Interrater and intrarater reliability in the measurement of kyphosis in postmenopausal women with osteoporosis

Abstract

Study design: A reliability study was performed using repeated random measurements involving three observers, 26 subjects and three instruments.

Objectives: To determine the most reliable, cost-effective, noninvasive, and clinically feasible method of measuring spinal kyphosis.

Summary of background data: The most clinically useful, noninvasive and reliable method of measuring postural deformity in spinal osteoporosis (kyphosis) remains unqualified. Despite traditional use of costly, invasive roentgenographs for the evaluation of spinal kyphosis, the reliability of this method remains questionable.

Methods: Twenty-six postmenopausal women with known bone mineral density and a diagnosis of osteoporosis were recruited from the Osteoporosis Program at Women's College Hospital, Toronto, Canada. Non-invasive measurements of thoracic kyphosis were obtained by three trained examiners using the DeBrunner's kyphometer and the flexicurve ruler. The intrarater and interrater reliability of and between each method was compared, using roentgenographic films obtained in the sagittal plane. Spinal posture was classified according to the method of Itoi (1990). Statistical computations were performed using SAS statistical software.

Results: Consistent measurements were obtained with the DeBrunner's kyphometer and the flexicurve ruler by each observer, according to the results of critical two-way analysis of variance (Intraclass Correlation Coefficient 2, 1). Measurements in two subgroups, healthy backs (n = 11) and rounded backs (n = 13), showed consistent use of each noninvasive instrument with some examiner preference for specific tools. There was marginally better intrarater and interrater reliability using the DeBrunner's kyphometer compared with that obtained with the flexicurve ruler. Two-way analysis of variance (Intraclass Correlation Coefficient 2, 1) of collapsed data showed no significant difference in the reliability of the kyphometer, flexicurve ruler, or roentgenographs in the measurement of thoracic kyphosis.

Conclusions: The flexicurve ruler and DeBrunner's kyphometer had the closest agreement in the measurement of spinal kyphosis. The kyphometer demonstrated the least variation in intrarater and interrater reliability when compared with the flexicurve ruler and roentgenographs. The flexicurve ruler permits qualitative assessment of posture, however, and is the most cost-effective instrument. The results of this study challenge the traditional belief that roentgenographic analysis is the best method for evaluating spinal kyphosis. The DeBrunner's kyphometer and flexible ruler may represent viable, cost-effective and noninvasive alternatives to roentgenographic evaluation of spinal kyphosis.