Ann Robinson reviews the latest research from the top medical journals

Hydroxychloroquine for covid-19. Time to give it up?

Could hydroxychloroquine be a useful prophylaxis for hospital based healthcare workers who care for patients with covid-19? Evidence suggests that, contrary to President Trump’s endorsement, hydroxychloroquine is not effective at treating people once they’ve been infected. A recent randomised study showed no clinical benefit when a five day course was given to people who had been exposed to the virus. So it beats me why it would work as a pre-exposure prophylactic—and indeed it doesn’t. This small (132 participants) but well designed trial of daily hydroxychloroquine for eight weeks was stopped early as there was no significant difference in positive antigen tests between exposed individuals given hydroxychloroquine or placebo (6.3% v 6.6%). The healthcare workers were generally young (median age 33 years) and female (69%), so results can’t be extrapolated to the population at large. Hydroxychloroquine was well tolerated except for an increased incidence of diarrhoea. The authors say that the study may have been underpowered to detect a clinically important difference; or perhaps it’s just time to accept that it doesn’t work.

JAMA Intern Med doi:10.1001/jamainternmed.2020.6319

ECGs are hard to interpret—even for cardiologists

A meta-analysis of 78 studies showed that medical students and doctors at all training levels find interpreting electrocardiograms (ECGs) difficult, even after specific teaching. Median pooled accuracy was 54% before training and 67% after training; medical students scored 42% pre-training, and cardiologists scored 74.9%, so even the experts got it wrong in a quarter of cases. The analysis included data collected over many years, and the tests used to assess accuracy were highly variable and often not very good. But there’s no doubt that ECGs are hard to interpret accurately, and other studies have shown that even automated computer interpretations are fallible. The interpretation of ECGs by computers may improve as AI driven algorithms are finessed by machine learning and human performance may benefit from new guidelines and training methods, say the authors optimistically.

JAMA Intern Med doi:10.1001/jamainternmed.2020.3989

No magic bullet for preventing wound infection after a caesarean

Negative pressure wound therapy (NPWT) seals the caesarean section wound with a dressing attached to a pump, which draws out excess fluid and bacterial contamination. Theoretically, it should reduce the rate of infection, but this multicentre randomised trial comparing NPWT with standard wound dressing found no significant difference in wound infection (3.6% v 3.4%), other wound complications, or pain. Adverse effects were low, but NPWT caused significantly higher rates of skin reactions. The unusually low infection rate is itself a limitation of this study; predicted risk was 10% and the lower rate in this study probably reflects good surgical technique and infection control. Patient satisfaction was high in both groups, and significantly higher in the group that had NPWT at discharge but not by one month later. There’s not enough evidence from this trial to support the extra intervention and expense of using NPWT after caesareans, even in women at higher risk of infection.

JAMA doi:10.1001/jama.2020.13361

Hope in non-small cell lung cancer

Treatment of metastatic non-small cell lung cancer (NSCLC) has improved, but the outlook is still grim, with five year survival rates of around 6%. This global, randomised, phase 3 trial showed that primary treatment with atezolizumab (a monoclonal antibody that reactivates the immune response to cancer cells by binding to the programmed death-ligand 1 (PD-L1)) extended survival by a median 7.1 months in patients with high PD-L1 expression compared with standard platinum-based chemotherapy. In the subgroup with the highest PD-L1 expression, median survival was 20.2 months with atezolizumab versus 13.1 months with chemotherapy, regardless of histological type. Adverse events were very high in both groups (>90%), with more severe (grade 3 or 4) events in the group given chemotherapy (52.5%) than the atezolizumab group (30.1%).

N Engl J Med doi:10.1056/NEJMoa1917346

No place for azithromycin for severe covid-19?

This small Brazilian study showed that adding the antibiotic azithromycin to standard care (which included hydroxychloroquine) did not improve clinical outcomes (or adverse events) in patients with severe covid-19. This is the first randomised clinical trial looking at patient centred outcomes when azithromycin was added to standard care among patients admitted to hospital with severe symptoms. Disappointingly, it didn’t work. Adding azithromycin didn’t improve clinical status at 15 days (measured by the need for hospitalisation, oxygen, or ventilation) or mortality at 29 days after randomisation. This study is problematic: there was no control group of patients who weren’t given either azithromycin or hydroxychloroquine; the patients had severe symptoms, so results can’t be extrapolated to those who are less ill in the community; and the open label design may have introduced bias. The results were clear enough, though: so far, there seems to be no good evidence for using azithromycin to treat severely ill people with covid-19.

Lancet doi:10.1016/S0140-6736(20)31862-6

Ann Robinson is an NHS GP and health writer and broadcaster.