A variability study of computerized sagittal spinopelvic radiologic measurements of trunk balance

The accurate measurement of spinal and pelvic alignment in the sagittal plane is of prime importance for various disorders. Pelvic incidence (PI) is a fundamental anatomic parameter that is specific and constant for each adult individual and is related to pelvic orientation as well as to the size of lumbar lordosis (LL). It is the summation of the sacral slope (SS) and pelvic tilt (PT), two position-dependent variables that determine pelvic orientation in the sagittal plane. The authors have proposed a computer software designed to measure PI, SS, PT, LL, and thoracic kyphosis (TK) on standardized standing lateral digitized x-rays of the spine and pelvis. The purpose of this study was to evaluate the inter- and intraobserver variability of measurements using this software, to determine if it can be used reliably in a clinical environment.

The standing lateral x-rays of 30 subjects were randomly selected from the database of two medical institutions. The normal population had standard radiographs on which the various pertinent landmarks were marked by one operator prior to digitization, whereas the scoliotic population had digital radiographs that obviated the need for prior marking of landmarks. Four individuals measured all variables on the 30 x-rays on two occasions, with a 15-day interval between the two sessions. Statistical analysis was done with intraclass correlation coefficients (ICCs).

The ICC measured within observers was between 0.93 and 0.99, whereas the ICC between observers varied between 0.92 and 0.99. The variations observed were similar for normal and scoliotic subjects, and prior marking of the x-rays had no significant influence.

We conclude that the variability of measurements with this method is lower than with similar radiologic measures done manually and that the use of this software can be recommended for future clinical and research studies of spinopelvic sagittal balance.