The accuracy of clinical kyphosis examination for detection of thoracic vertebral fractures: comparison of direct and indirect kyphosis measures.

OBJECTIVE To compare the accuracies of two simple physical examination maneuvers for detecting the presence of thoracic vertebral fractures (VF) diagnosed by radiography: direct measurement of kyphosis angle (KA, in degrees) and indirect measurement using wall-occiput distance (WOD, in cm). METHODS Subjects were 280 women (average age, 54.5 years; range, 18-92) referred for assessment of osteoporosis. KA was measured from T4 to T12 using a digital inclinometer while WOD was measured with the patient in a standardized position. VF were diagnosed on radiographs using semi-quantitative morphometry. RESULTS KA and WOD were moderately correlated (r = 0.72, p<10(-11)). KA increased by 3.7(o) (95% CI, 2.6-4.8(o)) for each VF (p = 4x 10(-11)) and WOD rose 1.3 cm (95% CI, 0.8-1.7 cm) per VF (p = 2 x 10(-11)). The areas under the receiver operating characteristic curves were 0.72 (95% CI, 0.65-0.79) for KA and 0.76 (95% CI, 0.69-0.82) for WOD, which were not significantly different (p = 0.13). CONCLUSIONS Given similar performances of direct and indirect measures of kyphosis, we propose that WOD should be used in clinical practice, with a clinical threshold of WOD>4.0 cm as an indication to consider spine radiography. At this WOD threshold, sensitivity was 41% (95% CI, 31-52%) and specificity was 92% (95% CI, 87-95%). WOD should be considered for use in the clinical assessment of osteoporosis patients.