In the present study, we conducted a cross sectional survey to assess the seroprevalence of SARS-CoV-2 in people living in two urban areas of Tunis: El Omrane and La Goulette. The two study areas were selected because they expressed contrasted incidence (intermediate to low versus high) of COVID-19, based on the cumulative incidence since the pandemic’s onset. The weighted prevalence of seropositivity in the study population, defined by the detection of IgG antibodies against the N and/or the S-RBD proteins, was equal to 38.0% [34.6-41.5]. In multivariate analysis, we found that younger age, smoking status, previous confirmed COVID-19 infection, history of COVID-19 related symptoms and contact with a COVID-19 case within the household were independently and significantly associated with SARS-CoV-2 seropositivity.
Our study was conducted on March-April 2021 corresponding to the end of the second epidemic COVID-19 wave in Tunisia and the beginning of the third one. Hence and up to the study dates, the population has been exposed mainly to the wild original SARS-CoV-2 virus and to the alpha variant. Since then, the country was hit again by two additional much higher waves: the fourth wave on May-October 2021 mainly due to the circulation of the delta virus variant  (Figure 1) and the fifth wave starting in January 2022 due to the emergence of the Omicron virus variant.
Our results reveal that a large fraction (almost 40%) of the population of the study areas got infected after being exposed to just the second epidemic wave. In addition, the estimated seroprevalences were 34.5 and 14.8 times greater than the reported number of confirmed COVID-19 cases in “El Omrane” and “La Goulette” respectively. These figures stress the key role played by asymptomatic infection in SARS COV2 transmission and also illustrate the limits of case detection and contact tracing in the study areas. These shortcomes likely had severely jeopardised the impacts of individual preventive measures including isolation, in term of virus circulation. The prevalence figures also illustrate the high level of infection reached after an epidemic wave that was, all in all, relatively modest as compared to the fourth and fifth waves that hit the country in the following 12 months.
Noteworthy, as the cumulative COVID-19 incidence in the two studied areas was four times higher at the end of the fourth wave than after the second wave (Table 5), it seems intuitive that the proportion of individuals seropositive would by now exceed 100% if one consider high (>10) seroprevalence to incidence ratio revealed by our study as well as by others ; This means that many of the previously infected individuals will be re-infected (most likely silently) by SARS CoV-2 during the following waves, hence participating to maintaining high level of virus circulation and questioning the extent of the herd immunity in the exposed populations.
COVID-19 incidence in the studied areas during the survey period and after the subsequent waves
End of the second wave (Survey Period)
End of the third wave
End of the fourth wave
|*: per 100000 population
Surprisingly, we found that the SARS-CoV-2 seroprevalence was higher in the area of “low to intermediate” COVID-19 incidence (“El Omrane”) compared to that in high COVID-19 incidence (“La Goulette”). One possible explanation could be that “La Goulette” is a seaside city and some Tunisians living abroad and other tourists came on summer 2021 after reopening of the borders, to spend holidays there. If they got infected in Tunisia during their temporary stay, they will be registered as cases by the regional health directorates. Still, they were not included in our survey as we only considered permanent Tunisian residents.
Our results are in keeping with those reported at the global level in population surveys conducted among unvaccinated people and before the high circulation of the delta variant, which is known to be more contagious than previous variants  . Seropositivity rates reported worldwide [14–23], ranged from less than 2.6% in Sierra Leone  to 70.0% in Iquitos (Peru) (Annex 2). Such heterogeneity likely reflects differences in surveys design, dates of epidemic onset, the adherence of exposed populations to social restrictions and individual preventive measures applied in each country  and the type of laboratory test used.
Our result also corroborate previous studies mainly in the African continent in which a high underdetection and /or under-reporting of COVID-19 cases was noted [14,23,25–27]. This could be explained by the high percentage of COVID-19 asymptomatic cases. Indeed, we found that a large majority of seropositive participants (79.2%) didn’t develop any kind of COVID-19 related symptoms. Such high percentage of asymptomatic COVID-19 cases was also found in some other studies [28,29]. Nevertheless, a memory bias, which can lead to an overestimation of asymptomatic forms, cannot be eliminated in this survey. Indeed, participants were asked about their symptoms since the beginning of the COVID-19 pandemic in Tunisia. Such a large gap between the true number of SARS-CoV-2 infected persons and the declared cases of COVID-19 can also be explained by limited testing, fear of the disease, infection related stigma and, in some cases, conviction that COVID-19 does not exists [29–31].This emphasizes the need of amplifying testing efforts, case finding and contact tracing  ,mainly with the circulation of the new Omicron variant characterized by a very high proportion of asymptomatic cases . This is key to generating accurate data on SARS-CoV-2 in Tunisia and to implement evidence based public health measures to flatten the COVID-19 curve.
In this study, we found that age was independently associated with seropositivity. Indeed, children (age<10) had the highest percentage of IgG antibodies and the same trend is observed in the next age range (10-20). According to literature, youth are more likely to have social contacts than adults  and may be less adherent to barrier measures such as masking, hand hygiene and social distancing . Contacts in schools were also found to be more physical than those at workplace . Another explanation could be that children seem to have higher and more sustainable immune responses than adults . However, the findings of the present study do not support most of previous research surveys that found either a lower seroprevalence among youngest participants [26,37–39] or a non-significant difference according to age [29,40]. Such differences in results may be explained by the fact that most of the aforementioned studies that found a lower seroprevalence among youngest participants were conducted during the first wave of the pandemic, when majority of schools were closed unlike our study which was conducted in spring 2021 after schools’ reopening. Previous longitudinal studies have found increased SARS-CoV-2 seropositivity among children along with the overall transmission of COVID-19 [41–43]. In addition, with the emergence of the Omicron variant, a rise in COVID-19 pediatric cases was noted . This raises the concern of the potential influence of variants emergence on the transmission of SARS-CoV-2 among children.
We also found that none of seropositive participants aged under 10 years were previously diagnosed with COVID-19 which shows that the spread of SARS-CoV-2 among children and adolescents is extremely underestimated in Tunisia. Public health measures to decrease SARS CoV-2 transmission should thus include the entire population, and not only focus on adults . Non pharmaceutical intervention, including masking, hand hygiene and ventilation of indoor settings, should also be strengthened mainly in schools  since the under 18 years old are still not a priority targeted group for COVID-19 vaccination in Tunisia.
The Seroprevalence to SARS COV2 did not differ significantly according to sex in line with results of previous studies [18,40,46,47]. Moreover, our study together with previous ones [18,24,37,40,48], found a higher prevalence of antibody seropositivity among participants who report a history of COVD-19 like symptoms. Consistent with other studies [40,48], previous diagnosis of COVID-19 infection and contact with a COVID-19 case within the household were also independently and significantly associated with a higher percentage of SARS-CoV-2 IgG antibodies. However, surprisingly, we found that tobacco smokers had lower SARS-CoV-2 seroprevalence than non-smokers. A similar result was found by Alsuwaidi et al and Paleiron et al [18,49]. One hypothesis is that nicotine decreases the expression of the angiotensin converting enzyme 2 (ACE 2) which is a receptor of SARS-CoV-2. Another hypothesis is that SARS-CoV-2 and nicotine compete for binding to the nicotine acetylcholine receptor (nAChR) which is possibly involved in the physiopathology of COVID-19 infection . However, our results should be interpreted with caution as we conducted an observational study. Also, participants may underreport their tobacco consumption introducing a social desirability bias to the survey. Indeed, a relationship between smoking and increased risk of COVID-19 infection was underlined by a British study that used mendelian randomization analysis . In addition, evidence suggest that tobacco increases the risk of severe illness and deaths due to COVID-19 .
Finally, a non-significant association was found between seropositivity and the used means of transport. In accordance, an online survey conducted in France assessing factors associated with a higher risk to COVID-19 contagion, found that public transport was not associated with a higher risk of virus transmission unlike restaurants and bars .
Strengths and limitations:
This is the first study in Tunisia that reports the extent of the COVID-19 infection among both children and adults. As well, our study is, to the best of our knowledge, the first seroprevalence survey reported from countries in North Africa Our study was conducted at the nadir of the second epidemic wave that peaked on January 4, 2021 and the start of the third epidemic wave that peaked on April 15 2021 (Figure 1) . Importantly it took place just before the beginning of the COVID-19 vaccination campaign in Tunisia and hence the detected antibodies could be unequivocally attributed to natural SARS CoV-2 infection and not to vaccine administration. Also, the serum samples were tested using two in-house ELISA tests developed by IPT that detect with a high sensitivity and specificity anti-N and anti-S-RBD IgG antibodies, respectively. Indeed, anti-N IgG antibodies may appear before the anti-S-RBD  and the latter tend to wane at a slower rate than the anti-N antibodies . In fact, Schoenhals et al found a decrease of more than 10% in the percentage of anti-N IgG antibodies among seropositive blood donors during a three months follow up in Toamasina (Madagascar) . The detection in our study of antibodies to the two viral proteins gives a better chance to detect more infected cases than when using only one antigen.
Our study has some limitations. The ELISA tests detect SARS-CoV-2 antibodies which are known to be evanescent after natural infection and vaccination  In fact, the antibody decay after natural infection  may have minored the seroprevalence rates in our study population. Therefore, long term cohort longitudinal serological studies are warranted to assess the temporal dynamics of prevalence rates that integrate the opposing effects of natural antibody decay and the successive reinfections by different variants.
Our study was conducted only in the capital city Tunis. Larger populational serosurveys including other regions in Tunisia, would best describe the actual dynamics of the epidemics in the country according to the diversity of local conditions (ie.in rural areas and in the various country eco-climatic stages, effect of transborder human movements, population density etc.). Besides, assessing in addition to serology, the protective virus neutralizing antibodies as well as the cellular immune responses to the COVID-19 infection, would certainly improve the estimation of the actual proportion of population immune to SARS-COV-2.