Alex Nowbar reviews the latest research from the top medical journals

Proper prevalence study in covid-19

What’s a proper prevalence study? I’d say it’s one that samples a population in an unbiased way. Anand and colleagues tested blood from randomly selected adults receiving dialysis in July 2020 (they had access to this blood from a central lab linked to 1300 dialysis facilities across the US). This gives a good estimate of prevalence of SARS-CoV-2 antibody formation in people receiving dialysis (8%), and only 9% of those with antibodies had had a diagnosis (which could have been due to lack of or deficiency of testing, lack of symptoms, or both). The estimate is less reflective of the prevalence of SARS-CoV-2 infection in the US adult population as a whole but isn’t a totally unreasonable surrogate. The researchers also compared seropositivity rates (essentially infection rates) by neighbourhood, confirming that rates were highest in the US northeast, in Black and Hispanic neighbourhoods, and in areas with the highest population density. This underlines the impact of social interactions on transmission of the virus.

Lancet doi:10.1016/S0140-6736(20)32009-2

Longitudinal antibodies in covid-19

Patel and colleagues tested over 200 healthcare professionals at Vanderbilt University Medical Center twice, 60 days apart. At the first test, 7.6% were seropositive for SARS-CoV-2 antibodies; at the retest, more than half of those who were initially seropositive were found to be seronegative. This is not the first time the reduction in antibodies has been observed, but this study does quantify the issue nicely. These data suggest that people mount antibodies to different extents, and those who mount less of a response become seronegative sooner because their levels weren’t that high to start with. There are several implications of this. First, antibody testing will underestimate the prevalence of immunity, although this study’s sample size is not large enough to predict how much by. Second, the window for donating convalescent plasma (if that were shown to be an effective treatment) is quite narrow.

JAMA doi:10.1001/jama.2020.18796

App-appropriate trial

Randomised trials of smartphone app interventions are good. But Bricker and colleagues go a little further by designing a double-blind one. How do you blind a participant to a smartphone intervention? Answer: use a smartphone intervention in the control group too. In this case the intervention is an app for smoking cessation based on acceptance therapy. This teaches skills for allowing urges to smoke to pass without smoking. The control app is based on avoidance of triggers. Over 2000 people were randomised. Abstinence rates at 30 days (the primary endpoint) were better in the intervention group (28%) compared with the control group (21%), with an odds ratio of 1.49 (95% confidence interval 1.22 to 1.83; P< 0.001). In other words, the acceptance-based intervention was more effective than the avoidance-based one for smoking cessation and would be worth adopting. One of the few limitations is that it does rely on people being smartphone users, which isn’t everyone.

JAMA Intern Med doi:10.1001/jamainternmed.2020.4055

Insulin icodec once a week

One of the medical holy grails is achieving glycaemic control with the least possible burden to people with diabetes. Rosenstock and colleagues randomised people with type 2 diabetes taking metformin to start either a new, once-weekly, long acting insulin (insulin icodec) or the traditionally used insulin glargine (which you may know better as Lantus) in a double-blind fashion titrated by the fasting glucose. For the insulin icodec group, this included daily dummy injections except for the one day a week when participants received the active drug. Insulin icodec seemed safe, but hypoglycaemic events were more common (16% had clinically significant or severe hypoglycaemia versus 9.8% in the insulin glargine group). The primary endpoint was a measure of glycaemic control, HbA1c, and there was no difference between groups at six months. While insulin icodec wasn’t superior clinically, it was superior in terms of needing fewer injections, which could enhance adherence in real life. Sadly, the rate of hypoglycaemic events means that insulin icodec hasn’t quite sealed the deal. I applaud the authors on the trial design though, double-blinding with double-dummy injections was key to getting valid results here.

N Engl J Med doi:10.1056/NEJMoa2022474

Dapagliflozin for chronic kidney disease

Sodium-glucose co-transporter-2 (SGLT2) inhibitors are looking to be the drug class of the 21st century. After stunning results in heart failure, they stun again in chronic kidney disease. Heerspink and colleagues randomised 4304 people with an estimated glomerular filtration rate (eGFR) between 25 and 75 mL/min to dapagliflozin or placebo in a double-blind fashion. Participants didn’t have to have diabetes to enter the trial. The primary endpoint was a composite of a 50% reduction in eGFR, end stage kidney disease, or death from renal or cardiovascular causes. Dapagliflozin reduced these events with a hazard ratio of 0.61 (95% confidence interval 0.51 to 0.72; P<0.001). Death from any cause was reduced. An outright win regardless of diabetes status. The trial was stopped early for benefit, which may have affected the results, but the treatment effect seems so strong that it is unlikely to be an error (especially in light of all the other positive evidence for this drug class).

N Engl J Med doi:10.1056/NEJMoa2024816

Alex Nowbar is a clinical research fellow at Imperial College London