35.3 Scores on the Novel Measure of Impulsive Aggression Associated With the Use of Medication for Aggression, Irritability, and ADHD

Abstract

ObjectivesThe aims of this presentation are to evaluate: 1) the reliability and criterion validity of a brief new scale measuring aggression and irritability in a large, nationally representative sample of youths; and 2) invariance across race and sex groups.MethodsA representative US sample of parents of children aged 6 to 18 years (N = 1163) completed a new measure of mood and behavior problems. Latent class analysis of expert ratings identified 43 items measuring impulsive aggression (IA). Exploratory factor analysis (EFA) assigned items to 2 highly correlated scales. Confirmatory factor analysis (CFA) and item-reliability curves were used to select a short form of items with representative content coverage. Area under the receiver operating characteristic curves (AUROCs) evaluated the scores’ predictive power to identify children medicated for ADHD, irritability, and aggressive behavior. Multigroup CFA evaluated invariance across biological sex and race.ResultsEFA guided the selection of 16 items; CFA indicated a good fit (Tucker-Lewis Index [TLI] = .984; root mean square error of approximation [RMSEA] = .068) for 2 highly correlated (r = .88) factors, each high internal consistency (α > .88). IRT analyses indicated good precision across severity levels ranging from high normal to severely clinically elevated. Treating the items as a single scale, higher scores predicted receiving medication for aggression (AUROC = .78), irritability (AUROC = .73), and ADHD (nonstimulant AUROC = .69, stimulant AUROC = .67). Finally, multigroup CFAs showed evidence of strong invariance across race and sex groups.ConclusionsThis new screener shows high internal consistency and good criterion validity against clinically relevant outcomes. It also shows promise in terms of measurement equivalence across race and biological sex groups, reducing potential measurement bias in treatment and services research. Having an unbiased, free tool would help clinicians identify children who might most benefit from treatment for aggression. Treatment sensitivity is the next feature to check.BRD, DEMF, OTH ObjectivesThe aims of this presentation are to evaluate: 1) the reliability and criterion validity of a brief new scale measuring aggression and irritability in a large, nationally representative sample of youths; and 2) invariance across race and sex groups. The aims of this presentation are to evaluate: 1) the reliability and criterion validity of a brief new scale measuring aggression and irritability in a large, nationally representative sample of youths; and 2) invariance across race and sex groups. MethodsA representative US sample of parents of children aged 6 to 18 years (N = 1163) completed a new measure of mood and behavior problems. Latent class analysis of expert ratings identified 43 items measuring impulsive aggression (IA). Exploratory factor analysis (EFA) assigned items to 2 highly correlated scales. Confirmatory factor analysis (CFA) and item-reliability curves were used to select a short form of items with representative content coverage. Area under the receiver operating characteristic curves (AUROCs) evaluated the scores’ predictive power to identify children medicated for ADHD, irritability, and aggressive behavior. Multigroup CFA evaluated invariance across biological sex and race. A representative US sample of parents of children aged 6 to 18 years (N = 1163) completed a new measure of mood and behavior problems. Latent class analysis of expert ratings identified 43 items measuring impulsive aggression (IA). Exploratory factor analysis (EFA) assigned items to 2 highly correlated scales. Confirmatory factor analysis (CFA) and item-reliability curves were used to select a short form of items with representative content coverage. Area under the receiver operating characteristic curves (AUROCs) evaluated the scores’ predictive power to identify children medicated for ADHD, irritability, and aggressive behavior. Multigroup CFA evaluated invariance across biological sex and race. ResultsEFA guided the selection of 16 items; CFA indicated a good fit (Tucker-Lewis Index [TLI] = .984; root mean square error of approximation [RMSEA] = .068) for 2 highly correlated (r = .88) factors, each high internal consistency (α > .88). IRT analyses indicated good precision across severity levels ranging from high normal to severely clinically elevated. Treating the items as a single scale, higher scores predicted receiving medication for aggression (AUROC = .78), irritability (AUROC = .73), and ADHD (nonstimulant AUROC = .69, stimulant AUROC = .67). Finally, multigroup CFAs showed evidence of strong invariance across race and sex groups. EFA guided the selection of 16 items; CFA indicated a good fit (Tucker-Lewis Index [TLI] = .984; root mean square error of approximation [RMSEA] = .068) for 2 highly correlated (r = .88) factors, each high internal consistency (α > .88). IRT analyses indicated good precision across severity levels ranging from high normal to severely clinically elevated. Treating the items as a single scale, higher scores predicted receiving medication for aggression (AUROC = .78), irritability (AUROC = .73), and ADHD (nonstimulant AUROC = .69, stimulant AUROC = .67). Finally, multigroup CFAs showed evidence of strong invariance across race and sex groups. ConclusionsThis new screener shows high internal consistency and good criterion validity against clinically relevant outcomes. It also shows promise in terms of measurement equivalence across race and biological sex groups, reducing potential measurement bias in treatment and services research. Having an unbiased, free tool would help clinicians identify children who might most benefit from treatment for aggression. Treatment sensitivity is the next feature to check.BRD, DEMF, OTH This new screener shows high internal consistency and good criterion validity against clinically relevant outcomes. It also shows promise in terms of measurement equivalence across race and biological sex groups, reducing potential measurement bias in treatment and services research. Having an unbiased, free tool would help clinicians identify children who might most benefit from treatment for aggression. Treatment sensitivity is the next feature to check.