Development And Validation Of A Scoring Model For Lumbar Spinal Stenosis Based On Logistic Regression Analysis

Objectives:To analyze the independent risk factors of lumbar spinal stenosis and construct a diagnostic scoring model to provide clinicians with a foundation for the early diagnosis and treatment of lumbar spinal stenosis.Methods:The clinical data of patients with lumbar spinal stenosis diagnosed by lumbar magnetic resonance examination from September 2019 to December2022 were collected and 336 subjects with a total of 864 lumbar segments were included.Two senior radiologists measured and evaluated the MRI images to collect imaging indices that may be associated with lumbar spinal stenosis.The data were randomly divided into a modeling group and a validation group in a7:3 ratio.Data from the modeling group were subjected to univariate Logistic regression analysis,and statistically significant variables were selected for multivariate Logistic regression analysis to establish a model,which was converted into a scoring scale based on the regression coefficient.The diagnostic scoring model was applied to score the two groups separately,and the diagnostic efficacy of the model was evaluated by calculating the area under the subject’s working characteristic(ROC)curve(AUC)and the HosmerLemeshow(H-L)test.A difference was considered statistically significant at P<0.05.Results:1.The 864 lumbar spine segments were randomly divided into 2 groups according to 7:3.70% of the 864 lumbar spine segments(N=604)were assigned to the modeling group,and the remaining 260 segments were included in the validation group.Clinicians made a comprehensive diagnosis of the 864 lumbar spine segments by clinical symptoms,signs,and imaging,and diagnosed 776 cases of lumbar spinal stenosis and 88 cases of undiagnosed lumbar spinal stenosis.2.Univariate Logistic regression analysis showed that 14 independent variables were age,sagittal diameter of spinal canal,transverse diameter of spinal canal,lumbar instability,lumbar spondylolisthesis,hypertrophy of facet joint,cross-sectional area of spinal canal,cauda equina nerve sedimentation sign,cross-sectional area of dural sac,degree of anterior cerebrospinal fluid spatial occlusion,Schizas classification,cauda equina redundancy sign,cauda equina nerve settlement sign,numbness and intermittent claudication(P < 0.05).3.The results of multivariate Logistic regression analysis showed that age,cross-sectional area of spinal canal,cross-sectional area of dural sac,lumbar spondylolisthesis,facet joint hypertrophy,numbness,intermittent claudication and cauda equina redundancy sign were independent risk factors.The corresponding scores of independent risk factors were 1,4,2(each grade),4,6,4,3,6,4 points.The area under the ROC curve(AUC)of the modeling group and the verification group using the scoring model is 0.913(95% CI = 0.887±0.935,P < 0.001)and 0.922(95% CI = 0.883 ±0.952,P < 0.001).The results of Hmurl test were 0.343 and 0.352 respectively(P > 0.05).The established scoring model has good diagnostic efficiency.Conclusions:The diagnostic scoring model of lumbar spinal stenosis is simple and easy,which is helpful for early diagnosis of lumbar spinal stenosis and optimizing the prognosis of patients.