select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
A study published in the BMJ looks at prognosis and persistence of smell and taste dysfunction in patients with COVID-19.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“I think this is a good, comprehensive study that uses appropriate statistical methods to draw some overall conclusions about long term smell and taste problems after Covid. I think there are some issues with the detailed numbers that the researchers report, and I’ll describe those below. But I don’t think that any of these issues detract from the finding that long-term problems with smell and taste, after Covid, are a major global issue affecting very large numbers of people. This raises major problems for health services as well as for the individuals who are affected.
“I’d question some numbers in the press release. Particularly, the top line says that ‘At least 27 million covid patients may have long term smell and taste problems’. I don’t see the 27 million mentioned in the actual research paper, and as far as I can make out it has come from adding together the estimates (which are in the research paper) that, globally, there are more than 15 million and 12 million patients with long term smell and taste problems, respectively. You can’t just add up the 15 million and 12 million to say that there are more than 27 million people with the problems, because that doesn’t take account of the fact that some individuals will, sadly, have long-term problems with both smell and taste (so that some of the 15 million are also included in the 12 million, and vice versa). And in fact I’d criticise the 15 million and 12 million figures too – see below.
“Further, I think the mention in the paper and in the press release about 5% of Covid patients developing long lasting changes to smell and/or taste is potentially misleading. In the form in the press release it reads, “About 5% of adults may develop long lasting changes to their sense of smell or taste after covid-19 infection, suggests research published by The BMJ today.” The research paper itself says much the same. But in my view this is not clear enough on what the 5% is a percentage of. Actually the new study, and the individual studies whose data it is based on, relate to recovery from smell or taste problems in people who had issues with their senses of smell and/or taste associated with their Covid disease. Not all Covid patients do lose their senses of smell or taste (or have other smell or taste problems) when they have the disease. The researchers say in their paper that 40-50% of Covid patients do have smell or taste problems, which makes sense at least in relation to the original variants of the virus – but it means that around half of Covid patients never have these issues. So the percentage of all Covid patients who may develop long-lasting changes to their smell or taste, taking into account those who never have smell or taste problems, is nearer to half of the 5% figure, so perhaps 2.5%, on the basis of this new research. That’s still a lot of people worldwide, though.
“One unavoidable issue with the statistical methods used in the research is that the conclusions about long-term persistent problems are based on data from studies that did not all follow patients up for a really long time. (In fact, since the search strategy used by the researchers included published studies and preprints only up to early October 2021, and it takes some time to do the statistical analysis even for a preprint after data collection ceased, and since almost nobody had Covid before early 2020, no study could have followed people up for more than about a year and a half, and most were considerably shorter.) So how do the researchers count people who did not recover their smell or taste if they don’t have data following the patients up for years on end? In fact they do this using a method called cure modelling, which has to assume that the curve describing how people recover over time has a shape drawn from a particular family of possible shapes. (The name comes, generally from the idea that some patients are ‘cured’ of their problem, while others aren’t.) The assumption that these researchers made is reasonable and pretty standard, but it’s a limitation that it can’t be checked in all its detail from the available data.
“One way in which this shows up in the findings is that these statistical models come up with estimates that 5.6% of Covid patients, who initially had smell problems, having persistent smell problems, and 4.4% of Covid patients, who initially had taste problems, having long-term taste problems, where in terms of the statistical model that was used, ‘long-term’ actually means that they won’t ever recover their taste or smell. However, the researchers also report that 4% still had problems with their sense of smell after 6 months, which is already less that 5.6%, and similarly that 2% still had taste problems after six months, considerably less than 4.4%. Part of the discrepancy could well be because the statistical margins of error around the 5.6% and 4.4% estimates are actually rather wide – the 5.6% figure for persistent problems with smell count plausible be anywhere between 2.7% and 11%, and the corresponding range for the 4.4% taste estimate runs from 1.2% right up to 14.6%.
“What I take from this is that any research like this, based to some extent on statistical modelling, can sometimes produce estimates with wide margins of error. Also, in any case the estimates depend on the statistical assumptions made in the modelling, which are never going to be a complete representation of reality, in the same way that no geographical map exactly represents the territory that it maps. But the general finding, without the detailed numbers, is that the great majority of people who have Covid either won’t have smell or taste problems at all, or if they do have them, will recover their smell or taste (though that might take months for some of them). That does leave a small minority who will have persistent problems – but a small minority, out of the hundreds of millions worldwide who have had Covid, is a lot of people and so an important health issue. I think this broad picture is confirmed by this new research, despite my issues with some of the detailed numbers. (The accompanying editorial does describe some very useful-sounding approaches to helping people with long-term problems, but this is beyond my own expertise to evaluate.)
“My issue with the 15 million and 12 million estimates of long-term smell and taste problems is as follows. The studies that these researchers put together to obtain their estimates of the percentages of people affected were all carried out before the Omicron variant and its sub-variants became dominant at the end of 2021. The researchers point out in their paper that more recent studies have found that the Omicron variants have been causing many fewer patients to report smell and taste problems than with earlier variants, even in the initial stage of their Covid disease. But, for instance, the estimate of 15 million people with long-term smell problems comes from applying the estimated percentage of people with long-term smelling problems, 5.6%, to the estimated global number of confirmed Covid cases as of July 2022, which was (over) 550 million, and taking account of the fact that around half of the 550 million never had smell or taste problems to begin with. (So the calculation is that around 275 million people had Covid with smell and/or taste problems up to July 2022, and 5.6% of 275 million is 15.4 million. The estimate of 12 million with long-term taste problems comes from a similar calculation but using 4.4% instead of 5.6%.)
“The snag is that many of the 550 million global confirmed cases would have had Covid after Omicron variants became dominant, and a much smaller percentage of those would have ever had smell or taste issues than for earlier variants. So it doesn’t make good sense to assume that half of the 550 million had smell or taste problems. (Arguably it also doesn’t make sense to apply the estimates for persistent smell and taste problems from this research, which also come from data before Omicron, to data that will include many Omicron cases anyway.) A further snag is that the 550 million is a cumulative estimate of confirmed cases, and doesn’t take account of the fact that some of these cases, indeed quite a lot of them since Omicron, are reinfections, so that the number of people who have had Covid is smaller than 550 million.
“All in all, yes, the 15 million and 12 million are impressively large numbers, but they seem to come from a rather back-of-the-envelope calculation, even though it’s partly based on the more careful findings of this research. My suspicion is that the true numbers should be considerably smaller – but are still likely to run well into millions of people. This is serious, even if some of the detailed numbers might not be very accurate.”
Prof Danny Altmann, Professor of Immunology, Imperial College London, said:
“This is a strong and important study, alerting us once again to the difficulties inherent in charting the scale of long-term damage caused by Covid-19. The authors conducted a rather rigorous meta-analysis across several cohorts, modelling the time to recovery of taste and smell. It goes without saying that problems with taste and smell are non-trivial for quality of life. This is part of a wider discussion about how we assess and address the persistent changes collectively termed Long Covid. Studies such as this alert us to the hidden burden out there of people suffering with persistent symptoms, but perhaps not having thought it worth contacting the GP on the assumption there wouldn’t be much to be done.”
‘Prognosis and persistence of smell and taste dysfunction in patients with covid-19: meta-analysis with parametric cure modelling of recovery curves’ by Benjamin Kye Jyn Tan et al. was published in the BMJ at 23:30 UK time on Wednesday 27 July 2022.
DOI: 10.1136/bmj-2021-069503
Declared interests
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. My quote above is in my capacity as an independent professional statistician.”
Prof Danny Altmann sits on the COVID-19 Vaccines Advisory Board of AstraZeneca. Altmann is co-author of the Penguin Handbook of Long Covid.