French Study Suggests That Long Covid is Mostly Psychosomatic

Long Covid was initially believed to affect one in every ten people who catch the virus. However, estimates have since come down considerably. In September of this year, the ONS published research indicating that only 2.5% of people still report symptoms after 12 weeks.

As I noted in a write-up for the Daily Sceptic, even this 2.5% figure is probably an overestimate, since it assumes that every participant reported their symptoms accurately. Due to media attention surrounding long Covid, some participants might have been inclined to exaggerate their symptoms – to report things they normally wouldn’t have done.

A new study published in JAMA Internal Medicine suggests that the 2.5% figure is an overestimate. Joane Matta and colleagues analysed data on a sample of about 27,000 French people, who were given serology (i.e., antibody) tests between May and November of 2020.

The same individuals took a questionnaire between December 2020 and January 2021. In that questionnaire, they were asked, “Since March, do you think you have been infected by the coronavirus (whether or not confirmed by a physician or a test)?”

Respondents who answered “Yes” were also asked when they caught the virus. Those who indicated that they caught it after their serology test were excluded from the analysis. Additionally, all respondents were asked to say whether they had experienced each of 18 different symptoms since March of 2020.

The main results are shown in the table below. There are four columns, corresponding to different combinations of belief and serology. The two on the left correspond to a negative test, while the two on the right correspond to a positive test. The belief columns indicate whether respondents believed they had been infected.

There are two main things to notice. First, if we compare the column on the left (for people with a negative test who believed they had not been infected) to the two columns on the right (for people with a positive test), we see that the percentages are about the same. The only exception is anosmia, shown in the final row.

In order to do this, you have to compute a ‘mental weighted average’ of the two columns on the right. For example, 7.3% of people in the first column had joint pain, while the corresponding percentage for the last two columns is 6.5% (the weighted average of 4.2 and 8.2).

These comparisons indicate that respondents who tested positive for Covid antibodies were not, in general, more likely to report symptoms than those who tested negative. (The only symptom they were more likely to report was anosmia.)

The second thing to notice is that, if we compare the two “Belief +” columns to the two “Belief –” columns, we see that the percentages tend to be higher in the former. This indicates that people who believed they had been infected reported were more likely to report symptoms, regardless of whether they actually had been infected.

The researchers estimated multivariate models that controlled for characteristics like age, sex and education, and observed the same pattern of results. Believing that one had had Covid was associated with reporting symptoms, but actually having had Covid was not (with the exception of anosmia).

Matta and colleagues’ findings are consistent with earlier studies based on young people, which found little or no difference in symptoms between those who were seropositive and those who were seronegative. Long Covid, it seems, is mostly psychosomatic.