Widening epidemiological data on the prevalence of child maltreatment: Validation of the German ICAST-R in a student sample and national household survey

Abstract

BackgroundA number of instruments for measuring child maltreatment (CM) prevalence have repeatedly been used across different countries. Although they hold the potential for providing benchmarks to tackle the gap of lacking comparability of CM prevalence across countries, contextual information about the adverse experiences such as perpetrator, chronicity, frequency, or severity are rarely covered. The ISPCAN Child Abuse Screening Tool - Retrospective (ICAST-R) covers these important dimensions. The German version increases the number of available versions to 21 different languages. Spoken by about 120 million people, German is one of the 20 most prevalent languages around the world. Moreover, the ICAST-R is intended to be used with young adults. This study further aims at adding towards the gap of psychometrics in older age groups. MethodsAnalyses are based on both a sample of German students (n = 333) and a nationally representative household survey (n = 2515). The validation process covered six steps: (1) Analyses of missing data on single items, (2) calculation of descriptive statistics to estimate the prevalence CM as well as subjective severity and main perpetrators. (3) Structural validity of the four conceptualized subtypes of CM (neglect, physical abuse, emotional abuse and sexual abuse) was tested using confirmatory factor analyses (CFA). Next (4), equivalence testing by multigroup confirmatory factor analyses (MGCFA) on age groups was conducted within the representative sample; (5) reliability was tested by determining internal consistencies for each subscale via the McDonald's Omega, Kuder-Richardson 20 (KR-20), and Cronbach's alpha. Lastly (6), criterion validity was tested in regression models comparing depressive/anxious symptomatology for single victimization and polyvictimization. ResultsThe German ICAST-R yielded low missing values items in both samples. 16 % of the participants in the national household survey reported neglect, 20.3 % physical abuse, 22.2 % emotional abuse, and 8.6 % sexual abuse. Polyvictimization was prevalent with 20.6 % of subjects reporting >2 types of CM. Students in the pilot-survey reported much higher prevalence estimates than participants in the nationally representative sample. The types of CM subjectively rated as most harmful were emotional abuse and sexual abuse. In both samples, structural validity was similarly confirmed as CFA was reproducing the four conceptualized subtypes of CM with adequate fit (household survey: CFI 0.919, TLI 0.907, RMSEA 0.017, SRMR 0.046). Internal consistency achieved acceptable and comparable values for all three types of coefficients; criterion validity was established with a significant dose-response effect of CM experiences on both anxiety and depressive symptoms/diagnoses. Age dependent analyses on structural validity (MGCFA) and reliability in the household survey revealed potential weaknesses of items. ConclusionThe German version of the ICAST-R both widens the possibility of international CM prevalence comparison and provides novel epidemiological data for Germany on subjective severity of CM and CM perpetrators. Even in the presence of a marked selection bias, the ICAST-R had similarly good psychometric properties in the student and nationally representative household sample. Except for issues with two items, equivalence testing was comparable across age groups.