They were supposed to be further proof that India is inching closer towards herd immunity against COVID-19, but the results of India’s latest serosurveys have turned out to be anything but.
Serosurveillance, considered the gold standard for measuring population immunity arising out of past infection or vaccination, can provide crucial insights into progression of a pandemic, its current prevalence level and its likely trajectory for the future.
Many were, therefore, hoping that the large number of serosurveys conducted in various parts of India in the August-September period would reveal how close the country is towards achieving herd immunity.
A distant dream?
Data from the most recent nationwide survey carried out by the Indian Council of Medical Research (ICMR) show that only a fraction of the general population is seropositive for the SARS-CoV-2 infection. At 7%, the figure fell way below the threshold needed for reaching herd immunity. According to most experts, viral transmission does not slow down until at least 60 percent of the population is infected.
The “revelation” that the great majority of the 1.3 billion-strong Indian population is still susceptible to the infection came as nothing less than a shock to those waiting for redemption from the pandemic by way of ‘herd immunity’.
They found the data at odds with the steady surge of daily case counts seen through early October which saw the country emerge as the world hotspot for COVID-19.
“The entire country is looking forward to achieving herd immunity,” said Dr Behram Pardiwala, Director, Internal Medicine and Academics, Wockhardt Hospitals, South Mumbai, echoing a widely held sentiment. “Once, herd immunity is reached, the pandemic will surely end.”
Repeated surveys in the cities of Mumbai and Delhi came out with results that seem at odds with early findings. Data from the second survey in Mumbai, for example, showed a 12% fall in the level of exposure to the virus in slum areas. The Brihanmumbai Municipal Corporation’s second round of survey in August, which covered 5,384 people in three wards, found that 45% of the people residing in the city’s slums were exposed to the coronavirus, compared to the nearly 57% recorded in the first survey in July. But for non-slum areas, seroprevalence rose to 18% from the 16% earlier.
The surveys also showed that levels of seroprevalence varies substantially from one locality to another within the city.
Similar trends were also seen in the third survey carried out in Delhi, which found seroprevalence down by almost 4%.
Flaws in methodology; erroneous sampling
Critics have pointed to several flaws in the sampling design and methodology of these surveys, while researchers explain that the discrepancies may be technical in nature as an entirely different set of people may have been surveyed the second time.
Another, more worrying possibility is that of a rapid drop in SARS-CoV-2 antibodies soon after recovery. “As time passes, antibodies to SARS-CoV-2 are known to decline rapidly,” says Dr Lancelot Pinto, Consultant Respirologist, P D Hinduja National Hospital and Medical Research Centre, Mumbai.
He pointed to a recent study that reported that concentrations of antibodies in a recovered individual are halved roughly every 36 days. This would imply that unless new infections continue to occur in a constant or increasing trend, the population prevalence of antibodies is likely to decline with time after large spikes of infections. Such “antibody decay” is likely the cause for the declining prevalence of antibodies in the population, according to Dr Pinto.
Others, however, beg to disagree. Some of them refuse to accept that a drop in antibody levels can be behind the survey results. Antibodies against the virus may drop to extremely low levels over a period of time, but it is highly improbable for them to vanish so quickly, they argue.
“The falling levels of seroprevalence is a technical error and not due to waning immunity,” says Dr V Anil Kumar, Clinical Professor & Head, Microbiology, Amrita Institute of Medical Sciences, Kochi, Kerala.
The microbiologist is convinced that the immune response to SARS-CoV-2 is similar to what was seen with its close cousin, SARS-CoV. In case of the SARS epidemic, antibodies remained in the system at detectable levels for at least 12 to 24 months.
Antibody levels in mild or asymptomatic cases can be low or even undetectable. Antibody titres are higher in critical patients when compared to non-critical patients due to different rates of maximal antibody response. Patients with severe disease become antibody positive earlier than those with mild disease. Longitudinal studies on 98 SARS-CoV patients have shown detectable levels of IgG for two years; however, the titres may have declined over this period. A similar study on 176 SARS CoV infected people found that antibody titres reduced by 33% within one year and 46% by two years. Antibodies to SARS CoV have been detected even 12 years after infection.
On the discrepancy in the numbers of different surveys, Prof Anil Kumar said it is impossible to accurately gauge true prevalence by targeting selected populations in a country like India which is so heterogeneous with respect to culture, lifestyle, beliefs, socioeconomic status, the density of population and family size. Further, various state governments follow different approaches.
Besides, seroprevalence is bound to be low when children and young. college-going adults — the most active members of the society — are confined to their homes. The virus may have reached every nook and corner of the country, he adds, but it will still take long to hit every household.
Epidemiologists say that the sampling methods adopted for these surveys could also influence their outcome. To ensure reliable results, surveys should make sure the cohort to be tested is representative of the larger population. This means the individuals to be surveyed are representative of the society at large by age, social and economic status, educational levels, areas of residence and other parameters. Likewise, the size of the sample should also be sufficiently large enough to proportionately represent all subgroups. Otherwise, such limitations will skew the intended distribution of the subjects.
The harmonisation of sampling rules and analytical methods is another major step. It is highly unlikely that ICMR and the various states followed the same set of guidelines for their studies. A comparison of results becomes impractical if different surveys adopt different sampling methods, point out experts. Having such a common set of guidelines will also allow multiple surveys to pool their findings.
They are calling upon organizations to publish their serosurvey protocols to bring about a fair understanding of the data and numbers being put out.
The WHO has made available serological survey guidelines for numerous vaccine-preventable diseases like measles, rubella, etc. As far as COVID-19 is concerned, a protocol for serosurveillance was published by ICMR in Indian Journal of Medical Research in June 2020. The article had WHO regional office personnel as collaborators. So, a framework exists on which serosurveys can be modelled in India.
Researchers, nevertheless, may face several challenges on this front. The selection of the population, points out Dr Anil Kumar, is a big challenge as the diversity of the Indian population is too high to achieve uniformity or a complete correspondence. This, for example, can be seen in the fact that serosurveillance of health care workers (HCW) in private hospitals may not show high prevalence when compared to those in government hospitals. Another reason is that, with each state’s epidemic curve behaving differently, it is very difficult to have consistent or uniform seroprevalence data for the country as a whole.
Moreover, says Dr Anil Kumar, it is also too early to attempt to bring in uniformity in serological surveys as the assays are continuously being modified to overcome shortcomings.
Implications of discordant results
Seroprevalence studies the world over show that they are liable to have considerable variations in the results. Data from initial serosurveys conducted in the worst-hit Spain, for instance, found only a small fraction of the population of this European nation seropositive.
Experience suggests that the underestimation of the true prevalence of the disease can have major implications for the epidemiological modelling of the infection as well as herd immunity.
Epidemiologists say that several factors can influence the accuracy of seroprevalence study findings, particularly, the type of diagnostic kits used for testing and the sensitivity of the assays deployed in detecting the antibodies. Chemiluminescence-based ELISA has the best sensitivity but is prone to false positives in low prevalence settings. The target used in the antibody assay also will affect the sensitivity.
Critics maintain that any kit with less than 99% specificity would be “pretty useless” in most seroprevalence studies.
“Initial serosurveys [in India] were done using a rapid card test which was found to be unreliable due to high false negatives and false positives. Subsequent serosurveys were conducted using manual ELISA, which suffer from technical bias,” quips Dr Anil Kumar.
The chemiluminescence based assays are very expensive and sensitive, but compromise on specificity to a small extent. The targets of the assays also vary, with each manufacturer making superior claims. The more commonly used nucleoprotein-based assays perform differently when compared to spike protein-based assays, he points out.
The trimeric spike glycoprotein has a higher ability to detect antibodies to the virus than the nucleocapsid in people with a low level of antibodies. Anti-spike, but not nucleocapsid, IgG, IgA and IgM antibody responses were readily detectable in symptomatic and asymptomatic individuals, studies show. However, several diagnostic tests which are currently authorised for emergency use still consider only the nucleocapsid.
Since the antibody response is directly proportional to the severity of the disease, the assays that target only nucleocapsids are likely to miss several mild and convalescent cases. Hence, the selection of the antigens that these tests target is critical.
Seroprevalence studies carried out in India provide a good template to show how surveys could go awry if the quality of the testing kits is not assured.
In the first serosurveillance study done by ICMR in May-June 2020, two IgG ELISA kits by two different manufacturers were used to confirm seropositivity. 290 samples were positive by the first kit. On retesting with the second kit, only 157 of these were found positive. In this case, the study design addressed the possibility of eliminating false positives by labelling a sample seropositive only if it was positive by both kits, experts point out.
“Evidently, the tools that we have to assess the serological status of a population are far from perfect,” comments Dr Sadia Khan, Consultant Microbiologist at Core Diagnostics, Gurgaon.
Most antibody detection kits, be it rapid or ELISA/CLIA based kits, claim to have sensitivity and specificity of over 90%. Hence accuracy, which is derived from the sensitivity and specificity of these antibody detection kits, is claimed to be high, as per the manufacturer kit inserts.
In a similar way, false negatives will skew our understanding of the spread of the disease, adds Dr Sadia.
The authors of the ICMR study clarified in their report that the sequential use of COVID Kavach and Euroimmun ELISA helped them potentially reduce false positives to as low as 0.01% by obtaining a serial specificity of 99.99%.
However, Dr Chandrakant Lahariya, a noted expert in epidemiology, public policy and health systems, has a slightly different take on this. While sensitivity and specificity are important, he says, these are not everything. In any case, he adds, antibody kits reflect past infections.
These factors limit the utility of standalone surveys. However, repeat surveys provide very useful guidance on trends.
The timing of the survey is another crucial determinant. Recent studies show that SARS-CoV-2-specific IgG levels drop rapidly following recovery. Hence, a delay in testing can give out false negatives. Demographic factors, including age and sex, can also affect the calibration of the results.
Guidance to public health policy; vaccine roll-out
A serological survey uses blood samples to test for antibodies as markers of pathogen exposure to estimate the proportion of the population that has been infected.
It is an important component of surveillance for infectious diseases. High-quality, representative serosurveys could provide an accurate assessment of immunity levels in the general population and are very useful to evaluate the susceptibility of a population to the disease. Using the susceptible, infectious and recovered (SIR model), they help to calculate the peak of a current or subsequent outbreak. No other tools are currently available to monitor an ongoing epidemic with the precision afforded by seroprevalence studies.
Governments can use vital data generated by such surveys to frame effective public health policies during times of pandemics. Without them, the authorities would never be able to make crucial decisions such as how and when to reopen schools, cinemas etc, especially given that not all cases are being recorded in official registers. In other words, the more serosurveys they do, the better equipped they become.
Health authorities can, by testing the same localities over time, follow the footprints of the virus moving through a community. A serological examination can also provide clues on ‘protective’ antibodies and their levels.
These serosurveys are even more useful in case of SARS-CoV-2 as a good proportion of people with COVID-19 show no outward symptoms. Despite the absence of symptoms, these studies can determine if a person has already been infected and recovered.
Dr Sadia clarifies that the field experience with serological tests is often used for decision-making in communicable diseases. In a pandemic, this could translate to the lifting of movement restrictions or determination of containment zones.
Most importantly, serosurveys can provide useful insights on herd immunity thresholds in different populations.
Besides providing useful data to make suitable interventions, these surveys can also guide the optimal rollout of any vaccine, points out Dr Anil Kumar. Individuals with detectable antibodies need not be vaccinated in the first phase and a natural immunity of 60% or above should be good enough to block further transmission. Finally, pre-existing immunity will diminish the potency of any live attenuated SARS-CoV-2 vaccine.
—With inputs from Prapti Shah from Mumbai