Opinion | The surveillance India actually needs to quell the virus
4 min read . Updated: 28 May 2020, 05:11 AM IST
PremiumScientists need much more detailed data to fully understand covid’s impact as our lockdown nears its end
As we exit the fourth phase of the national lockdown, it is time to do some reconnaissance: what worked, what didn’t and what is now needed. This exercise entails using data, of which there seems to be an abundance. Technology and consulting companies have raced to produce epidemiological models, contact tracing apps and mobility dashboards. Few have helped thwart the exponential rise of coronavirus infections. This is because the information we need is simply not available.
Predictions about the epidemic in India are based on data directly published by the Indian Council of Medical Research, or from private voluntary initiatives like covid19india.org. One of the early questions an epidemiologist attempting to inform policymaking will ask is how bad the outbreak is in a particular area. What share of the population the disease or has recovered from it? In the absence of tests for everyone, she would like to know how many were tested, and how many were positive. This is the first basic step. Over 60 days into the lockdown, this information is not available, precluding the scientific, clinical and public health community from making any sound estimations about the disease’s spread. Existing sources provide information on the total reported cases per district, but not the number of tests conducted per district. If Solapur had 100 cases last week and 200 this week, one would conclude that cases doubled, unless 100 out of 1,000 tested were positive in the first week, and 200 of 2,000 were positive in the second week, as testing capacity increased. It would then have entirely different ramifications for public health preparedness. Only state-wise tallies of tests done are available, rendering all projections to be educated guesses at best.
It is not sufficient to know how many were tested, but also who was tested. Testing criteria has changed weekly all over the world. We test some symptomatic patients and their contacts, some asymptomatic patients, some returning migrants, and some healthcare workers. This varies from district to district, state to state. Any robust calculation would require knowledge of the characteristics of this testing denominator to make sensible extrapolations for the rest of the population. Who do these positive tests represent. symptomatics, asymptomatics or travellers? What percentage of the population do they represent? Timely and transparent access to such information can greatly improve our understanding of the epidemic.
Scientists next need to know what happens to those that have a confirmed diagnosis. How many get admitted? What age groups? What other diseases do they have? The Indian response has depended on evidence from other countries, without contextualizing it to our population. The vulnerability profile in India (who is at risk?) will be unique, with its mix of malnutrition, stunting, non-communicable diseases and respiratory infections, as will be our demography (a younger population). Most local jurisdictions do not collect data on the age groups and disease profile of hospitalized patients, the length of stay of hospital admissions (2-3 weeks in the West, but shorter in India), the numbers transferred from one level of care to another, all of which hospital administrators would need to plan bed capacity. As India plans to unlock, it is critical to know who are the most at risk of falling sicker, needing hospitalization, and dying, so that they can be best protected.
These data sets will have limitations, and are best triangulated with other efforts like syndromic surveillance—a daily count of symptoms presented to clinics and hospitals to check if there is a spike in a certain region. The population-based self-reporting symptom trackers that have been promoted by agencies in India can serve as helpful adjuncts, but they must not be used in isolation. The consequences of forced institutional isolation for the entire family have discouraged many from reporting symptoms. Even with widespread compliance, the reporting of fever and diarrhoea in the monsoon will sky-rocket as seasonal malaria, dengue and water-borne diseases set in. Every spike will not simply mean covid-19. Containment may not always be the answer. Such data should then be compared with laboratory surveillance, which entails the counting of positive and negative tests at key laboratories, to provide early signals of impending zonal outbreaks. Governments already have access to data from several designated public labs. The pandemic is the impetus to mandate long-overdue reporting from private labs, provided that principles of data minimization, privacy and security are respected.
Additionally, mortality surveillance, or a count of deaths, is useful—as we saw in Malegaon, where few tests were conducted but mortality was high. Deaths will typically lag by about two weeks, and any spike in mortality numbers should raise an alarm. Local administrations must monitor funeral counts at crematoriums and graveyards, at least weekly. This is one of the more sensitive indicators of widespread contagion.
The need of the hour is not a sophisticated machine-learning model that sits idle for want of data. What India now needs is a simple count of how many of those tested are positive, and information about their demographic, socioeconomic and health status, so that we can take urgent steps to balance lives and livelihoods. Releasing the number of tests that have been conducted per district would be the first step in that direction.
Satchit Balsari and Manoj Mohanan are, respectively, assistant professor at Harvard Medical School and associate professor at Sanford School of Public Policy at Duke University.
