Abstract Rationale Ventilatory ratio (VR), calculated as (minute ventilation × arterial partial pressure of carbon dioxide PaCO2) / (predicted minute ventilation × 38.5), offers an advantage for estimating dead space because it does not rely on capnography. Prior research suggests that the capnography-based alveolar dead space fraction (AVDSf), defined as (PaCO2 - end-tidal carbon dioxide PETCO2) / PaCO2, is more strongly associated with mortality risk than VR in pediatric patients. We hypothesized that developing a pediatric-specific model for predicted minute ventilation would improve VR-based mortality risk stratification in critically ill children. Methods We conducted a single-site, retrospective observational cohort study of pediatric patients who required invasive mechanical ventilation in either the pediatric or cardiac intensive care unit (ICU) between January 2022 and December 2024. A predictive model for minute ventilation was derived from a first subgroup of patients with an oxygenation index (OI) 4, AVDSf 0.05, pH 7.35-7.50, and no new neurologic disease. Validation was performed in a second subgroup meeting the same criteria at first observation. VR was calculated 1) using the model-based predicted minute ventilation and 2) using the traditional method of age-based normal respiratory rate multiplied by 8 mL/kg tidal volume. A third subgroup of patients, excluding cardiac ICU cases, and with either an abnormal OI, AVDSf, or pH at first observation, was used to evaluate the association between model-based VR, traditional VR, and AVDSf with the primary outcome of mortality. Models were adjusted for OI. Results A total of 2,004 children were analyzed. The mixed-effects linear model developed in the derivation cohort of 1,210 patients (mean of 3.9 observations per patient) included age, body mass index (BMI) z-score, and heart rate, explaining 91% of the variance in minute ventilation (R² = 0.91). In validation, predicted and actual minute ventilation were strongly correlated (r = 0.94, 95% CI 0.91-0.96). In the outcomes cohort of 703 children (12.1% mortality), model-based VR was not associated with mortality in univariate analysis (OR 1.18 per SD change, 95% CI 0.99-1.40). In contrast, both traditional VR (OR 1.21 per SD change, 95% CI 1.02-1.45) and AVDSf (OR 1.40 per SD change, 95% CI 1.08-1.82) were significantly associated with mortality risk in univariate and multivariable analyses. Conclusions Refining predicted minute ventilation with a pediatric-specific model does not improve the utility of VR for mortality risk stratification in critically ill children. Traditional VR estimates based on age-predicted respiratory rate and tidal volume remain superior. This abstract is funded by: NIH
Whiteley et al. (Fri,) studied this question.