Higher SARS-CoV-2 seroprevalence in workers with lower socioeconomic status in Cape Town, South Africa

Background Inequality is rife throughout South Africa. The first wave of COVID-19 may have affected people in lower socioeconomic groups worse than the affluent. The SARS-CoV-2 seroprevalence and the specificity of anti-SARS-CoV-2 antibody tests in South Africa is not known. Methods We tested 405 volunteers representing all socioeconomic strata from the workforce of a popular shopping and tourist complex in central Cape Town with the Abbott SARS-CoV-2 IgG assay. We assessed the association between antibody positivity and COVID-19 symptom status, medical history, and sociodemographic variables. We tested 137 serum samples from healthy controls collected in Cape Town prior to the COVID-19 pandemic, to confirm the specificity of the assay in the local population. Results Of the 405 volunteers tested one month after the first peak of the epidemic in Cape Town, 96(23.7%) were SARS-CoV-2 IgG positive. Of those who tested positive, 46(47.9%) reported no symptoms of COVID-19 in the previous 6 months. Seropositivity was significantly associated with living in informal housing, residing in a subdistrict with low income-per household, and having a low-earning occupation. The specificity of the assay was 98.54%(95%CI 94.82%-99.82%) in the pre-COVID controls. Conclusions There is a high background seroprevalence in Cape Town, particularly in people of lower socioeconomic status. Almost half of cases are asymptomatic, and therefore undiagnosed by local testing strategies. These results cannot be explained by low assay specificity.


Please beef up substantially the description on the weakness of the study
Thank you for allowing us the opportunity to expand on this important section. The paragraph addressing the limitations of the study now reads: "An important limitation of this study is the moderate sample size which diminished the statistical significance of many of the findings. Furthermore, the design of the study was such that we could only sample selected participants from each sociodemographic and geographic group, who were not necessarily representative of the overall subpopulation they arose from. A further limitation of this sampling method is that it resulted in, unequal numbers from each district and occupational group, which made comparisons between groups difficult. In addition, our sample excluded those who are not part of the labour force, children, and the elderly and so can only be considered applicable to the economically active portion of the population. Our offer of serologic testing for SARS-CoV-2 before it was freely available may have introduced bias by selecting people who had a strong suspicion of having been infected. Nonetheless, we recruited a high proportion (405, 34%)  Thank you for this opportunity.
Our 'conclusion' paragraph has now been renamed 'Conclusion and recommendations'.
The following text has been added:

Please provide additional details regarding participant consent. In the ethics statement in the
Methods and online submission information, please ensure that you have specified: -whether consent was obtained -whether consent was informed -what type of consent you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed).
Thank you. The relevant paragraph of the Methods section now reads: "Each participant attended a single visit in which they underwent informed consent counselling,

provided written informed consent, and data on vitals and metrics, medical history, and COVID-19
exposure and symptoms were captured directly on to an electronic database." 3. Thank you for stating the following in the Acknowledgments Section of your manuscript: "We extend our thanks to the V&A Waterfront for proposing and funding the project, and to Mr Andre Theys, his administrative team, and the V&A community advisory board for the SU IRG: BMRI CT CoronaV&A Response to Reviewers P a g e 4 o f 15 warm welcome and continuous support throughout its execution."

We note that you have provided funding information that is not currently declared in your
Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.
Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows: "The authors received no specific funding for this work." Please include your amended statements within your cover letter; we will change the online submission form on your behalf.
Thank you for pointing this out. We welcome your advice on how to proceed with this declaration.
The V&A Waterfront is not a recognised funding agency, they are a commercial entity that assisted us in this study by providing the venue for recruitment and paying for the serology assays. Our own research group undertook half the cost of the study by providing the staff and logistical support. We share the publication costs.
We have amended the acknowledgements section to: "We extend our thanks to the V&A Waterfront for proposing the project, and to Mr Andre Theys, his administrative team…" With your approval, our new funding statement should read: "The project was co-funded by the V&A Waterfront Pty, Ltd and the Stellenbosch University Immunology Research Group." Overall, the paper is good and concise and brings valuable information.

Reviewers
We thank Reviewer 1 for the positive comments.

Reviewer #2:
In general, the paper is logically presented and is easy to follow. A few spelling and grammar mistakes are present as outlined in my comments below.
The title and abstract cover the main aspects of the work. The results uncover the high seroprevalence that are not detected by the current strategies, this is novel. This provides useful insights for strategies review. The methods are clear and replicable.
All the results presented match the methods described and statistical analysis were appropriate to the research question and study design. Data presentation was clear and easy to follow.

Control selection was appropriate for the study design.
There is correlation of results and conclusion. I am satisfied with the validity of the manuscript.
We thank Reviewer 2 for the positive comments.
A. Major Comment 1. No attempt has been made to address potential bias through analytic methods, e.g., sensitivity analysis.
Thank you for this comment. We performed a sensitivity analysis on the seven variables which were significantly associated with a positive serology test in the original analysis, i.e. 'fever or chills', 'muscle aches' 'loss of smell' and 'loss of taste', 'type of dwelling' and 'occupation'. We intentionally misclassified a proportion of the variables in five separate scenarios ranging between 5% and 25% misclassified.
We calculated the new p values using the harmonic mean p value method (HMP). In all of these five scenarios for all seven variables considered, only a single one became non-significant with intentional misclassification: 'muscle aches'. This variable was also identified when we originally adjusted for the multiple testing effect and was noted as being non-significant after applying the Holm method.
We did not initially perform a formal sensitivity analysis because it was felt that the likelihood of such bias in this study was very low. It was an observational study with no intervention, and a single time point, therefore no opportunity for protocol deviations which might affect the outcome. The variables were all categorical, apart from age and BMI -neither of which had serious outliers which might have biased the results. The accuracy of the antibody test we used is externally validated and the cut-off point predetermined, and so could not be manipulated by us.
In our analysis we considered both pooled results, and the subgroup analysis which we have reported in the paper for the following variables: medical history (all comorbidities pooled), symptoms of COVID-19, employment type and district of residence. The results were not significantly different between the pooled analyses and the subgroup analyses.
Moreover, the analysis was conducted with the assumption that all variables were non-normally distributed.
The proportion of missing data was minimal. The only variable which had any missing data was the 'district of residence', where 9 (2.2%) participants were unable to be classified.
Our initial and sensitivity analysis were performed using the Fisher's Exact Test, which was a deliberate choice as it is superior to Chi Squared when the cell numbers may be small.
We hope that the reviewers will be satisfied with the validity of our results on this basis. We have included the following sentence in the results section: Thank you for the comment. We have added the following text to the discussion: "…and the risks of continued use of public transport during lockdown, those infected were probably infected at home rather than the workplace. Thank you for the opportunity to add article which have been published subsequent to this study being written to our reference list and discussion.
The following addition has been made to the discussion: "A recent survey of 3098 Kenyan blood donors found an overall seroprevalence of 5.6% (population-weighted seroprevalence 4.3%), which was highest in those living in the three largest urban counties of Mombasa (8.0%), Nairobi (7.3%) and Kisumu (5.5%). (28)