Improving mobile COVID vaccination efforts through GIS mapping

Anaïs Ladoy, a PhD student at EPFL’s Laboratory of Geographic Information Systems (LASIG), was working on chronic disease prevention when her research took a completely unexpected turn in March 2021. That’s when someone from Vaud Canton’s General Health Office (DGS) asked if she and her supervisor could help the DGS plan the Canton’s mobile vaccination campaign.

“As part of my thesis, I had already met with members of the DGS,” says Ladoy. “I had already visited their headquarters, and they were familiar with my methodological approach.” Her thesis, which she began in January 2019, involved mapping chronic diseases and health problems – such as diabetes, heart disease and obesity – across Vaud Canton with the goal of improving the targeting of prevention campaigns.

Phase one
Ladoy and her supervisor Stéphane Joost, a senior lecturer and researcher, agreed to help. The pair had only ten days to develop a methodology and identify locations where the temporary vaccination centers should be deployed. Ladoy set aside her current research project and threw herself wholeheartedly into the task. A working group was set up between the two EPFL researchers and Louis-Henri Delarageaz, commander of the Vaud’s Office for Civil Protection, and Jérôme Mouton, head of the DGS’s COVID-19 pandemic response team.

“We needed to create an effective, flexible and easy-to-understand approach that could be up and running quickly,” says Ladoy. “It had to be ready in time and approved by the cantonal government committee in charge of vaccination.”

Vulnerability index
Ladoy compiled hundreds of thousands of anonymized records from the Swiss Federal Statistical Office and Vaud Canton. Her goal was to identify areas, by postal code, where vulnerable groups – such as elderly people residing alone, low-income households and non-French speakers – lived but were not in the vicinity of a permanent vaccination center. “I defined variables for each area, such as accessibility to a permanent vaccination center, median income, and residents hit hardest by COVID,” Ladoy says.

Based on these determinants, the civil protection office targeted sites and communities where mobile vaccination centers should be set up.

Ladoy then assigned a score to each variable and weighted them based on importance, with input from an immunization expert and taking into account the scientific literature and the Canton’s priorities. This culminated in the creation of a vulnerability index. “Based on these determinants, the civil protection office targeted sites and communities where mobile vaccination centers should be set up.” During this first phase, mobile facilities capable of administering 400 to 600 jabs per day were set up, by and large in community centers, and more than 14,500 people in 31 municipalities were vaccinated.

Phase two
In early June, buoyed by this initial success, the Canton asked the researchers to fine-tune their analyses down to hectare level, a more precise scale that would nevertheless not jeopardize data privacy. Ladoy used up-to-date vaccination statistics to determine which areas were lagging behind the cantonal average. By mid-June, 40% of the population had received at least one dose (about 340,000 people in all).

“This gap with the rest of the Canton has various causes,” says Ladoy. “The groups in question are living under difficult conditions that leave them isolated. They lack information and, for this reason, they often don’t make online appointments or travel to vaccination centers. But when a vaccination facility is accessible locally and information is provided in the appropriate language, people go ahead and get vaccinated. These groups are not vaccine-averse, and it’s well worth reaching out to them.”

The communication office of the Canton’s Department of Health and Social Action (DSAS), which was part of the strategic planning efforts, produced vaccination information in nine languages as well as in simplified French. Flyers were distributed to some 30 social assistance associations, as well as at hairdressers, grocery stores and places where undocumented workers and the homeless gather, and outside businesses in the targeted areas.

Under this new outreach campaign, which was launched in early July, nearly 2,000 people were vaccinated in mobile facilities that operated for 20 days at 36 different sites, including open-air markets, shopping centers and specific neighborhoods that were identified using LASIG’s recommendations.

The lab’s scientific input was critical in this planning stage of the cantonal vaccination campaign.

“Teaming up with LASIG allowed us to carry out a thorough, substantiated and fact-based assessment of the situation. This in turn clarified our thinking and allowed us to target our work more effectively on paper and then in the field. The lab’s scientific input was critical in this planning stage of the cantonal vaccination campaign,” says Louis-Henri Delarageaz.

A partnership for the future
In the end, Ladoy will devote part of her PhD thesis to COVID, and she has already published a paper on detecting COVID clusters in Vaud Canton during the first wave of the pandemic (link below). However, the initial focus of her research has not changed, and her observations quickly pointed to an interaction between chronic diseases and COVID-19. She found that people who suffer from a chronic illness and live in difficult circumstances are more likely to be severely affected by the virus. Experts refer to this as a syndemic.

“I was extremely fortunate that my doctorate involved such a high level of applied research. It also allowed me to show that targeted and effective measures can be implemented by factoring in the geographical aspects of public health,” says Ladoy. “We hope to take part in more collaborative opportunities like this in the future.”

Caption of the map below

Distance to the nearest vaccination center by road network for each post code area (ZIP) in the Vaud Canton. The darker the red, the less accessibility to the nearest vaccination center. Accessibility is one of the components of the vulnerability index used. © EPFL / LASIG A. Ladoy, S. Joost