Comparison of Ventilatory Ratios Calculated Using Predicted Versus Actual Body Weight

Ventilatory ratio (VR) is a surrogate marker of dead-space fraction that is increasingly used to assess ventilation impairment and predict clinical outcomes. However, the current VR equation uses on predicted body weight (PBW) to estimate reference minute ventilation, which may lead to an overestimation of dead-space fraction in patients with extreme body weights. We hypothesized that VR calculated using PBW and actual body weight (ABW) might differ in their ability to predict outcomes, particularly in obese patients.

This retrospective study compared the discrimination capacity of PBW-estimated VR (pVR), ABW-estimated VR (aVR), and corrected minute ventilation (V_Ecorr) for a composite outcome of death or weaning failure at ICU discharge across three body weight categories: underweight, normal-to-overweight, and obese. The area under the receiver operating characteristic curve (AUROC) was used to assess the discrimination capacity of the study outcome.

A total of 4,015 subjects who received mechanical ventilation were included. The primary composite outcome events occurred in 40.6% of subjects, including a mortality rate of 29.1%. In underweight subjects, no significant differences in discrimination capacity were observed among pVR, aVR, and V_Ecorr (AUROC, 0.599, 0.598, 0.595, P = .92). In normal-to-overweight subjects, aVR outperformed V_Ecorr (AUROC, 0.582 vs 0.571, P = .02), while pVR did not (AUROC, 0.576 vs 0.571, P = .39). In obese subjects, both aVR and V_Ecorr outperformed pVR (AUROC 0.559 vs 0.520, P = .003; and AUROC 0.557 vs 0.520, P = .041).

In nonobese subjects, there was no significant difference between using PBW and ABW to calculate VR for outcome prediction. However, in obese subjects, using ABW was associated with better predictive validity than using PBW. These findings suggest that PBW-based VR may be less reliable than ABW-based VR for assessing ventilation impairment in patients with higher body weight.