Seroprevalence of COVID-19 infection in a rural district of South India: A population-based seroepidemiological study

Objectives We aimed to estimate the seroprevalence of COVID-19 in a rural district of South India, six months after the index case. Methodology We conducted a cross-sectional study of 509 adults aged more than 18 years. From all the four subdistricts, two grampanchayats (administrative cluster of 5–8 villages) were randomly selected followed by one village through convenience. The participants were invited for the study to the community-based study kiosk set up in all the eight villages through village health committees. We collected socio-demographic characteristics and symptoms using a mobile application-based questionnaire, and we tested samples for the presence of IgG antibodies for SARS CoV-2 using an electro chemiluminescent immunoassay. We calculated age-gender adjusted and test performance adjusted seroprevalence. Results The age-and gender-adjusted seroprevalence was 8.5% (95% CI 6.9%- 10.8%). The unadjusted seroprevalence among participants with hypertension and diabetes was 16.3% (95% CI:9.2–25.8) and 10.7% (95% CI: 5.5–18.3) respectively. When we adjusted for the test performance, the seroprevalence was 6.1% (95% CI 4.02–8.17). The study estimated 7 (95% CI 1:4.5–1:9) undetected infected individuals for every RT-PCR confirmed case. Infection Fatality Rate (IFR) was calculated as 12.38 per 10000 infections as on 22 October 2020. History of self-reported symptoms and education were significantly associated with positive status (p < 0.05) Conclusion A significant proportion of the rural population in a district of south India remains susceptible to COVID-19. A higher proportion of susceptible, relatively higher IFR and a poor tertiary healthcare network stress the importance of sustaining the public health measures and promoting early access to the vaccine are crucial to preserving the health of this population. Low population density, good housing, adequate ventilation, limited urbanisation combined with public, private and local health leadership are critical components of curbing future respiratory pandemics.

This was not a research grant but a philanthropic support for COVID response.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. June 2020. Written Informed consent was taken from all the participants before data collection. If the data are held or will be held in a public repository, include URLs, accession numbers or DOIs. If this information will only be available after acceptance, indicate this by ticking the box below. For example: All XXX files are available from the XXX database (accession number(s) XXX, XXX. We conducted a cross-sectional study of 509 adults aged more than 18 years. From all the four 31 subdistricts, two grampanchayats (administrative cluster of 5-8 villages) were randomly 32 selected followed by one village through convenience. The participants were invited for the 33 study to the community-based study kiosk set up in all the eight villages through village health 34 committees. We collected socio-demographic characteristics and symptoms using a mobile 35 application-based questionnaire, and we tested samples for the presence of IgG antibodies for 36 SARS CoV-2 using an electro chemiluminescent immunoassay. We calculated age-gender 37 adjusted and test performance adjusted seroprevalence.

82
There has been substantial evidence that a large proportion of the people infected with SARS 83 CoV-2 are asymptomatic, but they can infect others. It has been reported based on an analysis 84 of 21 published reports that asymptomatic cases could account from 5 to 80% [5]. It is crucial 85 to recognise an infected person early and break the route of transmission to control COVID-86 19. However, in reality, they do not require or seek medical attention and contribute to the rapid 87 spread of the disease [6]. Hence, health authorities cannot totally rely on confirmed cases of 88 COVID-19 detected by RT-PCR as it could potentially miss asymptomatic and pre-89 symptomatic infections for containment measures. In order to overcome this challenge, WHO 90 and others have recommended population-based seroepidemiological studies to generate data 91 and to implement containment measures accordingly [7]. These surveys also can give us an 92 estimation of the proportion of the population still susceptible to the infection as it is assumed 93 that antibodies provide immunity.  Hence, we designed a community-based cross-sectional study to estimate a seroprevalence in 112 Bangalore rural district six months after the index case. We also hope the findings of this study 113 will help the health authorities in disease containment and add valuable data to researchers 114 across the globe.

117
The study was conducted in Bangalore Rural District of Karnataka, a South Indian State. This 118 district is located close to Bangalore city and is divided into four sub-districts (taluks) and 105-119 gram panchayats which are village administrative units (Fig 1). Each gram panchayat has a 120 cluster of 10-15 villages [11]. According to the Indian Census 2011, the population was 121 9,90,923, and the sex-ratio was 945 females for every male, which is lesser than the state of International Airport in this district, service and Information technology industries are also 126 booming in the past few years [12]. Health care in the district is delivered through the 127 Government health system and private practitioners. Apart from primary health care, BBH also   (adults ≥18years) from houses to give blood samples. If a household refused to participate, then 145 the next house was approached. In each cluster, mobilisation continued till the desired sample 146 size was achieved in each village. We aimed to include 63 adults from each of these villages, 147 adding up to 126 in each sub-district. 148 We met with the village health committees and discussed the purpose of the study and enlisted 149 their cooperation. Together with the community, we decided that kiosk-based recruitment of 150 the participants was more practical due to strict restriction on the movement of the people by 151 the state government. People were also apprehensive about the health team from cities visiting 152 their homes and increasing the risk of transmission of the disease. 153 We recruited people after explaining the purpose of the study, took written consent and then 154 interviewed people with a questionnaire by a trained research coordinator who had previous 155 training in data collection. An Epi-info 7.0 TM mobile application-based tool was used to 156 capture responses offline by the interviewer, and it was later downloaded for analysis.

157
The questionnaire contained questions about demographic information (age, gender, education,  Sample processing and analysis 165 The serum was separated and used to test for antibodies using the Elecsys Anti SARS CoV2 166 assay, an electro chemiluminescent immunoassay using a recombinant protein representing the nucleocapsid (N) antigen for the determination of high-affinity antibodies (including IgG) 168 against SARS CoV2 [15]. This assay employs a cut-off index (COI) that is automatically 169 calculated from two calibration standards-a COI of 1.0 or more is considered 170 reactive/positive, and a COI less than or equal to 1.0 is reported as nonreactive/negative. The 171 assay sensitivity and specificity were reported to be 97·2% (95·4-98·4) and 99·8% (99·3-100)   Table).

181
Case-to-undetected-infections ratio (CIR), was calculated as a ratio of the number of reported 182 RT-qPCR-confirmed COVID-19 cases two weeks before the imitation of serosurvey to the 183 number of people who have antibodies in our study. This was based on an earlier study reported 184 median seroconversion times for total antibodies, IgM and then IgG at day-11, day-12 and day-185 14, respectively based on hospitalised patients and seroconversion for IgG and IgM is reported 186 to occur simultaneously or sequentially [18,19]. Assuming a three-week lag time from  Hypertension (20.2%) and diabetes (16.9 %) were reported as the most common comorbidities. (Table 1)   The overall seroprevalence of COVID-19 was 12.4% (95 % CI 9.6-15.6) (      We conducted a cross-sectional study of 509 adults aged more than 18 years. From all the four 31 subdistricts, two grampanchayats (administrative cluster of 5-8 villages) were randomly 32 selected followed by one village through convenience. The participants were invited for the 33 study to the community-based study kiosk set up in all the eight villages through village health 34 committees. We collected socio-demographic characteristics and symptoms using a mobile 35 application-based questionnaire, and we tested samples for the presence of IgG antibodies for 36 SARS CoV-2 using an electro chemiluminescent immunoassay. We calculated age-gender 37 adjusted and test performance adjusted seroprevalence.

82
There has been substantial evidence that a large proportion of the people infected with SARS 83 CoV-2 are asymptomatic, but they can infect others. It has been reported based on an analysis 84 of 21 published reports that asymptomatic cases could account from 5 to 80% [5]. It is crucial  Hence, we designed a community-based cross-sectional study to estimate a seroprevalence in 112 Bangalore rural district six months after the index case. We also hope the findings of this study 113 will help the health authorities in disease containment and add valuable data to researchers 114 across the globe.

117
The study was conducted in Bangalore Rural District of Karnataka, a South Indian State. This 118 district is located close to Bangalore city and is divided into four sub-districts (taluks) and 105-119 gram panchayats which are village administrative units (Fig 1). Each gram panchayat has a 120 cluster of 10-15 villages [11]. According to the Indian Census 2011, the population was 121 9,90,923, and the sex-ratio was 945 females for every male, which is lesser than the state of   (adults ≥18years) from houses to give blood samples. If a household refused to participate, then 145 the next house was approached. In each cluster, mobilisation continued till the desired sample 146 size was achieved in each village. We aimed to include 63 adults from each of these villages, 147 adding up to 126 in each sub-district. 148 We met with the village health committees and discussed the purpose of the study and enlisted 149 their cooperation. Together with the community, we decided that kiosk-based recruitment of 150 the participants was more practical due to strict restriction on the movement of the people by 151 the state government. People were also apprehensive about the health team from cities visiting 152 their homes and increasing the risk of transmission of the disease. 153 We recruited people after explaining the purpose of the study, took written consent and then    Table). The overall seroprevalence of COVID-19 was 12.4% (95 % CI 9.6-15.6) (  However, rural areas are challenged by the poor health system and low cash economy, 307 distancing itself from urbanisation reaped overall health benefits to people in villages, in terms 308 of the number of infections. This is a reminder that guarded urbanisation preserving the natural 309 ecosystem is an essential determinant of health. Thirdly, low seroprevalence should be looked in two ways. One way to look at this 320 'achievement', success in preventing the spread and the other way to look at it as 'responsibility' 321 due to susceptibility. Since we assume that other villages in India have similar or a slightly  The study has potential biases. Though all the subdistricts were selected, and subsequently 328 villages were randomly selected, we employed convenience sampling at the village level.

329
Villages were apprehensive about the medical team from the city, and hence we enrolled based 330 on individual preference. This would have resulted in selection bias; however, we tried to 331 reduce the bias by calculating age-gender adjusted seroprevalence. Another possibility is the 332 occurrence of measurement bias in estimating seroprevalence. Since we have not done RT-333 PCR, we would have missed the current infection and underestimated the prevalence.

334
Measurement bias can also be due to validity parameters of the test, which we have addressed 335 through test performance adjusted seroprevalence.

336
There are many strengths to this study. This is one of the earliest population-based 337 seroprevalence study conducted in a rural district of India harbouring a million people. This closely is an added advantage as the results are discussed in relation to the contextual realities.

341
This study has a few limitations. We did not follow a strict probability sampling technique due 342 to feasibility reasons. Another limitation is that we did not estimate the current infection using 343 RT-PCR. Both these aspects have an effect on the true estimation of seroprevalence in this 344 community. Though a 15 days recall period is generally recommended for eliciting morbidity, 345 we have used a longer (3 months) recall. Our assumption was that people would recall their 346 symptoms related to COVID for a longer period due to the unusually significant nature of this 347 pandemic and the attention it had received from media. However, this could have resulted in 348 recall bias. Another limitation is that we have limited our research to one rural district; hence 349 the generalisation of the findings has to be done with caution.