Abstract
Objective: To assess the intra- and inter-observer reproducibility of a number of commonly used radiological measurements in pre- and post-operative patients with thoracic adolescent idiopathic scoliosis (AIS). Reproducibility of measures other than Cobb angle and vertebral rotation have not been studied and particularly there are no reports of reproducibility in patients after instrumentation.
Design: Repeat measurement of radiographs before and after surgery by 2 observers.
Subjects: 30 patients with thoracic AIS were selected from a scoliosis database at random: 15 treated with posterior USS and 15 with anterior instrumentation (8 Zielke / 7 anterior USS).
Outcome measures: The pre-operative AP radiograph, supine lateral bending radiograph and the post-operative AP radiograph at 6 months were selected for each patient. Two observers (MM beginner, AAC experienced) obtained the following measurements from the radiographs: Cobb angle, apical vertebral rotation (AVR, Perdriolle), apical vertebral translation (AVT) to the T1-S1 line, and frontal plane imbalance (FPI). With all marks removed, the radiographs were re-measured by each observer at least one week later. Repeatability was calculated using the method described by Bland and Altman (BMJ 1996). This method is a widely accepted anthropometrical technique but has not previously been used for assessing scoliosis measurements. It was assessed as 95% reproducibility. The co-efficient of reliability (r) expresses the proportion of the observed variability that is not due to error, i.e. higher is better. This was calculated as a means of assessing the usefulness of our measurements and to enable us to compare them.
Results: Intra-observer repeatability (MM vs. AAC): Whether the instrumentation was anterior or posterior had no effect on Cobb angle, AVT or FPI repeatability. AVR however was worse for posterior instrumentation 19° vs. 12°. “r” was > 90% for Cobb angle, AVT and PFI. But, for AVR r measured pre-op 52-92% and post-op 38–69%.
There was no relationship between repeatability and the measurement size.
Conclusions: Measurement reproducibility / error is slightly worse than previously suspected. E.g. a 5–6° curve progression is thought to be significant. We suggest that this could be due to measurement error and the figure should be 6–8°. There is no learning curve for the technique used to measure Cobb angle, AVT and FPI. AVR (Perdriolle) however requires experience. Cobb angle measurement error post-op is similar to pre-op. The Perdriolle method has greater error post-op especially in posterior instrumentation.
The abstracts were prepared by Mr Peter Millner. Correspondence should be addressed to Peter Millner, Consultant Spinal Surgeon, Orthopaedic Surgery, Chancellor Wing, Ward 28 Office Suite, St James’ University Hospital, Beckett Street, Leeds LS9 7TF.
