Fill up a small baseball stadium, like Fenway Park in Boston, with people. About 41,000 persons. Those are the number of people who kill themselves every year in the US. That comes out to over 100 persons daily, about 5 persons per hour. The World Health Organization estimates that about 1 million people in the world commit suicide yearly.
Suicide is to psychiatry what mortality is to medicine: our ultimate goal and responsibility in treating psychiatric illnesses. We may or may not succeed in making people happier, or more functional, or more satisfied with their lives. But we ought to at least help them stay alive. We can’t be satisfied with the old morbid surgical joke: The operation was a success, but the patient died.
Suicide is not solely a psychiatric matter. Albert Camus wrote that suicide is "the only really serious philosophical problem". Deciding whether or not life is worth living is to answer the fundamental question in philosophy. All other questions follow from that." Death, more broadly, is a basic human dilemma. We are born all of a sudden; we are alive. We don’t choose to live. But suicide raises the fact that we do have to choose to keep living. For most people, this is not a problem. We want to live, so we just keep living.
But there is a good deal of biological, and clinical, and sociological evidence that human beings are very aggressive creatures. In other words, as Freud put it, besides our Life instinct, we have an aggressive or destructive instinct. We not only want to live, we want to harm, to hurt, even to kill. Sometimes we even want to kill ourselves.
It is a basic fact, hard to deny, that most human beings throughout history have demonstrated a willingness, under the right circumstances, to kill others. We kill animals routinely, to eat them. And there has never been a period of human history without wars, where we kill each other.
So, at one level, it can be considered that Camus may have been right: there is a basic human instinct to aggression, to violence, usually directed to others, but sometimes to oneself.
The question then becomes: Bad as it is to kill others, to commit homicide, why do some people kill themselves?
Before we go on to psychiatric explanations, it is worth pointing out that there are other perspectives. A prominent one is the social explanation, the view that suicide has to do with problems in society. The French sociologist Emile Durkheim first described this view in the 19th century in his classic book, Suicide. Durkheim was the first person to describe population-based statistics on suicide. His impression was that suicide had increased in frequency in the 19th century compared to what was presumed to be the case in earlier centuries. He drew a simple conclusion: Suicide is the result of modern life. The big difference between modernity and the middle ages is industrialization; we work in factories and offices in cities where we avoid each other and where we are separated from families of origin.
“Anomie” was the phrase Durkheim coined: our modern sense of isolation and alienation from each other and from the verities of the past - religion, God, the King.
Many have debated Durkheim’s claim, but it’s clear that some related claims are wrong: they mistake correlation with cause. Some thought that living in cities caused suicide, since suicide rates are much higher in cities than in the countryside. Of course, there are more people in cities too, and those with psychiatric illnesses are often brought to cities, or decide to go there, to seek treatment at medical centers.
This latter point is important - a common mistake people make about suicide. Antidepressants were introduced decades ago; suicide rates went down. But you can’t draw a causal relation. Republicans took over the White House for most of three decades; suicide rates went down. The connection isn’t necessarily causal.
These are called “secular” factors in suicide: social and cultural changes that can affect the entire population, and possibly affect those more prone to suicide negatively.
Four relevant secular factors are poverty, political crisis, war, and sunlight.
John Kennedy identified, in his 1961 inaugural address as president, four major scourges of mankind: “poverty, tyranny, disease, and war itself”. We can think of each of these scourges as causes of suicide. We’ll discuss disease at length. Here let’s put together the other three major social factors: poverty, tyranny, and war.
One way of addressing these complex social factors is to examine national rates of suicide: Greenland has by far the highest rate of suicide in the world, followed by Lithuania, South Korea, Guyana, Kazakhstan, Slovenia, Japan, and then a number of other East European countries. The lowest rates in the world are in Nepal, preceded by Haiti, Grenada, other Caribbean islands, and then Egypt and multiple Middle Eastern states.
Examining this list, poverty doesn't stand out as a risk factor: Japan is wealthy but has a high suicide rate; Haiti is quite poor but has low suicide rates. The presence of multiple East European nations suggests that tyranny, or something related like political instability, could be a risk factor; but the very low rates in Middle Eastern states suggests otherwise. The countries with very high rates don't tend to have had recent wars, but there is no clear correlation. Bosnia suffered a major recent war but ranks similar to the US in suicide rates (28th in the world for Bosnia, 30th for US, out of 111 countries examined).
In short, these secular factors, which you might think would increase suicidality, don’t have strong correlations with suicide. Of course moderating factors, like religion perhaps, could be relevant, and deserve more study.
The last secular risk factor of sunlight turns out to have a much stronger effect, compared to poverty and war, on risk of suicide.
As described in the first issue of PL, there is a direct correlation, though small, between more sunlight and more suicide. There is a much larger correlation between season of the year and suicide, with springtime being the highest risk, followed by fall. In the spring, there is rapid increase in sunlight as winter transitions to summer; in the fall there is steep decline in light, as summer transitions in winter. In both cases, it seems that the slope maters: when there is a change in light, people become more suicidal. Specially, people with mood illnesses appear to be most sensitive to these seasonal changes in light. The slope upwards, from less to more light in April and May, seems to be the most harmful change: one can infer that winter depression is shifting to summer mania, with many people developing mixed mood states, which are the most likely mood states to produce suicide.
One way of identifying what causes suicide is to do “psychological autopsies.” This involves taking cases of suicide and trying to identify common correlates. In this research the following correlations have been found at the time of suicide: clinical depression; the use of antidepressants; anxiety; past suicide attempts; financial or marital/romantic problems; alcohol and substance abuse; male gender; access to firearms; social isolation; older age; white race; German or northern European ethnicity.
When you look at this list of risk factors, you might create the ultimate suicidal profile: an older white male of German ancestry who drinks alcohol excessively, owns a gun, lives alone, just got divorced, tried to kill himself previously, is anxious and depressed, and just got started on an antidepressant. This isn’t a serious claim of course. The idea is that the more of these types of features we see, the more we should be concerned about suicide. In studies where many suicidal risk factors are identified in at-risk persons, though, most such persons still don’t commit suicide. So these risk factors are about relative risk, not any predication that all or even most such persons would commit suicide.
It’s easy to overestimate or underestimate these risk factors. Take past suicide attempts: It’s rational to view them as a risk factor for suicide. But about one-half of persons who kill themselves do so with the first attempt. For half of all suicides, the first time is the only time. You can’t predict their cases by past attempts.
Take anxiety: As discussed in the last PL issue, anxiety is a basic personality trait, with a normal distribution. About one-fifth of the general population has “high” anxiety based on being in the top 2 standard deviations of the normal curve. But much less than 1% of the general population commits suicide. Anxiety itself is a poor suicide predictor.
Of all the risk factors from the psychological autopsies, the most common and predictive appears to be clinical depression. Not only is it present in the vast majority of completed suicides, it works the other way around: A good chunk of persons with severe clinical depression eventually commit suicide (in contrast to anxiety and being white or being northern European in ethnicity).
It turns out that up to 20% of persons with severe clinical depression eventually commit suicide. This is a very large number, in contrast to anxiety. It is more if people have bipolar illness, as opposed to unipolar depression, and more if they’ve ever been psychiatrically hospitalized as opposed to solely needing outpatient treatment. If only outpatient with less severe depressive episodes, and no manic periods, the suicide rate can go down to a 2% lifetime prevalence.
So a major risk factor for suicide is the presence of the bipolar or severe unipolar depressive diagnosis. Now we can start adding other important risk factors.
We mentioned that suicide attempts aren’t present in half of patients, but when present, suicide attempts are important predictors: almost 10% of persons who make a suicide attempt will kill themselves within the following decade.
All these risk factors have to come together in a specific time and place. That’s the final, least predictable, but most definitive description of suicide. Leston Havens gave it a classic formulation: “Suicide is the final common pathway of diverse circumstances, of an interdependent network rather than an isolated cause, a knot of circumstances tightening around a single time and place, with the result, sign, symptom, trait, or act.”
Now we can take the major risk factors, and place them in the mix as usually necessary for suicide. As philosophers put it, they are necessary but not sufficient conditions for suicide. Other factors - like access to a handgun, a recent relationship break-up, the failure of a clinician to return a phone call - are also needed. They provide the exact constellation that leads to the event: they are sufficient but not necessary conditions for suicide.
Havens always emphasized the importance of doing something about the sufficient but not necessary conditions: those are somewhat in our control. So, return all patients’ phone calls, ban handguns, help patients with their relationships. This is true, but it can be difficult to achieve these goals. Banning handguns in the US, for instance, is difficult politically.
PL suggests focusing on the necessary conditions for suicide: pay attention to past suicide attempts, be on high alert in all persons with severe depression or bipolar illness. Treat those conditions appropriately for suicide prevention, which, as we’ll see, doesn't involve most of the medications we use, like antidepressants, but rather a drug most clinicians avoid: lithium.
A special subgroup of persons who commit suicide are current or former members of the military. This problem has been in the news recently, as a result of suicide in those who return to the US from wars in Iraq or Afghanistan. More US soldiers have committed suicide than were killed in action in those wars. There has been some debate about what is causing these suicides. Deployment itself doesn't seem to be causal; but combat experience may be a key factor.
The experience of extensive research on PTSD in Vietnam veterans and in World War II is relevant. In that work, a major causal factor for PTSD, and resulting suicide in some veterans, was the premorbid presence of depression or high neuroticism as a personality trait. As with the civilian PTSD literature, it appears that one’s psychological make-up before experiencing trauma is a key predictor of harmful outcomes, whether it be PTSD or suicide (See PL website on PTSD).
These observations bring us back to the importance of biology: a predisposition to depressive disease and/or anxiety as a personality trait seems to determine who responds to combat trauma with severe PTSD and even suicide, and who doesn’t.
Another at-risk clinical group involves those who engage in self-cutting or other self-harm (burning themselves with cigarettes, head-banging). Called “parasuicide”, this kind of self-harm doesn't have intent to end one’s life. It’s most common in borderline personality, happening in about 60% of persons with that condition. It tends to be associated with a history of past sexual abuse, but not invariably so. It’s unclear whether these persons are at high risk of completed suicide, if they don’t make more typical suicide attempts like overdose. As noted previously, if they do also make standard suicide attempts, like overdose, they do have a notable risk of future suicide.
We come to the great debate about whether antidepressants cause or prevent suicide. In the classic article in this issue, we provide the evidence from the FDA database showing that SRIs increase the risk of suicide in young adults and children (although not in middle age or older persons).
The profession of psychiatry has worked hard to minimize the implications of this FDA analysis and its follow-up black box warning. It’s claimed that suicide rates went down when antidepressants were introduced widely in the rates went up after the black box warning when antidepressant use declined in adolescents and children. It’s claimed in some analyses of insurance records that suicide rates don’t go up after antidepressants are given, but rather decline.
All these claims are scientifically invalid, because they use non-randomized, observational data to try to contradict the FDA-based randomized data. This way of thinking ignores the concept of “confounding bias”, which is the central reason why clinicians need to pay more attention to randomized studies than to anything else, including their own clinical experience. (This concept is described further in the Statistical Concepts section of this issue). As noted above, social factors can affect suicide, and thus population-based rises or declines cannot be directly linked to antidepressant use.
In fact, randomized data with antidepressants repeatedly demonstrate little to no appreciable suicide prevention, although the harm of causing actual suicide also seems to be very low in frequency, as described in the Classic Article.
In sum, the PL view is that SRIs aren’t major benefits in prevention of suicide. In some people, they can be part of the mix that can be harmful.
If we are looking for drugs with a robust benefit shown to prevent completed suicide, there are only two agents shown to do so, lithium and clozapine. If focusing on randomized trials, clozapine has only one study and it showed reduction in suicide attempts but not in completed suicide. Lithium has four placebo-controlled randomized trials, which, taken together (as described in the Current Article of the Month), demonstrate almost 90% reduction in completed suicide.
In short, we have something like a cure, if we’re willing to use it.