What are the most and least important Covid-19 measures? Data and insights from over 160 countries

[Click here for an interactive dashboard of reported data and charts]

As countries are navigating their ways out of lockdown due to Covid-19 pandemic, it is critical to better understand the effectiveness of various measures, such as school and workspace closing, public event canceling, and restrictions on gatherings, imposed by governments. I looked at data, collected by the University of Oxford, of all measures imposed throughout the pandemic across over 160 countries to evaluate the effectiveness of each measure in slowing down the spread of Covid-19 virus.

The data show that restrictions on gatherings, public transport closing, and testing policy are the most effective measures in slowing down the Covid-19 spread while contact tracing and school closing are the least impactful measures (Figure 1). When comparing whether the measures were imposed nationally vs locally, an interesting insight emerged. An inexpensive local public information campaign by far has the biggest impact on reducing the Covid-19 transmission (Figure 2).

Figure 1: Transmission change ratio by different measures*. Lower score = Higher effectiveness, Higher score = Lower effectiveness.

I combined the data on all measures imposed locally and nationally with confirmed daily new cases, data collected by Johns Hopkins University, to calculate a daily effective transmission rate as a proportion of current day’s new cases and its previous day’s new cases. Note that this is different from reproduction rate (R) but they are proportional if the infectious duration of a patient stays the same. For example, if the number of new cases in March 1st is 120 and the number of new cases in March 2nd is 150, the defined effective transmission rate in March 1st is 1.25. Positive transmission rate means more people are getting the disease everyday and vice versa. The effective transmission rate is smoothed by a 7-day average since workforce shortages during weekends in some countries often create artificially lower numbers of reported new cases during weekends.

I calculated the changes in the transmission rate due to an imposed measure by dividing the 7-day averaged effective transmission rate on the 11th day after the measure is imposed by the 7-day averaged transmission rate on the day the measure is imposed (transmission change ratio). This approach allows some time for the measure to have an effect on new reported cases. Finally, all transmission change ratios that are more than 1.5 have been removed since the number of new cases can change significantly due to changes in testing policies.

Figure 2: Transmission change ratio by different measures imposed locally (0.0) or nationwide (1.0)*. Lower score = Higher effectiveness, Higher score = Lower effectiveness.

Figure 1 shows the average transmission change ratio, the effectiveness of each policy measure, across countries. Whereas, Figure 2 shows breakdown of the transmission change ratios based on the measures imposed locally or nationally to better understand the effectiveness of local and national policy measures on the Covid-19 transmission.

The data show that contrary to a common belief, contact tracing, which is widely applied in Asian countries, is the least effective measure. Similarly the much debated school closing, particularly in the UK, also has a small impact on the Covid-19 transmission rate. The results are in line with the common belief that coronavirus spreads more easily in closed and crowded spaces. Closing public transportation and imposing restrictions on gathering are amongst the most effective measures. Changes in testing policy also have a significant impact. Meanwhile, in most cases, there is no significant difference in whether measures are imposed locally or nationally. However, one notable exception is targeted public information campaigns, which are much more effective compared to blanket nationwide public information campaigns (Figure 2).

These measures are grouped into coarse groups. Each is implemented differently in different countries and its effectiveness is affected by different hidden factors. The need to look at these measures at a deep level is important to understand how to achieve maximum effectiveness of each measure.

A few simple calculations and visualisations give us some interesting insights to hopefully help governments in easing these measures. Governments need to base their lockdown and easing strategies on available data to help their citizens to get back to normal life safely and effectively.

* Details of statistical significance tests to be added.

Founder of Actable AI (https://actable.ai)