Risk factors for post-acute sequelae of COVID-19 in hospitalized patients: An observational study based on a survey in a tertiary care center in Türkiye

Abstract

Introduction: Long COVID is a multisystem disease with various symptoms and risk factors. We aim to investigate the post-acute sequelae of COVID-19 and related risk factors in a tertiary care center.

Materials and methods: In this observational study, based on a survey of 1.977 COVID-19 patients hospitalized from April 2020 to January 2021, a retrospective assessment was carried out on 1.050 individuals who were reachable via telephone to determine their eligibility for meeting the inclusion criteria.

Results: The data of 256 patients who reported at least one persistent symptom were analyzed. Long COVID prevalence was 24.3%. Among 256 patients (median age 52.8; 52.7% female; 56.63% had at least one comorbidity), dyspnea, fatigue, arthralgia-myalgia, cough, and back pain were the most common post-acute sequelae of COVID-19 (42.4%; 28.29%; 16.33%; 13.15% and 7.17%, respectively). The risk factors for the persistence of dyspnea included having lung diseases such as chronic obstructive pulmonary disease, a history of intensive care support, the requirement for long-term oxygen therapy, and a history of cytokine storm (p= 0.024, p= 0.026, p< 0.001, p= 0.036, p= 0.005, respectively). The correlation between lung involvement with post-discharge cough (p= 0.041) and dizziness (p= 0.038) was significant. No correlation between the symptoms with the severity of acute infection, age, and gender was found. When a multivariate regression analysis was conducted on the most common long COVID-related symptoms, several independent risk factors were identified. These included having lung disease for dyspnea (OR 5.81, 95% CI 1.08-31.07, p= 0.04); length of hospital stay for myalgia (OR 1.034, 95% CI 1.004-1.065, p= 0.024); and pulmonary involvement of over 50% during COVID-19 infection for cough (OR 3.793, 95% CI 1.184-12.147, p= 0.025).

Conclusion: COVID-19 survivors will require significant healthcare services due to their prolonged symptoms. We hope that our findings will guide the management of these patients in clinical settings towards best practices.

Figure 1

Flowchart of the study

Figure 2

The most common five post-acute sequelae of COVID-19 (PASC) according to the number of patients (n= 256)

Figure 3

ROC curves of multivariate logistic regression analyses for the risk factors of dyspnea, myalgia, and cough as a PASC. Having lung disease (OR 5.81, 95% CI 1.08-31.07, p= 0.04) was found to be an independent risk factor for the development of dyspnea. The Hosmer-Lemeshow test for goodness of fit indicates that the model adequately fits the data (χ2= 3.006, p= 0.223). The area under the ROC curve with a 95% confidence interval for predicted probabilities was found as 0.622 (0.501-743) indicating poor discrimination. Length of hospital stay (OR 1.034, 95% CI 1.004-1.065, p= 0.024) was found to be an independent risk factor for myalgia. The Hosmer-Lemeshow test for goodness of fit indicates that the model adequately fits the data (χ2= 12.359, p= 0.136). The area under the ROC curve with a 95% confidence interval for predicted probabilities was found to be 0.578 (0.478-0.678), indicating poor discrimination. Pulmonary involvement of >50% during COVID-19 infection (OR 3.793, 95% CI 1.184-12.147, p= 0.025) was found to be an independent risk factor for cough. The Hosmer-Lemeshow test for goodness of fit indicates that the model adequately fits the data (X²< 0.001, p> 0.999). The area under the ROC curve with a 95% confidence interval for predicted probabilities was found to be 0.608 (0.502-0.714), indicating poor discrimination.