Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Anomalous asthma and chronic obstructive pulmonary disease Google Trends patterns during the COVID-19 pandemic

Anomalous asthma and chronic obstructive pulmonary disease Google Trends patterns during the... Background: An increase in online searches on health topics may either mirror epidemiological changes or reflect media coverage. In the context of COVID‑19, this is particularly relevant, as COVID ‑19 symptoms may be mistaken for those of respiratory disease exacerbations. Therefore, we aimed to assess Internet search patterns on asthma and chronic obstructive pulmonary disease (COPD) in the context of COVID‑19, as compared to searches on other chronic diseases. Methods: We retrieved Google Trends (GTs) data on two respiratory (asthma and COPD) and three non‑respiratory (diabetes, hypertension, and Crohn’s disease) chronic diseases over the past 5 years (up to May 31, 2020). For 54 coun‑ tries, and for each disease, we built autoregressive integrated moving average (ARIMA) models to predict GTs for 2020 based on 2015–2019 search patterns. In addition, we estimated the proportion of searches in which COVID‑19‑related terms were used. To assess the potential impact of media coverage on online searches, we assessed whether weekly “asthma” GTs correlated with the number of Google News items on asthma. Results: Over the past 5 years, worldwide search volumes for asthma and COPD reached their maximum values in March 2020. Such was not observed for diabetes, hypertension and Crohn’s disease. In 38 (70%) countries, GTs on asthma were higher in March 2020 than the respective maximum predicted values. This compares to 19 countries for COPD, 23 for hypertension, 11 for Crohn’s disease, and 9 for diabetes. Queries with COVID‑19‑related terms repre ‑ sented up to 47.8% of the monthly searches on asthma, and up to 21.3% of COPD searches. In most of the assessed countries, moderate‑strong correlations were observed between “asthma” GTs and the number of news items on asthma. Conclusions: During March 2020, there was a peak in searches on asthma and COPD, which was probably mostly driven by media coverage, as suggested by their simultaneity in several countries with different epidemiological situations. Keywords: Asthma, Chronic diseases, Chronic obstructive pulmonary disease, COVID‑19, Google trends Introduction Google Trends (GTs), a web-based surveillance tool, can *Correspondence: bernardo@med.up.pt Bernardo Sousa‑Pinto and Enrico Heffler contributed equally to this provide insights into the real-life epidemiology of dis- manuscript eases and outbreaks. This tool provides information—on CINTESIS–Center for Health Technology and Services Research, a relative scale—on how often a certain keyword or query University of Porto, Rua Dr. Plácido da Costa, 4200‑450 Porto, Portugal Full list of author information is available at the end of the article is searched, allowing to compare different regions, time © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/publi cdoma in/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 2 of 14 periods, or keywords. However, as GTs assess individuals’ trend s.googl e.com/; Google, LLC, Mountain View, CA, health information-seeking behaviour, data do not often USA) for 54 countries identified by GTs as “major coun - reflect the true epidemiological situation of the searched tries”, including 23 in Europe, 19 in Asia and the Pacific, conditions [1, 2]. In fact, there are cases that describe ten in the Americas, and two in Africa. We adopted media coverage being associated with anomalously high a time series approach and assessed in detail how the online interest on many health topics, such as coronary COVID-19 pandemic impacted search patterns on these heart disease [3], pollen counts [4, 5] or COVID-19 [6, 7]. diseases. In the context of COVID-19, several GT-based stud- ies have been conducted with the aim of assessing Disease and keyword selection whether online search data correlated with the number We focused on asthma and COPD and included three of COVID-19 cases and deaths. Variable results have non-respiratory chronic diseases (diabetes, hyperten- been observed [8, 9]. In addition, GTs have been used to sion, and Crohn’s disease) for comparison. The three assess variations in online searches for health topics, with non-respiratory chronic diseases were selected on the particular focus on mental health and behaviour-related grounds that (i) diabetes and hypertension are common searches [10–14]. In fact, different studies consistently comorbid conditions that have been associated with a found a decrease in searches for suicide- and depression/ worse COVID-19 prognosis and (ii) Crohn’s disease—like anxiety-related terms in the initial phase of the COVID- asthma—is relatively frequent in young people (who are 19 pandemic [11, 12, 14]. Search patterns on respiratory the most active Internet users), and can manifest as diar- diseases, however, have been less often assessed. While rhoea (which may also occur in COVID-19). We did not a preliminary study visually identified anomalous online assess any other chronic disease, as GTs limit the number search interest for asthma occurring simultaneously in of simultaneously compared queries to five. In particular, several countries of the Northern and Southern Hemi- we did not assess rhinitis as it does not appear to be asso- spheres (Bousquet et  al., unpublished data), it is unclear ciated with COVID-19 searches (Bousquet, submitted). as to what has been driving such unparalleled search In addition, GTs for chronic respiratory diseases were interest, and whether similar search patterns also occur plotted along GTs for acute pneumonia. Searches for with other respiratory and non-respiratory chronic acute pneumonia were used as a proxy for searches for diseases. Understanding whether the perception of coronavirus/COVID-19 (as the search volume for the symptoms of chronic respiratory diseases may be mas- latter is so large that comparisons with chronic diseases querading those of COVID-19 [15], or whether searches are impossible), since searches on these two concepts are being driven mostly by users’ curiosity/concerns, reached their maximum values at the same time through- may have potentially relevant implications. Such implica- out 2020 (Bousquet, unpublished). tions concern, among others: (i) the usefulness of GTs in On account of the selected diseases, we retrieved GTs the epidemiological monitoring of chronic diseases, (ii) data on the following keywords (as “topics”): “asthma”, the way the occurrence of COVID-19 in patients with “chronic obstructive pulmonary disease”, “diabetes”, chronic respiratory diseases is being discussed in the “hypertension”, and “Crohn’s disease”. For pneumonia, media, or is being communicated to patients, and (iii) the the keyword “acute pneumonia” (as “topic”) was used pertinence of Google providing health screening ques- (of note, in GTs, “topics” are groups of search terms that tionnaires following searches on certain expressions [16]. share the same concept [17]; “asthma”, “chronic pulmo- Therefore, in this infodemiological study, we aimed to nary obstructive disease”, “Crohn’s disease” and “acute quantify whether there was an increased search activity pneumonia” are classified by Google as being “disease on two chronic respiratory diseases—asthma and chronic topics”; “diabetes” as a “disorder topic”; and “hyperten- obstructive pulmonary disease (COPD)—in the context sion” as a “medical condition topic”). Along with GTs of the COVID-19 pandemic. In addition, we aimed to on these keywords, we retrieved GTs data on searches assess whether such eventual abnormal search activity (i) involving each chronic disease and COVID-19-related could also be observed in other chronic diseases, and (ii) terms (this allowed us to quantify how much the 2020 was associated with COVID-19-related searches. GTs peaks on chronic diseases were driven by COVID- 19-related searches). For each country, we built a query Methods in its native language(s), consisting of terms specific to We assessed online searches for two respiratory diseases each chronic disease along with COVID-19-related terms (asthma and COPD) and three non-respiratory chronic (Additional file  1: Table S1). Whenever available, we used diseases over the past 5 years up until May 31, 2020. This top-related or rising query expressions (starting on the period includes the first months of the COVID-19 pan - most popular and until the character limit was reached). demic. Online searches were assessed using GTs (https :// In the absence of relevant top-related or rising queries, S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 3 of 14 we combined the most popular terms to search for each the observed GT exceeded the respective maximum pre- chronic disease along with the most popular terms to dicted value. search for COVID-19/coronavirus. Subsequently, for the year of 2020, we considered that GTs on each of the five selected chronic diseases (i.e., Data analysis total volume of searches in a given period of time) could Google Trends values represent the Google search inter- be divided into two components: (i) searches without any est over time for a given topic as a proportion of all COVID-19-related term, and (ii) online searches on each searches on all topics on Google at that time and loca- chronic disease along with COVID-19-related terms (i.e., tion. Values are indexed to 100, where 100 is the maxi- “COVID-19 related-searches” estimated for each coun- mum search interest for the time and location selected. try, using the queries listed in Additional file  1: Table S1). The values are re-indexed according to the selected time For each month, we calculated the average proportion period. that the latter represented among GTs on each chronic We started by analysing the worldwide search inter- disease. In addition, for each month, we subtracted the est patterns of these five chronic diseases over the past average GTs on each chronic disease + COVID-19-re- 5 years (up to May 31, 2020), to visually assess the pres- lated terms from the average total GTs on each chronic ence of spikes during the COVID-19 pandemic. For this disease. The difference was compared with the respective assessment, GTs on chronic diseases were plotted along predicted value as estimated by previously described sea- GTs for “acute pneumonia”. As a particular case, we com- sonal ARIMA models. This allowed us to assess whether pared the volume of searches subsequent to the thun- there might be an excess of searches on asthma beyond derstorm-asthma of Australia (November 2016) with that explained by queries including COVID-19-related COVID-19-associated searches in asthma, as the former terms. was the largest “asthma” spike that had previously been Finally, to preliminarily assess the impact of media retrieved worldwide [18]. coverage on online searches, we estimated the correla- Subsequently, we studied the search patterns of the tions between GTs and Google News (https ://news.googl five aforementioned chronic diseases during 2020 (Janu - e.com/; Google, LLC, Mountain View, CA, USA) items ary–May) in the 54 countries identified by GTs as “major on asthma in 19 different countries. For each country, countries”. Our aim was to assess whether the search we retrieved the weekly number of Google News search interest values on chronic diseases in each of these coun- results (i.e., searches in news items, which differ from tries exceeded those that would be expected based on the Google News aggregator service present in several patterns from the previous years. For this assessment, we countries) when searching the query “asthma” in the built seasonal autoregressive integrated moving average respective language and applying the respective coun- (ARIMA) models to predict GTs for 2020 based on the try and language restriction filters. Unrelated results GT patterns from 2015–2019. Seasonal ARIMA models (namely those which had only been retrieved because are defined by the parameters (p, d, q)(P, D, Q) , with p the respective websites advertised news for asthma) were corresponding to the order of autoregression, d to the not counted. Correlations were estimated by computing degree of difference, q to the order of the moving aver - Pearson correlation coefficients. age part, P to the seasonal order of autoregression, D to Data analysis was performed using software R version the seasonal integration, Q to the seasonal moving aver- 4.0.0 (R Foundation for Statistical Computing, Vienna, age, and s to the length of the seasonal period [19] (for an Austria). example of the use of seasonal ARIMA models for health forecasting, as well as for a discussion on their meth- Results odological strengths and limitations, please consult the Fivey ‑ ear searches for chronic diseases study of Song et  al. [19]). In this study, we applied sea- When visually analysing 5-year data from all countries sonal ARIMA(3,0,2)(0,1,1) models, using weekly GT combined, we observed that asthma and (on a lesser data (thus explaining the length of the seasonal period— scale) COPD searches reached their maximum values in s—being 52). For each model, we retrieved the maximum March 2020, simultaneously with a search spike on acute values—for the whole year of 2020, and for the month pneumonia (Fig.  1). In Australia, the maximum volume of March—of the upper bound of 95% confidence inter - of asthma searches in March 2020 was 23% lower than vals of predicted GTs (“maximum predicted values”). that observed in the week of November 20–26, 2016 Such maximum values were compared with the maxi- (associated with the thunderstorm-induced asthma). On mum observed GTs for the year of 2020 (January–May) the other hand, there were only two countries where the and for the month of March. A search peak was formally 2020 GTs for COPD reached higher values than those for defined as any situation in which, for a given search term, asthma: in Hungary, COPD maximum values occurred Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 4 of 14 Fig. 1 Worldwide 5‑ year Google Trends for respiratory and non‑respiratory chronic diseases 2  weeks after those for asthma, whereas in Turkey, they Peaks for non-respiratory chronic diseases were not as occurred simultaneously (Additional file 1: Fig. S1). geographically and/or temporally consistent as those for For non-respiratory chronic diseases, no worldwide asthma or COPD. Throughout 2020, the monthly average search spikes were visually identified in 2020 (Fig.  1). By of GTs on hypertension exceeded the predicted values in contrast, we identified annual spikes for diabetes asso - 23 countries (42.6%), mostly those in Europe (n = 12) and ciated with the World Diabetes Day. Visually assessing Latin America (n = 6). However, most GT peaks occurred 2020 data for each specific country (Additional file  1: in April or May (n = 14), including those observed in all Figs. S1, S2), large diabetes spikes were found for three Latin American countries. Peaks for diabetes mellitus specific countries, namely Italy (starting during the onset and Crohn’s disease were observed in fewer countries of the COVID-19 pandemic and having a 2-month dura- (n = 9 and n = 11, respectively), and were highly variable tion), Romania (week of March 29) and Sweden (week of depending on their region, month, and magnitude (of April 12). In these three cases, maximum 2020 GTs were note, Crohn’s disease peaks were frequently identified in greater than GTs in the “World Diabetes Day” weeks. Middle Eastern countries, a fact that might be related to typos in users’ queries, given the similitude of the Arabic Quantification of search peaks for chronic diseases in 2020 and Farsi words for “Crohn” and “corona”). In March 2020, search peaks for asthma GTs were identified in 38 out of the 54 studied countries (70.4%) Disentangling chronic diseases and COVID‑19 searches (Tables  1, 2). Such peaks were observed in the assessed Out of the 38 countries in which asthma search peaks countries of Europe (apart from Romania, Russia  and were identified, 28 (73.7%) had top-related or rising- Ukraine), in the Americas (apart from Mexico), in Aus- related queries involving COVID-19-related terms. On tralia and New Zealand, but only in one third of Asian the other hand, this occurred in five out of 22 (23%) countries. Search peaks for COPD were temporally con- countries for COPD, 11 out of 23 (48%) for hypertension, sistent with those of asthma, but were only observed for five out of nine (56%) for diabetes, and in 0 out of 11 for 19 countries, mostly those located in Central Europe, Crohn’s disease. North America, and the Pacific. However, COPD search In March 2020, asthma COVID-19-related searches peaks were overall smaller than those observed for were detected in all countries except Egypt, representing asthma. between 4.4% (for the Philippines and India) and 47.8% S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 5 of 14 Table 1 Observed and predicted maximum values for Google Trends (GT) on five chronic diseases (January–May 2020) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) (2020) Europe Austria 67 33 42 38 73 94 45 49 21 28 Belgium 67 32 33 24 83 91 41 47 28 32 Bulgaria 30 28 17 18 93 100 78 78 11 19 Czech Repub 37 35 30 18 87 92 32 35 30 43 lic Denmark 87 30 26 31 84 93 13 14 18 16 Finland 76 38 15 17 75 86 31 28 18 20 France 100 27 24 13 68 61 51 44 28 31 Germany 87 30 53 39 79 89 41 44 18 24 Greece 52 20 27 15 86 100 34 37 17 34 Hungary 34 20 100 14 64 68 50 47 16 20 Ireland 100 36 45 31 84 78 49 43 29 28 Italy 41 29 11 12 100 32 74 68 32 33 Netherlands 64 22 27 23 64 82 34 37 14 18 Norway 100 35 17 27 58 70 49 32 20 25 Poland 28 18 11 9 70 79 25 33 20 12 Portugal 50 25 9 15 59 75 28 31 16 24 Romania 19 22 7 10 100 97 42 31 6 11 Russia 26 32 9 10 68 97 34 41 7 10 Spain 62 21 15 23 57 62 66 67 17 59 Sweden 41 15 14 15 100 44 37 20 7 8 Switzerland 100 26 28 22 56 59 62 35 14 21 Ukraine 24 25 10 8 91 100 39 47 11 11 United King 100 13 35 13 58 43 30 18 12 14 dom Africa Egypt 16 8 3 4 87 86 13 15 69 4 South Africa 32 22 10 11 70 72 75 55 9 9 North America Canada 69 24 31 24 96 100 47 55 18 24 USA 60 28 28 27 99 100 61 66 16 22 Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 6 of 14 Table 1 (continued) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) (2020) Latin America Argentina 21 10 7 9 30 37 30 32 3 100 Brazil 52 29 11 9 87 96 72 65 8 29 Chile 49 28 12 19 69 87 73 74 8 12 Colombia 26 25 35 20 52 63 100 81 4 25 Ecuador 23 23 12 12 56 71 75 58 9 8 Mexico 19 21 12 12 76 87 69 67 4 5 Peru 42 33 7 9 76 89 74 60 5 6 Venezuela 31 32 13 12 54 63 71 71 5 10 Asia Hong Kong 25 33 14 16 87 100 47 63 8 14 India 24 20 10 11 78 79 37 34 3 6 Indonesia 53 54 7 10 95 100 93 100 2 2 Iran 78 82 1 2 20 72 16 19 45 4 Israel 27 22 10 11 85 98 23 29 23 25 Japan 70 48 12 9 90 94 29 34 6 11 Malaysia 36 45 11 13 90 100 71 80 4 5 Pakistan 36 42 17 21 100 100 69 77 8 7 Philippines 87 61 20 23 90 100 100 91 6 9 Saudi Arabia 13 16 3 4 86 84 9 12 46 5 Singapore 20 22 7 10 61 81 37 86 4 7 South Korea 17 23 12 13 80 93 29 45 32 13 Taiwan 27 29 9 13 95 100 59 68 4 5 Thailand 18 21 10 12 96 100 58 62 3 3 Turkey 65 35 100 26 65 79 46 31 7 10 UAE 22 21 6 9 64 82 30 34 12 9 Vietnam 18 20 6 9 100 89 34 36 2 2 Pacific Australia 77 53 28 25 88 94 46 47 17 24 New Zealand 91 42 39 27 76 100 34 52 35 28 Maximum predicted values correspond to the maximum values of the upper bound of the 95% confidence intervals for predicted asthma GT. Numbers in italic indicate cases in which maximum GT observed values were higher than maximum predicted values COPD Chronic Obstructive Pulmonary Disease, UAE United Arab Emirates, USA United States of America S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 7 of 14 Table 2 Observed and predicted maximum values for Google Trends (GT) on five chronic diseases (March 2020) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (March value (March value (March value (March value (March value (March value (March value (March value (March (March 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) Europe Austria 67 29 42 37 71 94 45 45 21 26 Belgium 67 30 21 20 69 88 38 42 25 27 Bulgaria 27 26 17 14 86 97 78 75 6 17 Czech Repub 37 35 14 17 63 92 32 34 18 40 lic Denmark 87 27 26 31 82 92 13 14 14 14 Finland 76 34 10 15 67 77 31 27 13 17 France 100 25 24 12 68 60 51 44 22 31 Germany 87 30 53 34 68 88 41 44 18 20 Greece 52 20 27 15 73 93 31 33 13 32 Hungary 34 18 100 12 37 65 39 47 10 20 Ireland 100 29 45 28 84 78 49 40 29 28 Italy 41 29 11 12 78 32 74 67 13 32 Netherlands 64 21 27 21 60 74 29 37 14 15 Norway 100 28 11 23 55 70 49 32 14 23 Poland 28 18 11 8 44 78 24 33 10 12 Portugal 50 25 9 15 52 70 18 26 15 22 Romania 18 18 7 9 100 94 42 31 4 11 Russia 26 29 9 10 61 97 33 41 6 9 Spain 62 19 15 23 56 61 65 64 13 39 Sweden 41 14 12 15 41 44 32 20 7 8 Switzerland 100 26 28 22 56 59 62 34 14 21 Ukraine 24 25 10 8 75 100 37 47 7 10 United King 100 13 35 12 58 42 30 18 12 14 dom Africa Egypt 8 8 3 4 55 86 10 14 69 4 South Africa 32 20 9 11 60 70 40 44 7 8 North America Canada 69 23 31 24 95 100 46 55 18 22 USA 60 28 28 24 99 100 59 64 16 22 Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 8 of 14 Table 2 (continued) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (March value (March value (March value (March value (March value (March value (March value (March value (March (March 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) Latin America Argentina 21 8 6 8 30 35 30 27 3 19 Brazil 52 25 11 8 87 96 72 59 8 25 Chile 49 26 11 17 64 87 64 69 8 10 Colombia 26 22 35 19 46 62 100 74 3 22 Ecuador 23 18 7 10 46 63 75 49 9 8 Mexico 19 19 12 12 76 87 69 67 3 5 Peru 39 26 5 9 64 87 57 51 4 6 Venezuela 31 28 5 12 48 63 71 71 3 10 Asia Hong Kong 23 33 14 15 69 100 43 63 5 14 India 22 19 10 10 61 79 31 33 3 4 Indonesia 53 53 7 10 87 100 83 100 1 2 Iran 68 82 1 2 16 34 16 16 6 4 Israel 27 20 9 10 52 98 18 25 19 20 Japan 51 43 12 8 68 88 26 30 4 9 Malaysia 35 40 11 13 90 100 59 79 2 5 Pakistan 30 37 12 18 66 100 55 69 7 7 Philippines 87 59 20 20 78 100 81 91 5 7 Saudi Arabia 13 16 2 3 59 83 8 12 46 5 Singapore 19 20 7 9 47 78 37 81 3 7 South Korea 17 20 10 12 63 92 27 45 7 11 Taiwan 24 29 9 12 82 100 45 68 2 5 Thailand 16 20 10 12 81 100 40 56 3 3 Turkey 65 34 100 22 60 76 40 30 6 10 UAE 22 18 6 6 55 73 22 33 6 8 Vietnam 16 20 9 9 95 88 29 36 1 2 Pacific Australia 77 48 28 24 85 94 45 42 17 23 New Zealand 91 36 39 25 76 100 33 42 19 28 Maximum predicted values correspond to the maximum values of the upper bound of the 95% confidence intervals for predicted asthma GT. Numbers in italic indicate cases in which maximum GT observed values were higher than maximum predicted values COPD Chronic obstructive pulmonary disease, USA United States of America S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 9 of 14 Table 3 Expected and excess Google Trends on asthma and chronic obstructive pulmonary disease (COPD) Expected baseline Excess searches Searches on asthma Expected Excess searches Searches on COPD searches on asthma on asthma beyond those with Covid‑19‑related baseline on COPD beyond those with Covid‑19‑related (%) including Covid‑19‑ terms (%) searches including Covid‑19‑ terms (%) related terms (%) on COPD (%) related terms (%) Europe Austria 52.0 1.6 46.4 75.5 6.6 17.9 Belgium 36.1 35.7 28.2 65.6 21.3 13.1 a a a b b b Bulgaria – – – 44.9 55.1 0 c c c Czech Republic 68.7 22.7 8.6 30.2 69.8 0 a a a Denmark 36.5 27.5 36.0 – – – a a a Finland 41.6 30.3 28.1 – – – France 32.9 32.9 34.2 41.1 41.6 17.3 Germany 38.8 20.4 40.8 53.1 28.8 18.1 Greece 41.7 47.9 10.4 49.6 44.1 6.3 Hungary 56.8 29.4 13.8 21.8 78.2 0 Ireland 24.2 45.5 30.3 57.1 29.7 13.2 a a a Italy 65.4 0 34.6 – – – Netherlands 40.3 20.3 39.4 58.8 19.9 21.3 a a a Norway 28.5 31.3 40.2 – – – Poland 62.5 16.2 21.3 69.6 30.4 0 a a a Portugal 41.8 38.5 19.8 – – – a a a Spain 33.6 18.6 47.8 – – – a a a Sweden 31.4 28.3 40.3 – – – Switzerland 29.1 33.5 37.4 67.9 32.1 0 a a a Ukraine – – – 64.9 35.1 0 United Kingdom 20.8 37.3 41.9 43.2 40.2 16.6 Africa c c c a a a Egypt 42.8 57.2 0 – – – a a a South Africa 54.0 26.6 19.4 – – – North America Canada 41.4 33.7 24.9 79.1 15.5 5.4 USA 51.0 20.6 28.4 89.9 2.8 7.3 Latin America a a a Argentina 28.9 41.7 29.4 – – – Brazil 53.7 25.3 21.0 78.5 16.9 4.6 a a a Chile 66.8 14.9 18.3 – – – b b b Colombia 83.5 5.8 10.7 51.2 40.3 8.5 a a a Ecuador 71.5 16.5 12.0 – – – c c c a a a Peru 71.8 20.3 7.9 – – – Venezuela 83.3 9.9 6.8 89.1 10.9 0 Asia b b b a a a India 64.5 31.1 4.4 – – – b b b a a a Israel 65.7 34.3 3.2 – – – b b b Japan 67.7 11.0 21.3 63.4 24.1 12.5 a a a Philippines 64.6 31.0 4.4 – – – Turkey 56.3 37.2 6.5 38.1 58.4 3.5 Pacific Australia 62.0 14.9 23.1 78.8 16.1 5.1 New Zealand 38.9 30.5 30.6 52.8 47.2 0 Percentages of Google Trends on asthma and COPD corresponding to (i) expected baseline searches, (ii) excess searches beyond those including Covid–19‑related terms, and (iii) searches with Covid‑19‑related terms. Unless otherwise indicated, search peaks were observed in March USA United States of America No search peak observed (of note, no search peak for either asthma or COPD was observed for Hong Kong, Indonesia, Iran, Malaysia, Mexico, Pakistan, Romania, Russia, Saudi Arabia, Singapore, South Korea, Taiwan, Thailand, United Arab Emirates or Vietnam) Search peak occurred in April Search peak occurred in May Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 10 of 14 (for Spain) of the GTs on that disease (Table  3; Fig.  2). Asthma primarily affects the respiratory system, and Overall, the percentage of COVID-19-related searches some of the main symptoms of COVID-19 are also res- was higher in European countries, reaching over 40% in piratory. This fact, along with a relative lack of infor - six of them. By contrast, apart from New Zealand, the mation (particularly when compared to diabetes or percentage of COVID-19-related searches did not exceed hypertension) on whether asthma can be associated 30% in any of the non-European countries. We also with a worse prognosis of COVID-19 [20], may partly observed a variable excess of searches on asthma beyond explain the particularly evident search increase observed those explained by queries including COVID-19-re- for asthma. Another possible explanation concerns the lated terms. In March 2020, such an excess represented fact that young adults, and especially parents, are par- between 0% (for Italy) and 47.9% (for Greece) of the GTs ticularly active Internet users [21]. In fact, asthma is on “asthma”. relatively common among young adults and even more The maximum percentage of COVID-19-related among children (in relation to whom, parents may wish searches was 21.3% for COPD (the Netherlands) (Table 3; to seek health information). Such an increase may be Fig.  3). For non-respiratory chronic diseases, 20.2% was smaller for COPD as the latter (i) is more frequent at a reached for diabetes (United Kingdom), 20.5% for hyper- more advanced age (with the elderly being less active on tension (Switzerland), and 4.7% for Crohn’s disease the Internet than the younger [1]), and (ii) is less known (Egypt) (Additional file  1: Table  S2, Figs. S2–S4). The among the general public. In fact, regarding COPD, some number of countries for which COVID-19-related search of the most frequent top-related queries consisted of just represented < 1% of all GTs was eight for COPD, com- asking what COPD was (data not shown). pared to four for diabetes, five for hypertension, and nine This study suggests that GTs alone may be inadequate for Crohn’ disease. for prospectively assessing the epidemiology of chronic In most countries, the number of Google News items diseases, and questions Google’s strategy of displaying on asthma reached their maximum value in March 2020 screening questionnaires when searching for key expres- (Additional file  1: Fig.  S5). The subsequent pattern was sions [16]. In fact, in this study, we found that asthma less consistent across countries, the number of asthma search peaks occurred simultaneously in several coun- news remaining high in some and decreasing in others tries in the Northern and Southern hemispheres, irre- (often with new rises). Therefore, while GTs and Google spective of the COVID-19 epidemiological situation (as News displayed moderate-strong correlations for the suggested by the relatively small Italian search peak) or whole of 2020 (often reaching their maximum values in of environmental phenomena. This suggests that media the same week), such correlations were stronger when coverage plays a major role in influencing GTs, as cor - specifically assessing the months of January to March roborated not only by the moderate-strong correlations (Additional file  1: Table  S3). While weaker correlations observed with the frequency of Google News items were found for countries in which small or no GT asthma (which should be carefully interpreted, as the amount peaks were observed (e.g., Italy and South Korea), this of news does not necessarily reflect their impact), but was not always the rule (as suggested for the correlations also by the observation of search spikes related to health observed in Colombia and the US). awareness campaigns (e.g., World Diabetes Day), or celebrity-related events. As an example, the death of the Swedish TV presenter Adam Alsing on April 15 2020— Discussion who died of COVID-19 and was known to be at risk of In this study, we found that, during the COVID-19 pan- developing diabetes [22]—prompted the largest Swed- demic in 2020, there was a consistent increase in web ish number of searches on diabetes of the past 5  years searches on “asthma” observed in several countries, par- (observed on April 15–17). The largest Turkish GTs on ticularly in March. Such an increase was variably associ- COPD (occurring in the second quarter of March) also ated with information-seeking on asthma and COVID-19 appear to be related to the death of the Turkish com- simultaneously, and resembled the thunderstorm- mander Aytaç Yalman (who suffered from COPD) on induced asthma-related searches in Australia (i.e., no March 15 [23], as well as to the widely mediatized state- other situation over the past years has prompted such an ments by respiratory clinicians, including members of increase of asthma searches in all countries). Smaller and the Turkish Thoracic Society [24, 25]. less frequent search peaks were observed for COPD, with This study has important limitations that are worth the role of queries involving COVID-19-related terms discussing. Firstly, we limited our comparison to five appearing to be smaller. Such increased search activity chronic diseases, as GTs are provided on a relative scale was not consistently observed for the assessed non-res- (i.e., on a 0 to 100 scale, with 100 corresponding to the piratory chronic diseases. maximum volume of searches registered for the included S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 11 of 14 Fig. 2 Monthly average Google Trends for “asthma” (as a disease) between January and May of 2020 Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 12 of 14 Fig. 3 Monthly average Google Trends for “chronic obstructive pulmonary disease” (COPD) (as a disease) (2020, January–May) keywords in the selected location and period of time) variations might have been missed (particularly in coun- and do not allow the comparison of more than five que - tries whose native language is not fluently spoken by any ries simultaneously. However, we tried to select chronic of the authors of this manuscript), the impact of missing conditions whose symptoms could masquerade those of those expressions is not expected to be particularly large, COVID-19 or which are widely known to be associated as otherwise they would have been listed as top-related or with a worse COVID-19 prognosis. Additional limita- rising queries. Finally, an important GT limitation con- tions concern the queries used for retrieving GTs on cerns the geographical and demographic representative- searches involving both chronic diseases and COVID- ness of Internet users. In fact, Internet use is still highly 19-related search terms, and which could have resulted asymmetrical across different regions of the globe. Of the in an underestimation of the percentage of searches that 54 countries identified by GTs as “major countries”, only were COVID-19-related. In fact, due to the GT limita- two are located in Africa, which is home to one-seventh tion of characters, we were not able to build queries using of the world population. In addition, in each country, the every combination of chronic disease and COVID-19-re- elderly (among whom diseases such as COPD, hyper- lated terms. For the cases in which we were not able to tension or diabetes are more frequent) are particularly include all relevant top-related and/or rising queries, we underrepresented among Internet users [1], and literacy made sure that we selected the most popular ones. On may also influence the topics of online searches. the other hand, for countries in which no relevant top- This study also has relevant strengths. We assessed over related or rising queries were available, we had to build 50 countries worldwide and took a 5-year period into expressions ourselves, combining both chronic diseases account. In addition, we applied a time series approach to and COVID-19-related terms. While important search estimate whether the number of observed searches was S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 13 of 14 Ethics approval and consent to participate higher than that predicted based on the data of previ- Not applicable. ous years. Finally, we quantified the proportion of excess searches that may be related to COVID-19. Consent for publication Not applicable. In conclusion, this study suggests that, during the COVID-19 pandemic, there was an anomalous increase Competing interests in online searches on chronic respiratory diseases, The authors declare that they have no competing interests. which was partly accounted for by searches on COVID- Author details 19-related terms. There was also a less evident peak MEDCIDS–Department of Community Medicine, Information and Health for COPD. Such peaks were not regularly observed for Decision Sciences, Faculty of Medicine, University of Porto, Porto, Portugal. CINTESIS–Center for Health Technology and Services Research, University other chronic diseases. This study points to the inad - of Porto, Rua Dr. Plácido da Costa, 4200‑450 Porto, Portugal. Personalized equacy of GTs as an isolated tool to assess the epidemi- Medicine, Asthma & Allergy, Humanitas Clinical and Research Center, IRCCS, ology of chronic diseases (and, most notably, to assess Rozzano, Italy. Department of Biomedical Sciences, Humanitas 5 6 University, Pieve Emanuele, Milan, Italy. MASK‑Air, Montpellier, France. Medi‑ it prospectively), as search patterns can be highly influ - cal Consulting Czarlewski, Levallois, France. MACVIA‑France, Montpellier, enced by users’ concerns and media coverage. France. Department of Pulmonary Diseases, Cerrahpasa Faculty of Medi‑ cine, Istanbul University‑ Cerrahpasa, Istanbul, Turkey. ISGlobAL, Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain. Uni‑ Supplementary information versitat Pompeu Fabra (UPF), Barcelona, Spain. CIBER Epidemiología y Salud Supplementary information accompanies this paper at https ://doi. Pública (CIBERESP), Barcelona, Spain. Charité, Universitätsmedizin org/10.1186/s1360 1‑020‑00352 ‑9. ``Berlin, Humboldt‑Universität Zu Berlin, Berlin, Germany. Department of Dermatology and Allergy, Comprehensive Allergy Center, Berlin Institute of Health, Berlin, Germany. Centre Hospitalier Universitaire, Montpellier, Additional file 1: Table S1. Queries used to retrieve, for each country, France. Google Trends on searches involving both chronic diseases and Covid‑ 19‑related search terms. Table S2. Percentages of Google Trends on Received: 26 July 2020 Accepted: 17 October 2020 non‑respiratory chronic diseases (diabetes, hypertension, and Crohn’s disease) corresponding to (i) expected baseline searches, (ii) excess searches beyond those including Covid‑19‑related terms, and (iii) searches with Covid‑19‑related terms. Unless otherwise indicated, search peaks were observed in March. Table S3. Pearson correlation coefficients References between Google Trends (GT ) and Google News items on asthma for the 1. Eysenbach G. Infodemiology and infoveillance: framework for an periods of January–May 2020 and January–March 2020. Figure S1. 2020 emerging set of public health informatics methods to analyze search, Google Trends for “acute pneumonia” (as a disease), “asthma” (as a disease), communication and publication behavior on the Internet. J Med Inter‑ “chronic obstructive pulmonary disease” (COPD) (as a disease), “diabetes” net Res. 2009;11(1):e11. (as a disorder), “hypertension” (as a medical condition). Figure S2. Monthly 2. Barbosa M, Morais‑Almeida M, Sousa C, Bousquet J. The, “Big Five” lung average Google Trends for “diabetes” (as a disorder) between January and diseases in CoViD‑19 pandemic—a Google Trends analysis. Pulmonol‑ May of 2020. Figure S3. Monthly average Google Trends for “hyperten‑ ogy. 2020. https ://doi.org/10.1016/j.pulmo e.2020.06.008. sion” (as a medical condition) between January and May of 2020. Figure 3. Pandey A, Abdullah K, Drazner MH. Impact of Vice President Cheney on S4. Monthly average Google Trends for “Crohn’s disease” (as a disease) public interest in left ventricular assist devices and heart transplanta‑ between January and May of 2020. Figure S5. Weekly Google Trends and tion. Am J Cardiol. 2014;113(9):1529–31. Google News data on “asthma” in 19 countries. 4. Bousquet J, Agache I, Berger U, Bergmann KC, Besancenot JP, Bous‑ quet PJ, et al. Differences in reporting the ragweed pollen season using Google Trends across 15 countries. Int Arch Allergy Immunol. Abbreviations 2018;176(3–4):181–8. ARIMA: Autoregressive integrated moving average; COPD: Chronic obstructive 5. Bousquet J, Agache I, Anto JM, Bergmann KC, Bachert C, Annesi‑Mae ‑ pulmonary disease; GTs: Google Trends. sano I, et al. Google Trends terms reporting rhinitis and related topics differ in European countries. Allergy. 2017;72(8):1261–6. Acknowledgements 6. Bento AI, Nguyen T, Wing C, Lozano‑Rojas F, Ahn YY, Simon K. Evidence Not applicable. from internet search data shows information‑seeking responses to news of local COVID‑19 cases. Proc Natl Acad Sci USA. 2020;117(21):11220–2. Authors’ contributions 7. Sousa‑Pinto B, Anto A, Czarlewski W, Anto JM, Fonseca JA, Bousquet BSP, EH, JAF and JB participated in the study design. BSP, EH and AA partici‑ J. Assessment of the impact of media coverage in coronavirus‑ pated in the data extraction. BSP, EH, AA, WC, AB, BG, GWC, JMA, JAF and JB related Google Trends: infodemiology study. J Med Internet Res. participated in the data analysis, manuscript writing and critical review of the 2020;22(8):e19611. manuscript. All authors read and approved the final manuscript. 8. Higgins TS, Wu AW, Sharma D, Illing EA, Rubel K, Ting JY, et al. Cor‑ relations of online search engine trends with coronavirus disease Funding (COVID‑19) incidence: infodemiology study. JMIR Public Health Surveill. This paper was written by five members of DigitalHealthEurope Grant Agree ‑ 2020;6(2):e19702. ment Number: 826353 Support to a Digital Health and Care Innovation initia‑ 9. Szmuda T, Ali S, Hetzger TV, Rosvall P, Sloniewski P. Are online searches for tive in the context of Digital Single Market strategy, SC1‑HCC‑05‑2018. Publica‑ the novel coronavirus (COVID‑19) related to media or epidemiology? A tion of this article was supported by National Funds through FCT ‑ Fundação cross‑sectional Study. Int J Infect Dis. 2020;97:386–90. para a Ciência e a Tecnologia, I.P., within CINTESIS, R&D Unit (reference 10. Heerfordt C, Heerfordt IM. Has there been an increased interest in smok‑ UIDB/4255/2020). ing cessation during the first months of the COVID ‑19 pandemic? A Google Trends study. Public Health. 2020;183:6–7. Availability of data and materials 11. Jacobson NC, Lekkas D, Price G, Heinz MV, Song M, O’Malley AJ, et al. The datasets used and/or analysed during the current study are available from Flattening the mental health curve: COVID‑19 stay‑at ‑home orders the corresponding author on reasonable request. Sousa‑Pinto  et al. Clin Transl Allergy (2020) 10:47 Page 14 of 14 are associated with alterations in mental health search behavior in the 20. Morais‑Almeida M, Pite H, Aguiar R, Ansotegui I, Bousquet J. Asthma United States. JMIR Ment Health. 2020;7(6):e19347. and the coronavirus disease 2019 pandemic: a literature review. Int Arch 12. Misiak B, Szczesniak D, Koczanowicz L, Rymaszewska J. The COVID‑19 Allergy Immunol. 2020;181:680–8. outbreak and Google searches: is it really the time to worry about global 21. Yardi S, Caldwell PH, Barnes EH, Scott KM. Determining parents’ patterns mental health? Brain Behav Immun. 2020;87:126–7. of behaviour when searching for online information on their child’s 13. Searle T, Al‑Niaimi F, Ali FR. Dermatological insights from Google Trends: health. J Paediatr Child Health. 2018;54(11):1246–54. what does the public think is important during COVID‑19 lockdown? Clin 22. Bäsén A. Smittan slår mot ung och gammal. Expressen. 2020. https :// Exp Dermatol. 2020;45(7):891–921.www.expre ssen.se/kroni korer /anna‑basen /smitt an‑slar‑mot‑ung‑och‑ 14. Sinyor M, Spittal MJ, Niederkrotenthaler T. Changes in suicide and gamma l/. Accessed 29 June 2020. resilience‑related google searches during the early stages of the COVID ‑ 23. Bakan Koca’dan Aytaç Yalman açıklaması. Cumhuriyet. 2020. https :// 19 pandemic. Can J Psychiatry. 2020;65(10):741–3.www.cumhu riyet .com.tr/haber /bakan ‑kocad an‑aytac ‑yalma n‑acikl amasi 15. Riggioni C, Comberiati P, Giovannini M, Agache I, Akdis M, Alves‑ Correia ‑17282 73/. Accessed 29 June 2020. M, et al. A compendium answering 150 questions on COVID‑19 and 24. Astim ve KOAH hastalari için koronavirüs uyarilari. Medikal Teknik. 2020. SARS‑ CoV‑2. Allergy. 2020. https ://doi.org/10.1111/all.14449 .https ://www.medik altek nik.com.tr/astim ‑ve‑koah‑hasta lari‑icin‑koron 16. Gillison DH. Learn more about anxiety with a self‑assessment on Search. aviru s‑uyari lari/. Accessed 29 June 2020. 2020. https ://blog.googl e/techn ology /healt h/anxie ty‑self‑asses sment 25. Yener D. KOAH hastalarına koronavirüs uyarısı. Anadolu Ajansı. 2020. https ‑searc h/. Accessed 29 June 2020.://www.aa.com.tr/tr/koron aviru s/koah‑hasta larin a‑koron aviru s‑uyari 17. Mavragani A, Ochoa G. Google Trends in infodemiology and infoveillance: si‑/17816 15. Accessed 29 June 2020. methodology framework. JMIR Public Health Surveill. 2019;5(2):e13439. 18. Bousquet J, O’Hehir RE, Anto JM, D’Amato G, Mosges R, Hellings PW, et al. Publisher’s Note Assessment of thunderstorm‑induced asthma using Google Trends. J Springer Nature remains neutral with regard to jurisdictional claims in pub‑ Allergy Clin Immunol. 2017;140(3):891–3. lished maps and institutional affiliations. 19. Song X, Xiao J, Deng J, Kang Q, Zhang Y, Xu J. Time series analysis of influenza incidence in Chinese provinces from 2004 to 2011. Medicine. 2016;95(26):e3929. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Clinical and Translational Allergy Springer Journals

Anomalous asthma and chronic obstructive pulmonary disease Google Trends patterns during the COVID-19 pandemic

Loading next page...
 
/lp/springer-journals/anomalous-asthma-and-chronic-obstructive-pulmonary-disease-google-yk0jskxNwA
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2020
eISSN
2045-7022
DOI
10.1186/s13601-020-00352-9
Publisher site
See Article on Publisher Site

Abstract

Background: An increase in online searches on health topics may either mirror epidemiological changes or reflect media coverage. In the context of COVID‑19, this is particularly relevant, as COVID ‑19 symptoms may be mistaken for those of respiratory disease exacerbations. Therefore, we aimed to assess Internet search patterns on asthma and chronic obstructive pulmonary disease (COPD) in the context of COVID‑19, as compared to searches on other chronic diseases. Methods: We retrieved Google Trends (GTs) data on two respiratory (asthma and COPD) and three non‑respiratory (diabetes, hypertension, and Crohn’s disease) chronic diseases over the past 5 years (up to May 31, 2020). For 54 coun‑ tries, and for each disease, we built autoregressive integrated moving average (ARIMA) models to predict GTs for 2020 based on 2015–2019 search patterns. In addition, we estimated the proportion of searches in which COVID‑19‑related terms were used. To assess the potential impact of media coverage on online searches, we assessed whether weekly “asthma” GTs correlated with the number of Google News items on asthma. Results: Over the past 5 years, worldwide search volumes for asthma and COPD reached their maximum values in March 2020. Such was not observed for diabetes, hypertension and Crohn’s disease. In 38 (70%) countries, GTs on asthma were higher in March 2020 than the respective maximum predicted values. This compares to 19 countries for COPD, 23 for hypertension, 11 for Crohn’s disease, and 9 for diabetes. Queries with COVID‑19‑related terms repre ‑ sented up to 47.8% of the monthly searches on asthma, and up to 21.3% of COPD searches. In most of the assessed countries, moderate‑strong correlations were observed between “asthma” GTs and the number of news items on asthma. Conclusions: During March 2020, there was a peak in searches on asthma and COPD, which was probably mostly driven by media coverage, as suggested by their simultaneity in several countries with different epidemiological situations. Keywords: Asthma, Chronic diseases, Chronic obstructive pulmonary disease, COVID‑19, Google trends Introduction Google Trends (GTs), a web-based surveillance tool, can *Correspondence: bernardo@med.up.pt Bernardo Sousa‑Pinto and Enrico Heffler contributed equally to this provide insights into the real-life epidemiology of dis- manuscript eases and outbreaks. This tool provides information—on CINTESIS–Center for Health Technology and Services Research, a relative scale—on how often a certain keyword or query University of Porto, Rua Dr. Plácido da Costa, 4200‑450 Porto, Portugal Full list of author information is available at the end of the article is searched, allowing to compare different regions, time © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/publi cdoma in/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 2 of 14 periods, or keywords. However, as GTs assess individuals’ trend s.googl e.com/; Google, LLC, Mountain View, CA, health information-seeking behaviour, data do not often USA) for 54 countries identified by GTs as “major coun - reflect the true epidemiological situation of the searched tries”, including 23 in Europe, 19 in Asia and the Pacific, conditions [1, 2]. In fact, there are cases that describe ten in the Americas, and two in Africa. We adopted media coverage being associated with anomalously high a time series approach and assessed in detail how the online interest on many health topics, such as coronary COVID-19 pandemic impacted search patterns on these heart disease [3], pollen counts [4, 5] or COVID-19 [6, 7]. diseases. In the context of COVID-19, several GT-based stud- ies have been conducted with the aim of assessing Disease and keyword selection whether online search data correlated with the number We focused on asthma and COPD and included three of COVID-19 cases and deaths. Variable results have non-respiratory chronic diseases (diabetes, hyperten- been observed [8, 9]. In addition, GTs have been used to sion, and Crohn’s disease) for comparison. The three assess variations in online searches for health topics, with non-respiratory chronic diseases were selected on the particular focus on mental health and behaviour-related grounds that (i) diabetes and hypertension are common searches [10–14]. In fact, different studies consistently comorbid conditions that have been associated with a found a decrease in searches for suicide- and depression/ worse COVID-19 prognosis and (ii) Crohn’s disease—like anxiety-related terms in the initial phase of the COVID- asthma—is relatively frequent in young people (who are 19 pandemic [11, 12, 14]. Search patterns on respiratory the most active Internet users), and can manifest as diar- diseases, however, have been less often assessed. While rhoea (which may also occur in COVID-19). We did not a preliminary study visually identified anomalous online assess any other chronic disease, as GTs limit the number search interest for asthma occurring simultaneously in of simultaneously compared queries to five. In particular, several countries of the Northern and Southern Hemi- we did not assess rhinitis as it does not appear to be asso- spheres (Bousquet et  al., unpublished data), it is unclear ciated with COVID-19 searches (Bousquet, submitted). as to what has been driving such unparalleled search In addition, GTs for chronic respiratory diseases were interest, and whether similar search patterns also occur plotted along GTs for acute pneumonia. Searches for with other respiratory and non-respiratory chronic acute pneumonia were used as a proxy for searches for diseases. Understanding whether the perception of coronavirus/COVID-19 (as the search volume for the symptoms of chronic respiratory diseases may be mas- latter is so large that comparisons with chronic diseases querading those of COVID-19 [15], or whether searches are impossible), since searches on these two concepts are being driven mostly by users’ curiosity/concerns, reached their maximum values at the same time through- may have potentially relevant implications. Such implica- out 2020 (Bousquet, unpublished). tions concern, among others: (i) the usefulness of GTs in On account of the selected diseases, we retrieved GTs the epidemiological monitoring of chronic diseases, (ii) data on the following keywords (as “topics”): “asthma”, the way the occurrence of COVID-19 in patients with “chronic obstructive pulmonary disease”, “diabetes”, chronic respiratory diseases is being discussed in the “hypertension”, and “Crohn’s disease”. For pneumonia, media, or is being communicated to patients, and (iii) the the keyword “acute pneumonia” (as “topic”) was used pertinence of Google providing health screening ques- (of note, in GTs, “topics” are groups of search terms that tionnaires following searches on certain expressions [16]. share the same concept [17]; “asthma”, “chronic pulmo- Therefore, in this infodemiological study, we aimed to nary obstructive disease”, “Crohn’s disease” and “acute quantify whether there was an increased search activity pneumonia” are classified by Google as being “disease on two chronic respiratory diseases—asthma and chronic topics”; “diabetes” as a “disorder topic”; and “hyperten- obstructive pulmonary disease (COPD)—in the context sion” as a “medical condition topic”). Along with GTs of the COVID-19 pandemic. In addition, we aimed to on these keywords, we retrieved GTs data on searches assess whether such eventual abnormal search activity (i) involving each chronic disease and COVID-19-related could also be observed in other chronic diseases, and (ii) terms (this allowed us to quantify how much the 2020 was associated with COVID-19-related searches. GTs peaks on chronic diseases were driven by COVID- 19-related searches). For each country, we built a query Methods in its native language(s), consisting of terms specific to We assessed online searches for two respiratory diseases each chronic disease along with COVID-19-related terms (asthma and COPD) and three non-respiratory chronic (Additional file  1: Table S1). Whenever available, we used diseases over the past 5 years up until May 31, 2020. This top-related or rising query expressions (starting on the period includes the first months of the COVID-19 pan - most popular and until the character limit was reached). demic. Online searches were assessed using GTs (https :// In the absence of relevant top-related or rising queries, S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 3 of 14 we combined the most popular terms to search for each the observed GT exceeded the respective maximum pre- chronic disease along with the most popular terms to dicted value. search for COVID-19/coronavirus. Subsequently, for the year of 2020, we considered that GTs on each of the five selected chronic diseases (i.e., Data analysis total volume of searches in a given period of time) could Google Trends values represent the Google search inter- be divided into two components: (i) searches without any est over time for a given topic as a proportion of all COVID-19-related term, and (ii) online searches on each searches on all topics on Google at that time and loca- chronic disease along with COVID-19-related terms (i.e., tion. Values are indexed to 100, where 100 is the maxi- “COVID-19 related-searches” estimated for each coun- mum search interest for the time and location selected. try, using the queries listed in Additional file  1: Table S1). The values are re-indexed according to the selected time For each month, we calculated the average proportion period. that the latter represented among GTs on each chronic We started by analysing the worldwide search inter- disease. In addition, for each month, we subtracted the est patterns of these five chronic diseases over the past average GTs on each chronic disease + COVID-19-re- 5 years (up to May 31, 2020), to visually assess the pres- lated terms from the average total GTs on each chronic ence of spikes during the COVID-19 pandemic. For this disease. The difference was compared with the respective assessment, GTs on chronic diseases were plotted along predicted value as estimated by previously described sea- GTs for “acute pneumonia”. As a particular case, we com- sonal ARIMA models. This allowed us to assess whether pared the volume of searches subsequent to the thun- there might be an excess of searches on asthma beyond derstorm-asthma of Australia (November 2016) with that explained by queries including COVID-19-related COVID-19-associated searches in asthma, as the former terms. was the largest “asthma” spike that had previously been Finally, to preliminarily assess the impact of media retrieved worldwide [18]. coverage on online searches, we estimated the correla- Subsequently, we studied the search patterns of the tions between GTs and Google News (https ://news.googl five aforementioned chronic diseases during 2020 (Janu - e.com/; Google, LLC, Mountain View, CA, USA) items ary–May) in the 54 countries identified by GTs as “major on asthma in 19 different countries. For each country, countries”. Our aim was to assess whether the search we retrieved the weekly number of Google News search interest values on chronic diseases in each of these coun- results (i.e., searches in news items, which differ from tries exceeded those that would be expected based on the Google News aggregator service present in several patterns from the previous years. For this assessment, we countries) when searching the query “asthma” in the built seasonal autoregressive integrated moving average respective language and applying the respective coun- (ARIMA) models to predict GTs for 2020 based on the try and language restriction filters. Unrelated results GT patterns from 2015–2019. Seasonal ARIMA models (namely those which had only been retrieved because are defined by the parameters (p, d, q)(P, D, Q) , with p the respective websites advertised news for asthma) were corresponding to the order of autoregression, d to the not counted. Correlations were estimated by computing degree of difference, q to the order of the moving aver - Pearson correlation coefficients. age part, P to the seasonal order of autoregression, D to Data analysis was performed using software R version the seasonal integration, Q to the seasonal moving aver- 4.0.0 (R Foundation for Statistical Computing, Vienna, age, and s to the length of the seasonal period [19] (for an Austria). example of the use of seasonal ARIMA models for health forecasting, as well as for a discussion on their meth- Results odological strengths and limitations, please consult the Fivey ‑ ear searches for chronic diseases study of Song et  al. [19]). In this study, we applied sea- When visually analysing 5-year data from all countries sonal ARIMA(3,0,2)(0,1,1) models, using weekly GT combined, we observed that asthma and (on a lesser data (thus explaining the length of the seasonal period— scale) COPD searches reached their maximum values in s—being 52). For each model, we retrieved the maximum March 2020, simultaneously with a search spike on acute values—for the whole year of 2020, and for the month pneumonia (Fig.  1). In Australia, the maximum volume of March—of the upper bound of 95% confidence inter - of asthma searches in March 2020 was 23% lower than vals of predicted GTs (“maximum predicted values”). that observed in the week of November 20–26, 2016 Such maximum values were compared with the maxi- (associated with the thunderstorm-induced asthma). On mum observed GTs for the year of 2020 (January–May) the other hand, there were only two countries where the and for the month of March. A search peak was formally 2020 GTs for COPD reached higher values than those for defined as any situation in which, for a given search term, asthma: in Hungary, COPD maximum values occurred Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 4 of 14 Fig. 1 Worldwide 5‑ year Google Trends for respiratory and non‑respiratory chronic diseases 2  weeks after those for asthma, whereas in Turkey, they Peaks for non-respiratory chronic diseases were not as occurred simultaneously (Additional file 1: Fig. S1). geographically and/or temporally consistent as those for For non-respiratory chronic diseases, no worldwide asthma or COPD. Throughout 2020, the monthly average search spikes were visually identified in 2020 (Fig.  1). By of GTs on hypertension exceeded the predicted values in contrast, we identified annual spikes for diabetes asso - 23 countries (42.6%), mostly those in Europe (n = 12) and ciated with the World Diabetes Day. Visually assessing Latin America (n = 6). However, most GT peaks occurred 2020 data for each specific country (Additional file  1: in April or May (n = 14), including those observed in all Figs. S1, S2), large diabetes spikes were found for three Latin American countries. Peaks for diabetes mellitus specific countries, namely Italy (starting during the onset and Crohn’s disease were observed in fewer countries of the COVID-19 pandemic and having a 2-month dura- (n = 9 and n = 11, respectively), and were highly variable tion), Romania (week of March 29) and Sweden (week of depending on their region, month, and magnitude (of April 12). In these three cases, maximum 2020 GTs were note, Crohn’s disease peaks were frequently identified in greater than GTs in the “World Diabetes Day” weeks. Middle Eastern countries, a fact that might be related to typos in users’ queries, given the similitude of the Arabic Quantification of search peaks for chronic diseases in 2020 and Farsi words for “Crohn” and “corona”). In March 2020, search peaks for asthma GTs were identified in 38 out of the 54 studied countries (70.4%) Disentangling chronic diseases and COVID‑19 searches (Tables  1, 2). Such peaks were observed in the assessed Out of the 38 countries in which asthma search peaks countries of Europe (apart from Romania, Russia  and were identified, 28 (73.7%) had top-related or rising- Ukraine), in the Americas (apart from Mexico), in Aus- related queries involving COVID-19-related terms. On tralia and New Zealand, but only in one third of Asian the other hand, this occurred in five out of 22 (23%) countries. Search peaks for COPD were temporally con- countries for COPD, 11 out of 23 (48%) for hypertension, sistent with those of asthma, but were only observed for five out of nine (56%) for diabetes, and in 0 out of 11 for 19 countries, mostly those located in Central Europe, Crohn’s disease. North America, and the Pacific. However, COPD search In March 2020, asthma COVID-19-related searches peaks were overall smaller than those observed for were detected in all countries except Egypt, representing asthma. between 4.4% (for the Philippines and India) and 47.8% S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 5 of 14 Table 1 Observed and predicted maximum values for Google Trends (GT) on five chronic diseases (January–May 2020) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) (2020) Europe Austria 67 33 42 38 73 94 45 49 21 28 Belgium 67 32 33 24 83 91 41 47 28 32 Bulgaria 30 28 17 18 93 100 78 78 11 19 Czech Repub 37 35 30 18 87 92 32 35 30 43 lic Denmark 87 30 26 31 84 93 13 14 18 16 Finland 76 38 15 17 75 86 31 28 18 20 France 100 27 24 13 68 61 51 44 28 31 Germany 87 30 53 39 79 89 41 44 18 24 Greece 52 20 27 15 86 100 34 37 17 34 Hungary 34 20 100 14 64 68 50 47 16 20 Ireland 100 36 45 31 84 78 49 43 29 28 Italy 41 29 11 12 100 32 74 68 32 33 Netherlands 64 22 27 23 64 82 34 37 14 18 Norway 100 35 17 27 58 70 49 32 20 25 Poland 28 18 11 9 70 79 25 33 20 12 Portugal 50 25 9 15 59 75 28 31 16 24 Romania 19 22 7 10 100 97 42 31 6 11 Russia 26 32 9 10 68 97 34 41 7 10 Spain 62 21 15 23 57 62 66 67 17 59 Sweden 41 15 14 15 100 44 37 20 7 8 Switzerland 100 26 28 22 56 59 62 35 14 21 Ukraine 24 25 10 8 91 100 39 47 11 11 United King 100 13 35 13 58 43 30 18 12 14 dom Africa Egypt 16 8 3 4 87 86 13 15 69 4 South Africa 32 22 10 11 70 72 75 55 9 9 North America Canada 69 24 31 24 96 100 47 55 18 24 USA 60 28 28 27 99 100 61 66 16 22 Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 6 of 14 Table 1 (continued) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) value (2020) (2020) Latin America Argentina 21 10 7 9 30 37 30 32 3 100 Brazil 52 29 11 9 87 96 72 65 8 29 Chile 49 28 12 19 69 87 73 74 8 12 Colombia 26 25 35 20 52 63 100 81 4 25 Ecuador 23 23 12 12 56 71 75 58 9 8 Mexico 19 21 12 12 76 87 69 67 4 5 Peru 42 33 7 9 76 89 74 60 5 6 Venezuela 31 32 13 12 54 63 71 71 5 10 Asia Hong Kong 25 33 14 16 87 100 47 63 8 14 India 24 20 10 11 78 79 37 34 3 6 Indonesia 53 54 7 10 95 100 93 100 2 2 Iran 78 82 1 2 20 72 16 19 45 4 Israel 27 22 10 11 85 98 23 29 23 25 Japan 70 48 12 9 90 94 29 34 6 11 Malaysia 36 45 11 13 90 100 71 80 4 5 Pakistan 36 42 17 21 100 100 69 77 8 7 Philippines 87 61 20 23 90 100 100 91 6 9 Saudi Arabia 13 16 3 4 86 84 9 12 46 5 Singapore 20 22 7 10 61 81 37 86 4 7 South Korea 17 23 12 13 80 93 29 45 32 13 Taiwan 27 29 9 13 95 100 59 68 4 5 Thailand 18 21 10 12 96 100 58 62 3 3 Turkey 65 35 100 26 65 79 46 31 7 10 UAE 22 21 6 9 64 82 30 34 12 9 Vietnam 18 20 6 9 100 89 34 36 2 2 Pacific Australia 77 53 28 25 88 94 46 47 17 24 New Zealand 91 42 39 27 76 100 34 52 35 28 Maximum predicted values correspond to the maximum values of the upper bound of the 95% confidence intervals for predicted asthma GT. Numbers in italic indicate cases in which maximum GT observed values were higher than maximum predicted values COPD Chronic Obstructive Pulmonary Disease, UAE United Arab Emirates, USA United States of America S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 7 of 14 Table 2 Observed and predicted maximum values for Google Trends (GT) on five chronic diseases (March 2020) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (March value (March value (March value (March value (March value (March value (March value (March value (March (March 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) Europe Austria 67 29 42 37 71 94 45 45 21 26 Belgium 67 30 21 20 69 88 38 42 25 27 Bulgaria 27 26 17 14 86 97 78 75 6 17 Czech Repub 37 35 14 17 63 92 32 34 18 40 lic Denmark 87 27 26 31 82 92 13 14 14 14 Finland 76 34 10 15 67 77 31 27 13 17 France 100 25 24 12 68 60 51 44 22 31 Germany 87 30 53 34 68 88 41 44 18 20 Greece 52 20 27 15 73 93 31 33 13 32 Hungary 34 18 100 12 37 65 39 47 10 20 Ireland 100 29 45 28 84 78 49 40 29 28 Italy 41 29 11 12 78 32 74 67 13 32 Netherlands 64 21 27 21 60 74 29 37 14 15 Norway 100 28 11 23 55 70 49 32 14 23 Poland 28 18 11 8 44 78 24 33 10 12 Portugal 50 25 9 15 52 70 18 26 15 22 Romania 18 18 7 9 100 94 42 31 4 11 Russia 26 29 9 10 61 97 33 41 6 9 Spain 62 19 15 23 56 61 65 64 13 39 Sweden 41 14 12 15 41 44 32 20 7 8 Switzerland 100 26 28 22 56 59 62 34 14 21 Ukraine 24 25 10 8 75 100 37 47 7 10 United King 100 13 35 12 58 42 30 18 12 14 dom Africa Egypt 8 8 3 4 55 86 10 14 69 4 South Africa 32 20 9 11 60 70 40 44 7 8 North America Canada 69 23 31 24 95 100 46 55 18 22 USA 60 28 28 24 99 100 59 64 16 22 Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 8 of 14 Table 2 (continued) Asthma COPD Diabetes mellitus Hypertension Crohn’s disease Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum Maximum GT GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed GT predicted GT observed predicted value value (March value (March value (March value (March value (March value (March value (March value (March value (March (March 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) 2020) Latin America Argentina 21 8 6 8 30 35 30 27 3 19 Brazil 52 25 11 8 87 96 72 59 8 25 Chile 49 26 11 17 64 87 64 69 8 10 Colombia 26 22 35 19 46 62 100 74 3 22 Ecuador 23 18 7 10 46 63 75 49 9 8 Mexico 19 19 12 12 76 87 69 67 3 5 Peru 39 26 5 9 64 87 57 51 4 6 Venezuela 31 28 5 12 48 63 71 71 3 10 Asia Hong Kong 23 33 14 15 69 100 43 63 5 14 India 22 19 10 10 61 79 31 33 3 4 Indonesia 53 53 7 10 87 100 83 100 1 2 Iran 68 82 1 2 16 34 16 16 6 4 Israel 27 20 9 10 52 98 18 25 19 20 Japan 51 43 12 8 68 88 26 30 4 9 Malaysia 35 40 11 13 90 100 59 79 2 5 Pakistan 30 37 12 18 66 100 55 69 7 7 Philippines 87 59 20 20 78 100 81 91 5 7 Saudi Arabia 13 16 2 3 59 83 8 12 46 5 Singapore 19 20 7 9 47 78 37 81 3 7 South Korea 17 20 10 12 63 92 27 45 7 11 Taiwan 24 29 9 12 82 100 45 68 2 5 Thailand 16 20 10 12 81 100 40 56 3 3 Turkey 65 34 100 22 60 76 40 30 6 10 UAE 22 18 6 6 55 73 22 33 6 8 Vietnam 16 20 9 9 95 88 29 36 1 2 Pacific Australia 77 48 28 24 85 94 45 42 17 23 New Zealand 91 36 39 25 76 100 33 42 19 28 Maximum predicted values correspond to the maximum values of the upper bound of the 95% confidence intervals for predicted asthma GT. Numbers in italic indicate cases in which maximum GT observed values were higher than maximum predicted values COPD Chronic obstructive pulmonary disease, USA United States of America S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 9 of 14 Table 3 Expected and excess Google Trends on asthma and chronic obstructive pulmonary disease (COPD) Expected baseline Excess searches Searches on asthma Expected Excess searches Searches on COPD searches on asthma on asthma beyond those with Covid‑19‑related baseline on COPD beyond those with Covid‑19‑related (%) including Covid‑19‑ terms (%) searches including Covid‑19‑ terms (%) related terms (%) on COPD (%) related terms (%) Europe Austria 52.0 1.6 46.4 75.5 6.6 17.9 Belgium 36.1 35.7 28.2 65.6 21.3 13.1 a a a b b b Bulgaria – – – 44.9 55.1 0 c c c Czech Republic 68.7 22.7 8.6 30.2 69.8 0 a a a Denmark 36.5 27.5 36.0 – – – a a a Finland 41.6 30.3 28.1 – – – France 32.9 32.9 34.2 41.1 41.6 17.3 Germany 38.8 20.4 40.8 53.1 28.8 18.1 Greece 41.7 47.9 10.4 49.6 44.1 6.3 Hungary 56.8 29.4 13.8 21.8 78.2 0 Ireland 24.2 45.5 30.3 57.1 29.7 13.2 a a a Italy 65.4 0 34.6 – – – Netherlands 40.3 20.3 39.4 58.8 19.9 21.3 a a a Norway 28.5 31.3 40.2 – – – Poland 62.5 16.2 21.3 69.6 30.4 0 a a a Portugal 41.8 38.5 19.8 – – – a a a Spain 33.6 18.6 47.8 – – – a a a Sweden 31.4 28.3 40.3 – – – Switzerland 29.1 33.5 37.4 67.9 32.1 0 a a a Ukraine – – – 64.9 35.1 0 United Kingdom 20.8 37.3 41.9 43.2 40.2 16.6 Africa c c c a a a Egypt 42.8 57.2 0 – – – a a a South Africa 54.0 26.6 19.4 – – – North America Canada 41.4 33.7 24.9 79.1 15.5 5.4 USA 51.0 20.6 28.4 89.9 2.8 7.3 Latin America a a a Argentina 28.9 41.7 29.4 – – – Brazil 53.7 25.3 21.0 78.5 16.9 4.6 a a a Chile 66.8 14.9 18.3 – – – b b b Colombia 83.5 5.8 10.7 51.2 40.3 8.5 a a a Ecuador 71.5 16.5 12.0 – – – c c c a a a Peru 71.8 20.3 7.9 – – – Venezuela 83.3 9.9 6.8 89.1 10.9 0 Asia b b b a a a India 64.5 31.1 4.4 – – – b b b a a a Israel 65.7 34.3 3.2 – – – b b b Japan 67.7 11.0 21.3 63.4 24.1 12.5 a a a Philippines 64.6 31.0 4.4 – – – Turkey 56.3 37.2 6.5 38.1 58.4 3.5 Pacific Australia 62.0 14.9 23.1 78.8 16.1 5.1 New Zealand 38.9 30.5 30.6 52.8 47.2 0 Percentages of Google Trends on asthma and COPD corresponding to (i) expected baseline searches, (ii) excess searches beyond those including Covid–19‑related terms, and (iii) searches with Covid‑19‑related terms. Unless otherwise indicated, search peaks were observed in March USA United States of America No search peak observed (of note, no search peak for either asthma or COPD was observed for Hong Kong, Indonesia, Iran, Malaysia, Mexico, Pakistan, Romania, Russia, Saudi Arabia, Singapore, South Korea, Taiwan, Thailand, United Arab Emirates or Vietnam) Search peak occurred in April Search peak occurred in May Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 10 of 14 (for Spain) of the GTs on that disease (Table  3; Fig.  2). Asthma primarily affects the respiratory system, and Overall, the percentage of COVID-19-related searches some of the main symptoms of COVID-19 are also res- was higher in European countries, reaching over 40% in piratory. This fact, along with a relative lack of infor - six of them. By contrast, apart from New Zealand, the mation (particularly when compared to diabetes or percentage of COVID-19-related searches did not exceed hypertension) on whether asthma can be associated 30% in any of the non-European countries. We also with a worse prognosis of COVID-19 [20], may partly observed a variable excess of searches on asthma beyond explain the particularly evident search increase observed those explained by queries including COVID-19-re- for asthma. Another possible explanation concerns the lated terms. In March 2020, such an excess represented fact that young adults, and especially parents, are par- between 0% (for Italy) and 47.9% (for Greece) of the GTs ticularly active Internet users [21]. In fact, asthma is on “asthma”. relatively common among young adults and even more The maximum percentage of COVID-19-related among children (in relation to whom, parents may wish searches was 21.3% for COPD (the Netherlands) (Table 3; to seek health information). Such an increase may be Fig.  3). For non-respiratory chronic diseases, 20.2% was smaller for COPD as the latter (i) is more frequent at a reached for diabetes (United Kingdom), 20.5% for hyper- more advanced age (with the elderly being less active on tension (Switzerland), and 4.7% for Crohn’s disease the Internet than the younger [1]), and (ii) is less known (Egypt) (Additional file  1: Table  S2, Figs. S2–S4). The among the general public. In fact, regarding COPD, some number of countries for which COVID-19-related search of the most frequent top-related queries consisted of just represented < 1% of all GTs was eight for COPD, com- asking what COPD was (data not shown). pared to four for diabetes, five for hypertension, and nine This study suggests that GTs alone may be inadequate for Crohn’ disease. for prospectively assessing the epidemiology of chronic In most countries, the number of Google News items diseases, and questions Google’s strategy of displaying on asthma reached their maximum value in March 2020 screening questionnaires when searching for key expres- (Additional file  1: Fig.  S5). The subsequent pattern was sions [16]. In fact, in this study, we found that asthma less consistent across countries, the number of asthma search peaks occurred simultaneously in several coun- news remaining high in some and decreasing in others tries in the Northern and Southern hemispheres, irre- (often with new rises). Therefore, while GTs and Google spective of the COVID-19 epidemiological situation (as News displayed moderate-strong correlations for the suggested by the relatively small Italian search peak) or whole of 2020 (often reaching their maximum values in of environmental phenomena. This suggests that media the same week), such correlations were stronger when coverage plays a major role in influencing GTs, as cor - specifically assessing the months of January to March roborated not only by the moderate-strong correlations (Additional file  1: Table  S3). While weaker correlations observed with the frequency of Google News items were found for countries in which small or no GT asthma (which should be carefully interpreted, as the amount peaks were observed (e.g., Italy and South Korea), this of news does not necessarily reflect their impact), but was not always the rule (as suggested for the correlations also by the observation of search spikes related to health observed in Colombia and the US). awareness campaigns (e.g., World Diabetes Day), or celebrity-related events. As an example, the death of the Swedish TV presenter Adam Alsing on April 15 2020— Discussion who died of COVID-19 and was known to be at risk of In this study, we found that, during the COVID-19 pan- developing diabetes [22]—prompted the largest Swed- demic in 2020, there was a consistent increase in web ish number of searches on diabetes of the past 5  years searches on “asthma” observed in several countries, par- (observed on April 15–17). The largest Turkish GTs on ticularly in March. Such an increase was variably associ- COPD (occurring in the second quarter of March) also ated with information-seeking on asthma and COVID-19 appear to be related to the death of the Turkish com- simultaneously, and resembled the thunderstorm- mander Aytaç Yalman (who suffered from COPD) on induced asthma-related searches in Australia (i.e., no March 15 [23], as well as to the widely mediatized state- other situation over the past years has prompted such an ments by respiratory clinicians, including members of increase of asthma searches in all countries). Smaller and the Turkish Thoracic Society [24, 25]. less frequent search peaks were observed for COPD, with This study has important limitations that are worth the role of queries involving COVID-19-related terms discussing. Firstly, we limited our comparison to five appearing to be smaller. Such increased search activity chronic diseases, as GTs are provided on a relative scale was not consistently observed for the assessed non-res- (i.e., on a 0 to 100 scale, with 100 corresponding to the piratory chronic diseases. maximum volume of searches registered for the included S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 11 of 14 Fig. 2 Monthly average Google Trends for “asthma” (as a disease) between January and May of 2020 Sousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 12 of 14 Fig. 3 Monthly average Google Trends for “chronic obstructive pulmonary disease” (COPD) (as a disease) (2020, January–May) keywords in the selected location and period of time) variations might have been missed (particularly in coun- and do not allow the comparison of more than five que - tries whose native language is not fluently spoken by any ries simultaneously. However, we tried to select chronic of the authors of this manuscript), the impact of missing conditions whose symptoms could masquerade those of those expressions is not expected to be particularly large, COVID-19 or which are widely known to be associated as otherwise they would have been listed as top-related or with a worse COVID-19 prognosis. Additional limita- rising queries. Finally, an important GT limitation con- tions concern the queries used for retrieving GTs on cerns the geographical and demographic representative- searches involving both chronic diseases and COVID- ness of Internet users. In fact, Internet use is still highly 19-related search terms, and which could have resulted asymmetrical across different regions of the globe. Of the in an underestimation of the percentage of searches that 54 countries identified by GTs as “major countries”, only were COVID-19-related. In fact, due to the GT limita- two are located in Africa, which is home to one-seventh tion of characters, we were not able to build queries using of the world population. In addition, in each country, the every combination of chronic disease and COVID-19-re- elderly (among whom diseases such as COPD, hyper- lated terms. For the cases in which we were not able to tension or diabetes are more frequent) are particularly include all relevant top-related and/or rising queries, we underrepresented among Internet users [1], and literacy made sure that we selected the most popular ones. On may also influence the topics of online searches. the other hand, for countries in which no relevant top- This study also has relevant strengths. We assessed over related or rising queries were available, we had to build 50 countries worldwide and took a 5-year period into expressions ourselves, combining both chronic diseases account. In addition, we applied a time series approach to and COVID-19-related terms. While important search estimate whether the number of observed searches was S ousa‑Pinto et al. Clin Transl Allergy (2020) 10:47 Page 13 of 14 Ethics approval and consent to participate higher than that predicted based on the data of previ- Not applicable. ous years. Finally, we quantified the proportion of excess searches that may be related to COVID-19. Consent for publication Not applicable. In conclusion, this study suggests that, during the COVID-19 pandemic, there was an anomalous increase Competing interests in online searches on chronic respiratory diseases, The authors declare that they have no competing interests. which was partly accounted for by searches on COVID- Author details 19-related terms. There was also a less evident peak MEDCIDS–Department of Community Medicine, Information and Health for COPD. Such peaks were not regularly observed for Decision Sciences, Faculty of Medicine, University of Porto, Porto, Portugal. CINTESIS–Center for Health Technology and Services Research, University other chronic diseases. This study points to the inad - of Porto, Rua Dr. Plácido da Costa, 4200‑450 Porto, Portugal. Personalized equacy of GTs as an isolated tool to assess the epidemi- Medicine, Asthma & Allergy, Humanitas Clinical and Research Center, IRCCS, ology of chronic diseases (and, most notably, to assess Rozzano, Italy. Department of Biomedical Sciences, Humanitas 5 6 University, Pieve Emanuele, Milan, Italy. MASK‑Air, Montpellier, France. Medi‑ it prospectively), as search patterns can be highly influ - cal Consulting Czarlewski, Levallois, France. MACVIA‑France, Montpellier, enced by users’ concerns and media coverage. France. Department of Pulmonary Diseases, Cerrahpasa Faculty of Medi‑ cine, Istanbul University‑ Cerrahpasa, Istanbul, Turkey. ISGlobAL, Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain. Uni‑ Supplementary information versitat Pompeu Fabra (UPF), Barcelona, Spain. CIBER Epidemiología y Salud Supplementary information accompanies this paper at https ://doi. Pública (CIBERESP), Barcelona, Spain. Charité, Universitätsmedizin org/10.1186/s1360 1‑020‑00352 ‑9. ``Berlin, Humboldt‑Universität Zu Berlin, Berlin, Germany. Department of Dermatology and Allergy, Comprehensive Allergy Center, Berlin Institute of Health, Berlin, Germany. Centre Hospitalier Universitaire, Montpellier, Additional file 1: Table S1. Queries used to retrieve, for each country, France. Google Trends on searches involving both chronic diseases and Covid‑ 19‑related search terms. Table S2. Percentages of Google Trends on Received: 26 July 2020 Accepted: 17 October 2020 non‑respiratory chronic diseases (diabetes, hypertension, and Crohn’s disease) corresponding to (i) expected baseline searches, (ii) excess searches beyond those including Covid‑19‑related terms, and (iii) searches with Covid‑19‑related terms. Unless otherwise indicated, search peaks were observed in March. Table S3. Pearson correlation coefficients References between Google Trends (GT ) and Google News items on asthma for the 1. Eysenbach G. Infodemiology and infoveillance: framework for an periods of January–May 2020 and January–March 2020. Figure S1. 2020 emerging set of public health informatics methods to analyze search, Google Trends for “acute pneumonia” (as a disease), “asthma” (as a disease), communication and publication behavior on the Internet. J Med Inter‑ “chronic obstructive pulmonary disease” (COPD) (as a disease), “diabetes” net Res. 2009;11(1):e11. (as a disorder), “hypertension” (as a medical condition). Figure S2. Monthly 2. Barbosa M, Morais‑Almeida M, Sousa C, Bousquet J. The, “Big Five” lung average Google Trends for “diabetes” (as a disorder) between January and diseases in CoViD‑19 pandemic—a Google Trends analysis. Pulmonol‑ May of 2020. Figure S3. Monthly average Google Trends for “hyperten‑ ogy. 2020. https ://doi.org/10.1016/j.pulmo e.2020.06.008. sion” (as a medical condition) between January and May of 2020. Figure 3. Pandey A, Abdullah K, Drazner MH. Impact of Vice President Cheney on S4. Monthly average Google Trends for “Crohn’s disease” (as a disease) public interest in left ventricular assist devices and heart transplanta‑ between January and May of 2020. Figure S5. Weekly Google Trends and tion. Am J Cardiol. 2014;113(9):1529–31. Google News data on “asthma” in 19 countries. 4. Bousquet J, Agache I, Berger U, Bergmann KC, Besancenot JP, Bous‑ quet PJ, et al. Differences in reporting the ragweed pollen season using Google Trends across 15 countries. Int Arch Allergy Immunol. Abbreviations 2018;176(3–4):181–8. ARIMA: Autoregressive integrated moving average; COPD: Chronic obstructive 5. Bousquet J, Agache I, Anto JM, Bergmann KC, Bachert C, Annesi‑Mae ‑ pulmonary disease; GTs: Google Trends. sano I, et al. Google Trends terms reporting rhinitis and related topics differ in European countries. Allergy. 2017;72(8):1261–6. Acknowledgements 6. Bento AI, Nguyen T, Wing C, Lozano‑Rojas F, Ahn YY, Simon K. Evidence Not applicable. from internet search data shows information‑seeking responses to news of local COVID‑19 cases. Proc Natl Acad Sci USA. 2020;117(21):11220–2. Authors’ contributions 7. Sousa‑Pinto B, Anto A, Czarlewski W, Anto JM, Fonseca JA, Bousquet BSP, EH, JAF and JB participated in the study design. BSP, EH and AA partici‑ J. Assessment of the impact of media coverage in coronavirus‑ pated in the data extraction. BSP, EH, AA, WC, AB, BG, GWC, JMA, JAF and JB related Google Trends: infodemiology study. J Med Internet Res. participated in the data analysis, manuscript writing and critical review of the 2020;22(8):e19611. manuscript. All authors read and approved the final manuscript. 8. Higgins TS, Wu AW, Sharma D, Illing EA, Rubel K, Ting JY, et al. Cor‑ relations of online search engine trends with coronavirus disease Funding (COVID‑19) incidence: infodemiology study. JMIR Public Health Surveill. This paper was written by five members of DigitalHealthEurope Grant Agree ‑ 2020;6(2):e19702. ment Number: 826353 Support to a Digital Health and Care Innovation initia‑ 9. Szmuda T, Ali S, Hetzger TV, Rosvall P, Sloniewski P. Are online searches for tive in the context of Digital Single Market strategy, SC1‑HCC‑05‑2018. Publica‑ the novel coronavirus (COVID‑19) related to media or epidemiology? A tion of this article was supported by National Funds through FCT ‑ Fundação cross‑sectional Study. Int J Infect Dis. 2020;97:386–90. para a Ciência e a Tecnologia, I.P., within CINTESIS, R&D Unit (reference 10. Heerfordt C, Heerfordt IM. Has there been an increased interest in smok‑ UIDB/4255/2020). ing cessation during the first months of the COVID ‑19 pandemic? A Google Trends study. Public Health. 2020;183:6–7. Availability of data and materials 11. Jacobson NC, Lekkas D, Price G, Heinz MV, Song M, O’Malley AJ, et al. The datasets used and/or analysed during the current study are available from Flattening the mental health curve: COVID‑19 stay‑at ‑home orders the corresponding author on reasonable request. Sousa‑Pinto  et al. Clin Transl Allergy (2020) 10:47 Page 14 of 14 are associated with alterations in mental health search behavior in the 20. Morais‑Almeida M, Pite H, Aguiar R, Ansotegui I, Bousquet J. Asthma United States. JMIR Ment Health. 2020;7(6):e19347. and the coronavirus disease 2019 pandemic: a literature review. Int Arch 12. Misiak B, Szczesniak D, Koczanowicz L, Rymaszewska J. The COVID‑19 Allergy Immunol. 2020;181:680–8. outbreak and Google searches: is it really the time to worry about global 21. Yardi S, Caldwell PH, Barnes EH, Scott KM. Determining parents’ patterns mental health? Brain Behav Immun. 2020;87:126–7. of behaviour when searching for online information on their child’s 13. Searle T, Al‑Niaimi F, Ali FR. Dermatological insights from Google Trends: health. J Paediatr Child Health. 2018;54(11):1246–54. what does the public think is important during COVID‑19 lockdown? Clin 22. Bäsén A. Smittan slår mot ung och gammal. Expressen. 2020. https :// Exp Dermatol. 2020;45(7):891–921.www.expre ssen.se/kroni korer /anna‑basen /smitt an‑slar‑mot‑ung‑och‑ 14. Sinyor M, Spittal MJ, Niederkrotenthaler T. Changes in suicide and gamma l/. Accessed 29 June 2020. resilience‑related google searches during the early stages of the COVID ‑ 23. Bakan Koca’dan Aytaç Yalman açıklaması. Cumhuriyet. 2020. https :// 19 pandemic. Can J Psychiatry. 2020;65(10):741–3.www.cumhu riyet .com.tr/haber /bakan ‑kocad an‑aytac ‑yalma n‑acikl amasi 15. Riggioni C, Comberiati P, Giovannini M, Agache I, Akdis M, Alves‑ Correia ‑17282 73/. Accessed 29 June 2020. M, et al. A compendium answering 150 questions on COVID‑19 and 24. Astim ve KOAH hastalari için koronavirüs uyarilari. Medikal Teknik. 2020. SARS‑ CoV‑2. Allergy. 2020. https ://doi.org/10.1111/all.14449 .https ://www.medik altek nik.com.tr/astim ‑ve‑koah‑hasta lari‑icin‑koron 16. Gillison DH. Learn more about anxiety with a self‑assessment on Search. aviru s‑uyari lari/. Accessed 29 June 2020. 2020. https ://blog.googl e/techn ology /healt h/anxie ty‑self‑asses sment 25. Yener D. KOAH hastalarına koronavirüs uyarısı. Anadolu Ajansı. 2020. https ‑searc h/. Accessed 29 June 2020.://www.aa.com.tr/tr/koron aviru s/koah‑hasta larin a‑koron aviru s‑uyari 17. Mavragani A, Ochoa G. Google Trends in infodemiology and infoveillance: si‑/17816 15. Accessed 29 June 2020. methodology framework. JMIR Public Health Surveill. 2019;5(2):e13439. 18. Bousquet J, O’Hehir RE, Anto JM, D’Amato G, Mosges R, Hellings PW, et al. Publisher’s Note Assessment of thunderstorm‑induced asthma using Google Trends. J Springer Nature remains neutral with regard to jurisdictional claims in pub‑ Allergy Clin Immunol. 2017;140(3):891–3. lished maps and institutional affiliations. 19. Song X, Xiao J, Deng J, Kang Q, Zhang Y, Xu J. Time series analysis of influenza incidence in Chinese provinces from 2004 to 2011. Medicine. 2016;95(26):e3929. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Clinical and Translational AllergySpringer Journals

Published: Nov 2, 2020

There are no references for this article.