A problematic issue in various studies comparing groups of adolescent suicide attempters with others groups (e.g., Cohen-Sandler, Berman, & King, 1982; Taylor & Stansfeld, 1984; Friedman et al., 1984; Brent et al., 1988; Spirito, Overholser, & Stark, 1989; Kosky, Silburn, & Zubrick, 1990; Kienhorst et al., 1990; De Wilde et al., 1992) stems from the underlying premise that these groups are considered to be homogeneous. But even in predefined risk groups, differences exist between individual subjects. It can be argued that, because of underlying differences in psychological and/or social circumstances in which the suicide attempt was made, some attempters are more in need of preventive care than others. This is also plausible in view of the fact that the majority of suicide attempts by adolescents are not repeated (Hawton et al., 1982; Goldacre & Hawton, 1985; Kotila & Lonnqvist, 1987; Diekstra, 1989). This is also true for other at-risk groups. For instance, the majority of depressed adolescents, a group acknowledged to be at risk for suicidal behavior, will never attempt suicide (Alessi et al., 1984; Pfeffer et al., 1988; Kienhorst et al., 1991). This implies that some of them must be in need of more preventive care than others. The question then arises: Which subjects in these predetermined groups are most in need of care to prevent a first attempt or repetition.
A previous study (De Wilde et al., 1993) investigated the differences in psychological characteristics between 48 adolescent suicide attempters (reported after the attempt), 66 depressed adolescents who never attempted suicide, and 43 "normal" adolescents. Of this entire sample, four new groups were constructed, each showing different psychological characteristics and a different risk for attempting suicide (again) at a one year follow-up. This new subdivision proved to have more predictive power regarding the attempts at follow-up than did the a priori group membership.
The classification of subjects in each of the four groups was established on the basis of discriminant scores of the subjects on two dimensions--the first ranging from high scores on measures of competence and low scores on hopelessness, anxiety, and inadequacy to the opposite; the second dimension ranged from reasonable self-esteem, a restrictive attitude toward suicide, and low psychiatric symptoms to the opposite. These two dimensions were orthogonal, and each subject score was to be found in any of the four resulting quadrants. After a one-year follow-up, it appeared that eight subjects of the entire sample (N = 157) attempted suicide (five repetitions and three first-evers). Seven of these were found in the quadrant where problematic scores on both dimensions were combined. This supported the assumption that the subjects in this quadrant (24 suicide attempters, 21 depressed adolescents, and one normal adolescent) were at high risk for attempting suicide (again). By analogy, those in the three other groups (or quadrants) were called subjects with a "low risk" for attempting suicide (again). The low-risk group with the most favorable psychological characteristics was predominantly formed by the normal adolescents; the other two low-risk groups consisted of a combination of depressed adolescents and suicide attempters with other psychological constellations.
The present study investigated three fields of characteristics with known relevance to adolescent suicidal behavior: social support, life events, and behavioral characteristics (for a review, see Spirito et al., 1989). The main question the study tried to answer was: Do these variables frequently associated with adolescent suicidal behavior, discriminate the high-risk, psychologically distressed group from the other, low-risk groups of adolescents who are less or not psychologically distressed?
As stated in the introduction, the four psychologically determined groups were derived from a database consisting of suicidal adolescents, depressed adolescents, and normal adolescents. …