Richard Lehman’s journal review—5 September 2016

richard_lehmanNEJM  1 Sep 2016  Vol 375
Adding LABAs to steroid inhalers
This week’s print NEJM contains two trials of adding inhaled long acting beta-adrenergic agents to inhaled corticosteroids. The first recruited 6208 children with asthma from the ages of 4 to 11, and compared fluticasone alone with fluticasone plus salmeterol. We know that salmeterol alone should never be given to children, as the accompanying editorial makes clear: “Monotherapy with a LABA in a child should be considered medical negligence, and we suggest that single LABA inhalers should carry a warning to that effect, as required in the United States by the FDA in 2010.” But in this large trial over 26 weeks, serious events did not differ between the group where salmeterol was added to fluticasone and fluticasone alone, and there were no deaths in either group. Still, as the editorial says, “There is no evidence for the use of a combined inhaler as first-line preventive therapy in children, and this fact needs to be emphasized because such use is increasingly creeping into practice.”

I find it odd that this editorial doesn’t cover the second trial, which recruited 11 693 asthma patients from the age of 12 upwards, and randomised them to inhaled budesonide alone or budesonide plus formoterol. This Astra-Zeneca funded trial reports its findings thus: “Among adolescents and adults with predominantly moderate-to-severe asthma, treatment with budesonide–formoterol was associated with a lower risk of asthma exacerbations than budesonide and a similar risk of serious asthma-related events.” Well, that’s technically correct, but there were in fact two asthma related deaths in 26 weeks, both in the 5846 patients randomised to receive the combined inhaler. It would be almost impossible to power a trial to make asthma related death the primary endpoint, but given what we know about the dangers of beta-agonists, this still sounds warning bells for me. Although observational studies cannot settle the question definitively, I’d be interested to know what the large databases show.

How to make your trial more positive
“The Primary Outcome Fails—What Next” is a provocative title that had some wondering if the piece was a deliberate attempt to encourage data torturing, or even written tongue in cheek. But I would encourage anyone with the time and interest to read this article by Stuart Pocock and Gregg Stone with an open mind. For a start, it’s a formidable piece of scholarship. Secondly, I don’t read it as undermining the principles of critical appraisal, even though I find some of the arguments tenuous, especially if you try to apply them to clinical reality and shared decision making. Thirdly, it will be followed by a second piece by the same authors explaining how we should question apparently positive trials. I’m told that Stuart gives a popular Christmas lecture at the London School of Hygiene & Tropical Medicine called “Crap Trials in Cardiovascular Medicine,” so this article should be worth waiting for. As for the NEJM‘s motivation in commissioning these papers, I have no idea, but they could not have provided a better argument for opening trial data up to independent scrutiny and interpretation. Parasites, there is work for you to do: be fruitful and multiply.

JAMA  23/30 Aug 2016  Vol 316
RNA to detect serious bacterial infection in kids
In developed countries, the number of children who are admitted to hospital with febrile illnesses continues to rise, while the proportion of those with serious bacterial infection (SBI) has been falling for decades and is now typically 2-3%. At the same time, antibiotic usage remains fairly steady. As Howard Bauchner, the paediatrician in charge of JAMA, puts it in his editorial:

“The substantial decline in the prevalence of SBI, following the introduction of conjugate vaccines, has made clinical decision making more difficult—the needle has become much smaller, and the haystack much larger, particularly in young infants.”

This issue of JAMA contains two reports of lengthy and painstaking international work that has been going on to improve discrimination between viral and bacterial infection in children, based on RNA expression. These reports are classed as “preliminary,” though they are very detailed and raise the exciting prospect that RNA signature tests will have an important place in clinical practice within a few years. They also both have “diagnostic accuracy” in their titles and quote sensitivity and specificity statistics, but none of this has any meaning independent of the populations studied and the comparisons used: a point that can never be made too often. The first study looked at hospital populations of small children (median 19 months): these included 52 with definite bacterial infection and 92 with definite viral infection: 69% and 35% respectively required intensive care. The second study looked at infants of median age 37 days, and bacteraemia by culture was one of the criteria by which the RNA test was judged. So these are not your ordinary babes and toddlers as seen in a GP surgery or out-of-hours centre, and it is pointless to speculate on the predictive characteristics of RNA tests in such a context. Nor do we know how this should affect the prescribing of antibiotics, or the consequences of withholding them: there are interesting times ahead for primary care paediatric research.

JAMA Intern Med  Sep 2016  Vol 176
Drug changes after fragility fracture
What should you start and what should you stop after a patient has had an osteoporotic fracture? I was lucky to work with a GP partner who was interested in such things long before they became fashionable: she made us do regular audits and medication reviews. I’m even prepared to believe that the UK Quality and Outcomes Framework may have subsequently improved practice in this area, because I find it hard to think that UK figures are as bad as the ones in this Medicare sample from the US.

Fewer than 25% of the patients who had fragility fractures received drugs to enhance bone mineral density, even after the event. Medications like opioids, benzodiazepines, and other sedatives associated with risk of falls were rarely discontinued. The situation was slightly better for drugs that can sometimes worsen bone fragility, such as proton pump inhibitors, thiazolidinediones, SSRI antidepressants, and antipsychotics.  Patients on oral steroids usually remained on them, at the same dose.

Docs in the wrong ball park
I don’t give many lectures, but when I do, I often ask medical audiences a few simple questions about the number needed to treat for common drugs. So I’m not entirely surprised by this survey of American, academic internal medicine physicians’ understanding of the benefits and harms of common medical interventions. For pretty well everything, their estimates of both were too high. Then the investigators looked at physicians’ use of statistical terms in patient communication and their awareness of high value healthcare campaigns. But this is embarrassing. I must stop, or else I’d be saying that we currently train doctors to be ignorant about what they do and also how to explain it to their patients. And this surely cannot be.

Depression treatment upside down
But now for something even spookier. “Most US adults who screen positive for depression did not receive treatment for depression, whereas most who were treated did not screen positive.” Questions: does this mean that screening for depression is a good thing? Does it mean that antidepressants are massively effective? Do we know what we are doing? Sorry, I asked that question the previous time.

Lancet  3 Sep 2016  Vol 388
Triple therapy for COPD
Triple therapy satisfies the desires of three parties: patients who find themselves getting worse and want something new to try, doctors who are running out of things to try, and drug companies who want a large market for unoriginal drugs. When so many people stand to be made happy, should we really look for other benefits? Chiesi Farmaceutici SpA funded this trial of their inhaler containing beclometasone dipropionate, formoterol fumarate, and glycopyrronium bromide, comparing it with beclometasone dipropionate and formoterol fumarate alone. “The funder of the study was responsible for the design and analysis of the study, oversaw its conduct, and was responsible for the study report preparation.” And their interpretation reads: “We provide evidence for the clinical benefits of stepping up patients with COPD from an inhaled corticosteroid/long-acting β2-agonist combination treatment to triple therapy using a single inhaler.” This clinical benefit amounted to a reduction in the annualised rate of moderate to severe exacerbations from 0.53 to 0.41, i.e. 0.12 of an annual flare-up of COPD across the treated population. But although this tiny marginal result reached statistical significance, the trial only lasted half a year. And there are other flaws in it too, as detailed in the accompanying editorial, which suggests we need more trials with other comparators. It concludes that “Until these next studies become available, we can be comforted by the knowledge that three therapies can be combined in a single inhaler which offers more effective therapy than at least one of the recommended treatment, regimens for patients with severe COPD.” Call me a grouch, but I stand uncomforted.

Pacing the diaphragm for sleep apnoea
Here’s a trial that was done mostly on fattish men of mean age 65 of German/Polish birth who had sleep apnoea. This is a fairly accurate description of me. But, in fact, over half had coronary heart disease and over 40% had atrial fibrillation, which isn’t yet true of me. And now the penny drops: these people had central sleep apnoea, not common or garden obstructive sleep apnoea like mine. Central sleep apnoea is a disorder of the respiratory control centre, common in heart failure and associated with a bad prognosis. In fact, seven of the 151 participants died in the two years covered by this report. Everyone was fitted with a pacing device, which transvenously stimulates a nerve causing diaphragmatic contraction similar to normal breathing. In the first six month phase of the trial only half the devices were activated. Retweaking was required to make the device perform comfortably in more than a third of the patients. But it worked very well in reducing the apnoea-hypopnoea index: long term hard endpoints won’t be in for a while.

The BMJ  3 Sep 2016  Vol 354
Ultrasound as fashion accessory
The BMJ (before it became The BMJ) once sported a cover showing an ultrasound device dangling around a doctor’s neck in place of the iconic stethoscope. Although ultrasound images used to look mostly like grey blurs to me, I’m told that younger and cleverer people can detect all sorts of things with them, such as joint surface erosions in rheumatoid arthritis. The Norwegian ARCTIC study set out to discover if ultrasound monitoring might improve the clinical management of early rheumatoid arthritis. It did not. “The findings highlight the need for randomised trials assessing the clinical application of medical technology.” Yes indeed.

Improved prescribing in general practice
During his 12 years as director of the UK National Prescribing Centre and the NICE Medicines and Prescribing Centre, Neal Maskrey achieved over £2bn in prescribing savings and patiently and successfully made British GPs aware of safety issues with a number of widely used drugs. As a distant recipient of his efforts, I look back with nostalgia at the discussions we used to have 15 years ago during visitations from prescribing advisers, who were invariably polite and well informed and worked with real data about our practice’s prescribing. This is a wheel that definitely needs reinventing, however unfashionable it has become to put actual individuals in direct conversation with each other. Bruce Guthrie’s trial does a bit of that, plus slightly more electronic feedback than we used to get those years ago. The primary outcome was a patient level composite of six prescribing measures relating to high risk use of antipsychotics, non-steroidal anti-inflammatories, and antiplatelets. The intervention worked. We already knew it would. The wretched NHS rediscovery cycle goes round and round.

Plant of the Week: Physalis peruviana
As its botanical name implies, this plant comes from Peru. It is a member of the nightshade/tomato/potato family. Here in England, we call it the “Cape Gooseberry,” which is wrong on every count. Britons brought it from South America to grow at home in the 18th century, and then took it to the Cape of Good Hope in South Africa, where it grew much better, and from there to Australia and New Zealand, where it grows best of all. In the US it’s known as the goldenberry, or more fancifully as the Pichuberry since Machu Picchu is the place in Peru that Americans know best. I doubt whether you will find many pichuberries there.

We have three in our garden: the parent plant and two nearby children. Alas, they are all runts. They appear above ground in late July and by now they are a few inches high. The fruit cases will be ornamental if you can trouble to bend over and get sight of them. There may be a small fruit or two to eat and make your nose pucker.

Physalis fruit are best bought in supermarkets and look good on the plate with almost anything, which is why they are so ubiquitous in restaurants. They also taste good with most things, though I have never properly mastered the art of cooking with them. In theory their tartness should go well with fatty fish such as mackerel and grey mullet, or with fatty meat like duck or belly pork. The nearest I came to success was with a sauce to accompany seared scallops, in which the physalis fruit were cooked with chopped tomato, a scrap of chilli, and a little white wine. The other accompaniment was lightly cooked cauliflower crushed with crême fraiche and pounded cardamom seeds. Only the best scallops (and guests) are worth this amount of bother.