July 28, 2020

expert reaction to preprint of a survey on transmission and height in relation to COVID-19

A preprint, an unpublished non peer-reviewed study posted on medRxiv, looked at transmission and height in relation to COVID-19.

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This research didn’t use any objective measures of viral spread.  The finding on height, like all the other findings, is based on a survey of 1000 adults of working age in the UK, and another similar survey in the USA.  The respondents were asked whether they had even had a medical diagnosis or positive test for COVID-19, together with a number of other questions about a very wide range of matters, including whether they were taller than six feet.  (No other questions about height were asked, so the researchers could not investigate anything about relationships between having had COVID-19 and other heights than six feet.)  The research was all done through electronic communication with the respondents (phones or computers), so what is being analysed is effectively simply what the respondents said.

“The association between height and COVID-19 diagnosis is just that, an association or correlation.  The patterns are not at all clear anyway – and the statistical evidence does not firmly establish that there’s any association at all.  The results couldn’t in any case show that the higher rate of COVID-19 diagnosis in people over six feet tall was actually caused by their heights, because no piece of survey research could do that on its own.  There are plenty of other differences between the populations of people under and over six feet high, apart from their height, even after allowing for their gender (which the researchers did).  It’s quite possible that one or more of these other differences is the cause of any difference in COVID-19 risk, and not the people’s heights at all.  The researchers did adjust their results to allow for possible differences that could be measured by the many other questions that the respondents answered, but they didn’t include all the plausible and possible other factors that might be involved.  And, importantly, the measure of association between the height factor and COVID-19 infection was not statistically significant, either in the UK or the USA, after adjustment for the other factors.  That is, it remains plausible that there is no genuine association between height and COVID risk, and all that is being seen in these data, in relation to heights, is the effect of chance variability.  Finally on this aspect, after making the adjustments for all the other factors, the association between height and COVID risk appears to go in different directions in the US and the UK.  Other things being equal on all the factors that were taken account of, people in the UK had a rather higher risk of having had a positive COVID-19 test if they were over six feet than if they were shorter, but in the US it was the other way round, the people over six feet had a rather lower risk.  The fact that the association goes in opposite directions in both countries would lead me to think that it probably doesn’t have a biological basis, if it really exists at all and isn’t just an effect of chance.

“If, and it’s a big if, an association between height and COVID-19 risk does exist, this study can’t say how it occurs physically, and the speculation that it might be related to the physical transmission route of the virus (aerosol or droplet) is just that, a speculation.  That’s because the study did not measure anything about airborne transmission.

“Finally, it’s worth pointing out that these results come from a preprint that has not yet gone through peer review.”

Prof Paul Hunter, Professor in Medicine, UEA, said:

“In my view this preprint should be interpreted with caution.  The authors analysed a substantial number of possible predictors for a range of outcome measures.  One of the biggest mistakes that people make with epidemiological surveys is multiple hypothesis testing.  The more hypothesis tests you do the more likely you are to flag up associations that have only arisen by chance.  This is a further problem when subgroups are analysed.

“In my view this analysis does not even provide convincing evidence that tall people are protected from the infection never mind whether this means that there is aerosol transmission.”

Preprint : ‘Work-related and Personal Predictors of COVID-19 Transmission’ by Paul Anand et al.  was posted on medRxiv. This work is not peer-reviewed.

All our previous output on this subject can be seen at this weblink:

www.sciencemediacentre.org/tag/covid-19

Declared interests

Prof Kevin McConway: “I do know Paul Anand, the lead author.  I am a member of the SMC Advisory Committee, but my quote above is in my capacity as a professional statistician.”

None others received.