How choices in classifications and data sources influence post-COVID syndrome frequency rates: a demonstration

In the early COVID pandemic years (2020-2021) there was no uniform diagnosis for post-COVID syndrome, nor a specific diagnostic code for recording it in electronic health records (EHRs). Yet, questions soon arose about the scale of the health problem and about the characteristics of patients affected by it. To increase insight into this, we compared various research methodologies, definitions and data sources from the early pandemic years. We found that especially frequency estimates of the occurrence of the syndrome varied significantly, while the patient characteristics, like age and sex, were fairly similar. These insights can help to advance research into post-COVID syndrome and contribute to pandemic preparedness.

Some of those infected with SARS-CoV-2 suffer from post-COVID syndrome (PCS). We aimed to improve understanding of PCS by operationalizing different classifications and to explore clinical subtypes.

Frequency rates differ while patient characteristics remain consistent

The frequency of PCS was 15% with on average 4.6 symptoms for which the GP was consulted using the narrow definition and 32% with on average 6.8 symptoms for the broad definition. Across all methods and classifications, the mean age of individuals with PCS was around 53 years and they were more often female. There were small sex differences in the type of symptoms and overall symptoms were persistent for 6 months. The community detection analysis revealed three possible clinical subtypes.

The research

We used data from Nivel Primary Care Database from 2019–2020 which consists of electronic health records (EHR) from general practices (GPs), combined with sociodemographic data of 10,313 individuals infected with the SARS-CoV-2. In addition, data from 276 individuals who had been infected with the SARS-CoV-2 in 2021, collected via a longitudinal survey, were used.