key: cord-0746941-0d4dunph
authors: Kleynhans, J.; Tempia, S.; Wolter, N.; von Gottberg, A.; Bhiman, J. N.; Buys, A.; Moyes, J.; McMorrow, M. L.; Kahn, K.; Gomez-Olive, F. X.; Tollman, S.; Martinson, N. A.; Wafawanaka, F.; Lebina, L.; du Toit, J.; Jassat, W.; Neti, M.; Brauer, M.; Cohen, C.; Group, the PHIRST-C
title: Longitudinal SARS-CoV-2 seroprevalence in a rural and urban community household cohort in South Africa, during the first and second waves July 2020-March 2021
date: 2021-05-29
journal: nan
DOI: 10.1101/2021.05.26.21257849
sha: 175afe584c8aca58cd1562c479ec09907b43045e
doc_id: 746941
cord_uid: 0d4dunph

Background SARS-CoV-2 infections may be underestimated due to limited testing access, particularly in sub-Saharan Africa. South Africa experienced two SARS-CoV-2 waves, the second associated with emergence of variant 501Y.V2. In this study, we report longitudinal SARS-CoV-2 seroprevalence in cohorts in two communities in South Africa. Methods We measured SARS-CoV-2 seroprevalence two monthly in randomly selected household cohorts in a rural and an urban community (July 2020-March 2021). We compared seroprevalence to laboratory-confirmed infections, hospitalisations and deaths reported in the districts to calculate infection-case (ICR), infection-hospitalisation (IHR) and infection-fatality ratio (IFR) in the two waves of infection. Findings Seroprevalence after the second wave ranged from 18% (95%CrI 10-26%) and 28% (95%CrI 17-41%) in children <5 years to 37% (95%CrI 28-47%) in adults aged 19-34 years and 59% (95%CrI 49-68%) in adults aged 35-59 years in the rural and urban community respectively. Individuals infected in the second wave were more likely to be from the rural site (aOR 4.7, 95%CI 2.9-7.6), and 5-12 years (aOR 2.1, 95%CI 1.1-4.2) or [≥]60 years (aOR 2.8, 95%CI 1.1-7.0), compared to 35-59 years. The in-hospital IFR in the urban site was significantly increased in the second wave 0.36% (95%CI 0.28-0.57%) compared to the first wave 0.17% (95%CI 0.15-0.20%). ICR ranged from 3.69% (95%CI 2.59-6.40%) in second wave at urban community, to 5.55% (95%CI 3.40-11.23%) in first wave in rural community. Interpretation The second wave was associated with a shift in age distribution of cases from individuals aged to 35-59 to individuals at the extremes of age, higher attack rates in the rural community and a higher IFR in the urban community. Approximately 95% of SARS-CoV-2 infections in these two communities were not reported to the national surveillance system, which has implications for contact tracing and infection containment. Funding US Centers for Disease Control and Prevention

When comparing the infection fatality ratios for the first and second SARS-CoV-2 waves, at the urban 50 site, the ratios for both in-hospital and excess deaths to cases were significantly higher in the second 51 wave (0.36%, 95%CI 0.28-0.57% in-hospital and 0.51%, 95%CI 0.34-0.93% excess deaths), compared 52 to the first wave in-hospital (0.17%, 95%CI 0.15-0.20%) and excess (0.13%, 95%CI 0.10-0.17%) 53 fatality ratios, p<0.001 and p<0.001, respectively). In the rural community, the point estimates for 54 infection-fatality ratios also increased in the second wave compared to the first wave for in-hospital 55 deaths, 0.13% (95%CI 0.10-0.23%) first wave vs 0.20% (95%CI 0.13%-0.28%) second wave, and 56 excess deaths (0.51%, 95%CI 0.30-1.06% vs 0.70%, 95%CI 0.49-1.12%), although neither change was 57 statistically significant. 58

In South Africa, the overall prevalence of SARS-CoV-2 infections is substantially underestimated, 60 resulting in many cases being undiagnosed and without the necessary public health action to isolate 61 and trace contacts to prevent further transmission. There were more infections during the first wave 62 in the urban community, and the second wave in the rural community. Although there were less 63 infections during the second wave in the urban community, the infection-fatality ratios were 64 significantly higher compared to the first wave. The lower infection-hospitalisation ratio and higher 65 excess infection-fatality ratio in the rural community likely reflect differences in access to care or 66 prevalence of risk factors for progression to severe disease in these two communities. In-hospital 67 infection-fatality ratios for both communities during the first wave were comparable with what was 68 experienced during the first wave in India (0.15%) for SARS-CoV-2 confirmed deaths. To our 69 knowledge, these are the first longitudinal seroprevalence data from a sub-Saharan Africa cohort, 70 and provide a more accurate understanding of the pandemic, allowing for serial comparisons of 71 antibody responses in relation to reported laboratory-confirmed SARS-CoV-2 infections within 72 diverse communities. 73 74 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 ABSTRACT 75

Background 76 SARS-CoV-2 infections may be underestimated due to limited testing access, particularly in sub-77 Saharan Africa. South Africa experienced two SARS-CoV-2 waves, the second associated with 78 emergence of variant 501Y.V2. In this study, we report longitudinal SARS-CoV-2 seroprevalence in 79 cohorts in two communities in South Africa. 80

We measured SARS-CoV-2 seroprevalence two monthly in randomly selected household cohorts in a 82 rural and an urban community (July 2020-March 2021). We compared seroprevalence to laboratory-83 confirmed infections, hospitalisations and deaths reported in the districts to calculate infection-case 84 (ICR), infection-hospitalisation (IHR) and infection-fatality ratio (IFR) in the two waves of infection. 85

Seroprevalence after the second wave ranged from 18% (95%CrI 10-26%) and 28% (95%CrI 17-41%) 87 in children <5 years to 37% (95%CrI 28-47%) in adults aged 19-34 years and 59% (95%CrI 49-68%) in 88 adults aged 35-59 years in the rural and urban community respectively. Individuals infected in the 89 second wave were more likely to be from the rural site (aOR 4.7, 95%CI 2.9-7.6), and 5-12 years (aOR 90 2.1, 95%CI 1.1-4.2) or ≥60 years (aOR 2.8, 95%CI 1.1-7.0), compared to 35-59 years. The in-hospital 91 IFR in the urban site was significantly increased in the second wave 0.36% (95%CI 0.28-0.57%) 92 compared to the first wave 0.17% (95%CI 0.15-0.20%). ICR ranged from 3.69% (95%CI 2.59-6.40%) in 93 second wave at urban community, to 5.55% (95%CI 3.40-11.23%) in first wave in rural community. 94

Introduction 113

The first laboratory-confirmed case of coronavirus disease 2019 in South Africa was 114 announced on March 5, 2020, and the country has since experienced two waves of COVID-19. 1 A 115 nationwide lockdown from 27 March -30 April 2020 confined all persons to their homes (excluding 116 essential services), which was followed by a gradual easing of restrictions. 2 The second wave of 117 infections began in November 2020, 1 and the country instituted a less restrictive lockdown from 28 118

December 2020 -1 March 2021, which prohibited all gatherings with the exception of funerals and 119 mandated a night-time curfew. 2 Across Africa, the second wave was more severe than the first, 3 and 120 specifically in South Africa higher weekly incidence, hospitalisations and deaths were reported for 121 the second wave, compared to the first. 4-6 The second wave in South Africa was coupled with the 122 emergence of a new variant of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), 123 501Y.V2, also known as B.1.351. 7 124 South Africa reported more than 1.6 million laboratory-confirmed cases (by reverse transcription 125 polymerase chain reaction (RT-PCR) or rapid antigen tests) by mid-May 2021, 4 but many cases may 126 go undiagnosed due to mild or absent symptoms or lack of, or reluctance to access care or testing. 127

Data on the proportion of people with serologic evidence of prior SARS-CoV-2 infection are critical to 128 assess infection rates, calculate infection-hospitalisation and infection-fatality ratios, compare 129 infection burden between waves of infection and to guide public health responses. 8 Previous studies 130 have shown that SARS-CoV-2 seroprevalence is higher in close contacts of cases and high-risk 131 healthcare-workers, and lower in individuals younger than 20 years, or 65 years and older, with no 132 differences between males and females. 9 It is still unclear whether HIV-infection increases the risk 133 for SARS-CoV-2 infection, and results from studies thus far have varied. 10,11 134

We describe the seroprevalence of SARS-CoV-2 by age and HIV-infection status in two household 135 cohorts in a rural and an urban community at five time points from July 2020, during the first wave, 136

to March 2021, after the second epidemic wave. We also compare disease burden between the first 137 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 and second wave by comparing the seroprevalence by wave to reported laboratory-confirmed 138 infections, hospitalisations and deaths within the respective districts in which these two 139 communities are located. 140

We conducted a prospective study on a randomly selected household cohort in a rural (Agincourt, 143 Ehlanzeni District, Mpumalanga Province) and an urban (Jouberton, Dr Kenneth Kaunda District, 144 North West Province) community as part of the PHIRST-C study (a Prospective Household study of 145 SARS-CoV-2, Influenza, and Respiratory Syncytial virus community burden, Transmission dynamics 146 and viral interaction in South Africa). Methods for the cohort study are detailed in the appendix. 147

Recruitment to this study began in July 2020 and follow-up will continue through July 2021. 148

Households which previously participated in the PHIRST study (Prospective Household observational 149 cohort study of Influenza, Respiratory Syncytial virus and other respiratory pathogens community 150 burden and Transmission dynamics in South Africa) during 2016-2018, 12,13 and additional randomly 151 selected households were eligible. Households with three or more household members of any age 152 were enrolled if ≥80% of members consented. 153

The study was approved by the University of the Witwatersrand Human Research Ethics Committee 154 (Reference 150808) and the US Centers for Disease Control and Prevention relied on local clearance 155 (IRB #6840). 156

Seroprevalence was confirmed from medical records (if HIV-infected), and by rapid test for participants with 162 unknown, or self-reported negative status. Prior SARS-CoV-2 infection was determined using the 163 Roche Elecsys® Anti-SARS-CoV-2 assay (Roche Diagnostics, Rotkreuz, Switzerland), using 164 recombinant nucleocapsid (N) protein. The assay was performed on the Cobas e601 instrument, and 165 a cut-off index (COI) of ≥1.0 was considered an indication of prior infection (seropositivity). Signal to 166 cut-off ratio was not considered. Data analysis was performed in Stata 14 (StataCorp, College 167 Station, Texas, USA), using six age groups: pre-school (<5 years), primary school (5-12 years), 168 secondary school (13-18 years), young adults (19-34 years), adults (35-59 years) and older adults 169 (≥60 years) who are prioritised for SARS-CoV-2 vaccination. 14 Seroprevalence estimates were 170 adjusted for sensitivity and specificity as previous described, 15 based on the manufacturers' reported 171 99.5% sensitivity and 99.8% specificity. 16 Seroprevalence 95% credible intervals (95%CrI) were 172 obtained using Bayesian inference with 10,000 posterior draws. 15 Pearson's chi-squared test was 173 used to assess the statistical significance of differences in SARS-CoV-2 seropositivity across blood 174 collection times, waves of infection and between the two communities. 175

To assess the burden of SARS-CoV-2, and compare the severity of illness between the first and 178 second waves, we performed an ecological study comparing estimated number of infections based 179 on seroprevalence in our cohort study, to reported number of cases, hospitalisations and in-hospital 180 and excess deaths in the same district for each wave. We calculated the age-adjusted total number 181 of infections, laboratory-confirmed cases, hospitalisations, deaths, ICR (number of infections 182 compared to laboratory-confirmed cases), IHR and in-hospital and excess death IFR for each wave of 183 infection as described in below equations. The first wave was defined as 1 March 2020 (week 11) to 184 21 November 2020 (week 47) coinciding with the first case of SARS-CoV-2 reported in South Africa, 185 and ending the week before blood draw 3 started, and the second wave as 22 November 2020 186 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Where si is the seroprevalence in the cohort in the respective community and wave for age group i 201 and pi(SA) is the South African population for age group i. Calculated for wave 1 as seroprevalence at 202 blood draw 3, and for wave 2 as seroprevalence at blood draw 5, excluding those who seroconverted 203 at blood draw 3. Estimates only included participants with a blood draw 3 and 5 pair, and adjusted 204 for sensitivity and specificity of test. 15 205

Where ci is the number of laboratory-confirmed cases (RT-PCR and antigen-based tests) from the 207 respective district reported to the NMCSS (wave 1: 3 March -21 November 2020, wave 2: 22 208

November 2020 -27 March 2021) in age group i, pi(d) is the district population for age group i and 209 pi(SA) is the South African population for age group i.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ;

Confidence intervals for infection ratios were calculated as ratios from the 95% confidence intervals 233 of infection, hospitalisation and death rates. For international comparisons, we repeated the 234 calculations to standardise to the WHO standard world population, 18 using Sprague multipliers 19 to 235 expand age groups to one-year bands, and aggregate to the age groups used in this study. 236

We further compared the characteristics of participants who seroconverted during the first wave of 238 infections, to those who seroconverted during the second wave of infections using unconditional 239 logistic regression. We compared site, age, sex, HIV status, CD4 count, viral load, other underlying 240 medial conditions, body mass index, employment status, smoking and alcohol use in participants 241 that seroconverted in wave 1 (blood draw 3) to those who seroconverted in wave 2 (blood draw 5, 242 excluding draw 3 seroconversions). For this analysis, we only included participants with a blood 3 243 and 5 paired serum sample. For the multivariable model we assessed all variables that were 244 significant at p < 0.2 on univariate analysis, and dropped non-significant factors (p ≥ 0.05) with 245 manual backward elimination. 246

For all participants for which 5 sera were collected, and who seroconverted during draw 2 to 5, we 248 plotted COI values with the blood draw at which seroconversion took place as point 0. For those who 249 were seropositive at baseline we plotted the COI results from each blood draw. We calculated the 250 mean COI and exact 95% confidence interval at each point using the Clopper-Pearson method. We 251 assessed the percentage of participants with a COI ≥1.0 at each subsequent blood draw as the 252 number of participants with COI ≥1 divided by the seroconverted participants with a serum sample 253 at the time point. 254

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

In the rural community, we approached 185 households, 118 (64%) were enrolled and 641/692 257 (92%) of household members consented and/or assented to participate. In the urban community, 258 352 households were approached, 114 (32%) enrolled and 570/607 (93%) of household members 259 consented and/or assented. In both communities, the percentage of children, females, and 260 unemployed individuals included in the cohort were higher than in district census data (Appendix 261 Table 1 ). Median age was 13 (IQR 7-29) and 21 (IQR 10-43) years, and HIV prevalence was 14% 262 (95%CI 11-17%) and 18% (95%CI 14-21%) in the rural and urban communities, respectively. 263

During blood draw 1-5, blood draw coverage was 67% -88% and 84-88% in the rural and urban 265 communities, respectively (Appendix Table 2 ). The majority, 83% (n=553) of participants who lived in 266 the rural and 83% (n=499) in the urban community had both blood draw 3 and blood draw 5 blood 267 collected, with 56% (n=377) and 72% (n=431) of participants at the rural and urban site having all 5 268 bloods collected, respectively. 269 Seroprevalence, adjusted for assay sensitivity and specificity, in the rural community was lower at 270 blood draw 1 than in the urban community (1%, 95%CI 0-2% vs 15%, 95%CI 12-18%, p<0.001), 271 increasing after the first wave of infections (at blood draw 3) to 7% (95%CI 5-9) in the rural 272 community and 27% (23-31%) in the urban community (p<0.001, Figure 1 , Appendix Table 3 ). After 273 the second wave (blood draw 5), seroprevalence increased by 19% to reach 26% (95%CI 22-29%, 274 p<0.001) in the rural community, and by 14% to reach 41% (95%CI 37-45%, p<0.001) in the urban 275 community (Figure 1, Appendix Table 3 ). 276

At blood draw 5, seroprevalence was highest in the 19-34 years age group (37%, 95% CrI 28-47%) in 277 the rural community and the 35-59 years age group (59%, 95%CrI 49-68%) in the urban community 278 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101/2021.05.26.21257849 doi: medRxiv preprint (Figure 1, Appendix Table 3 ). The seroprevalence was lowest in children <5 years, 18% (95% CrI 10-279 26%) and 28% (95% CrI 17-41%) in the rural and urban communities, respectively. 280

At blood draw 5, SARS-CoV-2 seroprevalence was similar between HIV-infected and HIV-uninfected 281 participants in all age groups and in both communities (Appendix Table 4 ). For adults aged 19-34 282 years, the seroprevalence in HIV-uninfected individuals in the rural and urban community was 34% 283 (21/66) and 36% (26/73), respectively, compared to HIV-infected individuals at 43% (10/23) and 37% 284 (7/19) in rural and urban community, respectively (p=0.41 and p=0.92, Appendix Table 4 ). 285

Weekly incidence of reported laboratory-confirmed SARS-CoV-2 infections in the rural site district 288 peaked in the first wave at 68 cases per 100,000 population in week 31 of 2020, (starting 26 July), 289 and again at 165 cases per 100,000 in week 2 of 2021 (starting 10 January). In the urban site district, 290 the first wave peaked at 106 cases per 100,000 in week 30 of 2020 (starting 19 July), the second 291 wave at 79 cases per 100,000 in week 2 of 2021 (starting 10 January, Figure 1) . 292

During the first wave of infections (blood draw 3) in the rural community, 40/553 participants had 293 seroconverted resulting in an age-adjusted seroprevalence of 10% (95%CrI 5-17%) or incidence of 294 10,041 (95%CrI 4,759-17,088) per 100,000 population. Within the rural district, standardised to the 295 South African population, 557 laboratory-confirmed SARS-CoV-2 infections per 100,000 population, 4 296 75 COVID-19-related hospitalisations per 100,000 population, and 14 in-hospital deaths per 100,000 297 population 5 were reported by end of week 47 of 2020 (starting 15 November). Excess deaths of 51 298 per 100,000 population were reported for Mpumalanga Province. 6 Considering the 10% 299 seroprevalence at blood draw 3 in the rural community as a proxy for the district, the ICR was only 300 5.55% (95%CI 3.40-11.23%). There was a 0.75% (95%CI 0.49-1.41%) IHR and an in-hospital IFR of 301 0.13% (95%CI 0.10-0.23%) and an excess deaths IFR of 0.51% (95%CI 0.30-1.06%, Figure 2-3) . 302 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10. 1101 /2021 If excluding participants who had seroconverted after wave 1, there were 100/553 participants from 303 the rural community that had seroconverted at blood draw 5, an age-adjusted seroprevalence of 304 21% (95%CrI 13-30%) or incidence of 20,756 (95%CrI 13,007-30,011) per 100,000 population for the 305 second wave. The ICR was 4.00% (95%CI 2.86-6.17%), IHR 0.66% (0.50-0.98%), in-hospital IFR 0.20% 306 (95%CI 0.16-0.28%) and excess deaths IFR 0.70% (95%CI 0.49-1.12%, Figure 2-3) . 307

In the urban community the age-adjusted seroprevalence at blood draw 3 was 29% (95%CrI 21-40%) 308 or incidence of 28,419 (95%CrI 20,902-38,840) per 100,000 population. There was a 3.68% (95%CI 309 2.82-4.79%) ICR and 2.01% (95%CI 1.57-2.57%) IHR. The in-hospital IFR was 0.17% (95%CI 0.15-310 0.20%) and excess deaths IFR 0.13% (95%CI 0.10-0.17%, Figure 2 -3). During the second wave, the 311 age-adjusted seroprevalence in the urban community was 15% (95%CrI 8%-23%) or incidence 15,132 312 (95%CrI 8,181-22,977) per 100,000 population, resulting in a ICR estimate of 3.69% (95%CI 2.59-313 6.40%), IHR of 2.29% (95%CI 1.63-3.94%), in-hospital IFR of 0.36% (95%CI 0.28-0.57%) and an excess 314 deaths IHR of 0.51% (95%CI 0.34-0.93%, Figure 2-3) . These estimates standardised to the WHO 315 world population are presented in Appendix Figure 2 . 316

Compared to the urban community, individuals in the rural community who seroconverted were 4.7 318 (95%CI 2.9-7.6) times more likely to seroconvert during the second wave. Compared to those aged 319 35-59 years, individuals aged 5-12 years and ≥60 years were 2.1 (95%CI 1.1-4.2) and 2.8 (95%CI 1.1-320 7.0) times more likely to seroconvert in the second wave (Table 1) . 321

Of the 72 participants seropositive at the baseline blood collection, and with blood draw 1-5 samples 323 collected, 99% (71/72) still had a COI ≥1 by blood draw 5. The mean COI at baseline for seropositive 324 participants was 64, which increased to 125 at blood draw 2 and reduced to 59 at blood draw 5 325 ( Figure 4a ). The participant who no longer had detectable SARS-CoV-2 antibodies at the fifth draw 326 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101/2021.05.26.21257849 doi: medRxiv preprint had a starting COI of 9, which subsequently declined to 0.7 six months after the first draw. Of the 327 210 participants with blood draw 1-5 samples, 99% (167/169), 99% (70/71%) and 93% (41/44) still 328 had a COI ≥1 in the first, second and third blood draw after initial seroconversion, respectively 329 ( Figure 4b) . The participants who seroreverted had starting COIs ranging from 2 -6. The mean COI at 330 the point of seroconversion was 48, which increased to 86 at the first blood post seroconversion and 331 reduced to 61 at the third draw post seroconversion. 332

We assessed SARS-CoV-2 seroprevalence in 1,211 individuals living in two diverse communities in 334

South Africa and show that laboratory-confirmed cases reported from study districts greatly 335 underestimate the true burden of SARS-CoV-2 infections. At baseline, seroprevalence was 1% and 336 15%, increasing to 7% and 27% after the first wave, and by March 2021. Following the second 337 epidemic wave, seroprevalence was 26% and 41% in the rural and urban communities, respectively. 338

The highest seroprevalence was 59% in adults aged 35-59 years in the urban community, and the 339 lowest was 18% in rural community children <5 years. During the second wave, compared to the first 340 wave, the rural site was more affected, and infections in the second wave more likely affected 341 children aged 5-12 years and adults ≥60 years. We observed no differences in seroprevalence by HIV 342 status. In the urban community, the IFR was higher in the second wave (0.36-0.51%), compared to 343 the first (0.13-0.17%), even though numbers of infections were lower, suggesting possible increased 344 severity associated with the emergence of novel variant 501Y.V2. Most individuals who 345 seroconverted maintained detectable SARS-CoV-2 antibodies in subsequent serum samples. 346

Low seropositivity was observed at the rural site at baseline, and the seroprevalence remained low 347 until blood draw 3, only reaching seroprevalence of 7% after the first wave of infections, which was 348 considerably lower than the seroprevalence of 27% at the urban site at the same time. This could 349 possibly be related to the relatively isolated location and lower population density in the rural 350 community compared to more densely populated urban community. The seroprevalence in the rural 351 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 site increased to 26% at blood draw 5, which was after the second wave of infections within the 352 district. This could have been due to possible increased transmissibility of the 501Y.V2 lineage that 353 was circulating in the second wave, 20 as well as additional transmission networks in the community 354 during the December holiday period when large scale urban-to-rural migration takes place as people 355 return home for the yearend holidays. The urban site had fewer seroconversions in the second wave 356 compared to the first, which may be due to existing immunity among individuals in the community 357 after the first wave. As seen in previous studies, 9 adults had the highest seroprevalence levels, 358

although there was still a relatively high seroprevalence of 18% and 28% in children <5 years at the 359 rural and urban community respectively. 360

There are few data available on seroprevalence in Africa, most studies have focused on specialised 361 groups. A study conducted among blood donors in South Africa during the second wave of the 362 pandemic found a seroprevalence of 32% to 63%, in five provinces of South Africa that have both 363 rural and urban communities. 21 In our study, enrolling a random sample of community members, we 364 observed a seroprevalence in adults ranging between 25-37% in rural households, and 35-59% in the 365 urban households, suggesting that seroprevalence is heterogeneous between communities. 366 Seroprevalence data from sub-Saharan Africa are also limited. In Kenya the seroprevalence in blood 367 donors during the country's first wave of infections was 4%, and was also higher in urban 368 communities. 22 In a population-level household sero-survey in Zambia during their first wave of 369 infections, 11% of individuals had evidence of SARS-CoV-2 infection. 23 After the first wave, we 370 estimated a higher seroprevalence in the rural (7%) and in the urban communities (27%) compared 371

to Kenya 22 and a higher seroprevalence in the urban community than in Zambia. 22,23 372 Based on our estimates, only 4-6% of cases were laboratory confirmed. Our study suggests 373 substantially higher prevalence of infection ascertained through serology and that the differences 374 may have been greater in the urban community than the rural community; however, more extensive 375 studies are needed to assess whether this is consistent in other areas. Compared to the urban 376 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101/2021.05.26.21257849 doi: medRxiv preprint community, the rural community had a higher proportion of SARS-CoV-2 infections reported to the 377 NMCSS (4-6% vs 4%) and less than half the rate of hospitalisation (0.8% vs 2-3%). These may be due 378 to differences in referral and testing policies, health-seeking behaviour, access to care and 379 differences in circulating lineages within these districts. 380

A study comparing the severity of the first and second waves of infections in South Africa in 381 hospitalised patients found a higher mortality rate in the second wave, compared to the first, even 382 controlling for increased pressure on health services which was also associated with increased 383 mortality. 24 At the urban site, the IFR was significantly higher in the second wave (0.43-0.50%) 384 compared to the first (0.13-0.20%), although no differences were observed in IHR between the two 385 waves. The lower overall number of infections in the second wave in this site means that our finding 386 of increased mortality is unlikely to be related to pressure on health services. The increased severity 387 of the second wave may be related to increased severity of the 501Y.V2/B.1.135 variant, but further 388 studies are needed to confirm this. The excess death IFR during the first wave in the urban site was 389 smaller than the in-hospital IFR. This may be due to uncertainty on the process for excess death 390 estimation, or that the 85% contribution of COVID-19 to excess deaths was an underestimation 391 within the province. However, the in-hospital IFR followed the same trend of increase between wave 392 1 and 2 (0.13% to 0.51%). 393

Considering the high seroprevalence observed, the age standardised IHR (wave 1: 0.8% rural, 2% 394 urban, wave 2: 0.8% rural, 2.58% urban) and IFR (wave 1: 0.2-0.5% rural, 0.1-0.2% urban, wave 2: 395 0.2-0.4% rural, 0.4-0.5% urban) were lower compared to the non-age standardised estimates from 396 the USA, where the IHR and IFR was estimated as 2% and 1% respectively, 25 and from Italy where IFR 397 was estimated as 0.9%. 26 Our first wave in-hospital IFR estimates (0.13% rural, 0.17% urban) were 398 more similar to the age-adjusted 0.15% reported from India for the first wave SARS-CoV-2-confirmed 399 deaths, 27 and our first wave excess death IFR was higher in the rural (0.51%) and lower in the urban 400 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101/2021.05.26.21257849 doi: medRxiv preprint (0.13%) community compared to the age-adjusted 0.28% IFR excess deaths reported from Brazil 401 during their first wave of infections. 28 402

We did not observe a difference in SARS-CoV-2 seroprevalence in HIV-infected and -uninfected 403 individuals at either site. Due to HIV causing immune suppression, there is a concern that HIV-404 infected individuals may be more susceptible to SARS-CoV-2 infection. 10,11 A meta-analysis including 405 data from North America, Africa, Europe, and Asia found a 24% higher risk of symptomatic SARS-406

CoV-2 infection in HIV-infected individuals, albeit with high heterogeneity between countries, 11 407 whereas a case-control study from the United States of America found a lower SARS-CoV-2 408 seroprevalence in HIV-infected individuals. Although HIV infection may not increase susceptibility to 409

infection, it has been demonstrated to be a risk factor for developing severe COVID-19 and death 410 following infection. 10,29 411 Our study is limited by a relatively small sample size, reducing the power for accurate 412 seroprevalence estimates in small age strata; and including only two geographic sites, and therefore 413 may not be representative of other districts and provinces in South Africa. The seroprevalence 414 reported here may be an under-estimate of population-level infection rates as not everyone infected 415 with SARS-CoV-2 develops antibodies. ICR, IHR and IFR formed part of an ecological analysis which is 416 inherently prone to biases. Excess deaths in the first wave may be underestimated since the 417 reporting period only stared in June, and were reported at provincial-level which may be different to 418 within the district. Similarly, transmission dynamics within our cohort may not be similar to the 419 district which formed the comparison point for our case, hospitalisation and in-hospital deaths. The 420 ELISA utilised to detect SARS-CoV-2 antibodies was qualitative, and not suited for quantitative 421 analysis of antibody levels. Ongoing follow-up of this cohort will track future infections and monitor 422 antibody waning, and compare these data to laboratory confirmed infections and symptoms from 423 twice weekly follow-up. A strength of our study is the collection of samples from prospectively 424 followed up individuals from randomly selected households within the study communities and 425 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 inclusion of individuals of all ages. To our knowledge, this is the first seroprevalence data from a 426 cohort in South Africa, which provides the advantage of serial comparisons of antibody responses in 427 relation to reported laboratory-confirmed SARS-CoV-2 infections within the community through two 428 successive SARS-CoV-2 waves. 429

We estimate that approximately 95% of SARS-CoV-2 infections in these two communities were not 430 laboratory-confirmed and reported to the national surveillance system, which has major implications 431 

The investigators welcome enquiries about possible collaborations and requests for access to the 531 dataset. Data will be shared after approval of a proposal and with a signed data access agreement. 532

Investigators interested in more details about this study, or in accessing these resources, should 533 contact the corresponding author. 534

The findings and conclusions in this report are those of the authors and do not necessarily represent 536 the official position of the CDC. 537

Cheryl Cohen reports receiving grant funds from US-Centers for Disease Control and Prevention, 539

Wellcome Trust and South African Medical Research Council. 540 541 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Vertical lines represent 95% confidence interval. Wave 1: 1 March -21 November 2020, wave 2: 22 580

November 2020 -27 March 2021. 581 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Figure 4 . Cut-off index (COI) on Roche Elecsys® Anti-SARS-CoV-2 assay for individuals with blood 583 draw 1 to 5 (BD1 to BD5) samples who a) were seropositive at baseline b) seroconverted during 584 blood draw 2-5, July 2020 -April 2021, South Africa. 585

Mean COI with 95% confidence interval in purple line. COI values in b aligned to first draw prior to 586 seroconversion. 587 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 29, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

COVID-19 Testing Summary

South African Government. COVID-19 Risk Adjusted Strategy

National Institute for Comminicable Diseases. COVID-19 Weekly Epidemiology Brief Week 445 19

National Institute for Comminicable Diseases. COVID-19 Sentinel Hospital Surveillance 448

Update Week 18

SARS-CoV-2 501Y.V2 escapes neutralization by South

African COVID-19 donor plasma. bioRxiv : the preprint server for biology

COVID-19 serosurveys for public health decision making. The 456

Serological evidence of human infection with SARS-CoV-2: a 458 systematic review and meta-analysis

SARS-CoV-2 seroprevalence, and IgG concentration and 460 pseudovirus neutralising antibody titres after infection, compared by HIV status: a matched case-461 control observational study

Epidemiology and outcomes of COVID-19 in 463

HIV-infected individuals: a systematic review and meta-analysis

Cohort Profile: a Prospective Household 465 cohort study of Influenza, Respiratory Syncytial virus, and other respiratory pathogens community 466 burden and Transmission dynamics in South Africa (PHIRST)

Asymptomatic transmission and high community 469 burden of seasonal influenza in an urban and a rural community in South Africa

Statistical release P0302 -Mid-year population estimates 2020

Age standardization of 482 rates: a new WHO standard. [Geneva]: World Health Organization

Department of Commerce BotC. The methods and 484 materials of demography

Emergence and rapid spread of a new severe 486 acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) lineage with multiple spike 487 mutations in South Africa

Prevalence of anti-SARS-CoV-2 antibodies among blood 489 donors in Northern Cape

Seroprevalence of anti-SARS-CoV-2 IgG antibodies in 492

Kenyan blood donors

Prevalence of SARS-CoV-2 in six districts in Zambia in 494

a cross-sectional cluster sample survey. The Lancet Global Health 2021

Increased mortality among individuals hospitalised 496 with COVID-19 during the second wave in South Africa

Estimation of US SARS-CoV-2 Infections

Infections, Hospitalizations, and Deaths Using Seroprevalence Surveys

All individuals participating in the study, field teams for their hard work and dedication to the study,