Comparison of the "Lifetime parasuicide count" and the "Structured clinical interview for suicide risk-correctional version" for predicting self-injurious behavior with offender/patients

This study investigated the reliability and validity of two suicide risk instruments, which focus on past self-injurious behavior. "The Lifetime Parasuicide Count" (LPC; Comtois &. Linehan, 1999) asks specific questions concerning past self-injurious behavior, while the "Structured Clinical Interview for Suicide Risk- Correctional Version" (SCISR-CV; Giles, 1997) asks one general question related to past self-injurious behavior. This study was a preliminary investigation of which assessment instrument provided the most accurate prediction of actual self-injurious behavior. Self-injurious behavior was determined via reports in the Texas Department of Criminal Justice (TDCJ) records. The participants were 45 male offender/patients. These participants were randomly chosen from the offender/patients admitted to the John T. Montford Psychiatric Unit in Lubbock, Texas. The participants had a documented history of self-injurious behavior within the Texas Department of Criminal Justice System (TDCJ).

Twelve participants were involved in a Pilot Study. Thirty-three participants were involved in two interview sessions. For Time I, participants with a known TDCJ history of self-injurious behavior were randomly chosen from the John T. Montford computer database and placed in two initial interview groups (14 LPC vs. 19 SCISRCV). Four weeks after the first interview, participants were retested with the same interview assessment. Research Question 1 addressed the inter-rater reliability of the LPC and the SCISR-CV. Thus, the relevant correlations were calculated regarding the total number of self-injurious behaviors, potentially lethal self-injurious behaviors, and types of self-injurious behaviors. These correlations were calculated for the LPC, SCISR-CV, and the chart reviews.

The results suggested that inter-rater reliability for ratio data was low to high (.09-1.0). The results suggested that the inter-rater reliability for nominal data and raters' clinical judgment decisions was low to moderate (17%-83%). Overall, the results suggested that inter-rater reliability was highest for the SCISR-CV Also, results indicated that the SCISR-CV yielded more consistent data than the LPC. Although hypotheses suggested that the LPC assessment would provide the most accurate information about past methods, intent and number of self-injurious behaviors, neither interview provided an accurate offender/patient report of overall self-harm behavior and nonlethal self-harm behavior. Only Lethal Self-Injurious Behaviors were accurately predicted by the SCISR-CV From this limited data set, predictions of self-harm behavior were moderate. As for treatment during the study, all offender/patients received some type of psychological intervention. Due to the small sample, caution is justified when evaluating the results. To improve these risk assessment instruments, more studies with large and diverse samples are needed.