Phase angle as a potential predictor for unplanned ICU readmission risk in critically ill surgical patients: a retrospective cohort study

Background: Surgical intensive care unit (ICU) readmission worsens outcomes; however, prediction remains difficult. Phase angle (PhA), a bioimpedance-derived marker of cellular integrity and nutritional status, may add prognostic value beyond routine clinical indices. Materials and Methods: We conducted a single-center retrospective cohort of ICU discharge-level observations (510 discharges) from a surgical ICU. PhA was measured by bioelectrical impedance analysis at ICU admission (PhA 0 ) and discharge (PhA dc ). The primary analysis used a mixed-effects logistic regression with a patient-level random intercept to account for multiple discharges per patient. Pre-specified cutoffs (PhA 0 ≤ 3.75°, PhA dc ≤ 2.55°) were derived by internal receiver operating characteristic/Youden methods and assessed with patient-wise fivefold cross-validation. Sensitivity analyses included stepwise multivariable models, first-discharge-only, cluster-robust standard errors, and a reduced generalized estimating equation. Results: Of 510 discharges, 68 (13.3%) were followed by unplanned ICU readmission. Low PhA 0 and low PhA dc were associated with readmission in bivariable analyses. In the primary mixed-effects model, low PhA 0 independently predicted readmission [adjusted odds ratio (aOR) = 4.00, 95% confidence interval (CI) 1.82–8.78; P < 0.001], while low PhA dc showed a positive trend (aOR = 1.80, 0.95–3.42; P = 0.07). Emergency surgery was also independently associated (aOR = 4.57, 2.00–10.40; P < 0.001). Findings were directionally consistent across sensitivity analyses. Conclusion: PhA measured at ICU admission and discharge provides an actionable, noninvasive prognostic indicator for surgical ICU readmission.