Caroline Favas
Refugee camps are exceptionally vulnerable to the COVID-19 pandemic, as they tend to have poor infrastructure around sanitation and health care. Typical distancing interventions (such as isolation and quarantine) are unlikely to be practical strategies both due to threats to the livelihoods of those living in the camps and the nature of the camp environment itself. However, without any intervention to stop or mitigate a pandemic, in only a few months, there will be a high mortality rate and a high healthcare burden. Caroline Favas of the London School of Hygiene and Tropical Medicine has given guidance on “targeted shielding,” an alternative strategy to those above. Targeted shielding involves selecting high risk and vulnerable individuals to be separated from the rest of the population, limiting contact between people that can result in transmission of the disease (effective contacts). This is done by placing the vulnerable in “green zones”, areas that are only used by these vulnerable individuals that other less vulnerable people may not enter. Green zones can be implemented at varying scales (household, neighbourhood, sector), and have flexibility in terms of disruption to daily life, as well as care and sanitation strategy. At the household level vulnerable members are sectioned into their own areas in their own home, and are provided with water and soap for handwashing. At the neighbourhood and section level, residents of the camps swap homes so that the vulnerable have dedicated accommodation, and it becomes possible to house showering and latrine facilities within the green zone. It also becomes possible to house physically able but low risk residents in these larger green zones to act as carers if needed.
Center for Social Complexity, George Mason University
ABM (Agent-Based Modelling) is a flexible approach for modelling human populations and the effects of displacement and disease over multiple spatial scales. This includes modelling areas with poor sanitation such as the Dadaab refugee camps on the Kenya-Somali border, and looking at the impacts this may have on disease outbreaks such as cholera epidemics. ABMs are also able to model fine details such as effective contact between individuals and to take into account heterogeneities in the population such as demographic differences in daily activities or behavioural responses. This makes it possible to model interactions between humans, disease and environment and to see how a disease might spread through a specific area like a refugee camp, where there might be areas of greater disease intensity, and which groups may act in ways to either cause or become victims of an outbreak. It also gives an idea of the levels of those who have been exposed, infected, recovered or removed from across the target population. There is a need for high quality data in ABMs. This is needed to model the geography of the area or camp correctly and to capture social heterogeneity along demographics. For the Dadaab refugee camps it was possible to get high quality map data, but socio-economic data of the population was lacking and had to be inferred from other sources.
Network Dynamics and Simulation Science Laboratory, Biocomplexity Institute, Virginia Tech
SEIR models and ABMs are able to model differences between slum populations and other urban areas. Slum areas are typified by poor sanitation and health infrastructure as well as high population density, which are issues similar to those faced in refugee camps. The effects of an influenza pandemic share some similarities with the current COVID-19 pandemic, with influenza contributing a significant amount to cases of acute respiratory infections in India (4-12%, approximately 1.5-4.5M/yr). Demographic data of a population can be acquired from social media and online retailers, supplementing other forms of data from surveys or census data. This includes more detailed data such as daily activities and movements which are not readily available from other sources. One of the strengths of an ABM is the ability to create a “synthetic population” which acts as a proxy to a real world population without revealing sensitive individual data. This allows a better evaluation of possible interventions such as isolation, closing school or the use of vaccination. The models found that when slum areas are taken into account, rate and peak of infection increased significantly for the entire population, and time to peak is sooner. In addition, the number of people initially infected and whether those people were from slum or other areas did not affect rate or peak of infection, but again time to peak infection was sooner where there were more people or they were residents of slum regions. This has significant implications for public health policy, showing that the existence of slum and similar regions can have a tremendous impact on the wider population’s health outcomes and therefore cannot be ignored in a concerted response to a pandemic, and that strategies that are trying to change the location and numbers of an initial outbreak, such as tighter border control, may only delay the pandemic rather than lessen its severity.
