Statistical Analysis
Patient and event characteristics were described using means or medians, and proportions. Sub-scores of TLS are expressed as mean values and as the mean % of TLS, and were examined both overall and by etiology of ARF. 
Mortality, ICU and hospital length of stay (LOS), PaO2/FiO2 ratio and ventilator hours were examined across tertiles of TLS, with Cuzick’s trend test used to assess trends across tertiles.
Lung scores were also examined by the common clinical respiratory disease categories of no pulmonary disease (e.g. altered level of consciousness, upper airway obstruction, and neuromuscular weakness), obstructive disease (e.g. asthma and chronic obstructive pulmonary disease (COPD)), and parenchymal disease (e.g. pneumonia, ARDS, aspiration, and CHF), with one-way analysis of variance (ANOVA) used to assess whether disease category explained a significant amount of the variation in scores.
kappa coefficient was used to assess inter-rater agreement in the pre-study training set. All analyses were conducted using Stata 14.1 (StataCorp, LP, College Station, TX).