Pointers from sero-surveys conducted in various cities

Print edition : September 11, 2020

Health workers taking blood samples from local residents at a serological survey site in New Delhi on August 6. The Delhi government, with technical support from Maulana Azad Medical College, carried out the second round of its sero-survey as a follow-up to the survey conducted between June 27 and July 10. Photo: T. Narayan/Bloomberg

Health workers carrying out throat swabs at the Pune Municipal Corporation’s Savitibai Phule school at Bhavanipeth on May 28. Photo: Jignesh Mistry

A Pune Municipal Corporation worker at Fergusson College’s boys hostel, which is being converted into an isolation ward for COVID-19 patients, on July 17. Photo: PTI

Pune and Berhampur follow the example of Delhi, Mumbai and Ahmedabad and conduct sero-surveys among their populations. But the utility of these studies is limited because they cannot be used to make comparisons or come up with a nationwide picture of the spread or severity of COVID-19.

Given the failure of the Indian Council of Medical Research (ICMR) to carry out a comprehensive nationwide survey of the seroprevalence of COVID-19 among the general population, which would have given States and regions timely guidance about the appropriate measures to take to control the spread of infection, it is a positive and gratifying sign that major academic institutions are joining hands with local governments and civic administrative bodies of big cities to carry out such surveys. Admittedly, these studies are belated but they do give some insights into the nature of the urban spread of COVID-19 in the country.

Following the example set by Delhi, Mumbai and Ahmedabad (“Far from herd immunity”, Frontline, August 28), Pune is the next major urban centre to conduct its own survey among its population. According to news reports, Berhampur in Odisha, too, seems to have carried out a sero-survey among its population. Pune has made the detailed results of its survey public, just as Mumbai had, but Delhi, Ahmedabad and Berhampur have not. Even without making public the complete data of its first round of sero-survey, the Delhi government launched its second round in early August but this time in association with Maulana Azad Medical College (MAMC), New Delhi. The previous round was conducted in association with the National Centre for Disease Control.

Pune sero-survey

The Pune Municipal Corporation conducted the survey between July 20 and August 5 in association with academic institutions that included the Savitribai Phule Pune University; the Indian Institute of Science Education and Research (IISER), Pune; the Translational Health Science and Technology Institute (THSTI), Faridabad, Haryana; and Christian Medical College, Vellore, Tamil Nadu. An IgG antibody-based sero-survey estimates the prevalence of infection in the past (about two weeks before) in the surveyed population. The Pune study estimated the seroprevalence of antibodies to the COVID-19 virus only in the high-incidence (>0.20 per cent of the population) areas of the city. (The incidence rates used for this classification were based on caseload or confirmed case data of June 1.) The survey found the seropositivity rate (SPR) in Pune to be 51.5 per cent. This would be the highest rate found in any of the sero-surveys conducted in the country so far. But since the survey was restricted to high-incidence areas, one cannot immediately extrapolate this figure to the entire city.

The study, however, disaggregated the SPRs according to age, gender and types of dwellings. These disaggregated numbers also cannot be extrapolated to the entire city. For the same reason, no meaningful comparison can be made with other urban sero-surveys, Mumbai’s in particular. While Mumbai too came came out with disaggregated rates, the data was given in a form that would not have allowed a direct comparison in any case.

Asked about the non-inclusion of areas of medium-incidence (0.10-0.19 per cent) and low-incidence (<0.10 per cent), Prof. Aurnab Ghose of the IISER, one of the principal investigators, said in an email interaction with Frontline: “In the absence of any serosurvey in Pune at the time of planning and execution, it was necessary to establish a baseline. The goal never was to estimate for [the] whole [of] Pune, but limit it to specific areas. This was a time when it was not clear what the seroprevalance might be—results of other serosurveys had not come out. So, it made sense to focus on high prevalence areas first.”

Although the number of infections per day was rising when the survey began, from mid July it began showing signs of declining. However, Pune’s total caseload (confirmed positive cases) has overshot Mumbai’s in terms of both the number of cumulative confirmed positive cases (1,37,601 compared with 1,31,542 as of August 20) and the number of active cases (40,665 compared with 17,914. The doubling time is also now showing an increasing trend. But the test-positivity rate in Pune has remained high, which implies that the testing strategy is not yet extensive enough to capture most of the infections.

Given this, and the fact of the survey found a high average SPR (which is indicative of prevalence) across all disaggregated groups, it is likely that the infection is much more widespread in Pune than in Mumbai and that the fraction of Pune’s population in areas with medium incidence and low incidence may not be high. Surprisingly, however, for reasons that are not clear, the case fatality rate (CFR)—the ratio of the number of deaths to the number of confirmed positive cases—in Pune (2.5 per cent) is much lower than that of Mumbai (5.5 per cent). Is it a case of undercounting or under-reporting of COVID-19 deaths? One cannot say.

The survey was conducted in five administrative prabhags (sub-wards) selected from the high-incidence areas of the city: Yerwada, Lohiyanagar-Kasewadi, Rastapeth-Raviwarpeth, Kasbapeth-Somwarpeth and Navipeth-Parvati. The sampling design was evolved by a team of geospatial experts who, assuming a minimum of 5 per cent seroprevalance, worked out that the minimum sample size had to be at least 1,520 individuals. The study finally included 1,664 recruits from these prabhags.

The total population of these prabhags together is 3,66,984, which is about 10 per cent of the city’s population. The sample size then is about 0.45 per cent of the population of the five prabhags. Asked about the fraction of Pune’s population in the high-incidence areas according to June 1 data on which the survey classification was based, Aurnab Ghose said: “We have the raw data, and [we] are curating and estimating numbers such as these now.” He also added that the study team was continuously monitoring the incidence rates and, at the time of sample collection, these prabhags continued to be high-incidence areas. The sampling methodology was as follows: The selected prabhags were gridded and the survey randomly chose 53 grid elements of roughly equal area. In each such grid element, all types of dwellings—hutments, tenements, apartments and bungalows—were covered, and ensuring that the age/gender balance was maintained, a blood sample was taken from one adult individual (>18 years of age) from every fifth home. As with the other sero-surveys, no sampling was done in active containment zones.

The category hutments included traditional slums, whereas the category tenements were proper constructions—which are called chawls in urban Maharashtra—where not all families have access to independent toilets. Even though the peths included in the survey prabhags have a high population density, they do not have what would be traditionally termed slums. But they have many chawls, Aurnab Ghose explained. Lohiyanagar has the biggest slum in Pune, which is often called Pune’s Dharavi. While Navipeth itself is an affluent part of the city and may not have hutments or chawls, the same sub-ward covers the Parvati area as well, which has fairly large slum dwellings.

The survey used the THSTI-RBD-ELISA (enzyme-linked immunosorbent assay) kit that the THSTI developed for the detection of IgG antibodies to the receptor-binding domain (RBD) of the spike protein of the causative virus SARS-CoV-2. It has 100 per cent specificity and 84.7 per cent sensitivity. According to the survey report, this kit was extensively characterised and compared with other commercially available SARS-CoV-2 IgG test kits. When Aurnab Ghose was asked about the low sensitivity of the kit, he said: “Currently, there are no gold standard serum panels or tests for SARS-CoV-2 antibodies against which other tests can be validated. Therefore, the reported numbers for sensitivity are contingent on the panel of sera used.”

As an example, he pointed to the case of a kit developed by a company called DiaSorin that Public Health England (an executive agency of the United Kingdom’s Department of Health and Social Care) determined had a high sensitivity. However, when THSTI colleagues compared the DiaSorin kit head-to-head with the test used in their study on their panel, the THSTI test was marginally better in sensitivity than DiaSorin, Aurnab Ghose said. “So, currently, these [sensitivity] numbers mean little. Suffice to say [that], our test, the one used in Mumbai [CLIA] and DiaSorin have comparable specificity. The reported [sensitivity] numbers have little meaning unless compared head-to-head using the same panel of sera. The sensitivity of Zydus-Covid Kavach, commonly used in ICMR studies, is slightly lower in head-to-head comparisons with both our test and DiaSorin’s,” he added.

Table 1 shows the SPRs the survey found in the five prabhags. The number sampled in each prabhag was about the same, but there is a noticeable variation in the SPR. The higher rates in Yerwada, Lohiyanagar and Parvati are perhaps attributable to the hutment/slum areas there. (According to the survey report, these rates were not corrected for “IgG detection sensitivity, population parameters or sample structure or cluster design”.) The corrected figures were not available yet, Ghosh said, but added that since no test had 100 per cent sensitivity underestimation was very likely but not overestimation because the test had 100 per cent specificity. This means that the corrected overall average SPR is likely to be higher than 51.5 per cent.

The survey included samples from 863 men and 801 women and did not find any significant difference in the SPRs between men and women unlike other surveys (including Delhi’s second round survey), which found the rates among women to be higher. The SPRs determined among males and females in the high-prevalence areas were 52.8 and 50.1 respectively. Table 2 shows SPRs in the five high-incidence prabhags in an age-stratified form. Prevalence was lower in the older (66+) age group but was similar across all the lower (<66) age groups.

As was the case with the findings of the Mumbai survey, the Pune survey found the SPRs among residents of hutments and tenements, which include slums or slum-like habitations with high-people density and shared conveniences such as toilets, to be higher compared with those among residents of other types of residences such as apartments and bungalows (Table 3). However, it is to be noted that the difference in the seropositivity among residents of slum and non-slum residences was not as wide as in the Mumbai survey (56.5 per cent and 15.5 per cent respectively), though the caveat mentioned above, that such a comparison would not be entirely meaningful, needs to be kept in mind.

Interestingly, the study found a lower SPR (33.2 per cent) among residents of apartment dwellings compared with bungalows. A possible reason for this could be that in apartment blocks and housing societies, restrictions and precautionary measures imposed by the collective community get more strictly enforced than in large independent dwellings and bungalows.

The study also found that people with access to independent toilets had a lower SPR compared with those who had to share toilets. The sampling included 1,045 people with access to independent toilets and 629 who used common toilets; the rates were 45.3 per cent and 62.3 per cent respectively. This is, of course, understandable because of the lack of adequate hygiene and cleanliness in shared toilets. But, as the report notes, the SPR among those with independent toilets is also significantly high. This is another indicator that infection spread in Pune is perhaps much more extensive than in Mumbai.

The high SPR found is an indication that a large fraction of the infections in the surveyed areas were asymptomatic. As the report notes, even though the SPRs are fairly high across the different disaggregated groups, this is no indication that the people carrying IgG antibodies will be immune to subsequent infection. Nor is it an indication that general population-level herd immunity may be achieved sooner rather than later. The dropping daily caseload is also not an indication of that.

Asked about epidemiological observations that could be made by comparing the Pune sero-survey data with those of the other city-based surveys, in particular Mumbai, Aurnab Ghose said: “We do not have much detail on the Delhi/Ahmedabad/Behrampur [surveys] but really only the Mumbai one to be able to able to genuinely compare. Mumbai’s publicly released data is also unadjusted for sensitivity etc., but a common theme appears to be that the infection has spread extensively in the areas studied, especially in crowded communities. Population structures and behaviour of the population and the infection may be different in these communities and, until these can be adjusted for, it will be unfair to claim anything more.”

As to the way this work would be carried forward, Aurnab Ghose said that the aim was to characterise the nature of the immune response further. “We will test if ‘protective’ activity is found in the IgG positive sera. However, since this is one-time sampling, we cannot estimate the duration of the ‘protection’. This [characterisation] does not require sampling [the] whole of Pune; a smaller, tightly characterised set is best. These studies are ongoing and will provide more nuanced characterisation of the response to infection,” he added.

As for the possibility of more rounds of sero-survey just as Delhi and Mumbai were doing, he said: “There are at least two other surveys planned for Pune. It is important that such surveys are expanded to other areas and also carried out at different time points. We are not involved in these two follow-ups but hope data will be freely shared in a timely manner. We will make ours freely available so that the other studies can learn and improve. I think we have set the baseline for Pune.” According to news reports, B.J. Medical College will be leading a second sero-survey, which would be mainly for front-line health care workers, including nurses and the police, and D.Y. Patil Medical College will be conducting a sero-survey in the Pimpri-Chinchwad area of Pune.

Berhampur sero-survey

According to news reports, a community-based sero-survey conducted in Berhampur city in Ganjam district of Odisha, the worst hit by COVID-19, between August 3 and 7 found that nearly one in three people had been infected by SARS-CoV-2 in the past. The survey was carried out jointly by the Regional Medical Research Centre (RMRC), the M.K.C.G. Medical College and Hospital and the Berhampur Municipal Corporation and covered 25 wards and five slums of the city. It found that 31.3 per cent of the surveyed population had IgG antibodies to the virus and about 90 per cent of those found to carry antibodies were either asymptomatic or had no major symptoms. The city’s population is about 4.5 lakh.

According to the survey results that the RMRC made public on August 14 through the media, to estimate the spread of the infection in the general population, a total of 2,830 samples were collected from randomly selected individuals across different categories and high-risk groups. The seropositivity found in the survey ranged from 7 per cent to 60 per cent in the different wards surveyed. In four wards, the estimated seropositivity was more than 50 per cent. The average seropositivity in slums was 35 per cent and that in the high-risk group—people with co-morbidity conditions such as diabetes, hypertension, cardiac and kidney problems—was found to be 23 per cent. On the basis of this data, the infection fatality rate—the ratio of the number of deaths to the actual number of infections—was estimated to be 0.02 per cent as against Berhampur’s CFR of 1.6 per cent. From the above data, which are sparse in details, one can neither comment on the nature of the prevalence or spread of infection in Berhampur city nor can one make any meaningful comparison with other urban survey results. The second round of the survey is planned to begin after 21 days.

Delhi sero-survey: Second Round

The Delhi government, with technical support from the MAMC, carried out the second round of its sero-survey in the first week of August as a follow-up to the survey conducted between June 27 and July 10, which found an average SPR of 23.48 per cent (22.86 per cent after adjusting for kit sensitivity, population and sampling design factors, and so on). On the basis of this result, it was calculated that a minimum sample size of 15,000 would be needed. As in the case of the earlier round, multistage sampling was done from all the 11 districts of Delhi. A total of 15,239 individuals were selected for the survey. This time around, too, the sampling was age stratified as follows: 5-17, 18-49 and >50. The sampling was distributed among these groups in the ratio of 25:50:25 and age-ordered participants were selected by an established statistical procedure. While this time the age-stratified results have been made public, in the last round they were not.

The survey found an SPR of 28.35 per cent, which, after adjusting for the specificity (97.7 per cent) and sensitivity (92.1 per cent) of the IgG antibody assay test used, yields a true SPR of 29.1 per cent. Table 4 gives the comparative SPR figures for each of the 11 districts. All the districts have shown an increase in the SPR compared with the first round. The range of percentage increases in the SPRs is wide: the highest increase in South East district (50.09 per cent), followed by South (46.21 per cent) and then West (38.6 per cent). In terms of the gender- and age-disaggregated data, while the average SPR among males was found to be 28.3 per cent, among females it was significantly higher, 32.2 per cent. This was the case in the first round as well (21.63 per cent and 24.2 per cent respectively). The age-wise SPRs are 34.7 per cent (<18 age group), 28.5 per cent (18-49) and 31.2 per cent (> 50) respectively. Like in the other city-based sero-surveys, the youngest age bracket seems to have a high SPR, which is not all that surprising given that they are likely to get exposed to the virus more than other age groups.

The Delhi government will carry out two more follow-up rounds of the survey: in the first week of September and in the first week of October.

This article is closed for comments.
Please Email the Editor