JMIR Publications

ABSTRACT

Background:

Contrary to air pollution and pollen exposure, data on occurrence of common cold is difficult to incorporate in models predicting asthma hospitalisations.

Objective:

To assess whether online searches on “common cold” would correlate and help to predict asthma hospitalisations.

Methods:

We analysed all hospitalisations with a main diagnosis of asthma occurring in five different countries (Portugal, Spain, Finland, Norway, and Brazil) for a period of approximately five years (January 1, 2012-December 17, 2016). Data on online searches on “common cold” were retrieved from Google Trends (using the “pseudo-influenza syndrome” topic as well as local language search terms for “common cold”) for the same countries and time period. We applied time series analysis methods to estimate the correlation between Google Trends and hospitalisation data. In addition, we built autoregressive models to forecast the weekly number of asthma hospitalisations for a period of one year (June 2015-June 2016) based on admissions and Google Trends data from the three previous years.

Results:

In time series analyses, Google Trends data on common cold displayed strong correlations with asthma hospitalisations occurring in Portugal (ρ=0.63-0.73), Spain (ρ=0.82-0.84) and Brazil (ρ=0.77-0.83), and moderate correlations with those occurring in Norway (ρ=0.32-0.35) and Finland (ρ=0.44-0.47). Similar patterns were observed in the correlation between forecasted and observed asthma hospitalisations for the period June 2015-June 2016, with the number of forecasted hospitalisations differing on average between 12% (Spain) and 33% (Norway) from observed hospitalisations.

Conclusions:

Common cold-related online searches display moderate-strong correlation with asthma hospitalisations and may be useful in forecasting them.