Detection of Suicidal Risk among Hospitalized Veterans: Preliminary Experience with a Suicide Prediction Scale

Article excerpt

The development of prediction scales which aim to improve, or at least complement the clinician's assessment of suicide risk, has a relatively short history. Research in this area has generally been developed using three different approaches to the problem. The first involves assessment via lethality scales which are behavioral and demographic in origin; the second deals with traditional risk assessment evaluations through the use of psychological tests; and the third has developed out of information gleaned from intensive survey researches of the characteristics of suicidal individuals.

Early work on the development of suicide prediction scales has been confined to a rather small range of sociodemographic and clinical data and has shown that variables such as older age (Kreitman, 1976), male sex, broken marriage or widowhood, social isolation or disorganization, and psychiatric illness, notably depression, are linked with suicide (Brown & Sheran, 1972; Lester, 1972). By reviewing 378 studies, Devries (1968) proposed a three dimensional model for predicting suicidal behavior and saw the need for comprehensive investigations along three major variables: the person's psychological characteristics, his social background and relationships, and his physical determinants. Nevertheless, even some of the more promising studies have continued to rely on police records (Tuckman & Youngman, 1968), on records of hospitalized psychiatric patients (Farberow & MacKinnon, 1974) or on the records of suicide prevention centers (Lettieri, 1974) instead of on detailed investigations of the lives, problems and mental states of those most at risk for suicide. Pallis and colleagues (1982) attempted to incorporate this information in a discriminate function analysis designed to separate a sample of 75 suicides from a sample of 146 attempted suicides on which comprehensive clinical and social data were recorded on an identical schedule. Two sets of discriminating items (with 18 and 6 variables) correctly classified 91% and 83% of the two samples in their respective groups.

There exist significant problems of a methodological nature which make the accurate prediction of suicidal behavior difficult. To date there has been only one study to have used a sample on which comprehensive data were available and which was large enough to have employed advanced statistical procedures for developing a predictive scale (Present Depressive State and Resources Scale) (Motto & Heilbron, 1976). This study also assessed the scale's efficiency by subsequent follow-up. Although the predictive accuracy of this particular scale was discouragingly low, a subsequent scale for a selected high risk group has produced more encouraging results (Motto, 1978). More recent attempts to develop a valid measure of suicidal behavior include the Scale for Suicidal Ideation (SSI) (Beck, Kovacs & Weisman, 1979), the Suicide Probability Scale (SPS) (Cull & Gill, 1982), and the modified Scale for Suicidal Ideation (MSSI) (Miller, Norman, Bishop, et al., 1986).

Importance of Assessing Suicide Lethality

Among Rehabilitation Patients

Epidemiologic studies have related suicide to many emotional states: hostility, despair, shame, dependency, hopelessness, ennui. It is well documented that persons with physical disabilities typically experience such symptomatology during the adjustment process. However, while there exists a limited data base which identifies the prevalence/incidence of active suicide among persons with disabilities, health care professionals generally are of the opinion that suicide among this population by indirect means and through self-neglect is a significant issue which must be addressed.

Committing physiological suicide by ceasing to care for one's body has been examined by several investigators. For example, Sainsbury (1955) found a relationship between the diagnosis of cancer and suicide rates among persons with the disease. …