BACKGROUND The potential clinical value of driving pressure (DP) and mechanical power (MP) as digital biomarkers for predicting postoperative pulmonary complications (PPC) has been emphasized. Although several studies have explored this association, evidence from clinical cohorts involving large patient populations remains limited.

METHODS A retrospective cohort study was conducted between October 2004 and May 2023 on patients who underwent OLV surgery. The association between time-weighted median dynamic DP or MP and PPC was analyzed using multivariate logistic regression models adjusted for confounders. Additionally, risk threshold analysis was conducted to propose thresholds for an increased risk of PPC.

RESULTS Among the 3386 (using plateau pressure; Pplat) or 4951 (using peak inspiratory pressure; PIP) patients included, PPC occurrence was 19.31 % and 17.43 %, respectively. The mean of the time-weighted median DP derived from Pplat (DP-Pplat) and MP derived from Pplat (MP-Pplat) were 14.6 cmH₂O and 7.1 J/min, while DP-PIP and MP-PIP were 17.2 cmH₂O and 6.4 J/min. Both calculation methods showed that DP had an independent association with the occurrence of PPC, with an odds ratio of 1.047 [95 % CI 1.019-1.075, p < 0.05] and 1.036 [95 % CI 1.013-1.059, p < 0.05] using Pplat and PIP, respectively. However, MP was not found to be independently associated with PPC using either method, with an odds ratio of 1.033 [95 % CI 0.980-1.089, p = 0.226] and 1.048 [95 % CI 0.992-1.106, p = 0.092] using Pplat and PIP, respectively. The risk threshold for DP-Pplat was 15 cmH₂O, whereas for DP-PIP, it was 18 cmH₂O.

CONCLUSIONS In this OLV surgery population, a DP-Pplat-limited mechanical ventilation strategy of 15 cmH₂O or DP-PIP of 18 cmH₂O was associated with lower risk of PPC.