Around the end of August in Punjab’s Chatha Nanhera village, samples from two individuals tested positive for Covid-19. But when asked to quarantine at a health centre, the patients refused.
Later in the day, a local gurudwara announced that no one in the village would be tested for Covid-19 and health teams would be prevented from working. On September 9, the BBC reported that the problems had spread with “rumours flying fast in Punjab that the virus is a hoax, that people who do not have Covid-19 are being taken away to care centres, where they are being killed for their organs, and that bodies are being swapped to allay suspicion”.
These episodes raise a critical question. While much of the debate around testing in India (and other countries) revolves around the supply of tests and labs, what if people just do not want to get tested? The low demand for testing should not come as a surprise. Low take-up remains one of the most persistent barriers to public health interventions around the world.
Unfortunately, opposition to testing implies not only severe problems for containing Covid-19 but also for interpreting any population-based surveillance statistics where non-response is high.
Extreme reluctance to test?
To understand the extent of this problem, we collaborated with Punjab government on a pilot project during the first two weeks of July in Kharar, Punjab.
At that time, Covid-19 was still in its early stages, compared to now, and Kharar had reported only a few cases. Given a large number of asymptomatic cases and the likelihood that asymptomatic carriers can infect others, we were particularly interested in identifying the social dimensions of Covid-19 risk to inform testing strategies. Perhaps those who had met many other people were more likely to be infected at an early stage.
We created a random sample of 500 households across 50 polling booths, stratified by proximity to booths with known Covid-19 cases, closeness to mobile towers reporting high presence of positive cases and distance to the testing centre.
The Punjab government sent staff to approach each household and complete a short survey with a person self-identified as the most mobile individual in the household. The survey included questions on their age, occupation, the number of elderly people in their household and a list of symptoms and co-morbidities.
At the end of the survey, the interview subjects were given a free voucher that would enable them to visit the testing centre for a free Covid-19 test. The voucher could be redeemed for up to seven days within the date of the survey. With the voucher, individuals were given a map that showed the location of the testing site. The government requested full consent for each survey and the decision whether or not to test was completely voluntary, so non-response or non-testing was a distinct possibility.
Every person who was approached (regardless of their consent for participation in the survey) was given this voucher. All told, the government reached 477 households, of which 465 consented to the survey. The mean age of individuals in the sample was 44 and 44% were women.
As it turns out, our attempts to identify individual risk predictors for Covid-19 were a spectacular failure. This is because only four of 465 individuals or 0.86% chose to take a Covid-19 test. (In a different later smaller pilot, with intense communication, over half of the individuals chose to test). Given the very small number of individuals who chose to be tested, inferring the correlates of a positive test or even just what determines the demand for testing was a lost cause.
Not only did they choose not to be tested, they may have also concealed symptoms. Only eight individuals or 1.7% reported any of the symptoms associated with Covid-19 and only 12, i.e., 2.6% reported having hypertension and eight, i.e., 1.7% reported having diabetes (three had both), both of which are known to increase the mortality and hospitalisation risks for the disease. These data potentially reflect systematic under-reporting.
Our analysis of the National Family Health Survey shows that prevalence rates of hypertension and diabetes rates in the kind of population we interviewed (adjusted for age and sex) are much higher at 12% and 8%, respectively. It is likely that individuals chose not to report any symptoms or co-morbidities, perhaps because they believed that this would increase their likelihood of being tested in the future.
Why do people not want to be tested and why are so few reporting symptoms or risky co-morbidities? One possibility is a fear of quarantine or hospitalisation, even when individuals are asymptomatic (as many individuals in India have turned out to be). For lower-income households, a three-week quarantine for the main breadwinner could lead to significant loss of income amid rising hunger.
Another possibility is that individuals worried about the additional risk of visiting a testing site. A third possibility is fear and stigma – people may worry that if they are positive, they would be stigmatised and shunned by their neighbours.
We do not know why the demand for testing is so low in this population.
Hampering epidemic management
But the fact of it being so low is itself a marker of serious impediments towards better epidemic management. Several studies point out that there is low trust in government healthcare in India. One manifestation of the historical neglect of health might be that at this critical juncture, individuals may be choosing not to interact with the government for epidemic management. From the long-lasting impacts of the infamous Tuskegee experiment in the US to instances of vaccine hesitancy even before CoVID-19, lack of trust in public health institutions is often the largest barrier for successful public health interventions.
Addressing these barriers is difficult, but not insurmountable. In mid-September, the Indian Council for Medical Research issued new guidelines, allowing self-testing without a doctor’s prescription. Given the reluctance to test, those who choose to self-test must have strong reasons to do so–and such “self-selection” mechanisms may increase the efficacy of testing or select risk-averse individuals, who are less likely to get infected.
Similarly, stressing that a survey is being conducted in order to promote testing may lead to different types of testing behavior. In another differently designed, smaller pilot in Kharar with intense communication, over half of the individuals chose to test–but the fraction who agreed to be surveyed in the first place was smaller. The statistical and epidemiological implications of these different testing regimes have yet to be rigorously evaluated and we just don’t know what they imply for our understanding of Covid-19 and its management.
What is clear, nevertheless, is that while the growth of cases has slowed in recent weeks, it may rise again soon, as festivals and weather facilitate infections. Addressing the issue of testing hesitancy should thus be an urgent priority, both for the management of Covid-19 and for understanding any data emerging from India.
Yamini Aiyar is the President and Chief Executive and Partha Mukhopadhyay is a Senior Fellow at the Center for Policy Research, New Delhi.
Neelanjan Sircar is an Assistant Professor at Ashoka University and Jishnu Das is a professor at Georgetown University. Both are and Senior Visiting Fellows at the Center for Policy Research.
Respond to this article with a post
Share your perspective on this article with a post on ScrollStack, and send it to your followers.