Richard Lehman reviews the latest research in the top medical journals

NEJM  3 Aug 2017  Vol 377

“Beyond” randomised controlled trials?

These weekly reviews are now entering their 20th year, and over that time I’ve done lots of other writing and reading and reviewing, but I refuse to become a methodology dweeb. I write solely from the point of view of a generalist clinician trying to share decisions with individual patients in ten-minute chunks, even though I’ve now retired from doing that. Are randomised controlled trials an ideal method for informing clinical practice? Often they’re not. Is there a better way? Yes: it’s some kind of n-of-1 double-blinded process with rigorous long-term follow-up across a wide range of patient groups: something incredibly hard or perhaps impossible to carry out in real life as we currently know it. Here’s a single-author piece with this mission statement: “I describe the use of RCTs and alternative (and sometimes superior) data sources from the vantage point of public health, illustrate key limitations of RCTs, and suggest ways to improve the use of multiple data sources for health decision making.” Writing from the vantage point (or disadvantage point, as far as clinicians are concerned) of public health means that you can ignore shared decision making with individuals, because it is the population effect that counts. A wonderful peace then steals over the soul, as you contemplate how rational beings might be more physically active, eat the things you think are good for them, ride bikes, reduce salt and sugar, take statins etc. All of which may be true, but feelings of moral purpose don’t mean that you can ignore the inherent limitations of observational evidence, or apply it directly to individuals. I think there is much that could be done to make its quality better. At the moment seemingly barn-door figures such as death due to myocardial infarction can vary by 25+% depending on which database you use: it’s a whole order worse when it comes to “pharmacovigilance” in its current form. Yes, big data may—in some cases, and depending on the quality of the inputs even more than the rigour of the analysis—yield information which is so clear that it tips the hypothesis from null effect to the likelihood of an effect. But most hypotheses will still need to be tested. There are plenty of evidence sources beside randomised controlled trials, but they do not lie “beyond” them.

JAMA 1 Aug 2017  Vol 318

Statins: thrice more unto the breach, dear friends

For me, the chief adverse effect of statins is fatigue. I cannot read the word without an involuntary sigh. My heart aches, and a drowsy numbness pains my sense, As though of hemlock I had drunk… (Keats: Ode to a Nightingale). The first nightingale to warble upon the subject of statins this week is Harlan Krumholz, who with Rod Hayward in 2012-3 persuaded the American College of Cardiology/American Heart Association to adopt a risk-based rather than a lipid-level-based approach to the use of statins. Nonetheless the title of his viewpoint is “Treatment of Cholesterol in 2017”. I won’t try to explain: everybody should read this piece for themselves.

Nightingale number two is John Ioannidis, whose title promises an article about low-density-lipid cholesterol lowering using PCSK9 inhibitors. The most striking thing to me about these new drugs (apart from their cost) is the fact the while statins confer protection in proportion to lowering LDL-C, the proportionate effect is far, far less with the new drugs. But anyway, Ioannidis writes mainly about statins: again, read this for yourself.

Our final statin nightingale this week is Larry Husten, whose Cardiobrief website is essential reading for everyone interested in cardiovascular medicine. You won’t find a better account anywhere of how the statin controversy has played out and where it is currently heading. Personally I think a measure of resolution will come about through a novel combination of n-of-1 trials and a better understanding of human behaviour and the framing of risk. But I suspect there will always remain two camps who can scarcely understand the other’s point of view.

Do you need a clot retriever to retrieve clot in stroke?

Retrieval of clot from the anterior cerebral circulation is a major advance in the management of stroke, as illustrated in a number of well-conducted randomised controlled trials using stent-retrievers made of platinum or nitinol mesh. These devices don’t come cheap, and in the case of platinum I’m surprised at the use of such a rare and useful metal for such a lowly and transient purpose. A French trial decided to test whether simple clot suction could yield results equal to the use of these retrievers. It did. This could be practice-changing and warrants a large replication study.

JAMA Intern Med  Aug 2017

Sharing the process of diagnostic decision making

There aren’t many books I would say that every doctor should read, let alone read more than once. Osler said that of Don Quixote, with which I very much agree: and I would say that another is John Brush’s beautifully clear and well-illustrated guide to diagnostic thinking, The Science of the Art of Medicine. Set aside an afternoon each year to reread it and check how well you are doing in this vital aspect of practice. I was delighted to be able to invite John and his colleague James Brophy to write on Sharing the Process of Diagnostic Decision Making for the JAMA Internal Medicine Sharing Medicine series: they do not disappoint.

Ann Intern Med 1 Aug 2017

Treating latent TB

Last week’s piece in The BMJ about antibiotic courses fingered Mycobacterium tuberculosis as a “professional pathogen”. I’ve sometimes described it in these columns as a slouchy old killer, because most times it just lies around snoozing like a Mexican bandit on a hot afternoon in a bad film. In the last decade we’ve got better at detecting latent tuberculosis infection (LTBI), a proportion of which can later become invasive. A common definition is essential if you’re going to do a systematic review of treatment regimens, but I don’t see one in this NIHR-funded network meta-analysis. It’s surprising conclusion is that single agent regimens appear quite effective, though triple-agent regimens are more efficacious.

Whole-genome sequencing: more than we need to know

Over the last year, I took part in the close analysis of two genome-based diagnostic packages, and quickly reached the conclusion that they defied all possible criteria of diagnostic validity. For most of the components, let alone a combination of them, there was no way to derive a meaningful predictive value for the individual. With whole-genome sequencing, the capacity to create anxiety becomes infinite while the unravelling of meaning for most components is impossible, even theoretically. Technology has outstripped any possibility of full interpretation, like a universe of Heisenberg measurements. That’s my reading, anyway: the problem is somewhat more modestly stated in a study of whole-genome sequencing in a US primary care setting.

The Lancet 5 Aug 2017  Vol 390

Bevacizumab for advanced cervical cancer

“The benefit conferred by incorporation of bevacizumab [in advanced cervical cancer] is sustained with extended follow-up as evidenced by the overall survival curves remaining separated.” This is true, though the absolute difference is small and represents an average of three and half months for most patients: longer if they had not had pelvic radiotherapy.

Non-ST elevation MI: is quicker intervention better?

Non ST-elevation myocardial infarction emerged as a diagnostic category with the coming of troponin testing around the year 2000. It’s taking a while to determine how urgently patients with this syndrome need intervention, because most interventional trials were underpowered to show subgroup differences. Here’s a meta-analysis with an emergent message, which could do with further confirmation: “An early invasive strategy does not reduce mortality compared with a delayed invasive strategy in all patients with NSTE-ACS. However, an early invasive strategy might reduce mortality in high-risk patients.”

Is less more for low-risk breast cancer?

When a medical condition is a moving target, that is a good thing and a sign of progress. “Early breast cancer” means something different now compared with ten years ago, when the UK IMPORT LOW trial first started recruiting patients. (I used to mock acronyms on a regular basis once, but like UK/US politics, they have moved beyond parody now). “We showed non-inferiority of partial-breast and reduced-dose radiotherapy compared with the standard whole-breast radiotherapy in terms of local relapse in a cohort of patients with early breast cancer, and equivalent or fewer late normal-tissue adverse effects were seen. This simple radiotherapy technique is implementable in radiotherapy centres worldwide.” I won’t try and comment, because in the meantime there have also been trials of intra-operative local radiotherapy and a whole reclassification of the subtypes of “early breast cancer”. If somebody could come up with a shared decision tool for patients, I would commend them as a genius, but also defy them to keep it up to date.

The BMJ 5 Aug 2017  Vol 359

Opioids and psychotropics in pregnancy: pity the neonate

“In the US, about every 25 minutes an infant is born with signs of drug withdrawal.” And these are just the ones that are recognized, usually because the mother is known to have been taking opioids. The real scale of the problem is likely to be much larger. I don’t think this study of a subgroup of Medicare mothers giving birth between 2000-2010 can claim generalizable accuracy, but it does contain plenty of worrying data, and I hope it will stimulate prospective studies. More importantly, I hope it will make doctors more hesitant to prescribe opioids, antidepressants, benzodiazepines, and gabapentin to pregnant women except where the need is urgent.

Fungus of the Week: Agaricus langei

The rainy weather of the last weeks has been very welcome to a fungus-hunting, garden-loving misery-guts like me. I haven’t been out a lot but just an hour looking around the road-verges of Banbury yielded five different species of Agaricus, and I know where I could find a lot more.

There is only one common species of brown-gilled, ring-bearing mushroom that needs to be avoided, and that is the yellow-stainer. The name says all you need to know. Various field mushrooms can develop slight yellow patches on their caps where you touch them, but this species goes a bright chrome-yellow as soon as it is cut, most markedly at the base of the stem. It also smells inky and off-putting, whereas all good agarics spp. smell mushroomy and/or fennelly.

Agaricus langei  has a white squamous cap tending slightly towards grey, and declares itself within a minute of cutting, by turning red (fading later to brown). It is a mushroom of good texture and flavour, but one must not overstate the virtues of these wild agarics. They can be maggoty or sodden, and need to be eaten on the day of picking. Don’t say it too loudly, but cultivated mushrooms from a supermarket generally taste and keep better.

Mushroom-hunting is a bit like golf: a good pretext for a walk outdoors, but with the added bonus of wearing an anorak, enjoying the woods and coming across interesting things. Eat them at your own risk.