The validity of sacral ratios to identify sacral abnormalities.

Abstract

To test the repeatability and validity of a previously described sacral ratio measurement as a method for detecting sacral abnormalities, as the association between sacral abnormalities and neuropathic bladder is well known but the diagnosis of subtle sacral anomalies is often delayed. Fifty sacral radiographs (from 30 patients with anorectal anomalies and 20 normal children) were viewed and reported as normal or abnormal by an expert spinal radiologist (reference standard). Sacral ratios (anteroposterior, AP, and lateral) were measured while unaware of origin by four investigators and the inter- and intra-observer variability examined. Validity was assessed by comparing individual sacral ratio values with the radiological diagnosis, using Pena's criteria for a normal sacrum (normal >/= 0.74 AP; >/= 0.77 lateral). Sacral ratio estimates were possible in 48 of the 50 selected radiographs. There was good repeatability of measurements and moderate variability among investigators (P > 0.1). For the 20 normal children the mean (sd) AP sacral ratio was 0.74 (0.156); in the 14 with anorectal conditions and a normal sacrum it was 0.87 (0.16) and in the 12 with an abnormal sacrum 0.64 (0.27). There was a difference (P < 0.02) among the three groups (analysis of variance) but there was wide variability in individual sacral ratios within each group, and considerable overlap of values between normal and abnormal sacra. When all sacra were defined as normal or abnormal using Pena's criteria, there was poor agreement using Cohen's kappa (AP and lateral view) with the radiological diagnosis by all four investigators. The sacral ratio has good inter- and intra-observer repeatability. The mean value for a normal AP ratio concurred with that described previously by Pena but the variability of values among similar patients was large, suggesting this single value is of limited value in discriminating a normal from an abnormal sacrum.