In the year 2000, the
National Center for Patient Safety asked all local VA Patient Safety
Committees to track patient parasuicidal behaviors at their respective
facilities and then to suggest interventions to reduce the frequency
of these behaviors. This request generated 94 aggregated root cause
analyses for parasuicidal behavior that were submitted by 59 different
VA facilities between October 1, 2000, and March 31, 2003. The
report that synthesized this data was entitled Parasuicidal Aggregate Review
and Selected Suicide Root Cause Analysis: Actions and Implementation
Strategies to Reduce Suicidal Events in the VA System: Final report.
It concluded that “the reduction of suicidal and parasuicidal
behavior is a challenge because it is under-reported and there are a
variety of definitions of parasuicide, making outcome difficult to measure”
(p. 28). The VA’s National Center for Patient Safety had thus
encountered what those engaged in the study of suicidal and self-injurious
behavior have struggled with for decades if not centuries; or, in the
words of O’Carroll, et al, (1996),
Despite hundreds of years of writing and
thinking about suicide, and many decades of focused suicide research,
there is to this day no generally accepted nomenclature for referring
to suicide-related behaviors, not even at the most basic conversational
level (p. 238).
The term parasuicide was first suggested
by Kreitman, et al, (1969) when he and others grappled with the observation
that the term attempted suicide was being used to refer to patients
who injured themselves, but, in fact, were not attempting suicide. Kreitman
thought of the term parasuicide as referring to self-injurious behavior
that simulates or mimics attempted suicide, but is not suicidal in intent.
His hope seemed to be that a variety of behaviors, sometimes referred
to as suicidal gestures, manipulative suicide attempts, and/or self-mutilation,
could be categorized as parasuicidal events and that this classification
would help us to distinguish those behaviors that were more truly suicidal.
The concept parasuicide, however, was never more clearly defined, and
it became used by different investigators with different meanings and
inclusive of different behaviors. Just as an example, Linehan (1993)
used it to refer to actual suicide attempts as well as to selfinjuries
such as self-mutilation, while Diekstra and Garnefski (1995) excluded
habitual self-injury (i.e., self-mutilation) from their definition.
Why have these behaviors
been so difficult to define and to study? First of all, human
intentionality is often complex, mixed, and fluid. With suicidal
or self-injurious behavior, there can be degrees of intent to die,
intent to influence others, and intent to regulate self. Second,
research on suicidal behavior seldom, if ever, includes patients
who are clearly suicidal. If a study subject becomes actively suicidal,
ethics demands that we intervene to save the patient and remove
him or her from the study. Third, we are attempting to study an
exceedingly rare event. In 2002, for example, suicide had a base
rate of 11.0 per 100,000 in the general U.S. population. Estimates
of the suicide rate in the psychiatric population are about five-fold
that of the general population, or approximately 55 per 100,000.
We don’t have the clinical acumen
or assessment instruments to accurately detect such rare events
and to know who the truly suicidal are.
Given these difficulties
in attempting to know and understand suicidal and self-injurious
behaviors, the Suicidal and Parasuicidal Aggregate Review Committee
at VA Boston decided to forego an “a priori” category
system and to launch a quality improvement-type investigation of
the range of self-injurious behaviors that we see in our healthcare
system. We have taken self-injurious behavior to mean any intentional,
self-inflicted, physical harm to the individual. The behaviors
that are included as self-injurious can range from the individual
who inflicts a superficial scratch to relieve tension to the individual
who takes a lethal amount of medication and is accidentally found
unconscious. We also track any completed suicides in the system.
Since instances of self-harm or self-injury are recorded in the
morning reports and in medical center incident reports, it is possible
to conduct a post self-injury debriefing interview within a few
days of the particular event. We have developed a semi-structured
debriefing interview that is based on the available evidence in
the literature on characterizing and assessing the seriousness
of suicidal and self-injurious behavior. It is available as an
overprinted form in the electronic progress note package of the
VA Boston medical record. Fundamental to the interview is that
self-injurious behaviors are discriminated from each other by the
intent of the individual. Thus, there is an effort in the interview
to obtain a measure of intent to die and a measure of the perceived
lethality of the act.
We have piloted the
debriefing interview over the past year. It consists of three sections.
In Section I, the patient is asked for (1) the date, time, and
day of the week when the event occurred, (2) a brief description
of what he or she did, and (3) a Likert scale rating of his or
her intent to die (ranging from “6
“– definitely wanted to die to “0” – no
intent to die). If the patient gives a rating of greater than zero
intent to die, the interviewer completes Section II (Self-Injury
with Intent to Die) with the patient. If the patient gives a rating
of zero intent to die, the interviewer skips to Section III (Self-Injury
with No Intent to Die).
In Section II of the
interview, the pt is asked: (1) to state a reason for his or her
self-injury; (2) to describe the method used; (3) to give a Likert
scale rating of perceived lethality (ranging from “6” – definitely thought it would be
lethal to “0” definitely thought it would not be lethal);
(4) to state how much physical damage was done; (5) to state whether
consciousness was lost or disturbed; (6) to indicate what level of
medical response was needed to deal with the injury (admission to
ICU, treatment in Urgent Care Clinic, etc.); (7) to state whether
he or she was under the influence of alcohol or drugs at the time;
(8) to indicate if he or she had made a plan or if this action was
an impulsive decision; (9) to state whether he or she had told others
of his or her suicidal or self-injurious thoughts; (10) to indicate
if other people were close by and/or likely to contact him or her;
(11) to state whether he or she hoped to be found or saved; and (12)
to report how he or she feels about the outcome.
As noted above, if the patient indicates
that he or she had no intent to die, the interviewer skips to Section
III. Section III has a number of items in common with Section II (i.e.,
a reason for the self-injury, the method used, how much physical damage
was done, whether consciousness was lost, the level of medical response,
and whether the patient was under the influence of alcohol or drugs
at the time). Section III deviates from Section II, however, in that
the patient is asked: (1) to indicate if this self-injury was a single
event or part of a cluster of acts (e.g., cutting, burning, etc.); (2)
if it was part of a cluster, how many acts were in the series and over
what period of time; and (3) if it was a single event, how many times
these isolated events have occurred in the past.
In the pilot year with
the debriefing interview, we have decided that it will be necessary
to add two items to the interview. For comparative purposes, we
are adding an item that asks for the interviewer’s
rating of the patient’s intent to die. This item will also
allow us to obtain a measure of inter-rater reliability. We are adding
a second item that asks the interviewer to indicate if he or she
felt that the patient gave reliable information, or, if not, what
the reason was that it was not reliable (e.g., psychosis, sedation,
Within the first year of this quality improvement
project, we have collected data on 40 instances of self-injury in our
healthcare system. We are beginning to analyze this initial data. It
is hoped that in the future, with our now refined post self-injury interview,
we will be able to generate information about such things as the frequency
of suicide attempts in our system in which there was serious intent
to die, the frequency of self-injuries in which there was a lesser intent,
the frequency of self-injuries that had no suicidal intent, the methods
of self-injury that we are likely to encounter, the extent of injuries,
how frequently alcohol or drugs may contribute to self-injurious behavior,
what sorts of self-injuries were likely to be planned and what sort
were more likely to be impulsive, and so forth. From a quality management
perspective, it is hoped that this improved method of tracking and monitoring
self-injurious behavior will result in the discovery of systemic changes
that may reduce its incidence, particularly the incidence of those behaviors
that result in serious damage to the individual. In the future, I hope
to report further on the results of this endeavor.
Diekstra, R., and Garnefski, N. (1995).
On the nature, magnitude, and causality of suicidal behaviors: An international
perspective. In M. Silverman and R. Maris (Eds.): Suicide
Prevention: Toward the year 2000. NY: Guilford Press.
Kreitman, N., Philip, A., Greer, S., and
Bagley, C. (1969). Parasuicide. British Journal
of Psychiatry, 115,
Linehan, M. (1993). Cognitive-behavioral
treatment of borderline personality disorder. NY: Guilford Press.
Mills, P., Neily, J., Luan, D., Osborne,
A., and Howard-Hirschler, K. (2004). Parasuicidal
aggregate review and selected suicide root cause analysis: Actions
and implementation strategies to reduce suicidal events in the VA
system: Final report. Unpublished
manuscript, VHA National Center for Patient Safety.
Berman, A., Maris, R., Moscicki, E., Tanney, B., and Silverman, M.
(1996). Beyond the Tower of Babel: A
nomenclature for suicidology. Suicide and Life-Threatening
Behavior, 26, 237-252.