The clinical utility of measured kyphosis as a predictor of the presence of vertebral deformities

Abstract

Meaured spinal kyphosis, as a predictor of prevalent and incident vertebral deformity, was examined in older women (>70 years) and found to not have sufficient sensitivity or specificity to justify its use as a predictor of present or future vertebral deformity risk. Kyphosis may be attributable to vertebral deformity and was investigated as a clinical tool for predicting the presence and future risk of vertebral deformity. Kyphosis was measured in 434 women aged 70 years or older and the kyphosis index (KI) calculated. Prevalent and incident vertebral deformities were assessed by morphometric X-ray absorptiometry (MXA). The predictive value of KI was examined. Severity of kyphosis was categorised by tertile of KI; 65% of anterior thoracic deformities occurred in the 33% of subjects in the highest (most kyphotic) tertile. Using this tertile as a predictor of anterior thoracic deformity, the probability for a positive test rose from 14% for the whole population to 28% and for a negative test the probability fell to 8%. For any spinal deformity the highest tertile of KI increased the probability of a positive test from 34% to 42% and reduced the probability for a negative test to 30%. The incidence of new deformities was 6% over 4 years; a high KI tertile did not increase the probability of any vertebral deformity. Severe kyphosis does not increase the probability of detection of a prevalent or incident spinal deformity sufficiently to make it a useful method of selecting patients for further evaluation of spinal deformity.