Richard Lehman’s journal review—11 January 2016

richard_lehmanNEJM 7 Jan 2016 Vol 374
Predicting next week’s PET
13 This week begins with a toughie. Diagnostic test studies are always tricky to analyse. What matters to you as a clinician is the downstream value of the test—how it will help your management and improve the outcomes of patients. In the PROGNOSIS study a combination of tests was used to predict the short term onset of pre-eclamptic toxaemia in women past the 24th week of gestation who were suspected of being about to get PET. The primary outcome measure was whether they did get PET. And in this study, designed and closely supervised by Roche Diagnostics, the combination test shows pretty useful predictive value for the development of PET in the coming week: it is less good for predicting PET thereafter. But what you as a clinician need to know is if this is any better than sending the patient home to rest and measure her BP and urine protein daily. And if these go up and stay up, you have a diagnosis of “PET” and therefore get her to hospital and deliver the baby when you think safest. So where does the Roche test fit into routine practice? It is, by the way, the ratio of soluble fms-like tyrosine kinase 1 (sFlt-1) to placental growth factor (PlGF). Personally, I can’t see that it has any potential to improve the outcomes of real life PET. And, in fact, you can’t tell from this study because participating clinicians were blinded to the test results. There are data about neonatal and maternal outcomes in the supplements but it would take a day to try and back-calculate from them. And if you wanted to delve further, it might be difficult, because “the research contract between the sponsor and the institutions participating in the study included a confidentiality agreement.”

Antimalarials for Ebola
23 The Ebola epidemic is over and in a properly run world there should never be another one. But if there is, we now know a bit more about the disease and its management. In 2014, 382 patients with confirmed EVD were admitted to the Ebola treatment centre in Foya, Liberia. At admission, 194 patients were prescribed artemether–lumefantrine and 71 were prescribed artesunate–amodiaquine. This wasn’t intentional: they simply ran out of the first drug. But this natural experiment showed that patients who were prescribed artesunate–amodiaquine had a lower risk of death from EVD than did patients who were prescribed artemether–lumefantrine, despite having similar characteristics.

Convalescent plasma for Ebola fails
33 During the Ebola outbreak in Guinea, 99 patients were given two infusions of plasma from recovering Ebola patients. In this non-randomised case-series, they fared no better than those who received no plasma.

Globulin for graft v host disease
43 The NEJM has printed a lot of strangely worded abstract conclusions so far this year. Here is a typical example: “The inclusion of ATG resulted in a significantly lower rate of chronic GVHD after allogeneic transplantation than the rate without ATG. The survival rate was similar in the two groups, but the rate of a composite end point of chronic GVHD–free survival and relapse-free survival was higher with ATG.” The plain English version would read: “In this two year, open-label trial, our company’s antihuman T-lymphocyte immune globulin (ATG) halved the rate of chronic graft-versus-host disease (GVHD) in people who had previously undergone allogeneic haematopoietic stem-cell transplantation, but made no difference to mortality.” I guess these conclusion sections are the most contested bits of any paper, especially when sales may depend on them. I see this example as evidence of the NEJM taking pains to ensure that only the prespecified primary outcome measures appear in this section—hence the upside-downiness of the language. All very commendable, and from now on Ben Goldacre’s COMpare team will be keeping a close eye on the reporting of non-prespecified outcomes. See Annals below.

JAMA 5 Jan 2016 Vol 315
Too late for diet & exercise
36 When people over 65 are seriously obese—the mean BMI here was 39.3—they often become breathless on exertion and develop a degree of ankle oedema (especially if they are women), and they may even have audible crackles at their lung bases. This is one of the varieties of HFPEF—so called heart failure with preserved ejection fraction. There are various confirmatory tests, including BNP and a restrictive pattern of diastolic filling, but there is still a lot of diagnostic fuzziness. Here’s a trial of calorie restriction and/or exercise in 100 such patients (81 of whom were women). The interventions made no difference to symptom scores but improved the peak amount of oxygen used during exercise (peak V̇o2). Sometimes it’s too late for half measures: maybe a trial of bariatric surgery would be a better idea. I’m not joking: the mean age of these people was 67, and if they really have HEFPF then it is as life shortening as many cancers that you wouldn’t hesitate to operate on.

Genomic alarms & arrhythmias
47 Two studies in this week’s JAMA provide a contrast between the new genomics with the old genetics. Genotyping is widely touted as a way to discover rare variants which carry substantial risks for individuals, the most serious of which is dropping down dead. Now most people do not drop down dead, so if these variants turn out to be common, something has to be wrong. Dropping dead may be due to a cardiac arrhythmia: often it’s impossible to tell. But associational studies have identified a number of allegedly arrhythmogenic genetic variants. So if you send your blood to a lab offering genome sequencing for arrhythmia risk, what will happen? In this study, 2022 samples were sent to three such labs. “There was low concordance in designating SCN5A and KCNH2 variants as pathogenic. In an unselected population, the putatively pathogenic genetic variants were not associated with an abnormal phenotype. These findings raise questions about the implications of notifying patients of incidental genetic findings.” Yes indeed. Just don’t go there.

How much cancer is written in genes?
68 Now for the traditional approach: twin studies from Scandinavian public registries. They looked at 80 309 monozygotic and 123 382 same sex dizygotic twin individuals: wonderful stuff. “For most cancer types, there were significant familial risks and the cumulative risks were higher in monozygotic than dizygotic twins. Heritability of cancer overall was 33%. Significant heritability was observed for the cancer types of skin melanoma (58%), prostate (57%),nonmelanoma skin (43%), ovary (39%), kidney (38%), breast (31%), corpus uteri (27%).”

Anns Intern Med 5 Jan 2015 Vol 164
Outcome switching: editors know best?
OL The most interesting things on the Annals website at the moment are a letter from Ben Goldacre and the COMpare team and a reply from the Annals editors. These are open access and are the first shots fired in what will no doubt be a long and interesting battle over what counts as good science and good reporting in the outcomes of clinical trials. The first phase of COMpare looks at how the outcomes reported in journal articles relate to the outcomes prespecified when the trial was designed. The primary purpose of an interventional trial must always be to investigate whether the intervention makes a significant difference to some predefined outcome of interest. These outcomes should be clearly stated and be of importance to clinical practice. The more of them you specify, or the more you merge together into a composite category, the more likely it is that you will find a spurious association. And if at the end of the trial, you discover that the intervention has done something you didn’t expect, then you cannot claim causality: for that, you need to repeat the trial looking for that outcome as its primary purpose. This is the way medical science has to work in order to get to a 19 out of 20 chance of doing the right thing for the greatest number of patients. There are no shortcuts. In their letter, Ben and colleagues point out that in one trial report “One prespecified primary outcome was reported incorrectly as a secondary outcome. In addition, the article reports 5 ‘primary outcomes’ and 9 secondary outcomes that were not prespecified without flagging them as such.” They ask for a correction. But the editors question the basis of the criticism on the grounds that prespecification often doesn’t matter. “On the basis of our long experience reviewing research articles, we have learned that prespecified outcomes or analytic methods can be suboptimal or wrong. Regardless of prespecification, we sometimes require the published article to improve on the prespecified methods or not emphasize an end point that misrepresents the health effect of an intervention.” I think they are plain wrong.

Lancet 9 Jan 2016 Vol 387
Totalled: thrombus suction for STEMI
127 The results of the TOTAL trial appeared on the Lancet website three months ago. If you are an interventional cardiologist, you will know by now that this demonstrated that routine thrombus aspiration for ST elevation myocardial infarction does not improve outcomes and may lead to more strokes.

Ranolazine flops
136 Interventional cardiologists will also know that ranolazine was tested in a trial published electronically at the same time. This orally available drug is a piperazine derivative that is supposed to reduce intracellular calcium overload during ischaemia by blocking the late sodium channel. The patients recruited had a history of chronic angina and angiographic evidence of incomplete revascularisation after percutaneous coronary intervention. Giving half of them ranolazine made no difference to the composite rate of ischaemia-driven revascularisation or hospitalisation without revascularisation.

Three types of IBD
156 I became a doctor 40 years ago this month. During that time there have been some notable advances in knowledge and treatment, but sadly not many of these relate to disabling chronic inflammatory diseases, such as multiple sclerosis or chronic inflammatory bowel disease. Here, a magnificent genetic association study questions the way we categorise the latter. “After quality control, the primary analysis included 29 838 patients (16 902 with Crohn’s disease, 12 597 with ulcerative colitis). Three loci (NOD2, MHC, and MST1 3p21) were associated with subphenotypes of inflammatory bowel disease, mainly disease location (essentially fixed over time; median follow-up of 10.5 years). Little or no genetic association with disease behaviour (which changed dramatically over time) remained after conditioning on disease location and age at onset.” Mull over that for a minute, and then try to reconcile it with this: “Our data support a continuum of disorders within inflammatory bowel disease, much better explained by three groups (ileal Crohn’s disease, colonic Crohn’s disease, and ulcerative colitis) than by Crohn’s disease and ulcerative colitis as currently defined. Disease location is an intrinsic aspect of a patient’s disease, in part genetically determined, and the major driver to changes in disease behaviour over time.” What they are saying is that the location of disease is partly genetically determined and that this governs the progression of disease. The wording that continues to fox me is “conditioning on”: does that mean “adjusted for”?

The BMJ 9 Jan 2016 Vol 352
Cancer screening & lives saved
I get quite lost reading The BMJ magazine these days, and what is worse, I often end up reading stuff I have written. However, there are many compensations. This week the chief one is an article by Vinay Prasad and colleagues about the need to judge cancer screening by the criterion of reduced all-cause mortality. In “Why cancer screening has never been shown to ‘save lives’—and what we can do about it” he says outright what most have only hinted at. But the narrative of “my life was saved by early detection” is so much easier to sell than the message “this test may alter what you die from, but overall it won’t alter the fact of your death.” There is an excellent editorial from Gerd Gigerenzer showing how the facts might be presented to people so they can make their own decisions.

Cystitis: take ibuprofen & come back?
The BMJ as distributed to superannuated British GPs in the sticks has a tiny vestigial original research section. This week’s offerings come from Denmark, Mexico, and Germany. I mourn the fact that my working lifetime has seen the complete extinction of the hobbyist researcher: a clinician deeply interested in filling knowledge gaps that matter to patients, by looking at her or his local community. The BMJ used to encourage this with a “Short Reports” section, where I once appeared, reporting how levels of BNP correlated with a diagnostic label of heart failure in general practice. The most interesting study this week was a German trial of ibuprofen versus fosfomycin for treating women with uncomplicated cystitis. In the ibuprofen group, symptoms persisted for longer and there were more cases of pyelonephritis, but the rate of antibiotic usage was two thirds less. This is interesting, but hard to evaluate in the context of British practice, not least because fosfomycin is not available here. Some keen GP should design a real life trial based on the local guideline, which is probably to use trimethoprim or nitrofurantoin. In the real world, you would have to test that against open label ibuprofen, using a shared decision aid. But good luck finding a funder, and don’t expect any major journal to publish what you discover.

Plant of the Week: Helleborus x hybridus “Harvington Yellow Redspot”

In our garden, we haven’t yet seen any snowdrops, let alone daffodils, but our hellebores are all in flower a couple of months early. This is a sovereign antidote for seasonal affective disorder, and the yellow Harvington hybrids are the cheeriest of all. They glow beautifully on the dullest of days, with a yellow that is clearer than a primrose’s: subtly greenish as they emerge, becoming pure and radiant when the petals are fully expanded.

You should try and grow these treasures on a raised bed where you can admire them at close quarters from the moment their nodding flowers begin to open. You cannot cut them and bring them indoors because they will fall apart after a day or two. Outside, they will gladden your eyes for weeks.