life. This has nothing to do with research trials, but rather practical trials of treatment to see if they make dying patients better. The key point is to have an explicit plan for stopping them if they don’t. We all automatically do such trials of treatment in the hope of helping the dying in real life, I hope: the difference here is the clarity with the process is made explicit, and that is something America tends do better than Britain.

NEJM  6 Oct 2011  Vol 365
1273   The main papers in this week’s NEJM give me déjà-vu like I can’t remember. Will there ever be an end to trials of new chemo combinations for breast cancer and new drugs for relapsing multiple sclerosis with tantalizingly small benefits? I suspect that as long as I go on reviewing, I shall have to keep telling you about such things. Incremental advances in combination treatment have led to complete cures for certain types of cancer, but this trial does not get us any nearer to that in HER-2 positive early breast cancer. The title of the editorial, “Steady progress in HER-2 positive breast cancer” belies its message which is that trastuzumab-containing regimens do not improve on existing anthracycline-containing regimens. Anthracyclines can induce irreversible heart failure and some leukemias, while trastuzumab does not: but the final balance sheet of mortality is the same.

1293   What is the current best treatment for relapsing multiple sclerosis? If it is not placebo, then this trial was arguably unethical. Although I’ve read all the main MS trials over the last 13 years, I wouldn’t be sure what to opt for if I had MS. I don’t think I’d bother with teriflunomide on the basis of this pharma-funded trial. Hair loss: that I can live with (or I would be dead by now); diarrhoea and nausea: OK, I’ve lived with those from time to time, but I don’t want to more than I have to. A 6% absolute reduction in disability over two years might result, which is a benefit I would never be aware of. Personally, I would say no.

This is a great teaching paper, to be used in conjunction with Know Your Chances by Woloshin, Schwarz and Welch. I’ve put myself in the position of the patient here and said no thanks. But real patients might want to make a variety of decisions, assuming that NICE approves the drug and your local consortium decides it can be afforded. The manufacturers, of course, herald this as a 30% reduction, whereas from the patient’s point of view it is a reduction from a one quarter chance of progression to a one fifth chance in two years. The patient needs to know the detail and have time to mull it over. Discuss how the content of this paper would need to be processed to enable this decision. Would a trial to assess tolerability be a good first step?

Lancet  8 Oct 2011  Vol 378
Once again there are no clinical trials in this week’s printed Lancet, and such studies as there are I have already told you about from the website. As some readers predicted, that makes it difficult for me to know when to talk about what. I could talk about the general decline of The Lancet, and Richard Horton’s Offline page, which now exerts a horrible fascination, like the poetry in JAMA, or watching Fox News. But I do not want to advertise these deplorable aberrations. Let’s turn to the one trial of interest to general practice on The Lancet’s website:

Nobody could describe The Lancet as the friend of British general practice, but when studies from UK primary care do make it into these supposedly august pages, they tend to be very good. This one is outstanding, and should inform daily practice. Patients with low back pain were assessed and randomized as they presented to ten GP surgeries. The physiotherapist assessors divided them into three groups: low, medium or high risk for lasting disability. They were then allocated to further care at the discretion of the physio, or tiered care according to their risk of chronic incapacity. Tiered care proved better. The notable features of this trial were (a) that the comparator was better-than-usual care (i.e. immediate triage and physio) and (b) that it stratified patients logically and avoided the one-size-fits-all approach that discredits most research in the eyes of the jobbing practitioner. As for the scoring instrument, economic assessments etc, I have no doubt that criticism is possible, but it will not come from me. General practice needs many more trials for like this, generating questions from the point of view of the patient, using interventions that are cheap and practical, and providing answers in terms of the outcomes that matter.

BMJ  8 Oct 2011  Vol 343
On my first clinical “firm” as a medical student, the consultant asked the student next to me to describe arcus senilis. He looked the ageing consultant in the eye and described it perfectly. It seems that we must call it arcus corneae these days, and it predicts nothing, according to the Copenhagen City Heart Study. By contrast, xanthelasmata are a marker for accelerated atheroma, independently of other risk factors. This is not significantly mediated by total cholesterol. The take-home message therefore is not “well, we all knew that already” but that we can measure cholesterol in people with xanthelasmata out of curiosity, but prescribe statins irrespective of what it turns out to be. And above all offer them help to stop smoking, if (as is likely) they are smokers.

Always approach modelling studies with scepticism, and that includes the “standard error” figures they give. You and I have to take these things on faith, and faith is not a virtue when reading the medical literature. I only draw your attention to this modelling paper on smoking and tuberculosis to emphasize that these are two evils which combine to kill millions in the resource-poor world, despite the fact that the first is totally avoidable and the second is largely treatable. If the toll goes up in the next 40 years, as this paper predicts, then the world will have messed up big-time.

Ann Intern Med  4 Oct 2011  Vol 155
Clinical decision rules abound, and to me they are rather like most works of classical music written after about 1920: I know I should try to understand and like them, but all they do is fill me with a vague sense of unease. They seem to work against intuition, rather than with it: they mess with the smooth inner workings of the mind. The classic playground for decision rule-makers is venous thromboembolism, and pulmonary embolism in particular, where Wells has gifted us a Schoenberg-like series of scores. I exaggerate: Wells is more like common sense slowed down. So what if you just take out the score, speed up common sense, and then do a D-dimer? According to this meta-analysis comparing clinical decision rules with what the Americans call gestalt, the results are just the same.

Plant of the Week: Colchicum autumnale

I was reminded of this plant by a report in this week’s Annals of a randomized trial of colchicine as prophylaxis for recurrent pericarditis.

It reduces recurrences by about a third.

I first came across colchicine in the school biology lab, where it was used to arrest mitosis, so we could pretend to see and admire the spindles of a dividing nucleus under a high-power microscope.

It is still a great treatment for gout in people who can tolerate it, and is the only safe agent in the gout we induce in heart failure patients by giving them loop diuretics. But hey, I am supposed to be talking about the beauties of the autumn crocus (or “naked ladies” as they are sometimes called).

All colchicum flowers are purple in various states of paleness: they rise and then sag in various poignant poses among the bare ground and the leaves of autumn. They are plants for the larger garden, to be wandered among as the mists rise and the sun gives a watery glow. Stuck in a small flowerbed, they just look out of place, and sad in all the wrong ways.

You can buy bags of the corms for planting earlier in the year, usually giving the name of a variety and a misleading description and picture. To me they all look much the same when they come up. Do not eat any part of this plant: Wikipedia informs us curtly that “Murderess Catherine Wilson is thought to have used it to poison a number of victims in the 19th century.”