The reliability of lumbar motion palpation using continuous analysis and confidence ratings: choosing a relevant index of agreement.

INTRODUCTION Most studies show motion palpation unreliable. This study's primary objective was assessing its reliability using a continuous measure methods, most-fixated level paradigm, stratified by examiners' confidence; and the secondary objective was comparing various indices of examiner agreement. METHODS Thirty-four minimally symptomatic participants were palpated in side posture by two experienced examiners. Interexaminer differences in identifying the most-fixated level and degree of examiner confidence were recorded. Indices of agreement were: Intraclass correlation coefficient, Mean and Median Examiner Absolute Examiner Differences, Root-Mean-Square Error and Bland-Altman Limits of Agreement. RESULTS Three of four reliability indices (excluding intraclass correlation) suggested on average examiners agreed on the most fixated motion segment, and agreement increased with confidence. Statistical measures of data dispersion were low. The analyses of subgroups were "fragile" due to small sample size. DISCUSSION Although subject homogeneity lowered ICC levels, the other reliability measures were not similarly impacted. Continuous measures statistical analysis demonstrates examiner agreement in situations where discrete analysis with kappa may not. CONCLUSION Continuous analysis for the lumbar most-fixated level is reliable. Future studies will need a larger sample size to properly analyze subgroups based on examiner confidence.