Deriving and validating a risk prediction model for long COVID-19: protocol for an observational cohort study using linked Scottish data

• Published in: BMJ open 2022-07, Vol.12 (7), p.e059385-e059385
• Main Authors: Daines, Luke, Mulholland, Rachel H, Vasileiou, Eleftheria, Hammersley, Vicky, Weatherill, David, Katikireddi, Srinivasa Vittal, Kerr, Steven, Moore, Emily, Pesenti, Elisa, Quint, Jennifer K, Shah, Syed Ahmar, Shi, Ting, Simpson, Colin R, Robertson, Chris, Sheikh, Aziz
• Format: Article
• Language: English
• Subjects: Codes; Coding standards; Cohort analysis; Cohort Studies; Computerized physician order entry; Coronaviruses; COVID-19; COVID-19 - complications; COVID-19 - epidemiology; COVID-19 diagnostic tests; COVID-19 vaccines; Datasets; Electronic health records; Health care; Hospitalization; Hospitals; Humans; Illnesses; Immunization; Infections; Intensive care; Laboratories; Long COVID; Mortality; Observational studies; Observational Studies as Topic; Pandemics; Population; Post-Acute COVID-19 Syndrome; Primary care; protocols & guidelines; Public Health; SARS-CoV-2; Severe acute respiratory syndrome coronavirus 2; Surveillance; Telemedicine
• ISSN: 2044-6055; 2044-6055
• DOI: 10.1136/bmjopen-2021-059385

IntroductionCOVID-19 is commonly experienced as an acute illness, yet some people continue to have symptoms that persist for weeks, or months (commonly referred to as ‘long-COVID’). It remains unclear which patients are at highest risk of developing long-COVID. In this protocol, we describe plans to develop a prediction model to identify individuals at risk of developing long-COVID.Methods and analysisWe will use the national Early Pandemic Evaluation and Enhanced Surveillance of COVID-19 (EAVE II) platform, a population-level linked dataset of routine electronic healthcare data from 5.4 million individuals in Scotland. We will identify potential indicators for long-COVID by identifying patterns in primary care data linked to information from out-of-hours general practitioner encounters, accident and emergency visits, hospital admissions, outpatient visits, medication prescribing/dispensing and mortality. We will investigate the potential indicators of long-COVID by performing a matched analysis between those with a positive reverse transcriptase PCR (RT-PCR) test for SARS-CoV-2 infection and two control groups: (1) individuals with at least one negative RT-PCR test and never tested positive; (2) the general population (everyone who did not test positive) of Scotland. Cluster analysis will then be used to determine the final definition of the outcome measure for long-COVID. We will then derive, internally and externally validate a prediction model to identify the epidemiological risk factors associated with long-COVID.Ethics and disseminationThe EAVE II study has obtained approvals from the Research Ethics Committee (reference: 12/SS/0201), and the Public Benefit and Privacy Panel for Health and Social Care (reference: 1920-0279). Study findings will be published in peer-reviewed journals and presented at conferences. Understanding the predictors for long-COVID and identifying the patient groups at greatest risk of persisting symptoms will inform future treatments and preventative strategies for long-COVID.