Comparing the impact of various interventions to control the spread of COVID-19 in twelve countries

Sir,

Since the start of the coronavirus disease 2019 (COVID-19) pandemic (caused by severe acute respiratory syndrome coronavirus 2, SARS-CoV-2), multiple countries have implemented various interventions to reduce the disease transmission into their populations [1]. These have included border closure, travel restrictions, and compulsory quarantine for foreigners entering these countries [2,3].

Additional interventions to reduce the local spread of the virus have included the wearing of surgical masks, mass-testing for SARS-CoV-2, travel restrictions, the banning of mass gatherings (of various sizes), the closure of mosques, bars and restaurants, schools and universities, curfews and stay-at-home orders, and the stoppage of all ‘non-essential’ work [2,3].

Here we examine the impact of these interventions across 12 countries with the most complete data available from either media or government websites: the USA, Canada, the UK, France, Germany, Italy, Denmark, India, Thailand, Malaysia, Singapore, and Australia.

For each country, the daily number of COVID-19 cases was extracted and plotted over time, along with the number of diagnostic tests or people tested for SARS-CoV-2 performed, the proportion of the population wearing masks (percentage masking), and the degree of travel restriction by examining the change in volume of people walking in their communities to provide a country-based mobility trend (‘mobility index’) [[4], [5], [6], [7]]. Next, we indicated the dates where individual or combined intervention was implemented, and plotted the estimated daily effective reproductive number, R _t (an average number of secondary cases generated by a typical case at time t under the control measures in place), with 90% confidence intervals plotted on the same timeline (Figure 1 ) [8,9]. R _t is a useful measure to evaluate the effectiveness of public health interventions at a regular interval. In Italy and the UK (Figure 1), the R _t values have decreased to ≤1, from a previously higher value of ∼2, which may be attributed to the interventions that were implemented. While these two study populations became more mobilized after lockdown easing, R _t remained slightly smaller than 1 with improving testing capacity.

Figure 1.

Impact of masking, testing, mobility restrictions and other interventions on daily numbers of COVID-19 cases and value of R_t (dotted lines showing the corresponding 90% confidence interval), for example in (A) Italy, (B) UK.

From each of the countries' plots, we can observe several trends: (i) the testing pattern follows the daily number of new COVID-19 cases, which is not unexpected – you will only identify more cases by testing; (ii) similarly, the percentage masking also follows the number of cases (or R _t) (Supplementary Figure S1), which may be a likely population response to the increasing numbers of cases, where sufficient masks are available; (iii) the mobility index and R _t are strongly correlated – this is reasonable as R _t will depend on the number of contacts with susceptibles, and this will be reduced if mobility is reduced when the social interventions are in place (Supplementary Figure S2), though the relationship of both these parameters with the number of cases is unclear; (iv) any decrease in the number of cases due to any intervention (Supplementary Table S1) tends to be seen two to four weeks later, but this is not consistent and seems not to depend on the country's population size. For example, Australia, Italy, and Germany appear to show a decrease in the number of daily COVID-19 cases approximately two weeks after their first intervention (Supplementary Figure S3), i.e. a 14-day quarantine imposed in Australia, a ‘lockdown’ in Italy, where citizens had to stay at home except for essential needs, and the closure of mosques in Germany. The R _t values for these three countries are currently <1 (at the time of analysis), though that for Australia appears to be drifting upwards.

The USA, India, and Singapore show similar patterns of daily new cases in the early phase despite their massively different population sizes (Supplementary Figure S3), with the USA first banning public events and state-wide stay-at-home orders being issued in the majority of states, India first closing restaurants, and Singapore first imposing a 14-day quarantine for all foreign visitors. India has not yet reached its peak of cases, though its R _t values are decreasing towards 1 (at the time of analysis). The number of cases in Singapore and the USA are now both decreasing with R _t at ∼1. No single intervention was observed to produce a definitive and immediate reduction in the number of daily COVID-19 cases.

For the other countries, R _t values of Canada, France, Denmark, Malaysia, and Thailand have reached or have fallen to <1, though worryingly (at the time of analysis), some are showing signs of rising beyond this threshold again (France, Denmark, Malaysia, Thailand) – which may be due to a fluctuating degree of compliance with the interventions in a proportion of these populations or the easing of lockdown (Supplementary Figure S3).

Despite the data uncertainty, the different intervention combinations implemented within each country, and a variable level compliance with them within each population, we have observed some useful and encouraging trends in the multi-country COVID-19 data presented here. Nation- or territory-specific modelling should be conducted individually to quantify the impact of single or bundled measures on COVID-19 activity and to determine the level of lockdown easing that would trigger another phase of the epidemic.

Conflict of interest statement

None declared.

Funding sources

None.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Supplementary Figure S1

(A–M; there is no I). Temporal variation of population masking percentage and R_t, in: (A) USA, (B) Canada, (C) UK, (D) France, (E) Germany, (F) Denmark, G) Italy, (H) India, (J) Malaysia, (K) Singapore, (L) Thailand, (M) Australia.

Supplementary Figure S2

(A–M; there is no I). Temporal variation of mobility index and R_t under different implemented control measures, in: (A) USA, (B) Canada, (C) UK, (D) France, (E) Germany, (F) Denmark, (G) Italy, (H) India, (J) Malaysia, (K) Singapore, (L) Thailand, (M) Australia.

Supplementary Figure S3

(A–K; there is no I). Impact of masking, testing, mobility restrictions and other interventions on daily numbers of COVID-19 cases and value of R_t, in: (A) USA, (B) Canada, (C) France, (D) Germany, (E) Denmark, (F) India, (G) Malaysia, (H) Singapore, (J) Thailand, (K) Australia.