Early Identification of Severe COVID-19 Cases and the Need for ICU Care Based on Clinical and Laboratory Risk Factors

Background and objective Treatment in ICUs became extremely difficult due to the growing number of coronavirus disease 2019 (COVID-19) patients at the height of the pandemic. Consequently, prompt patient triage depends on the early categorization of severe cases in such scenarios. This study aimed to provide an evidence-based strategy to ensure the best use of resources by triaging patients based on objective risk factors. Methods This retrospective observational study comprised 500 inpatients (>age 18 years) who were hospitalized between March 20 and April 19, 2020, at the Khyber Teaching Hospital (KTH) and Hayatabad Medical Complex (HMC) in Peshawar, Pakistan. The clinical, laboratory, and radiological parameters were assessed. Real-time polymerase chain reaction (RT-PCR) findings were used to confirm the diagnosis of COVID-19. Results A total of 19 potential clinical and laboratory risk factors associated with ICU admissions were identified. At least one comorbidity among chronic lung disease, cardiovascular disease (CVD), and diabetes was the factor with the strongest association with ICU admission with a univariable odds ratio (OR) of over 27, followed by renal disease and other COVID-19 sequelae such as diarrhea, respiratory rate (>24 breaths/minute), and positive RT-PCR (vs. negative) with an univariable OR between 9 and 15. Furthermore, a multivariate logistic regression model was further developed with five risk factors, including comorbidity, presence of chronic lung disease, presence of diabetes, and RT-PCR (positive vs. negative), male sex (vs. female), and older age (65.0-80.5 years), suggesting a good fit of the model to the data shown by the area under the receiver operator characteristic curve (AUC) of 0.943 (95% CI: 0.917, 0.969). Additionally, a chest CT scan showed the typical COVID-19 pneumonia with pulmonary involvement of 30-40%, which was further evaluated by the COVID-19 Reporting and Data System (CO-RADS). The typical COVID-19 pneumonia was on a scale of four (15/25) or five (19/25) lung lesions. Conclusions Based on our findings, this approach could be used to screen the severe cases of COVID-19 patients and help them to be treated in ICUs on time while preventing others from unnecessarily using ICUs in the setting of limited medical resources, such as the outbreak of a pandemic.