University of Twente Student Theses


Risk factors for ICU admission, long-term stay and mortality in hospitalized COVID-19 patients

Eskes, Harm and Schilderink, Jasper L.M. (2021) Risk factors for ICU admission, long-term stay and mortality in hospitalized COVID-19 patients.

[img] PDF
Abstract:OBJECTIVE: The aim of this study is to develop a prediction model based on demographic and clinical characteristics at hospital admission to identify risk factors for intensive care unit (ICU) admission, long-term stay, and mortality in hospitalized COVID-19 patients. OUTCOMES: Primary outcomes were ICU admission, long-term stay, and death while hospitalized. METHOD A total of 334 eligible patients (208 [62.3%] male, median age 65.5 [58-76] years), admitted to either a general ward or ICU of Medisch Spectrum Twente were enrolled in this study. Data such as demographics, comorbidities, vital signs, respiratory condition and laboratory test results, was obtained from the electronic medical records and the Castor database of our COVID-19 cohort at hospital admission. Logistic regression was used to identify independent variables predicting the three outcomes. All three models were validated by randomly splitting the data into 80% for training and 20% for validating. Performance accuracy was evaluated using area under the receiver operating characteristic curve (AUC-ROC) analysis. FINDINGS: Eight independently significant predictors for ICU admission were BMI between 25-30 kg·m−2 compared to BMI ≤25 kg·m−2, respiratory rate ≥20/min, pH value 7.35-7.45 or ≤7.35 compared to pH value ≥7.45, HCO3 ≥23.8 mEq/L, neutrophils ≥5.18 cells/µL, C-Reactive Protein ≥79.5 mg/L and D-dimer ≥2286 µg/L. For long-term stay (≥7 days), 6 predictors were independently significant: age ≥70 years compared to age ≤50 years, use of immunosuppressive therapy, respiratory rate ≥20/min, pH value ≤7.35 compared to pH value ≥7.45, HCO3 ≥23.8 mEq/L and D-dimer ≥2286 µg/L. In case of death while hospitalized, age ≥70 years compared to age ≤50 years, male gender, sodium ≥136 mmol/L, D-dimer ≥2286 µg/L, and oxygen therapy ≤4 L compared to no oxygen therapy were found as the 5 best predictors. The prediction model of the validation set yielded an AUC-ROC of 0.94 (95% CI [0.88-1.00], p <0.001) for ICU admission, a AUC-ROC of 0.63 (95% CI [0.49-0.77], p = 0.081) for long-term stay and a AUC-ROC of 0.68 (95% CI [0.48-0.89], p = 0.093) for death while hospitalized. CONCLUSION: This study identified key independent predictors for ICU admission, long-term stay and death while hospitalized with COVID-19. These predictors offer the potential to stratify patients based on risk factors so that they can triage COVID-19 patients more effectively.
Item Type:Essay (Bachelor)
Medisch Spectrum Twente, Enschede, Nederland
Faculty:TNW: Science and Technology
Subject:44 medicine
Programme:Health Sciences BSc (56553)
Keywords:COVID-19, SARS-CoV-2, Prediction model, Risk factors, Outcomes
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page