Interexaminer reliability of the Johnston and Friedman percussion scan of the thoracic spine: secondary data analysis using modified methods.

OBJECTIVE: The purpose of this study is to perform a secondary analysis using modified methods of previously reported data to analyze the amount of examiner concordance in the Johnston and Friedman percussion scan of the most fixated spinal level.
METHOD: A 2001 study evaluated interexaminer reliability of the percussive method of Johnston and Friedman for detecting altered segmental mobility (somatic dysfunction, spinal/segmental dysfunction, or chiropractic subluxation) in the thoracic spine. The original reported level of agreement using the κ statistic for discrete measures was only 0.07, judged "slight." The data were reformatted to permit recalculating the degree of interexaminer agreement using the intraclass correlation coefficient statistic, which uses continuous analysis, unlike κ that performs discrete analysis. Following an initial calculation, the data were modified to reflect the caudally increasing vertebral height of the thoracic vertebrae.
RESULTS: The reformatted and modified data, intraclass correlation coefficient (2,1) = 0.253 (0.100,0.482), showed the findings as "poor," which is better interexaminer agreement for percussion motion palpation than the original reported κ value judged as "slight."
CONCLUSIONS: Reanalyzing the data using an alternative statistical method showed greater interexaminer reliability than was originally reported. This secondary analysis demonstrates how study results may vary depending on the experimental design and statistical methods chosen for analysis.