The rate of completed suicide among older persons continues to be the highest of any age group in the United States. This trend has led to a call by the Surgeon General for a "national strategy" of prevention and intervention (1). Many questions need to be considered. What are the characteristics of this form of ending life and of the persons who employ it? In addition to age, are there cohort and period effects? Are there warning signs or risk factors? Are all people who commit suicide "mentally unbalanced" or of "unsound mind," or might this act sometimes be considered "rational"? Most important, how does the suicidal act arise as "the final common pathway of diverse circumstances" (2)? This review, which is based predominantly on data from the United States, considers recent attempts to find answers to these questions.
Suicide is the seventh leading cause of potential years lost in the United States (3). In 2000 the suicide rate in the 65- to 74-year-old group was 12.6 per 100,000 people, rising to 19.4 per 100,000 for the 85-year-and-older group, with men having a rate of approximately 60 per 100,000 (4). The challenge posed by these observations is compounded by the fact that a majority of elderly persons who take their own lives are clinically depressed and have seen physicians, usually in primary care, shortly before committing suicide (5). For both these reasons successful identification of suicidal risks and ensuing intervention should be the rule. Yet they are the exceptions.
Methodologic difficulties abound in researching this complex area (6). Problems include the low base rate, absence of standardized informant interview protocols, lack of control groups and prospective studies, variability in the criteria for making a diagnosis of suicidal intent, and the varying qualifications of those doing so. Most investigations into late-life suicide employ variants and combinations of two data-gathering methods: reviews of coroners' or medical examiners' records and psychological autopsies (7). The latter technique centers on interviewing informants concerning the decedent's history and the suicidal act. Several investigators have shown that this method has validity and interrater reliability (8).
In addition to age, gender and ethnicity are influential variables. The rate of suicide for women plateaus or decreases slightly after peaking around age 60 (3). A possible reason for the lower suicide rate among elderly women is their wider social network compared with men's, and their lower rate of suicide may also reflect their less frequent use of lethal means, especially firearms (9), although this may be changing (10).
Race is also associated with the incidence and prevalence of completed suicides. The rate of suicide in the Caucasian group markedly outweighs rates in other groups, although a rise in the rate of suicide by black males has begun to appear (4). In 1998 4,655 Caucasian men aged 65 years or older died by suicide (33.1 per 100,000), and 132 African-American men did so (11.7 per 100,000). A closer analysis of rates by age, gender, and ethnicity delivers small enough numbers that the results are considered unreliable.
How aging is experienced varies from person to person, with individuals reacting differently to the impact of loss and bereavement, physical illness, changes in psychosocial roles, diminution of income, and other factors.
The increasing social isolation believed to be an inevitable concomitant of advancing age has been thought to be a contributing factor to the rising rate of suicide as age increases, although this is still not established (11). Widowhood among men, but not among women, appears to carry particular risks, and the corollary also appears to be the case—marriage confers a protective effect against suicide. Loss or separation in early life could be a mediating influence on suicide by bereaved persons after the death of a spouse (12). Social support as perceived by the individual does not always correspond with objectively measured support, with depression and hopelessness being determinants of how such support is experienced (13).
Murder of a spouse or partner followed by the perpetrator's suicide has recently received research attention (14). On the basis of a series of cases, the incidence is estimated at 1,000 to 1,500 cases per year (14). The authors suggested a relationship typology in which most perpetrators are of the dependent-protective type and the others are divided among aggressive and mutually dependent, symbiotic types.
Age is accompanied by an increase in the degree of suicidal intent and, consequently, mortality. This observation is based on the fact that the ratio of completed to attempted suicide is higher for persons aged 65 years or older than for any other age group and especially so for elderly men. The methods employed contribute to the relationship of age with the dangerousness of the suicidal act, with men more prone to using firearms (9).
The association between physical illness and suicide is especially marked among the elderly (15). One study found a specific association between visual impairment, neurologic disease, and malignant illness with suicide, and the association was especially marked in men (16). However, the role that terminal illness plays in suicide is controversial. Although it is commonly believed that terminal illness, especially when accompanied by pain and the loss of independence and hope, provides grounds for suicidality, several investigators have not found this to be the case. Conwell and Caine (17) found that only one person among 85 persons of all ages who completed suicide had a terminal illness. There is evidence that the link between physical illness and suicide is mediated by depression and hopelessness (18). Because much illness during old age is of a chronic nature, expectations of admission to long-term care may arise, and this anticipation appears to be a factor associated with completed suicide among the elderly (19). However, admission to a long-term care facility is no guarantee that the risk of suicide has passed (20).
An extensive array of investigations has established that it is depression, usually unipolar, and to a much lesser extent alcohol abuse that are the most common axis I diagnoses associated with completed suicide among the elderly (21).
Also, a body of evidence is growing concerning the contribution of personality to completed suicide among the elderly. Efforts to apply the available axis II framework have not been very fruitful. Recently, however, the construct of "openness to experience" has been found useful in typifying the personality of elderly persons who commit suicide (22). This personality is characterized by such terms as "constricted," "shut in," and "[showing] diminished affective intensity, preference for the familiar (that is, find change very stressful)" (22). Pearson and Brown (23) suggested that "the personality characteristics of the majority of elderly male suicides include the appearance of being overly strong and self-sufficient, having difficulty with being dependent and accepting help, and [having] a rigid, inflexible coping style, similar to a lack of openness to new experience."
Contact with medical services
Numerous investigations have found that suicidal elderly people have frequently consulted medical services, usually a primary care physician, shortly before the suicidal act, sometimes in the last few days or hours before the suicidal act (5). Also, the diagnosis of depression and suicidal intent is infrequently made among the elderly (24,25). One investigation found that 41 percent of 1,397 persons who committed suicide had contact with a health care professional within 28 days of death, 47 percent within one week, and 18 percent on the day of death. Suicidal intent was discussed in only 22 percent of these meetings (26).
Several initiatives have been started in order to improve the situation pertaining to primary care (27,28). The Prevention of Suicide in Primary Care Elderly: Collaborative Trial provided depression screening of randomly selected patients aged over 60 years who were being seen in 20 primary care practices in three cities. In certain cases intervention consisting of psychopharmacologic treatment and psychotherapy was more rapidly successful in reducing the Hamilton Depression Rating Scale score and suicidal ideation than in the control group, which received customary care.
Although the pervasive influence of depression on suicidality has been established, a growing body of evidence indicates that depression is a common but in many cases not sufficient diagnosis in the understanding, evaluation, prevention, and treatment of suicidality (7). These observations have been followed by attempts to conduct the prospective, multivariate studies that are needed to illustrate and explore them (29,30). In a community-based prospective study of aging, the Established Populations for Epidemiologic Studies of the Elderly, 21 persons out of 14,456 people over age 65 committed suicide over the ten-year observation period. Depressive symptoms, perceived poor health, poor sleep quality, and absence of a confidant predicted suicide. Alcohol use, medical illness, and physical impairment did not. The authors make the point that events occurring in the immediate period before suicide ("triggers") were not assessed, so the factors that were identified represent longer-term propensities ("will to live") and as such offer more opportunity for recognition and intervention (11). Personality was not assessed.
Multifactorial causation requires broadly directed and multimodal interventions. Improved diagnosis of depression, better treatment of depression, better control of possible means of suicide, increased supervision (31), and further education of primary care physicians (32) have not accomplished the sustained, larger-scale reduction of suicide prevalence in old age that is the goal. These considerations have led to a new approach based on a public health rather than a disease paradigm (33). This change requires a layered approach to prevention (34). One example of such a multipronged approach that shows robust effects has been reported by the U.S. Air Force (35). Key components are commitment, communication, destigmatization, collaboration, and training of "gatekeepers" or informants.
Freud wrote in 1917 "we have never been able to explain what interplay of forces can carry such (suicidal impulses) through to execution" (36). In the case of elderly people, the "interplay of forces" has arguably even more components than for younger persons. These components are gradually beginning to yield to enquiry.
Dr. Loebel is clinical professor emeritus of psychiatry and behavioral sciences at the University of Washington, 5505 Coniston Road, NE, Seattle, Washington 98105 (e-mail, firstname.lastname@example.org). Marion Zucker Goldstein, M.D., and Olivera J. Bogunovic, M.D., are editors of this column.