Prevention of Suicide in Older Persons: Lessons and Limitations of Evidence-Based Interventions

Gary Kennedy, MD, Albert Einstein College of Medicine, Montefiore Medical Center, New York, pinpointed suicide as a major public health problem that is still not sufficiently addressed either in the research arena or in clinical practice. Traditionally, suicide has been viewed by two competing arguments about what is the cause of death, whether in younger or older persons, but particularly in late life. The belief that isolation and social factors are major contributors to suicide in late life was expressed a century ago by Emile Durkheim,¹ and more recently by Barraclough² and Batchelor and Napier.³ There are also period and cohort effects, and this is a secular phenomenon in that suicide rates vary by generation and by historical epoch, depending on the local stressors as well as community relations. “There are other factors beyond clinical factors that affect this phenomenon,” Dr. Kennedy explained. “Others like Marcia Angell⁴ have argued that the excesses of modern medicine are contributing to older adult suicide.” There is an argument that claims that with modern medicine, physicians can prolong death for days, if not years. “However, older adults don’t necessarily want that kind of autonomy and independence presented to them,” said Dr. Kennedy. “Etzersdorfer et al⁵ have argued that it is press coverage: suicide is being portrayed as a heroic escape from intolerable circumstances. This idea also influences others to adopt suicide as an acceptable means of terminating a problem rather than viewing suicide as a missed opportunity for help and for treatment, both in the patient and the family.” There is also the availability of lethal means. In certain areas like New York City, leaping from tall buildings is a frequent cause of death among older persons. In other areas nationwide, the average means are firearms. Reducing the lethality of the means may reduce the mortality rates and could alter the morbidity of the phenomenon. In competition, there is an argument that it is psychopathology that drives late-life suicide. “Both Esquirol and Kraepelin⁶ were major proponents of this particular message,” explained Dr. Kennedy. van Praag⁷ argued that monoamine, specifically serotonin, represents the break on suicidal behavior; alcohol is the accelerator for violence and suicidal behavior. On the subject of unrecognized depression, Beautrais⁸ suggested that suicide could be reduced by nearly one-quarter by improving the social network. Goldney et al⁹ observed that only about 50% of the attributable risk of suicide was due to depression in the specific group they looked at. He argued that one-third of the risk was attributable to traumatic events in earlier life. Cavanagh et al’s¹⁰ 2003 study of 74 psychological autopsy studies suggested that 25-50% of the attributable risk for suicide was due to social factors rather than mental illness, for example. “In contrast, if the argument is depression is the driving force behind suicide, then one might expect an increasing prevalence in the use of antidepressants to be associated with a marked reduction in the prevalence of suicidal behavior,” said Dr. Kennedy. “A number of studies suggest that that may or may not be the case. In Sweden, spanning the introduction of the selective serotonin reuptake inhibitors (SSRIs) between 1977 and 1997, there was a substantial increase in antidepressant prescribing and a decrease in the slope of suicides among both men and women. In Australia—in a different timeframe, but nonetheless countering the time when the SSRIs began to take off and were widely used in the community—there was a twofold increase in antidepressants among men and a threefold increase among women, a 200% and a 300% increase in the use of antidepressants. However, there was only a 16-23% decrease in suicides among women and 30-41% decline in suicide among men. In contrast, in Italy, about the time the SSRIs were coming in, there was an increase in antidepressants, but no discernible population-based change in the suicide rates. Essentially, there was a decrease of suicide in women and an increase in men. “I think the appropriate way to view those data is that we are treating depression more, but we are not necessarily treating it better,” said Dr. Kennedy. “The national mortality follow-up survey, a 2004 published study, was conducted by Yeates Conwell and other investigators at Rochester. To reduce the psychological autopsy phenomenon, researchers looked at both accidental deaths and suicidal deaths, and then talked with the next of kin about the decedent’s behavior prior to death. They were interested in antisocial behavior, substance abuse, symptoms of depression, symptoms of psychosis, and the comparison is 1,400 persons who committed suicide. But, there are more than 1,000 older adults in this sample and 4,000 persons that were victims of accidental deaths. They discriminated characteristics that distinguished suicide from accidental death. Those with death ideation wished for death but did not say they wanted to commit suicide. Those with suicidal ideation had thoughts of taking their lives, of bizarre behavior, and of threatened violence. Depression did not distinguish suicidal deaths from accidental deaths in this study.” Crude death rates have declined over the last two decades, but the present number of suicides approaches that observed prior to the use of SSRIs. Without a more substantial reduction in rates of late-life suicide, the number of deaths in the families with a legacy of suicide is going to continue to rise. It is thought that a suicide in the family in one parent conveys a sixfold risk of suicide to the children, so reducing the rates of suicide in late life are likely to have an impact on the rates of suicide in young life. Advances in the treatment of depression continue to progress incrementally, leaving substantial numbers of seniors at risk. Physicians are treating more people, but simply prescribing an antidepressant is not enough. “Our public policy also continues to lag behind the obvious need. We still are trying to get parity from mental health services both for older adults and younger adults. The argument is that we need to integrate mental health services into primary care, and we need to re-engineer our primary care patterns. So, we have major policy problems,” said Dr. Kennedy. Dr. Ira Katz, Professor and Director, Section on Geriatric Psychiatry, Department of Psychiatry, University of Pennsylvania, Philadelphia, a principal investigator for the PROSPECT¹¹ study, which seeks to reduce suicidal ideation and suicidal risk in primary care elders, presented data that were recently published in The Journal of the American Medical Association. He discussed the background of the PROSPECT study and its components. The PROSPECT study began at a hearing from the Senate Select Committee on Aging in late 1996 on clinical neuroscience and on late-life depression. The people presenting talked about the malignancy of late-life depression and its association with medical illness and with major morbidity including suicide. At this hearing, Senator Harry Reed of Nevada, a member of the Committee, rose to speak about the importance of the issue. Subsequently, he became a major champion for research on the prevention of suicide and for the importance of recognition and treatment of depression in medical care settings as a vehicle for doing this. The conceptual background is that the elderly have the highest suicide rates in the United States, with older white males at greatest risk. Late-life suicide victims typically see their primary care doctors in the months prior to death, and the majority of older suicide victims have depression. “They have had the first depressive episode in late life, and although effective treatments exist,” explained Dr. Katz, “depression is often not detected or treated by the primary care doctors. The background question for the PROSPECT study was whether a program that improved the recognition and treatment of late-life depression could decrease risk factors and markers for suicide risk. “To conduct a randomized clinical trial looking for a decrease of completed suicide, we would have had to randomize by state and look at state-by-state differences,” said Dr. Katz. As an initial feasible approach we looked at whether we could decrease markers of suicide, specifically suicidal ideation, by treatment focusing on depression. We randomized practices to receive either somewhat enhanced usual care or an intervention. Care managers, in most cases nurses, in some cases social workers or psychologists, went into practices, teaming up with the primary care doctor to give guideline-adherent care for depression. Practices were more or less matched, urban or suburban, and other factors in the three cities, and delivered the interventions on a practice basis.” The basic premise for the intervention was that treatments for depression that are effective exist, but that the availability of effective treatment and guidelines alone don’t ensure appropriate care. Adding a depression specialist, an information system, and education for patients to the mix in order to assist the physician in providing timely and targeted patient-specific strategies is key. The PROSPECT investigators approached 24 primary care practices. Twenty of these practices agreed to randomization, 10 to the intervention, and 10 to usual care. Researchers selected from these practices about 16,000 or 17,000 older individuals eligible for screening, and screened 9,000 individuals to get the study sample. Three hundred and twenty individuals were included in the intervention arm, and about 280 in the usual care arm. There was modest attrition during the course of this study. Most of the subjects were admitted because they had Center for Epidemiologic Studies Depression Scale scores greater than 20. They were then evaluated, and those found to have either major depression or persistent minor depression were entered into this study. The subject population consisted of older adults, with about 30% of people age 75 or above and cognitively intact. About 10% of the people had Mini-Mental State Examination scores lower than 24. Two-thirds of the sample had major depressive disorder and others had minor depression. The mean severity on the Hamilton Rating Scale for Depression was 18, a respectable degree of depression. “One of the things that we noticed at baseline,” explained Dr. Katz, “was that there was an imbalance in suicidal ideation between the intervention and usual care groups. This tells us that in addition to the national variability in rates of suicidal markers, there is also local microheterogeneity that we have to learn more about. The differences were controlled for in analysis.” What were the treatments that were delivered by the end of 1 year in intervention and usual care? Medication was the most commonly delivered treatment, with 66% of the intervention patients receiving it, relative to 44% of usual care patients. Major differences in the delivery of psychotherapy exist, which was a part of the component of care delivered by the depression specialists. “What about the outcomes with respect to suicidal ideation? If one looks at the percent response in ideation across the groups, we very much see an effect, such that the PROSPECT intervention, by targeting depression, does reduce suicidal ideation as an important symptom and a marker of suicide risk. The PROSPECT intervention works. Treatment of depression not only impacts on depressive symptoms in general, but also in suicidality.” Remaining questions include: How can the intervention be implemented and sustained in real life? Can population-based care for late-life depression that involves delivering algorithms beyond the first step, changing treatment when the first approaches aren’t effective, and adding psychotherapy when appropriate decrease actual suicide rates? Targeting the depression can reduce suicidal markers like suicidal ideation, but what else should be done to decrease suicide? What added value is there to targeting hopelessness as another risk factor, to getting the guns out of the house, to making the acetaminophen in blister packs instead of bottles, and to alerting families to issues? There are some subgroup analyses about intervention effects that Dr. Katz and associates are beginning to develop. Among those patients with minor depression, 13% had suicidal ideation, and these patients showed improved outcome when they received the PROSPECT intervention. Therefore, this study demonstrated that the intervention is useful both for those with major depression and those with minor depression complicated by suicidal ideation. Another subgroup analysis looked at individuals who had diabetes. In this group, the usual care versus intervention differences are much greater. This strongly suggests the value of combining depression care management with Medicare-allowable diabetes education. “These issues are in the zeitgeist,” explained Dr. Katz. “The Hartford Foundation-supported IMPACT¹² study found very similar outcomes of an analogous care management intervention. The Substance Abuse and Mental Health Services Administration and VA-supported PRISM-E¹³ study was somewhat different in design. It randomized persons not to care management versus usual care, but to different strategies for integrating mental health services with primary care versus referral to mental health specialists. According to emerging data, 75% or more of those who were randomized to integrated care actually began the intervention with the integrated provider; whereas among those who were referred, only about one-half ever made it into the door of specialty mental health settings. This clearly demonstrated that treatment of late-life depression must be a primary care issue.” Yeates Conwell, MD, at the University of Rochester Center for the Study and Prevention of Suicide, Rochester, NY, has been a major contributor to the investigation of depressive disorders and depressive symptoms in persons who have committed suicide through psychological autopsy and methodology. Dr. Conwell began by providing a framework for thinking about the relationship between risk factors for suicide in later life, and how that can translate into prevention programming. Some characteristics of suicide in older adulthood consist of a dangerous set of behaviors, suicidal ideation, and attempted suicide. In the general population, for every suicide death, perhaps five people are hospitalized with suicide attempts, many more are seen in emergency rooms, and certainly many more people are at risk for suicide with ideation and even attempts that go unrecognized. “That is not the case for later life,” explained Dr. Conwell, “where for every completed suicide in an older adult, there may be only two or four suicide attempts that reach clinical recognition, and probably fewer that escape recognition. Suicidal ideation and suicide attempts are dangerous for older people, and physicians need to be aggressive clinically in working with senior citizens who may be at risk.” Older people tend to be more isolated, less likely to be rescued in the face of a self-destructive act, and so more likely to commit acts that result in death. Physicians need to be cautious in the prevention programming in order to prevent the development of high-risk states, because one might argue that once a senior becomes suicidal, physicians have fewer opportunities and tools available to prevent that person from taking his or her own life. Researchers have conducted psychological autopsy studies, of which there have been a handful of well-done, case-controlled studies comparing seniors who have taken their own lives with comparison groups, typically older adults still living in the community or people who have died from natural deaths. The information in both cases is coming from next-of-kin. These studies have consistently and repeatedly shown that psychiatric illness is present in 85% to nearly 100% of older adults who take their own lives, and the majority of the time it is in the form of an affective disorder, typically major depression. “Studies differ with regard to whether that is a recurrent major depression or the first episode,” Dr. Conwell said, “which may be more likely to go unrecognized. It also shows that medical illness does appear to be a risk factor for suicide, but that in large measure, it has its effect mediated by depression. People develop serious physical illnesses with associated functional impairment, and then become depressed and suicidal in that context.” The role of pain, functioning decline, social dependency, and isolation were themes in Dr. Conwell’s speech. Older adults often become isolated because of depression. Family discord in these studies also comes up as an independent risk factor for suicide in older adults. Other risk factors include stressful life events, most often losses in their lives, as well as background factors, personality, inflexibility, neuroticism, low openness to experience as a characteristic in the five-factor model of personality, and rigid coping styles in the face of a range of other stressful life events associated with aging. “It is our job as physicians,“ explained Dr. Conwell, “to figure out how to address one or more of these in a way that is going to change that person’s options.” “There are a number of terms when we start to think about design of prevention strategies that are pertinent,” said Dr. Conwell. “The one that we are used to is called indicated preventive interventions. These are interventions that address people who are symptomatic, marked as having high-risk factors, individuals with, for example, depressive illness or symptoms that could be diagnosable and treatable. Interventions to prevent full-blown disorders would be the focus of indicated interventions. Selective high-risk groups and individuals, though not all people in that cohort would actually have markers of suicide risk, and prevention would happen through reduction of those risks within those groups. Then, universal preventive interventions address an entire population as the target, irrespective of the risk level within any individual, reducing risk and enhancing health for a population.” After Dr. Conwell discussed the public perspective, he looked at the developmental perspective. “You can think of a developmental model in which, typically, we recognize somebody as in trouble toward this end of the spectrum. But, they got there progressively in later life, a series of steps that began with a background of social ecology, of cultural values and perceptions and personality factors that people, through their experiences, have brought to this critical stage in later life. Things begin to mount up in terms of role, medical illness burden, functional status, and other acute and chronic stresses typically within the family or social environment. People develop symptoms then, resiliency may diminish, depression and hopelessness may become manifest syndromes, and symptoms on top of this basis, perisuicidal state develops, and they become suicidal and take action.” “There is an interesting argument to be made that having suicidal ideation is protective, in a way, because it is a little flag that says “I need help,” and there are a lot of older adults that can be in desperate situations. So, there may be other important ways for physicians to develop strategies to back this preventive intervention program up and to prevent people from getting to that situation to begin with.” The public health approach is called Rose’s theorem,¹⁴ which says that a large number of people at small risk may give rise to more cases of disease than a small number who are at high risk. “In other words,” explained Dr. Conwell, “the focus is not solely in addressing the needs of people at high risk, but finding ways to address entire populations prior to the development of acute situations. The corollary then is that many approaches to therapeutics are insufficient for fundamentally reducing suicide-promoting life situations. Treatments of diagnoses and symptoms are necessary, a very important element of this package, but they are not sufficient at reducing the burden of psychopathology, and don’t specifically reduce turmoil-filled life situations. That is the social context again, and we would argue are comprehensive suicide prevention strategies at a social level.” “Behavioral and social implications of this approach are many, but as a society, we do have some examples of how we have been successful at identifying ‘normative’ behaviors that can be modified at a social level and have great benefit for public health,” offered Dr. Conwell. “Examples of these might be our gradual redefinition over time of what constitutes a healthy blood pressure or a satisfactory cholesterol level, which would result in reduced deaths and morbidity. However, it is a complicated issue, because in our society we value autonomy, and there is resistance to institutional authority defining and limiting our behavior. There is a dilemma as to how to balance institutional authority and moral authority.” Alternative models of risk are complex, and there are many barriers to mental health treatment. Many are social in nature and relate to the stigma that people, including older adults and their families, experience with regard to accessing mental health and the notion of suicide. For older adults who have smaller social networks and are more isolated, social interventions may be less feasible. “We know from research that more time would allow us to review in more detail that social factors place older adults at risk for depression and are determinants, in fact, of outcomes. There is this complex interaction between risk for suicide that is determined by underlying psychiatric illness and depression in particular, and the social context in which older adults live their lives,” observed Dr. Conwell. The speaker explained that manifestations of selective and universal preventive intervention come into play. A selective intervention is one that is population- and high-risk focused, asymptomatic or pre-symptomatic individuals or subgroups who may be at higher than average risk, and focusing on more proximal risk factors. In a study conducted in Padua, Italy, De Leo and colleagues¹⁵ mounted a service that was essentially a bank of social workers available on a 24/7 basis for support to a group of significantly functionally-impaired and at-risk seniors in that region. These were older adults with a mean age of 80 years, of whom a great majority were women. This is an issue given the relative risk for suicide being so much greater among men than women; however, the great majority lived alone. What they found over 10 years was that there were many fewer observed suicides than would have been expected, a standardized mortality ratio of only 29%. This was conceived as a continuous support model, largely a social intervention, but also with the ability to link these vulnerable seniors to additional services as needed. Universal interventions are purely population-based irrespective of individual risk. One example of that might be handgun legislation. Regarding the Brady Handgun Violence Prevention Act, in 1993 Ludwig and Cook¹⁶ looked at 18 states that met requirements at the time that the new legislation was implemented and compared them to 32 states who required a new waiting period for the access to the handgun. What they found was that there was no difference subsequently between that intervention and control states with regard to homicide rates at any age or with regard to suicide rates for younger people. However, there was a significantly greater reduction in suicide rates in the intervention states, firearm suicides specifically, for people age 55 years and over. “In summary, a public health model as opposed to or in combination with a high-risk approach is necessary,” said Dr. Conwell. Physicians typically treat signs of illness, but they do not typically treat risk. Nonmedical risk reduction might be an essential component to add to preventing suicide through recurrence of illness. Shifting population averages rather than focusing exclusively on the worst case requires distinctive approaches. Prevention of disease expression is the desired outcome. “We should combine a medical model or high-risk approach with a public health approach through work with aging services providers,” said the speaker. “And we need to find creative ways to surmount policy and financial barriers to comprehensive suicide risk reduction strategies.”