Intraoperative Hemodynamic Trajectories and Postoperative Acute Kidney Injury After Total Knee Arthroplasty

Background and Objectives: Conventional intraoperative hypotension measures may not capture temporal blood pressure patterns. We aimed to identify data-derived intraoperative MAP trajectory patterns and exploratorily characterize their relationship with postoperative acute kidney injury (AKI) after elective total knee arthroplasty (TKA). Materials and Methods: This retrospective cohort study used the INSPIRE perioperative database. Intraoperative MAP measurements were normalized to anesthesia duration, interpolated to 21 time points, centered to early MAP, and classified using outcome-independent K-means clustering. AKI was defined using KDIGO serum creatinine criteria. AKI frequencies were compared descriptively, and exploratory odds ratios were estimated using Firth penalized logistic regression. Results: Among 855 eligible procedures, 716 met trajectory completeness criteria, and 713 were evaluable for AKI. Two data-derived trajectory patterns were identified: lesser relative decline (320/716, 44.7%) and greater relative decline (396/716, 55.3%). The greater-relative-decline phenotype had a higher initial MAP but lower median and minimum MAP, more frequent MAP < 65 mmHg, and a nearly twofold greater maximum MAP decline. AKI occurred in 9/713 procedures (1.3%): 1/320 (0.3%) in the lesser-relative-decline group and 8/393 (2.0%) in the greater-relative-decline group. The absolute risk difference was 1.72 percentage points (95% CI, −0.03 to 3.67). The unadjusted Firth odds ratio was 4.68 (95% CI, 0.82–26.79). The between-group comparison was descriptive, and the regression estimate was highly imprecise; these findings were considered exploratory and hypothesis-generating. Conclusions: Two data-derived relative MAP trajectory patterns were identified in patients undergoing elective TKA. Most AKI events occurred in the greater-relative-decline group, but this was a descriptive observation, and the regression estimates were highly imprecise. However, given the small number of AKI events, modest cluster separation, and sensitivity of patient-level classification to trajectory representation, these findings should be considered exploratory and require validation in larger external cohorts.