On April 15, India’s health ministry identified 170 districts of the country as ‘COVID-19 hot-spots’. Of these, 123 districts have been classified as ‘hotspots with large outbreaks’ – the ‘red zones’, and 47 districts as ‘hotspots with disease-clusters’. The government has urged extra containment measures for such districts.
This is indeed a welcome development in the country’s fight against the novel coronavirus, given that many have urged (here and here) the government to recognise the uneven spread of COVID-19 across India and the emergence of affected district clusters. However, the logic behind the government’s red-zone categorisation is not quite clear regarding either the stated “inclusion criteria” or the announced “red-zone list”.
The government has stipulated that red-zone districts must be the “highest case load districts contributing to more than 80% of cases in India, or highest case load districts contributing to more than 80% of cases for each state in India, or districts with doubling rate (in) less than 4 days” (as reported; independent corroboration of this government proclamation also exists).
The inclusion criteria require a district to be designated a red zone if it satisfies any one of the three stated conditions. But the second and the third conditions make little sense. Is it appropriate to a tag a district a red zone in either of the following cases: (a) it has a total of 4 infections while the state is fortunate to have a total of only 5, or (b) district disease-load increases from 1 to 2 in four days?
The only defensible condition seems to be the first one, and the way to implement it is to list all 736 districts of India in descending order of disease-load (at some point in time), and then select the ‘top set of the districts’ whose aggregate number of COVID-19 cases just crosses 80% of the county’s aggregate. While this approach is clear-cut, there are two concerns. First, it is very likely that there won’t be much outcome-difference between some ‘barely included’ and some ‘barely excluded’ districts. Though it will be hard to accept, luck will indeed play a role in the process.
A second problem is more serious. Is it justified to focus only on disease-load to categorise districts, while ignoring disease growth-rate altogether? Many have argued against such a ‘blinkered’ view, and have suggested a more balanced approach (see here and here).
The government has not only introduced the concept of COVID-19 -hotspot districts, it has also specified the red zones in the country. Given that, should we not focus on how sensibly these districts have been selected, ignoring any weakness in the stated inclusion criteria (which, in any case, seems not to have been followed to the letter)? Indeed we should, if the full list of red-zone districts look quite appropriate to us; if that is not the case, then we might worry that the shortcomings in the inclusion criteria have led to ad hoc selections.
At this point, let me clearly state that I have no issues with about 80-90% of the 123 districts tagged to be red-zones by the government. However, the logic behind the selection of 10-20% of the ‘borderline districts’ is not so clear, and there are some ‘obvious’ districts that have not been selected.
Consider the disease-load (cumulation of all infections on April 17, 12 noon), and disease-growth (increase in the number of cases between midnight on April 5 and 12 noon on April 17), in the following eight districts. [The data is taken from patient records posted in covid19india.org.]
To me, it is clear that if a single district between Yavatmal and Ranipet is to be tagged a red zone, it must be the latter. But the former is included in the government’s red-zone list, and the latter does not find its place in the 170-district ‘hotspot list’. Further, if significant disease-growth is a major concern, then among Pathanamthitta, Thiruvananthapuram, East Godavari, Anand, Raigarh and Thanjavur, if I am to tag exactly three districts to be red zones, I will unhesitatingly choose the latter three. In the government’s list, however, while the former three are designated red zones, the latter three are excluded from the ‘hotspot list’.
Have I cherry-picked districts to make my point? Certainly, I have (and if I could, I would cherry-pick more by getting hold of data for many districts that are currently unavailable in the public domain). The issue of identifying severely COVID-19 infected districts for enhanced containment efforts is so critical that we should not overlook even a single district that requires additional assistance.
I am not arguing for excluding any of the districts that already exist in the red-zone list. My only plea is that a few districts that look to be prime candidates for extra containment assistance should be included in the list. We must err on the side of inclusivity. And if the publicly-available data is incomplete, thus biasing my judgement, I will correct myself and endorse the government’s list. We are being forced to play god without the associated omniscience or omnipotence, so it’s important that we be adequately informed and appropriately cautious in our actions and announcements.
Arijit Sen is a professor of economics at IIM Calcutta. The views expressed here are his own.