Prevalence of SARS-CoV-2 antibodies in the Mozambican population: a cross-sectional Serologic study in three cities, July-August 2020

Abstract Background The extent of population exposure to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was uncertain in many African countries during the onset of the pandemic. Methods We conducted a cross-sectional study and randomly selected and surveyed general population and occupational groups from July 6 to August 24, 2020, in three cities in Mozambique. Anti-SARS-CoV-2 specific immunoglobulins M and G antibodies were measured using a point-of-care rapid test. The prevalence was weighted for population (by age, sex, and city) and adjusted for test sensitivity and specificity. Results A total of 21,183 participants, including 11,143 from the general population and 10,040 from occupational groups, were included across all three cities. General population seropositivity (immunoglobulins M or G) prevalence was 3.0% (95% CI, 1.0–6.6) in Pemba, 2.1% (95% CI, 1.2–3.3) in Maputo City, and 0.9% (95% CI, 0.1–1.9) in Quelimane. The prevalence in occupational groups ranged from 2.8% (95% CI, 1.3–5.2) to 5.9% (95% CI, 4.3–8.0) in Pemba, 0.3% (95% CI, 0.0–2.2) to 4.0% (95% CI, 2.6–5.7) in Maputo City, 0.0% (95% CI, 0.0–0.7) to 6.6% (95% CI, 3.8–10.5) in Quelimane, and showed variations between the groups tested. Conclusions Exposure to SARS-CoV-2 was extensive during the first pandemic wave, and transmission may have been more intense among occupational groups. These data have been of utmost importance to inform public health intervention to control and respond to pandemic in Mozambique. Previous presentations of findings Results from this study were presented (in Portuguese) at the Mozambican Jornadas Nacionais de Saúde in Maputo, Mozambique on Aug 10, 2021 (abstract #108, title, Prevalência da exposição ao novo coronavírus em três cidades de Moçambique, Julho-Agosto de 2020).

importance to inform public health intervention to control and respond to pandemic in 23 Mozambique. On March 11 th , 2020 the World Health Organization announced that COVID-19 met 2 the definition of a pandemic [1,2]. As of March 1 st , 2022, there were more than 433 million 3 confirmed cases of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) 4 infections and over 5.9 million deaths globally [3]. In Mozambique, more than 225,000 5 individuals had been confirmed positive and 2,192 COVID-19-related deaths were reported 6 as of Mach 1 st , 2022 [4]. Despite the implementation of various interventions to control its 7 spread, SARS-CoV-2 infection has continued to steadily expand. The full burden of 8 infections is potentially underestimated due to mild or absent symptoms and limited country 9 molecular testing capacity [5]. conducting seroprevalence surveys to better understand the level of prior population exposure 16 to the virus, and identify higher-risk populations for prioritization for vaccination [10,12,13]. 17 However, representative studies about the prevalence of SARS-CoV-2 infection in African 18 countries remain limited [14-16] and so far no data are available for Mozambique. 19 To assess the seroprevalence of SARS-CoV-2 in the general population living in 20 locations of higher population density in Mozambique, as well as among key groups believed 21 to be at increased risk of infection due to their work or living conditions, we implemented 22 serological surveillance based on cross-sectional sampling of individuals from randomly 23 selected households in three provincial capitals and from occupational groups in those cities. 24

A C C E P T E D M A N U S C R I P T
We report the estimated prevalence of anti-SARS-CoV-2 immunoglobulins G (IgG) and M 1 (IgM) antibodies in three cities during the first pandemic wave.

3
Study design, population and sampling 4 We conducted a cross-sectional survey in the general population and selected 5 occupational groups believed to be at increased risk of SARS-CoV-2 infection (high-risk 6 groups) from three cities in Mozambique, in July and August, 2020. The selected cities were 7 Pemba (6-13 July), Maputo City (4-24 August) and Quelimane (10-21 August) in the 8 provinces of Cabo Delgado, Maputo City, and Zambézia, respectively. 9 In each city, in addition to including participants from households, we recruited 10 participants from occupational groups at their workplace. The general population sample was 11 selected through multistage sampling by the Mozambique National Institute of Statistics.

12
From every neighborhood in each city, a first stage sample of two to four blocks or segments 13 per neighborhood were selected with equal probability. This was followed by listing of all the 14 occupied households in the block or segment by the survey team. Immediately following the 15 listing, in the second stage, a fixed number of 16 households were randomly sampled within 16 each selected block using interval sampling with a random starting point. Any household 17 refusals led to selection of the next household in the list, and so on until 16 households were 18 included. From each head of household, the total household size was obtained and one 19 individual in each of three target age groups (0-17, 18-54, 55+ years) present at time of 20 interview was selected through convenience sampling. If individuals in the target age group 21 were not available for sampling or refused to provide a sample or to be interviewed, a 22 replacement member was selected from the household, or if unavailable or unwilling to 23 participate, from the next household in the list. Convenience sampling was used for the 24 occupational groups. First, sampling points for each occupational group were enumerated with assistance from local informants. Members of each occupational group were then 1 sampled among those present at the sampling point, at the time of the survey. These groups 2 comprised health professionals (physicians, nurses, health-care workers, pharmacy staff, 3 administrative staff, laboratory technicians, service agents, etc.), transport workers (bicycle 4 taxi, motorcycle taxi and car taxi drivers, urban, semi-collective transporter drivers, and their 5 ticket collectors, district/provincial transporters, and truck drivers), market vendors, 6 supermarket staff and defense and security forces. If participants refused to provide consent 7 for the interview or blood collection, they were excluded. Interviews and testing were done at 8 the place of work from which individuals were selected. 9 Sample Design 10 Generally, the goal was to obtain representative estimates at the neighborhood level 11 for each city, however sample size per city was determined based on availability of tests and 12 other resources such as lab supplies and human resources. The resulting sample design called 13 for 1,344, 9,360 and 4,800 individuals to be sampled in Pemba, Maputo City and Quelimane, 14 respectively. For key populations, the initial sample size of 2,800, 2,856 and 2,914 in Pemba, 15 Maputo City and Quelimane, respectively was defined based on available resources such as 16 rapid tests, laboratory supplies and interviewers in each city after accounting for the 17 community sample requirements. 19 Fieldwork was carried out by trained health workers from each of the study sites of all staff and fieldworkers. All field data collectors were tested three days before fieldwork 23 using the rt-PCR test and only participated in the study if the result was negative, and were 24 provided with personal protective equipment (gloves, surgical face masks, and hair covers

Study coordination and COVID-19 prevention measures
and face-shield) that were discarded and managed as hospital waste after each interview. 1 Study personnel were advised to conduct all study procedures outdoors where feasible.  After informed consent, participants were interviewed using pretested electronic 10 questionnaires to assess demographic characteristics, recent self-reported COVID-19-related 11 symptoms (i.e., fever, chills, severe tiredness, sore throat, cough, shortness of breath,   Table S1).  For the general population sample, design weights were developed separately for each 14 city using the population for each neighborhood by age from the 2017 population census.

15
Three weighting classes based on the recruitment age bands (0-17, 18-54, 55+ years) were 16 used. Due to the use of sampling with replacement at the household and individual level, non-17 response adjustments were not performed. Final weights were calibrated, post-stratified to the 18 total city population by age and sex and normalized. 19 The general characteristics of the study population were described for each individual 20 city. We estimated the population prevalence of exposure to SARS-CoV-2 for the general 21 population and at-risk populations by occupation and by city. The crude prevalence of 22 exposure was estimated as the proportion of individuals who had a positive rapid test result 23 for IgM, IgG or both. In the general population, the crude prevalence was then weighted to 24 adjust for population structure and adjusted for the corresponding locally derived test performance for the specific test used in each city. As a sensitivity analysis, these adjustments 1 were also done using the manufacturers' reported test sensitivity and specificity (see 2 Supplementary Appendix 1). The prevalence in occupational groups was reported as crude 3 and adjusted for test performance by city, by population. Percentages are reported to two 4 significant figures. Analyses were done using Stata 16.1 (StataCorp, College Station, TX, 5 USA) and R 4.0.2 (R Core Team, Vienna, Austria).

7
Of 15,504 sampled residents in the general population sampling, 11,143 participants 8 (72%) were recruited, of whom 59% were female. Although participants were sampled with 9 replacement, in some cases, particularly in Maputo City, it was not possible to reach the 10 target sample size due to higher rates of absence or refusal of household members. The  (Table 1). During fieldwork, the number of participants recruited from occupational groups in 17 Maputo was greater than originally planned, resulting in a total of 10,859 recruited 18 participants. Of these, 10,040 (92%) were included in the analysis across all three cities. 19 Eight-hundred and nineteen (7.5%) were excluded from analysis due to withdrawal of 20 consent after enrollment, having age (i.e., too young) inconsistent with the occupational 21 group they were enrolled in, or being from a group of fishermen, port, airports staff, reception 22 centers because they were only present in one of the cities (Figure 1).

11
Compared with that observed in the general population in each city, adjusted 12 seroprevalence was greater for most occupational groups, ranging from 2.8% to 5.9% in 13 Pemba, 0.8% to 4.0% in Maputo City and 0.0% to 6.6% in Quelimane (Table 3). In Pemba 14 and Maputo, cities with ongoing community transmission at the time of survey, the adjusted 15 prevalence among market vendors, 5.9% (95% CI 4.3-8.0) in Pemba and 4.0% (95% CI 2.6-16 5.7) in Maputo City, and health professionals, 5.0% (95% CI 3.0-7.7) in Pemba, were higher 17 than that observed in the community. In Quelimane, which had no community transmission 18 declared at the time of the survey, and with the apparent low prevalence in the general 19 population, transport workers were highly exposed to SARS-CoV-2 (6.6% [95% CI 3.8-20 10.5]) ( Table 3).

22
This study was the first of its kind in Mozambique and was conducted in the context 23 of urgent needs for epidemiological data to inform intervention strategies during the first 24 wave of the COVID-19 pandemic in the country. The seroprevalence of SARS-CoV-2- respectively. In Pemba, the first SARS-CoV-2 outbreak was observed in mining camps, 24 which may have contributed to increased community transmission in that city.  [36,37]. In this study, the adjusted seroprevalence of SARS-CoV-2 infection ranged from 5 0.0% to 6.6% among occupational groups across the study cities and in many cases was 6 found to be higher than that observed in the general population, which may be partially  The authors would like to warmly thank all study participants for their involvement. This    Contributions. PA conceived the study, study analysis plan, and wrote the manuscript. NM 22 co-conceived the study, secured seroprevalence testing, and supervised sample processing 23 and data preparation, co-wrote the manuscript. PWY assisted with data cleaning and analysis 24 planning, and manuscript writing. TT assisted with data cleaning and analysis planning, and 25 A C C E P T E D M A N U S C R I P T manuscript writing. IC undertook sample processing. NS co-conceived the study, selected 1 seroprevalence testing. AN conceived data collection tools, undertook data cleaning and 2 analysis. NI selected seroprevalence testing, supervised sample processing. AJ assisted with 3 data cleaning and analysis planning, contributed to manuscript writing. BC contributed to 4 sample estimation and selection of study sites. OFI contributed to conception of the study.

5
EG conceived the study, supervised data collection, and contributed to data interpretation. IJ 6 supervised the study conception, analysis plan, identified relevant external data, contributed 7 to data interpretation, and supervised manuscript writing.       Table 3. Seropositivity to SARS-CoV-2 in general population and occupational groups by 1 city, July-August 2020 (N=21,183).