Inter-rater reliability in the medical diagnosis of child sexual abuse.

Abstract

To examine inter- and intra-rater reliability in the assessment of genital findings in cases of suspected child sexual abuse. Colposcopic photographs of the external genitalia of 70 female children were independently assessed by child sexual abuse teams in Auckland and Sydney. For the Auckland centre, intra-rater reliability was assessed by making a second independent assessment 6 months following the first. Reliability was quantified using per cent of agreement and Cohen's Kappa statistic. There were high levels of inter- and intra-rater agreement. When photographs were classified as normal/non specific or strongly indicative of child sexual abuse, there was 93% agreement between the Auckland and Sydney teams with a kappa score of 0.70. For the two separate ratings made by the Auckland team there was 94% agreement with a kappa score of 0.75. The high levels of inter- and intra-rater agreement obtained in this study were reassuring. The results obtained compare favourably with the results of reliability studies in other areas of medical practice.