suicide in the past, but unlike Gayle, she had no history of repeated
painful and provocative experiences through which she might have
acquired the ability to enact lethal self-injury. She thus did not have
the setting condition for serious suicidal behavior, even though, as it
turns out, she did have the other factors important in the current
theory. That is, she felt she was a burden on others and felt discon-
nected from them. These feelings, combined with statements like,
“I’d be better off dead” and with symptoms like sleep difficulty,
clearly indicated a mood disorder, but her risk for suicide was slight.
The thought never occurred to me that she should be hospitalized.
Indeed, though she clearly had a mood disorder, it was of relatively
moderate severity, and she remitted with less than two months of
psychotherapy and stayed remitted for at least two years thereafter,
which was the last time I contacted her.
The cases of Gayle and Sharon, especially when viewed through
the lens of this book’s theory on suicide, are informative regarding
suicide risk assessment. Generally speaking, someone like Gayle is at
chronically elevated risk, at least to some degree, because the capacity
for serious self-injury already is in place. All that is needed for Gayle
to engage in serious suicidal behavior if she chooses is a quick change
in her feelings of connection and effectiveness. Accordingly, routine
assessment of risk status is required with someone like Gayle. By
contrast, someone like Sharon is unlikely to engage in serious self-
harm because she has not beaten down the instinct to live. Even if
What We Know and Don’t Know about Suicide ● 25
Sharon feels disconnected from others and ineffective, she lacks the
capacity to translate the desire for death into action. These points
will be expanded on in a later chapter on clinical implications.
Notes from Scientific Research
The science about suicide is not especially well developed and has
certainly not permeated the public consciousness. I was reminded of
this the other day at my sons’ soccer game. There were five or so full-
field games going on—approximately 150 people out on the fields.
Off in the distance, lightning struck, and the field administrators de-
cided to cancel the games. There was some grumbling about this de-
cision of course, but everyone understood the rationale—lightning
can be lethal.
But just how lethal is lightning? In other words, how many people
die from lightning strikes? In fact, from 1980 to 1995, there were ap-
proximately eighty deaths per year from lightning strikes in the
United States. During this same time period, there were more than
eighty deaths per day from suicide.
Why do people scramble to prevent death by lightning strike but
don’t scramble in the same way to prevent death by suicide? The
latter is approximately 365 times more common than the former.
One could invoke bias or stigma against mental health problems,
but I think a more mundane answer is available. It is fairly easy
to understand how and why people die by lightning strike, and pre-
vention is straightforward too—you just get out from under the
weather. By contrast, it is not at all easy for people to understand how
and why people die by suicide, and prevention is not clear-cut at all.
To make the prevention of suicide more like the prevention of light-
ning strikes, people need a clearer understanding of how and why
people die by suicide. This book is intended to provide such an un-
derstanding.
The example of lightning strikes does not really illustrate bias
26 ● WHY PEOPLE DIE BY SUICIDE
against suicide; rather, it simply indicates that lightning is a well-
characterized phenomenon and its prevention is straightforward.
But in other examples, bias and stigma are detectable. In Tad Friend’s
2003 New Yorker article on suicide at the Golden Gate Bridge, he
points out that a main reason for community resistance to a suicide
barrier fence (which would clearly save lives) is aesthetics. For the
past twenty-five years, however, a large section of the bridge has been
festooned with an eight-foot-tall cyclone fence directly above a site
people dropping things—including, to take a real example, bowling
balls—on other tourists below. Friend cites the bridge’s former chief
engineer as saying that the fence is needed because “It’s a public-
safety issue.” True enough, it is a public safety issue, but not one that has ever killed anyone, bowling balls notwithstanding. By contrast,
around thirty people die by suicide each year by jumping from the
bridge. The acceptability of a debris fence coupled with the un-
acceptability of a suicide barrier seems misguided and unfair.
To digress a bit, the stigma and taboo of suicide are topics that
warrant their own book. The stigma, pervasive and enduring, can be
found even in the seventh circle of Dante’s Inferno. As A. Alvarez4
summarizes, “In the seventh circle, below the burning heretics and
the murderers stewing in their river of hot blood, is a dark pathless
wood where the souls of suicides grow for eternity in the shape of
warped poisonous thorns . . . At the Day of Judgment, when bodies
and souls are reunited, the bodies of suicides will hang from the
branches of the [thorns], since divine justice will not bestow again
on their owners the bodies they have willfully thrown away.” Accord-
ing to Dante, my dad is, as I write this, below the murderers, and will hang from thorns for eternity—stigma indeed.
To return to the Golden Gate Bridge, aesthetics does not really
provide a convincing explanation for the lack of a suicide barrier, but
what about cost? As Friend points out, cost did not prevent the re-
What We Know and Don’t Know about Suicide ● 27
cent construction of a barrier between the bridge’s walkway and
traffic, designed to separate bicyclists from traffic. This barrier cost
5 million dollars, and yet no bicyclist has ever been killed on the
bridge. Five million dollars and zero deaths for bicyclists; zero dollars
and over a thousand deaths by suicide: it is difficult to avoid the con-
clusion of stigma and bias.
Regarding knowledge about suicide and its prevention, much re-
mains to be learned and to be done. Some facts are established, but
even for these, fitting the facts into a coherent overarching theory has
proven elusive. This book provides the outlines of one such theory.
Any compelling explanation of suicide should shed at least some
light on various established facts, including prevalence of suicide; the
associations of suicide with age, gender, race, neurobiological indices,
mental disorders, and substance abuse; impulsivity; and childhood
adversity, as well as issues like treatment and prevention efforts and
the clustering and “contagion” of suicide. Each of these topics is de-
fined here and accounted for later, at least in part, by the theory pro-
p
osed in this book.
Definition
One might imagine that defining suicide is relatively easy. Indeed, the
dictionary definition could not be clearer—“the act of killing oneself
intentionally.” This definition seems to apply to my dad and many
others who will be mentioned throughout the book, like the poets
Hart Crane and Weldon Kees and the musician Kurt Cobain. But
what about people on the upper floors of the World Trade Center
who jumped to their deaths on September 11, 2001? At least fifty
people died in this way, and the actual number is probably closer to
200. Did they die by suicide? According to the dictionary definition,
they did, but according to the New York medical examiner and intu-
itively to many of us, they did not. All September 11 deaths at the
World Trade Center were classified as homicides. What about the
28 ● WHY PEOPLE DIE BY SUICIDE
September 11 terrorists, whose actions, in addition to all of their
horrible consequences, caused the terrorists’ own deaths? Did they
die by suicide? Again, according to dictionary definitions, they did.
But the terrorists themselves would more likely have characterized
their deaths as martyrdom or casualties of holy war than as suicide.
Difficulties in defining suicide arise in other situations. Did Mari-
lyn Monroe die by suicide, or was she killed? Virtually all the evi-
dence points to suicide, but the idea of homicide resurfaces, often for
spurious reasons.
dents who are later found to have been intoxicated at the time of
death? We cannot know with certainty whether these deaths were in-
tentional or accidental. One basis on which to make the designation
might involve the facts of the accident, such as the angle at which the
car was driven into a tree or the pattern of skid marks. Someone who
brakes or swerves at the last instant could be viewed as simply having
fallen asleep at the wheel. This is certainly possible, but it is also pos-
sible that someone intended suicide and changed his or her mind too
late. This appears to happen relatively frequently, as seen in cases of
those who jump from high places, survive, and report that they re-
gretted their decision in midair.
There are still other ambiguities regarding the definition of sui-
cidal behavior. One of my adolescent patients took a regular sewing
needle, inserted it in the side of her wrist a millimeter or two, and
immediately told her mother she had attempted suicide. This sce-
nario is of clinical concern (and of course was of great concern to
the mother), but does this qualify as a real suicide attempt? Is it
of the same quality as more serious attempts, such as what my
patient “Gayle” had in mind (severing her hand and bleeding to
death)?
I am aware of no current theory that adequately handles all of
What We Know and Don’t Know about Suicide ● 29
these definitional problems, but in this book I will at least address
each one.
Prevalence
Though rates vary somewhat from year to year, approximately
30,000 people in the United States and more than half a million peo-
ple worldwide die by suicide each year. A useful common metric for
death rate is deaths per 100,000 in the population. The rate of death
by suicide has been between 10 per 100,000 and 15 per 100,000 for
decades. In 2001, suicide was the eleventh leading cause of death
overall in the United States.
On the one hand, 30,000 U.S. deaths per year—one every eighteen
minutes or so—is a lot. On the other hand, relatively speaking, sui-
cide is a rare cause of death. For example, given that a person has
died, the chance that the cause was heart disease or cancer is 52 per-
cent. The chance that the cause of death was suicide is a little over 1
percent.
Suicide is thus a relatively rare form of death, and any compelling
theory should be able to account for this fact. Many theories of sui-
cide run aground on the simple shores of prevalence rates—for ex-
ample, they propose a cause that is very common, yet do not fully ex-
plain why relatively few die by suicide. The theory to be developed in
this book has something to say about the relative rarity of death by
suicide.
Gender
Men are approximately four times more likely than women to die by
suicide; women are approximately three times as likely as men to at-
tempt suicide. This pattern of male lethality is partly related to a ten-
dency toward violent behavior more common in men than women.
Women’s attempts are more frequent but less violent. Two of three
30 ● WHY PEOPLE DIE BY SUICIDE
male suicide victims in the United States die by firearm, as compared
to one of three for women—the most common cause of death by
suicide in women is overdose or poisoning. With one key exception,
men are more likely to die by suicide than women in every country
of the world.
The exception is China, where roughly as many women as men die
by suicide. A persuasive account of death by suicide will need to ex-
plain the overall pattern of male lethality and address the interesting
exception of China.
Suicide “Contagion” and Mass Suicide in Cults
From time to time, completed suicides cluster in space and time. For
example, in a high school of approximately 1,500 students, two stu-
dents died by suicide within four days. During an eighteen-day span
that included the two completed suicides, seven other students at-
tempted suicide. Occasionally a spate of deaths by suicide will occur
following a well-publicized suicide, especially if media coverage in
any way glorifies it. What is the mechanism underlying suicide clus-
ters?
In 1999, I proposed an explanation that involves the concept of
assortative relating, which means that people form relationships
nonrandomly—they assort based on shared interests, characteristics,
even shared problems, such as substance abuse. I believe that suicide
clusters are, in a sense, pre-arranged, in that vulnerable people
assortatively relate and then are simultaneously impinged upon by
some serious stressor, which activates the suicide potential of each
member of the potential cluster.5 I provide some empirical support
for this view in a study of college roommates, some of whom related
assortatively in that they chose to room together, others of whom related randomly, in that they were paired with one another by the uni-
versity housing agency. If my explanation of suicide clusters holds
water, the suicide potentials of roommates who chose to room to-
What We Know and Don’t Know about Suicide ● 31
gether should be more similar than the suicide potentials of those
who were randomly assigned to one another. This was in fact the
finding.6
The Internet provides another, sometimes more pernicious way in
which people can assortatively relate. Consider, for example, the
“pro-suicide” website alt.suicide.hol
iday (or ASH), where suicide is
construed favorably. Visitors to the site are instructed on the best
methods for suicide. As many as twenty-four completed suicides
have been connected to the site. Other web forums have been docu-
mented to encourage self-destructive behavior too. On January 12,
2003, a twenty-one-year-old man died in his bedroom after ingesting
huge amounts of prescription drugs, marijuana, and alcohol. It was
later determined that many people had witnessed the death through
the man’s webcam, and that several of the onlookers had typed
things like, “Eat more!” as the man ingested obviously dangerous
numbers of pills. The man and the onlookers were part of an ongo-
ing chat room that regularly discussed substance use. Excessive sub-
stance use brought these people together (assortatively related them),
and in this case, self-destructive behavior was explicitly encouraged
from within the group. Indeed, experimental studies have evaluated
the connection between group norms and self-aggression. These
studies use a self-aggression paradigm (self-administered shock dur-
ing a task disguised as a reaction-time game, with self-aggression de-
fined by the intensity of shock chosen). In this research, high levels of
self-administered shock occurred when group norms were manipu-
lated to encourage self-aggression.
In his 2003 New Yorker article, Tad Friend documents the death of
a fourteen-year-old girl who left her high school by taxi, rode to the
Golden Gate Bridge, and jumped to her death. The girl’s mother later
discovered that she had been visiting a website that offered advice
about completing suicide and showed graphic autopsy photos. The
site discourages methods like poison, drug overdose, and wrist cut-
32 ● WHY PEOPLE DIE BY SUICIDE
ting, since less than 15 percent of people who attempt by these meth-
ods die. By contrast, the site recommends jumping from a high place
because “jumps from higher than . . . 250 feet over water are almost
always fatal.”
Suicides occasionally do cluster in space and time; concepts like
assortative relating and group norms may partially explain them.
These concepts are starkly illustrated in some of the most horrific
suicide clusters, those occurring in cults. In Jonestown, where 914
people died, the majority died from drinking a grape-flavored drink