Suicide is a major cause of mortality among adolescents; it has been estimated that up to 25% of young people have had suicidal ideation, and approximately 2–12% have attempted suicide at some time in their lives [1
]. Standardized clinical assessments of adolescent outpatient samples have revealed that up to 60% have suicidal ideation, and up to 20% have made suicide attempts [2
]. In a sample of adolescent depressed outpatients more than half had made suicide attempts [4
]. According to a review by Safer [5
] anonymous surveys of suicidal behavior have yielded lifetime prevalences of 7% to 10% for adolescents, whereas studies using structured interviews have found lifetime prevalences of 3% to 4%. Regardless of the great variability in the estimations of its prevalence, suicidality in its different forms seems surprisingly prevalent in the adolescent general population. Adolescent suicide occurs mostly in the context of an active, often treatable, but unrecognized or untreated mental illness, such as depression or substance abuse [6
]. The increase in antidepressant treatments of adolescents [8
] have been suggested to at least partly explain the decline in the incidence of suicide [9
] in many Western countries during the past decade. Recently, though, some reports have connected SSRI-treatment in adolescents to an increase in suicidality [10
Suicide attempts are complex acts for which no single set of clinical characteristics can be expected to be a good predictor [12
]. Although the domain of suicidal behavior is multidimensional [14
], a continuum from suicide ideation to suicide attempts has been reported in clinical adolescent populations [15
]. Although only a minority of patients with suicidal ideation attempt suicide, and only a minority of attempters die, a previous suicide attempt has been shown to be one of the most significant risk factors for suicide [17
]. Research concerning the role of suicidal ideation as a risk factor for suicide is less consistent, but many studies suggest that suicidal ideation predicts suicide attempts and suicides [e.g. [2
]]. Thus, accurate assessment of suicidality is of major importance in both clinical and research settings.
The ability of clinicians to evaluate suicidality has been addressed in a few publications. Pelkonen et al. [22
], for example, found that previous and current suicidal behavior was more common than referring persons were able to recognize, and could be detected by the clinician's systematic, structured, and documented inquiring about suicidality of all adolescent psychiatric outpatients. In a study by Malone et al. [23
] fewer suicide attempts were clinically reported than in concurrently and independently completed research data. These studies suggest that a significant degree of past and present suicidal behavior is not recognized and recorded during routine clinical assessment. Thus, easy-to-use instruments are needed to improve the clinicians' ability to recognize suicidality.
Numerous instruments have been developed with the aim of measuring different factors involved in the complex clinical task of suicide risk evaluation, but the use of them is often restricted to research settings [14
]. A three-class mutually exclusive grouping of suicidality (non-suicidal, suicide ideation, suicide attempts) assessed by a clinician is a simplified version of the 5-item "Spectrum of Suicidal Behavior Scale" [24
], and has previously been used in both research and clinical settings [3
]. It consists of two structured questions, which are asked during a routine clinical interview, and the documentation of the answers to them. There is some evidence supporting the predictive validity of this grouping [3
] but there have never been attempts to compare it with more structured measures.
We aimed to evaluate this simple and straightforward assessment of suicidality by training clinicians to ask about suicidality and document the answers, and compare this data with data obtained from the suicidality items of the K-SADS-PL Screen Interview. Although we expected these measures to converge with each other, we hypothesized that the clinical evaluation might under-detect suicidality compared with the structured assessment performed by trained raters.