Identifying scoliosis in population-based cohorts: development and validation of a novel method based on total-body dual-energy x-ray absorptiometric scans.

The purpose of this study was to develop and validate a novel method of identifying scoliosis on total-body dual energy X-ray absorptiometric (DXA) scans. Scoliosis was identified on total-body DXA scans by triaging to distinguish true curves from positioning errors, followed by a modified Ferguson method to measure angles. Precision was assessed on 174 children from the Avon Longitudinal Study of Parents and Children (ALSPAC), who underwent repeat DXA scans at age 15, 2-6 weeks apart. In addition, precision of angle estimation was evaluated on 20 scans measured five times. To evaluate accuracy, angle size was compared to spinal radiographs in 13 individuals with known scoliosis. Subsequently, this method was applied to estimate scoliosis prevalence rates and curve patterns from DXA scans previously obtained in 7,298 ALSPAC participants at age 9 and 5,122 at age 15. There was substantial agreement in identifying those with scoliosis on repeat DXA scans taken 2-6 weeks apart (kappa 0.74, 95 % CI 0.59-0.89). Of repeat angle measures, 95 % were within 5°. Angle size was underestimated by approximately 40 %. Prevalence of scoliosis ≥10° in the ALSPAC was 0.3 % at age 9 and 3.5 % at age 15 and was higher in girls at both time points. The mean ± SD curve size was 12 ± 4° at age 9 years and 15 ± 7° at age 15. We have developed and validated a novel method for identifying scoliosis from DXA scans. Comparison with prevalence data using more established techniques suggests our method provides valid estimates of scoliosis prevalence in population-based cohorts.