Physical functional recovery in the early period after hospitalization for COVID-19: impact of hypertension and outcome prediction model

The aim – to study the influence of hypertension (HT) on the dynamics of natural recovery of physical functional status in the early period after hospitalization for COVID-19 and to develop a model for predicting recovery results at 1 month after discharge.Materials and methods. 221 hospitalized patients with COVID-19 (age 53.4±13.6 years, 53 % women) were included in the study, 176 of whom underwent the 6-minute walk test (6MWT) using an extended protocol within 1-2 days before discharge. A repeat visit to assess the dynamics of natural recovery was performed at 1 month after discharge.Results and discussion. Pre-discharge 6-minute walk distance was 378±57 m in patients with HT and 418±75 m without it, p=0.001, during the second visit – 440±52 versus 478±68, p=0.002; the achieved percentage of the individually predicted distance was 67.4±10.5 vs. 69.5±13.6 % and 81.6±9.9 vs. 81.9±15.7 %, respectively, p>0.05 for both visits. The increase in heart rate during the test at visit 1 was 18.5±8.3 versus 30.1±19.3 bpm, p<0.001, the percentage of chronotropic reserve utilizatoin was 21.3±9.6 % versus 29.2±11.4 %, p<0.001. During the second visit, residual manifestations of this trend were observed, with an increase in HR by 24.0±9.5 vs. 30.8±12.1, p=0.003 and the use of chronotropic reserve of 28.1±10.1 % vs. 33.4±12.4 %, respectively, p=0.029. The developed multivariate linear regression model explained 59 % of the variability in the achieved percentage of the individually predicted 6-minute walk distance at 1 month after discharge. The use of machine learning allowed to create an artificial neural network based regression model that used age, height, use of remdesivir in treatment, and SBP and DBP values at the time of discharge as predictors, and explained 90 % of observed variability.Conclusions. Hospitalized patients with COVID-19 were characterized by a decrease in the general physical functional status as assessed by 6MWT at the time of discharge and incomplete recovery after 1 month. Presence of hypertension was associated with more pronounced disturbances of the autonomic regulation of heart rate, but did not affect the reached percentage of the distance walked. The proposed artificial neural network based regression model allows for a high accuracy prediction of the 6MWT results at 1 month after discharge, which can be used in the selection of candidates for cardiopulmonary rehabilitation programs.