Elder Abuse Decision Support System: Field test outcomes, abuse measure validation, and lessons learned

Abstract

The Elder Abuse Decision Support System was designed to meet the critical need for improved methods for assessment and substantiation of elder mistreatment, using a web-based system with standardized measures. Six Illinois agencies participated in the field test. One-year pre/post analyses assessed substantiation results, using Illinois' standard investigation procedure as a comparison. Pre/post acceptability was assessed with caseworkers in focus groups with adult protective service staff. Validity of measures was assessed using Cronbach's alpha and receiver operator characteristic curve analyses with final substantiation decision as a criterion. Increased substantiation of abuse was found. Regarding acceptability, the two systems were found to have differing strengths and weaknesses. Outcome measures had high validity estimates, while focus groups indicated directions for improvement. This study was a successful proof of concept that data collected in the field would be useful for clinical purposes as well as for research.

Development of short-form measures to assess four types of elder mistreatment: Findings from an evidence-based study of APS elder abuse substantiation decisions

“…The validity of each short form, i.e., financial, emotional, physical, and neglect, was judged against the criterion of the final yes/no substantiation decision of abuse for each type, respectively. In this study, we compared the area under the curve (AUC) of the short forms to the AUC of the long forms (Conrad et al, 2017) using the preceding standards.…”

“…However, these broad instruments are inevitably less effective for detecting specific manifestations of each abuse type, leading to higher false-negative rates. Incorrect assessments can have negative unintended consequences for providers, caregivers, and patients (Beach, Carpenter, Rosen, Sharps, & Gelles, 2016, p. 194), yet comprehensive questionnaires, such as the rigorously constructed, comprehensive Elder Abuse Decision Support System (EADSS) interview guides (Conrad, Iris, & Liu, 2017), and other extensive measures of each abuse type (e.g., Acierno et al, 2010), are less feasible because they require substantially more time to administer. Indeed, even though the EADSS questionnaires were demonstrably quite effective at helping APS caseworkers substantiate reports of alleged abuse, follow-up interviews with caseworkers in Illinois revealed that the entire battery of EADSS abuse measures was too time consuming to administer given their caseloads.…”

“…The EADSS is a theory-based system developed through extensive literature review and testing in the field (Conrad et al, 2017; Conrad, Liu, & Iris, 2017). It includes comprehensive, structured interview guides to assess risk of abuse and abuser/victim characteristics and has detailed measures of five types of elder mistreatment: (1) the Older Adult Financial Exploitation Measure (OAFEM); (2) the Older Adult Emotional Abuse Measure (OAEAM); (3) the Older Adult Physical Abuse Measure (OAPAM); (4) the Older Adult Neglect Measure (OANM); and (5) the Older Adult Sexual Abuse Measure (OASAM).…”

“…Construct validation of the full-length EADSS abuse measures (Conrad et al, 2017) showed high areas under the ROC curve (McNeil & Hanley, 1984) for FE (96.6%), emotional abuse (EA) (97.7%), PA (97.0%), and neglect (96.2%), using the final substantiation decision as the validity criterion for each abuse type. In ROC analysis, values close to 100% indicate that a measure is highly accurate in detecting a particular phenomenon; in this case, substantiated abuse and/or neglect.…”

“…Use of the EADSS during the field test also increased rates of abuse substantiation (Conrad et al, 2017) because caseworkers were instructed to administer all assessment forms regardless of what abuse type had been reported. Requiring administration of all the EADSS long forms yielded information about all types of abuse (not just the abuse type initially reported).…”