Explained: How many Covid-19 deaths prevented by lockdown?

These estimates cover a wide range. However, we know next to nothing about how they were obtained.

Unfortunately, such a lack of transparency and unwillingness to communicate meaningfully is pervasive. Most data collected by the government, such as the numbers of patients reporting with influenza-like symptoms across the first few months of the year, is simply unavailable to those outside it. The Indian Council of Medical Research has chosen to ignore questions about the quality of the data it provides. The Ministry of Health and Family Welfare has shifted to reporting data once a day from twice in the initial stages while also sometimes cancelling its scheduled briefings with little notice.

Alternative estimates

We provide alternative estimates from the Indian Scientists Response to COVID19 (ISRC) collective. The ISRC is an initiative which combines about 600 scientists from both India and abroad, together with artists, science communicators, doctors and other members of the civil society. It was formed to ensure that accurate, evidence-based information about Covid-19 reaches the public. The ISRC has developed a detailed India-specific epidemiological model, called INDSCI-SIM. This model is well suited to address questions such as those of deaths or cases averted. The model contains up-to-date clinical information specific to Covid-19, as well as demographic details, including age-specific probabilities of mortality and the distribution of ages within the population in each state. The model also describes the effects of interventions such as the lockdown and associated measures. These measures include the increased ability to test those with symptoms and their contacts in a lockdown situation. In effect, testing some proportion of those infected and quarantining them effectively reduces contacts between infected and susceptible people. This reduced contact implies that the disease spreads more slowly.

To estimate the number of deaths or cases averted, we ensured that we could adequately describe the publicly-available aggregated national data on the cumulative number of deaths. (Data for fatalities are more likely to be accurate as a measure of the scale of the infection than numbers for those positive when tested.) We use reasonable estimates of the numbers of exposed and infected persons at the time of lockdown. We can then obtain predictions from our model, both with and without mitigation strategies in place.

Following the lockdown, the model predicts a range of 8,000-32,000 for the number of deaths averted by May 15. Compared to the government’s numbers, this is a more conservative but likely more robust estimate, since it is obtained from what is the most detailed epidemiological model available currently for India. Importantly, all information regarding what we do is publicly available and therefore open to general scrutiny and criticism.

📣 Express Explained is now on Telegram. Click here to join our channel (@ieexplained) and stay updated with the latest

We find that a substantial number of cases must have gone undetected. Our best estimates suggest that between about 0.2-1% of the Indian population has by now been infected, with a likely ratio of 20-30 undetected for every detected infection. Of course, a good fraction of these cases would have recovered by now. Whether someone was infected in the past will turn up in antibody testing. The limited, but well-planned, serological surveys to be initiated by the ICMR shortly should help settle this issue.

The implications

What are the implications of this analysis for the efficacy of the lockdown? This is certainly evidence that the lockdown can be said to have worked, at least in the limited sense of averting Covid-19 deaths that would have otherwise happened if the epidemic spread had not been restricted. However, our numbers for averted deaths are smaller than those quoted by the government. Thus, the gains from the lockdown, unless sustained, could be cancelled by losses elsewhere. These could come from increased difficulties for those with chronic disease in accessing health care and the long-term impacts of not following up on regular immunisation programmes. Deaths in the wake of the movement of migrant workers, and the increased risk of infection spread due to their return to relatively less affected regions of the country, are also corollaries.

There are uncertainties in the recorded numbers of deaths. These would change our results, especially if the number of deaths attributed to Covid-19 was larger by a factor of 2 or 3 over what is recorded. But they do not change them significantly and our estimate that the number of true infections must outnumber figures quoted for reported cases by about a factor of 20-30 remains robust. While such national aggregation provides an overall picture, estimates of the gains from mitigation must be made at local levels, since the epidemic is spreading at different rates in different parts of the country. The INDSCI-SIM model can be used to make such predictions, both at the state and district level.

Actually, it is not even a question of whose estimates are better. The point is that the government is choosing to not make precisely those details public that could help us understand how the estimates were made. That, even in these simple matters, openness should be a casualty, is unacceptable.

(Pinaki Chaudhuri and Gautam I Menon are at the Institute of Mathematical Sciences, Chennai and Ashoka University, Sonepat respectively. The views expressed here are their own.)