Covid-19 and Gridded Population Data: New Models, Data Platforms, and Research Highlights
Written by Hayden Dahmm and Maggie Smith
As we continue to confront the Covid-19 pandemic, one could argue that never before has timely and accurate population data been so critical. Population data is necessary to understanding everything from the spread of the virus and who is impacted, to where vulnerable populations are located and levels of population density for establishing effective social-distancing measures. And gridded population data are playing an important role in helping researchers access this type of information. Gridded population maps distribute data using grid cells, combining census results with additional information, such as geospatial data from satellites to ensure more frequent and granular estimates, which are vital during a constantly evolving epidemic. In this blog, we explore how gridded population data are being used by researchers to inform the Covid-19 response.
Data Platforms:
Many of the leading models and indices of COVID-19 spread are using WorldPop data, which provide high-resolution age and sex disaggregated population figures. These include:
The UN Population Fund (UNFPA) - Has established a COVID-19 Population Vulnerability Dashboard that uses data from the World Health Organization and WorldPop to calculate different aspects of population vulnerability, such as population density and health sector readiness. With these calculations, it then visualizes risk factors at the national and subnational levels.
The Institute for Health Metrics and Evaluation (IHME) - Has built a model that combines data on population density, mobility, mask use, COVID-19 testing rates, seasonal pneumonia mortality, and other variables to project the spread of COVID infections, deaths, and hospital resource use globally through November 2020.
The Imperial College’s MRC Centre for Global Infectious Disease Analysis - Developed its COVID-19 model by simulating a spatially and socially-structured population. Individual people in the model are assigned demographic attributes and places where they would interact with others, including at home or at work. The researchers use this simulated population to examine how different policies and social behaviors will impact COVID-19 transmission dynamics.
The Surgo Foundation - Has created the Africa COVID-19 Community Vulnerability Index that looks at 48 African countries and 751 subnational regions and ranks them in terms of health and economic risk. In addition to population density, it considers factors such as socioeconomic status, housing and transportation availability, exposure to other epidemiological risks, and access to health facilities to construct a composite indicator.
Recent Studies:
Country Preparedness and Air Travel
In addition to supporting various data platforms, gridded population data have been used in a variety of studies to assess COVID-19 preparedness, risk factors, and spread. For example, earlier in the course of the pandemic, a study using Landscan predicted that Sweden would reach its peak number of COVID-19 cases in early May, demonstrating that additional preparations would be required. Although the study’s results suggested that Sweden’s mitigation policies were helping to prevent up to 15% of COVID-related deaths, it showed that intensive care bed availability at the time would have to increase at least ten-fold to meet the demand. Another study assessed the risk of spreading the virus through air travel by using air travel records, data on confirmed COVID cases, and WorldPop data to measure population density in areas around airports. The researchers concluded that China, the Middle East, Europe, and the U.S. had the highest risks of spreading the virus via air travel, and they called for strict air travel limits to manage this risk. An additional analysis by the World Bank used WorldPop data on population density in urban areas combined with measures of total livable floor space and sanitation services to identify potential COVID-19 hotspots, highlighting the compounding risks of Covid-19 in slum communities.
Additional Factors Influencing Covid-19 Spread
Research has also looked retrospectively at what factors have both contributed to and helped to preclude the spread of the virus, with much of the research focused on the U.S. For example, one recent study combined Landscan data with the locations of COVID cases during March and May to estimate the velocity of the pandemic in the U.S., calculating that it had spread more than 3 km per day. As part of the study, the researchers also examined mobile phone data and shelter-in-place announcement times to estimate the impact of containment measures, finding that the government response time to outbreaks and individual compliance with social distancing requirements were valuable for mitigating the spread of the virus. Another study examined 28 variables and employed machine learning to identify which factors had been most important to determining the spread of COVID-19 across the 50 U.S. states throughout the month of May. The results suggest that population density (measured with WorldPop data) and mobility (measured with Google’s COVID-19 Community Mobility Reports) were critical predictors, which demonstrates the need for lockdown measures to be tailored to given locations. Other work has also shown how different age groups have influenced COVID-19’s spread. For instance, researchers used Worldpop data to model the frequency and main drivers of super-spreading events in the U.S. state of Georgia, finding that 2% of infected individuals (primarily children and younger adults) were causing 20% of new infections.
Environmental Factors and Covid-19
There has been speculation about the role seasonal conditions and other environmental factors might play in the virus transmission, and combining gridded population data with environmental data has enabled researchers to explore these relationships. For example, one study combined Gridded Population of the World (GPW) with global climate databases to consider the impact of climate variables – such as temperature, precipitation and wind speed - on COVID-19 spread. By controlling for population density, quality of human life, and travel restrictions, the researchers concluded that a lower COVID-19 growth rate is associated with high precipitation seasonality and a lower local rate of climate change. However, travel restrictions were found to have a more significant impact on COVID-19 spread than any specific climate characteristics. Another study combined Landscan with a spatial dataset of COVID-19 cases and compared it with local environmental conditions, determining that increased exposure to UV light had a statistically significant effect on reducing spread of the virus, although temperature and humidity were not found to have a significant relationship.
Additional research has also modeled the intersecting effects of environmental conditions and human activity on the spread of the virus. For example, Gianpaolo Coro created a localized risk index for COVID-19 spread by calculating population density from GPW and pollution levels (represented by satellite-derived CO2 emissions) along with geophysical variables, including surface air temperature, precipitation, and elevation. The model was trained on data from Italy, but was able to predict most of the high-infection areas around the world, with the results indicating that temperature and precipitation are significant to defining regions that may experience high infection rates, although local pollution was actually the most influential parameter in increasing the possible number of infections.
Beyond the immediate public health consequences, COVID-19 is impacting populations in other ways. Using WorldPop demographic data to identify the population density of people over the age of 70, GPW to measure overall population density, and other variables for hazard, exposure, vulnerability, and resilience, a study examined the overall economic risk of Covid-19 by country, showing that the risk was particularly high in Sub-Saharan Africa and South East Asia. Yet, the reduction of economic activity globally has also helped to reduce air pollution. A study of 27 countries, mostly concentrated in Europe, quantified the reduction in air pollution resulting from COVID-19 economic disruptions and correlated these declines with potential public health benefits. Using Global Human Settlement Layer (GHSL) and GPW data multiplied with baseline mortality and asthma incidence rates to calculate daily health burdens for countries’ population, air pollution was shown to have decreased by 20% as a result of lockdown measures. This is estimated to have helped avoid 7,400 premature deaths and 6,600 cases of pediatric asthma.
As the global community continues to combat the health, social, and economic threats of the pandemic, increased awareness and understanding of population dynamics is necessary to make sure no one is left behind in our collective response. Moreover, additional research on the role population dynamics has in determining infections will help develop informed policies and enhance response efforts, and groups like the POPGRID Data Collaborative are working to further this research. Gridded population data continue to prove to be a critical tool for researchers and policymakers seeking to get a better handle on Covid-19.
For more information on the importance of gridded population data and an analysis of major global gridded datasets, read TReNDS’ recent report, Leaving No One Off the Map: A Guide For Gridded Population Data for Sustainable Development.