Materials and Methods
Data relating to Google Internet searching on the single search term
’Coronavirus’ was downloaded for a range of European countries from the
Google Trends website
(https://trends.google.com/trends)
(Google Trends, 2020)
on the 14 March 2020. GTD indexes the volume of search interest against
a benchmark index of 100. Data was collected for several European
countries where COVID-19 cases have been confirmed. ’Coronavirus’ was
selected as a search term due to the ubiquitous use of this name in
popular parlance across Europe. ’Coronavirus’ is part of the official
definition of this condition, ’coronavirus disease 2019 (COVID-19)’
(
World Health
Organization, 2020). Google translate showed that ’Coronavirus’ was
commonly used across the majority of European countries involved in our
analyses.
Data was collected on 15 Mar 2020 for a 51-day period running from 23
Jan 2020 to 13 Mar 2020. This encompassed the initial phases of the
outbreak, from the potential threat from COVID-19 being highlighted by
WHO in a statement on 30 January 2020
(WHO, 2020).
Corresponding incidence data was obtained from the GitHub database
(https://github.com/CSSEGISandData/COVID-19) of the Coronavirus
COVID-19 Global Cases by the Center for Systems Science and Engineering
(CSSE) at Johns Hopkins University (JHU) homepage, which is being
updated daily based on
WHO,
CDC,
ECDC,
NHC and
DXY
and local media reports
(Dong et al., 2020).
Data for the decreasing phase of the European epidemic was obtained
similarly on 27 July 2020 for a 90-day period from 27 Apr to 25 Jul
2020. Spearman‘s rank cross-correlation analysis between incident case
number and corresponding Google search volumes was performed using a
+-40 days lag.
Time series modelling of incident cases was performed using generalized
additive models. The spline described non-linear effect of numerical
date was added to the model as an independent variable with or without
lag-distributed country specific GTD. Incident case numbers were added
as the explained variable. Models were compared with the Akaike
Information Criteria. All analyses were performed with R (version 3.6.3)
(The R Core Team,
2020) using the mcgv
(Wood, 2004) and dlnm
package (Gasparrini,
2011).