key: cord-0830708-gzptpo6u
authors: Diegoli, H.; Conzatti, V. S.; Mazin, S. C.; Safanelli, J.; Delatorre, L. D. C.; Bett, K.; Radtke, R. E.; Wollmann, G. M.; Lima, H. d. N.; Franca, P. H. C. d.; Silva, J. R. d.
title: Population-Based Study of anti-SARS-CoV-2, Social Distancing and Government Responses in Joinville, Brazil
date: 2021-02-10
journal: nan
DOI: 10.1101/2021.02.08.21251009
sha: 8dcd97089cb316f3d6e769cfc920c227947ea907
doc_id: 830708
cord_uid: gzptpo6u

Background: The city of Joinville had been mildly affected by the COVID-19 pandemic until June 2020. This study aimed to longitudinally assess the prevalence of exposure to the virus and social distancing practices in the local population. Methods: A randomized selection of households stratified by region was created. From June 15 to August 7, 2020, a dweller was randomized in each household, answered a questionnaire, and performed a test for the detection of SARS-CoV-2 antibodies. The prevalence of positive tests was calculated for each week and adjusted for the test's sensitivity and specificity. Results: The adjusted proportion of positive results increased from 1.4% in the first week (margin of error [ME] 0% to 2.87%) to 13.38% in the eighth week (ME 10.22% to 16.54%). Among the 213 participants that tested positive, 55 (25.82%) were asymptomatic. Only 37 (17.37%) sought medical consultation for any symptom. Among the 77 (36.15%) that were leaving home to work or study, only 18 (23.38%) stopped due to any symptom. The proportion that referred going to bars, restaurants, or making non-essential shopping decreased from 20.56% in the first week to 8.61% during the peak of diagnoses. Conclusion: The low proportion of participants that sought medical consultation or stopped leaving home indicates strategies directed to isolate only those symptomatic reach a low proportion of infected patients.

Since March 2020, the SARS-CoV-2 infection became a disease with great concern for the Brazilian public health care system. In the first months, the country was more severely affected by the disease in the regions Southeast and North, with the cities of São Paulo and Manaus having a high death toll 1 Government responses to COVID-19 have varied worldwide, including enforced social distancing, testing, and contact-tracing, with varying drees of success [3] [4] [5] [6] .

During the early pandemic phase, the Brazilian Health Ministry's leadership was unstable, and there was a change of minister of health in April and another in May.

Decisions about social distancing were the responsibility of governors and mayors.

There was a serious concern with the possibility of pandemic spreading and uncertainties about when and how restrictive public measures could impact each city's epidemiological situation.

Considering the high uncertainty about the prevalence of exposure to the virus in developing countries, and that comprehensive testing can contribute to a better understanding of the local epidemiological scenario and the planning of public health care interventions, we developed a study to evaluate the serial prevalence of antibodies to SARS-CoV-2 and social distancing practices in representative samples of Joinville, Brazil.

JoinCOVID was a serial study composed of eight weekly cross-sectional studies that estimated the prevalence of contact with the SARS-CoV-2 virus through serologic tests from June 17 to August 7, 2020, in Joinville, a city in southern Brazil. Joinville is the third most populous city in the three southern states of Brazil. According to the last census, the city had about 598 thousand inhabitants in 2020 7 .

Every week, health care professionals and trained volunteers made telephone calls to households previously randomly selected. A resident was invited to answer a questionnaire that included social distancing practices and symptoms of COVID-19

and perform a serologic test in one of 13 health care centers. The study was approved by a local ethics committee (protocol number 37962620.6.0000.8062).

The epidemiological scenario and social distancing interventions before the study The first case of COVID-19 in Brazil was registered on February 25 1 . In a provisional measure taken by the federal judiciary in March, the state's and city's governments were considered responsible for social distancing decisions 8 public transportation, and other public services considered "non-essential". The use of face masks was made mandatory for work and commercial activities 9 . There was no prohibition of people's circulation in the city (what is often called "lockdown").

From March 23, the Municipal Health Secretariat started to offer medical evaluations to all inhabitants through telephone calls and messages, for suspicion of or any other medical reason, without out-of-pocket costs. A substantial part of Family Health Units and emergency departments' activities were directed for caring for people with symptoms of COVID-19, services that are also provided without cost.

Multi-stage sampling was used to select participants for the study. We used information provided by the city public institution that supplies water and sewage to define a sampling frame of the households. The institution contains data about 98,3% of the city's households 10 , being the most reliable database for the researchers.

The households were divided among the strata that correspond to the eight regions of the city and randomized proportionate to the number of inhabitants in each region.

A telephone call was performed to each household, and a resident was selected by simple randomization among those older than 18 months. Another qualified resident provided the answers in the case of children or participants with cognitive . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted February 10, 2021. ; impairments. A list of substitute households was ordered by zip code to guarantee the proximity of the initially randomized household. In case the phone number was invalid, the call was not answered, or the dwellers did not want to participate, the next household in the list of substitutes was called.

A team composed of volunteer healthcare students performed telephone calls, questionnaires, and serological test scheduling, supervised by healthcare professionals from the Health Secretariat. The questionnaire followed a protocol proposed by the World Health Organization 11 , with the addition of some questions to explore aspects related to social distancing practices. The questionnaire included symptoms, self-reported comorbidities, and practices of social distancing. In case the participant reported coughing, anosmia, coryza, or fever in the last three days, a new interview was scheduled after ten days to avoid exposure to other people during the transportation and serological testing. If the individual continued to present any symptoms in the follow-up call, the participant was excluded from the study and oriented to seek medical evaluation.

The tests were performed in 12 Family Health Units and the Center for COVID-19

Screening of the city.

We used the One Step COVID-19 Test . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint

The prevalence of positive cases for COVID-19 was adjusted using the formula 13 :

where the crude prevalence was the proportion of positive results, specificity was 99.57%, and sensitivity was 86.43%.

We performed deterministic sensitivity analyses to assess the extent to which the results were modified by changes in the test's sensitivity or specificity.

The chi-square test was applied to evaluate the difference between participants' characteristics with positive and negative tests. We considered p values below 0.05

to be statistically significant.

The data originating from JoinCOVID were graphically displayed and correlated with the local incidence of diagnosis and deaths by COVID-19 provided by the Health Department of Joinville 2 . Health professionals are required by law to report cases of COVID-19 confirmed by laboratory tests, and the data for correlation included all reported cases. The statistical analyses were conducted using Microsoft Excel 365 ® . . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint

The crude prevalence of positive tests varied from 1.63% in the first week (margin of error [ME] 0.05% to 3.22%) to 11.94% in the eighth week (ME 8.92% to 14.95%).

After adjusting for the test's sensitivity and specificity, the estimated prevalence changed to 1.4% (ME 0 to 2.89%) in the first week and 13.38% (ME 10.22% to 16.54%, Figure 1 ) in the last week. The observed increase in seroconversion presented a high correlation with the city's count of diagnosis and deaths by COVID- 19 . We estimate that one in every 5 to 10 of all estimated infections were reported. The deterministic sensitivity analysis indicated that the test's specificity was 98% or less, the prevalence of positives would be below zero in the first two weeks, which supports the test's high specificity. Of all individuals who tested positive, 77 (35.81%) referred they were working or studying outside their homes. Of these, only 18 (23.38%) stopped leaving their homes to work or study due to any symptoms since the pandemic. Of the 45 who . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint reported fever, coughing, anosmia or coryza, only 17 (37,77%) stopped leaving their homes to work or study at any moment.

The proportion of individuals that referred leaving home without a mask oscillated between 3.39% (ME 1.81% to 4.98%) in the first week and 1.69% (ME 0.59% to 2.78%) in the final week, reaching a maximum of 3.78% (ME 2.43% to 5.13%) in the third week (Figure 2) . The proportion of people that referred going to bars, restaurants, or non-essential shopping was 20.56% (ME 17.02% to 24.1%) in the first

week. This proportion started to reduce in the fourth week, reaching 8.61% (ME 6.23% to 10.99%) in the last week. Figure 4 displays the ICU bed occupancy and the date of governmental decrees with restrictions related to social distancing. The restrictions were imposed with the primary objective of avoiding a demand for ICU beds higher than the city's supply.

The first five weeks indicated a low prevalence of antibodies, with a rapid increase in the last three weeks. This increase accompanied a change in the epidemiological scenario observed through the number of new cases and deaths for COVID-19 and the demand for ICU beds.

A quarter of the participants who tested positive for SARS-CoV-2 did not report COVID-19 symptoms since March 2020, and only 15.49% had a previous laboratory diagnosis. Only 17.37% had sought medical consultation, even though they were available without out-of-pocket payment through various services. Also, less than one in every four participants who tested positive stopped leaving their homes to work or study when they had symptoms suggestive of COVID-19.

Most published studies reported antibodies in less than 4% of the population [14] [15] [16] [17] [18] . In some cities, a prevalence between 4 and 10% have been reported [19] [20] [21] , and some more heavily affected regions reported a prevalence above 10% [22] [23] [24] . One of the most . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint comprehensive studies aimed at obtaining a representative sample of the whole Spanish territory 25 . The study demonstrated that an average of 5% had developed antibodies to COVID-19, with a substantial variation among regions, ranging from 1.4% to 14.4%. 27 . It is estimated that around 40-45% of COVID-19 infections are transmitted by persons who do not exhibit symptoms 28 . However, to our knowledge, the prevalence of symptomatic patients that continue leaving their homes to work or study is not reported elsewhere.

The current study provides evidence that most people did not stop leaving their homes to work or study, even when they had symptoms of COVID-19. This finding may be explained because a high proportion of infected patients are asymptomatic, have unspecific symptoms, or because patients do not recognize mild symptoms as suggestive of COVID-19. Another significant issue is that workers may be concerned about stopping to work because of mild symptoms, either because they are paid per service or are worried about losing their jobs. Those concerns may be aggravated during an economic crisis. Of note, official estimates report that around 40% of workers in Brazil are informal 29 .

Those findings suggest that public health strategies directed only towards testing and isolating persons with a suspected infection are likely to reach a small portion of potential transmitters. Strategies that target all individuals, such as social distancing, face coverings, and hand hygiene, have more potential to reduce the virus's . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint transmission since they also encompass asymptomatic or symptomatic residents who do not seek health services.

Our study also provides an example of the weekly use of an epidemiological study for decision-making at the municipal level in a developing country. In a scenario where the mayors have a high degree of responsibility for social distancing practices and the funding, management, and provision of health care resources, JoinCOVID was a useful tool for better understanding the city's epidemiological scenario.

The information that we still had less than 2% of people exposed to COVID-19 by the end of June, three months after the first case, was critical in providing a picture of the long-term necessity of resources and that the worst period of transmission had not arrived yet. The proportion of positives also helped in estimating a benchmark for the number of recovered, allowing for the creation of more reliable Susceptible-Exposed-Recovered (SIR) models used for the estimation of ICU bed occupancy in the following weeks 30 .

By the end of June and during July, an increase in the transmission rate led to a maximum occupancy of ICU beds, and the region adopted new social distancing measures (Figure 4 ). Municipal and state public transportation was suspended on July 20. The city also imposed restrictions on activities in restaurants and limited the operation of commerce. The peak in diagnosis occurred on July 28, when JoinCOVID indicated that 8.37% of the sample had developed antibodies. On August 7, when the present study indicated a prevalence of 13.38%, the total number of deaths for COVID-19 was 158 2 . The efficacy of each government response implemented in the city is beyond the present article's scope. However, having longitudinal populationbased data about the prevalence of antibodies and the population's behavior was crucial in the decision-making process.

A limitation was that the scheduling process was carried out through telephone calls.

The process may generate a selection bias in favor of those who have telephone sets and are available to answer the calls. Besides, there was an . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; underrepresentation of residents between 0 and 29 years, and those between 30 and 69 years had a higher representation. On the other hand, an adjustment for age did not substantially change the proportion of residents with a positive test. Besides, the proportion that refused to participate remained similar between regions. The data's reliability is also strengthened by the high correlation between the total number of positive tests in the city and the prevalence of antibodies in the study.

In a population-based study of seroprevalence of antibodies to SARS-CoV-2, we found a prevalence ranging from 1.4% in the first week to 13.38% in the eighth week of testing.

Most residents who tested positive did not seek medical attention and continued leaving their homes to work or study. These data indicate a substantial difficulty in controlling the disease's spread through strategies targeted primarily at diagnosing and isolating residents with suspected disease, justifying more comprehensive measures that increase the social distance between all individuals.

JoinCOVID was a valuable tool to provide a clearer picture of the local epidemiological scenario for decision-makers and to justify the need and timing of decrees related to social distancing practices. 

The study follows all the recommendations from the Declaration of Helsinki and was approved by the corresponding ethics committee by the number 37962620.6.0000.8062.

The anonymized data supporting the findings of the study are available from the corresponding author upon reasonable request.

. CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The letters A to E in the graph indicate the time when each governmental decree was issued. The content of each decree was:

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint

Ministério da Saúde do Brasil. Coronavírus Brasil

Dados Casos Coronavírus Município de Joinville

Evaluation on different nonpharmaceutical interventions during COVID-19 pandemic: An analysis of 139 countries

Strong social distancing measures in the United States reduced the COVID-19 growth rate. Health Aff

On the effectiveness of containment measures in controlling the COVID-19 pandemic: the role of labour market characteristics and governance

An Examination on the Transmission of COVID-19 and the Effect of Response Strategies: A Comparative Analysis

Relatório Anual de Gestão

Câmara dos Deputados. Medida Provisória 926/2020

Diário Oficial dos Municípios de Santa Catarina

Cidade em Dados

Population-based age-stratified seroepidemiological investigation protocol for COVID-19 virus infection

Especial Vigilância Epidemiológica Laboratorial

Adjusting Coronavirus Prevalence Estimates for Laboratory Test Kit Error

Low prevalence of active COVID-19 in Slovenia: a nationwide population study of a probability-based sample

Performance characteristics of the abbott architect sars-cov-2 igg assay and seroprevalence in

Spread of SARS-CoV-2 in the Icelandic Population

Population Point Prevalence of SARS-CoV-2 Infection Based on a Statewide Random Sample -Indiana

Seroprevalence of immunoglobulin M and G antibodies against SARS-CoV-2 in China

Seroprevalence of SARS-CoV-2-Specific Antibodies Among Adults

Seroprevalence of SARS-CoV-2 and Infection Fatality Ratio, Orleans and Jefferson Parishes

Seroprevalence of Antibodies to SARS-CoV-2 in 10 Sites in the United States

Prevalence of Antibodies to SARS-CoV-2 in Italian Adults and Associated Risk Factors

High seroprevalence of anti-SARS-CoV-2 antibodies in Chelsea

Seroprevalence of anti-SARS-CoV-2 IgG antibodies in Geneva, Switzerland (SEROCoV-POP): a population-based study

Prevalence of SARS-CoV-2 in Spain (ENE-COVID): a nationwide, populationbased seroepidemiological study

SARS-CoV-2 antibody prevalence in Brazil: results from two successive nationwide serological household surveys

Proportion of asymptomatic coronavirus disease 2019: A systematic review and meta-analysis

Prevalence of Asymptomatic SARS-CoV-2 Infection: A Narrative Review

Instituto Brasileiro de Geografia e Estatística (IBGE)

COVID-19 Scenario Analysis Tool

. CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a 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 February 10, 2021. ; https://doi.org/10.1101/2021.02.08.21251009 doi: medRxiv preprint