Deriving a Novel Score for the Stratification of Risk Progression in Early-onset Scoliosis: A Multicenter Initiative

This was a retrospective multicenter study.To develop a novel progression risk stratification scoring system for early-onset scoliosis.There is a lack of investigations into variables affecting the risk of curve progression in early-onset scoliosis, which prevents stratification. A novel risk score system is needed to help in progression risk estimation.A retrospective analysis was done at three centers, from 1995 to 2020. Scoliosis cases before the age of 10 years, were included. Medical identifier, date of birth, sex, primary diagnosis, curve type, date/modality of treatment, date of follow-up appointments, and Cobb angles, were collected. Five ranks were selected for stratification. Categories with the same ranks were discarded. Point scores started at 0, for the lowest risk, and ended at 4, for highest risk. Iterations of variable combinations were conducted and clinical relevance was determined by evaluating sensitivity, specificity, positive predictive value, and negative predictive value based on score ranges for low and high risk of progression.A total of 476 (230 males, 246 females) early-onset scoliosis patients were collected. The average age at diagnosis was 4.8 years (SD±2.8 yr). The average follow-up duration was 9.3 years (SD±6.9 yr, range: 5 mo-38 yr). Appointments totaled 2911, giving 2182 observations for the analysis. Patient observations numbered: 800 (36.7%) ending in progression, 1265 (58.0%) for nonprogression, 117 (5.4%) for inadequate follow-up, and 368 (16.9%) for rapid progression. The risk scoring system contained four categories: etiology, age, curve magnitude, and curve type. Categorized point combinations totaled 755, giving 1975 iterations. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated to be 85.8%, 96.5%, 89.7%, and 95.1%, respectively.A novel progression risk score for early-onset scoliosis was derived. The system can reliably differentiate between low and high-risk cases in clinical settings. Further validation in other regions may be important for verifying clinical relevance.Level 3.