A precise score for the regular monitoring of COVID-19 patients condition validated within the first two waves of the pandemic

PurposeThe sudden outbreak of COVID-19 pandemic have shown that medical community needs an accurate and interpretable aggregated score not only for an outcome prediction but also for a daily patients condition assessment. Due to a continuously changing pandemic landscape, a robustness becomes a crucial additional requirement for the score. Materials and methodsIn this research a real-world data collected within the first two waves of COVID-19 pandemic was used. The first wave data (1349 cases collected from 27.04.2020 to 03.08.2020) was used as a training set for the score development, while the second wave data (1453 cases collected from 01.11.2020 to 19.01.2021) was used as a validating set. For all the available patients features we tested their association with an outcome using a robust linear regression. Statistically significant features were taken to the further analysis for each of which their partial sensitivity, specificity and promptness were estimated. The sensitivity and the specificity were further combined into a feature informativeness index. ResultsThe developed score was derived as a weighted sum of the following 9 features showed the best trade-off between informativeness and promptness: APTT (> 42 sec, 4 points), CRP (> 146 mg/L, 3 points), D-dimer (> 2149 mkg/L, 4 points), Glucose (> 9 mmol/L, 4 points), Hemoglobin (< 115 g/L, 3 points), Lymphocytes (< 0,7*10^9/L, 3 points), Total protein (< 61 g/L, 6 points), Urea (> 11 mmol/L, 5 points) and WBC (> 13,5*10^9/L, 4 points). Thus, the proposed score ranges between 0 and 36 points. Internal and temporal validation showed that sensitivity and specificity over 90% may be achieved with an expected prediction range >7 days. Moreover, we demonstrated a high robustness of the score to the varying peculiarities of the pandemic. For the additional simplicity of application we split the full range of the score into five grades delimited with 9, 14, 19 and 24 points which determine expected death:discharge odds 1:100, 1:25, 1:5 and 1:1 correspondingly. ConclusionsAn extensive application of the score within the second wave of COVID-19 pandemic showed its potential for the optimization of patients management as well as improvement of medical staff attentiveness during a high workload stress. The transparent structure of the score as well as tractable cut-off bounds simplified its implementation into a clinical practice.