As we continue to analyse the spread of COVID-19 across different countries, we have noticed a widely varying impact from country to country in terms of total case detection and deaths. In this article, we analyse and correlate online web search trends with the COVID-19 case statistics, to determine if there was timely and consistent public awareness of the on-going pandemic from country to country. The hypothesis here is that a more 'clued in' and informed public would likely be better prepared to brace for the impact of the virus' spread; and may well take greater precautions early rather than wait for authorities to institutionalise harder controls. What we found is that the proximity of Asian countries to the Wuhan outbreak around mid-January may have helped build awareness early, which was non-existent in their European and North American counterparts.
Some countries, namely those in Southeast Asia, may have been better prepared for the COVID-19 outbreak due to their closer proximity to China. Online search trend patterns suggest a critical mass of online audiences had picked up news on the Coronavirus and outbreak in Wuhan well ahead of the rest of the world, even as early as January 2020
Meanwhile, COVID-19 related searches picked up only in mid to end March for European countries and the US, suggesting that a large portion of the population was less aware about the virus' spread
This may have also made policy control actions such as lock-downs more acceptable by the general public in those countries, whereas Western countries had to debate their pros and cons with stakeholders before rolling them out
Key findings and discussions
Opening note: see Annexure 2 below for relevant data.
Based on the data gathered, there was clearly much earlier awareness of COVID-19 across Hong Kong, Malaysia, Singapore presumably due to closer proximity to China and frequent travel by Chinese tourists into these countries. The keywords 'Wuhan' and 'Coronavirus' started appearing with significant predominance in these countries as early as early to mid January.
Meanwhile, for European countries such as France and UK, as well as the US, the keyword 'Coronavirus' only started being predominant much later, around early to mid March. At this point, cases and deaths had already started growing significantly compared to the same threshold points for the earlier countries.
We have seen in a previous article how the initial growth rate for countries like Hong Kong, Singapore and Malaysia have been distinctly lower vs. the European countries and the US, at 4-5 days to double vs. 2-3 days to double (in deaths). It is possible that given that the news and information of the COVID-19 spread in Wuhan had reached these countries sooner, there was better preparedness to face the incoming pandemic.
This also underscores the major role of media and public authorities to communicate more effectively to the general public both in terms of breadth and depth; warning them of the virus' spread, its dangers and precautionary measures that could be taken early. We had previously highlighted the need to better publicise theoretical projections of infection figures as opposed to detected cases in this article here; and this point is underscored again given our findings here.
Another positive byproduct of early information and syndication was also perhaps an improved public acceptance towards control and lock-down measures that eventually were implemented by governments to taper the growth of COVID-19. While the UK and US for example took time to debate and deliberate the merits of such a strategy, many Asian countries took more decisive action and immediately put in place such controls. Malaysia for example put in place a movement control order when detected cases reached less than 700. For the UK on the other hand, the equivalent figure was 5,000.
Summary table: Google keyword search predominance and timing in each country, contrasted against COVID-19 case statistics
Annexure 1: A brief note on the analysis conducted within this article.
In this analysis we utilise Google Trends to find search predominance for a range of keywords related to COVID-19 via Google web search (trends.google.com). This data was sourced for a 90 day period up to 10th April 2020. The following graphic from the website is included as an example.
We make two simplifying assumptions here; (i) that for all countries under study, peak Google search has been reached for any of the COVID-19 related topics listed. We believe this to be a fair assumption as this pandemic has been one of the greatest calamities faced by the world today, and it is reasonable to assume that the online population in each country has already met maximum search volumes for COVID-19 related topics over the past 90 days, be it for symptoms, statistics, cures news and more. And (ii) local language specific word searches are ignored, and we assume that terms like 'COVID-19' and 'Coronavirus' are in fact used at the local level.
We are aware that this method may not comprehensively map the latent 'search' interest across any country; at best it only represents the active online population. It is however a rather powerful method to get an objective and quantitative view of macro-level interests across the population.
COVID-19 case detection and death statistics were sourced from Johns Hopkins University's online repository at: https://github.com/CSSEGISandData/COVID-19
Annexure 2: Reference datasets and figures: Google search patterns for COVID-19 related keywords across countries.
<i> Comparison across countries
<ii> Country specific data