Childhood maltreatment (CM), widely held as a risk factor for substance use disorders (SUDs), is commonly assessed using the Childhood Trauma Questionnaire (CTQ). Retrospective self-reports are, however, potentially subject to bias. We used a unique patient sample with prospectively documented CM to examine the performance of the CTQ and how this is affected by the presence of SUD.
Methods
Analysis was based on a total of 104 individuals. Subjects with prospectively recorded CM were identified from a specialized childhood trauma unit in Linköping, Sweden (n = 55; 31 with SUD, 61% females; 24 without SUD, 71% females). Clinical controls had SUD but no CM (n = 25, 48% females). Healthy controls had neither SUD nor CM (n = 24, 54% females). We analyzed the agreement between retrospective CTQ scores and prospectively documented CM by κ analysis and assessed the performance of the CTQ to identify CM exposure using receiver operating characteristic (ROC) analysis.
Results
Agreement between prospectively and retrospectively recorded CM exposure was poor for sexual abuse (36.6%, Cohen κ = 0.32, P = 0.008) and physical abuse (67.3%, κ = 0.35, P = 0.007). Overall CTQ performance was fair (ROC: area under the ROC curve = 0.78, optimal cutoff = 36.5, sensitivity = 0.65, specificity = 0.75). However, performance was excellent in the absence of SUD (area under the ROC curve = 0.93, cutoff = 32.0, sensitivity = 0.88, specificity = 0.88), but poor in participants with lifetime SUD (area under the ROC curve = 0.62, cutoff = 42.0, sensitivity = 0.60, specificity = 0.36).
Conclusions
These data support the CTQ as a tool to assess CM exposure but suggest that it may be less useful in patients with SUD.