COMPREHENSIVE PRE-DISCHARGE SYMPTOMS ASSESSMENT IN HYPERTENSIVE PATIENTS HOSPITALIZED FOR COVID-19 ALLOWS TO PREDICT PERSISTENCE OF SYMPTOMS AT 3 MONTHS

Abstract

Objective: To study the prognostic value of survey-based respiratory, physical, emotional, and cognitive symptoms assessment in hypertensive patients hospitalized for COVID-19 in prediction of symptoms persistence at 3 months after discharge. Design and method: 88 hospitalized patients with COVID-19 and a history of hypertension (mean age 57.8+-11.8 years, 53% female) underwent a comprehensive survey-based symptoms assessment 1-2 days prior to discharge that included MRC Dyspnea scale, CAT and CCQ questionnaires, HADS, the physical symptoms subscale of the EFTER-COVID questionnaire, and the Memory, Thinking, and Communication subscale of the SBQ-LC questionnaire. Marginal effects in logistic regression analysis were used to gauge the predictive value of resulting scores and separate symptoms. 500 SANN-based classification models were subsequently trained for various combinations of significant predictors, with an automatic ranging of obtained models by their predictive accuracy. Results: 77% of participants have reported an incomplete ecovery at 3 month after hospital discharge. In the order of decreasing predictive value for persistence of symptoms, significant (p < 0,05) risk factors included EFTER-COVID physical symptoms score (Somers’ D-statistic = 0,63), HADS depression subscale (D = 0,58), SBQ-LC Memory, Thinking, and Communication score (D = 0,52), MRC dyspnea class (D = 0,45), and CAT score (D = 0,30); among the specific symptoms, significant predictors included dyspnea (D = 0,42), dizziness (D = 0,38), muscles and joints pain (D = 0,38), and difficulties in remembering things (D = 0,40). Out of the derived machine-learning based classification models, the optimal performance was observed for the one using gender, pre-discharge MRC dyspnea class, and summary scores on EFTER-COVID physical symptoms and SBQ-LC Memory, Thinking, and Communication subscales. The model exhibited a 100% predictive acccuracy in classifying the pre-specified test/validation subset of the study group into those who would subsequently report persistence of symptoms or complete recovery. Conclusions: Survey-based pre-discharge assessment of symptoms in hypertensive patients hospitalized for acute COVID-19 yields a high prognostic value. The combination of gender and survey-based measures of dyspnea, physical and cognitive symptoms may be used in this setting to predict the persistence of symptoms at 3 months after hospital discharge.