Use of non-invasive mechanical ventilation in the Emergency Department, clinical outcomes and correlates of failure

Abstract

Background: Despite several studies having been carried in this organizational context, there is an absence of information about the effectiveness of non-invasive mechanical ventilation (NIV) in Emergency Departments (ED), based on a number of suitable patients with acute respiratory failure (ARF) of different aetiology. In particular, it has not yet been defined as to whether the context of the ED suits the necessary requirement of quality for the correct application of the method and if the obtained results are different from those taken in other studies in general or respiratory intensive care unit. Finally there are few data related to the predictive factors to NIV failure (endotracheal intubation, in-hospital mortality) when applied in the emergency setting. Methods: To answer these questions we have retrospectively studied a population of 210 patients (95 with COPD exsacerbation; 92 with acute cardiogenic pulmonary oedema; 23 with severe community acquired pneumonia) treated for ARF in the "critical area" of four Italian level II Emergency Departments. For all patients demographic data; some comorbidities (diabetes, dementia, sopraventricular arrhythmias, obesity); the physiological scores (Kelly, SAPS II, Apache II); the need for pharmacological sedation; vital and blood gas parameters (evaluated at entry, after one hour of treatment and before its suspension); the ventilatory modality applied (CPAP or PSV+PEEP) and some parameters of in-hospital stay (duration of the hospitalization in the critical area, duration of ventilation, compliance to the treatment, patient's refusal to continue it, development of skin necrosis, need for endotracheal intubation, in-hospital mortality) were considered. Finally demographic, event of death with Cox regression or to the need for ETI through linear regression analysis. Results: Globally, in-hospital mortality reached 13,3%, the percentage of failure with consequent endotracheal intubation amounted to 10,4%. Considering the single aetiologic groups in the patients with COPD, exsacerbation mortality and ETI percentage were 12,6% and 13,7% respectively; in ACPO patients these data respectively amounted to 3,3% and 4,3%; while for patients with severe CAP they respectively reached 34,7% and 21,7%. The following factors were independently correlated with in-hospital mortality: SAPS II>35, presence of dementia for COPD patients; SAPS II>35; presence of dementia, presence of sopraventricular arrhythmias for ACPO patients; SAPS II>35, presence of sopraventricular arrhythmias, presence of dementia for CAP patients. Considering the whole population of 210 patients, the predictive factors of in-hospital mortality were the following: SAPS II>35; presence of dementia; presence of sopraventricular arrhythmias; maintenance of a respiratory rate above 24 bpm during tratment. The following were factors independently correlated with the need for endotracheal intubation: male gender, pH<7,32, respiratory rate>24 bpm, mean arterial pressure>96 mmHg, all measured at one hour of treatment, for COPD patients; male gender sex, pH<7,32, respiratory rate>24 bpm, PaCO2>54,5 mm Hg all measured after one hour treatment for ACPO patients. Given the low number of patients, it was not possible to perform the logistic regression and to calculate the matrix of covariance of the parameter for the CAP group. Considering the whole series of patients, the factors independently correlated to ETI resulting in the following: male gender; diagnosis of COPD; pH<7,32, respiratory rate>24 bpm, mean arterial pressure>96 mm Hg, PaCO2>54,5 mm Hg, all measured after one hour treatment. Conclusions: In conclusion, our study shows that NIV is practicable in the ED with safety and clinical results comparable to those obtained in the general or respiratory intensive care units. In addition, it points out some factors, quickly obtainable, correlated to a greater probability of in-hospital death or need for ETI. Such factors can contribute more properly to a better selection of the patient candidate to NIV in the ED with obvious effects on organization and saving of resources. The validation of these results on the general plan would suggest the need for future prospective studies.