Fixed internet quality in Colombia, Ecuador, and El Salvador (2019/2020). Average fixed internet download speed (expressed in Mbps) computed from Ookla® Speedtest Intelligence® data for different departments of Colombia, Ecuador, and El Salvador

The adoption of non-pharmaceutical interventions and the role of digital infrastructure during the COVID-19 pandemic in Colombia, Ecuador, and El Salvador

Nicolò Gozzi, Niccolò Comini & Nicola Perra for EPJ Data Science


Adherence to the non-pharmaceutical interventions (NPIs) put in place to mitigate the spreading of infectious diseases is a multifaceted problem. Several factors, including socio-demographic and socio-economic attributes, can influence the perceived susceptibility and risk which are known to affect behavior. Furthermore, the adoption of NPIs is dependent upon the barriers, real or perceived, associated with their implementation. Here, we study the determinants of NPIs adherence during the first wave of the COVID-19 Pandemic in Colombia, Ecuador, and El Salvador. Analyses are performed at the level of municipalities and include socio-economic, socio-demographic, and epidemiological indicators. Furthermore, by leveraging a unique dataset comprising tens of millions of internet Speedtest® measurements from Ookla®, we investigate the quality of the digital infrastructure as a possible barrier to adoption. We use mobility changes provided by Meta as a proxy of adherence to NPIs and find a significant correlation between mobility drops and digital infrastructure quality. The relationship remains significant after controlling for several factors. This finding suggests that municipalities with better internet connectivity were able to afford higher mobility reductions. We also find that mobility reductions were more pronounced in larger, denser, and wealthier municipalities.


Stay-at-home mandates, travel bans, face masks, curfews, remote working, and closure of non-essential shops are just some examples of the non-pharmaceutical interventions (NPIs) implemented worldwide to contain the spread of SARS-CoV-2 [15]. Though extremely successful public health measures, NPIs are associated with high socio-economic costs, might bring economic activities to a halt, and disrupt social life [67].

The literature on the subject suggests that compliance with such measures is a multi-faced problem driven by individual and societal factors. Socio-demographic (e.g., age, gender, educational attainment, country of residence, population density, age pyramid), and epidemiological (e.g., number of cases, deaths, and vaccination rates) indicators play an important role in adherence. They modulate the perceived risk, severity, and susceptibility to the threat and can ultimately affect individual behaviors [1810]. Adherence to NPIs strongly correlates also with socio-economic determinants [1114]. From low to high-income countries, disadvantaged communities are affected by structural inequities and as result had markedly fewer opportunities to adopt NPIs during the acute phases of the Pandemic [131521]. Indeed, while restrictive measures have disrupted the lives of everyone, the challenges and barriers to adoption faced in implementing them are extremely different across socio-economic strata. Informal jobs and several types of occupations, for example, made it extremely hard to implement many forms of NPIs [12223].

Widespread adoption of mobile devices and of the Internet allowed to continue remotely some economic, educational, and social activities [2425]. These new tools, while opening unprecedented opportunities for many, constituted new barriers for others. Limited and unequal access to a reliable digital infrastructures can influence teleworking, e-commerce adoption, distance learning, use of telehealth platforms, remote access to financial services, and more in general the ability to carry out activities from home thus limiting travels and mobility [2531]. Arguably, individuals with better internet connections faced far less obstacles while transitioning to online activities with respect to those with poorer digital connectivity which, whenever possible, may have continued to attend essential activities in person.

However, these factors have not been yet extensively explored via a quantitative data-driven approach. Furthermore, the current literature on the subject, apart from a report focused on social distancing in the US during the early phases of 2020 and its relation with access to high-speed internet [32], is mainly focused on specific contexts such as education and telemedicine probed via surveys [3338].

Here, we tackle this limitation investigating a wide range of variables possibly affecting the adoption of NPIs in the first COVID-19 wave in one lower-middle-income country (El Salvador) and two upper middle-income countries (Colombia and Ecuador) in Latin America. In March 2020, all three countries implemented a series of measures to control the spread of SARS-CoV-2. According to data from the Oxford COVID-19 Government Response Tracker [39], Ecuador was the first to introduce measures restricting internal movement across regions and cities on March 17, 2020, followed by El Salvador on March 18 and Colombia on March 25. At around the same time, the three countries implemented also stay-at-home orders with exceptions for essential trips, closed schools and non-essential/specific sectors workplaces. We explore socio-demographic (i.e., population size, population density, the fraction of the population above 60), socio-economic (i.e., wealth index, GDP per capita), and epidemiological (i.e., number of reported COVID-19 cases) indicators that might affect risk, severity, and susceptibility perception as well as impose barriers to the adoption of NPIs. Additionally, we investigate the quality of the digital infrastructure as a form of barrier that might affect NPIs adherence. To this end, we leverage a unique dataset containing tens of millions of geolocalized internet Speedtest® results from Ookla® [40] that we use as a proxy for quality of internet connection. Measuring the quality of internet connections using Internet Speedtests® or similar tools has become a widely used method, utilized not only by researchers but also by private companies, governments, and other authorities [4143]. This approach is particularly valuable in identifying disparities in access to reliable and high-quality internet connections across communities [424445].

We characterize NPIs adherence using a publicly available dataset from Meta’s Data for Good program that provides high spatio-temporal resolution information about mobility changes [46]. Aggregated mobility indicators, obtained from digital crumbs we leave while using or simply carrying mobile phones in our pockets, together with ad-hoc surveys, have been used to characterize adherence with NPIs at the population level [1]. The scale of the emergency and the strictness of the measures disrupted such a broad range of activities and behaviors that a simple comparison of aggregated mobility volumes (obtained from mobile phones) with respect to a pre-pandemic baseline shows marked differences and allows to quantify adoption of many types of NPIs.

We focus our analyses at the municipal level on three Latin American countries: Colombia, Ecuador, and El Salvador. Latin America is often regarded as the most unequal region in the world [4748]. The profound disparities that afflict the region are also reflected in high rates of infection and deaths observed during the Pandemic [49] as well as in the access, use, and quality of digital tools [5051]. Analyses conducted within these countries show large spatial heterogeneities and urban-rural divides in many indicators including digital literacy, skills, and access to broadband [25]. Unfortunately, these points, together with the limited number of studies in this region of the world, make Latin American countries good case studies to investigate and expand our knowledge about NPIs adoption.

We find that, during 2020, NPIs significantly affected mobility in the three countries, causing a maximum drop in movements, from pre-pandemic baselines, of 53% in Colombia and 64% in Ecuador and El Salvador. Using different statistical analyses, we first show that municipalities with access to a better internet connectivity – measured with Speedtest® proxy data – also feature more consistent reductions in mobility. Such association is preserved when controlling for possible confounders. We estimate that, for every 10 Megabits per second (Mbps) increase in average fixed download speed, movement reduction increases by another 13%, 4%, 19% in, respectively, Colombia, Ecuador, and El Salvador. We then adopt a regression approach that, besides digital infrastructure quality, accounts also for many other factors possibly influencing mobility reductions in each municipality. We find that digital infrastructure quality is still a significant predictor of NPIs adoption. Population size, density, and socio-economic status are also associated with higher mobility reductions. As a supplementary analysis, we include also regressors on employment and labor force structure (for Colombia only, due to limited data availability). Even in this case, we find that internet quality remains a significant predictor of NPIs adoption.

Notwithstanding clear progress made, much is still unknown about the adherence to NPIs especially when it comes to the effects of different socio-economic barriers, such as the access to a reliable and high-quality internet connection. We aim to extend the literature on the subject offering a quantitative investigation about the role of digital infrastructure quality, along with many other variables, in NPIs compliance during the first wave of COVID-19 Pandemic in three Latin American countries. Understanding such phenomenon is key to informing policies aimed at increasing the resilience to external shocks as well as equity in communities, cities, and countries. Additionally, our research highlights the possibility of utilizing non-conventional data sources, such as Speedtest®, to examine the compliance to NPIs in countries with limited information availability.


2.1 Proxies of NPIs adoption
We characterize NPIs adoption by quantifying how aggregated mobility changed during 2020 in the three countries…

Continue reading at :