Suicide: Can We Predict It?

Richard Balon

Prediction of suicide is the most important factor for suicide prevention and treatment. Methodolog-
ical issues of suicide prediction studies are reviewed as well as demographic, clinical and biological
predictors of suicide. These data suggest that suicide cannot be reliably predicted.
o 1987 by Grune & Stratton, Inc.

UICIDE, the act of killing oneself intentionally, has been a puzzling phenome-
S non throughout the history of mankind. The definition of suicide as either a
behavioral deviation or as a disorder or disease has varied through time.’ The first
documented suicide note is allegedly in an ancient papyrus.* The ancient Hebrews
recorded only five suicides in the Bible (all related to defeat in war); the only suicide
in the New Testament is that of Judas Iscariot.3
Suicide was tolerated as a “social disease” until Christian views began to
influence social and legal attitudes, and suicide was considered a sin. With the
coming of age of enlightenment, the focus moved away from sin and its emphasis on
personal disorder to the impersonal influence of environment on behavior.
Although 1,000 people commit suicide every day, experts do not agree on a
classification of suicide. Suicide has become a focus of interest for anthropologists,
sociologists, philosophers, psychiatrists, psychologists, theologians, and lay people.
Each discipline brings its own view and explanation. The famous French sociologist,
E. Durkheim,4 considered suicide a social phenomenon and classified suicide into
four basic types: altruistic, egoistic, anemic, and fatalistic.5 The father of modern
thanatology, ES. Shneidman,’ suggested that all committed suicides should be
viewed as being one of three types: egotic, dyadic or ageneratic. The International
Classification of Diseases lists suicide in its classification but the Diagnostic and
Statistical Manual of Mental Disorders (ed 3) (DSM-III) does not. The list of
classifications and explanations is quite long and reflects only our poor knowledge of
this phenomenon.
While many have been thinking and theorizing about suicide, detection and
prevention of suicide have become more and more the responsibility of the mental
health professionals, mainly psychiatrists. This is an enormous responsibility which
has become a serious burden, especially in the recent litigious atmosphere.
However, the very nature of suicide makes prediction difficult. Suicide is the end
result of a process, not a process itself. Like most complex behavior, suicide appears
to have multiple determinants. The etiology of suicide is likely to be multifactorial.
In other behavior disorders, the process is ongoing and available for examination. In
suicide, all we usually have is the end result arrived at by a variety of paths.3
Understanding of the development of the phenomenon is very important for
prediction.

From the Lafayette Clinic, Detroit and the Department of Psychiatry, Wayne State University School
of Medicine, Detroit.
Address reprint requests to Richard Balon. M.D., Wayne State University School of Medicine,
Department of Psychiatry, Lafayette Clinic, 951 E Lafayette, Detroit, MI 48207.
o 1987 by Grune & Stratton, Inc.
0010-440X/87/2803-0005%03.00/0

236 Comprehensive Psychiatry, Vol. 28, No. 3 (May/June), 1987: 236-241

SUICIDE PREDICTION 237

There have been two basic approaches to the improvement of our understanding
of suicide. One of them is a statistical approach, collecting usually demographic
data. These data may provide important clues to environmental influences on the
incidence and population distribution of suicidal behavior. On the other hand, the
medical and psychiatric approaches started to study the biology of suicide and the
role of individual psychopathology. This approach is clinical or individual, focused
on individual reasons and endogenous forces.

METHODOLOGICAL ISSUES
Studies of prediction of suicide have to deal with many methodological problems:
1. Definition of the phenomenon. Do we deal with a committed suicide, suicidal attempt, or with a
suicidal gesture? Suicide may be defined as a single lethal act, or more broadly as a pattern of behavior,
such as drug abuse, likely to result in death.
2. Suicide is still a somewhat rare phenomenon.
3. The problem of false-positives and false-negatives. The latter may he a particular problem because
of the social stigma associated with suicide.
4. We need to clarify if we consider all suicidal persons as mentally ill or if we separate “normal”
persons committing suicide and persons suffering from any form of mental illness (affective disorders,
schizophrenia, etc.) and committing suicide.
5. The ethical and practical difficulties of collecting cases of future suicide prospectively prevent us
from designing of a classical study of prediction of suicide.
6. The separation of the short (immediate) term and the long term suicide risk and prediction.
7. The collection, even retrospectively, of representative data from in- and outpatient populations
might be different.
8. The impact of retrospective distortion of evaluators of suicide after the fact6 seems to he an
important issue. Finally, as Murphy’ pointed out, primary prevention occurs both at the level of removing
a substrate of suicidal thinking and at the level of direct interruption of the developed thought. If it is
successful, the patient will live. A suicide will be prevented. Yet to quantify this effect is difficult. It is
important to realize that the absence of a suicide generates no data in the absence of a matched control
group with a predictable rate of suicide. Thus, we can never prove what has been accomplished. Yet we
can hardly doubt it occurs.

DEMOGRAPHIC PREDICTORS
The bulk of literature on suicide prediction deals with demographic predictors.
Demographic data basically delineate population at risk. Most authors agree that
the main demographic risk factors are affective disorder, alcoholism, sex, race, age,
marital status, lack of social support, history of previous suicidal attempts, and
mental illness. Some of these factors are controversial.
Some of the crisis situations triggering suicide are death in the family (or loss of
the key person), infidelity of the partner, breaking up with the partner, loss of
employment, and social failure.
Mental illness is a major demographic factor. Depression has been implicated
most often. Borg et a1.8 in their prospective study of 2,184 inpatients and outpatients
found that from 34 patients who committed suicide, the majority were drug and
alcohol abusers and neurotics with symptoms of depression. Siani et a1.9 found
several items discriminating suicide repeaters and nonrepeaters. Those items were
sociopathy, change of domicile in the last year, previous inpatient psychiatric
treatment, previous parasuicide resulting in hospital admission, unemployment, and
criminal record.
Results of the study by Stallone et al” indicated that female depressives who
score high on social isolation, have a positive family history of suicidal behavior, and
238 RICHARD BALON

are not diagnosed as bipolar, are likely to be attempters. Humphrey et al” tried to
determine sociological and psychological differences of suicidal attempters and
thinkers among suicidal inpatients. They found that attempters tended to be
younger, unmarried, urban residents with higher income and education than
thinkers. Evenson et al.‘* found that the history of psychiatric treatment increases
the suicide risk more for women than for men, although male patients are still about
twice as likely to commit suicide than female patients are.
Noreik13 found that the greatest number of attempted suicides was among
patients with psychoses of a depressive nature. A study by McIntosh14 emphasized
that suicide rates for aged have declined over time but that the elderly are more
successful in suicide attempts than young.
Rosen,” in his study of 886 patients concluded that suicidal patients with
depression or insomnia, or both, plus three or more other high risk signs (older than
40 years, married, recently separated, widowed, retired or living alone, middle class,
good employment record) should be taken seriously, treated, and hospitalized if
necessary. Black et a1.16found that among, 5,142 psychiatric inpatients significant
excesses in suicide occurred among males and females with schizophrenia, affective
disorder and alcohol or drug abuse, among males with neuroses or personality
disorders and among females with depressive neuroses.
Roy in his studyI found that significantly more of the suicides suffered from
chronic schizophrenia (33.3%) or recurrent affective disorder (18.8%) and had
made a previous suicide attempt (46.6%). The suicides were young and significantly
more were unmarried (84.5%), unemployed (66%), living alone (55.5%), and
depressed (65.5%) and 44.4% had a primary diagnosis of depressive disorder.
Computed data in Morrison’s study’* emphasized that personality disorder, alcohol-
ism, multiple diagnoses, and psychoses all were associated with increased risk of
suicide.
Completed suicide is most common in white, elderly male Protestants who live
alone. This is a large group in most communities. According to World Health
Organization (WHO) statistics (1969), the majority of suicides are still married.
The application of even a number of demographic variables will not specify a
person very accurately. Demographic predictors ignore individual characteristics
and are often better long term than short term predictors,” but they seem to fail in
case of accurate individual prediction. One might have to solve the problem like this
in the emergency room: is the man going to kill himself tonight? The fact that the
man was white, Protestant, and living alone is not going to help much to decide
whether to hospitalize the patient.
Some authors, e.g., Barraclough and Shepherd,” mention also the high-risk
period of suicide. An example of this is the “birthday blues”: an excess of death in
the birth month and the following months for persons aged 75 and over was found.

CLINICAL (PSYCHOLOGICAL) PREDICTORS AND SCALES

Clinical factors seem to be more relevant to the individual prediction of suicide.
When talking about clinical prediction we usually mean use of some psychiatric and
psychological data quite often obtained from different scales and tests, but very
often we mean some kind of intuition.
Clinical signs most frequently mentioned in the literature are suicidal ideation or
SUICIDE PREDICTION 239

concrete intention, depression, hopelessness, assaultive behavior, psychosis, marital

or economic crisis, serious physical illness, previous alcohol intake, agitation, and
anger. Hopelessness has been reported as an important sign and scale item.*‘***
Several suicide scales are available3*23-27
as are commonly used psychological tests
and scales validated for prediction of suicidality.
The literature on usefulness of scales and psychological tests in prediction of
suicide is rather contradictory and confusing. Some authors**.*’ did not find the
MMPI to be useful in the prediction of suicide; some authors, e.g., Leonard,30 claim
that the Minnesota Multiphasic Personality Test (MMPI) patterns appear to have
usefulness in predicting suicide potential.
The literature on the Rorschach’s usefulness (31-33) presents the same picture.
Statements that clinicians could not identify suicide and nonsuicide patients from
their MMPI profiles34 and that the Rorschach may be an inappropriate technique
for assessing suicidal activity33 seems to be the most prudent and wise assessment of
the usefulness of these tests.
In their review in 1972, Brown and Sheran3’ indicated that none of the predictors,
e.g., single signs, standard psychological tests, specially devised tests, and clinical
judgments and scales, were able to predict suicide at useful levels. The situation has
not changed much. According to them, scales are shown to offer the best predictive
potential, but their construction has not been systematic.

BIOLOGICAL FACTORS
Biological factors are a new category of suicide predictors emerging during the
last two decades.36+44
Different authors tried to find a specific biological correlate or marker of suicidal
behavior. Results of these studies are again inconclusive, sometimes contradictory
and so far not very useful.
Krieger3’ reported that inpatients who later committed suicide tended to have a
higher mean plasma cortisol level measured at 8:30 AM and he suggested that
cortisol level above 20mg% in the absence of other possible causes for such a high
level should be recognized as a sign of high risk of suicide. Asberg et a13’ found that
patients with lower 5-HIAA in the cerebrospinal fluid attempted suicide more often
than those with higher levels and they used more violent means. Ostroff et a1.42
considered a low norepinephrine-to-epinephrine ratio a risk factor for suicidal
behavior. They also found a higher 24-hour urinary cortisol level in suicidal
patients.43 Ennis et a1.44did not find the dexamethasone suppression test (DST) test
to be useful in identifying suicidal patients and they warned that routine use of DST
in such patients could be highly misleading.
Basically, none of the attempts to find a biological correlate of suicidal risk or
behavior has earned clinical significance.

CONCLUSION
The presented overview of literature revealed a rather negative picture for the
prediction of suicide. This overview does not intend to discourage mental health
professionals, but merely presents a realistic picture of the recent state of art.
Prediction of an individual suicide is quite difficult and unreliable at the present
level of knowledge. There are many methodological obstacles for suicide studies,
240 RICHARD BALON

and even well-designed studies have limited utility for the prediction of an
individual suicide. A prospective study confirming our ability to predict suicide is
also impossible from an ethical point of view.
There are three groups of suicide prediction factors: demographic, clinical, and
biological. Demographic factors do not seem to be very useful in that the
demographic factors might not even be relevant to the “true” prediction of suicide.
But this might be caused by our poor understanding of these factors. Clinical factors
might be more useful but they are not able to predict suicide at significant levels.
Suicidal scales, closely connected with clinical factors, have not been found very
useful. None of the biological factors reached the level of clinical acceptance and
relevance.
We can conclude that suicide is a violent behavior with a relatively low incidence
which we are unable to predict. Reliable prediction of this phenomenon is out of
question and may never be possible, partly because of methodological difficulties
inherent to suicide studies, and partly due to the complexity of suicidal behavior.

ACKNOWLEDGMENT
The author is grateful to Robert B. Pohl, M.D. for reviewing of this manuscript and for his valuable
comments.

REFERENCES
1, Thomas C: First suicide note? Br Med J 261:284-285, 1980
2. Rosen Cl: History in the study of suicide. Psycho1 Med 1:267-285, 1971
3. Zubin J: Observations on nosological issues in the classification of suicidal behavior, in Beck AT,
Resnik HLP, Lettieri DJ (eds): The Prediction of Suicide. Bowie, MD, Charles Press, 1974, pp 3-25
4. Durkheim E: Suicide, Spaulding JA, Simpson G (trans). Glencoe, IL, Free Press, 1951
5. Shneidman ES: Classifications of suicidal phenomena. Bull Suicidology 2: 1-9, 1968
6. Hood RW Jr: Effects of foreknowledge of death in the assessment from case history material of
intent to die. J Consult Clin Psycho] 34: 129-133, 1970
7. Murphy GE: The prediciton of suicide: Why is it so difficult? Am J Psychother 38:341-349, 1984
8. Borg SE, Stahl M: Prediction of suicide: A prospective study of suicides and controls among
psychiatric patients. Acta Psychiatr Stand 65:221-232, 1982
9. Siani R, Garzotto N, Zimmermann-Tansella C, et al: Predictive scales for parasuicide repetition:
Further results. Acta Psychiatr Stand 59:17-23, 1979
10. Stallone F, Dunner DL, Ahearn J, et al: Statistical prediction of suicide in depressives. Compr
Psychiatry 21:381-387, 1980
I 1. Humphrey JA, Nishwander GD, Casey TM: A comparison of suicidal thinkers and attempters:
Interim findings. Dis Nerv Syst 32:825-830, 1971
12. Evenson RC, Wood JB, Nuttall EA, et al: Suicide rates among public mental health patients.
Acta Psychiatr Stand 66:254-264, 1982
13. Noreik K: Attempted suicide and suicide in functional psychoses. Acta Psychiatr Stand
52:81-106, 1975
14. McIntosh JL: Suicide among the elderly: Levels and trends. Am J Orthopsychiatry 55:288-293,
1985
15. Rosen DH: The serious suicide attempt: Epidemiological and follow-up study of 886 patients. Am
J Psychiatry 1271764-770, 1970
16. Black DW, Warrack G, Winokur G: The Iowa record-linkage study. I. Suicides and accidental
deaths among psychiatric patients. Arch Gen Psychiatry 42:71-75, 1985
17. Roy A: Risk factors for suicide in psychiatric patients. Arch Gen Psychiatry 39:1089-1095, 1982
18. Morrison J: Suicide in psychiatric practice population. J Clin Psychiatry 43:348-352, 1982
19. Lester D: Demographic versus clinical predictors of suicidal behavior: A look at some issues, in
Beck AT, Resnik HLP, Lettieri DJ (eds): The Prediction of Suicide. Bowie, Maryland, Charles Press,
1974, pp 71-84
SUICIDE PREDICTION 241

20. Barraclough BM, Shepherd DM: Birthday blues: The association of birthday with self-inflicted
death in the elderly. Acta Psychiatr Stand 54:146-149, 1976
21. Beck AT, Steer RA, Kovacs M, et al: Hopelessness and eventual suicide: A IO-year prospective
study of patients hospitalized with suicidal ideation. Am J Psychiatry 142:559-563, 1985
22. Wetzel R: Hopelessness, depression and suicide intent. Arch Gen Psychiatry 33:1069-1073,
1976
23. Pallis DJ, Barraclough BM, Levey AB, et al: Estimating suicide risk among attempted suicides. I.
The development of new clinical scales. Br J Psychiatry 141:37-44, 1982
24. Pallis DJ, Gibbons JS, Pierce DW: Estimating suicide risk among attempted suicides. 11.
Efficiency of predictive scales after the attempt. Br J Psychiatry 144: 139-l 48, 1984
25. Pierce DW: The predictive validation of a suicide intent scale: A five year follow-up. Br J
Psychiatry 139:391-396, 1981
26. Buglass D, Horton J: A scale for predicting subsequent suicidal behaviour. Br J Psychiatry
124:573-578, 1974
27. Resnick JH, Kendra JM: Predictive value of the “Scale of Assessing Suicide Risk” (SASR) with
hospitalized psychiatric patients. J Clin Psycho] 29:187-190, 1973
28. Watson CG, Klett WG, Walters C: Suicide and the MMPI: A cross validation of predictors. J
Clin Psychology 40:115-l 19, 1984
29. Clopton JR, Jones WC: Use of the MMPI in the prediction of suicide. J Clin Psychology 3 I :52-54,
1975
30. Leonard CV: The MMPI as a suicide predictor. J Consult Clin Psycho1 45:367-377, 1977
3 1. Kestenbaum JM, Lynch D: Rorschach suicide predictors: A cross validation study. J Clin Psycho]
34:754-758, 1978
32. Exner JE, Wylie J: Some Rorschach data concerning suicide. J Pers Assess 41:339-348, 1977
33. Neuringer C: Suicide and Rorschach: A rueful postscript. J Pers Assess 38:535-539, 1974
34. Clopton JR, Baucom DH: MMPI ratings of suicide risk. J Pers Assess 43:293-296, 1979
35. Brown TR, Sheran TJ: Suicide prediction: A review. Life-Threatening Behav 2:67-98, 1972
36. Krieger G: Biochemical predictors of suicide. Dis Nerv Syst 31:478-482, 1970
37. Krieger G: The plasma level of cortisol as a predictor of suicide. Dis Nerv Syst 35:237-240, 1974
38. Bunney WE, Jr, Fawcett JA: Possibility of a biochemical test for suicidal potential. Arch Gen
Psychiatry 13:232-239. 1965
39. Asberg M, Trlskman L, Thoren P: 5-HIAA in the cerebrospinal fluid: A biochemical suicide
predictor? Arch Gen Psychiatry 33:1193-1197, 1976b
40. Trlskman L, Asberg M, Bertilsson L, et al: Monoamine metabolites in cerebrospinal fluid and
suicidal behavior. Arch Gen Psychiatry 38:631-636, 1981
41. Van Praag HM: Depression, suicide and the metabolism of serotonin in the brain. J Affective
Disord 4:275-290, 1982
42. Ostroff RB, Giller E, Harkness L, et al: The norepinephrine-to-epinephrine ratio in patients with a
history of suicide attempts. Am J Psychiatry 142:224-227, 1985
43. Ostroff RB, Giller E, Bonese K, et al: Neuroendocrine risk factors of suicidal behavior. Am J
Psychiatry 139:1323-1325, 1982
44. Ennis J, Barnes RA, Kennedy S: The Dexamethasone Suppression Test and suicidal patients. Br J
Psychiatry 147:419-423, 1985