Related Experiment Video
Updated: Jul 12, 2025
Minimally Invasive Treatment for Thoracolumbar Burst Fracture Using Sagittal Alignment Screws and A Trauma Reduction Device
Published on: November 8, 2024
Addressing the Challenge of Spine Patient Triage: Development of a Simple Algorithm for Identification of Potential
Justin J Turcotte1, Jane C Brennan, Andrea H Johnson
1From the Department of Orthopaedics, Luminis Health Anne Arundel Medical Center, Annapolis, MD.
Introduction:
Given that most spine conditions do not require surgical intervention, using surgeons to manage the subset of patients potentially requiring surgery is the most efficient resource allocation strategy. The purpose of this study was to develop a simple algorithm for identifying patients most likely to require spine surgery that could be used to appropriately triage this population to surgeons.
Methods:
A retrospective review of 5,886 consecutive new patients presenting to a multidisciplinary spine clinic from March 2021 to September 2022 was conducted. The primary outcome was whether a patient underwent spine surgery during the study period. A total of 64 independent variables were recorded from patient intake and the first visit. A gradient boosted model was generated to identify the independent variables most associated with undergoing surgery. The five most important variables were entered into a multiple logistic regression model, and a simplified decision support tool was generated and assessed.
Results:
Overall, 440 of 5886 patients (7.5%) underwent surgery during the study period. The following variables were identified as the top five predictors of spine surgery: patient goal of interest in learning about spine surgery, history of spine injections, difficulty walking a mile, radicular symptoms greater than axial symptoms, and increased age. Each of these variables was confirmed to be independently associated with undergoing surgery (all P < 0.001). The decision support tool yielded a sensitivity of 60.0%, specificity of 76.6%, likelihood ratio of 2.56, positive predictive value of 17.2%, and negative predictive value of 96.0% for predicting surgery. An AUC of 0.683 was achieved.
Conclusion:
A simple 5-question algorithm incorporating patient demographics, symptoms, treatment history, physical function, and patient goals may improve the ability of practices to identify potential spine surgery candidates before their first visit. Prospective application and evaluation of the algorithm to evaluate whether it improves the triage of appropriate patients to spine surgeons is warranted.
