SESSION TITLE: Critical Care Posters II
SESSION TYPE: Poster Presentations
PRESENTED ON: Saturday, March 22, 2014 at 01:15 PM - 02:15 PM
PURPOSE: Our aim was to describe an easy-to-use and pragmatic index based on clinical variables that reliably predicts the need for mechanical ventilation (MV) in patients treated with high flow nasal cannula (HFNC).
METHODS: This is a prospective cohort observational study including patients with pneumonia requiring ICU admission over a 2 year period. HFNC failure was defined as further need for MV. Respiratory variables that were significantly different among groups (success vs failure) were used to create a score to predict the need of MV, named FLOR as the ratio of SpO2/FIO2 to respiratory rate (RR) and administered flow [(SpO2/FIO2)x100/(RRxFlow)]. Multivariate logistic regression models were used to adjust for potential confounding. Discrimination of each significant variable and FLOR index (FLORi) was tested by calculating the area of receiver operating characteristic curve (AUROC).
RESULTS: Seventy-seven patients with pneumonia were treated with HFNC, being successful in 56 (72.7%). Patients who succeed on HFNC had higher SpO2/FIO2 (140[95%CI 129-151] vs 119[95%CI 104-133];p=0.04), decreased RR (22[95%CI 20-24]bpm vs 26[95%CI 24-28]bpm;p=0.01) and needed lower flow rates (29[95%CI 28-30]lpm vs 32[95%CI 29-34]lpm;p=0.02) after 12h of the treatment onset. These differences increased throughout the study period. FLORi was higher in HFNC success from 6h of treatment and differences increased over time. (21 [95% CI 18-24] vs 16 [95%CI 14-19];p=0.04). In a logistic regression model, only FLORi estimated from 12h after HFNC onset (OR 1.24 [95%CI 1.08-1.43]; p<0.01) to the subsequent time points was significantly predicted the need for MV, even after adjustment for potential confounding. Compared with other variables, FLORi had the best discrimination for the need for MV. The AUROC improved from 0.80 (95%CI 0.68-0.91) after 12h of initiation of HFNC to 0.90 (95%CI 0.81-0.99) at the end of the first day of treatment (p=0.04).
CONCLUSIONS: FLORi can identify patients at risk of MV in patients with pneumonia treated with HFNC.
CLINICAL IMPLICATIONS: Early identification of patients who fail on HFNC may avoid the apparition of complications associated with delayed intubation.
DISCLOSURE: Oriol Roca: Other: Supported in part by Ventilung,(r) (National Course of Mechanical Ventilation). Our Critical Care Department has received research funding from Fisher & Paykel not related with the present study. The company did not take part in the design of the study, the interpretation of the results or the abstract preparation. Joan Ramon Masclans: Other: Supported in part by Ventilung,(r) (National Course of Mechanical Ventilation). Our Critical Care Department has received research funding from Fisher & Paykel not related with the present study. The company did not take part in the design of the study, the interpretation of the results or the abstract preparation. The following authors have nothing to disclose: Berta Caralt Ramisa, Marina García de Acilu, Jordi Rello
No Product/Research Disclosure Information