The Rorschach Test for Predicting Suicide Among Depressed Adolescent Inpatients

Abstract

With the decreasing length of psychiatric hospitalizations, identification of test indicators of suicide risk becomes critically important. This Rorschach study was designed to model a clinical decision-making scenario concerning adolescent suicide risk. Using Psychiatric Evaluation Form (PEF) scores, we selected a sample of 25 severely depressed and suicidal adolescents; 26 severely depressed, not suicidal adolescents; and 28 not suicidal, not depressed adolescent inpatients at The Sheppard and Enoch Pratt Hospital. A Rorschach Index using the Exner (1986) Comprehensive System for scoring was developed to predict group membership. Four of six of the features on this index selected 64% of suicidal subjects. This constellation included traditional affective variables (vista responses, color-shading blends, color dominated responses, and morbid content) as well as measures of cognitive distortion (inaccurately perceived human movement responses [M-] and special scores). We discuss the implications of these findings for the diagnosis and treatment of the suicidal adolescent.

CSF 5-HIAA and the Rorschach Test in Patients Who Have Attempted Suicide

“…variables in the suicide constellation for adolescents (Silberg & Armstrong, 1992). Our finding is not contradictory to the results in the study by Rydin et al (1982).…”

“…The false negative rate is now at least 15% and the false positive rate is more than 10% (Exner, 1993a). Silberg and Armstrong (1992) focused their attention on the need to search for an adolescent Rorschach suicide index. Inpatient adolescents were rated, according to the Psychiatric Evaluation Form (Endicott & Spitzer, 1972) as depressed with suicidal behavior, depressed without suicidal behavior and as showing neither depression nor suicidal behavior.…”

Empirical support for interrater reliability of Rorschach Comprehensive System coding

“…% agreement reported in study ranged from 70-93% for 11 categories with only two categories below 80% Lerner & St. Peter (1984b, p. 346) a % agreement for both human differentiation and content above 90% Meyer (1993, p. 157) % agreement calculated for response categories. Agreement ranged from 74-97%, with a mean of 89.2% Perry & Braff (1994, p. 365) ⌲s ranged between .88 and .97 for EII and subcomponents Perry, McDougall, & Viglione (1995, p. 115 Silberg & Armstrong (1992) % agreement on a sample of 10 protocols ranged from 91-95%; % agreement for special scores was 94% Singer & Brabender (1993) % agreement across response categories ranged from 84-93%, with a mean of 86.9% Sloan, Arsenault, Hilsenroth, Harvill, & Handler (1995) 20 protocols were chosen randomly and scored independently, achieving an overall % agreement of 88%; range 84-93% Sloan, Arsenault, Hilsenroth, Handler & Harvill (1996) 20 protocols were chosen randomly and scored independently, with an overall % agreement of 88% Stuart, Westen, Lohr, Benjamin, Becker, Vorus, & Silk (1990) a ⌲, ranging from .62-.96; above .80 for 9 of 13 scales.…”

An Empirical Study of Seriously Disturbed Suicidal Patients

“…Morbid images of dead, damaged, decaying, and injured objects (Exner 1993;Schafer 1954) may best capture representations of the self as damaged, depleted, and attacked, all of which may be related to experiences of self-hatred, despair, and hopelessness. The morbid response is one element of the Suicide Constellation (Exner 1993) and has been empirically associated with completed and attempted suicides (Affra 1982;Exner 1993;Silberg and Armstrong 1992). A high number of morbid responses have also been linked to the presence of major depression (Exner 1993).…”