Remote Covid Assessment in Primary Care (RECAP) risk prediction tool: derivation and real-world validation studies

Abstract

Background Accurate assessment of COVID-19 severity in the community is essential for best patient care and efficient use of services and requires a risk prediction score that is COVID-19 specific and adequately validated in a community setting. Following a qualitative phase to identify signs, symptoms and risk factors, we sought to develop and validate two COVID-19-specific risk prediction scores RECAP-GP (without peripheral oxygen saturation (SpO2)) and RECAP-O2 (with SpO2). Methods Prospective cohort study using multivariable logistic regression for model development. Data on signs and symptoms (model predictors) were collected on community-based patients with suspected COVID-19 via primary care electronic health records systems and linked with secondary data on hospital admission (primary outcome) within 28 days of symptom onset. Data sources: RECAP-GP: Oxford-Royal College of General Practitioners Research and Surveillance Centre (RSC) primary care practices (development), Northwest London (NWL) primary care practices, NHS COVID-19 Clinical Assessment Service (CCAS) (validation). RECAP-O2: Doctaly Assist platform (development, and validation in subsequent sample). Estimated sample size was 2,880 per model. Findings Data were available from 8,311 individuals. Observations, such SpO2, were mostly missing in NWL, RSC, and CCAS data; however, SpO2 was available for around 70% of Doctaly patients. In the final predictive models, RECAP-GP included sex, age, degree of breathlessness, temperature symptoms, and presence of hypertension (Area Under the Curve (AUC): 0.802, Validation Negative Predictive Value (NPV) of low risk 98.8%. RECAP-O2 included age, degree of breathlessness, fatigue, and SpO2 at rest (AUC: 0.843), Validation NPV of low risk 99.4%. Interpretation Both RECAP models are a valid tool in the assessment of COVID-19 patients in the community. RECAP-GP can be used initially, without need for observations, to identify patients who require monitoring. If the patient is monitored at home and SpO2 is available, RECAP-O2 is useful to assess the need for further treatment escalation.

Home monitoring of coronavirus disease 2019 patients in different phases of disease

“…In the future, selection of high-risk COVID-19 patients, for example, kidney transplant and other immunocompromised patients, could increase effectiveness [40,41]. Scoring systems could select high-risk patients, for example, based on dyspnoea, sex, temperature and hypertension [42]. Home monitoring systems must be adapted to this population.…”

“…This also meets the current healthcare needs. Algorithms could increase efficiency by predicting the need for clinical assessment and escalation of care, for example, by using age, degree of breathlessness, fatigue and oxygen saturation at rest or predict desaturation by serial oxygen saturation and heart rate monitoring [42,43]. However, these algorithm first need to be tested among different patient groups.…”