The assessment of intraobserver and interobserver error in the measurement of noncongenital scoliosis in children < or = 10 years of age.

Abstract

Retrospective review of scoliosis radiographs. To determine measurement variability in children < or = 10 years of age with noncongenital scoliosis. Measurement variability in congenital and adolescent idiopathic scoliosis has been studied. There is no study of measurement variability in young children with noncongenital scoliosis. A retrospective review of children < or = 10 years of age followed for noncongenital scoliosis was performed. End vertebrae were identified on radiographs, and the curves were measured (Cobb method) twice by each of six observers. The same soft lead pencil and goniometer was used. Intraobserver and interobserver variability for continuous data was determined. There were 64 children. The diagnosis was infantile/juvenile idiopathic scoliosis in 42, neuromuscular scoliosis in 7, scoliosis associated with mesenchymal disorders or other syndromes in 12, and unknown in 3 children. The curve was thoracic in 54, thoracolumbar in 8, and lumbar in 2. There were 19 left and 45 right curves. The average age was 6.6 +/- 2.6 years. There were a total of 768 Cobb angle measurements with an average Cobb angle of 38 +/- 22 degrees (range, 10 degrees -115 degrees ). Intraobserver variability was +/- 6 degrees; interobserver variability was +/- 7 degrees. In children < or =10 years of age with noncongenital scoliosis, intraobserver measurement variability in Cobb angle measurement is +/- 6 degrees and interobserver variability is +/-7 degrees. To be certain that there is a significant difference between Cobb angle measurements in children with noncongenital scoliosis and < or = 10 years of age there must be a change of at least +/-7 degrees.