Richard Lehman reviews the latest research in the top medical journals

NEJM 22 Feb 2018

Rivaroxaban aspires to beat aspirin

Sometimes it doesn’t require much ingenuity to advance medicine. In days gone by, major surgery to the lower limb was the business of orthopaedic surgeons, while physicians (or pathologists) picked up the resulting cases of pulmonary embolism. These classes of men (almost exclusively) did not communicate with one another except via comments in the medical notes. Eventually it became obvious that most of these PEs were preventable. Now a direct oral anticoagulant like rivaroxaban is often used—as in this ASPIRE trial—for five days after knee or hip replacement surgery. But still venous thromboembolism continues to occur after that period—especially after hip operations. The trial compared low-dose aspirin (81mg) with rivaroxaban 10 mg, for a further nine days after total knee replacement and a further month after total hip replacement. There was no difference between groups in proximal vessel deep vein thrombosis or pulmonary embolism. The old cheap drug, first synthesised in 1853, proves equal to the latest targeted treatment.

Thrombectomy for stroke: a bigger window

When I read this report of the DEFUSE-3 trial it began to DAWN on me that it was very similar to another trial of late endovascular treatment for stroke that had appeared in the NEJM. DAWN was a multinational trial of thrombectomy for stroke caused by internal carotid or middle cerebral artery occlusion in patients known to be well 6-24 hours earlier, in whom there was evidence of ischaemic, but viable brain tissue. DEFUSE-3 is a very similar trial where the window is 6-16 hours, and it was modified and then terminated for benefit once the preliminary results of DAWN came through. This was smart work: although I don’t usually like early termination of trials, in this case it makes sense. The pooled data, merely eyeballed, make a strong case for offering thrombectomy for strokes of this kind, where there is evidence of ischaemic, but salvageable brain well after the traditional thrombolysis window.

Neonatal rotavirus vaccine

What I learned from this trial of oral rotavirus vaccination in Indonesia is that around the world, many vaccination schedules start on the day of birth. So giving the anti-diarrhoeal vaccine on day 1 is not in itself a big deal, and it makes obvious sense since the tinier you are when you get diarrhoea, the more likely you are to die from it. There were no deaths in this comparison between neonatal and infant (8 weeks on) vaccination, though there was one case of intussusception in the latter group (following the last dose), consistent with baseline incidence. With a protection rate of 94% in the first year, the vaccine RV3-BB used in this trial seems to be the best yet.

JAMA 20 Feb 2018

Low carbs v low fats: the latest trial

“In this 12-month weight loss diet study, there was no significant difference in weight change between a healthy low-fat diet vs a healthy low-carbohydrate diet, and neither genotype pattern nor baseline insulin secretion was associated with the dietary effects on weight loss. In the context of these two common weight loss diet approaches, neither of the two hypothesized predisposing factors was helpful in identifying which diet was better for whom.” This beautifully clear summary of the DIETFITS trial will please nobody. Carb-haters will argue that they used the wrong fats and that the timescale and the surrogate markers were all wrong; fat haters will no doubt use the same arguments turned on their head. I would be perfectly happy except that they use the word “healthy” to describe these diets. Not working for the marketing department of a supermarket, I don’t know what that means.

Deliriously negative

Very sick people in an acute confusional state use up energy, tear out tubes, look terrified, and generally do themselves a mischief. That is the logic for giving haloperidol prophylactically to those at high risk of delirium who are going to be in an ICU for two days or more. On the other hand, haloperidol is not a harmless drug. Given a dearth of good evidence, it was worth a placebo-controlled randomised trial: and this Dutch one was adequately powered (n=739) and had a primary outcome measure of mortality at 28 days. Among critically ill adults at high risk of delirium, survival (around 80%) was the same in both groups. Is more research needed? I doubt it.  

JAMA Intern Med Feb 2018

Tapering opioids for chronic pain

In the USA just before Trump, many more people died from opioids than from gun violence (64K vs 38K, 2016 figures). It’s hardly likely to have improved. But here is a glimmer of hope. A pain clinic in Stanford offered a slow programme of opioid reduction to patients who had been taking opioids for chronic non-cancer pain for six years or longer. It was based on a high degree of communication between patients and physicians. About a third of patients dropped out, but for those who persisted, results were good: “Our data challenge common notions that patients taking high-dose opioids will fail outpatient opioid tapers or that duration of opioid use predicts taper success. Combining patient education about the benefits of opioid reduction with a plan that reduces opioids more slowly than current tapering algorithms with close clinician follow-up may help patients engage and succeed.” All good: but for the future, could we please use “shared understanding” instead of “patient education”? After all, these patients are educating us in what works for them, not the other way round.

The Lancet 24 Feb 2018

CABG vs PCI: individual patient data

I am a great advocate for complex things that I will never have to do myself, and individual patient data meta-analysis is one of them. In principle, it can reveal subgroup effects that can directly inform patient care, or at least generate strong hypotheses for further testing. So if you are wondering if coronary bypass surgery still has a place in the management of coronary artery disease, here is the result of a pooled analysis of individual patient data from 11 randomised trials involving 11 518 patients: “CABG had a mortality benefit over PCI in patients with multivessel disease, particularly those with diabetes and higher coronary complexity. No benefit for CABG over PCI was seen in patients with left main disease. Longer follow-up is needed to better define mortality differences between the revascularisation strategies.”

Those antidepressants

I quite like network meta-analyses too, though they rarely provide a good match for clinical practice. Every doctor who has tried to treat depressed patients knows that it is a matter of offering—and delivering—personal support and understanding, sometimes coupled with a trial of antidepressant drugs. This can go on for many months, or years, and if the patient improves we tend to assume that they are on the “right” drug for them individually. This individual trial-and-error approach cannot be mapped using traditional randomised trial methodology. The main message in the press about this massive review is that the drugs all work to some extent. It tells us little about the variation in individual responses both to the 21 drugs at initiation, and to withdrawal from serotonin reuptake inhibitors in particular: a massive problem affecting millions of people. The infographic summarizing the network results has become an instant classic. It’s the perfect illustration of Samuel Johnson’s famous definition of a network: “Anything reticulated or decussated, at equal distances, with interstices between the intersections.” Dictionary of the English Language (1755).

The BMJ 24 Feb 2018

Maternal thyroid function and kids’ outcomes

Lots of British rivers are called “Avon” for the simple reason that “avon” is the British name for river. The Avon Longitudinal Study of Parents and Children is a prospective birth cohort study set up in 1991, at the heyday of such studies: it is named after the gorgeous river which flows through Bristol. Avon ladies who bore children had their thyroid function tested in the first trimester: 4.3% were newly identified as having hypothyroidism or subclinical hypothyroidism and 2.1% subclinical hyperthyroidism or hyperthyroidism. These Avonian children are now grown up, and their educational results can be tracked. They don’t differ in anyway from those of kids born to euthyroid mothers.

Non-invasive coronary testing

Here is another meta-analysis reticulated or decussated, at equal distances, with interstices between the intersections. This network meta-analysis deals with “diagnostic randomised controlled trials comparing non-invasive diagnostic modalities in patients presenting with symptoms suggestive of low risk acute coronary syndrome or stable coronary artery disease.” Am I succeeding in putting you off? This is not very fair to the authors, who have done a good job within the confines of the material at their disposal. But I have deep reservations about all meta-analyses of diagnostic tests, and even more about indirect comparisons which are dependent on population characteristics and clinicians applying Bayesian logic. Since these points would need many pages to explain, and I would do it badly, I hereby dismiss this week’s class. Go out and play in the snow or the sunshine. Wrap up warm. Follow Stephen Senn on Twitter for the perfect mixture of winter walks and Bayesian enlightenment.  

Plant of the Week: Sarcococca ruscifolia

The little evergreen shrubs called “Christmas boxes” do indeed belong to the box family, though they rarely flower at Christmas. Instead, they lurk unobtrusively amid the mud, frost, and snow of January until they begin to call attention to themselves by their wafts of honey scent in February.

These are amongst the very best plants for dry shade, and they don’t mind being amongst tree and shrub roots. In a way they are “plant and forget” ground-coverers, but on the other hand their sweetness is wasted unless you have them somewhere close to where you can lean over and sniff.

The flowers are just little tassels of stamens, and if they happen to be female they will produce red berries later. That’s the case with ruscifolia, whereas the one we grow, S hookeriana “Digyna” is male and has no berries, despite its name. Any of these plants can be lifted and divided to produce many more, though they are pretty slow to get going.