Richard Lehman’s journal review—12 September 2016

richard_lehmanNEJM  8 Sep 2016  Vol 375
CPAP & CV events
I adore CPAP and bless the night when she arrived in my life. I embrace her—or rather she embraces me—for seven or eight hours of sleepy bliss. I will not hear a word said against her. Does she reduce my risk of cardiovascular events? Probably not, but I couldn’t care less. Here is a trial that randomly assigned 2717 eligible adults between 45 and 75 years of age who had moderate to severe obstructive sleep apnoea and coronary or cerebrovascular disease to receive continuous positive airways pressure treatment plus usual care or usual care alone. Over 3.7 years, “No significant effect on any individual or other composite cardiovascular end point was observed. CPAP significantly reduced snoring and daytime sleepiness and improved health-related quality of life and mood.” It’s remarkable that it even did that, because the mean duration of adherence to CPAP therapy was just 3.3 hours per night. In the NHS, you wouldn’t be allowed a machine if you used it so little. Additionally, I wonder if the selection process distinguished accurately between obstructive sleep apnoea and central sleep apnoea. I don’t think this trial tells us a great deal.

Cord blood for minimal residual disease
The disease referred to here—and in the same way in the NEJM title— is acute leukaemia or the myelodysplastic syndrome. The 582 consecutive patients underwent myeloablation and then had a hematopoietic-cell transplant from an unrelated cord-blood donor (140 patients), an HLA-matched unrelated donor (344), or an HLA-mismatched unrelated donor (98). The problem in this area is that it can be hard to find an HLA-matched donor in a hurry, as we know from newspaper stories. So the relative success of cord-blood transfusion in this series is a hopeful signal, though short of definitive: “Our data suggest that among patients with pretransplantation minimal residual disease, the probability of overall survival after receipt of a transplant from a cord-blood donor was at least as favorable as that after receipt of a transplant from an HLA-matched unrelated donor and was significantly higher than the probability after receipt of a transplant from an HLA-mismatched unrelated donor. Furthermore, the probability of relapse was lower in the cord-blood group than in either of the other groups.”

Pocock & Stone part two
Pocock & Stone, you will remember, are not a worthy firm of Wiltshire solicitors but a pair of eminent medical statisticians, who caused a flutter in the dovecotes last week by suggesting ways that data might be reworked to put a positive slant on seemingly negative trial results. Here, under the title “The Primary Outcome Is Positive—Is That Good Enough?”, they operate in the opposite direction. This is a good account of mainstream critical appraisal, and to my surprise I found that I’d reviewed most of the trials they cite over the years and come to similar conclusions, as far as I remember. I don’t suppose I should be surprised, and I suppose I should remember; but hey, I’m old.

Stents, bare & eluting
There are stent wars, and statin wars, and Star Wars. What is it about the ancient Indo-European ST root that provokes men to war? I say men because you don’t find many women in these sagas. The very words “bare metal” have a masculine character. “Eluting” perhaps less so. But the fact is that drug eluting stents pretty well destroyed the bare metal tribe in the early 2000s, for reasons I could never quite understand. The eluters then started fighting among themselves, like characters in a Roman play by Shakespeare: Sirolimus slew Zotarolimus (or was it the other way round?) while Everolimus fought Paclitaxel with a great clatter of stage swords. Slowly, the audience trickled out. I was left reporting on trials that even interventional cardiologists hardly cared about. And now, at last, comes a large, long term randomised trial comparing bare metal stents with eluters: the very thing we needed all along. “In patients undergoing PCI, there were no significant differences between those receiving drug eluting stents and those receiving bare metal stents in the composite outcome of death from any cause and nonfatal spontaneous myocardial infarction. Rates of repeat revascularization were lower in the group receiving drug eluting stents. (Funded by the Norwegian Research Council and others; NORSTENT.)” Imagine if health systems around the world had waited for the actual evidence: tens of billions of pounds/dollars might have been saved in needlessly expensive stents and antiplatelet agents.

JAMA  6 Sep 2016  Vol 316
MRI in early pregnancy
Magnetic resonance imaging is safe in the first trimester of pregnancy. In a case-control study of the whole population of Ontario, investigators found no difference in the risk of stillbirth or neonatal death within 28 days of birth and any congenital anomaly, neoplasm, and hearing or vision loss evaluated from birth to age 4 years. Add in gadolinium contrast agent, however, and the picture gets nasty: “Gadolinium MRI at any time during pregnancy was associated with an increased risk of a broad set of rheumatological, inflammatory, or infiltrative skin conditions and for stillbirth or neonatal death.” I don’t know to what extent this was already known, but it sounds as if it would now be medical negligence to give gadolinium to any woman of childbearing age without first excluding pregnancy.

JAMA Intern Med  Sep 2016
Gluten free fashions
A few weeks ago, the New Yorker ran a cartoon of two girls meeting for lunch and one of them saying “I’ve only been gluten free for a week, but I’m already really annoying.” I’m afraid that I may be distantly implicated in this: around 1997 I heard about a new blood test (anti-endomysial antibody) for gluten enteropathy and alerted my work partner Harold Hin to its potential for detecting coeliac disease in the community. His was the first primary care study in the UK to show that histologically confirmed CD had a prevalence 10-100 times greater than previously thought, and that its symptoms could be almost anything, or nothing. So now people with almost any symptoms, or none, have taken to gluten free diets, even though they test negative for coeliac disease. This is confirmed in the US National Health and Nutrition Examination Surveys (NHANESs) 2009-2014.

Lancet  10 Sep 2016  Vol 388
Early PCI for NSTEMI
One consolation for spending every weekend for the past 18 years writing these reviews is to see the results of long term trials that I hoped somebody would do a decade or two ago. “Non-ST elevation myocardial infarction” barely existed as a diagnostic category until the coming of troponin testing around 2002: this was also the time when immediate percutaneous intervention took over from thrombolysis as the preferred option for ST-elevation MI, and possibly NSTEMI too. It was a time of massive change, into which drug eluting stents slipped as part of a rich mix. They now turn out to have been largely a waste of money. Can the same be said of service reorganisation to provide immediate PCI for NSTEMI? Few countries can have struggled more to provide it than Sweden, with its widely dispersed range of small towns and communities. But they did the right study (FRISC) early on, so that now we can know the answer to that question from a prospective randomised trial in 58 Scandinavian centres:

“During 15 years of follow-up, an early invasive treatment strategy postponed the occurrence of death or next myocardial infarction by an average of 18 months, and the next readmission to hospital for ischaemic heart disease by 37 months, compared with a non-invasive strategy in patients with non-ST-elevation acute coronary syndrome. This remaining lifetime perspective supports that an early invasive treatment strategy should be the preferred option in most patients with non-ST-elevation acute coronary syndrome.” This is how medical knowledge gathering should work. No shortcuts, no surrogates, good long term data reported in a way that informs patients and providers.

Achieved BP in CVD & outcomes
Many have written eloquently on the need for randomised controlled trials rather than observational studies to determine the effects of treatment—none more so than Rory Collins and colleagues in their massive narrative review of statin trials conducted largely by themselves. And I largely agree with them on that issue, as you will see from a separate BMJ blog. But sometimes it is almost impossible to conduct a meaningful RCT of a treatment strategy because of confounding and the multiplicity of agents involved. This is true of blood pressure lowering in patients with cardiovascular disease: here is an area where observational evidence is as good as we’re likely to get. Data were taken from 22 672 patients with stable coronary artery disease enrolled (from 26 Nov 2009, to 30 June 2010) in the CLARIFY registry (including patients from 45 countries) and treated for hypertension.

After a median follow-up of 5.0 years, increased systolic blood pressure of 140 mm Hg or more and diastolic blood pressure of 80 mm Hg or more were each associated with increased risk of cardiovascular events; but systolic blood pressure of less than 120 mm Hg and diastolic blood pressure of less than 70 mm Hg were also associated with adverse cardiovascular outcomes. So here again is the J-shaped curve, which has appeared in almost every study—interventional or observational—except the recently reported SPRINT. No wonder people are sceptical about the methods and findings of that study.

The BMJ  10 Sep 2016  Vol 354
Injuries and cancer diagnosis
This is a fruitful week for Scandinavian data. This time it’s a Swedish study relating to 720 901 patients with a diagnosis of cancer between 1991-2009. The investigators interrogate national databases to determine the risk of injury 16 weeks before and after a cancer diagnosis. Full marks for singling out an intriguing and important question, and for a great analysis. It shows that patients with cancer have highly increased risks of both iatrogenic and non-iatrogenic injuries requiring inpatient care shortly before and after their diagnosis. So here is a major safety issue of which I, for one, was largely unaware.

Alcohol & “fecundability”
My inner pedant, always ready to spring into action, took over completely when I spotted the word “fecundability” in the title of this paper. “Fecund” is obviously a Latin gerund, and no Latinist would ever append the suffix -abilis to a gerund: it would smack of pleonasm. Although this study comes from Denmark, we can make no allowances on that account: the Danish nation has been ornamented by many fine Latin stylists from the time of Frederik II (1559-1588) onwards. I am sure that many readers will have shared these thoughts. But we must overcome our misgivings in order to penetrate to the message of this unfortunately worded communication. The splendidly named “SnartGravid” (Soon Pregnant) study tells us that consumption of less than 14 servings of alcohol per week seemed to have no discernible effect on female fertility. In fact, no appreciable difference in fecundability was observed according to level of consumption of beer and wine. In vino fecundabilitatis, to worsen the Latin.

Plant of the Week: Luma apiculata
The Luma tree of the temperate Chilean Andes is a most beautiful plant and an important source of honey. If you live in a wet part of Britain with mild winters, you can grow it to full height and enjoy its wonderful bark, as well as its fragrant late summer flowers and aromatic autumn fruits.

I think the one we have in our little front garden was sold under the pretext of being a dwarf form. So far it’s a densely branched shrub about a metre high, covered in white tufted flowers with a unique scent. One frosty winter it got cut back quite a bit and looked brown and ugly until pruned, but it sprang back. Currently it is both shapely and vigorous. It’s probably not really hardy north of Nottingham unless you have a favourable microclimate.

I’ve only ever tasted a single luma fruit and it was nice. This year I’m hoping we will have lots: perhaps enough to garnish a small bowl of yoghourt.