Wang H, Liu C, Meng Z, Zhou W, Chen T, Zhang K, Wu A. Real-world study for identifying the predictive factors of surgical intervention and the value of magnetic resonance imaging in patients with low back pain.
Quant Imaging Med Surg 2022;
12:1830-1843. [PMID:
35284262 PMCID:
PMC8899926 DOI:
10.21037/qims-21-584]
[Citation(s) in RCA: 2] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/02/2021] [Accepted: 11/04/2021] [Indexed: 09/18/2023]
Abstract
BACKGROUND
Low back pain (LBP) is a prevalent disease and can be disabling. Currently, many patients with LBP with or without radiculopathy commonly undergo magnetic resonance imaging (MRI) for diagnosis and therapeutic assessment, yet the final intervention is mainly centered around nonoperative treatment. This study's aim was to identify the predictive factors of surgical treatment and the value of MRI in patients with LBP with or without radiculopathy.
METHODS
The study included a training cohort that consisted of 461 patients with MRI from January 2014 to December 2018. Demographic characteristics and MRI findings were collected from our medical records. We developed and validated 2 nomograms to predict the possibility of receiving surgical treatment in LBP patients, based on multivariable logistic regression analysis. The performance of the 2 nomograms was assessed in terms of their calibration, discrimination, and clinical usefulness. An independent validation cohort containing 163 patients was comparatively analyzed.
RESULTS
The baseline model incorporated 6 clinicopathological variables, while the MRI model consisted of 9 variables including several MRI findings. Internal validation revealed the good performance of the 2 nomograms in discrimination and calibration, with a concordance index (C-index) of 0.799 (95% CI: 0.743-0.855) for the baseline model and 0.834 (95% CI: 0.783-0.884) for the MRI model, which showed that the addition of MRI findings to the nomogram failed to achieve better prognostic value (Z statistic =-1.509; P=0.131). Application of the 2 models in the validation cohort also showed good discrimination (baseline model: C-index 0.75, 95% CI: 0.671-0.829; MRI model: C-index 0.777, 95% CI: 0.696-0.857) and calibration. No significant predictive benefit was found in the MRI model in the validation cohort (Z statistic =-0.588; P=0.557).
CONCLUSIONS
This study showed that clinical demographic characteristics provide good prognostic value to determine whether LBP patients with or without radiculopathy require surgical treatment. The addition of MRI findings yielded no significantly incremental prognostic value.
Collapse