Real-world data on daily confirmed COVID-19 cases and deaths were retrieved on July 9, 2020, from the WHO’s COVID-19 dashboard from January 1, 2020, until July 7, 2020 [1]. Worldwide data, as well as individual country-level data for the 50 countries with the highest cumulative confirmed COVID-19 case numbers (as of July 7, 2020), were also retrieved. Key dates of the COVID-19 pandemic were retrieved from the WHO’s COVID-19 timeline and news reports of regional COVID-19–related events [20,21].

GT provides data on volumes and patterns in online search behaviors of internet users [15]. It tracks keyword search queries that users enter into the Google search engine and presents information on the search query according to the specified time period and geographical location [22]. The search volume results are normalized and presented as an RSV index, wherein each data point is divided by the total searches performed in a specified geography within a given time range to provide relative comparisons [22]. The resulting output ranges from 0 to 100, with 100 indicating the maximum search interest in the selected time period and location. To comprehensively capture trends in online search behavior and infectious burden over an extended period, daily worldwide and country-specific GT data were retrieved over the first 6 months of the outbreak, from January 1 to July 7, 2020.

In addition to GT, the Baidu Index and the Yandex Wordstat Keyword Statistics Service were used to retrieve data on search queries in China and Russia, respectively. Baidu and Yandex are the predominant search engines used in China and Russia, respectively [14,23]. To facilitate direct visualization and comparisons with the RSV index obtained from GT, data from the Baidu Index and Yandex were similarly scaled to range from 0 to 100 [18]. In this work, total-RSV denotes the cumulative RSVs over the entire evaluation period, while average RSV represents an RSV value that was averaged over a specified period (eg, pre– or post–COVID-19 period; further descriptions are provided below).

For the GT analysis, we followed the detailed methodology described by Mavragani et al [10] for our keyword selection. First, different permutations of search terms and topics related to “telehealth” and “telemedicine” were searched to understand overarching trends in worldwide interest and to optimize keyword search combinations. We then conducted worldwide and country-specific GT searches using a baseline combination of English, Spanish, Russian, and French translations (chosen from the list of most commonly spoken languages worldwide) [24], in addition to translations in the native or official language of that particular country. Mandarin Chinese is the second most spoken language in the world [24] but was not included in the baseline search combination since the majority of native Mandarin Chinese speakers reside in China and do not use Google as their main internet search tool. For the Baidu and Yandex search indexes, a combination of keywords in Mandarin Chinese (both traditional and simplified) and Russian, respectively, were used. The detailed keyword search strategy can be found in Table S1 in Multimedia Appendix 1.

Data regarding the key dates of major public health responses such as lockdowns for each country were obtained from internet sources and news reports [25-27]. In addition, ICT data from the ICT adoption pillar of the Global Competitiveness Index 4.0 framework were obtained from the World Economic Forum Global Competitiveness Report 2019 and used to compare ICT level across countries [28]. The extracted ICT adoption index (ICT index) scores range from 0 to 100 (highest), wherein a higher score represents greater levels of networked infrastructure and higher regional usage and access to such infrastructure. Additional country-specific indices including GDP (gross domestic product) per capita ($US), literacy rates, and World Bank regional income groups were obtained from the World Bank DataBank [29].

As a proxy measure of the existing telehealth capacity of the respective countries, the Crunchbase (CB) [30] and CB Insights (CBI) [31] business analytic platforms were used to search for the prevalence of prominent telehealth providers within each of the 50 countries, and was used to define a CB/CBI score.

First, to provide a more accurate observation of underlying trends and eliminate short-term fluctuations in data, the time trends for telehealth-related RSVs were smoothened by 7-day rolling intervals [32]. For countries with lower total-RSVs (<5), 14-day interval smoothing was used instead, as these countries are more susceptible to daily fluctuations (thus more “noise” in the trend data). Having a longer smoothing interval helps to minimize errors in estimations caused by these fluctuations over a short period. The RSVs were then plotted against daily COVID-19 confirmed cases and deaths (two separate Y axes), both worldwide and for each of the 50 countries.

Mean (SD) and median (IQR) values were used to provide summary statistics for key country parameters including total-RSVs, COVID-19 cases and deaths, and ICT indexes. The Pearson and Spearman correlation tests (when applicable) were used to further investigate associations between key parameters. The kurtosis/skewness and Shapiro-Wilk tests were used to determine normality of data distribution. P values of <.05 were considered as statistically significant.

The lockdown periods for each country were incorporated along with two key dates for reference—January 23, 2020, when China first imposed a lockdown in Hubei Province, and March 11, 2020, when the WHO declared COVID-19 a global pandemic. When evaluating changes in average RSV levels in the pre– and post–COVID-19 periods (ie, RSVs averaged over the period before and after COVID-19, respectively), we defined these periods based on the landmark date of March 11, 2020 (the WHO’s COVID-19 pandemic declaration), except for China, where the pre–COVID-19 period was defined as before January 23, 2020. The ratios of the average pre– and post–COVID-19 RSV levels for each country were then calculated.

Lastly, bubble plots were used to illustrate the relationships between the total-RSVs of individual countries and various telehealth-related national indicators (GDP per capita [$US], literacy rates, the ICT index, and the CB/CBI score). Countries were grouped and color-coded according to World Bank regional classifications. All analyses and visualizations were conducted using Python (Python Software Foundation, version 3.7.4).

In this study, we used data from GT, the Baidu Index, and Yandex Keyword Statistics to evaluate trends in telehealth demand during the first 6 months of the COVID-19 pandemic. To our knowledge, this is the first study to apply an infodemiological approach to investigate the potential public demand for telehealth in the 50 countries most affected by COVID-19. Our study further unraveled trends of demands alongside key COVID-19 events and varying levels of ICT infrastructure. Our findings suggest a general trend of increased demand for telehealth services across the evaluated countries during the COVID-19 pandemic. This trend was largely sustained beyond the initial wave, country lockdown periods, and subsequent reopenings. Our results suggest an ongoing and possible increased interest in telehealth services in the future as we enter a post–COVID-19 new normal phase. We also observed that current ICT infrastructure in several developing countries may lag behind this surging demand for telehealth. Our findings collectively indicate a pressing need to scale up telehealth capabalities in response to this growth in telehealth demand.

Our results demonstrate increased RSVs in most of the 50 evaluated countries. Among them, Canada and the United States had the highest total-RSVs and displayed a large spike in interest for telehealth services compared to most other countries. It is noteworthy that this pattern was also found in Australia (though not ranked within the top 50 countries), which had similarly high total-RSV values (RSV=98.9; data not shown in tables). Most other countries had less well-defined but observable increases in RSVs.

Interestingly, in many countries, RSV trends over time did not closely follow local COVID-19 case and death counts. However, RSVs often increased in early March, suggesting that the global announcement by the WHO on March 11, 2020, possibly had a great impact on the public’s change in behavior to seek for telehealth options. Other studies investigating COVID-19–related search trends have reported that increases in RSVs often preceeded local COVID-19 cases and deaths [17,33]. It has been suggested that RSV trends tend to change in response to particular “index events” [18]. For instance, Rovetta et al [33] reported that significant increases in COVID-19–related RSVs were only observed after the WHO’s COVID-19 pandemic announcement and the imposition of strict social distancing rules by governments.

Our study also found fair correlations between total-RSVs and a country’s COVID-19 cases and deaths. We posit two possible explanations for this. First, larger numbers of COVID-19 cases and deaths could result in increased risk perception in the general population, thus driving greater interest in telehealth to minimize risk exposures. This is in line with other studies that have also reported that changes in search volumes toward other risk-minimizing activities, such as social distancing, hand washing, and face mask wearing, corresponded with increases in COVID-19 cases [19,33]. Another potential explanation would be that the increase in telehealth-related RSVs reflected actual needs of affected patients. The timing of the RSV spikes for each country often preceded the actual spike in COVID-19 cases, suggesting that changes in search activity were probably more likely due to changes in public perception as the pandemic situation evolved. Nevertheless, further real-world investigations in the form of custom-designed questionnaires would be needed to further elucidate this aspect.

The bubble plots presented in this report further investigated the relationships between total-RSVs and various telehealth-related indices (ICT index, GDP per capita, literacy rates). These indices were evaluated as proxies for the capacity of a country to adopt and operationalize new telehealth services [34-36]. The clustering patterns observed in Figure 4 and Figures S2-S4 in Multimedia Appendix 1 may enable the classification of regions or countries based on their relative demands and capacity for telehealth services. For example, regions with the potential for rapid growth and adoption of telehealth are those with higher literacy levels, better ICT infrastructure, and higher RSVs (Figure S4, Multimedia Appendix 1). These countries include Argentina, Chile, Qatar, the United Arab Emirates, Saudi Arabia, and Singapore [37]. On the other hand, countries such as Colombia, Peru, Ecuador, Bangladesh, Pakistan, and the Philippines demonstrated growing interest in telehealth (moderate to high RSVs) but are limited by existing ICT capability (low to moderate ICT index).

Despite high ICT indexes and strong existing telehealth capacity (high CB/CBI scores; Figure S4, Multimedia Appendix 1) in a majority of European countries [38], the RSVs of these countries were relatively low compared to other developed western countries such as Canada and the United States. This observation may be explained by the availability and easy accessibility of existing telehealth services, which may also impact the population’s information-seeking behavior. For instance, in European countries that are at a more advanced stage of telehealth adoption, telehealth awareness and literacy may already be present, and patients may be directly seeking telehealth services from providers instead of conducting online queries in search engines [38]. Other factors that may also influence search volumes include different models of health care systems. For example, patients in the United Kingdom may contact their general practitioner or seek online consultations readily and directly through the NHS App [39]. Public health communications play a role as well. Sweden, for instance, contrarily downplayed the significance of the pandemic [40], eschewing lockdown measures, which may also explain why Sweden had the lowest RSVs among the evaluated European countries [40,41].

Our study offers useful insights into the short- and medium-term trends in telehealth demand in response to the COVID-19 pandemic. The trend of sustained increase observed in the Baidu Index RSV in China provides preliminary indications that telehealth demand will likely remain higher than in the pre–COVID-19 era. This trend is especially likely given that China can be considered to have entered its post–COVID-19 recovery phase, with minimal new cases reported over the last 3 months of our study period. On the other hand, given the resurgence of second or third waves of COVID-19 in several countries in recent months [1,42], it is foreseeable that telehealth demand may surge yet again or remain higher than the pre–COVID-19 period. Nonetheless, to better ascertain the long-term impact of COVID-19 on telehealth interest, further evaluation over a longer period is required.

The key strengths of our study include our unique infoveillance approach to evaluate the public demand for telehealth services, capturing real-time responses to key COVID-19 pandemic events [11,16,43]. Second, our study provided extensive coverage of the 50 most affected countries worldwide, and evaluated these countries over a long duration (6 months), thus providing more concrete insights on trends. Third, we also included the additional use of the Baidu Index and Yandex Keyword Statistics to further investigate RSVs in China and Russia. The high degree of correlation between GT and Yandex Keyword Statistics for RSVs in Russia (r=0.875, P<.001) further confers a degree of reliability to our results.

This study also has a few limitations. First, it should be acknowledged that infodemiological approaches can only serve as a proxy for estimating the true demand for telehealth services. Second, as suggested earlier, although search engines provide a good catchment of overall interest, there may be alternative channels for the public to seek health care–related information, including their general practitioners, insurance services, or directly from telehealth service providers. In addition, it should also be acknowledgd that it is not possible to include all terms utilized by internet users to search for information on telehealth. Third, not all countries use Google as their primary search engine. Hence, using GT as a main proxy for overall demand when comparing across countries may be subjected to bias for some countries. Fourth, overall RSV trends may be confounded by different sampling profiles between countries, wherein factors including education and internet access may skew the representativeness of the RSV samples for each country. Additionally, infodemiology platforms such as GT and the Baidu Index present data as normalized RSVs rather than as absolute search counts, thus limiting direct comparisons of RSV data extracted from different sources of search engines [10]. Fifth, to reduce subjectivity, across all included countries (except for China), we standardized the definition of the post–COVID-19 period based on the date when the WHO declared COVID-19 as a global pandemic. Although it might have been ideal to individually quantify each country’s pre– and post–COVID-19 periods, it was difficult to do so, considering each country had a different community transmission trajectory. Furthermore, changes in search volumes were often in response to not just local events (lockdowns, local increases in COVID-19 cases, etc) but also global events such as the WHO’s pandemic announcement. Lastly, to better ascertain the long-term impact of COVID-19 on telehealth interest, further evaluation over a longer period is required.

Telehealth is a major health technology solution that has gained further traction during the COVID-19 pandemic. We identified increased demand for telehealth services across the 50 countries most affected by COVID-19. We also found indications that several developing countries may still have suboptimal ICT infrastructure to cope with this surge in telehealth demand. These findings underscore a pressing need for policy makers and health care providers to scale up telehealth infrastructure and capability, during and beyond COVID-19.