Alex Nowbar reviews the latest research from the top medical journals.

JAMA

Convalescent confusion

This Chinese trial on the effect of convalescent plasma therapy for covid-19 had to be stopped early as it ran out of patients. It is a shame that infrastructure was not available to continue recruitment in areas that still have lots of covid-19 patients, and that the UK, for example, has to start from scratch. But what did the Chinese find in the 103 people with severe covid-19 that they were able to randomise? The primary endpoint was time to clinical improvement in 28 days with improvement defined as discharged alive or a reduction of 2 points on a 6-point disease severity scale. The scale ranged from 1 for discharge and 6 for death. In the convalescent plasma group, clinical improvement within 28 days occurred in 51.9%. In the control group, 43.1% had clinical improvement. This difference was not statistically significant. This is no reason to abandon convalescent plasma as a potential therapeutic avenue because it was likely underpowered due to the early termination. However, I would also say that ideally, this trial would not have been open-label. If it was double-blind it would have ensured that patients were not treated differently in other ways based on knowledge of the allocation arm.

Psychological distress and loneliness

McGinty et al studied a critical issue in their survey of almost 1500 US adults. They assessed levels of psychological distress using the Kessler scale and levels of loneliness by asking “how often do you feel lonely?” with response options always, often, sometimes, rarely, and never. They compared the distress levels with national data from 2018. In 2018, the prevalence of serious psychological distress was 3.9%. In April 2020 it was a whopping 13.6%. The authors note one worrying implication of these findings, that since the Kessler scale is predictive of serious mental illness, the distress during the pandemic could transfer to longer-term psychiatric disorders. This is not outside the realms of possibility especially since the social and economic impact of the pandemic is expected to be felt for years to come. The authors should be commended both for their methodology and for their upfront discussion of its limitations, namely the potential for sampling bias. People might have been more likely to respond to such a survey in April 2020 compared with 2018, therefore these two populations aren’t directly comparable and the 2020 figures could be an overestimate.

Lancet

Observational operating data

The COVIDSurg Collaborative conducted a cohort study spanning 24 countries. They analysed the outcomes of over 1000 patients who underwent surgery in the first 3 months of 2020 and had confirmed SARS-CoV-2 infection (either 7 days before or less than 30 days after). Almost three quarters were emergency surgeries. Half had post-operative pulmonary complications. Almost a quarter of the cohort died within 30 days. Most of the deaths were due to pulmonary complications. Risk factors for death were male gender, older age and higher ASA grade (i.e. comorbidities). These not-even-fully-prospective data are interesting, but there was no control group, let alone randomisation. Therefore this cannot and should not inform the decision whether or not to operate in the presence of the virus. Absence of evidence does not mean we should rely on low quality data. I don’t blame the authors for the lack of control group. Not only is it hard to collect such data, it may not even exist as people without the virus may have avoided presenting with their surgical issue and therefore not been undergoing surgery at this time. However, I do disagree the authors for suggesting that surgery might need to be deferred. It is dangerous. For starters, a lot of these operations would probably not have gone ahead during the pandemic unless their clinicians felt it was absolutely necessary. Who says that death rates would have been lower if surgery had been deferred? Is it even possible to defer “emergency surgery”?

Masks, distancing and more

Has anyone else noticed the surreptitious drift from mask to “face covering” lately? Anyway, let’s talk about masks. In their systematic of 172 observational studies, Chu et al have brought together the evidence on masks, respirators, eye protection and physical distancing for the pandemic viruses, SARS-CoV-2, SARS-CoV, and MERS-CoV. Forty four of the studies were comparative and thus included in their meta-analysis. They found that masks including N95 respirators (check out “Masks unmasked” for a reminder of mask types) and surgical masks were associated with a lower risk of infection compared to no mask. They report that N95 respirators might be more protective than surgical masks and that both types might be more protective than a single-layer mask. They estimate that using a mask could result in 143 fewer infections per 1000 people compared with no mask, 102 fewer by distancing over 1 metre compared to less than 1 metre, and 108 fewer with eye protection. Unfortunately the studies on which these estimates were based were not randomised and several were at high risk of bias. I expect policymakers reading this study will still feel stuck between a rock and a hard place, but the estimates are still informative to an extent.

NEJM

Post exposure prophylaxis with hydroxychloroquine

Boulware et al performed a double-blind randomised controlled trial of hydroxychloroquine in the US and Canada. The aim was to see if, instead of watching and waiting after a person has been exposed to someone with covid-19, could a 5-day course of the drug lower the risk of getting covid-19 if given within 4 days of the exposure? It didn’t. They found no difference in infection rates between those who received the drug compared with placebo. This study had a number of atypical features. For example, recruitment was primarily via social media and patients enrolled themselves online. Secondly the primary endpoint (illness consistent with covid-19 or confirmed covid-19) was often based on self-reported symptoms as PCR testing was not widely available. Thirdly, they randomised 921 people, but only included 821 in the final analysis because the others became symptomatic which rendered them ineligible. However, these features do not invalidate the conclusion that this strategy was not effective.

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

Competing interests: None declared