A systematic review published in JAMA Network Open looks at short-term and long-term rates of postacute sequelae of SARS-CoV-2 infection.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This is a potentially valuable contribution to the burgeoning research literature about what’s commonly called ‘long Covid’. To some extent, what it demonstrates most clearly is the huge variability in definitions of what the condition is (or the conditions are – it’s still not at all clear whether it makes any sense to think of it as a single condition or syndrome). Because of that variability in definitions, and because there are many ways to attempt to estimate how common the condition(s) is or are, the estimates of the prevalence of long Covid from different studies are very variable indeed. Because of that, and rightly in my view, the researchers did not attempt a formal statistical meta-analysis, that would have combined all the different estimates from the studies they examined into an overall estimate. Therefore the researchers’ overall conclusion, that “more than half of COVID-19 survivors experienced PASC [postacute sequelae of COVID-19] 6 months after recovery”, does look dramatic but is not very solid statistically, I’d say. But I do agree with them that having to deal with these long-term effects could have a major impact on health care systems, particularly in low- and middle-income countries. Indeed I think it’s likely to have a considerable effect in the UK, a high-income country, for instance.
“Perhaps we’re going to have to get used to PASC, standing for postacute sequelae of COVID-19, as another term for long Covid, alongside such terms as “ongoing symptomatic Covid-19” and “post-Covid-19 syndrome” (the terms currently recommended by NICE*, depending on how long the symptoms have persisted), “ongoing symptoms following coronavirus (COVID-19) infection” (ONS**), and many more. PASC really just means something like “conditions that are a consequence of a Covid-19 infection, after the initial (acute) stage of the disease has ended. In this new study, the conditions or symptoms had to go on at least a month after either diagnosis of Covid-19 or discharge from hospital to be counted as PASC, and the researchers distinguished between short-term PASC (at least 1 month but less than 2), medium-term PASC (2 to 5 months), and long-term (6 or more months). These don’t, for example, match the NICE distinction (which distinguishes between signs and symptoms going on for 4 to 12 weeks, and going on for more than 12 weeks), but the researchers couldn’t possibly come up with a split based on time that matched most of the studies they looked at, because the studies that they reviewed are so variable in what they did.
“The new study does appear to have been carried out carefully and generally appropriately. It reviewed in all 57 different studies from 18 different countries, most of which are from high-income countries (though the list does also include studies from Bangladesh and India, for instance). And this is certainly not going to be the last word on the matter. The cut-off date for the search for studies was March 2021. This means that long Covid studies that have been in the news in the UK recently are simply not there, because they came out too recently. So there’s nothing on the ONS work*** based on their infection survey, or on any of the UK-based studies with which ONS make comparisons in that article, or with (for instance) the recently-published CLoCk study**** of long Covid in children and young people. Those other studies probably all came out after the search cut-off date (though a couple did first appear in March 2021). Also most of them are preprints, rather than peer-reviewed published papers, and only three of the 57 studies reviewed in this new research are preprints. And I’ve only looked here at UK-based studies – doubtless research on long Covid is also going on apace in many other countries.
“Because the reviewed studies are so diverse – different definitions of the condition(s), concentration on different types of signs or symptoms, different populations (people hospitalised with Covid-19, people diagnosed in the community, and other groups), and so on – the conclusion that “more than half of COVID-19 survivors experienced PASC 6 months after recovery” is a bit dubious, and indeed doesn’t really follow from the analysis. What they actually found was that the median proportion that still had at least one sign or symptom, at least 6 months after they entered each of the individual studies, was 54%. In other words, if you lined up the proportion still having at least one sign or symptom in order, for the nine studies that they reviewed that included data going out to 6 or more months, 54% would be in the middle of that line. This doesn’t relate to the individuals in the studies, but only to the overall percentages in the studies, and in particular it takes no account of the fact that the nine studies included different numbers of individuals – so it’s not talking about a percentage of individual Covid-19 survivors, it’s talking about a kind of average property of those nine studies. The percentage with at least one sign or symptom after at least 6 months ranged from 31% to 67% in those nine studies. So I don’t attach much importance to the specific “more than half” conclusion.
“But however you look at it, it’s very clear that signs and symptoms going on for some considerable time after a Covid-19 infection are common, in people who have had Covid-19, and that health care of various different sorts will be required to cope with this, on a major scale. This new research can’t really give details on exactly how Covid-19 infection actually causes these long-term effects, though it does list some possible mechanisms that have been discussed. Because the range of possible long-term symptoms is very wide, it seems unlikely that just one or two causal mechanisms are involved. It also can’t really say much on how long these long-lasting symptoms are likely to go on in individuals.
“It’s also true that many of the possible long-term symptoms can have other causes, that may turn out to have nothing direct to do with the previous Covid-19 infection. Some other long Covid studies have compared a group of people with known infection with a separate, control group that very probably were never infected, and in some of those studies, the prevalence of symptoms like those in the cases who definitely had the disease is quite high (though lower than in the cases). This new review does not generally consider controls of that kind.”
‘Short-term and Long-term Rates of Postacute Sequelae of SARS-CoV-2 Infection – A Systematic Review’ by Destin Groff et al. was published in JAMA Network Open at 16:00 UK time on Wednesday 13 October.
All our previous output on this subject can be seen at this weblink:
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. I am also a member of the Public Data Advisory Group, which provides expert advice to the Cabinet Office on aspects of public understanding of data during the pandemic. My quote above is in my capacity as an independent professional statistician.”