Politics

Second Wave Update: Predicted Patterns From Three Weeks Back Holding Good

  • The latest data reconfirms two important aspects:
  • One, that the peak life of a well-managed cluster is about three to four weeks, as previously estimated.
  • Two, a broadly quantifiable correlation exists between epidemic progression and district population size, which would be useful to administrators in their material procurement and crisis-planning processes.

Venu Gopal NarayananMay 05, 2021, 02:11 PM | Updated 02:11 PM IST
Fighting Covid.

Fighting Covid.


The latest epidemic data indicates that the spread of the Wuhan virus in most clusters is either reaching a plateau, or mercifully entering into consistent decline. This conforms to Swarajya’s prediction made three weeks ago, that the peak life of a cluster is around three to four weeks.

That assessment was based on historical data spanning multiple administrative units, including national, provincial and district geographies. Some cities have, however, not conformed to this, and the situation there remains grave.


These would be useful in understanding how the contagion is developing, or regressing, in various jurisdictions, and provide a gauge of how successful the authorities have been in containing transmissions in their areas.

The first step of the workflow involved generating a cross-plot of active cases by district, as on date, versus the district’s population. This is the current situation for all districts in the country:

Chart 1: Active cases by district versus district population cross-plot as on 4 May 2021

Four big cities are major outliers — Bengaluru, Pune, Hyderabad and Delhi. Of these, Delhi merits discounting, since its government issues daily data for the National Capital Region as a whole, rather than by district, which is the norm; which means that if one were to segregate the state’s data by district, they would conform more closely to the cluster of blue dots in the lower left quadrant of the chart above.


In addition, data up to 28 April appeared to indicate that there were two distinct trends — one, a high trend among the top half-dozen clusters, and another, for the balance bulk of the districts:

Chart 2: Active cases by district versus district population cross plot as on 28 April 2021 with top four worst-affected cities removed.


Chart 3: Active cases by district versus district population cross plot as on 4 May 2021 with top four worst-affected cities removed.

Nonetheless, the low trend correlation — the black dotted line passing through the blue dots in charts 2 and 3 above — remains consistent, and its angle has reduced, as it should, in step with an overall decline in cases.

See how Mumbai — the blue dot on the extreme right — stays within the scatter envelope in both charts 2 and 3.

This correlation is important since it offers administrators a rough quantification, of the scale of the crisis they are expected to tackle in their districts.

This would aid district magistrates in forward planning and resource management, especially in areas where fresh clusters are developing.


In fact, ideally, administrators and public health officials could also generate such cross plots at smaller units, like at the taluk or block level, by date, for better understanding of epidemic dynamics.

Now, having analysed the current situation by freezing time (called the instantaneous value), the next step was to study the progression of the epidemic chronologically, to see if there were other useful trends which emerged.

For this, the daily district-wise case count of the 16 worst-affected clusters were plotted against delta time, with day zero being the one on which the epidemic began to surge exponentially.

For ease of visualisation, a starting point of 100 cases per day has been used. Hyderabad and Guwahati have not been plotted for data reconciliation reasons.

Chart 4: Daily cases by district versus delta time

The most important observation is that clusters have an average peak life of three to four weeks. The caveat, as the exceptions in chart 4 above show, is that this time frame is heavily contingent upon local authorities being nimble enough to stem the surge.

The best example is Lucknow (red line in chart 4), the capital of Uttar Pradesh, where an alarming rise in cases commenced on day 15.

However, by day 36, three weeks later, the situation had been firmly contained, and the epidemic had firmly entered into a phase of consistent decline.

What is sad to note is that though Lucknow and Bengaluru followed the same high trend for roughly the same duration, Lucknow was able to effect a flattening, while Bengaluru couldn’t. Indeed, Bengaluru is just beginning to enter a plateau phase, about 50 days after its surge commenced.

Ahmedabad (black dashed line in chart 4) was unlucky, in that it had in fact brought the second wave under control, restricted the daily case counts to under a thousand, and even entered into a plateau, when new clusters mushroomed, and the fateful surge recommenced.

Nonetheless, we see that both phases lasted for about four weeks each, thereby conforming to our predicted timeframes.


Similarly, Delhi has taken the longest time to reach a plateau­ — over 50 excruciating days; and here too, the root causes would have to be sought in the administrative domain.

A third exception is Nasik where a plateau of sorts seems evident, but with too high spiking to confidently believe that the epidemic there is indeed being brought under control.

The strangest pattern observed is in Chennai, where the curve rises steadily, and without large fluctuations. Authorities there would do well to consider the testing levels they are at presently, and whether they need to augment tracing and isolation protocols.

There is little other reason why such a relatively-low trend (see how the yellow Chennai curve lies below the others) did not enter into plateau and decline earlier.

In conclusion, the latest data reconfirms two important aspects: one, that the peak life of a well-managed cluster is about three to four weeks, as previously estimated; and two, that a broadly quantifiable correlation exists between epidemic progression and district population size, which would be useful to administrators in their material procurement and crisis-planning processes.

All data from Covid19india.org

Join our WhatsApp channel - no spam, only sharp analysis