SARS-CoV-2 Seroprevalence in Tamil Nadu in October-November 2020

Abstract

A population-representative serological study was conducted in all districts of the state of Tamil Nadu (population 72 million), India, in October-November 2020. State-level seroprevalence was 31.6%. However, this masks substantial variation across the state. Seroprevalence ranged from just 11.1% in The Nilgris to 51.0% in Perambalur district. Seroprevalence in urban areas (36.9%) was higher than in rural areas (26.9%). Females (30.8%) had similar seroprevalence to males (30.3%). However, working age populations (age 40-49: 31.6%) have significantly higher seroprevalence than the youth (age 18-29: 30.7%) or elderly (age 70+: 25.8%). Estimated seroprevalence implies that at least 22.6 million persons were infected by the end of November, roughly 36 times the number of confirmed cases. Estimated seroprevalence implies an infection fatality rate of 0.052%.

Full text

1
SARS-CoV-2 Seroprevalence in Tamil Nadu in October-November 2020
Anup Malani, Sabareesh Ramachandran, Vaidehi Tandel, Rajeswari Parasa,
S. Sudharshini, V. Prakash, Y. Yogananth, S. Raju, T.S. Selvavinayagam*
Abstract
A population-representative serological study was conducted in all districts of the state of Tamil
Nadu (population 72 million), India, in October-November 2020. State-level seroprevalence
was 31.6%. However, this masks substantial variation across the state. Seroprevalence ranged
from just 11.1% in The Nilgris to 51.0% in Perambalur district. Seroprevalence in urban areas
(36.9%) was higher than in rural areas (26.9%). Females (30.8%) had similar seroprevalence to
males (30.3%). However, working age populations (age 40-49: 31.6%) have significantly higher
seroprevalence than the youth (age 18-29: 30.7%) or elderly (age 70+: 25.8%). Estimated
seroprevalence implies that at least 22.6 million persons were infected by the end of
November, roughly 36 times the number of confirmed cases. Estimated seroprevalence implies
an infection fatality rate of 0.052%.
Introduction
Knowledge of population-level immunity is critical for understanding the epidemiology of SARS-
CoV-2 (COVID-19) and formulating effective infection control, including the allocation of scarce
vaccines. Tamil Nadu is the 6th most populous state in India, with roughly 72 million persons1. It
has reported roughly 820,000 COVID-19 cases and 12,000 deaths, ranked 4th and 2nd highest,
respectively, among Indian states2. Reported cases are not, however, gathered from
population-representative samples.
The state conducted a population-level seroprevalence survey of 26,640 adults across the 37
districts of the state in October-November 2020. We report seroprevalence estimates from this
survey by district, by demographic groups, and by urban status. We also compare the results of
the survey to estimates from reported cases to measure the degree to which serological
surveys underestimate population immunity.
Methods
* Malani: University of Chicago, USA; Ramachandran: University of California San Diego, USA; Tandel: independent;
Parasa: IDFC Institute, India; Sudharshini, Prakash, Yogananth, Raju, and Selvavinayagam: Directorate of Public
Health & Preventative Medicine, Government of Tamil Nadu.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

2
This study was approved by the Directorate of Public Health & Preventative Medicine,
Government of Tamil Nadu, and the Institutional Ethics Committee of Madras Medical College,
Chennai, India.
Outcomes. The first primary endpoint of the study is the rate of positive results on CLIA
antibody tests at the district-level. The second primary endpoint is seroprevalence rates at the
district-level once seropositivity rates are adjusted for test inaccuracy.
There are multiple secondary endpoints. One set is seroprevalence (a) by age categories and
sex, (b) by rural or urban status, and (c) at the state level. A final secondary outcome is the
difference between population immunity estimated by serological survey and by reported
cases.
Sample and location. Individuals residing in Tamil Nadu and ages 18 years and older were
eligible for this study. The exclusion criteria were refusal to consent and contraindication to
venipuncture.
Sample size. The state sought to sample roughly 300 persons per 1 million population based on
the 2011 Indian Census. The state’s population is organized into districts, districts are
organized into health unit districts (HUD), and HUDs are organized into clusters, defined as a
street in urban areas and habitations in rural areas. Within each cluster the study aimed to
sample 30 individuals. Therefore, the sample rate can be converted into a target number of
clusters per HUD. In total 888 clusters were sampled.
Sample selection. The study selected participants within a HUD in three steps. First, within
each HUD, the study randomly selected clusters. Second, within each cluster, we selected a
random GPS starting point. Third, we sampled one participant from households adjacent to
that starting point until 30 persons consented within a cluster. Within each household, the
member asked to provide a biosample was selected via the Kish method3.
Data collection and timing. Data was collected between October 19 - November 30, 2020.
Each participant was asked to complete a health questionnaire and provide 5ml venous blood
collected in EDTA vacutainers. Serum was analyzed for IgG antibodies to the SARS-CoV-2 spike
protein using either the iFlash-SARS-CoV-2 IgG (Shenzhen YHLO Biotech; sensitivity of 95.9%
and specificity of 95.7% per manufacturer)4 or the Vitros anti-SARS-CoV-2 IgG CLIA kit (Ortho-
Clinical Diagnostics; sensitivity of 90% and specificity of 100% per manufacturer)5. Clusters
were characterized as rural if they were in Census-defined villages. The government shared
data on each reported COVID-19 case and death, including date and demographics.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

3
Statistical analysis. We estimate the proportion of positive CLIA tests by district by estimating
a weighted logit regression of test result on district indicators and reporting the inverse logit of
the coefficient for each district indicator. Observations are weighted by the inverse of sampling
probability for their age and gender groups. Standard errors calculated account for correlations
at the cluster level.
We estimate the seroprevalence by district in two steps. First, we calculate the weighted
proportion of positive tests at the district level everywhere except Chennai, where we calculate
it at the health unit district (HUD), a subset of districts. All samples in a district use the same
CLIA kit, except in Chennai, where all samples in a HUD do so. We estimate a weighted logit
regressions of test results on district indicators outside Chennai and HUD indicators in Chennai
and take the inverse logit of the coefficient for each jurisdiction indicator. Observations are
weighted by the inverse of sampling probability for their age and gender groups. Standard
errors are clustered at the cluster level. Second, for each jurisdiction, we predict
seroprevalence using the Rogan-Gladden formula, test parameters for the kit used in each
jurisdiction, and regression estimates of seropositive proportion by jurisdiction. In Chennai
district, we calculate seroprevalence at the district level as a weighted average of
seroprevalence at the HUD level using as weights the share of clusters in each HUD. (We
employ this approach to Chennai in the estimators below.)
We estimate the seroprevalence in the state by aggregating the seroprevalence across districts
weighted by 2011 census data on the relative populations of districts.
We estimate seroprevalence by demographic group in three steps. First, we calculate the
proportion of positive tests at the jurisdiction-by-demographic group level using logit
regressions of test results on jurisdiction-by-demographic group indicators. Demographic
groups indicators are sex x age for 6 age bins. Standard errors are clustered at the cluster level.
Second, we predict district-by-demographic group level seroprevalence using the Rogan-
Gladden formula. Third, we compute the weighted average of seroprevalence at the
demographic-group level using as weights the share of demographic-group population in each
district using data from the 2011 Indian census.
We estimate seroprevalence by urban status in the same manner we estimate it for
demographic groups with two changes. First, we use the urban status of a cluster in lieu of
demographic status of an individual at each step. Second, observations in our regression are
weighted by inverse of the sampling probability for their urban status.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

4
We estimate the infection fatality rate (IFR) for a population (defined by state, district,
demographic group, or urban status) by dividing total SARS-CoV-2-related deaths reported as of
two days after the last date of sampling in a district by the estimated size of previously infected
persons in that population. We obtain data on deaths by district and demographic group from
daily reports from the Tamil Nadu government. The date of the death count reflects that fact
that the delay the delay between infection and death is on average two days longer than the
delay between infection and seroconversion.6,7
We estimate the size of a population that was previously infected by multiplying our
seroprevalence estimates for that population by the size of those populations as reported in
the 2011 census. We estimate the degree of undercounting of cases in a population by dividing
the estimated number previously infected in the population by the number of officially
reported cases in that population as of 1 week before the median sampling date (median
October 23, 2020). We obtain data on officially reported cases from the government of Tamil
Nadu. The lag accounts for the delay both between infection and seropositive status and
between infection and prevalence testing.
We calculate the Pearson’s correlation coefficient between IFR and age at the individual level,
between undercounting rate and testing rate (tests per million as of median date of testing) by
district, and between the number of SARS-CoV-2-related deaths and the testing rate by district.
Statistical tests comparing groups are performed using a two-sided Wald test with 95%.
All statistical analyses were conducted with Microsoft Excel 365 (Microsoft, USA) and Stata 16
(StataCorp, USA). All plots were generated in R.
Results
One person aged 16 was incorrectly consented and dropped from the analysis. The study was
unable to obtain lab results for 287 additional observations. The study employs 26,135 test
results. Because numerous clusters had fewer than 30 persons consenting, the study yielded
292 samples per million population.
Table 1 reports the demographic characteristics of the sample. Age is missing for 8 persons and
6 persons are transgender. The sample has substantially more females and fewer persons age
18-29 than the general population.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

5
Seropositivity varies dramatically across the state, from 12.1% in The Nilgris to 49.3% in
Perambalur district (Figure 1). Seroprevalence has a similar pattern to seropositivity (Figure 2).
Seroprevalence is significantly greater in urban areas (36.9%; rural, 26.9%; p<0.001) (Table 2).
State level seroprevalence is 31.6% (95% CI: 30.4-32.8%).
Seroprevalence is not significantly different across sexes (females, 30.8%; males; 30.3%; p=0.25)
(Table 2). Seroprevalence among the elderly (70+: 25.8%) is significantly lower than among the
working age populations (age 40-49: 31.6%; p<0.001) or the young (18-29: 30.7%; p<0.001),
respectively (Table 2).
The IFR varies substantially across the state, from 0.007% in Perambalur to 0.203% in Chennai
(Table 3). The IFR increases with age (𝜌=0.8791; p=0.021) and is higher among males (0.11%)
than among females (0.04%; p=<0.001) (Table 4).
Ratio of actual cases to confirmed cases ranges from 9 in Chennai to 144 in Perambalur (Table
3). There is a negative and significant (𝜌=-0.58; p<0.001) correlation between testing rate and
the undercount rate (Figure 3).
Discussion
Overall seroprevalence (31.6%) implies that at least 22.7 million persons were infected by
November 30, 2020, the last day of serological sampling. Thus, the actual number of infections
is roughly 36 times larger than the number of confirmed cases, which totaled 670.392 by 15
October 2020, 7 days before the date the median biosample is collected.
Seroprevalence is highest among working age populations. The lower seroprevalence among
the young is not informative about the value of closing schools because the young in our
sample were over age 18. Moreover, the difference in seroprevalence among the young and
working ages is not significant. The significantly lower seroprevalence among the old is
different than what was found in a recent Karnataka seroprevalence study8 and suggests that
the elderly in Tamil Nadu have been somewhat protected even in multigenerational
households.
Higher seroprevalence in urban areas is consistent with higher density in urban districts (Figure
4). It does not seem positively correlated with mobility as measured by average decline in non-
residential Google mobility measures.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

6
The IFR across the state is 0.052%, nominally lower than that in Karnataka8 and Mumbai9. The
lower IFR is not due to lower infection rate amongst the elderly because IFR is nominally lower
even among the elderly. Higher IFR among males is consistent with the literature10.
Our study has several limitations. One is that, because antibody concentrations in infected
persons decline over time11, our estimate of seroprevalence may underestimate the level of
prior infection and perhaps natural immunity.
Second, we may underestimate IFR. The number of deaths per million is positively correlated
(𝜌=0.96; p<0.001) with testing rate per million at the district level (Table 3). Perhaps increasing
the testing rate would show greater deaths from SARS-CoV-2.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

7
Figures and tables
Figure 1. Proportion of positive CLIA tests by district.
Figure 2. Seroprevalence by district.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

8
Figure 3. Relationship between rate of undercounting and testing rate.
Notes. Each point represents a district. The x-axis presents
the number of tests in a district divided by the 2011 Census
population in that district. The y-axis presents the ratio of
actual cases to confirmed cases. Actual cases are the
estimated seroprevalence (%) in the district times its 2011
Census population. The confirmed cases are counts up to 7
days before the median date of serological sampling in the
district.
Figure 4. Relationship between seroprevalence, population density, and mobility.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

9
Table 1. Demographics of sample, as compared to 2011 Census.
Sample
Mean
Lower
bound
Upper
bound
2011
Census
Gender
Female
61%
60%
61%
50%
Male
39%
39%
40%
50%
Age
18-29
23%
22%
23%
27%
30-39
23%
23%
24%
24%
40-49
20%
20%
21%
20%
50-59
16%
16%
17%
14%
60-69
11%
11%
12%
10%
70+
6%
6%
6%
6%
Obs.
26,135
72,147,030
Note. Census 2011 number for ages 18-29 includes only
those ages 20-29.
Table 2. Seroprevalence by type of region, sex, and age.
Variable
Seropre-
valence
CI lower
bound
CI upper
bound
Region
Rural
0.251
0.242
0.261
Urban
0.367
0.357
0.377
Sex
Male
0.304
0.296
0.312
Female
0.321
0.311
0.33
Age
18-29
0.311
0.303
0.318
30-39
0.32
0.312
0.327
40-49
0.333
0.325
0.34
50-59
0.332
0.324
0.339
60-69
0.284
0.277
0.291
70+
0.252
0.245
0.258
Note. Calculated over a sample representative of the
entire state.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

10
Table 3. Infection fatality rate, undercount of infections, and testing by district.
District Deaths
Confirmed
cases
Tests
conducted
Seropre-
valence (%)
Population IFR (%)
Ratio of
actual to
confirmed
cases
Ariyalur 47 4191 76279 26.52% 754894 0.0235% 48
Chengalpattu 673 40241 399263 34.19% 2556244 0.0770% 22
Chennai 3862 207390 1350491 40.94% 4646732 0.2030% 9
Coimbatore 557 37932 458054 20.43% 3458045 0.0789% 19
Cuddalore 271 22170 263947 33.37% 2605914 0.0312% 39
Dharmapuri 48 4954 100746 19.06% 1506843 0.0167% 58
Dindigul 186 9465 168422 26.88% 2159775 0.0320% 61
Erode 124 8644 207777 18.88% 2251744 0.0292% 49
Kallakurichi 103 9802 133384 38.66% 1370281 0.0194% 54
Kancheepuram 385 23941 328692 34.30% 1166401 0.0962% 17
Kanniyakumari 246 14158 226184 35.40% 1870374 0.0372% 47
Karur 44 3664 65578 16.16% 1064493 0.0256% 47
Krishnagiri 106 5857 68403 18.92% 1883731 0.0297% 61
Madurai 424 18014 364893 38.00% 3038252 0.0367% 64
Nagapattinam 110 5959 93702 21.99% 1616450 0.0309% 60
Namakkal 94 7845 125838 17.04% 1726601 0.0320% 37
Perambalur 21 2010 43049 51.05% 565223 0.0073% 144
Pudukkottai 148 10001 137332 25.21% 1618345 0.0363% 41
Ramanathapuram 127 5778 105173 35.30% 1353445 0.0266% 83
Ranipet 177 14326 128022 45.09% 1210277 0.0324% 38
Salem 425 25144 377688 22.44% 3482056 0.0544% 31
Sivagangai 126 5580 107979 26.68% 1339101 0.0353% 64
Tenkasi 153 7702 108215 48.24% 1407627 0.0225% 88
Thanjavur 221 14486 257680 26.58% 2405890 0.0346% 44
The Nilgiris 38 5510 120113 11.12% 735394 0.0465% 15
Theni 193 15888 170530 44.33% 1245899 0.0349% 35
Thiruchirappalli 168 11645 196207 32.79% 2722290 0.0188% 77
Thiruvarur 99 8620 133411 21.56% 1264277 0.0363% 32
Thoothukudi 132 14391 198278 37.91% 1750176 0.0199% 46
Tirunelveli 208 13684 198389 43.47% 1665253 0.0287% 53
Tirupathur 117 5843 113189 23.93% 1111812 0.0440% 46
Tiruppur 176 10209 187803 19.71% 2479052 0.0360% 48
Tiruvallur 619 35320 450341 34.85% 3728104 0.0476% 37
Tiruvannamalai 262 16804 216551 36.18% 2464875 0.0294% 53
Vellore 306 16854 208871 27.72% 1614242 0.0684% 27
Villupuram 105 12712 205430 32.25% 2093003 0.0156% 53
Virudhunagar 221 14980 266562 37.92% 1942288 0.0300% 49
Notes. Death counts are up to 2 days after date of serological sampling. Test (RT-PCR, not serological)
and confirmed case counts are up to 7 days before the median date of serological sampling.
Population is from 2011 Census. Ratio of actual cases to confimred cases uses seroprevlanece times
the population as the numerator and confirmed cases as the denominator.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

11
Table 4. Infection fatality rate by sex and age.
Age
Sex
Deaths
Seropre-
valence
IFR
0-17
Female
18
N/A
18-29
Female
50
29.15%
0.002%
30-39
Female
119
32.65%
0.006%
40-49
Female
301
32.58%
0.019%
50-59
Female
660
32.67%
0.060%
60-69
Female
974
28.65%
0.143%
70+
Female
1091
27.84%
0.266%
0-17
Male
14
N/A
18-29
Male
71
32.21%
0.003%
30-39
Male
249
28.87%
0.015%
40-49
Male
681
30.53%
0.045%
50-59
Male
1761
31.35%
0.164%
60-69
Male
2491
28.81%
0.380%
70+
Male
3235
25.28%
0.923%
Notes. Death numbers are total up to 2 days after last date of
surveillance. IFR for ages -17 is unavailable because this group is
excluded from the sero-surveillance sampling.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

12
SUPPLEMENT
Seroprevalence in Tamil Nadu in October-November 2020
Anup Malani, Vaidehi Tandel, Rajeswari Parasa, Sabareesh Ramachandran,
S. Sudharshini, V. Prakash, Y. Yoganathan, S. Raju, T.S. Selvavinayagam
Methods
Sampling. Numerous clusters had less or more than 30 persons sampled. We report those in
Table S 1. We retain all samples because it is unclear which samples to drop from clusters with
>30 observations.
Table S 1. Number of samples per cluster.
Samples
per cluster
Number
of clusters
1
4
4
1
8
1
10
1
11
1
20
1
24
1
26
1
28
10
29
31
30
818
31
16
32
2
36
1
Sample size. Sampling 300 per million would lead to different sample sizes per district.
Because some clusters yielded less than 30 persons per cluster, the study produced just 292
samples per 1 million population. The following table presents the sample size obtained and
the resulting minimum detectable effect in each district. These are reported in Table S 2.
Assuming a design effect of 2, the implied minimum detectable effect (MDE) per district varies
from 3.3 (Chennai) to 13.6 (Nagapattinam) percentage points.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

13
Table S 2. Sample size obtained per district and implied minimum detectable effect.
Sample. Suspected or confirmed current or prior COVID-19 infection was not an exclusion
criterion. If a participant was currently receiving medical care for COVID-19, a family member
or proxy was used to complete the questionnaire on the participant’s behalf; however, the
blood sample was taken from the participant.
Data collection. Blood was collected in EDTA vacutainers. Serum was isolated and stored in
Eppendorf tubes. Serum was analyzed using either of two chemiluminescent immunoassay
(CLIA) kits.
The first kit was the iFlash-SARS-CoV-2 IgG kit from Shenzhen YHLO Biotech. Per the
manufacturer, it has a sensitivity of 95.9% (95% CI: 93.3-97.5%) and specificity of 95.7% (95% CI:
92.5-97.6%)4. Independent analysis estimated a sensitivity of 93% (95% CI: 84.3–97.7%) and
specificity of 92.9% (95% CI: 85.3–97.4%)12.
District
Sample
size
MDE District
Sample
size
MDE
Ariyalur 270 0.119 Ramanathapuram 480 0.089
Chengalpattu 720 0.073 Ranipet 420 0.096
Chennai 3613 0.033 Salem 1260 0.055
Coimbatore 1182 0.057 Sivagangai 450 0.092
Cuddalore 870 0.066 Tenkasi 420 0.096
Dharmapuri 567 0.082 Thanjavur 834 0.068
Dindigul 717 0.073 The Nilgiris 223 0.131
Erode 745 0.072 Theni 420 0.096
Kallakurichi 596 0.080 Thiruchirappalli 957 0.063
Kancheepuram 480 0.089 Thiruvarur 420 0.096
Kanniyakumari 659 0.076 Thoothukudi 547 0.084
Karur 390 0.099 Tirunelveli 630 0.078
Krishnagiri 690 0.075 Tirupathur 329 0.108
Madurai 1140 0.058 Tiruppur 740 0.072
Mayiladuthurai 320 0.110 Tiruvallur 745 0.072
Nagapattinam 209 0.136 Tiruvannamalai 810 0.069
Namakkal 600 0.080 Vellore 592 0.081
Perambalur 209 0.136 Villupuram 594 0.080
Pudukkottai 599 0.080 Virudhunagar 688 0.075
Note. MDE is calculated assuming a prevalence of 0.5, confidence level of
95%, and a design effect of 2.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

14
The second kit was the Vitros anti-SARS-CoV-2 IgG CLIA from Ortho-Clinical Diagnostics. Per the
manufacturer it has 90% sensitivity (95% CI: 76.3-97.2%) and 100% specificity (95% CI: 99.1–
100.0%)5. FDA evaluation suggests it has 100% sensitivity (95% CI: 88.7-100%) and 100%
specificity (95% CI: 95.4-100%)13. Independent analysis estimated that it has a sensitivity of
98.8% (95% CI: 92.9-100%) and specificity of 97.3% (95% CI: 85-100%)14.
All the samples in a district are analyzed using the same kit, with the exception of the Chennai.
In Chennai 2 HUDs used 1 kit, one used the other. Table S 3 reports the test kit used in each
district.
Table S 3. Test kit used in each district.
District
Type of Kit
District
Type of Kit
Ariyalur
CPC Kit
Ramanathapuram
CPC Kit
Chengalpattu
Ortho Kit
Ranipet
Ortho Kit
Chennai
CPC & Ortho kits*
Salem
Ortho Kit
Coimbatore
Ortho Kit
Sivagangai
CPC Kit
Cuddalore
CPC Kit
Tenkasi
CPC Kit
Dharmapuri
Ortho Kit
Thanjavur
CPC Kit
Dindigul
CPC Kit
The Nilgiris
Ortho Kit
Erode
Ortho Kit
Theni
CPC Kit
Kallakurichi
CPC Kit
Thiruchirappalli
CPC Kit
Kancheepuram
Ortho Kit
Thiruvarur
CPC Kit
Kanniyakumari
CPC Kit
Thoothukudi
CPC Kit
Karur
CPC Kit
Tirunelveli
CPC Kit
Krishnagiri
Ortho Kit
Tirupathur
Ortho Kit
Madurai
CPC Kit
Tiruppur
Ortho Kit
Mayiladuthurai
CPC Kit
Tiruvallur
Ortho Kit
Nagapattinam
CPC Kit
Tiruvannamalai
Ortho Kit
Namakkal
Ortho Kit
Vellore
Ortho Kit
Perambalur
CPC Kit
Villupuram
CPC Kit
Pudukkottai
CPC Kit
Virudhunagar
CPC Kit
Notes. 33 out of 122 clusters in Chennai used the CPC test kit.
Statistical analysis.
Nagapattinam district was split into Nagapattinam and Mayiladuthurai districts in March 2020,
after the state started reported data on confirmed cases but before we conducted our
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

15
serological survey. We aggregate these two districts together in our estimates of seropositivity
and seroprevalence.
In Chennai, we do not have the population by HUDs. Since the samples were drawn
proportional to population, we divide the district population across the HUDs in proportion to
the sample size.
When estimating our district-level seroprevalence, the weights for our regression analysis
employ data from the 2011 Census for the population in each age x gender category in each
district. We estimate the sampling probability for demographic group (age category x sex) as
the number of observations in that group in the sample in a district divided by the census
population in that group in a district.
When estimating our urban- and rural-level seroprevalence, the weights for our regression
analysis employ data from the 2011 Census for the population in each urban/rural category in
each district. We estimate the sampling probability for urban/rural group as the number of
observations in that group in the sample in a district divided by the census population in that
group in a district.
We calculate the sampling probabilities for each regression observation at the level of 2011-
defined districts (of which there are 32) rather than the 2020-defined districts (of which there
are 38), HUDs or clusters because the population is available only at the level of the old 32
districts. Likewise, we calculate district weights when we aggregate estimates across districts
using the 32, 2001 districts. The 38, 2020 districts are all the same or bifurcations of the 32,
2011 districts. Fortunately, in all bifurcated districts, the same kit was used. Therefore, we can
combine all bifurcated districts into older 2011 districts for purposes of calculating sampling
probabilities in regression analyses or weights when aggregating estimates.
Results
Table S 4 provides the data behind Figure 2, which reports seroprevalence by district. Table S 5
reports seroprevalence by district and demographics.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

16
Table S 4. Seroprevalence by district.
District
Seropre-
valence
CI lower
bound
CI upper
bound
District
Seropre-
valence
CI lower
bound
CI upper
bound
Ariyalur
0.275 0.213 0.337
Ramanathapuram
0.358 0.255 0.461
Chengalpattu
0.346 0.289 0.402 Ranipet 0.453 0.38 0.526
Chennai
0.41 0.382 0.437 Salem 0.224 0.176 0.272
Coimbatore
0.21 0.157 0.263 Sivagangai 0.263 0.175 0.351
Cuddalore
0.351 0.291 0.41 Tenkasi 0.472 0.39 0.555
Dharmapuri
0.195 0.124 0.266 Thanjavur 0.271 0.221 0.321
Dindigul
0.257 0.174 0.339 The Nilgiris 0.114 0.021 0.206
Erode
0.165 0.116 0.214 Theni 0.449 0.344 0.554
Kallakurichi
0.384 0.314 0.454 Thiruchirappalli 0.328 0.274 0.383
Kancheepuram
0.345 0.279 0.411 Thiruvarur 0.215 0.144 0.286
Kanniyakumari
0.343 0.261 0.425 Thoothukudi 0.392 0.325 0.46
Karur
0.16 0.059 0.26 Tirunelveli 0.446 0.368 0.524
Krishnagiri
0.191 0.126 0.256 Tirupathur 0.234 0.138 0.33
Madurai
0.388 0.337 0.439 Tiruppur 0.191 0.138 0.244
Mayiladuthurai
0.273 0.165 0.38 Tiruvallur 0.341 0.291 0.391
Nagapattinam
0.128 -0.023 0.278 Tiruvannamalai 0.359 0.287 0.431
Namakkal
0.173 0.102 0.244 Vellore 0.293 0.225 0.362
Perambalur
0.493 0.386 0.6 Villupuram 0.334 0.273 0.395
Pudukkottai
0.254 0.185 0.323 Virudhunagar 0.388 0.288 0.489
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

17
Table S 5. Seroprevalence by district and demographic group.
Female Male
District 18-29 30-39 40-49 50-59 60-69 70+ 18-29 30-39 40-49 50-59 60-69 70+
Ariyalur 16.8% 29.3% 24.6% 22.8% 24.3% 50.1% 29.8% 37.5% 19.8% 39.2% 50.1% 15.4%
Chengalpattu 34.1% 31.8% 40.8% 36.3% 32.6% 18.3% 43.3% 25.6% 39.8% 39.4% 31.1% 10.8%
Chennai 44.3% 41.7% 49.1% 44.0% 30.7% 43.7% 39.2% 35.4% 40.3% 47.0% 33.0% 29.8%
Coimbatore 15.9% 25.9% 21.9% 25.5% 21.6% 21.6% 21.5% 15.5% 22.1% 23.2% 14.6% 11.4%
Cuddalore 27.6% 43.6% 40.3% 51.4% 40.7% 25.8% 35.8% 29.6% 21.6% 34.5% 24.4% 5.9%
Dharmapuri 10.7% 16.5% 22.1% 26.8% 20.8% 16.1% 19.0% 26.6% 21.0% 22.4% 10.9% 16.1%
Dindigul 19.2% 33.7% 17.4% 24.0% 16.5% 28.7% 36.1% 28.9% 28.7% 31.9% 27.4% 21.2%
Erode 20.5% 13.9% 17.9% 13.8% 15.0% 18.4% 33.4% 26.7% 11.8% 13.6% 17.2% 3.7%
Kallakurichi 37.1% 29.0% 44.0% 58.7% 45.9% 19.8% 34.6% 38.7% 48.8% 39.2% 31.9% 39.2%
Kancheepuram 30.7% 50.8% 42.1% 26.1% 27.5% 68.1% 28.6% 40.4% 34.4% 26.8% 40.4% 17.7%
Kanniyakumari 34.8% 40.1% 42.0% 28.8% 24.5% 19.8% 39.7% 38.0% 39.2% 29.6% 40.7% 18.1%
Karur 11.3% 12.1% 26.4% 17.4% 29.1% 1.9% 15.4% 13.7% 14.8% 7.2% 9.2% 56.2%
Krishnagiri 21.5% 20.3% 18.5% 19.8% 37.8% N/A 17.8% 17.7% 18.0% 17.7% 10.3% 50.8%
Madurai 37.9% 34.7% 36.7% 48.9% 40.2% 38.0% 39.1% 25.8% 36.7% 46.9% 39.2% 54.0%
Nagapattinam 14.1% 16.9% 23.3% 22.8% 14.5% 16.0% 24.5% 24.1% 28.9% 38.6% 22.8% 12.8%
Namakkal 19.0% 19.3% 13.6% 17.4% 7.8% 4.5% 20.3% 18.0% 18.7% 16.1% 24.8% 7.4%
Perambalur 55.6% 50.1% 63.8% 60.0% 50.1% 9.2% 47.6% 54.3% 65.0% 44.0% 20.7% 39.2%
Pudukkottai 31.2% 26.6% 20.6% 23.5% 13.7% 54.3% 19.0% 18.9% 31.1% 30.4% 14.3% 42.3%
Ramanathapuram 44.8% 30.9% 33.0% 39.6% 51.6% 40.0% 23.8% 38.2% 41.7% 22.8% 28.9% 36.5%
Ranipet 47.3% 50.8% 41.1% 44.3% 63.8% 40.4% 56.2% 23.5% 38.8% 46.4% 47.7% 24.8%
Salem 9.3% 25.4% 24.4% 27.6% 18.6% 12.8% 24.3% 23.7% 29.8% 31.1% 23.8% 13.0%
Sivagangai 26.7% 26.7% 32.9% 22.8% 22.2% 16.5% 27.2% 22.8% 34.8% 20.3% 37.8% 14.5%
Tenkasi 50.8% 47.8% 55.6% 46.2% 38.7% 41.0% 51.6% 44.7% 44.0% 46.2% 65.0% 31.9%
Thanjavur 22.1% 28.8% 27.9% 27.2% 17.4% 31.9% 26.4% 24.9% 36.3% 22.8% 24.9% 31.9%
The Nilgiris 9.2% 24.8% 8.8% 13.6% 50.8% 50.8% 6.6% 9.2% 13.6% 16.1% 5.3% 13.6%
Theni 41.0% 45.3% 59.9% 62.7% 37.7% 41.5% 50.1% 50.1% 18.9% 31.9% 53.7% 24.3%
Thiruchirappalli 29.1% 32.8% 31.0% 34.4% 33.6% 38.4% 32.3% 30.2% 44.4% 32.6% 20.1% 42.3%
Thiruvarur 17.8% 21.5% 31.9% 25.0% 10.1% 50.1% 23.3% 19.8% 22.8% 17.4% 34.8% 0.5%
Thoothukudi 36.5% 42.6% 41.5% 36.8% 42.3% 13.7% 38.7% 39.2% 35.0% 51.7% 34.5% 13.7%
Tirunelveli 36.2% 46.7% 42.8% 52.1% 49.2% 27.2% 49.0% 37.8% 41.3% 44.4% 50.1% 48.3%
Tirupathur 19.2% 37.8% 20.0% 11.2% 32.0% 28.5% 25.9% 30.4% 9.2% 31.9% 50.8% 21.9%
Tiruppur 15.2% 18.3% 17.6% 19.6% 12.5% 6.8% 26.2% 23.4% 19.2% 26.8% 23.1% 14.1%
Tiruvallur 38.7% 39.8% 30.5% 31.8% 28.8% 47.3% 44.2% 32.0% 24.8% 24.2% 36.6% 22.8%
Tiruvannamalai 40.1% 35.5% 35.1% 30.3% 36.2% 17.1% 31.2% 37.0% 45.0% 37.2% 35.9% 48.3%
Vellore 28.1% 38.8% 35.5% 34.4% 14.4% 26.8% 35.7% 9.5% 29.6% 10.6% 28.9% 23.9%
Villupuram 36.1% 36.9% 41.7% 31.9% 45.7% 36.5% 26.7% 28.8% 17.4% 33.1% 22.8% 40.5%
Virudhunagar 34.6% 41.6% 47.5% 41.7% 30.4% 33.7% 30.7% 45.9% 35.5% 48.5% 22.0% 33.3%
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint

18
References
1. Wikipedia. List of states and union territories of India by population. 2020.
https://en.wikipedia.org/wiki/List_of_states_and_union_territories_of_India_by_population
(accessed January 4, 2021 2021).
2. Worldometer. COVID-19 CORONAVIRUS PANDEMIC. 2020.
https://www.worldometers.info/coronavirus/ (accessed August 3, 2020.
3. Kish L. A Procedure for Objective Respondent Selection within the Household. Journal of
the American Statistical Association 1949; 44(247): 380-7.
4. Shenzhen YHLO Biotech Co. Ltd. Customer Notification: Sensitivity and Specificity of
iFlash-SARS-Cov-2 IgG and IgM kits from Clinical Trials 2020.
5. Ortho Clinical Diagnostics. INSTRUCTIONS FOR USE - CoV2G (Version 4.2), 2020.
6. Linton NM, Kobayashi T, Yang Y, et al. Incubation Period and Other Epidemiological
Characteristics of 2019 Novel Coronavirus Infections with Right Truncation: A Statistical Analysis
of Publicly Available Case Data. Journal of Clinical Medicine 2020; 9(2).
7. Stringhini S, Wisniak A, Piumatti G, et al. Seroprevalence of anti-SARS-CoV-2 IgG
antibodies in Geneva, Switzerland (SEROCoV-POP): a population-based study. The Lancet 2020;
396(10247): 313-9.
8. Mohanan M, Malani A, Krishnan K, Acharya A. Prevalence of COVID-19 In Rural Versus
Urban Areas in a Low-Income Country: Findings from a State-Wide Study in Karnataka, India.
medRxiv 2020: 2020.11.02.20224782.
9. Malani A, Shah D, Kang G, et al. Seroprevalence of SARS-CoV-2 in slums versus non-
slums in Mumbai, India. The Lancet Global Health 2020.
10. Cai R, Novosad P, Tandel V, Asher S, Malani A. Representative Estimates of COVID-19
Infection Fatality Rates from Three Locations in India. medRxiv 2021: 2021.01.05.21249264.
11. Post N, Eddy D, Huntley C, et al. Antibody response to SARS-CoV-2 infection in humans:
A systematic review. PLOS ONE 2021; 15(12): e0244126.
12. Plebani M, Padoan A, Negrini D, Carpinteri B, Sciacovelli L. Diagnostic performances and
thresholds: The key to harmonization in serological SARS-CoV-2 assays? Clinica Chimica Acta
2020; 509: 1-7.
13. U.S. Food and Drug Administration. Serology Test Evaluation Report for “VITROS
Immunodiagnostic Products Anti-SARS-CoV-2 IgG Reagent Pack” from Ortho-Clinical
Diagnostics, Inc., 2020.
14. Theel ES, Harring J, Hilgart H, Granger D. Performance Characteristics of Four High-
Throughput Immunoassays for Detection of IgG Antibodies against SARS-CoV-2. Journal of
Clinical Microbiology 2020; 58(8): e01243-20.
. CC-BY-NC-ND 4.0 International licenseIt 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 8, 2021. ; https://doi.org/10.1101/2021.02.03.21250949doi: medRxiv preprint