On March 25, 2020, India imposed a 21-day nationwide lockdown to combat the deadly spread of covid-19. As we near the end of that period, it is time to evaluate whether the lockdown has worked to inform future police measures, and to be able to comment intelligently about any measures that are taken.

The lockdown has undoubtedly imposed great human and economic cost, but in this analysis, I will confine myself to its effect on the spread of the disease itself.

The Government of India has claimed that there would have been over 200,000 cases without containment or a lockdown and over 45,000 cases if there was no lockdown as opposed to 7,447 cases that had been observed until that time. The first number extrapolates the number of cases from March 25 to April 11 assuming a growth rate of 41% per day. There is no explanation of how a daily growth rate of cases higher than observed at any time in India was obtained. The second number extrapolates the number of cases from March 25 to April 11 assuming a daily growth rate of cases equal to the highest ever daily growth rate that existed in India, although that growth rate was observed only for one particular day. Clearly, the methodology is highly questionable to say the least. So how should we evaluate the effect of the lockdown instead?

The median incubation period of covid-19 is 5 days. Therefore, we have to compare the growth rate of the number of cases before and after March 30, which is 5 days after the lockdown came into effect.

The figure below shows the cumulative number of cases each day. The points are the observed number of cases. The solid curve is the best-fit extrapolation using the trend in the number of observed cases up to March 30. The shaded region is the 95% confidence interval.

Extrapolating the pre-March 30 trend gives 9,499 cases as opposed to the 7,599 cases that have been recorded until April 10. As a result of the lockdown, there were 1,900 fewer cases than would have been the case if previous trends had continued. Assuming a mortality rate of 2%, this means 38 fewer deaths.

But it is important to remember that many states had mobility restrictions of various kinds before the nationwide lockdown. So the pre-lockdown period does not represent a period when absolutely no measures were being taken.

Deaths

We can repeat the same analysis using the number of deaths.

However, the number of deaths post-lockdown remains within the 95% confidence interval of what we would expect by extrapolating the pre-lockdown data. This is likely to be because there is time lag before changes in the number of cases are reflected in the number of deaths.

Summary

What conclusions can be drawn? Do the relatively modest gains achieved by the lockdown imply that it has failed and must therefore be lifted? Not necessarily.

Adjusting for the incubation period of the virus, for the first few days post-lockdown, the number of cases actually grew at a faster rate than pre-lockdown. But after April 6, growth in the number of cases has slowed down. The intial spurt was perhaps due to the chaos and mass migration that accompanied the lockdown. Once that was stabilised, the growth rate of cases began to fall.

Regardless of one’s assessment of the success or failure of the lockdown, the question of whether it must be lifted must be assessed on its own merits, paying heed to the old adage that one cannot make another mistake to correct a mistake. It is likely that lifting the lockdown will lead to further migrations and crowding at particular locations and transport facilities leading to yet another spurt in the growth rate of the number of cases.