Abstract: BACKGROUND: Current studies only predict the loss of cervical lordosis after cervical surgery through imaging measurement indicators and other indicators at present. There are a few articles summarizing these prediction indicators. This paper establishes a prediction model to summarize these prediction indicators.
OBJECTIVE: To investigate the risk variables for severe loss of cervical lordosis following posterior cervical spondylotic myelopathy surgery, as well as to develop and validate the prediction model. 
METHODS: Retrospective analysis was performed on the cervical spondylotic myelopathy patients who underwent posterior approach of cervical surgery in the Affiliated Hospital of Xuzhou Medical University from January 2015 to January 2020 and met the inclusion criteria. The observation indexes included age, sex, body mass index, surgical technique chosen, the number of operation segments, accumulation of C2 or C7, C2-7 Cobb angle prior to operation, Cobb angle of operation segment, C7 slope angle, sagittal vertical angle of the cervical spine, C2-C7 curvature, extension range of motion, and flexion range of motion. The difference between the cervical spine’s C2-7 Cobb angle before and after surgery (ΔCL) was used to calculate the change in cervical lordosis. Those with ΔCL≤-10° had significant loss of cervical lordosis, while those with ΔCL > -10° had less severe loss of cervical lordosis. Prediction models were created and validated by doing single-factor and multi-factor analyses on these parameters to identify pertinent risk factors. 
RESULTS AND CONCLUSION: 117 patients in all, 48 females and 69 males, met the inclusion criteria. The follow-up time ranged from 12 to 26 months. Among these patients, 30 experienced a severe loss of cervical lordosis following surgery, while 87 patients did not have a severe loss of cervical lordosis. Statistical analysis showed that the choice of procedure, whether it involved the C2 or C7 vertebral bodies, the C2-7 Cobb angle, the C7 slope angle, the C2-C7 curvature, and flexion range of motion prior to the procedure were independent risk factors linked to serious loss of cervical lordosis following posterior cervical surgery. Most obviously, whether the surgical segment involved the C2 or C7 segment (OR=3.524, 95% CI:1.127-11.013), and the surgical approach chosen (OR=3.165, 95% CI: 1.013-9.889) were the factors that enhanced the probability of significant postoperative curvature loss. Further foundations were laid for the clinical prediction model (nomogram) and its validation. The model has an excellent capacity for prediction, as evidenced by the C-index of internal validation, which was 0.91, and the C-index of external validation in the validation group, which was 0.87. It is indicated that after posterior cervical surgery, the choice of operation method, whether the operation segment involves the C2 or C7 segment, the preoperative C2-7 Cobb angle, the C7 slope angle, and flexion mobility all are high risks of severe loss of cervical lordosis. 

Key words: multi-segment cervical spondylotic myelopathy, cervical lordosis, posterior cervical surgery, clinical prediction model, nomogram