Validation of the RADA (RAD and DSED Assessment): A Semi-Structured Interview for Attachment Disorders.

• In Event: Poster Session 08
  In Poster Session: PS 08 Topic - Developmental Psychopathology

Abstract

Introduction.
Attachment disorders (AD), namely the Reactive Attachment Disorder (RAD) and the Disinhibited Social Engagement Disorder (DSED), are psychiatric disorders that children may develop as a result of severe early adversity (e.g. maltreatment, institutionalization, repeated changes of caregivers, Zeanah & Gleason, 2015). Several measures of AD are currently available, but none has been fully validated. The well-known CAPA-RAD interview (Minnis et al., 2009) was recently revised into the RADA (RAD and DSED Assessment semi-structured interview, Lehmann et al., 2020) to consider the latest RAD and DSED DSM-5 criteria (APA, 2013). Recently, the RADA was used in a sample of 320 foster care (FC) adolescents and showed a three-factor organization, with two scales for RAD and one scale for DSED (Lehmann et al., 2020). However, data for younger children of the school-age period are scarce. This study documents the RADA’s factorial structure and other psychometric properties in school-age children.

Method.

Sample. The sample includes three subgroups of school-aged children (N =305; M= 8.8 SD= 2.2): 1) Typically developing children (TDC; n=95), 2) Children in foster care (FC; n=117), and 3) Children in residential care (RC; n=93).

Measures. Primary caregivers were interviewed using the RADA semi-structured interview. The RADA is an “interviewer-based” measures (Angold et al., 2012) which means that the interviewer makes the final judgment about the presence of the symptom (0 = absent, 1 = somewhat present, 2 = symptom definitely present), based on examples of behavior reported by the caregiver. The RADA interview has 20 core items (DSED: 9 items, RAD: 11 items) and each item comprise 2 to 8 questions about a specific symptom.

Results and discussion.
Confirmatory factor analysis (see Figure 1) with WLSMV estimator on the RADA interview items indicates that the best fitting model includes three factors: 1) a DSED factor (all DSED items), 2) a RAD AB12 factor (items measuring RAD criteria A1, A2, B1 and B2) and a RAD B3 factor (items measuring RAD B3 criterion). Fit indices indicate a good fit of the model (X2 = 352.975, df = 167, RMSEA = .060, CFI = .934, TLI = .924). Removing an item that was cross-loading on all factors (items 9: Attention-seeking) improved the fit indices of the model significantly (X2 = 266.121, df = 149, RMSEA = .051, CFI = .956, TLI = .949). The RAD AB12 and RAD B3 factor correlated strongly (r = .70) and they both correlated moderately with the DSED factor (r =.25 and .42, respectively). Items from both RAD factors were combined to form a RAD scale. Both the RAD and DSED scales showed good internal consistency (McDonald omega .82 and .81 respectively). ANOVAs, with Scheffe comparisons, revealed differences between the three groups of children for the RAD (TDC.90 for both scales and kappa > .60 on all items).

Authors

• Marie Taillefer, Université de Montréal (UdeM)
  Presenting Author

• Maude Archambault
  Non-Presenting Author

• Rosalie Guillemette, Université de Sherbrooke
  Non-Presenting Author

• Sebastien Monette, Université du Québec à Montréal (UQAM)
  Non-Presenting Author

• Chantal Cyr, Université du Québec à Montréal (UQAM)
  Non-Presenting Author