Lung-heart pressure index is a risk factor for acute respiratory distress syndrome (ARDS): A machine learning and propensity score-matching study.

Background Ventilator-induced lung injury (VILI) and haemodynamic instability play vital roles in acute respiratory distress syndrome (ARDS). The principle of driving pressure (DP) is the response to “volutrauma”, and the mean arterial pressure (MAP) is a haemodynamic systemic manifestation. In this study, we explored a novel lung-heart pressure index (LHPI) based on DP and MAP and explored its prognostic value in patients with ARDS.Methods This is a retrospective study based on the MIMIC-III database. ARDS patients who had undergone mechanical ventilation for more than 48 hours were selected through structured query language. The focus of the study was whether a high LHPI was associated with 30-day mortality and whether its predictive power was better than that of DP and mechanical power (MP). We used random forest, propensity-score matching, and logistic regression to test our hypothesis.Results A total of 448 ICU ARDS patients were enrolled. The mortality rate of ARDS patients was 29.02%. The LHPI was more important than DP and MP in the random forest. A significant adverse effect of high LHPI on 30-day mortality was observed in the high-LHPI group compared to the effect observed in the low-LHPI group (OR=1.86, 95% CI 1.08–3.26, p =0.027). More importantly, LHPI was significantly better than DP (NRI=0.054, 95% CI (0.014-0.094), P=0.008; IDI=0.011, 95% CI (0.002-0.019), P=0.014) and MP (NRI=0.061, 95% CI (0.001-0.122), P=0.049; IDI=0.047, 95% CI (0.022-0.071), P<0.001) in predicting mortality.Conclusions The study showed that the LHPI was a powerful prognostic indicator of 30-day mortality in ARDS patients, and its predictive discrimination was better than that of DP and MP. Further experimental trials are needed to investigate whether adjusting treatment decisions according to the LHPI will significantly improve clinical outcomes.