The parade of COVID-19 papers is relentless enough (fluvoxamine, anyone?) it’s almost impossible to try and keep up, but this one is a little different. This trial, rather than just another inflammatory mediator striving for its day in the sun, tries to inform something we can all do every day: wear a mask.
So, this is the mask trial everyone’s been talking about – in which 4,862 patients completed a study where the intervention arm was given 50 masks and asked to wear them outside the home. The primary outcome at 1 month was SARS-CoV-2 infection, as measured by seroconversion, PCR, or hospital clinical diagnosis.
At the end of the day, this was a “negative” trial – there was no statistically significant difference in SARS-CoV-2 infection between the mask (1.8%) and control (2.1%) participants. C’est la vie.
While the findings of this trial are being covered by news outlets in a variety of ways, the more important takeaways from this research are what it does not show:
- It does not show masks are ineffective for preventing the wearer from becoming infected with SARS-CoV-2. The 95% confidence interval for the between group difference is -1.2 to 0.4 percentage points. This can be strictly interpreted, in a frequentist sense, to indicate a range of possible true effect sizes from this study. Most of these true effect sizes favor the intervention, but, most likely not to the effect size meeting the authors’ definition of a clinically meaningful difference.
- It does not have invalid results because of imperfect adherence with mask use. Research results need to be applicable to the world in which we live. A pragmatic trial taking into account individual behaviors reflects (and likely even overstates) their typical real-world use, and this can greatly inform public policy.
- It does not, finally, inform any aspect of current mask recommendations, which are designed, primarily, to prevent the wearer from spreading infection. This is why surgeons wear masks while operating – it isn’t for their own protection, but for the patient.
There are also additional points to make regarding the right censoring of outcomes. Patients received their final testing kits at one month, but sensitivity limitations inherent to both PCR and lateral flow immunoassays could have missed infections present at time of final testing. This would have the effect of attenuating the observed effect size. Other limitations of these tests are clearly applicable, but their inaccuracies ought to be evenly distributed between groups. Many others have also pointed out issues with study attrition, and its inevitable effect on outcomes.
Lastly, a fair bit of discussion on Twitter pertains to whether this study ought to have been published at all, considering it may be disseminated with headlines lacking the nuance of the study findings. I certainly do not have a problem with letting light shine upon data, and it being the responsibility of authors and editors to educate through their contextual presentation. I would also go as far to say if this trial had been clearly positive (instead of just inconclusive based on their sample and power), it would have been potentially a strong motivator for mask use. We should be wary of the scientific harms of publication bias and encourage the completion and dissemination of “negative” studies.
“Effectiveness of Adding a Mask Recommendation to Other Public Health Measures to Prevent SARS-CoV-2 Infection in Danish Mask Wearers: A Randomized Controlled Trial”
https://www.acpjournals.org/doi/10.7326/M20-6817