Perioperative hemodynamic disturbances as potential statistical intermediate pathways in the association between preoperative pulmonary hypertension and acute kidney injury after cardiopulmonary bypass-assisted cardiac surgery: a retrospective cohort study

Patients with preoperative pulmonary hypertension (PH) often face a higher risk of acute kidney injury (AKI) following cardiac surgery, but the potential perioperative intermediate pathways remain insufficiently characterized. We aimed to investigate whether and to what extent intraoperative hypotension (IOH) and support with vasoactive-inotropic agents were statistically consistent with mediation of the observed association between PH and AKI. This retrospective cohort study enrolled adult patients undergoing cardiac surgery with cardiopulmonary bypass at Nanjing First Hospital (June 2018–June 2023). Intraoperative blood pressure data were extracted at 5-min intervals. At four prespecified mean arterial pressure (MAP) thresholds (< 65/60/55/50 mmHg), cumulative duration below the threshold (TIME) and the corresponding area between the threshold and the MAP curve (AUT) were calculated. Vasoactive-inotropic score (VIS) was assessed intraoperatively and within the first 24 h postoperatively. The bootstrap mediation analysis using continuous mediators was the primary inferential framework. Baron–Kenny analysis and receiver operating characteristic (ROC)-based dichotomization of the mediators were additionally performed as exploratory supplementary analyses. Missing data were imputed using k-nearest neighbors, and a complete-case sensitivity analysis assessed robustness to missing-data handling. Among 2242 patients, 751 (33.5%) developed AKI. In the prespecified bootstrap models, postoperative VIS showed the largest estimated indirect association (estimated proportion mediated, 23.9% when PASP was modeled continuously and 27.6% when severe PH was modeled as a binary exposure), followed by MAP_TIME_65 (5.0% and 5.3%, respectively). No statistically significant indirect association was observed for intraoperative VIS (1.9%; P = 0.120). In the complete-case analysis ( n = 1981), results for MAP_TIME_65 and postoperative VIS were consistent with the primary analysis. Among patients with PH undergoing cardiac surgery, postoperative vasopressor-inotropic requirements and intraoperative hypotension represent statistical intermediate factors associated with AKI risk. These findings may inform future research on a bundled, phase-specific hemodynamic management strategy spanning the entire perioperative continuum.