What epidemic modelling can teach us about the spread of disease in refugee settlements

6 min read
Eighteen-year-old Rohingya refugee and Burmese language instructor Nur Ayna teaches a young student at Kutupalong camp in Cox’s Bazar, Bangladesh. Photo: UNHCR
Share on facebook
Share on twitter
Share on linkedin
Share on email

The spread of COVID-19 has presented many challenges to healthcare systems worldwide, often affecting the world’s most vulnerable populations the most. In settlements for refugees and internally-displaced persons (IDPs), which suffer from overcrowding, insufficient sanitation, and particularly rapid disease spread, the pandemic presents a significant threat—as will any future epidemics or other health crises.

In a paper which came out today, scientists and public health officials from UN Global Pulse, UNHCR, WHO, Durham University, London School of Hygiene and Tropical Medicine, New York University, and MIT-IBM Watson AI Lab describe an agent-based model that simulates the potential effects of various interventions against the spread of COVID-19 in Cox’s Bazar refugee settlement in Bangladesh. The model is based on the JUNE epidemiological modelling framework, which is informed by data on geography, demographics, comorbidities, physical infrastructure and other parameters obtained from real-world observations and previous literature. The team looked at various scenarios that could support more effective and efficient decision-making and advocacy, which we summarise in this article. 

Scenario 1: Defining the Baseline

What epidemic modelling can teach us about the spread of disease in refugee settlements 1
This graph represents the simulated spread of new COVID-19 cases over time.

Before assessing the effectiveness of different possible interventions inside the refugee camp, the team developed a baseline model that shows how COVID-19 would spread among the settlement’s population. Because exact figures would be difficult to predict given incomplete data on COVID-19 and other variables, the team focused on assessing relative changes to the baseline, rather than focusing on an exact number of cases.

The baseline scenario makes several assumptions in order to create a “standard” for what the disease spread could look like with no additional interventions. The model starts out with 88 infected individuals across the settlement, as approximated from the WHO Early Warning, Alert and Response System; mild cases are assumed to self-isolate about 30 per cent of the time, given that many individuals must leave their shelters for basic needs, while more severe cases are assumed to stay inside the shelters due to the severity of their symptoms.

Scenario 2: Home-Care vs. Hospitalization

What epidemic modelling can teach us about the spread of disease in refugee settlements 2
This graph shows the difference between home-based care vs hospitalization

The first type of intervention that the team modelled investigated what the difference would be between treating all positive COVID-19 cases in treatment centers versus encouraging home isolation for cases that don’t require hospitalisation. Therefore, two scenarios were constructed. In the first, patients presenting mild to more severe symptoms were treated at home (the baseline model) while only people requiring hospitalization were treated in a treatment center. In the second, all symptomatic patients, regardless of the severity of their symptoms, sought care in treatment centers.

After running both simulations multiple times, it was determined that the two scenarios didn’t significantly differ. Therefore, encouraging home-based care would not negatively affect the spread of COVID-19, and in addition it could help alleviate pressure on treatment centers, which  might otherwise become overwhelmed. 

Scenario 3: Mask or No Mask

What epidemic modelling can teach us about the spread of disease in refugee settlements 3
This model shows the reduction in the number of COVID-19 cases over time based on compliance with mask wearing.

The second intervention that the team modelled looked at the effects of mask-wearing on the spread of COVID-19 inside the Cox’s Bazar settlement. For example, in a scenario where half of the population in the settlement was wearing cotton-type masks, the model showed about 20 percent fewer infections overall,  and a 50 percent reduction in the peak number of daily infections.

Nearly every scenario the team modelled showed that wearing a mask reduced the spread of the disease and the results allowed policy makers to understand the tradeoffs between investing in higher efficacy masks versus increasing compliance with mask wearing. Ultimately, encouraging settlement residents to wear masks could result in a significantly lower number of COVID-19 cases spread across a longer period of time, which would allow for more efficient treatment of patients. 

Scenario 4: Re-opening Learning Centers

What epidemic modelling can teach us about the spread of disease in refugee settlements 4
This graph shows the differences between re-opening learning centers with no additonal measures against COVID-19, versus reduced attendance.

This final intervention explored the different possibilities for reopening learning centers, which are vital to the community and which were closed in March 2020 in order to mitigate the initial spread of the virus. Keeping them closed could result in significant educational delays for children in the settlement, as well as additional concerns of children contracting and spreading COVID-19 while not in classrooms.

The simulation showed that reopening learning centers in their pre-COVID state—with no additional health and safety measures taken—could lead to an increase in the speed of the spread of the disease. Therefore, the team simulated the effects of various mitigation strategies that might be used to reduce the risks of reopening these centers. In particular, results showed that reducing attendance, encouraging physical distancing, as well as mask wearing by the children and teachers in the classroom could effectively lower the risk of remaining indoors for extended lengths of time, thereby having the potential to allow for the safe reopening of learning centers.

Scaling the Experiment

The inter-disciplinary team working on this project has designed this specific modelling technique in the hope that it will be generalisable. This work contributes more widely to the Secretary-General’s Data Strategy, Strategy on New Technologies, and Innovation Agenda which emphasize the need for data-driven collaborations inside and ouside the UN to enable better decisions and stronger support for the people we serve.

UN Global Pulse is undergoing conversations with other teams around the world who have also focused on modelling disease spread in similar geographies. The hope is that from our collective knowledge and understanding we can better assess the lessons learned and ensure the best provision for these vulnerable populations during and beyond COVID-19.

Download the paper

Did you enjoy this blog post? Share it with your networks!

Share on facebook
Share on twitter
Share on linkedin
Share on email

News, thoughts and ideas about big data and AI, data privacy and ethics from across the Pulse Lab Network. Read more on the blog.

Pulse Lab Jakarta

Expanding the Definition of Scaling Up

The “pilot to scale-up model” is a widely used approach when introducing social interventions or new policies. The premise of the model is very logical

Scroll to Top