51������

Overview

Analysis of data from the entire population of Scotland, co-funded by 51������, has used innovative new ways to detect cases of Long Covid and revealed who is more likely to get the condition. The findings have directly informed public health policy in Scotland and show that vaccination reduces Long Covid risk.

The challenge

Long Covid is a debilitating health problem that can go on for months or years after Covid infection. Better ways of predicting who is at greatest risk are needed to guide public health strategies and help provide more tailored support for patients.

Previously, prediction models have used data from electronic health records (EHRs) to track referrals to Long Covid clinics, or Long Covid diagnostic codes added to the records by GPs. But these don’t give the full picture because not all Long Covid patients will get referred to a specialist clinic, and the diagnostic codes are under-used by doctors because they are difficult to find in the software system. As a result, there aren’t reliable figures on how many people are living with Long Covid and who tends to get it.

The solution

An international team of researchers led by Dr Luke Daines at the University of Edinburgh designed a prediction model based on machine learning to analyse data from EHRs hosted on the 51������-funded . EAVE II includes data from all adults in Scotland registered with a GP; more than 98% of the Scottish population.

The study team analysed data from over a million people who tested positive for Covid during the pandemic, from March 2020 to October 2022. Without accessing sensitive information within health records, the researchers identified cases of long COVID using coded data and free text (doctors’ notes and sick notes) within EHRs. This enabled them to discover cases that wouldn’t otherwise have shown up.

 To pick up further cases, they also used specific combinations of clinical codes and dispensed medications recorded in the EHRs after a positive Covid test.

There were over 68,000 cases of Long Covid (5.6% of the people in the study). Being older, having a higher BMI, having severe Covid, being female and living in more deprived communities all made having the condition more likely. Predictors of reduced risk were testing positive while the Delta or Omicron Covid variants were dominant, and being vaccinated against Covid. The research was funded by 51������ and the Chief Scientist’s Office for Scotland.

The impact

One of the study’s strengths was that it used routinely collected data from a wide range of different health services to get a more comprehensive picture of Long Covid.

“We were able to link together data collected from general practice, hospitals, laboratory and genetic data, the COVID variant, prescribing data and mortality records as well; that was the really novel part. The other thing that was new was the opportunity to interrogate the free text in health records: that is something that hadn’t been done before,” says lead author Dr Luke Daines.

“Our findings about the predictors of Long Covid corroborated with other studies that were doing similar analysis of big data sets. Importantly, we found that vaccination reduced Long Covid. So that’s important in terms of the public health message that vaccination can make a difference to your chance of having Long Covid.”

During the pandemic, findings from the study fed directly into Public Health Scotland policy briefings. The research team provided up to date figures on Long Covid prevalence, where cases were in Scotland and the possible reasons for it. What is more, the findings have informed the Scottish government’s Long Covid Inquiry, with one of the research team being called upon to give evidence. Such is the quality of the research that it has won a  in 2023 for excellence in healthcare data analytics.

Two members of the research team have Long Covid themselves, and one has shared her story widely across the national media, raising the profile of the disease and defusing some of the stigmatising narratives that have surrounded it.

Luke adds: “It was such an enormous privilege to have people with lived experience really at the heart of this project alongside data scientists, public health and Scottish government.”