shinyDLRs: A Dashboard to Facilitate Derivation of Diagnostic Likelihood Ratios

Despite increased recognition of the importance of evidence-based assessment in clinical psychology, utilization of gold-standard practices remains low, including during diagnostic assessments. One avenue to streamline evidence-based diagnostic assessment is to increase the use of diagnostic likelihood ratios (DLRs), derived from receiver operating characteristic curve analyses. DLRs allow for the adjustment of the likelihood that an individual has a disorder based on self-report data (e.g., questionnaires, psychosocial, family history). Although DLRs provide strong and readily implementable psychometric data to guide diagnostic decision-making, analyses necessary to derive DLRs are not commonplace in psychological curriculum and available resources require familiarity with specialized statistical methodologies and software. We developed a free, researcher-oriented dashboard, shinyDLRs (https://dlrs.shinyapps.io/shinyDLRs/), to facilitate the derivation of DLRs. shinyDLRs allows researchers to carry out multiple analyses while providing descriptive interpretations of statistics derived from receiver operating characteristic curves. We present the utility of this interface as applied to several freely available measures of mood and anxiety for the purposes of guiding diagnosis of psychopathology. The sample leveraged to accomplish this goal included 576 youth, 4-19 years of age, and a parent informant, both of whom completed several questionnaires and semi-structured interviews prior to participating in treatment at a university-based research clinic. Lastly, we provide recommendations for inclusion of DLRs in future research investigating the psychometric properties and diagnostic utility of assessments. Public Significance Statement This study developed a free online interface to facilitate the derivation of diagnostic likelihood ratios (DLR) for application to the psychometric evaluation of evidence-based assessment. This article details how researchers may apply this interface to calculate and evaluate DLRs, to facilitate EBA and statistically guided decision-making techniques.