Search

The study assesses how an educational intervention describing uncertainty in child sexual abuse assessments affects estimates of sexual abuse probability by non-physician child abuse professionals (CAP). It evaluates whether CAP incorporate medical information into abuse estimates in concordance with Bayes' Theorem. Eighty-nine CAP estimated the abuse probability for a hypothetical preadolescent female: (1) randomly selected; (2) disclosing abuse; (3) with physical evidence of abuse; and (4) disclosing abuse but with a normal physical examination. CAP then attended a workshop that included discussion of uncertainty in abuse assessment. Post-lecture questionnaire, identical to pre-lecture questionnaires except for estimates of the examination sensitivity and specificity, were administered. Expected responses for post-lecture Scenarios (3) and (4) were generated using Bayes' Theorem and compared to actual responses. Respondents estimated a high abuse prevalence (average 32%, range 5 to 75%). Respondents incorporated medical information into their estimates in a Bayesian manner. However, they undervalued the medical exam findings relative to Bayes' Theorem. These findings suggest CAP had difficulty adjusting for medical uncertainty. Further research into approaches for more easily adopting Bayesian approaches to child abuse decisions may improve the quality of decisions made by child abuse professionals. [ABSTRACT FROM AUTHOR]