Objective: To analyze the concurrent validity of the Digital Image-based Postural Assessment (DIPA) method for identifying the magnitude and classification of thoracic kyphosis in adults.

Methodology: On the same day and in the same place, thoracic kyphosis was assessed in 68 adults using 2 methods: the DIPA software protocol and radiography. The DIPA software provided angular values of thoracic kyphosis based on trigonometric relations, while with the radiograph, the curvature was calculated using the Cobb method. The following tests were applied in the statistical analysis: Pearson's correlation, Bland-Altman's graphic representation, root mean square error, and receiver operating characteristic (ROC) curve; α = 0.05. The reference angular values for the standard thoracic posture used in DIPA were determined with the ROC curve based on the Cobb angles.

Results: The correlation between the angles obtained for thoracic kyphosis using the DIPA and Cobb methods was found to be high (r = 0.813, P < .001), and the accuracy was ±4°. According to Bland-Altman's representation, the magnitudes provided by the DIPA software were in agreement with those of the Cobb method. In reference values for determining the standard posture of the thoracic spine, the ROC curve indicated good accuracy in diagnosing a decrease in thoracic kyphosis (with a value of 33.9°) and excellent accuracy in diagnosing thoracic hyperkyphosis (with a value 39.9°) when using DIPA.

Conclusion: The DIPA postural assessment method is valid in the sagittal plane for identifying the magnitude of thoracic kyphosis in adults. Furthermore, it is accurate in diagnosing alterations in thoracic kyphosis.

Author keywords: Kyphosis, Validity of Results, Photogrammetry

Author affiliations: School of Physical Education, Physiotherapy and Dance, Federal University of the Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil.

This abstract is reproduced with the permission of the publisher; full text is available by subscription. Click on the above link and select a publisher from PubMed's LinkOut feature.