Alex Nowbar’s weekly review—1 November 2019

Alex Nowbar reviews the latest research from the top medical journals


Which first line antihypertensive?

Finally, a broadly applicable study to write about. LEGEND-HTN is a large scale, observational analysis using nine databases to identify the class of antihypertensive with the best cardiovascular outcomes when used singly. Thiazides and thiazide-like diuretics were the best, but you can be reassured ACE inhibitors, angiotensin receptor blockers, and calcium channel blockers weren’t far behind. The advantages of this study are its size and its real-world nature, as both these factors increase the generalisability of the findings. The disadvantages are that the data aren’t randomised, and so are subject to confounding. For example, people who tend to be prescribed thiazides could be more likely to have better outcomes for other reasons like lower baseline blood pressure, better adherence or fewer comorbidities. Even with propensity score modelling, these data will never replace randomised controlled trials for evidence based decision making. 

Deep or light anaesthesia by bispectral index

Bispectral index (BIS) is an EEG-based measure of consciousness (or “depth” of anaesthesia) ranging from 0 to 100. The Balanced Anaesthesia Study was an international, blinded, randomised trial of 6644 patients comparing two anaesthesia levels (BIS 35 to BIS 50) to define whether deep anaesthesia is linked to post-operative mortality (as suggested by observational data). They studied a high-risk population—people over 60 with comorbidities (ASA 3 or 4)—undergoing major surgery lasting at least two hours and anticipated to be in hospital for at least two days afterwards. The primary outcome was all-cause mortality at one year and there was no difference between the two groups. Recovery from anaesthetic was also similar in both groups. While this result is generalisable to many patients, it only applies to volatile anaesthetic use (not IV propofol which is a commonly used agent). The idea of balancing the anaesthetic to the patient is laudable and it is a shame that one BIS threshold wasn’t better than the other. It is difficult to know how useful BIS really is (it’s hard to validate a technology about consciousness level) and whether there is an optimal BIS. Anaesthesia is no less of a dark art than before this study unfortunately.

Breast cancer surgery outcomes with different types of anaesthesia

Sessler et al compared regional anaesthesia (paravertebral blocks and propofol) with general anaesthesia (sevoflurane and opioid analgesia) for primary breast cancer resections in an international randomised controlled trial of 2132 women. The primary outcome was cancer recurrence, and the secondary outcome was persistent incisional pain. The idea that the type of anaesthesia would affect cancer recurrence is biologically appealing, but it seems implausible to me. Indeed, this study found a 10% rate of recurrence in both groups. The secondary hypothesis that the anaesthesia would affect postoperative pain seems far more plausible, but again there was no difference between groups. This trial is clinically useful because it provides the evidence to support whichever option the operating team prefer. Unfortunately, to reduce cancer recurrence and incisional pain, researchers may need to look at other mechanisms.  

Annals of Internal Medicine

Who to screen for lung cancer

Screening has an automatic appeal because it implies preventing cancer deaths. To that end, you might design a screening programme aimed at the people most likely to get cancer and die. Is this an oversimplification though? These “high-risk” people are also those at high risk for death from other causes. Screening for lung cancer may not lengthen their life when they are likely to die from something else first. 

Instead of risk-based selection for screening, Cheung et al propose using a life-gained strategy. This study compares the outcomes of the two strategies for lung cancer CT screening using survey-based modelling in people aged 40 to 84 who smoke or have smoked. They found that the life-gained strategy maximised the life-years gained from screening in the population, but prevented fewer lung cancer deaths than the risk-based strategy.

If the life-gained strategy is the best way to select people, is denying screening to high risk people palettable? It feels like pigeon-holing people based on their risk profile (previous smoking, co-morbidities etc) and saying you may be quite likely to have cancer, but let’s not look for it because we won’t be able to extend your life as much as someone with a longer life expectancy. Deciding who to screen is a tricky area because no strategy is perfect and no screening is without harm.


Fluoxetine for obsessive compulsive behaviours in autism

Reddihough et al performed a blinded randomised controlled trial of fluoxetine in children and adolescents with autism at three centres in Australia. The severity of the obsessive compulsive behaviours was reduced in the fluoxetine group compared with placebo, but this became null in the prespecified adjusted analysis. Treatment discontinuation rates were very high and no primary outcome data were available for 25% of participants, which make it difficult to know what value to place on the results. The researchers should be commended, though, because a trial in this population has several challenges: for example, reluctance to be randomised to placebo when the family could access fluoxetine elsewhere, drop out due to perceived failure of treatment, and adverse events being (probably wrongly) attributed to treatment. Meanwhile, the role of medication in this population remains unclear—which is no surprise really because I suspect there is little room for improvement without psychosocial intervention.

JAMA Internal Medicine

Reducing post-operative delirium

The Tailored Family-Involved Hospital Elder Life Program sounds boring but, by jove, was it effective! Also known as t-HELP (which coincidentally is the kind of help I get from my partner when I’m writing – cups of tea), the program was tested against usual care in a cluster randomised trial in a Chinese hospital involving 281 participants. The randomisation units were six surgical floors rather than individual patients. The Confusion Assessment Method was used to assess for the primary outcome, post-operative delirium. The effect of t-HELP was incredible with a delirium rate of 2.6% in those who received t-HELP compared to 19.4% in the control group. Functional and cognitive decline at 30 days was also less in the t-HELP group. It is great to have this supporting evidence for what is already being done (to a greater or lesser extent) in the NHS. In fact, the intervention contained several components from the NICE guidelines. What makes these results a little hard to believe is not the large reduction in postoperative delirium (that could be attributed to the intervention), but that the incidence of delirium could be as low as 2.6% in the intervention group. This is remarkable. It makes me wonder if they excluded the highest risk patients which does then make this result less generalisable.


β blockade in COPD

β receptor agonists open the airways. β blockers slow the heart, reduce blood pressure, and in theory make the airways less open. This theory has frightened us into “caution” when prescribing β blockers in people with chronic obstructive pulmonary disease. The downside of this is that people with COPD often have a prognostic indication for a β blocker, such as heart failure (because of a common risk factor, smoking). But one day, it was proposed that β blockers might be protective in COPD. The BLOCK COPD trial randomised people with COPD, moderate airflow limitation, and increased risk of exacerbations (that is, had had an exacerbation in the previous year or were prescribed supplemental oxygen) to extended release metoprolol or placebo in a double blind fashion (although blinding was probably ineffective because of inevitable reduction in heart rate and blood pressure). There were higher rates of exacerbation requiring hospitalisation in those who received metoprolol, although the time to first exacerbation was no different between groups. The trial was terminated due to futility and concerns regarding safety, which seems appropriate but does make the results difficult to interpret. But there were no objective signs of worsening lung function in those who received metoprolol. For now, better go back to the “caution” approach, whatever that means.

Antiplatelets in acute coronary syndrome

Aspirin is the cart and clopidogrel was the donkey of dual antiplatelet therapy for acute coronary syndrome. Prasugrel was the pretty pony that replaced the donkey with better results than clopidogrel when PCI was planned. Then ticagrelor was hailed as the horse to pull the cart into lower event rates. And so the cardiologists duly got on the ticagrelor bandwagon. Now to the horse-seller’s displeasure, the pony has proved better than the horse in the ISAR-REACT 5 trial which randomised over 4000 people with acute coronary syndrome planned for invasive evaluation to either ticagrelor or prasugrel. There was a marked reduction in death, stroke and myocardial infarction at 1 year with prasugrel (6.9%) compared to ticagrelor (9.3%) with similar bleeding rates. The early and persistent difference in outcomes in the 2 arms is quite convincing.  

Surgical treatment for heartburn

Spechler et al have performed a smart trial of people with proton pump inhibitor refractory heartburn. The beauty of the trial design was the careful assessment of patients before randomisation, including endoscopy, manometry, and pH monitoring to get objective confirmation of reflux and a trial of omeprazole 20 mg twice daily to get confirmation of symptoms despite proton pump inhibitor treatment. These tests led to many patients being excluded, but the remaining 78 patients were randomised to one of three groups: laparoscopic Nissen fundoplication, omeprazole and baclofen with or without desipramine, or omeprazole plus placebo. The primary endpoint was achievement of a 50% reduction in reflux-related quality of life score at 1 year. This was achieved in 67% in the surgery group, 28% in the omeprazole and baclofen group, and 12% in the omeprazole and placebo group. This demonstrates the effectiveness of surgical treatment in this highly selected but clearly defined population. However, without sham surgery it is not possible to know how much the placebo effect of a procedure contributed to the effect size.

Alex Nowbar is a clinical research fellow at Imperial College London