Covid-19: What you need to know today
What explains the coronavirus disease’s trajectory in Mumbai, Chennai, Delhi, Hyderabad, and Bengaluru? All five cities, among India’s largest and most important, have been badly hit by the viral disease – and all five would appear to be witnessing a prolonged run of the disease, with some ebbs and flows. This columnist has hypothesised about this in the past, proffering factors such as population density as a possible explanation.
A recent (October 5) article in peer reviewed journal Nature Medicine by Benjamin Rader from the Boston University’s School of Public Health, Samuel V Scarpino from the Network Science Institute at Boston’s Northeastern University, Moritz Kramer at the Department of Zoology at Oxford University, and others, claims that the “degree to which cases of Covid-19 are compressed into a short period of time (peakedness of the epidemic) is strongly shaped by population aggregation and heterogeneity”. It adds that “epidemics in crowded cities have larger total attack rates than [in] less populated cities”, that “in general, epidemics in coastal cities were less peaked and larger and more prolonged”, and that infection trajectories in rural areas were likely to be peaked.
This could explain why some cities see sharp spikes (or peakedness as the researchers term it; this typically happens in less crowded and less heterogeneous cities, according to the study), while others, such as the Indian cities named above (crowded, and where a heterogeneous population tends to move about), see prolonged attacks. In some ways, this is vindication of the density hypothesis – densely populated cities do tend to see extended runs of the pandemic according to the study – but it is also much more. That is because the study also looks at so-called mean crowding (a measure of both density and its variation across an area), mobility data (sourced from Google in some cases), and applies a mathematical model to calculate the epidemic’s peakedness within cities. The model even took into account the impact of lockdowns or other restrictions put in place to slow the spread of the disease.
The study’s authors explain that this “multivariate model” successfully explained trajectories of the coronavirus disease in Chinese cities and Italian provinces. But that’s not the most interesting part of the study. That would be the fact that the researchers went on to apply the model to 310 cities around the world and calculated their “predicted epidemic peakedness”. This is a number between 0 and 1, where a number closer to 1 shows high peakedness, and a number closer to zero a prolonged run of the pandemic. In this case, spike (or peakedness) should not be confused for actual numbers. A spike refers to a sudden rise and an equally sudden fall in cases; a prolonged outbreak refers to an extended run for the disease. It’s easy to see how the latter could witness higher daily cases, say, than the former.
So, what do the predictive scores show?
New York has a score of 0.0035. Mumbai’s is 0.011. Delhi’s is actually lower than Mumbai’s at 0.008. Hyderabad’s score is 0.012. Chennai is at 0.016. Kolkata at 0.011. The researchers do not appear to have calculated the score for Bengaluru, but given the pattern of scores of other Indian cities, it is easy to see what it could be.
One of my first reactions after going through the study was that it would be interesting to see it redone with more data (from more cities). The researchers seem to think so too. “As with all modelling studies, further data generated during the epidemic might change our parameter estimates, and large-scale serological data would help verify our findings.”
That it would – and it’s another reason why India should carry out widespread antibody tests at regular intervals to assess the spread of the disease.