Evaluating the generalisability of formulas used to set tidal volumes in mechanically ventilated patients: an observational, multicohort, retrospective study.

Abstract

Lung-protective mechanical ventilation with low tidal volume (V_T) decreased mortality in the ARMA trial. V_T in that trial was adjusted to predicted bodyweight (PBW) as a proxy for preinjury lung volumes, an approach that is now commonly used to set target V_T in patients who are mechanically ventilated. In other clinical contexts, PBW is not used to estimate lung volumes and spirometric reference equations are used instead. We aimed to compare predicted forced vital capacity (PFVC), calculated using the Global Lung Initiative 2012 spirometric reference equations, with PBW as a proxy for preinjury lung volume. In this observational, multicohort, retrospective study we compared ARMA PBW with PFVC estimated with the Global Lung Initiative 2012 reference equations for two cohorts: Medical Information Mart for Intensive Care-IV (MIMIC-IV), a cohort of patients from a single academic medical centre in Boston, MA, USA, and electronic Intensive Care Unit Collaborative Research Database (eICU-CRD), a cohort of patients from 208 hospitals in the USA. Patients who had lung-protective ventilation, hypoxaemia, and height in the range 150-210 cm were included. The data were collected from publicly available, deidentified datasets. We then studied the association between normalised V_T (V_T per PBW or V_T per PFVC) and the ratio of lung volume estimates (PBW:PFVC) and mortality, driving pressure, and normalised elastance in patients receiving lung-protective ventilation with V_T per PBW of 6-8 mL/kg. 9152 patients were included from MIMIC-IV (3064 [33·5%] female and 6088 [66·5%] male; 645 [7·0%] Black, 5885 [64·3%] White, and 2622 [28·6%] other race; mean age 63 years [SD 16]) and 12 420 patients were included from eICU-CRD (4262 [34·3%] female and 8158 [65·7%] male; 1023 [8·2%] Black, 10 093 [81·3%] White, and 1304 [10·5%] other race; mean age 62 years [SD 16]). Patients who were younger, male, and White had greater PFVCs than patients who were older, female, and non-White with the same PBW. A 1 SD increase in PBW:PFVC ratio was associated with 1·43-fold (95% CI 1·17-1·76) higher odds of death in MIMIC-IV and similar results were observed in the eICU-CRD cohort. Adjusting V_T using PBW formulas leads to overestimation of preinjury tidal lung volumes in patients who are older, shorter, female, and non-White compared with other patients. This algorithmic bias could contribute to mortality disparities in patients who are ventilated. Further research is required to establish the optimal tidal volumes for patients who are mechanically ventilated. National Heart, Lung, and Blood Institute.