Richard Lehman’s journal review—30 May 2017

Richard Lehman reviews the latest research in the top medical journals

richard_lehmanNEJM  25 May 2017  Vol 376

Assisted dying in hospital

Assisted dying has now been legalized in some form in five European countries (the Netherlands, Belgium, Switzerland, Germany, and Luxembourg), six US states (Oregon, Washington, Vermont, Montana, California, and Colorado), Colombia, and most recently Canada. If the day comes when the wishes of the majority in the UK are respected by legislators, it will become legal here too. This very clear and practical article describes how it is being implemented in hospitals in Canada. In fact this is quite a rich week for those interested in this topic. There is a Viewpoint article in JAMA IM which discusses the difficulties encountered from conscientious objection by clinicians in California, and The BMJ contains a discussion about extending assisted dying to old people who are not terminally ill in the Netherlands. For those who want to reach an informed view of what should happen here, these are all required reading.

Getting in fast for sepsis

In medicine, what seems obvious is not always true, but in the case of antibiotics for sepsis, the obvious is indeed likely to be true: get them in as soon as possible. Since nobody is going to do a randomised trial of delayed antibiotics for sepsis, this observational study from New York is as close as we will get to proof. The mandated intervention consisted of a 3-hour bundle of care for patients with sepsis that involved obtaining blood cultures, administering broad-spectrum antibiotic agents, and lactate measurement—all to be performed within 3 hours. This was achieved in 82.5% of patients across 149 hospitals. They were given a bolus of fluid too, but the timing of this made no difference to outcomes, whereas those who were given their antibiotics more quickly showed a small reduction in risk-adjusted mortality.

Angiotensin II for shock

Vasodilatory shock is the name given to the most common type of shock, in which there is peripheral vasodilation and a large drop in blood pressure despite preserved cardiac output. To an old doc like me this seems counterintuitive, as we tend to think of patients in shock as “shut down,” cold, and clammy. Anyway, the kind of doctors who look after such patients are principally concerned with preventing irreversible organ damage by keeping blood pressure up, using adrenaline/epinephrine. Physiologically, you could also do this with angiotensin II, and this is what was tried here in shocked patients who were not responding to maximal doses of high-dose vasopressors. In this industry-funded trial, angiotensin proved better at maintaining blood pressure in this group than placebo. This often allowed doses of catecholamines to be reduced, but the trial was not powered to measure effects on mortality or organ failure.

Benralizumab for asthma requiring oral steroids

Some people have terrible asthma with lots of eosinophils in their airways, and cannot get by without continuous oral steroids. Benralizumab is an antibody designed to kill eosinophils, or, to put it more amply, a humanized, afucosylated (engineered to eliminate fucose sugars from the oligosaccharides in the Fc region) monoclonal antibody directed against the alpha subunit of the interleukin-5 receptor that induces direct, rapid, and nearly complete depletion of eosinophils by means of natural killer cell–mediated antibody-dependent cellular cytotoxic effects. In this Astra Zeneca funded trial, most of these severely affected patients who were given benralizumab rather than placebo were able to reduce their oral steroids without any loss of lung function at 28 weeks—but without any gain either. It’s hard to know how this drug may or may not fit into the future treatment of people with severe eosinophilic asthma: you could call it an example of uncertain benralizability.

JAMA  23/30 May 2017  Vol 317

Combining antibiotics for cellulitis

Between-country variations in drug availability are a complete mystery to me. In the UK we use flucloxacillin as first-line treatment for most skin infections, but when I mentioned this agent to US doctor friends, they looked blankly. It isn’t available there. There as here, most non-purulent cellulitis is caused by streptococci which remain sensitive to most beta-lactam antibiotics, but in contrast with British practice, it seems that cefalexin is a popular choice in America. This randomised controlled trial of adding trimethoprim-sulfamethoxazole to cefalexin for uncomplicated cellulitis seems to be predicated on the idea that some of the infections are caused by meticillin-resistant Staphylococcus aureus. But within the limits of the trial, this combination did not seem to make any difference to cure rates. To be honest, I can’t quite understand the rationale of this study.

Teaching hospitals: the difference shows

Teaching hospitals took to calling themselves “centres of excellence” in the 1990s and the term seems to have persisted. Sadly for those of us who like to prick such bubbles of self-congratulation, it seems that teaching hospitals actually are better than other hospitals in the USA. In this survey, the primary outcome was 30-day mortality rate for all hospitalizations and for 15 common medical and 6 surgical conditions. Teaching hospitals had significantly better results across the board, albeit not by huge margins.

JAMA Intern Med  May 2017  Vol 177

Looking at the elderly in a statin trial

Post-hoc subgroup analysis is always a sin, but some sins are more excusable than others. Here we have a post-hoc analysis of the effects of pravastatin in people over 65 included in ALLHAT-LLT. This is so large a group that no real excuses are needed, and I think we can accept the conclusion that pravastatin had no effect when given for primary prevention to over-65s with moderately elevated lipids and blood pressure. But in discussing actual mortality increases in over-75s given statins, I think we need to be a bit more circumspect. This would require individual participant data meta-analysis of greater scale and sophistication: an urgently necessary task which some trialists are arguing against even attempting, despite its huge implications for clinical practice.

Ann Intern Med  23 May 2017  Vol 166

MR CLEAN stroke treatment and extra cranial atheroma

I consider it my job to be sceptical in these reviews, but even I have been convinced by the growing evidence that intra-arterial procedures to retrieve clots from the cerebral circulation represent a real advance in stroke management. One key trial was MR CLEAN, which included a prespecified subgroup of stroke patients with a >50% occlusion of the carotid artery in the neck. It might be supposed that those with external carotid stenosis would benefit less from clot retrieval within the brain arteries, but this was not the case. Here is a procedure which can help a lot of disparate stroke patients.

The Lancet  27 May 2017  Vol 389

Helping children detect unreliable health claims

Last Monday I spent a joyful and fascinating afternoon listening to the work of a team from Uganda and Norway who had worked for years on ways to teach African children to detect unreliable health claims. The twin aspects of their work are described in two papers on The Lancet website. One was a cluster-randomised trial of a teaching package which had been carefully designed over 6 years. “The results show that it is possible to teach primary school children to think critically in schools with large student to teacher ratios and few resources.” They certainly do, and it’s only a pity that readers can’t share the wonderful video clips which the researchers used in their presentation, especially a girl of about ten summarising the essence of shared decision-making with a broad smile: “Health choices are mine. They are mine. Nobody else’s.” The second study involved a podcast given to parents which led to a large improvement in the ability of parents to assess claims about the effects of treatments. “Future studies should assess the long-term effects of use of the podcast, the effects on actual health choices and outcomes, and how transferable our findings are to other countries.” Congratulations.

Excluding pulmonary embolism with the YEARS rule

The trouble with decision tools is that they are erratically used, and when they are tested, they often add nothing to clinical judgement. When a diagnosis carries serious implications for management (not to mention the possibility of litigation) the temptation is to skip both rules and judgement and go straight to a “definitive” investigation. In the case of pulmonary embolism, this is computed tomography pulmonary angiography (CTPA). A new decision tool for suspected PE called YEARS was developed and tested in 12 Dutch hospitals. If you are used to relying on Wells’ rule and D-dimer, then you will carry out 14% more unnecessary CTPAs than if you use YEARS, the authors conclude.

Post-mortem CT angiography for sudden death

The reason for avoiding CTPA on people who are alive is that it exposes them to quite high doses of ionizing radiation. But CT scanning can be freely used on the dead, and here is a single-centre study which shows that post-mortem CT (PMCT), enhanced with targeted coronary angiography (PMCTA) provided a cause of death in 92% of coroner-referred cases.

I was surprised to learn from the introduction of this article that England and Wales have one of the highest frequencies of autopsy in the world. It had always seemed to me that there were too few, and that doctors by and large continue to bury their mistakes.

The BMJ  27 May 2017  Vol 357

Antibiotics and cough complications

For the last 40 years, I must have read hundreds of articles about antibiotic prescribing for respiratory infections in general practice. This is the gift which keeps on giving. Here is a study of 28 779 adult patients with lower respiratory tract infections.

Subsequent hospital admission or death occurred in 26/7332 (0.3%) after no antibiotic prescription, 156/17 628 (0.9%) after prescription for immediate antibiotics, and 14/3819 (0.4%) after a prescription for delayed antibiotics. These raw data are quite interesting: they show that the patients given immediate antibiotics by GPs really are sicker, because the difference disappears with multivariable analysis. In fact with this adjustment all the group differences become non-significant. I cannot say what this implies for practice. Last year I was worried to read a study of my own locality (Oxfordshire) in Thorax showing a rapid increase in admissions for community-acquired pneumonia since 1998, especially from 2008-2014. Oxfordshire GPs are of course the best in the world, and prescribe fewer antibiotics than most. I left their ranks in 2010, further reducing the antibiotic prescribing rate.


In his terrific book, The Science of the Art of Medicine, John Brush comments that most people find it impossible to conceptualise risk and probability. But if we are to share decisions with patients, we need not only to learn how to do it ourselves, but also to impart understanding simply to patients. Measuring CV risk is important, and Julia Hippisley-Cox’s QRISK scores are widely recognised as the best there are for the UK. The latest, QRISK3, incorporates new clinical variables and improves on the predictive value of earlier versions. But in the end, we are trying to apply this knowledge to millions of free-living individuals, each with their own cluster of modifiable and non-modifiable risk factors. In a desperately undermanned NHS, we are going to have to find ways of converting our understanding into helpful advice and support for each of them. The alternative—to be decided next week—is to let the devil take the hindmost.

Plant of the Week: Rosa “Gloire de Dijon”

In 1926, at the age of 17, Graham Stuart Thomas became an assistant gardener at Cambridge University Botanic Garden. He had to buy his own boots, which cost more than a week’s salary. So he took good care of his boots. He went on to become the greatest scholar of garden plants that England has produced, writing to the age of 90 and travelling around National Trust Gardens as their chief adviser. His books cover every class of hardy garden plant, but his greatest work of scholarship consisted of three books about roses, collected together later as the Graham Stuart Thomas Rose Book, in which his own watercolours appear together with an abundance of other beautiful illustrations.

Ironically, his work was a leading inspiration to David Austin, whose modern English hybrids now dominate the rose trade to the exclusion of many of the lovely old roses that Thomas describes. One of Austin’s first big successes was a scented buff yellow rose which he named Graham Thomas, and which you can find in every garden centre. Meanwhile, the marvellous and versatile old French climber called “Gloire de Dijon”, which used to be common, is getting seriously hard to find. But it’s very much worth the effort if you are looking for a tough but elegant rose for a north wall or fence. Here is Thomas on the subject: note the mixture of precise  scholarship with telling description and authoritative advice:

‘Gloire de Dijon’ Jacotot, France, 1853. Far removed from the original Noisettes, and as near to a Tea Rose as ‘Lamarque’: it was the result of crossing a vigorous tea rose with the Bourbon ‘Souvenir de la Malmaison’… It is the most popular and satisfactory of all old climbing roses; in constant flower in sun and shade, admirable for wall training, and the giver of a rich fragrance from its deep buff-yellow flowers suffused in warm weather with pink and apricot. It is rather leggy in growth, but this can be dealt with by planting with a small-growing Clematis, or growing it between two shrubs on the wall: it certainly needs support…(It) is in many respects the best yellow climber still.