Introduction
Since the onset of the COVID-19 pandemic, concerns have been raised about its impact on mental health (Holmes et al., Reference Holmes, O'Connor, Perry, Tracey, Wessely, Arseneault and Bullmore2020; Xiang et al., Reference Xiang, Yang, Li, Zhang, Zhang, Cheung and Ng2020). Early studies conducted in the general population suggested a significant and global increase in depressive and anxiety symptoms between the pre- and mid-pandemic periods (Rogers et al., Reference Rogers, Chesney, Oliver, Pollak, McGuire, Fusar-Poli and David2020; Santomauro et al., Reference Santomauro, Herrera, Shadid, Zheng, Ashbaugh, Pigott and Ferrari2021). Acute infectious diseases are recognized risk factors of psychiatric disorders (Köhler-Forsberg et al., Reference Köhler-Forsberg, Petersen, Gasse, Mortensen, Dalsgaard, Yolken and Benros2019; Orlovska et al., Reference Orlovska, Vestergaard, Bech, Nordentoft, Vestergaard and Benros2017), particularly mood disorders (Benros et al., Reference Benros, Waltoft, Nordentoft, Østergaard, Eaton, Krogh and Mortensen2013). One of the main proposed explanations for these associations was that the immune system response may induce neuroinflammation that contributes to the emergence of psychiatric disorders (Leboyer et al., Reference Leboyer, Berk, Yolken, Tamouza, Kupfer and Groc2016; Pape, Tamouza, Leboyer, & Zipp, Reference Pape, Tamouza, Leboyer and Zipp2019). A similar hypothesis has been put forward for COVID-19 in particular (Bottemanne, Delaigue, & Lemogne, Reference Bottemanne, Delaigue and Lemogne2021; Han et al., Reference Han, Yuan, Wang, Liu, Lu, Liu and Lu2021).
Despite evidence that SARS-CoV-2 may have an impact on brain tissues (Douaud et al., Reference Douaud, Lee, Alfaro-Almagro, Arthofer, Wang, McCarthy and Smith2022; Penninx, Benros, Klein, & Vinkers, Reference Penninx, Benros, Klein and Vinkers2022), evidence from epidemiology remains elusive as to establish a direct association between SARS-CoV-2 infection and depression in the general population, with inconsistent results. Analyzing the risk of mood disorders following COVID-19 from electronic health records in different countries with different follow-up times (90 days, 6 months, 2 years), Taquet et al., found significant associations between COVID-19 and further mood disorders, in comparison to other infectious diseases (Taquet, Geddes, Husain, Luciano, & Harrison, Reference Taquet, Geddes, Husain, Luciano and Harrison2021a; Taquet, Luciano, Geddes, & Harrison, Reference Taquet, Luciano, Geddes and Harrison2021b; Taquet et al., Reference Taquet, Sillett, Zhu, Mendel, Camplisson, Dercon and Harrison2022). In a study of the risk of mood disorders after an hospitalization for COVID-19 in France during the year 2020, in comparison to hospitalization for another reason, Decio et al. (Reference Decio, Pirard, Pignon, Bouaziz, Perduca, Chin and Tebeka2022) and Geoffroy et al. (Reference Geoffroy, Decio, Pirard, Bouaziz, Corruble, Kovess-Masfety and Tebeka2024) found crude increases of the risk of mood disorders and depression. However, these associations were no longer significant following adjustments for confounders, in particular history of mood disorders. In a prospective study of the evolution of mental health between April 2020 and 2021 in eleven longitudinal population-based studies from the United Kingdom, Thompson et al. (Reference Thompson, Stafford, Moltrecht, Huggins, Kwong, Shaw and Patalay2022) found an association of small magnitude between COVID-19 and depressive symptoms. However, analyzing the association according to self-report and serology status, they displayed significant variations: self-reported COVID-19 without positive serology were associated with depressive symptoms, while participants with positive serology without self-reported COVID-19 had unchanged mental health (including depression) during the study period. Strikingly, those with both self-reported COVID-19 and positive serology, thus presenting with the highest probability of having been infected with SARS-CoV-2, also had unchanged mental health (including depression) during the study period. Interestingly, Davisse-Paturet et al. (Reference Davisse-Paturet, Orri, Legleye, Florence, Hazo and Warszawski2023) found similar results when examining suicidal ideation.
In addition, the critical issue of overlapping symptoms – specifically somatic symptoms – between acute infectious diseases, enduring systemic inflammation, and depression remains unaddressed in the context of COVID-19. Fatigue, cognitive impairment, sleep disorders, loss of appetite, or anhedonia, which are core symptoms of a major depressive episode, are also core symptoms of ‘sickness behavior’ (Hart, Reference Hart1988). Sickness behavior is characterized by a set of behavioral changes that occur in response to an infection and helps combat the infection and prevent its spread. Importantly, sickness behavior, which is mediated by pro-inflammatory cytokines, mainly interleukins 1β and 6 (IL-1β and IL-6) and tumor necrosis factor-α (TNF-α), also encompasses other depressive symptoms than somatic ones, such as social withdrawal (Shattuck & Muehlenbein, Reference Shattuck and Muehlenbein2016). In addition, persistent symptoms may impair the quality of life of many patients months after a COVID-19 episode (Ballering, van Zon, Hartman, & Rosmalen, Reference Ballering, van Zon, olde Hartman and Rosmalen2022; Robineau et al., Reference Robineau, Zins, Touvier, Wiernik, Lemogne and de Lamballerie2022). Among frequent symptoms of this ‘post-COVID-19 condition’, frequently referred to as ‘long COVID’, fatigue, poor attention/concentration, or sleep disorders may also overlap with depressive symptoms and thus lead to overestimate the association between SARS-CoV-2 infection and subsequent depression in epidemiological studies (Soriano, Murthy, Marshall, Relan, & Diaz, Reference Soriano, Murthy, Marshall, Relan and Diaz2022).
This longitudinal study aimed to examine the association of SARS-CoV-2 infection with subsequent depressive symptoms in a large population-based sample. We investigated particularly the combination of self-reported COVID-19 with biological confirmation of the infection on this association, while considering various dimensions of depressive symptoms, including somatic symptoms. Regarding self-report v. biological confirmation, we hypothesized that we would replicate and extend the results of Thompson et al. (Reference Thompson, Stafford, Moltrecht, Huggins, Kwong, Shaw and Patalay2022) and show stronger association of depressive symptoms with self-reported infection than with biologically-confirmed infection. We further hypothesized that much of the association between COVID-19 and depressive symptoms would be explained by a specific association with somatic symptoms of depression.
Methods
CONSTANCES and SAPRIS studies
The French CONSTANCES population-based cohort study received ethical approval by the institutional review board of the National Institute for Medical Research (Authorization number 910486) and included more than 200 000 volunteers aged 18–69 years at inclusion (i.e. between 2012 and 2019) who gave informed consent to be followed-up through annual questionnaires (Zins & Goldberg, Reference Zins and Goldberg2015). Participants were selected among individuals covered by the general insurance scheme or partner health mutual societies (i.e. 85% of the French population) using a random sampling system stratified on place of residence, age, sex, occupation, and socioeconomic status. At inclusion, volunteers completed a self-administered questionnaire on lifestyle and health status and attended a Health Screening Center for a comprehensive evaluation including a physical examination and laboratory tests.
Between 2 April 2020 and 12 May 2020, a total of 63 471 volunteers of the CONSTANCES cohort responding to annual questionnaires through the internet were invited to take part in the nested SAPRIS (‘Santé, pratiques, relations et inégalités sociales en population générale pendant la crise COVID-19’) and SAPRIS-Sérologie (SAPRIS-SERO) surveys, which concerned specifically the impact of the COVID-19 pandemic (Carrat et al., Reference Carrat, de Lamballerie, Rahib, Blanché, Lapidus and Artaud2021; Matta et al., Reference Matta, Wiernik, Robineau, Carrat, Touvier and Severi2022, Reference Matta, Robineau, Wiernik, Carrat, Severi, Touvier and Lemogne2023). The SAPRIS survey was approved by the French Institute of Health and Medical Research ethics committee, and the SAPRIS-SERO survey was approved by the Sud-Méditerranée III ethics committee. Participants were not offered any kind of incentive for participating neither in CONSTANCES nor in SAPRIS.
The present study is a retrospective longitudinal analysis of data from the French CONSTANCES cohort and from the SAPRIS and SAPRIS-SERO surveys nested in the CONSTANCES cohort. Figure 1 shows the different steps of inclusion and data collection.
Figure 1. Time frame of the study based on data from CONSTANCES cohort and nested SAPRIS and SAPRIS-SERO surveys. Inclusion in the CONSTANCES cohort occurred between 2012 and 2019. Among CONSTANCES volunteers, inclusion in SAPRIS and SAPRIS-SERO surveys occurred between 2 April 2020 and 12 May 2020. SAPRIS ad hoc SARS-CoV-2 serological tests were conducted between May and October 2020. Self-reported COVID-19 was measured retrospectively between December 2020 and February 2021. Center of Epidemiologic Studies-Depression Scale (CES-D) was completed between December 2020 and February 2021.
SARS-CoV-2 infection
SAPRIS serology
Between May and October 2020 (i.e. a period without a wide availability of PCR tests), self-sampling dried blood spot kits were mailed to each participant, with kit material and printed instructions to mail a blood spot to a centralized biobank. Received blood spots were visually assessed, registered, punched, and stored in tubes (0.5 mL, FluidX 96-Format 2D code; Brooks Life Sciences) at −30 °C. Eluates were processed with an enzyme-linked immunosorbent assay (Euroimmun) to detect anti-SARS-CoV-2 antibodies (IgG) directed against the S1 domain of the virus spike protein. A test was considered positive for SARS-CoV-2 when the results indicated an optical density ratio of 1.1 or greater (sensitivity/specificity: 0.870/0.975). To reduce the risk of false-negative results, participants with indeterminate results (i.e. optical density ratio ≥ 0.8 and <1.1) and without declaration of an otherwise positive test were excluded.
Self-reported infection
Between December 2020 and February 2021, participants answered the question: ‘Since March, do you think you have been infected with the coronavirus (whether or not confirmed by a physician or a test)?’. Participants answered ‘Yes’, ‘No’, or ‘I don't know’, and participants who answered ‘Yes’ were additionally asked whether their infection had been confirmed, with two possible responses indicating biological confirmation: ‘Yes, by virological or PCR test (based on nose swab, results provided after at least 24 h)’, ‘Yes, by serological test (based on a blood test, results provided after at least 24 h)’.
Assessment of depressive symptoms
Between December 2020 and February 2021, simultaneously with the assessment of self-reported infection, depressive symptoms were assessed using the self-administered Center of Epidemiologic Studies-Depression Scale (CES-D) (Morin et al., Reference Morin, Moullec, Maïano, Layet, Just and Ninot2011). This 20-item scale assesses the frequency of depressive symptoms during the previous week. Reponses range from 0 (mostly never) to 3 (most of the time), resulting in a total score ranging from 0 to 60. According to the validated cutoff of the French version, ‘clinically significant depressive symptoms’ were defined as having a total score ≥ 19 (sensitivity/specificity for major depression: 0.85/0.86). Based on the dimensional structure of the French version of the CES-D, the four following dimensions of the CES-D were also assessed: depressed affect (7 items, scores between 0 and 21), positive affect (4 items, scores between 0 and 12), somatic complaints (7 items, scores between 0 and 21), and disturbed interpersonal relationships (2 items, scores between 0 and 6).
Adjustment variables
Sex and age were obtained from the national personal identification directory. Educational level, household income, and current tobacco smoking status were self-reported at inclusion in the CONSTANCES cohort. Body mass index (BMI) was categorized in four categories (<18.5 kg/m2: underweighted, 18.5–25: normal, 25–30: overweighted, >30: obese) from weight and height measured in Health Screening Centers at inclusion. Self-rated health (‘How do you rate your overall health?’) was assessed according to a score between 1 to 8 (1: ‘very bad’, 8: ‘very good’) at the same time as CES-D.
Statistical analyses
Six dependent variables were considered for depressive symptoms, including five continuous variables (i.e. total CES-D score and its four dimensions) and one binary variable (i.e. clinically significant depressive symptoms defined by a total CES-D score ≥ 19). For the total CES-D score, person mean imputation was performed if 2 items or less were missing (Hawthorne, Hawthorne, & Elliott, Reference Hawthorne, Hawthorne and Elliott2005). If more than 2 items were missing, the subjects were excluded from the current analyses.
Unadjusted analyses first considered COVID-19 status as a 3-category variable: no COVID-19/self-reported unconfirmed COVID-19 (i.e. neither a positive SAPRIS serology nor self-reported serology or PCR test results)/biologically confirmed COVID-19, self-reported or not. Pairwise mean comparisons were performed with t-Student tests and χ2 test, for continuous and binary dependent variables, respectively.
Adjusted linear and/or binary logistic regression models were then computed for continuous and/or binary dependent variables previously associated with the 3-category COVID-19 status variable in unadjusted analyses. All a priori selected covariates were included in the models (i.e. sex, age, educational level, household income, smoking status, BMI, and self-rated health). The continuous dependent variables (i.e. CES-D total score and its four dimensions) were divided by their interquartile range (IQR), to yield interpretable regression coefficients and 95% confidence intervals (CIs).
Finally, in order to replicate the findings by Thompson et al. (Reference Thompson, Stafford, Moltrecht, Huggins, Kwong, Shaw and Patalay2022), these adjusted analyses were repeated with COVID-19 status as a 4-category variable based on self-report and SAPRIS serology only: no self-reported COVID-19 and SAPRIS serology negative test results/self-reported COVID-19 and SAPRIS serology negative test results/no self-reported COVID-19 and SAPRIS serology positive test results/self-reported COVID-19 and SAPRIS serology positive test results. For these analyses, participants with SAPRIS serology negative test results who later reported positive PCR or serology test results were excluded.
Results
Participants
Among the 63 541 volunteers from CONSTANCES cohort who were included in the SAPRIS survey, 28 704 participants were not included in this study because they were not included in SAPRIS-SERO study and 786 because they had indeterminate SAPRIS-SERO serology test results without self-reported serology or PCR positive test results. Of the 34 041 remaining participants, 2034 had unavailable data concerning self-reported COVID-19. A total of 32 007 participants were thus included in the present study. Among these participants, 925 reported positive PCR or serology test results after negative SAPRIS serology and were not included in the analyses based on the 4-category COVID-19 status variable.
The characteristics of the participants (total sample, samples with and without self-reported and/or biologically confirmed COVID-19) are displayed in Table 1. Women were overrepresented (51.6%). Overall mean age was 52.5 years (standard deviation (s.d.) = 13.3). While 28 659 participants (89.5%) did not have any self-reported or biologically confirmed COVID-19 episode, 666 (2.1%) self-reported a COVID-19 episode without biological confirmation, and 2682 (8.4%) had a biologically confirmed COVID-19.
Table 1. Characteristics of the participants according to COVID-19 status
Abbreviations: s.d., standard-deviation; CES-D, Center of Epidemiologic Studies-Depression scale.
a According to CES-D binarized score: participants with a score ≥ 19 are considered with clinically significant depressive symptoms.
b Educational level: 1 = without diploma, without high school diploma, or with certificate of vocational aptitude or vocational studies; 2 = high school diploma or equivalent; 3 = 2 or 3 years of post-secondary education, 4 = 4 years of post-secondary education; 5 = 5 years or more of post-secondary education.
c At the inclusion in CONSTANCES cohort.
d Self-rated health 1: ‘very bad’ to 8: ‘very good’.
Unadjusted analyses
Pairwise mean comparisons revealed an increasing gradient of the mean scores for the CES-D total, depressed affect and somatic complaints dimensions, across the three categories as follows: first, no COVID-19, then biologically confirmed COVID-19, and third self-reported unconfirmed COVID-19 (Table 2). The same gradient was found for the prevalence of ‘clinically significant depressive symptoms’, but not for the positive affect or disturbed interpersonal relationships CES-D dimension scores.
Table 2. Comparisons of meana of CES-D scores or proportion of clinically significant depressive symptomsb according to the 3-category COVID-19 status
Abbreviations: CES-D, Center of Epidemiologic Studies-Depression Scale; df, degrees of freedom.
a Using t-Student tests.
b Using χ2 tests.
c According to CES-D binarized score: participants with a score ≥ 19 are considered with clinically significant depressive symptoms.
Adjusted analyses
In unadjusted analyses, CES-D total score (Table 3), as well as depressed affect and somatic complaints dimension scores, were the lowest in non-infected participants, intermediate in participants with biologically confirmed COVID-19 and highest in participants with self-reported unconfirmed COVID-19. Adjusted analyses were consistent for total score and somatic complaints dimension. Only participants with self-reported unconfirmed COVID-19 had higher depressed affect dimension score, in comparison to participants without prior infection. Concerning depressed affect dimension score, in adjusted analyses, no significant variations were observed. The positive affect and the disturbed interpersonal relationships dimensions score were not associated with the COVID-19 status. Clinically significant depressive symptoms were more frequent in participants with biologically confirmed COVID-19 and, with a greater effect size, in those with self-reported unconfirmed COVID-19 in both unadjusted and adjusted analyses.
Table 3. Analyses of the association of the 3-category COVID-19 status and CES-D scores or clinically significant depressive symptoms
Abbreviations: CI−, lower confidence limit; CI+, upper confidence limit; IQR, interquartile range; OR, odds ratio; ref., reference value.
a Adjusted on sex, age, educational level, household income, smoking status, BMI, and self-rated health.
b CES-D scores are divided par the IQR.
*p value < 0.05, **p value < 0.01, ***p value < 0.001.
Among the adjustment factors, in the different models, female sex, higher educational level, tobacco smoking, younger age, lower household income, obese status, and self-rated health were associated with greater depressive symptoms (online Supplementary Table S1).
Finally, regarding SAPRIS serology test results specifically, 323 participants did not report having been infected with SARS-CoV-2 while they had positive test results, 822 reported COVID-19 while they had negative test results, and 632 reported COVID-19 with positive test results, while 28 659 did not report COVID-19 with negative test results. In comparison to the absence of COVID-19 history (according to both self-report and SAPRIS serology test results), only participants with self-reported COVID-19 with negative SAPRIS serology test results had higher CES-D total scores (Table 4, see online Supplementary Table S2 for details concerning adjustment factors). Participants with self-reported COVID-19 had higher depressed affect and somatic complaints dimension scores, especially those with negative SAPRIS serology test results. Positive affect and disturbed interpersonal relationship dimension scores did not significantly vary according to the four-category COVID-19 variable. The proportion of ‘clinically significant depressive symptoms’ was higher among participants who self-reported COVID-19, whatever the SAPRIS serology test results.
Table 4. Adjusteda association between CES-D scores, clinically significant depressive symptoms, and the 4-category COVID-19 status
Abbreviations: CI−, lower confidence limit; CI+, upper confidence limit; IQR, interquartile range; OR, odds ratio; ref., reference value.
a Adjusted on sex, age, educational level, household income, smoking status, BMI, and self-rated health.
b CES-D scores are divided par the IQR.
*p value < 0.05, **p value < 0.01, ***p value < 0.001.
Discussion
This longitudinal population-based study aimed to examine the relationships between a history of SARS-CoV-2 infection and depressive symptoms according to self-report v. biological confirmation and across different dimensions of depressive symptoms. Taking into account both PCR and serology test results, participants with self-reported COVID-19 without biological confirmation and those with biologically confirmed COVID-19 had both higher levels of subsequent depressive symptoms. A gradient of depressive symptoms was observed from participants with no history of COVID-19 (lowest) to those with biologically confirmed COVID-19 (intermediate), and those with self-reported unconfirmed COVID-19 (highest). In addition, among the four CES-D specific dimensions, only the somatic complaints dimension was associated with a biologically confirmed history of COVID-19. The depressed affect dimension was associated with self-reported unconfirmed COVID-19 only. Altogether, these results are not consistent with the hypothesis of a direct impact of SARS-CoV-2 infection on the risk of depression.
This study has several strengths, including its population-based nature, the fairly large sample, and the longitudinal design. The dimensional assessment of depressive symptoms, with both total and specific dimensions, is another major strength of this study. Moreover, participants' SARS-CoV-2 infection status was measured with different methods (ad hoc serologic testing and/or self-reported positive PCR/serology test results). The time frame of the study makes the results relatively independent of the effects of viral variants that followed the wild type after the second wave of the pandemic.
This study has also limitations. First, the date of SARS-CoV-2 infection could not be precisely determined. Second, misclassification may have occurred regarding COVID-19 status. For instance, false positives may result from the combination of serologic testing low specificity with low prevalence of prior SARS-CoV-2 infection at the time of the study, especially in those not reporting COVID-19, as well as from low specificity of self-report, since other viruses may cause similar symptoms. On the other hand, false negatives may result from low sensitivity of PCR test results (encompassing participants who did not had PCR testing at the time of infection), especially in those reporting COVID-19, as well as from low sensitivity of self-report since SARS-CoV-2 infection could be asymptomatic in many cases (Sah et al., Reference Sah, Fitzpatrick, Zimmer, Abdollahi, Juden-Kelly, Moghadas and Galvani2021). Other limitations included the reliance on the CES-D, which is a well-validated scale in population-based studies but does not allow diagnosing major depression. Finally, of note, our sample could not be considered as representative of the general population, and the rate of SARS-CoV-2 infection in our sample was slightly lower than in the general population at the same time (10.5% in our sample v. 13.7% in the general population) (Salje et al., Reference Salje, Tran Kiem, Lefrancq, Courtejoie, Bosetti, Paireau and Cauchemez2020; Santé Publique France, 2023).
The fact that the somatic complaints dimension was, among the four considered depressive symptom dimensions, the most strongly associated with prior COVID-19, suggests that this association may be partially explained by overlapping somatic persistent symptoms following SARS-CoV-2 infection, such as those typically observed in long COVID (e.g. fatigue, poor attention/concentration, sleep disorders) (Pignon et al., Reference Pignon, Matta, Wiernik, Toussaint, Loewe, Robineau and Lemogne2024). For instance, the sickness behavior associated with circulating pro-inflammatory cytokines, which are under intense scrutiny in long COVID (Klein et al., Reference Klein, Wood, Jaycox, Dhodapkar, Lu, Gehlhausen and Iwasaki2023), also includes typical depressive symptoms such as fatigue, poor attention/concentration, social withdrawal, sleep disorders, loss of appetite, or anhedonia (Hart, Reference Hart1988; Shattuck & Muehlenbein, Reference Shattuck and Muehlenbein2016). However, the results of immunological studies in long COVID are not consistent regarding the nature of circulating pro-inflammatory cytokines or dysregulated immune cells (Patterson et al., Reference Patterson, Guevara-Coto, Yogendra, Francisco, Long, Pise and Mora-Rodríguez2021; Phetsouphanh et al., Reference Phetsouphanh, Darley, Wilson, Howe, Munier, Patel and Matthews2022; Schultheiß et al., Reference Schultheiß, Willscher, Paschold, Gottschick, Klee, Bosurgi and Binder2023). Besides, it is unclear whether the observed low grade inflammation relates to prior SARS-CoV-2 infection irrespective of the presence of persistent symptoms (Lund Berven et al., Reference Lund Berven, Selvakumar, Havdal, Stiansen-Sonerud, Einvik, Leegaard and Wyller2022; Sommen et al., Reference Sommen, Havdal, Selvakumar, Einvik, Leegaard, Lund-Johansen and Berven2022), long COVID (Klein et al., Reference Klein, Wood, Jaycox, Dhodapkar, Lu, Gehlhausen and Iwasaki2023), or associated features such as low physical activity (Gleeson et al., Reference Gleeson, Bishop, Stensel, Lindley, Mastana and Nimmo2011) or sleep disturbances (Irwin, Olmstead, & Carroll, Reference Irwin, Olmstead and Carroll2016). Furthermore, the results of the present study are indeed not consistent with a direct biological effect of SARS-CoV-2 infection on the risk of depression.
According to the hypothesis of a direct link between SARS-CoV-2 infection and depression, the gradient of depressive symptoms across the different COVID-19 status categories would have followed the probability of prior SARS-CoV-2 infection, which is arguably the lowest in participants who did not report COVID-19 (either suspected or biologically confirmed) and had negative SAPRIS serology test results, and the highest in those with biologically confirmed COVID-19, with intermediate probability in those with self-reported, yet unconfirmed COVID-19. However, depressive symptoms were higher among subjects with self-reported unconfirmed COVID-19 than among subjects with biologically confirmed COVID-19. One might argue than among those with biologically confirmed COVID-19, false positives among those who did not report COVID-19 but had a positive SAPRIS serology test results could have diluted the effect. However, when splitting our population into four groups according to self-report and SAPRIS serology test results, only participants with self-reported COVID-19 but negative SAPRIS serology test results displayed higher levels of depressive symptoms than those without COVID-19, while those with self-reported COVID-19 and positive SAPRIS serology test results did not, despite an undoubtedly higher probability of prior SARS-CoV-2 infection. These results are consistent with Thompson et al. (Reference Thompson, Stafford, Moltrecht, Huggins, Kwong, Shaw and Patalay2022) and Davisse-Paturet et al. (Reference Davisse-Paturet, Orri, Legleye, Florence, Hazo and Warszawski2023) studies.
One should keep in mind the clinical heterogeneity of the sample and the fact that few participants with SARS-CoV-2 infection and severe COVID-19 may have experienced encephalopathy (Liotta et al., Reference Liotta, Batra, Clark, Shlobin, Hoffman, Orban and Koralnik2020). These patients are at higher risk of subsequent psychiatric disorders, including depression (Granerod et al., Reference Granerod, Davies, Ramanuj, Easton, Brown and Thomas2017; Varatharaj et al., Reference Varatharaj, Thomas, Ellul, Davies, Pollak, Tenorio and Michael2020).
Overall, our results suggest that depressive episodes occurring after COVID-19 may not be attributable to a direct impact of SARS-CoV-2. Regarding clinical implications, they suggest that post-COVID-19 depressive episodes should be treated following current international guidelines for depressive disorder, without being misguided by biologically plausible but epidemiologically unsubstantiated hypotheses. Regarding implications for research, these results suggest that the exclusion of depressive symptoms from the definition of the post-COVID-19 condition might be worth considering. Informed by accumulating evidence from epidemiological studies (Tebeka et al., Reference Tebeka, Carcaillon-Bentata, Decio, Alleaume, Beltzer, Gallay and Coste2023), such exclusion would diminish the heterogeneity of this condition, and help the search for potential biomarkers and/or adequate treatments.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291724002435.
Author contributions
B. P. and C. L. participated in the conception and design of the study; E. W., F. C., G. S., M. T., S. K., M. G., and M. Z. participated in the acquisition of data; B. P. and C. L. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis; B. P. performed the analyses; B. P. and C. L. wrote the first draft of the manuscript. All authors participated in the interpretation of results, writing and revision of the successive drafts of the manuscript and approved the final version.
Funding statement
The CONSTANCES cohort benefits from grant ANR-11-INBS-0002 from the French National Research Agency. CONSTANCES is supported by the Caisse Nationale d'Assurance Maladie, the French Ministry of Health, the Ministry of Research, and the Institut National de la Santé et de la Recherche Médicale (INSERM). CONSTANCES is also partly funded by AstraZeneca, Lundbeck, L'Oréal, and Merck Sharp & Dohme Corp. The Santé, Pratiques, Relations et Inégalités Socials en Population Générale Pendant la Crise COVID-19 (SAPRIS) and SAPRIS-Sérologie (SERO) study was supported by grants ANR-10-COHO-06 and ANR-20-COVI-000 from the Agence Nationale de la Recherche; grant 20DMIA014-0 from Santé Publique France; grant 20RR052-00 from the Fondation pour la Recherche Médicale; and grant C20-26 from INSERM. The present study was supported by a grant ‘AAP Covid long 2022-1’ from the Agence nationale de recherches sur le sida et les hépatites virales (ANRS) | Maladies infectieuses émergentes. Dr Gouraud and Prof. Lemogne were supported by a grant from ‘la Fondation de l'Assistance Publique – Hôpitaux de Paris’.
Competing interests
Dr Robineau reported personal fees and nonfinancial support from Gilead, ViiV Healthcare, and Merck Sharp & Dohme Corp outside the submitted work. Dr Pitron reported personal fees from Grunenthal, outside the submitted work. Dr Lemogne reported non-financial support from Nordic Pharma, outside the submitted work. No other disclosures were reported.
Open access
