It was one of the biggest questions of 2020: when will I be vaccinated from Covid? We tried to answer it, building a tool to tell our readers their vaccine priority based on their age and situation. As well as telling our readers when they might expect to get the jab, we illustrated how many people they might be behind in their local area to give a sense of how they fit in compared to their family and friends, as well as the unique pressures on specific parts of the country that have more vulnerable people than average.
In its first weekend alone there were 131,000 searches of the tool – more than any other interactive we’ve ever produced – including more than 40,000 unique combinations of postcode, age and other input. Because the piece had nearly 20,000 hits from search in that first weekend alone, we were encouraged by the SEO team to republish it several times as more data came in (hence the later 2021 timestamp on the article). The project gained traction in the days after we published it and was featured on Times Radio the following week.
We also built an automatic share function for twitter which let people share a tweet with a sentence of their queue position pre-filled in. This helped us to track how people were interacting with it.
Our project was partly inspired by the New York TImes’ piece which estimated vaccine priority in different parts of the US
We started with the list of vaccine ‘priority groups’ from the Joint Committee of Vaccination and Immunisation in the UK, which prioritises care home staff and residents and over-80s, and moves down the age groups with special priority given to those with underlying health conditions. This gave us the ‘order’ in which people would be getting vaccinated.
However, because we wanted to show people their queue position within their local area, we had to source data on the number of people who were in each age group, in frontline health professions, in care homes as both residents and workers, and those who were clinically vulnerable, all by local authority – a Herculean task.
Fortunately, we found most of the data we needed from the Census which, where required, was then scaled up to take into account population growth since 2011. The data analysis was all done using R.
The tool itself was built by our brilliant developers using React.js. When readers entered a postcode, it was matched using postcodes.io to their local authority. Based on their inputs, their queue position was calculated from the number of people in each area that were in each priority group.
What was the hardest part of this project?
The most challenging part of the project was the data gathering, especially in trying to find data on population health for local authorities in all four nations of the UK. While the number of people who were “clinically extremely vulnerable”, and then those with “underlying health conditions”, is published for the whole of the UK, that data just did not exist at the local authority level.
We had to improvise. Insteaad we found data from the 2011 census by five-year age group and local authority, crucially split into whether they reportred being in “good”, “very good”, “bad” and “very bad” health, along with whether they had problems with their mobility. We used these categories to estimate how many people in each area were likely to be “clinically extremely vulnerable” as well as having “underlying health conditions”. We checked our estimates based on how many people nationally were known to be in these vulnerable categories. It was not a perfect solution, but it fit our purposes.
More than 1,000 readers in the first weekend opened the ‘methodology’ tab, suggesting their was a strong enthusiasm for the inner workings of the tool.
What can others learn from this project?
Don’t let perfect be the enemy of the good! When you’re trying to explain or show something to readers, sometimes a model with sensible assumptions but imperfect data can be just as effective. We didn’t have the exact data on vulernablle people in each area, but we improvised using the data we had.
More specifically, the tool drew people’s attention to the fascinating issue of varying demographics when it comes to vaccine distrirbution. It reminded us and other journalists that areas had more than twice the proportion of older and more vulnerable people than others, a point which became clear when readers in London compared their results to readers in ‘older’ parts of the country. Sure enough, several stories about the ‘postcode lottery’ of vaccine distribution followed in the weeks after.