Three diagrams representing:
- A single condition, where many symptoms are encapsulated by a single grouping
- A cluster of overlapping symptoms (syndromes), where many symptoms are encapsulated by several different but overlapping groupings
- Individual symptoms, which are described separately
Moving from a single condition, to clusters of symptoms, to individual symptoms represents increasing differentiation.
An illustration demonstrating some of the reasons why the estimated prevalence of Long Covid is not comparable between studies. The illustration is based on 27 articles that estimate Long Covid prevalence (published or pre-prints) between May and October 2020.
The illustration notes that the prevalence of Long Covid has been stated as between 2.3% and 91% and suggests you can’t compare estimates for four reasons:
- The number of participants in the study – which varies between less than 50 and more than 1000
- Bias in the study – there is variation in recruitment (e.g. only recruiting cohorts from hospitals or support groups), participants (e.g. only including older individuals, or those who have had a positive PCR test) and symptoms (e.g. only including those who were symptomatic or have less than 5 symptoms)
- The care the patients in the study have received – there is variation across the studies in whether patients were hospitalised, non-hospitalised or a mixed cohort
- Timing of the study – the point at which symptoms were measured varies from a few weeks to 6 months after acute infection
Scatter plot plotting studies by their estimate of Long Covid prevelance against the length of follow-up. The prevalence of one or more ‘Long Covid’ symptoms (from 0% to 100%) is on the Y axis. The X axis shows the length of follow up (average number of days, as reported) from 0 days to 250 days, in increments of 50.
Studies have then been plotted on the chart, and differentiated by whether they are based in hospital, outpatient, or community settings. A majority of the studies followed-up for less than 100 days, but a handful (Yeoh, Huang, the NIHR CED survey and Munblit) followed-up over 150 days. The estimates of prevalence are very varied across the studies, even across those with a similar timeframe, ranging from 10% to 89%.