Alex Nowbar’s journal reviews—29 May 2019

Alex Nowbar reviews the latest research from the top medical journals

Annals of Internal Medicine

Operative infection risk in rheumatoid drugs

A group of US researchers assessed infection rates after hip or knee surgery in patients with rheumatoid arthritis who were taking biologic therapies. The retrospectively collected, pharma funded, observational data tell us the risk for each of six biologic drugs and concludes that no one biologic has a particularly different risk compared with the others. The authors also report the infection outcomes on different steroid doses (worse on higher doses, of course). The introduction states: “Understanding risk for postoperative infection with different immunosuppressive therapies is important to optimizing perioperative management.” Nice try. This study design cannot help with this because it isn’t a randomised controlled trial and gives no indication of whether to stop biologic treatment before surgery and, if so, when to stop and when to restart.


Groupiness in Sepsis

Seymour et al define four clinical phenotypes for sepsis which they have named alpha, beta, gamma and delta. They derived these from retrospective data from 16,552 unique patients in hospitals in Pennsylvania, USA, using unsupervised machine learning. They used data from a further 31,160 patients and 4737 patients from randomised controlled trials in sepsis to validate the phenotypes with clinical outcomes including mortality.

Patients included were those with suspected infection and organ dysfunction using the SOFA score, both within the first 6 hours of admission. The worst mortality was seen in the delta group. The delta group also featured liver and septic shock. The least mortality was seen in the alpha group, characterised by fewer laboratory abnormalities and less septic shock. This was also the most common group.

Humans like classification even if it’s pointless, hence the hype around this study. The groupings have no implications for treatment so currently will only be useful for patient selection for researching interventions. The differences between groups are an inkling that the best treatment may differ depending on the phenotype. If so, and if this treatment was identified, this could prove to be a pivotal piece of work.

Recombinant human soluble thrombomodulin in sepsis

This randomised double-blind trial of 800 people in intensive care looked at whether the administration of a recombinant human soluble thrombomodulin reduced the mortality of critically ill patients with sepsis-associated coagulopathy. Sepsis-associated coagulopathy was defined as prolonged international normalised ratio (INR) >1.40 and low platelets. Thrombomodulin was thought to have some impact on pathways to inflammation that may mediate organ failure in sepsis rather than just affecting coagulation pathways, but it is no surprise to me that this treatment did not improve 28 day mortality. A limitation of this trial was that many patients did not have a prolonged INR at the time of randomisation, even though they did at screening, which suggests they might already have been getting better. This reduces the ability of the study to show a difference in mortality. Patients also received heparin, which has overlapping effects with the thrombomodulin, again limiting the ability of this study to show a mortality benefit.

This whole area is very challenging because by the time someone with sepsis is sick enough to be in intensive care, it may be too late for an intervention to make a difference above current standards of care. We need to predict who, amongst people with infection, is going to do badly and target them with interventions. This is why Seymour et al’s phenotypic work could be so critical.

For thrombomodulin, there is a small chance that a subgroup of patients might benefit but this may never be identified. Anyhow, recruitment for a trial of such patients would be very difficult given that the broader SCARLET trial required 159 sites in 26 countries.


Stroke arm therapy robot on the rocks

Interventions to improve function after disabling stroke beyond current therapy regimens are a bit of a holy grail. The much awaited three-arm RATULS trial is unfortunately negative. 770 people were randomised to robot-assisted training, enhanced upper limb therapy, or usual care. Upper limb function tended to increase in all groups, but there was no improvement with the interventions compared with usual care. Intuitively, more therapy should equate to more functional improvement, so you might be surprised by this result, but there are several possible explanations. Firstly, the interventions were mostly started months after the stroke, so if there was potential for function to improve, the gains might already have been made, although the authors found no evidence in subgroup analysis that time since stroke affected the outcome. Secondly, the primary end point was “upper limb function success,” which is a high bar to achieve. This outcome was based on the action research arm test, which tests various aspects of upper limb function. Lastly, and most likely, perhaps people engage better with therapists than with equipment.

Some alteplase answers?

Stroke is a particularly emotive medical condition, with life-changing treatment available. Making sure that we deliver treatment appropriately is much easier said than done, as the treatment may do more harm than good if delivered too late or to someone whose brain tissue is not salvageable. This meta-analysis, looking at thrombolysis given from 4.5 to 9 hours after stroke (or for wake-up stroke), is a good start, even if it includes only 414 patients from three randomised controlled trials. Campbell et al found that more patients had an excellent functional outcome with alteplase than with placebo, but the alteplase group also had a higher rate of symptomatic intracerebral haemorrhage and a higher mortality rate (although the increase in deaths was not statistically significant). Thus, the authors conclude that the risks did not negate the overall net benefit of thrombolysis. Just because they write this, does not make it true, though. If alteplase delivers an increase in achievement of good functional outcome from 29% to 36% with an intracerebral haemorrhage rate increased from 0.5% to 5% and death rate from 9% to 14%, is it worth the risk? Answer: there are definitely no definites here.

Antiplatelets after intracerebral haemorrhage

The RESTART trial randomised people with previous spontaneous intracerebral haemorrhage to either starting or avoiding antiplatelet therapy. They had to have been taking antiplatelets or antithrombotics before the haemorrhage and had the drug stopped afterwards. It must have been a hard sell asking people to take part in this trial. It is difficult to admit that we honestly didn’t know which is the safer option. Fortunately, we now have some evidence to help make the right recommendation. This 537 participant randomised controlled trial showed that it is safe to start the antithrombotic therapy; there was no increase in risk of recurrent intracerebral haemorrhage. In fact, there was even a suggestion it might be better to start it than to avoid it. I think the open-label nature of the trial is a serious disadvantage in the methodology because knowledge of the treatment arm will have influenced other behaviour that could have affected the endpoint, even though the endpoint assessment was blinded. When designing trials, we should always be asking, “Why can’t this be double-blinded?” and the reason had better be good.


Treatment for mild asthma

As-needed use of salbutamol is a common form of asthma treatment but it is widely perceived that preventer (steroid-containing) therapy is a better option. But what about using the steroid-containing therapy as the reliever treatment? Astra-Zeneca tried this in the SYGMA trial published last year and there was some promise when they compared the steroid-containing reliever to turbutaline in a double-blinded trial. Now, the Novel-START trial compared three treatment strategies, as-needed salbutamol, as-needed budesonide-formeterol, and regular budesonide-formeterol with as-need salbutamol. The latter two strategies had much fewer exacerbations and the as-needed budesonide-formeterol was as good as the maintenance strateegy. This is a good result in theory because using steroids as a reliever inhaler means reduced steroid intake compared to maintenance steroids and yet the outcomes are similar. However, this study was not double-blinded. The primary outcome was rate of exacerbations which is susceptible to bias. Patients and clinicians need to know the benefits beyond placebo. Otherwise, we might be undermining adherence to preventative therapy for no good reason.


Length of steroids for nephrotic syndrome

Webb et al compared an 8-week course of prednisolone with a 16-week course in 237 UK children with steroid-sensitive nephrotic syndrome in a double-blinded fashion and followed them up for 2 years. The primary endpoint was time to first relapse. This was no different between the 8 week and 16 week groups. Reassuringly the adverse effects weren’t worse in the 16 week group though. The authors report better quality of life with the 16 week group which is important but I do not think justifies giving additional steroids. Nor does the estimate of significant cost savings. Clinical outcomes were no better with the 16-week course so even performing an economic evaluation is a little daft.

Why education provides cardiovascular protection

Why does education provide cardiovascular protection? A good answer is that the protective effects of education are mediated by lower BMI, lower blood pressure and less smoking which are associated with higher education levels. If it’s more complex than that, I’m not sure I really want to know but here goes.

Carter et al derived educational attainment from over a million individuals of European ancestry and used the UK Biobank to do a mendelian randomisation analysis. The study confirmed the protective effect of education on cardiovascular outcomes and that BMI, blood pressure and smoking increased risk. The key new finding was that these three risk factors only contribute 40% of the increased cardiovascular risk seen with low education. What contributes the other 60%? Perhaps psychosocial stress, diet, exercise? Stay in school, kids. Not sure how it helps but it does.

Alex Nowbar is a clinical research fellow at Imperial College London.