Thresholds and accuracy in screening tools for early detection of psychopathology

• DOI: 10.1111/jcpp.12442
• Publication Date: 2015-06-19T08:57:29Z
• Authors: R. Christopher Sh...; James C. Benneyan; Ivy Giserman Kiss; Margaret J. Brigg...; William Copeland; Alice S. Carter
• Keywords: Gold standard (test); Odds

The accuracy of any screening instrument designed to detect psychopathology among children is ideally assessed through rigorous comparison 'gold standard' tests and interviews. Such comparisons typically yield estimates what we refer as 'standard indices diagnostic accuracy', including sensitivity, specificity, positive predictive value (PPV), negative value. However, whereas these statistics were originally binary signals (e.g., diagnosis present or absent), questionnaires commonly used in psychology, psychiatry, pediatrics result ordinal scores. Thus, a threshold 'cut score' must be applied scores before can evaluated using such standard indices. To better understand the tradeoffs inherent choosing particular threshold, discuss concept 'threshold probability'. In contrast PPV, which reflects probability that child whose score falls at above has condition interest, refers specifically likelihood equal interest.The two well-validated behavioral assessment instruments, Child Behavior Checklist Total Problem Scale Strengths Difficulties Questionnaire total scale examined relation structured psychiatric interview three de-identified datasets.Although both measures effective identifying groups elevated risk for all samples (odds ratios ranged from 5.2 9.7), who scored near clinical thresholds optimized sensitivity specificity unlikely meet criteria on gold interviews.Our results are consistent with view instruments should interpreted probabilistically, attention where along continuum an individual falls.