Abstract
Objective
To examine the association between trust in different sources of information on COVID-19 at the beginning of the pandemic and the burden of incident persistent symptoms.
Methods
This prospective study used data from the SAPRIS and SAPRIS-Sérologie surveys nested in the French CONSTANCES population-based cohort. Trust in different information sources was measured between April 6 and May 4, 2020. Persistent symptoms that emerged afterwards were self-reported between December 2020 and January 2021. The associated psychological burden was measured with the somatic symptom disorder B criteria scale (SSD-12). The analyses were adjusted for gender, age, education, income, self-rated health, SARS-CoV-2 serology tests, and self-reported COVID-19.
Results
Among 20,985 participants [mean age (SD), 49.0 years (12.7); 50.2% women], those with higher trust in government/journalists at baseline had fewer incident persistent symptoms at follow-up (estimate (SE) for one IQR increase: −0.21 (0.03), p < 0.001). Participants with higher trust in government/journalists and medical doctors/scientists were less likely to have ≥1 symptom (odds ratio (95% confidence interval) for one IQR increase: 0.87 (0.82–0.91) and 0.91 (0.85–0.98), respectively). Among 3372 participants (16.1%) who reported ≥1 symptom, higher trust in government/journalists and medical doctors/scientists predicted lower SSD-12 scores (−0.39 (0.17), p = 0.02 and − 0.85 (0.24), p < 0.001, respectively), whereas higher trust in social media predicted higher scores in those with lower trust in government/journalists (0.90 (0.34), p = 0.008). These associations did not depend upon surrogate markers of infection with SARS-CoV-2.
Conclusions
Trust in information sources on COVID-19 may be associated with incident persistent symptoms and associated psychological burden, regardless of infection with SARS-CoV-2.
Keywords: Risk factors, Persistent symptoms, Post-COVID condition, Media
1. Introduction
Many patients affected by the coronavirus disease 2019 (COVID-19) suffer from persistent symptoms that may impair their quality of life months after being infected with SARS-CoV-2 [1,2]. The World Health Organization defined the ‘Post-COVID condition’ as the presence of symptoms that occur in the three months from the onset of a SARS-CoV-2 infection, persist for at least two months, affect everyday functioning, and cannot be explained by another diagnosis [2]. The promulgation of this definition is an important milestone in acknowledging the importance of ‘long COVID’. However, the lack of specificity of the symptoms and their default attribution to SARS-CoV-2 infection makes long COVID a very heterogeneous condition [1]. Some persistent symptoms may not be related to SARS-CoV-2 [[3], [4], [5]] and mechanisms not specific to SARS-CoV-2 should also be considered [6]. Furthermore, although even mild COVID-19 may be associated with long-term organ damages or physiological disturbances, there is often little or even no correlation between these findings and self-reported persistent symptoms [[7], [8], [9], [10], [11], [12]], suggesting that mechanisms other than biomedical could also be involved, at least for some symptoms in some patients [13,14].
Patients' beliefs and expectations may influence the experience of symptoms either directly, by affecting their perception [[15], [16], [17]] or indirectly, by shaping health behaviors [18,19]. In the context of COVID-19, it has been shown that beliefs regarding own's vulnerability may be associated with the likelihood of subsequent physical symptoms [20,21]. Since exposure to and trust in different information sources may influence patients' beliefs and expectations, we hypothesized that trust in different information sources regarding COVID-19 could be associated with subsequent persistent symptoms, beyond their potential association with actual infection (e.g., through adherence to preventive strategies) [19]. Such knowledge could usefully inform both preventive and therapeutic strategies.
This population-based prospective study examined the association between the degree of trust people had in different sources of information on COVID-19 (i.e., medical doctors, government, scientists, journalists, and social media) at the beginning of the pandemic first wave and the burden of incident persistent symptoms that emerged afterwards. We examined two main outcomes at follow-up, up to ten months after the assessment of trust in information sources: first, the number of self-reported incident persistent symptoms, which is typically high in patients with long COVID and correlated with the impact on their quality of life [22,23]; second, the associated psychological burden measured with the somatic symptom disorder B criteria scale (SSD-12). In addition, we aimed at examining whether any association between trust in sources of information on COVID-19 and the burden of incident persistent symptoms would depend upon surrogate markers of infection with SARS-CoV-2.
2. Methods
2.1. Data source
The French CONSTANCES population-based cohort study received ethical approval and included approximately 200,000 volunteers aged 18–69 years at inclusion between 2012 and 2019 and who gave informed consent to be followed up through annual questionnaires and linked administrative databases [24]. Volunteers were selected among individuals covered by the general insurance scheme or partner health mutual societies (in all, 85% of the French population) using a random sampling scheme stratified on place of residence, age, gender, occupation and socioeconomic status. Eligible individuals were invited to participate in the study by mail. Volunteers completed a self-administered questionnaire on socio-professional status, and attended a Health Screening Center for a comprehensive evaluation including a physical examination and laboratory tests. The CONSTANCES cohort study has received the authorization of the French Data Protection Authority (Commission Nationale de l'Informatique et des Libertés, CNIL) and the institutional review board of the National Institute for Medical Research (Inserm) (Authorization number 910486) and all methods were performed in accordance with the relevant guidelines and regulations.
A total of 35,852 volunteers 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” [25] and SAPRIS-Sérologie (SERO) surveys [26]. The SAPRIS survey was approved by the French Institute of Health and Medical Research ethics committee, and the SAPRIS-SERO study was approved by the Sud-Mediterranée III ethics committee. No one received compensation or was offered any incentive for participating in this study.
The present study is a longitudinal analysis of data from the SAPRIS and SAPRIS-SERO surveys nested in the French CONSTANCES cohort.
2.2. Trust in sources of information on COVID-19
Between April 6, 2020 and May 4, 2020, that is at the beginning of the pandemic in France, during the first lockdown (from March 17 to May 11, 2020), trust in five sources of information on COVID-19 was measured using the following question:
“To inform yourself on the coronavirus crisis, do you trust what they say: medical doctors, government, scientists, journalists, and social media?” For each of these five sources, answering options were: “Yes, absolutely”, “Rather yes”, “Rather not”, “No, not at all” and “I do not know”.
2.3. Serologic testing
Between May and November 2020, self-sampling dried-blood spot kits were mailed to each participant. Each kit included material (a dried-blood spot card, lancets, and a pad), printed instructions, and an addressed, stamped, and padded envelope to be returned with the card to a centralized biobank (CEPH 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, 87%; specificity, 97.5%). To reduce the risk of false-negative results, samples with indeterminate results (ie, optical density ratio ≥ 0.8 and < 1.1) were discarded. The participants received their serology test results by mail or email.
2.4. Self-reported COVID-19
Between December 2020 and January 2021, the participants answered this 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.” At the time they answered this question, the participants were aware of their serology test results. The participants who answered “Yes” additionally answered this question: “When did you get the coronavirus? Between March and June; In July or August; Between September and now.”
2.5. Incident persistent symptoms
In the same questionnaire (i.e., 7 to 10 months after collecting information on trust in information sources), incident symptoms were measured by the following question: “Since March 2020, have you had any of the following symptoms that you did not have before?” Based on early reports on ‘post-acute sequelae of COVID-19’ and ‘long COVID’ at the time of the study design, later confirmed in the literature [[27], [28], [29]], the following symptoms were explored: sleeping problems, joint pain, back pain, muscular pain, sore muscle, fatigue, poor attention/concentration, skin problems, sensory symptoms (pins and needles, tingling or burning sensation), hearing impairment, constipation, stomach pain, headache, breathing difficulties, palpitations, vertigo, chest pain, cough, diarrhea, anosmia, and other symptoms.
Two additional questions were asked for each reported symptom: “Has this symptom been present in the past four weeks?” Participants answered ‘yes, but not present anymore’, ‘yes, and still present’, or ‘no’; “How much time did this symptom last? Or how long has it been since you have had this symptom (if it is still present).” with possible responses ranging from ‘less than a week’ to ‘more than eight weeks’. Persistent symptoms were defined to a ‘yes’ and a ‘more than eight weeks’ response to these two questions. Participants with missing data for a particular symptom were considered as belonging to the ‘no’ category for this symptom.
2.6. Somatic symptom disorder B criteria scale (SSD-12)
In the same questionnaire, participants having reported at least one incident symptom were invited to fill the SSD-12 [30]. This 12-item scale was designed to assess the three aspects of the DSM-5 somatic symptom disorder B criteria, namely cognitive (e.g., “I think that my physical symptoms are signs of a serious illness”), affective (e.g., “I am worried that my physical symptoms will continue into the future”) and behavioral (e.g., “My health concerns hinder me in everyday life”). A ‘somatic symptom disorder’ is diagnosed when a person has a significant focus on physical symptoms, which may or may not be associated with a diagnosed medical condition, to a level that results in subtantial distress or functional impairment [30]. Here, we used the SSD-12 total score to assess the psychological burden associated with persistent symptoms (Cronbach's α = 0.88 in the current sample).
2.7. Covariates
Self-rated health at the beginning of the pandemic, as well as gender, age, education and income at inclusion in the CONSTANCES cohort were collected since they may influence trust in information sources.
2.8. Statistical analysis
Since we expect the five information source raw scores to be correlated, we a priori planned to use a principal component analysis with varimax rotation to obtain a smaller set of uncorrelated principal components. The categories of information sources were coded from 1 (“No, not at all”) to 4 (“Yes, absolutely”), so that a higher value corresponds to higher trust, excluding participants who answered “I don't know” for at least one source. The retained factors were divided by their interquartile ranges to ease the interpretation of the results.
First, we examined the association between trust in information sources at baseline and the number of incident persistent symptoms at follow-up (i.e., seven to ten months later). To focus on symptoms associated with long COVID, only symptoms previously associated with either self-reported COVID-19 or positive SARS-CoV-2 serology tests in the CONSTANCES cohort were considered [5]. Since reporting zero symptoms may result either from having no medical condition or from having an asymptomatic condition, we used zero-inflated negative binomial (ZINB) models to simultaneously examine the associations of the predictors with the number of persistent symptoms and the likelihood of zeros to be due to having no condition rather than an asymptomatic condition (Supplementary information 1) [31]. In the context of the present study, the ZINB regression distinguishes between two groups of participants: a group of participants who had the condition but may be asymptomatic (i.e., non-certain zeros) and another group of participants who did not have the condition at all (i.e., certain or ‘excess’ zeros). The ZINB regression models two separate processes so it produces two sets of coefficients: one for the ‘count’ part of the model, which may be interpreted as one would interpret coefficients from a standard negative binomial model (i.e., the expected number of symptom changes by a factor of exp(estimate) for each unit increase in the corresponding predictor); another for the ‘zero’ part of the model, which may be interpreted as one would interpret coefficients from a standard binary logistic regression model (i.e., the expected change in log-odds for each unit increase in the corresponding predictor variable). In other words, the zero model simultaneously estimates the associations of the predictors with zeros, in order to estimate the likelihood of zeros to be ‘certain’ or ‘excess’ zeros, that is, in the context of the present study, to be due to having no condition rather than an asymptomatic condition.
Second, the association between trust in information sources at baseline and the SSD-12 score at follow-up was examined with general linear models, among participants reporting at least one incident persistent symptom, adding the number of symptoms at follow-up as a covariate.
For both outcomes, different models were computed. Model 1 included trust in information sources (as factors derived from the principal component analysis), gender, age, education, income and self-rated health. Then interactions between the trust in information sources were tested.
Since some incident persistent symptoms may not be related to any infection with SARS-CoV-2 [[3], [4], [5]], we also aimed to examine whether the associations depend upon SARS-CoV-2 serological status or self-reported COVID-19. To this end, two other models were computed: Model 2A and 2B further adjusting Model 1 for serology test results or self-reported COVID-19, respectively. Then interactions of the trust in information sources with serology test results or self-reported COVID-19 were tested.
To test the robustness of our findings, sensitivity analyses were performed. First, we repeated the analyses, including principal component analysis, coding the “I don't know” response given to the questions about trust in information sources as intermediate responses between “Rather no” and “Rather yes”. Second, using incident persistent symptoms as a binary variable (i.e., at least one symptom versus none), we examined its association with trust in information sources with binary logistic regression models. Third, we repeated the analyses using the five information source raw scores instead of factors obtained through principal component analysis.
A two-sided value of P < 0.05 was considered statistically significant. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc).
3. Results
3.1. Participants
A total of 20,985 participants with complete data on information sources, gender, age, education, income, self-rated health, serology test results, and self-reported COVID-19, were included in the analyses [mean (SD) age, 49.0 (12.7) years; 10,530 women (50.1%) and 10,455 men (49.8%)] (Supplementary Fig. 1, Table 1 ). Among these participants, 702 (3.35%) reported having had COVID-19, 880 (4.19%) had positive serology test results, and 3372 (16.07%) reported at least one incident persistent symptom at follow-up. Among the 702 participants who reported having had COVID-19, 357 (50.85%) had a negative serology test result, while 345 (49.15%) had a positive serology test result. On the other hand, among the 880 participants who had a positive serology test result, 535 (60.80%) did not report having had COVID-19, while 345 (39.20%) did.
Table 1.
Descriptive characteristics of the sample (N = 20,985).
| Characteristic | N (%) of the population |
|---|---|
| Age, mean (SD), y | 49.00 (12.79) |
| Gender | |
| Female | 10,530 (50.18) |
| Male | 10,455 (49.82) |
| Monthly income, euros | |
| <450 | 63 (0.30) |
| 450 to <1000 | 230 (1.10) |
| 1000 to <1500 | 601 (2.86) |
| 1500 to <2100 | 1478 (7.04) |
| 2100 to <2800 | 2552 (12.16) |
| 2800 to <4200 | 6978 (33.25) |
| 4200 or higher | 9083 (43.28) |
| Educational level | |
| No diploma | 113 (0.54) |
| General education certificate, primary education certificate, school-leaving certificate | 649 (3.09) |
| Certificate of professional competence, vocational training certificate | 1873 (8.93) |
| Baccalaureate or equivalent diploma | 2852 (13.59) |
| Baccalaureate plus 2 or 3 y | 5839 (27.82) |
| Baccalaureate plus 4 y | 2305 (10.98) |
| Baccalaureate plus 5 y and more | 7354 (35.04) |
| Self-rated health (scale from 1 to 8), higher is better | |
| 1 | 59 (0.28) |
| 2 | 192 (0.91) |
| 3 | 293 (1.40) |
| 4 | 398 (1.90) |
| 5 | 821 (3.91) |
| 6 | 2950 (14.06) |
| 7 | 10,267 (48.93) |
| 8 | 6005 (28.62) |
| Sources of information on COVID-19, trust in: | |
| Medical doctors | |
| No, not at all | 65 (0.31) |
| Rather not | 440 (2.10) |
| Rather yes | 10,989 (52.37) |
| Yes, absolutely | 9491 (45.23) |
| Government | |
| No, not at all | 2626 (12.51) |
| Rather not | 5260 (25.07) |
| Rather yes | 10,972 (52.28) |
| Yes, absolutely | 2127 (10.14) |
| Journalists | |
| No, not at all | 3254 (15.51) |
| Rather not | 8305 (39.58) |
| Rather yes | 8816 (42.01) |
| Yes, absolutely | 610 (2.91) |
| Scientists | |
| No, not at all | 88 (0.42) |
| Rather not | 479 (2.28) |
| Rather yes | 11,115 (52.97) |
| Yes, absolutely | 9303 (44.33 |
| Social media | |
| No, not at all | 13,651 (65.05) |
| Rather not | 6630 (31.59) |
| Rather yes | 677 (3.23) |
| Yes, absolutely | 27 (0.13) |
| Self-reported COVID-19 | |
| Yes | 702 (3.35) |
| No | 20,283 (96.65) |
| Serology test results | |
| Positive | 880 (4.19) |
| Negative | 20,105 (95.81) |
| Number of incident persistent symptoms | |
| 0 | 17,606 (83.93) |
| 1 | 2119 (10.10) |
| 2 | 742 (3.54) |
| 3 | 258 (1.23) |
| 4 | 144 (0.69) |
| 5 and more | 109 (0.52) |
| Missing | 7 (0.03) |
| Incident persistent symptoms | |
| Cough (missing = 155) | 135 (0.62) |
| Breathing difficulties (missing = 262) | 208 (0.96) |
| Chest pain (missing = 269) | 139 (0.64) |
| Palpitations (missing = 375) | 172 (0.80) |
| Back pain (missing = 440) | 1309 (6.10) |
| Muscular pain, sore muscle (missing = 371) | 702 (3.26) |
| Headache (missing = 379) | 295 (1.37) |
| Anomaly of the facial nerves (missing = 396) | 12 (0.06) |
| Sensory symptoms (missing = 261) | 397 (1.84) |
| Speech problems (missing = 316) | 46 (0.21) |
| Nausea (missing = 230) | 54 (0.25) |
| Diarrhea (missing = 296) | 119 (0.55) |
| Constipation (missing = 571) | 316 (1.48) |
| Stomach pain (missing = 350) | 315 (1.46) |
| Anosmia (missing = 328) | 115 (0.53) |
| Fever or fever sensation (missing = 274) | 19 (0.09) |
| Fatigue (missing = 325) | 605 (2.81) |
| Poor attention or concentration (missing = 209) | 521 (2.40) |
| Dizziness (missing = 327) | 132 (0.61) |
| Discomfort (missing = 356) | 7 (0.03) |
| Other symptoms (missing = 135) | 337 (1.55) |
3.2. Trust in information sources on COVID-19 at baseline
As expected, raw scores of trust in information sources were correlated to some extent (Supplementary Table 1). After a principal component analysis with varimax rotation, the visual inspection of the scree plot and the interpretability of the rotated factor loadings suggested a three-factor solution (Table 2 ): medical doctors/scientists, government/journalists, and social media, which explained 22%, 11% and 8% of the variance, respectively. Sensitivity analyses accounting for the “I don't know” responses yielded similar results.
Table 2.
Trust in information sources on COVID-19 at the beginning of the pandemic: Factor patterns of principal component analysis with varimax rotation (N = 20,985).
| Factor 1 | Factor 2 | Factor 3 | |
|---|---|---|---|
| Trust in medical doctors | 0.89285 | 0.17495 | 0.01834 |
| Trust in government | 0.21456 | 0.85293 | −0.12489 |
| Trust in scientists | 0.88885 | 0.18599 | 0.01498 |
| Trust in journalists | 0.15314 | 0.78843 | 0.30640 |
| Trust in social media | 0.01319 | 0.07261 | 0.97263 |
Variance explained by factor 1: 22.0%.
Variance explained by factor 2: 11.0%.
Variance explained by factor 3: 8.2%.
3.3. Association with the number of incident persistent symptoms at follow-up
Participants with higher trust in government/journalists at baseline had fewer incident persistent symptoms at follow-up, whereas there was no association with trust in medical doctors/scientists or social media (Model 1, Table 3 ). Female gender and poorer self-rated health were also associated with more symptoms. Moreover, male gender and better self-rated health were associated with a higher likelihood for zeros to be due to having no condition rather than an asymptomatic condition. There was no interaction between the three factors of trust in information sources (all p > 0.10).
Table 3.
Association between trust in information sources on COVID-19 at the beginning of the pandemic and the number of incident persistent symptoms 7 to 10 months later; zero-inflated negative binomial regression models with the number of symptoms as the outcome (n = 20,978).
| Model 1 |
Model 2A |
Model 2B |
||||
|---|---|---|---|---|---|---|
| β (95% CI) | p | β (95% CI) | p | β (95% CI) | p | |
| Count modela | ||||||
| Medical doctors/scientists | −0.03 (−0.13, 0.06) | 0.50 | −0.03 (−0.13, 0.06) | 0.52 | −0.04 (−0.14, 0.05) | 0.38 |
| Government/journalists | −0.21 (−0.28, −0.15) | <0.001 | −0.21 (−0.28, −0.15) | <0.001 | −0.20 (−0.27, −0.14) | <0.001 |
| Social media | 0.01 (−0.08, 0.10) | 0.83 | 0.01 (−0.08, 0.10) | 0.80 | 0.005 (−0.09, 0.10) | 0.91 |
| Female gender | 0.30 (0.19, 0.41) | <0.001 | 0.29 (0.18, 0.41) | <0.001 | 0.30 (0.19, 0.41) | <0.001 |
| Age | 0.001 (−0.002, 0.006) | 0.43 | 0.003 (−0.001, 0.008) | 0.16 | 0.004 (−0.0004, 0.009) | 0.07 |
| Education | −0.008 (−0.04, 0.02) | 0.67 | −0.008 (−0.04, 0.02) | 0.66 | −0.01 (−0.05, 0.02) | 0.47 |
| Income | −0.03 (−0.07, 0.01) | 0.18 | −0.03 (−0.07, 0.01) | 0.16 | −0.03 (−0.08, 0.006) | 0.10 |
| Self-rated health | −0.22 (−0.26, −0.17) | <0.001 | −0.22 (−0.26, −0.17) | <0.001 | −0.20 (−0.25, −0.16) | <0.001 |
| Positive serology test results | 0.53 (0.31, 0.74) | <0.001 | ||||
| Self-reported COVID-19 | 0.73 (0.54, 0.93) | <0.001 | ||||
| Dispersionb | 2.37 (2.05, 2.75) | 2.31 (1.99, 2.69) | 2.20 (1.89, 2.55) | |||
| Zero modelc | ||||||
| Medical doctors/scientists | 0.12 (−0.16, 0.42) | 0.39 | 0.13 (−0.15, 0.42) | 0.36 | 0.10 (−0.17, 0.38) | 0.47 |
| Government/journalists | −0.17 (−0.35, 0.01) | 0.07 | −0.17 (−0.35, 0.01) | 0.07 | −0.15 (−0.33, 0.02) | 0.09 |
| Social media | 0.02 (−0.25, 0.30) | 0.87 | 0.03 (−0.24, 0.30) | 0.83 | 0.04 (−0.22, 0.30) | 0.77 |
| Female gender | −0.42 (−0.76, −0.07) | 0.02 | −0.40 (−0.75, −0.06) | 0.02 | −0.36 (−0.69, −0.04) | 0.03 |
| Age | −0.004 (−0.01, 0.009) | 0.54 | −0.003 (−0.01, 0.01) | 0.60 | −0.003 (−0.01, 0.01) | 0.63 |
| Education | −0.09 (−0.19, 0.01) | 0.08 | −0.08 (−0.19, 0.01) | 0.09 | −0.09 (−0.19, −0.0005) | 0.05 |
| Income | −0.02 (−0.15, 0.10) | 0.71 | −0.02 (−0.15, 0.10) | 0.68 | −0.02 (−0.14, 0.09) | 0.70 |
| Self-rated health | 1.11 (0.88, 1.34) | <0.001 | 1.09 (0.87, 1.32) | <0.001 | 1.04 (0.83, 1.26) | <0.001 |
| Positive serology test results | −0.31 (−0.95, 0.32) | 0.33 | 0.38 | |||
| Self-reported COVID-19 | −1.46 (−2.70, −0.22) | <0.001 | ||||
Interactions between medical doctors/scientists, government/journalists and social media factors were not significant (all p > 0.10).
Interactions between serology test results and information source factors were not significant (all p > 0.10).
Interactions between self-reported COVID-19 and information source factors were not significant (all p > 0.10).
Trust in information source factors have been divided by their IQR.
The count model estimates the associations of the predictors with the number of incident persistent symptoms, including zeros. Estimates and standard errors from this section may be interpreted as one would interpret coefficients from a standard negative binomial regression model (i.e., the expected number of symptoms changes by a factor of exp (estimate) for each unit increase in the corresponding predictor).
The variance of a negative binomial distribution is a function of its mean and of a dispersion parameter, which measures how much a sample fluctuates around a mean value.
The zero model simultaneously estimates the associations of the predictors with zeros, in order to estimate the likelihood of zeros to be ‘certain’ or ‘excess’ zero, that is, in the context of the present study, to be due to having no condition rather than an asymptomatic condition. Estimates and standard errors from this section may be interpreted as one would interpret coefficients from a standard binary logistic regression model (i.e., the expected change in log-odds for each unit increase in the corresponding predictor variable).
The association between higher trust in government/journalists at baseline and fewer symptoms at follow-up remained virtually unchanged when accounting for serology test results or self-reported COVID-19, whereas positive serology test results and self-reported COVID-19 were associated with more persistent symptoms and a lower likelihood for zeros to be due to having no condition (Model 2A and 2B, Table 3). There was no significant interaction between each factor of trust in information sources and serology test results or self-reported COVID-19 (all p > 0.10), suggesting that the magnitude of the association did not significantly differ according to surrogate markers of SARS-CoV-2 infection. In other words, participants with higher trust in government/journalists at baseline had fewer incident persistent symptoms at follow-up, regardless of infection with SARS-CoV-2.
Sensitivity analyses accounting for the “I don't know” responses regarding trust in information sources yielded similar results (Supplementary Table 2). Using incident persistent symptoms as a binary variable, participants with higher trust in medical doctors/scientists and government/journalists at baseline were less likely to have at least one symptom at follow-up (Table 4 ). Analyses using the raw score of the five trust in information sources showed that participants with higher trust in the government and journalists at baseline had fewer persistent symptoms at follow-up (Supplementary Table 3).
Table 4.
Association between trust in information sources on COVID-19 at the beginning of the pandemic and having at least one incident persistent symptom 7 to 10 months later, binary logistic regression models with having at least one symptom as the outcome (N event/participants = 3372/20,978).
| Model 1 |
Model 2A |
Model 2B |
||||
|---|---|---|---|---|---|---|
| OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | |
| Medical doctors/scientists | 0.91 (0.85, 0.98) | 0.01 | 0.91 (0.85, 0.98) | 0.01 | 0.91 (0.85, 0.99) | 0.01 |
| Government/journalists | 0.87 (0.82, 0.91) | <0.001 | 0.86 (0.82, 0.91) | <0.001 | 0.87 (0.83, 0.91) | <0.001 |
| Social media | 0.97 (0.91, 1.04) | 0.40 | 0.97 (0.90, 1.04) | 0.37 | 0.97 (0.90, 1.04) | 0.32 |
| Female gender | 1.40 (1.30, 1.51) | <0.001 | 1.39 (1.29, 1.50) | <0.001 | 1.39 (1.29, 1.50) | <0.001 |
| Age | 1.01 (1.00, 1.01) | 0.003 | 1.01 (1.00, 1.01) | <0.001 | 1.01 (1.00, 1.01) | <0.001 |
| Education | 1.03 (1.01, 1.06) | 0.02 | 1.03 (1.00, 1.06) | 0.03 | 1.03 (1.00, 1.06) | 0.03 |
| Income | 0.98 (0.95, 1.02) | 0.27 | 0.98 (0.95, 1.02) | 0.28 | 0.98 (0.94, 1.01) | 0.18 |
| Self-rated health | 0.72 (0.70, 0.74) | <0.001 | 0.72 (0.70, 0.74) | <0.001 | 0.72 (0.70, 0.75) | <0.001 |
| Positive serology test results | 1.76 (1.49, 2.08) | <0.001 | ||||
| Self-reported COVID-19 | 2.67 (2.26, 3.15) | <0.001 |
Interactions between the 3 factors in model 1 were non-significant (all p > 0.10).
Interactions between positive serology test result or self-reported COVID-19 and the three factors were non-significant in models 2A or 2B, respectively (all p > 0.10).
3.4. Association with the psychological burden of persistent symptoms at follow-up
Among participants with at least one incident persistent symptom at follow-up, those with higher trust in medical doctors/scientists and government/journalists at baseline had lower SSD-12 scores at follow-up (Model 1, Table 5 ). Being older, having a better self-rated health and fewer symptoms were also associated with lower SSD-12 scores. There was no interaction between trust in medical doctors/scientists and the two other factors (all p > 0.10). However, there was a significant negative interaction between trust in government/journalists and trust in social media (estimate (SE): −0.84 (0.31), p = 0.007). Indeed, higher trust in social media at baseline was associated with higher SSD-12 score at follow-up (estimate (SE): 0.90 (0.34), p = 0.008) in participants with lower trust in government/journalists (i.e., below the median value) but not in those with higher trust in government/journalists (i.e., above the median value).
Table 5.
Association between trust in information sources on COVID-19 at the beginning of the pandemic and psychological burden associated with incident persistent symptoms 7 to 10 months later; general linear regression models with the SSD-12 score as the outcome among participants with at least one incident persistent symptom (n = 3367).
| Model 1 |
Model 2A |
Model 2B |
||||
|---|---|---|---|---|---|---|
| Estimate (SE) | p | Estimate (SE) | p | Estimate (SE) | p | |
| Model 1 | ||||||
| Medical doctors/scientists | −0.85 (0.24) | <0.001 | −0.85 (0.24) | <0.001 | −0.85 (0.24) | <0.001 |
| Government/journalists | −0.39 (0.17) | 0.02 | −0.39 (0.17) | 0.02 | −0.40 (0.17) | 0.02 |
| Social media | 0.35 (0.23) | 0.14 | 0.35 (0.23) | 0.14 | 0.36 (0.23) | 0.13 |
| Female gender | 0.18 (0.26) | 0.49 | 0.18 (0.26) | 0.48 | 0.18 (0.26) | 0.49 |
| Age | −0.04 (0.01) | <0.001 | −0.04 (0.01) | <0.001 | −0.04 (0.01) | <0.001 |
| Education | −0.16 (0.09) | 0.08 | −0.16 (0.09) | 0.08 | −0.16 (0.09) | 0.08 |
| Income | 0.03 (0.11) | 0.77 | 0.036 (0.11) | 0.75 | 0.04 (0.11) | 0.71 |
| Self-rated health | −1.96 (0.10) | <0.001 | −1.96 (0.10) | <0.001 | −1.96 (0.10) | <0.001 |
| Number of persistent symptoms | 1.86 (0.10) | <0.001 | 1.87 (0.11) | <0.001 | 1.88 (0.11) | <0.001 |
| Positive serology test results | −0.26 (0.54) | 0.63 | ||||
| Self-reported COVID-19 | −0.78 (0.51) | 0.13 | ||||
Interactions between trust in medical doctors/scientists and the two other factors were not significant (p > 0.10).
There was a significant negative interaction between trust in government/journalists and social media (estimate (SE): −0.84 (0.31), p = 0.007 in Model 1).
There was no significant interaction between trust in information sources and serology test results or self-reported COVID-19 (all p > 0.10).
Trust in information source factors have been divided by their IQR.
Accounting for serology test results or self-reported COVID-19 did not change the magnitude of the estimates (Model 2A and 2B, Table 5). There was no significant interaction between trust in information source and serology test results or self-reported COVID-19 (all p > 0.10). In other words, participants with higher trust in medical doctors/scientists and government/journalists at baseline had lower SSD-12 scores at follow-up, regardless of infection with SARS-CoV-2.
Sensitivity analyses accounting for the “I don't know” responses regarding trust in information sources yielded similar results (Supplementary Table 4). Analyses using the raw scores of the five information sources showed that participants with higher trust in the government and medical doctors at baseline had lower SSD-12 scores at follow-up (Supplementary Table 5).
4. Discussion
This population-based prospective study aimed to examine the association between trust in different sources of information on COVID-19 at the beginning of the pandemic and the burden of incident persistent symptoms seven to ten months later. Participants with higher trust in government/journalists at baseline had fewer symptoms at follow-up. Among participants who reported incident persistent symptoms at follow-up, those with higher trust in government/journalists and medical doctors/scientists at baseline had a lower psychological burden, independently of the number of these symptoms. Furthermore, higher trust in social media was associated with higher psychological burden in those reporting lower trust in government/journalists. Even though these data were gathered in the context of the COVID-19 pandemic, it is noteworthy that these associations were independent of surrogate markers of past infection with SARS-CoV-2. Although effect sizes were relatively modest regarding the risk of incident persistent symptoms, the present results suggest that attitudes toward information sources on COVID-19 may be associated with the burden of incident persistent symptoms regardless of exposure to SARS-CoV-2, thus going beyond the issue of long COVID.
Interestingly, although trust in medical doctors/scientists was not associated with the number of incident persistent symptoms, it was strongly and negatively associated with the psychological burden associated with these symptoms among afflicted participants. A possible explanation is that trust in medical doctors/scientists may be more relevant for people actually facing persistent symptoms, thus more likely to seek medical care, than in other individuals. Trust in medical doctors/scientists was nonetheless associated with a lower risk of having at least one incident persistent symptom.
In addition, the psychological burden associated with incident persistent symptoms was also positively associated with trust in social media in afflicted participants reporting lower levels of trust in government/journalists. Since people with persistent symptoms may be more likely to consult and contribute to social media sources than those who recovered, health-related contents from social media may indeed promote negative expectations about the course of symptoms as well as maladaptive behaviors. Further studies are needed to examine whether this interaction might be explained by differences in trustworthiness of the social media sources participants were following according to their level of trust in government/journalists.
Even if our results are consistent with an influence of health-related information on the burden of incident persistent symptoms, this prospective association might be confounded by unmeasured variables. For instance, personality traits associated with a general tendency toward distrust have been linked to poor health outcomes [32,33]. Interestingly, when analysing the trust in information source raw scores, only trust in government was negatively associated with both outcomes. We hypothesize that this seemingly protective effect might be explained by individual differences in tolerance to uncertainty, as suggested by two unique features of government sources. First, government sources generally attempt to deliver consistent informations while other sources may convey more debated and contradictory messages (e.g., among scientists). Second, the government is expected to make statements only when a consensus has been reached, leaving areas of uncertainty as such. Uncertainty is a well-established cause of subjective distress that our brain is hardwired to reduce [34,35]. In the context of ambiguous information, attempts to reduce uncertainty may increase adherence to the information sources providing the most definitive claims, even at the expense of evidence [18]. Likewise, in the context of ambiguous perceptive situations, attempts to reduce uncertainty may lead our brain to overweight expectations at the expense of actual sensory inputs [16], leading to nocebo effects [15,20]. Higher intolerance to uncertainty may thus lead to both lower trust in government sources and higher burden of incident persistent symptoms. It may also explain why higher trust in social media was associated with heavier psychological burden only in individuals with lower trust in government/journalists. Finally, it could also partially be related to the excess of post-COVID condition observed after the first wave, arguably the most stressful period of the pandemic [36].
Incident persistent symptoms were quite frequent in this population, with one participant out of six having at least one incident persistent symptom at follow-up, about three times as many as those who either reported having COVID-19 or had a positive serology test. Although some infected participants may not have been identified, these figures show that most participants who experienced incident persistent symptoms during the first year of the COVID-19 pandemic were not infected with SARS-CoV-2 at the time of the study. This result is an important reminder that incident persistent symptoms may occur without any infection with SARS-CoV-2. However, should such symptoms occur within three months of COVID-19 infection, affect everyday functioning, and be of otherwise unknown origin, they would meet the definition of the ‘post-COVID condition’ [2]. Such default attribution to infection with SARS-CoV-2 is thus likely to result in an important clinical heterogeneity, which has to be overcome before any biomarker could be found [13,14].
The strengths of our study include the large size and the population-based nature of the sample, and the prospective design that allowed us to link data on trust in information sources at the beginning of the pandemic, when long COVID was barely known, to the burden of incident persistent symptoms seven to ten months later. Moreover, we have measured two complementaty aspects of this burden, that is both the number of symptoms and the associated psychological burden with a standardized questionnaire. In addition, adjusting for self-rated health, a sensitive indicator of global health, at the beginning of the pandemic, allowed us to partially control for this potentially confounding variable. To this end, self-rated health may be a better adjustment variable than any composite variable based on self-reported conditions that would not integrate perceived severity or could miss conditions that are not listed. It is noteworthy that vaccination adherence was unlikely to mediate the associations since the vaccination campaign started on December 27, 2020 in France. Finally, based on either SARS-CoV-2 serology test results or self-reported COVID-19, we were able to examine whether the associations were specific to post-acute symptoms of COVID-19 or rather a general phenomenon unrelated to any infection with SARS-CoV-2.
Limitations include the observational nature of the data which prevents drawing causal conclusions because of possible residual confounding by unmeasured variables. Second, the specificity of the results regarding trust in information about COVID-19 is unknown as no control question regarding trust in other contents was available. Third, actual exposure to the different information sources was not measured, nor was the type of social media consulted. Fourth, the three factors retained after the principal component analysis with varimax rotation accounted for only 41% of the variance. However, this a priori planned procedure allowed searching for interactions between uncorrelated factors and sensitivity analyses based on raw scores yielded consistent results. Fifth, the SSD-12 was designed to assess the DSM-5 Somatic symptom disorder B criteria, that is “excessive thoughts, feelings or behaviors related to the physical symptoms or health concerns” [30,31] so other aspects of the psychological burden associated with these symptoms might have been overlooked. Sixth, selection biases limit the representativeness of our sample. In addition, trust in sources of information on COVID-19 was measured during the first lockdown, which was a very particular period. Seventh, misclassification regarding the history of COVID-19 may have occured but using either self-reported COVID-19 or SARS-CoV-2 serology test results yielded similar results.
In summary, this study found a prospective association between trust in various sources of information and the incidence of persistent symptoms independently of surrogate markers of past infection with SARS-CoV-2. While observational studies may examine potential moderators and mediators of these associations, such as beliefs or health behaviors, experimental studies are needed to explore the relationships between exposure to and trust in different information sources and the burden of persistent symptoms in the general population, especially during major health crises. Such knowledge may eventually be useful for preventive or therapeutic strategies. It may also help to reduce the clinical heterogeneity of the post-COVID condition that is likely to encompass symptoms of various origins [13,14].
Funding
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 Dr. Lemogne were supported by a grant from “la Fondation de l'Assistance Publique - Hôpitaux de Paris”.
Role of funder statement
The funding source had no involvement in the study design; in the collection, analysis and interpretation of the data; in the writing of the report; and in the decision to submit the paper for publication.
Access to data and data analysis
JM and CL had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Authors contributions
JM and CL take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. CL designed the study; MG, MT, GS, and MZ acquired the data; JM and CL performed statistical analysis. All authors contributed to the interpretation of data. JM and CL drafted the article; EW, OR, GS, MT, CG, COV, VP, BR, NH, OVDB, MW, SK, MG, MZ revised it critically for important intellectual content.
All authors have read and approved the final manuscript. All authors confirm that they had full access to all the data in the study and accept responsibility to submit for publication.
Declaration of Competing Interest
All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf and declare that Olivier Robineau have reported personal fees and nonfinancial support from Gilead, ViiV Healthcare, and Merck Sharp & Dohme Corp outside the submitted work. The other authors have no competing interests to report.
Footnotes
Department where the work was conducted: Université Paris Cité, « Population-based cohorts Unit », INSERM, Paris Saclay University, UVSQ, UMS 011, Paris, France and Service de Psychiatrie de l'adulte, AP-HP, Hôpital Hôtel-Dieu, F-75004 Paris, France.
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jpsychores.2023.111326.
Appendix A. Supplementary data
Selection of the study population
Supplementary material 2
References
- 1.Wisk L.E., Nichol G., Elmore J.G. Toward unbiased evaluation of postacute sequelae of SARS-CoV-2 infection: challenges and solutions for the long haul ahead. Ann. Intern. Med. 2022;175(5):740–743. doi: 10.7326/M21-4664. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Soriano J.B., Murthy S., Marshall J.C., Relan P., Diaz J.V., Group WCCDW A clinical case definition of post-COVID-19 condition by a Delphi consensus. Lancet Infect. Dis. 2022 Apr;22(4):e102–e107. doi: 10.1016/S1473-3099(21)00703-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wisk L.E., Gottlieb M.A., Spatz E.S., Yu H., Wang R.C., Slovis B.H., et al. Association of Initial SARS-CoV-2 test positivity with patient-reported well-being 3 months after a symptomatic illness. JAMA Netw. Open. 2022;5(12) doi: 10.1001/jamanetworkopen.2022.44486. e2244486-e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Behnood S.A., Shafran R., Bennett S., Zhang A.X., O’Mahoney L.L., Stephenson T.J., et al. Persistent symptoms following SARS-CoV-2 infection among children and young people: a meta-analysis of controlled and uncontrolled studies. J. Inf. Secur. 2021;84(2):158–170. doi: 10.1016/j.jinf.2021.11.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Matta J., Wiernik E., Robineau O., Carrat F., Touvier M., Severi G., et al. Association of self-reported COVID-19 infection and SARS-CoV-2 serology test results with persistent physical symptoms among French adults during the COVID-19 pandemic. JAMA Intern. Med. 2022;182(1):19–25. doi: 10.1001/jamainternmed.2021.6454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Choutka J., Jansari V., Hornig M., Iwasaki A. Unexplained post-acute infection syndromes. Nat. Med. 2022;28(5):911–923. doi: 10.1038/s41591-022-01810-6. [DOI] [PubMed] [Google Scholar]
- 7.Lund Berven L., Selvakumar J., Havdal L., Stiansen-Sonerud T., Einvik G., Leegaard T.M., et al. Inflammatory markers, pulmonary function, and clinical symptoms in acute COVID-19 among non-hospitalized adolescents and young adults. Front. Immunol. 2022;13:837288. doi: 10.3389/fimmu.2022.837288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Douaud G., Lee S., Alfaro-Almagro F., Arthofer C., Wang C., McCarthy P., et al. SARS-CoV-2 is associated with changes in brain structure in UK biobank. Nature. 2022;604(7907):697–707. doi: 10.1038/s41586-022-04569-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Petersen E.L., Goßling A., Adam G., Aepfelbacher M., Behrendt C.-A., Cavus E., et al. Multi-organ assessment in mainly non-hospitalized individuals after SARS-CoV-2 infection: The Hamburg City Health Study COVID programme. Eur. Heart J. 2022;43(11):1124–1137. doi: 10.1093/eurheartj/ehab914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ladlow P., O’Sullivan O., Houston A., Barker-Davies R., May S., Mills D., et al. Dysautonomia following COVID-19 is not associated with subjective limitations or symptoms but is associated with objective functional limitations. Heart Rhythm. 2022;19(4):613–620. doi: 10.1016/j.hrthm.2021.12.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Havdal L.B., Berven L.L., Selvakumar J., Stiansen-Sonerud T., Leegaard T.M., Tjade T., et al. Neurological involvement in COVID-19 among non-hospitalized adolescents and Young adults. Front. Neurol. 2022;1127 doi: 10.3389/fneur.2022.915712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sneller M.C., Liang C.J., Marques A.R., Chung J.Y., Shanbhag S.M., Fontana J.R., et al. A longitudinal study of COVID-19 sequelae and immunity: baseline findings. Ann. Intern. Med. 2022 Jul;175(7):969–979. doi: 10.7326/M21-4905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Saunders C., Sperling S., Bendstrup E. A new paradigm is needed to explain long COVID. The lancet. Respir. Med. 2023;11(2) doi: 10.1016/S2213-2600(22)00501-X. [DOI] [PubMed] [Google Scholar]
- 14.Lemogne C., Gouraud C., Pitron V., Ranque B. Why the hypothesis of psychological mechanisms in long COVID is worth considering. J. Psychosom. Res. 2023;165:111135. doi: 10.1016/j.jpsychores.2022.111135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Bräscher A.-K., Raymaekers K., Van den Bergh O., Witthöft M. Are media reports able to cause somatic symptoms attributed to WiFi radiation? An experimental test of the negative expectation hypothesis. Environ. Res. 2017;156:265–271. doi: 10.1016/j.envres.2017.03.040. [DOI] [PubMed] [Google Scholar]
- 16.Van den Bergh O.W.M., Petersen S., Brown R.J. Symptoms and the body: taking the inferential leap. Neurosci. Biobehav. Rev. 2017;74(Pt A):185–203. doi: 10.1016/j.neubiorev.2017.01.015. [DOI] [PubMed] [Google Scholar]
- 17.Henningsen P., Gündel H., Kop W.J., Löwe B., Martin A., Rief W., et al. Persistent physical symptoms as perceptual dysregulation: a neuropsychobehavioral model and its clinical implications. Psychosom. Med. 2018;80(5):422–431. doi: 10.1097/PSY.0000000000000588. [DOI] [PubMed] [Google Scholar]
- 18.van der Linden S. Misinformation: susceptibility, spread, and interventions to immunize the public. Nat. Med. 2022;28(3):460–467. doi: 10.1038/s41591-022-01713-6. [DOI] [PubMed] [Google Scholar]
- 19.Moore C.A., Ruisch B.C., Granados Samayoa J.A., Boggs S.T., Ladanyi J.T., Fazio R.H. Contracting COVID-19: a longitudinal investigation of the impact of beliefs and knowledge. Sci. Rep. 2021;11(1):1–12. doi: 10.1038/s41598-021-99981-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rozenkrantz L., Kube T., Bernstein M.H., Gabrieli J.D. How beliefs about coronavirus disease (COVID) influence COVID-like symptoms?–a longitudinal study. Health Psychol. 2022;41(8):519. doi: 10.1037/hea0001219. [DOI] [PubMed] [Google Scholar]
- 21.Engelmann P., Löwe B., Brehm T.T., Weigel A., Ullrich F., Addo M.M., et al. Risk factors for worsening of somatic symptom burden in a prospective cohort during the COVID-19 pandemic. Front. Psychol. 2023:6730. doi: 10.3389/fpsyg.2022.1022203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Davis H.E., Assaf G.S., McCorkell L., Wei H., Low R.J., Re’em Y., et al. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EClinicalMedicine. 2021;38 doi: 10.1016/j.eclinm.2021.101019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tran V.-T., Riveros C., Clepier B., Desvarieux M., Collet C., Yordanov Y., et al. Development and validation of the long coronavirus disease (COVID) symptom and impact tools: a set of patient-reported instruments constructed from patients’ lived experience. Clin. Infect. Dis. 2022;74(2):278–287. doi: 10.1093/cid/ciab352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Zins M., Goldberg M. The French CONSTANCES population-based cohort: design, inclusion and follow-up. Eur. J. Epidemiol. 2015;30(12):1317–1328. doi: 10.1007/s10654-015-0096-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Carrat F., de Lamballerie X., Rahib D., Blanché H., Lapidus N., Artaud F., et al. Antibody status and cumulative incidence of SARS-CoV-2 infection among adults in three regions of France following the first lockdown and associated risk factors: a multicohort study. Int. J. Epidemiol. 2021;50(5):1458–1472. doi: 10.1093/ije/dyab110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Carrat F., Touvier M., Severi G., Meyer L., Jusot F., Lapidus N., et al. Incidence and risk factors of COVID-19-like symptoms in the French general population during the lockdown period: a multi-cohort study. BMC Infect. Dis. 2021;21(1):1–13. doi: 10.1186/s12879-021-05864-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Nehme M., Braillard O., Alcoba G., Aebischer Perone S., Courvoisier D., Chappuis F., et al. COVID-19 symptoms: longitudinal evolution and persistence in outpatient settings. Ann. Intern. Med. 2021;174(5):723–725. doi: 10.7326/M20-5926. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Al-Aly Z., Xie Y., Bowe B. High-dimensional characterization of post-acute sequelae of COVID-19. Nature. 2021;594(7862):259–264. doi: 10.1038/s41586-021-03553-9. [DOI] [PubMed] [Google Scholar]
- 29.Crook H., Raza S., Nowell J., Young M., Edison P. Long covid—mechanisms, risk factors, and management. BMJ. 2021:374. doi: 10.1136/bmj.n1648. [DOI] [PubMed] [Google Scholar]
- 30.Toussaint A., Murray A.M., Voigt K., Herzog A., Gierk B., Kroenke K., et al. Development and validation of the somatic symptom disorder–b criteria scale (SSD-12) Psychosom. Med. 2016;78(1):5–12. doi: 10.1097/PSY.0000000000000240. [DOI] [PubMed] [Google Scholar]
- 31.Hu M.-C., Pavlicova M., Nunes E.V. Zero-inflated and hurdle models of count data with extra zeros: examples from an HIV-risk reduction intervention trial. Am. J. Drug Alcohol Abuse. 2011;37(5):367–375. doi: 10.3109/00952990.2011.597280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Wong J.M., Na B., Regan M.C., Whooley M.A. Hostility, health behaviors, and risk of recurrent events in patients with stable coronary heart disease: findings from the heart and soul study. J. Am. Heart Assoc. 2013;2(5) doi: 10.1161/JAHA.113.000052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Iribarren C., Sidney S., Bild D.E., Liu K., Markovitz J.H., Roseman J.M., et al. Association of hostility with coronary artery calcification in young adults: the CARDIA study. Jama. 2000;283(19):2546–2551. doi: 10.1001/jama.283.19.2546. [DOI] [PubMed] [Google Scholar]
- 34.De Berker A.O., Rutledge R.B., Mathys C., Marshall L., Cross G.F., Dolan R.J., et al. Computations of uncertainty mediate acute stress responses in humans. Nat. Commun. 2016;7(1):1–11. doi: 10.1038/ncomms10996. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Anderson E.C., Carleton R.N., Diefenbach M., Han P.K. The relationship between uncertainty and affect. Front. Psychol. 2019;10:2504. doi: 10.3389/fpsyg.2019.02504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Scherlinger M., Lemogne C., Felten R., Sibilia J. Excess of post-acute sequelae of COVID-19 after the first wave of the pandemic. Infect. Dis. Ther. 2022:1–7. doi: 10.1007/s40121-022-00698-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Selection of the study population
Supplementary material 2