• Mirza Kovačević, Nermina Rizvanović, and Šabanović AdilovićAdisaADepartment of Anaesthesiology and Intensive Care Unit, Cantonal Hospital Zenica, Bosnia and Herzegovina..
• Department of Anaesthesiology and Intensive Care Unit, Cantonal Hospital Zenica, Bosnia and Herzegovina.
• Med Glas (Zenica). 2021 Aug 1; 18 (2): 362-369.

AbstractAim To identify predictive factors related with noninvasive ventilation (NIV) failure that are not based on the patient's respiratory status or acid base gas analyses in COVID-19 critically ill patients, and to create a predictive model of NIV failure. Methods A total of 73 COVID-19 critically ill patients who developed acute respiratory failure and underwent NIV were divided into two groups: Group 1, patients who required endotracheal intubation and invasive mechanical ventilation after NIV and Group 2, patients with successful weaning from NIV. Demographic data, clinical symptoms and signs, clinical index and scores, duration indicators and laboratory data were analysed. Predictive factors of NIV failure were assessed using univariate and multivariate regression analyses followed by the receiver operating characteristic (ROC) curve. Results In the Group 1 (NIV failure) there were 54 (73.97%) patients. Predictive factors for NIV failure were: the presence of dyspnoea on the day of admission at hospital (p<0.05; sensitivity 44.40%; specificity 84.20%), higher radiographic assessment of lung oedema score (RALES) on the day of starting NIV (p<0.009; sensitivity 70.40%; specificity 73.75%), higher length of NIV (p<0.014; sensitivity 48%; specificity 84.10%) and higher urea on the day of starting NIV (p<0.004; sensitivity 70.44%; specificity 73.72%) Conclusion NIV treatment in COVID-19 critically ill patients has a high failure rate. In addition to respiratory parameters, dyspnoea, higher RALES, higher length of NIV and increased urea value could predict NIV failure. These factors should be considered in treatment decision making.Copyright© by the Medical Assotiation of Zenica-Doboj Canton.

17 keywords...

hide…