NHS records data including CPRD have been used to study the epidemiology of COVID-19 in the last two years. Incidence (new cases) and prevalence (existing cases) are usual metrics of this. Understanding the reliability and usefulness of NHS records to monitor these will impact future decisions on the need for testing and other measures to control the current and any future pandemics.
Our objective is to assess the impact of different analytic choices when estimating incidence and prevalence of COVID-19 infections and reinfections.
We will calculate prevalence as the number of with COVID-19 in a period or specific date, and incidence as the number of people newly diagnosed with a condition in a time period. We will calculate these in specific populations of interest based on age and sex, and over calendar years or months to look at trends over time.
Finally, we will look at the resulting figures and compare them with official numbers from government sources and dashboards.
Background: Primary care records provide a unique source of data for estimating the incidence and prevalence of health conditions, medication use, and other health-related events of interest. CPRD has been used for the study of COVID-19 infections, but little work has been done on reinfections.
Objective: We aim to estimate the incidence and/or prevalence of COVID-19 primoinfections and reinfections. We will compare the resulting figures with those obtained from official dashboards.
Study design: Population-based cohort
Population: All people in CPRD GOLD or AURUM with >=1 year of prior history will be eligible, with sensitivity analyses using different prior history requirements.
Variables: COVID-19 infections will be identified based on clinical diagnoses and/or positive PCR or other test records. Reinfections will be identified by imposing a 6-week washout period after a first infection. Age (categorised) and sex will be studied for stratification. Monthly figures will be estimated to observe time trends and waves of infection.
Analyses: Point and period prevalence will be estimated for COVID-19, including partial prevalence for COVID-19, where individuals will be considered as a prevalent case if they have had a record in the prior 1 month. Incidence rates will also be estimated using a 6-week washout. Analyses will be stratified by age, sex, and calendar month. We will plot official figures from UKHSA/PHE dashboards for comparison.
1. incident COVID-19 identified by a recorded clinical diagnosis or a positive PCR
2. Covid-19 re-infection, identified as #1 but at least 6 weeks after a first infection
Daniel Prieto-Alhambra - Chief Investigator - University of Oxford
Martí Català Sabaté - Corresponding Applicant - University of Oxford
Antonella Delmestri - Collaborator - University of Oxford
Edward Burn - Collaborator - University of Oxford
Hezekiah Omulo - Collaborator - University of Oxford
Wai Yi Man - Collaborator - University of Oxford
Xintong Li - Collaborator - University of Oxford