Study design: Prospective comparative study of measuring pelvic incidence (PI) among standing radiographs of whole spine and pelvis and computed tomographic (CT) scans in a cohort of patients.
Objective: To analyze accuracies in measuring PI and other spinopelvic parameters.
Summary of background data: Previous reports indicated relatively low agreement in measuring PI even among experienced spinal surgeons; intra- and inter-rater reliability in manually measuring PI were 0.69 (0.62-0.74) and 0.41 (0.36-0.45), respectively; the mean interclass correlation coefficient value of manually measuring PI was 0.881. No study compared PI on standing radiographs with that measured on CT scans.
Methods: A total of 120 consecutive patients with spinal disease (38 patients had history of hip arthroplasty) who admitted to our hospital from April 2012 for 6 months were enrolled. Subjects had obtained full-spine lateral standing radiograph, standing radiograph of pelvis, and CT scans. Pelvic incidence on full-spine lateral standing radiograph and that on pelvis lateral standing radiograph were measured manually by 2 experienced spinal surgeons. Intra- and interobserver reliability of the measurements were analyzed by using interclass correlation coefficient. On CT scans, PI was measured using 3-dimensional CT scan software (CT-PI). PI among 3 different imaging modalities was evaluated using correlation coefficients.
Results: In whole-spine radiographs, the intra- and interobserver agreement rates with measurements in PI (0.84 and 0.79, respectively) and sacral slope (0.87 and 0.83, respectively) were lower than those in pelvic tilt (0.98 and 0.96, respectively) and PI-lumbar lordosis (0.97 and 0.97, respectively). The correlation coefficient between P-PI and CT-PI was higher (0.95) than that between FS-PI and CT-PI (0.81) and between FS-PI and P-PI (0.85).
Conclusion: The reliability of measuring PI is comparatively lower than that of other spinopelvic parameters, and the variability of PI measurement is mainly due to difficulty of precisely identifying sacral endplate.
Level of evidence: 2.