J Manipulative Physiol Ther. 1998 Sep; 21(7): 460-7.

Further analysis of the reliability of the posterior tangent lateral lumbar radiographic mensuration procedure: concurrent validity of computer-aided X-ray digitization.

Troyanovich SJ, Harrison DE, Harrison DD, Holland B, Janik TJ.

Department of Statistics, Temple University, Philadelphia, PA, USA.

OBJECTIVE: To investigate the reliability of a specific method of radiographic analysis of the geometric configuration of the lumbopelvic spine in the sagittal plane, and to investigate the concurrent validity of a computer-aided digitization procedure designed to replace the more tedious and time-consuming manual measurement process. DESIGN: A blind, repeated-measures design was used. The results of radiographic measures derived through the traditional manual marking method were compared with measures derived by computer-aided digitization of lateral lumbopelvic radiographs. SETTING: Private chiropractic clinic. MAIN OUTCOME MEASURES: Pearson's product-moment correlation coefficients, paired sample t tests and intraclass correlation co-efficients (ICC) were used to examine intraexaminer reliability, and repeated measures of analysis of variance were used to examine interexaminer reliability for relative rotation angles for T12-L1, L1-L2, L2-L3, L3-L4, L4-L5, L5-S1, overall lordosis measurement [absolute rotation angle (ARA)] from L1-L5 and Cobb angle of overall lordosis measured from the inferior surface of T12 to the superior surface of S1, Ferguson's sacral base angle to horizontal, angle of pelvic tilt (arcuate angle) to horizontal and anteroposterior thoracic translation (Sz) in millimeters. RESULTS: ICC estimates for intraexaminer reliability were in the range of 0.96-0.98 for the L1-L5 ARA, a range of 0.87-0.99 for the arcuate angle measurement, 0.83-0.94 for the Ferguson's angle measurement, 0.88-0.95 for the Cobb angle measurement from the inferior surface of T12 compared with the superior surface of S1 and 0.98-1.00 for the translation measurement of the lower thoracic spine to S1 (Sz). The intersegmental measurement's (T12-L1, L1-L2, L2-L3, L3-L4, L4-L5, L5-S1) correlations ranged from a low of 0.55 to a high of 0.97. Examination of these findings suggests that the reliability for the three doctors is acceptable with only the T12-L1 intersegmental measure falling below 0.70 for the least experienced examiner. Average ICC of interexaminer reliability for manual and computer-aided digitizing examiners were the following: 0.96 for the L1-L5 ARA; 0.84 for the arcuate angle measurement; 0.82 for the Ferguson's angle measurement; 0.88 for the Cobb angle measurement; 1.00 for the Sz translation measurement; and values of 0.65, 0.73, 0.74, 0.75, 0.89 and 0.81 for relative rotation angle measurements T12-L1, L1-L2, L2-L3, L3-L4, L4-L5 and L5-S1, respectively. CONCLUSION: The data tend to support the reliability of this method of radiographic analysis of the geometric configuration of the lumbopelvic spine as viewed on lateral lumbopelvic radiographs. The additional data presented here tend to support the concurrent validity of the computer-aided digitization method of analysis inasmuch as the measures determined by the digitizing examiners are essentially identical to those determined by the manual method plus or minus the average standard error of measure of each value.