Construction of a predictive model for postoperative intervertebral space infection following lumbar disc hernia-tion surgery based on propensity score matching
MA Yuan
TANG Li-xin
YANG Guo-zhi
LIANG Yu-zhu
SONG Chun-xia
PENG Xiao-he
SUN Xiu-qin
LI Sen
WANG Hai-yu
Abstract：Objective To analyze the risk factors for postoperative intervertebral space infection(ISI)following lumbar disc herniation(LDH)surgery using propensity score matching(PSM)and to develop a risk prediction model.Methods This retrospective study included 1 527 patients who underwent LDH surgery at Nanyang Central Hospital from September 2021 to September 2024.Among them,48 patients developed postoperative ISI,yielding an incidence rate of 3.14％.Based on sex,disease duration,length of hospital stay,and American Society of Anesthesiologists(ASA)classi-fication,PSM was performed at a 1∶1 ratio,resulting in 45 matched pairs.Clinical data were collected to identify inde-pendent risk factors for postoperative ISI.A risk prediction model was constructed based on these factors,and its perform-ance was evaluated using the receiver operating characteristic(ROC)curve.Results Univariate analysis revealed signif-icant differences between groups in age,diabetes status,body mass index(BMI),operative time,intraoperative blood loss,serum albumin(ALB),and procalcitonin(PCT)levels(P＜0.05).Multivariate logistic regression identified oper-ative time,intraoperative blood loss,ALB,and PCT as independent risk factors for ISI(P＜0.05).The predictive model built using these four variables achieved an area under the ROC curve(AUC)of 0.978(95％CI:0.953-1.000),with a sensitivity of 88.9％and specificity of 97.8％.Conclusion Operative time,intraoperative blood loss,serum ALB,and PCT levels are independently associated with postoperative intervertebral space infection in LDH patients.The devel-oped predictive model demonstrates high accuracy and may provide clinical value for early identification and prevention.
Keywords：lumbar disc herniationintervertebral space infectionpropensity score matchingprediction model
Publication Date：2025-07-15
Online Publishing Date：2025-08-25(First online date of this platform, not the publication date of the document)
Pages：6( 1067-1072 )