Predicting the necessity of routine postoperative laboratory tests after primary total hip arthroplasty.

Abstract

We aimed to develop a predictive model for abnormal postoperative laboratory tests in patients undergoing primary total hip arthroplasty (THA) to aid clinical decision-making. We analyzed 4,000 patients undergoing primary THA from 1 January 2016 to 31 December 2020, excluding 1,773 due to incomplete records and 63 for being one-stage bilateral THAs. The final cohort comprised 2,164 patients (60.55% female; mean age 70 years (SD 13)). Abnormal laboratory results were defined as values requiring medical intervention. Univariate analyses were performed, followed by model generation with a generation cohort (n = 1,499) and a validation cohort (n = 665). The model's discriminatory performance was measured using the time-dependent area under the receiver operating characteristic (AUROC) curve. The calibration-in-the-large (CITL) was calculated as the logistic regression model intercept. A score was created calculating each point's sensitivity and specificity to determine the best cut-off value. Variables such as American Society of Anesthesiologists (ASA) grade III to IV, diabetes, and preoperatively altered potassium, sodium, blood glucose, and serum creatinine were statistically significant predictors (all p < 0.05) of postoperative abnormal laboratory tests. The model had AUROCs of 0.74 and 0.71 in the generation and validation cohorts, respectively. The Cox calibration had a slope of 1 (95% CI 0.86 to 1.13) and a CITL of 0 (95% CI -0.14 to 0.14) in the generation cohort, and a slope of 1 (95% CI 0.77 to 1.22) and a CITL of 0 (95% CI -0.024 to 0.24) in the validation cohort. A scoring system was developed (AUROC, 0.71; slope, 1; CITL, 0) to assess risk based on these variables. A score of 1.5 yielded 58.22% sensitivity and 76.71% specificity for predicting laboratory abnormalities. The proposed model accurately predicted abnormal postoperative laboratory findings in THA patients, highlighting the significance of specific preoperative factors together. It seems appropriate to recommend a postoperative laboratory test whenever a patient's score is ≥ 1.5 points.