Acute renal dysfunction is defined by the maximum increase in serum creatinine (sCr) without considering the pattern of change in sCr. We aimed to identify longitudinal patterns (trajectories) of postoperative sCr concentrations and investigate their association with long-term outcomes in patients undergoing valvular heart surgery. In this retrospective review of 3436 patients who underwent valvular heart surgery, we applied trajectory projection cluster analysis to identify the trajectories of sCr changes from baseline during the 7 postoperative days. Primary and secondary endpoints were to investigate the associations of sCr trajectories with mortality using Kaplan-Meier curves and Cox proportional hazards regression analysis and a composite of major adverse kidney events (MAKEs) at 1 year after surgery, respectively. Four clusters were identified: Clusters 1 and 2, a minimal change in sCr (90.1% of patients); Cluster 3, a significant and persistent increase in sCr (4.1% of patients); and Cluster 4, a significant but transient increase in sCr (5.8% of patients). The 1-year postoperative mortality rate was higher in Cluster 3 (18.4%) and Cluster 4 (11.6%) than in Cluster 1+2 (2.7%). The Kaplan-Meier survival curve demonstrated significant differences in mortality rates among the clusters (log-rank test, P <0.001). In the multivariable Cox analysis, the sCr trajectory cluster was an independent prognostic factor for mortality. Cluster 3 had a higher prevalence of MAKEs (37.6%) compared with Cluster 1+2 (6.8%, P <0.001) and Cluster 4 (24.1%, P =0.045). The cluster was an independent prognostic factor for MAKEs. The sCr trajectory clusters exhibited significantly different risks of mortality and MAKEs 1 year after surgery. Through these sCr trajectories, we confirmed that both the extent of sCr increase and its sustainability during the first 7 postoperative days were closely associated with the long-term prognosis after valvular heart surgery.
Read full abstract