Rethinking Suicide Risk Assessment and Risk Formulation

Section 1
Chapter
Rethinking suicide risk assessment
4 and risk formulation

Alan L. Berman and Morton M. Silverman
A thorough SRA is a necessary precursor to an SRF.

Introduction The SRF is a more abstruse process forged by clinical
Approximately one in three American, and one in four judgment about the dynamic contributions of this data
British, citizens contact a mental health care provider to determine the patient’s level of risk that, then,
or service in the year prior to dying by suicide (Luoma dictates safety planning, triage decisions, and treat-
et al., 2002; Appleby et al., 2012). About one in ten ment planning and management.
patients are seen in emergency departments within 2
months of dying by suicide (Knesper, 2010). A con-
servatively estimated 31.3% of individuals who died by Suicide risk assessment
suicide were receiving mental health treatment at the No method of SRA has been empirically tested for
time of their death (Centers for Disease Control and reliability and validity (Busch et al., 2003) or has suffi-
Prevention, 2012). cient sensitivity and specificity to be effective (Simon,
Clearly, health care providers have significant 2012). Yet, the standard of care1 requires mental health
opportunity to identify at risk individuals and engage professionals to recognize and foresee the possibility
them in treatment to reduce the risk of suicidal behav- that a patient might engage in suicidal behavior, hence
ior, yet many lack the requisite training and skills to to conduct an SRA and make an SRF sufficient to
appropriately assess for suicide risk (Schulberg et al., guide triage and treatment planning.
2004; Schmitz et al., 2012). Tragically, the Joint The SRA, ideally, incorporates separate but inte-
Commission’s Sentinel Event Report denotes failures grated components, the first two of which are:
in conducting an assessment to be the leading root 1. Identifying perpetuating (non-modifiable),
cause of hospital-based suicides, associated with 84% predisposing (modifiable), acute (warning signs),
(N = 502) of the 600 suicides reported between 2004 precipitating (triggering), contributing, and
and 2011(www.jointcommission.org). protective risk factors (see www.suicidology.org)
We see the assessment process as encompassing 2. Exploring the presence/absence of suicidal
two key steps: first to determine whether an individual ideation, motivations, intent, planning, and
is at risk for suicide; second to determine the level of behavior.
risk. These steps then inform appropriate interven-
tions (ideally evidence-based) to reduce the level of
risk and clinical judgments about modifying or tailor- A focus on suicide ideation
ing interventions to the specific individual and setting. The presence of communicated SI has traditionally
This chapter addresses suicide risk assessment been the gateway to the investigation of whether an
(SRA) and suicide risk formulation (SRF) commonly individual might be “suicidal.” Hence the normative
employed by mental health practitioners. We will chal- approach to an SRA is to first inquire about the pres-
lenge the current thinking and the lack of an evidence ence of SI. The unfortunate corollary has been the
base that appear to guide these processes. clinical assumption that the absence of SI is
An SRA involves the collection of data, of observ-
able and reported factors associated with suicide risk 1
The standard of care is generally defined as that duty to exercise
and ascertained by mental health interview, collateral that degree of skill and care ordinarily employed in similar circum-
information gathering, screening tools, and/or scales. stances by members of the same profession.
A Concise Guide to Understanding Suicide, ed. Stephen H. Koslow, Pedro Ruiz, and Charles B. Nemeroff. Published by
Cambridge University Press. © Cambridge University Press 2014.
Downloaded from https://www.cambridge.org/core. University of Birmingham, on 18 Nov 2019 at 12:11:16, subject to the Cambridge Core terms of use, available at 33
https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9781139519502.006
Section 1: Understanding Suicide
tantamount to the absence of any suicide risk, hence (aged 15–54) was 13.5%. The probability of transition-
no need to pursue a line of questioning that might ing from SI to plan was 34%, from a plan to an attempt
reveal the presence of other factors that help define was 72%, and from SI to an unplanned attempt, 26%.
presence of risk. As noted by Simon (2012), “the extent Identifying which persons with SI are at greatest risk of
of suicide risk assessment usually is no more than the progressing to a suicide attempt, however, has been
statement ‘Patient denies HI, SI, CFS (homicidal idea- elusive (May et al., 2012).
tion, suicidal ideation, contracts for safety).’”
If a patient states that SI is present, the clinician is Does the presence of suicide ideation
obligated to explore SI further regarding questions of
intent, content (active thoughts of suicide versus passive predict anything?
wishes for death; planning), duration, frequency, inten- Hence,
sity, controllability, etc. Received wisdom has dictated * the majority of individuals reporting SI do not go
that if SI is deemed to be active and with intent or plans, on to make a plan (~66%)
risk should be judged to be greater than if SI is chronic, * SI with plan is no more predictive of a suicide
passive and/or without intent (Rudd et al., 2006). attempt than is SI without a plan (~19–25%)
The reality is the following: * the majority of individuals reporting SI do not go
* Among those who die by suicide, SI is more likely on to attempt suicide (~66%). However, within a
to be denied than admitted to upon clinical inquiry year of onset of SI, there is an increased risk for
conducted reasonably proximate to the date of suicide attempts.
death (see below).
Fawcett et al. (1990) followed a mostly inpatient sam-
* Active SI is no more associated with death by ple of over 950 patients for 10 years and found that
suicide than is passive SI (Simon, 2006). hopelessness and SI were significantly associated with
* The association between currently expressed SI and suicide in those patients who died by suicide (N = 32),
subsequent death by suicide is less clear (Beck et al., but more than a year after the intake evaluation.
1999). Hence, it appears that acknowledgment of SI is a
more significant chronic risk factor for eventual
Determining the presence of suicide death by suicide than it is an acute (near-term) risk
ideation factor (Rudd et al., 2006).
Determining the presence of suicide ideation depends Moreover, the presence of current SI may not be
upon how the question is asked, how the respondent the most important focus in an SRA. Beck et al. (1999)
understands the question, and whether the respondent found that a retrospective report of SI at the worst
feels safe in providing an honest answer. point in a patient’s life was a better predictor of even-
tual death by suicide than was current SI or
SI in the past year hopelessness.
The prevalence of SI in the past 12 months in the general
population (18 years of age or older) varies from 3.3% to Does the denial of suicide ideation
3.8% (Kessler et al., 2005; www.samhsa.gov/data/ predict death by suicide?
NSDUH/2k10State/NSDUHsae2010). However, how
It is common sense to believe that suicidal behavior
the question gets asked determines what responses are
must be preceded by SI – and it may well be – but in
given. Asking questions that are imprecise, vague, and
fact the majority of patients who die by suicide actually
non-specific does not help identify those at risk for deny having suicidal thoughts when last asked prior to
suicidal behaviors in the long- or near-term. For exam-
their death (or communicate their risk behaviorally)
ple, the response to the question, “Have you seriously
(Isometsa et al., 1995; Hall et al., 1999; Busch et al.,
considered ending your life?” will be determined by how
2003). As but one example, the US National Violent
a respondent interprets the term “seriously.”
Death Reporting System informs us that only 28.3% of
almost 10 000 persons who died by suicide disclosed
SI across the lifespan their intent to die by suicide (Karch et al., 2012).
Kessler et al. (1999) reported that the lifetime preva- For the clinician, inquiring about SI is seen as the key
lence of suicide ideation in the general population data point in the decision to admit or discharge a patient,
34
Downloaded from https://www.cambridge.org/core. University of Birmingham, on 18 Nov 2019 at 12:11:16, subject to the Cambridge Core terms of use, available at
Chapter 4: Rethinking suicide risk assessment and risk formulation
extend hospitalization, or refer a patient to specialized detection of SI in primary care is abysmally low (cf.
mental health treatment. But, for the patient, denying SI Schulberg et al., 2004). Rates of inquiry about SI, even
may be seen as essential to maintaining autonomy and when the patient presents with symptoms of major
control over one’s life and choices, avoiding hospital- depression, remain low (Feldman et al., 2007).
ization, facilitating discharge, and a consequent freedom Similarly, Claassen & Larkin (2005) found that SI
to act on a desire and intent to die by suicide, etc. was common in emergency department (ED) patients
Similarly, taking active precautions against discovery who present for medical disorders. In a study of 1590
during an attempt is an important predictor for eventual ED patients, 11.6% reported SI. However, there is wide
death by suicide (Wenzel et al., 2011). variance between and within EDs surrounding suicide
Accordingly, it is incumbent upon the clinician to risk screening and assessment and most patients who
go beyond a cursory exploration of SI to be reassured present to the ED are not assessed for suicide risk, due
about the level of risk. Quite simply stated, a “No” to time constraints, inadequate training, and a lack of
response to a clinical query regarding the presence of screening instruments (Larkin et al., 2009).
SI does not suffice to constitute an SRA (Simon, 2012). In summary, simply inquiring about SI does not
constitute an adequate SRA. Terminating the SRA or
indicating that risk is low because SI is denied are tragic
Is there a difference between “active” mistakes too often made by clinicians. Correspondingly,
and “passive” SI? If there is, does the the presence of SI does not indicate high-risk for suicide
difference make a difference? and active SI is no more associated with risk for death by
suicide than is passive SI.
SI may be described as passive (“I wonder what it would
be like if I were dead”) to active (“I’m thinking of killing
myself”). Nevertheless, there are no clear definitions or A note about screening
guidelines to differentiate active vs. passive SI. Both the Joint Commission and the American
When SI extends to suicide planning and/or prepa- Academy of Pediatrics have called for suicide screen-
ration, the level of risk is presumed to be greater, i.e., the ing for all patients in health care settings (www.join-
more detailed and specific the plan, the greater the risk tcommission.org; American Academy of Child and
(American Psychiatric Association, 2003). However, this Adolescent Psychiatry, 2009). Screening is not synon-
guideline has developed an unintended corollary, i.e., ymous with assessment. As commented upon by
when passive ideation is expressed, clinicians see less Simon (2009, p. 292), “forms tend to be an event,
risk than when active ideation is expressed, often mini- whereas clinical assessment is a process.”
mizing the risk altogether. A screening tool is a brief instrument, typically
This thinking carries over to often used scales. The consisting of a list of evidence-based risk (and,
HAM-D, for example, has a suicide item that scales possibly, protective) factors, designed to provide a
“suicidal ideas” (whatever these may be) at a score of short-cut to identify patients at risk for suicide.
“3,” hence more significant than “wishes he were dead” at Because these tools are brief, they, by necessity,
a score of “2,” which is more significant than “feels life is must exclude a large number of chronic and acute
not worth living” at a score of “1” (Hamilton, 1960). factors associated with elevated risk for suicide.
There is simply no empirical support for this supposi- Most often, their focus is on the presence of suicide
tion, hence the difference between active and passive sui- ideation and behavior, with little, if any, attention
cide ideation has no clinical utility (Szanto et al., 1996). paid to the variety of symptoms, behaviors, and
other signs of increased and acute risk. They are
SI in the primary care setting and rarely tested for reliability and validity; too often are
scored, as if the derived score has some validated
emergency department relationship to a reasonable judgment of risk; and
The primary care physician is typically the first, and are widely variable, institution to institution, as
often the last, medical provider visited by those who there is no universally accepted instrument.2
die by suicide (Luoma et al., 2002; Schulberg et al.,
2004). 2
The Joint Commission makes each facility responsible for devel-
Although SI is present in 2–7% of all primary care oping its own screening protocol (www.jointcommission.org/
patients (Olfson et al., 1996), the rate of inquiry and assets/1/18/NPSG_Chapter_Jan2013_HAP.pdf)
The function of a positive score on a screening attempts) essentially hinge on worst-point lifetime
instrument is solely to alert the clinician to the need suicide ideation and lifetime history of severe suicide
for a more extensive SRA. A negative score on a ideation (Posner et al., 2011). As these measures of SI
screening instrument should not alleviate the need are strongly correlated with suicide attempts, these
for an SRA, if there is any indication of potential risk findings are expected. Notably, the C-SSRS makes no
for suicide. In short, clinicians need to have a working claims to predict risk for death by suicide.
understanding of and sensitivity to acute risk factors The C-SSRS risk assessment version is a checklist
for suicide that should automatically trigger an SRA, of empirically based risk and protective factors, typ-
regardless of how someone “scores” on a screening ical of many other checklists. The C-SSRS screening
instrument. tool is entirely based on past week and lifetime obser-
Moreover, screening instruments can generate vations of SI (rated on a 1–5 scale with active SI
false negatives, in part because they rely on the indi- scoring as more severe than passive SI) and suicidal
vidual to self-report responses to often ill-defined, behavior. Of interest, the screen differentiates suici-
undefined, and/or idiosyncratically defined terms; in dal behavior into classifications inclusive of actual
part because individuals at risk may not respond with attempts, interrupted or aborted attempts, prepara-
full and accurate disclosure because of their pathology, tory behaviors, and non-suicidal self-injurious
their embarrassment, etc. Of course, these same bar- behaviors. The categories of recent preparatory
riers may present in a face-to-face SRA, but the clini- behaviors, interrupted, and aborted attempts are
cian has the advantage of a trained eye and ear to claimed to be associated, empirically, with increased
discern and question non-verbal cues, to gently con- risk (http://www.commondataelements.ninds.nih.gov/
front denials, etc. in the face of other available evidence doc/CSSRS%20Additional%20Information.pdf), but
of possible risk. upon closer examination are based on studies with
The Suicide Ideation Questionnaire (SIQ; Reynolds, very small Ns (recent preparatory behaviors: N = 4
1998) and, more recently, a variation on the SIQ, the Ask cases out of 30 [G. Brown, personal communication,
Suicide Screening Questions (ASQ) (Horowitz et al., January 3, 2013]; interrupted attempts: N = 6 cases
2012), are two examples of such rapid screening tools, out of 41 [Steer et al., 1988]; and aborted attempts:
the former developed with the intent to measure the N = 22 cases out of 71 [Barber et al., 1998]). Clearly,
severity of SI and the latter, mostly a briefer version of more research with larger datasets is called for in
the former, suggesting that it has predictive value regard- this regard.
ing risk for suicide. This claim, however, has been ques- Simon (2009) has written forcefully about the pitfalls
tioned, essentially, again, on the limited available data on of screening, arguing, appropriately we believe, that
the relationship between SI and death by suicide assessment forms and checklists cannot – and should
(Wintersteen et al., 2013). not – replace a face-to-face clinical risk assessment.
Other screening instruments, primarily developed Moreover, the US Preventive Services Task Force has
for the retrospective classification of suicidal events in found insufficient evidence that screening for suicide
clinical trials increasingly have been widely promoted risk in the primary care setting reduces suicide attempts
as prospective clinical measures of suicide risk. Two or deaths by suicide (O’Connor et al., 2013).
examples of these are the Sheehan Suicidality Tracking
Scale (Sheehan-STS: Coric et al., 2009) and the
Columbia-Suicide Severity Rating Scale (C-SSRS: A brief guide to a better assessment
Posner et al., 2011). of suicide ideation
The STS was designed to track treatment-emergent First: Ask about SI:
suicide ideation and behavior and has been found to be
more sensitive at this than the screen for these obser- * Always frame and ask (and document) at least two
vations that historically served this function, item 3 different questions regarding the presence of SI and
from the HAM-D. ask patients to explain their responses.
The C-SSRS, similarly, was developed to track * Always ask about SI at the worst point in the
changes in SI and behavior in the Treatment of patient’s life.
Adolescent Suicide Attempters (TASA) study (Brent * Refrain from using vague temporal references such
et al., 2009). Predictive validity data (for suicide as “recently.”
36
* Be specific about what you are asking the patient. Surprisingly, there is little in the clinical literature
Define what you mean by “suicidal thoughts” or that offers a model for an SRF, i.e., a guideline for
“thinking of killing yourself.” establishing levels of risk. What was noted by Motto in
* Explore, understand, and document expressed SI – 1989 remains true almost a quarter of a century later:
its frequency, intensity, duration, temporality,
“. . .to date we have no established and generally
intent, controllability, context, motivation for, etc.
accepted procedure to guide us . . . Innumerable
* Explore the patient’s history of impulsivity or decisions regarding risk are made and imple-
predisposition to be poorly controlled (e.g., mented every day – the job gets done – but how
alcoholic). Any type of SI, in the context of poor it gets done is determined primarily by the skills
impulse control, may be just as lethal as SI with and philosophy of the individual clinician.”
resolved plans and preparations. (Motto, 1989, p. 245).
* If you choose to categorize SI as passive versus
The detection of suicide risk (SRA) and formulation of
active, do not assume that one is more, versus less,
suicide risk (SRF), and the consequent triage and treat-
indicative of future suicidal behavior.
ment decisions that are dependent on these procedures
Second: If SI is denied: are, perhaps, the most significant actions a clinician must
make. The failure to reasonably accomplish these tasks
* If motivation (unfulfilled need states) for leaving
has the potential for significant negative outcomes for
life is present, continue to conduct an SRA.
both the patient (a possibly preventable death by suicide)
* Use multiple screening methods, ranging from self-
and the clinician (the premature death of a patient; a
report scales to computer-administered screens to
preventable tort action). It is time to move beyond the
direct psychiatric interview.
intuition and philosophy of the clinician and to develop
* Ask, “When you were at your worst point and train to models for conducting an SRA and SRF.
psychologically, did you have any thoughts of
suicide? If so, what was going on at that time and
what kind of thoughts did you have?” What is meant by “level
* Confront (gently) the denial of SI, when other signs of suicide risk”?
of acute risk are present. Traditionally, the clinician evaluates a patient’s risk on
* It is only appropriate to accept a denial of SI as a an ordinal scale ranging from No Risk to High-Risk;
sufficient SRA when there is no presence of and, within this last category, determines whether “High-
predisposing risk factors or acute risk factors, Risk” might be “Imminent.” Standardized, operational
independent reports of suicide-related activities, definitions of these levels of risk do not exist, and some
and/or potential motivation for suicide. question the concept of “Imminent Suicide” (Simon,
2006). What is generally understood by the clinician is
Suicide risk formulation merely that higher versus lower levels of judged risk
The data collected via an SRA informs an SRF. The carry more significant demands for aggressive treatment
SRF requires an understanding of how risk factors planning, triage, and intervention (and potentially civil
interact, exacerbate, and otherwise fuel heightened or commitment).
lowered risk for suicide, no less how they inform a Imminent risk. Imminent risk determinations are
clinical formulation about level of risk. clinical and temporally related predictions of behavior
The clinical literature abounds with reference to an in the ill-defined near future.
SRF as an intuitive enterprise. Motto (1992) wrote about In the USA, statutes for assisted treatment (invol-
formulating suicide risk, “. . .there is no avoiding sub- untary hospitalization) vary considerably, but, typically,
jectivity. . .leaving us with what I have referred to above employ language to guide the clinician as follows:
as an intuitive judgment” (p. 633). Maltsberger (1992) * Real and present threat of substantial harm
wrote, “Because no one knows how to do this [integrate a * Likely to injure if not thwarted
large mass of data] precisely and with empirical certainty, * . . .in the reasonable future
the clinician is forced to rely on clinical experience and
on inductive reasoning. He does not know how impor- As Simon (2006) wrote, this near-term prediction of
tant each datum may be in comparison to the others that imminent risk “imposes an illusory time frame on an
have been collected” (p. 38). unpredictable act” (p. 296).
Just how reliable are judgments of risk? Cahill and Imminent risk may be determined if an individual
Rakow (2011) presented 35 hypothetical case examples states (or is reported to have stated by a person
to seven practitioners and found all but two potential risk believed to be a reliable informant) both a desire
factors were related to risk and priority judgments. They and intent to die and has the capability of carrying
through his/her intent.
noted that “low risk” might be particularly subject to
(Joiner et al., 2007).
variability in its interpretation, noting that the term
“risk” could reflect “the probability of occurrence for This model asserts that desire, intent, and acquired
some,” and “the severity of occurrence to others.” It capability, in a context of low buffers or connected-
should come as no surprise that Appleby et al. (2012) ness, are associated with imminent risk. While the
reported that more than 80% of patients who die by individual components of these factors have support
suicide are seen by their clinical teams as having no or for their association with greater lifetime risk for sui-
low risk. cide, no test of the asserted combination of desire,
intent, and capability and imminent or even near-
Suicide risk scales term risk has yet been made.
Beginning in the early 1960s, clinical researchers began An alternative model is outlined in The American
developing suicide potential scales to help evaluate a Association of Suicidology’s Recognizing and
patient or caller who identified him/herself as suicidal. Responding to Suicide Risk (www.suicidology.org)
For example, Tuckman and Youngman (1968) devel- clinical training program. This model asserts that the
oped a scale for differentiating high versus low suicide presence of warning signs (acute risk factors) (Rudd
risk among those who had made an attempt, defining et al., 2006)3 in the context of the patient’s vulnerabil-
high-risk operationally as a predicted death from suicide ity to be suicidal across his/her lifetime (chronic risk)
within 1 year of the attempt. Perhaps, the most signifi- informs imminent risk. Although no actuarial state-
cant contribution to developing a SRF was offered by ment is made about cut-off scores, the presence of
suicide researchers at the NIMH in the early 1970s, who three or more acute risk factors has been shown to
developed sophisticated models for formulating risk describe the majority of 55 suicidal decedents on
among four demographic groups defined by age and whom a psychological autopsy was conducted
gender (old–young, male–female). Weighted items and (Berman, 2011). Unfortunately, current warning
associated ratings of risk were established, with scaled signs (Rudd et al., 2006) are associated with suicide
scores indicating high-risk equating to the likelihood of over the subsequent 12 months, not the next 30 or 7
being dead by suicide within 2 years (Beck et al., 1974). days or, for that matter, 48 hours. There is an impor-
None of these early efforts to predict suicide were ever tant need for research on observable risk factors in the
replicated or validated. last days of life of those who die by suicide.
Given the low base rate of suicide, predictive scales A third model is offered by the SAFE-T pocket card
will produce far more false positives than could be (www.sprc.org/library/safe_t_pcktcrd_edc.pdf). The
clinically tolerated. Indeed, attempts to actuarially downside to the SAFE-T model is that it maintains a
integrate suicide risk factors into some sort of formula focus on current SI or behavior as integral to inform-
for judging level of risk have been shown to have poor ing that judgment of level of risk.
specificity (Haney et al., 2012). There is a need for The outcome of an SRF is a disposition, for exam-
clinical models to inform clinical judgments about ple in an emergency department: release without
levels of risk. Moreover, questions remain as to follow-up or recommendation for further treatment,
whether clinical judgments of risk are to be based release with recommendation for follow-up outpatient
reasonably on science, intuition, or both, and whether care, release with crisis management and safety plan-
good clinical judgment regarding a patient’s relative ning interventions, release with arranged outpatient
level of suicide risk can be trained. interventions (psychotherapy and/or medications), or
hospitalization (for safety and further assessment). In
Current models for suicide risk this case, triage decisions would have to include an
assessment of the patient’s support system, evidence of
formulation
The National Suicide Prevention Lifeline (www.suici-
depreventionlifeline.org/crisiscenters/bestpractices) has 3
see www.suicidology.org for an extensive list of chronic and acute
offered a guide to determine imminent risk, as follows: risk factors.
38
controls over SI and/or subsequent attempt, ability to science into this process as is possible to arrive at the best
cooperate, and degree of resolution of life crises that possible approach to risk formulation. Indeed, evidence
prompted the presentation to the ED, in addition to an that this is possible exists (Swets et al., 2000).
assessment of current acute risk factors and warning
signs. These same variables might be considered in an Conclusion
outpatient setting to help guide dispositional decision- An SRA is a data gathering enterprise that may be
making about the need for hospitalization. approached initially by asking questions regarding the
presence of SI, but must not be limited to inquiries about
A final note on clinical judgment SI. Suicide risk may exist and may even be imminent
As noted, there is a need for research to inform an SRF: with or without communicated SI. Moreover, SI has an
* On risk factors associated with very near-term risk inconclusive association with death by suicide; it is a
* On which of the many components of the SRA – and better predictor of suicide attempt, but, again, is neither
to what degree the frequency, duration, intensity, necessary, nor sufficient, for the assessment of that risk.
timing, etc., of each of these components, as well as SI comes in a variety of forms and little to no research has
their combinations and permutations – informs the been conducted to assure clinicians that one or another
appropriate level of risk in an SRF form, such as passive versus active SI, is more, versus less,
associated with near-term or long-term risk of death by
* On validating models for an SRF
suicide.
* On evaluating the appropriateness of interventions
An SRF is akin to a recipe to piece together the
linked to an SRF.
ingredients developed from an SRA. The types and
Clinicians deserve a better roadmap or clinical tool to amounts of the ingredients will dictate the final product.
assist in assigning a level of suicide risk to an individ- Clinical intuition and judgment will always be necessary
ual. Without such a guide, the clinician is left to his/her to an SRF, as purely actuarial models will never attain the
education, training, intuition, judgment, and prior level of specificity and sensitivity desired. That said,
experience to determine how best to intervene and models for an SRF and guidelines to better inform clin-
manage the patient. Clinical judgments and intuitions ical judgment are called for and need to be empirically
are influenced by knowledge and experience (Bittner & tested. It is time to establish empirically whether good
Tobin, 1998), but greater experience does not necessa- clinical judgment in the assessment and formulation of
rily equate to better judgment. Clinical “expertise” suicide risk can be trained. Additionally we need to better
adds little to the accuracy of clinical judgment understand how decisions about dispositions are made
(Hilton et al., 2006). based on the SRF.
Clinical errors are both universal and inevitable in
clinical practice. Clinical judgments, particularly those References
made under the stress of seeking to be life-saving, may American Academy of Child and Adolescent Psychiatry,
be especially prone to error. As noted by Karthikeyan Committee on Health Care Access and Economics Task
and Pais (2010), “clinical judgment conjures up visions Force on Mental Health (2009). Improving mental health
of the archetypal clinician endowed with infinite wis- services in primary care: reducing administrative and
dom and breathtaking clairvoyance.” We recognize that financial barriers to access and collaboration. Pediatrics,
123(4), 1248–1251.
creating clinicians in this mold is an unattainable goal.
No doubt that sound clinical judgment must be brought American Psychiatric Association (2003). Practice guidelines
to the task of formulating risk, but the questions of for the assessment and treatment of patients with suicidal
behaviors. American Journal of Psychiatry, 160(Suppl 11),
whether good clinical judgment can be improved 1–60.
upon, no less trained, remain outstanding.
Appleby, L., Kapur, N., Shaw, J., et al (2012). The National
Moreover, despite the obvious appeal, it remains
Confidential Inquiry into Suicide and Homicide by People
dangerous and unpredictable to rely on “clinical intu- with Mental Illness Annual Report, July 2012, London:
ition” in arriving at a judged risk level for an individual, Department of Health.
when life or death hangs in the balance. In the end Barber, M. E., Marzuk, P. M., Leon, A. C., & Portera, L.
actuarial models for predicting a patient’s risk of vio- (1998). Aborted suicide attempts: A new classification of
lence, in contrast to clinical judgment, may not be supe- suicidal behavior. American Journal of Psychiatry, 155(3),
rior (Litwack, 2001). The need now is to bring as much 385–389.
Beck, A. T., Resnik, H. L., & Lettieri, D. J. (1974). The Hilton, N. Z., Harris, G. T., & Rice, M. E. (2006). Sixty-six
Prediction of Suicide. Philadelphia, PA: Charles Press. years of research on the clinical versus actuarial
Beck, A. T., Brown, G. K., Steer, R. A., Dahlsgaard, K. K., & prediction of violence. The Counseling Psychologist,
Grisham, J. R. (1999). Suicide ideation at its worst point: A 34(3), 400–409.
predictor of eventual suicide in psychiatric outpatients. Horowitz, L. M., Bridge, J. A., Teach, S. J., et al. (2012). Ask
Suicide & Life-Threatening Behavior, 29(1), 1–9. Suicide-Screening Questions (ASQ): A brief instrument
Berman, A. L. (2011). Countermeasures to Reduce Suicides on for the pediatric emergency department. Archives
Railway Rights-of-Way in the United States. Washington, Pediatric and Adolescent Medicine, 166(12), 1170–1176.
DC: American Association of Suicidology. Isometsa, E. T., Heikkinen, M. E., Marttunen, M. J., et al. (1995).
Bittner, M. P., & Tobin, E. (1998). Critical thinking: Strategies The last appointment before suicide: Is suicide intent
for clinical practice. Journal for Nurses Staff Development, communicated? American Journal of Psychiatry,
14(6), 267–272. 152(6), 919–922.
Brent, D. A., Greenhill, L. L., Compton, S., et al. (2009). The Joiner, T. E., Kalafat, J., Draper, J., et al. (2007).
Treatment of Adolescent Suicide Attempters Study (TASA): Establishing standards for the assessment of suicide
Predictors of suicidal events in an open treatment trial. risk among callers to the National Suicide Prevention
Journal of the American Academy of Child and Adolescent Lifeline. Suicide and Life-Threatening Behavior,
Psychiatry, 48(10), 987–996. 37(3), 353–365.
Busch, K. A., Fawcett, J., & Jacobs, D. G. (2003). Clinical Karch, D. L., Logan, J., McDaniel, D., Parks, S., & Patel, N.
correlates of inpatient suicide. Journal of Clinical (2012). Surveillance for violent deaths – National
Psychiatry, 64, 14–19. Violent Death Reporting System, 16 States, 2009.
Morbidity and Mortality Weekly Report (MMWR),
Cahill, S., & Rakow, T. (2011). Assessing risk and prioritizing 61 (6), 1–43.
referral for self-harm: When and why is my judgment
different from yours? Clinical Psychology & Psychotherapy, Karthikeyan, G., & Pais, P. (2010). Clinical judgment and
19(5), 399–410. evidence-based medicine: Time for reconciliation.
Indian Journal of Medical Research, 132(5), 623–626.
Centers for Disease Control and Prevention (2012).
Surveillance for violent deaths – National Violent Death Kessler, R. C., Borges, G., & Walters, E. E. (1999). Prevalence
Reporting System, 16 states, 2009. Morbidity and Mortality of and risk factors for lifetime suicide attempts in the
Weekly, 61(6), 1–43. National Comorbidity Survey. Archives of General
Psychiatry, 56(7), 617–626.
Claassen, C. A., & Larkin, G. L. (2005). Occult suicidality in an
emergency department population. British Journal of Kessler, R. C., Berglund, P., Borge, G., Nock, M., &
Psychiatry, 186, 352–353. Wang, P. S. (2005). Trends in suicide ideation, plans,
gestures, and attempts in the United States, 1990–92 to
Coric, V., Stock, E. G., Pultz, J., & Marcus, R. (2009). 2001–03. Journal of the American Medical Association,
Sheehan Suicidality Tracking Scale (Sheehan-STS). 293, 2487–2495.
Psychiatry, 6(1), 26–31.
Knesper, D. J. (2010): Continuity of Care for Suicide
Fawcett, J., Scheftner, W., Fogg, L., et al. (1990). Time-related Prevention and Research: Suicide Attempts and Suicide
predictors of suicide in major affective disorder. American Deaths Subsequent to Discharge from the Emergency
Journal of Psychiatry, 147, 1189–1194. Department or Psychiatry or Psychiatry Inpatient Unit.
Feldman, M. D., Frank, P., Duberstein, P. R., et al. (2007). Newton, MA: Education Development Center, Inc.
Let’s not talk about it: Suicide inquiry in primary care. Larkin, G. L., Beautrais, A. L., Spirito, A., et al. (2009).
Annals of Family Medicine, 5(5), 412–418. Mental health and emergency medicine: a research
Hall, R. C., Platt, D. E., & Hall, R. C. (1999). Suicide risk agenda. Academic Emergency Medicine, 16(11),
assessment: a review of risk factors for suicide in 100 1110–1119.
patients who made severe suicide attempts. Evaluation of Litwack, T. R. (2001). Actuarial versus clinical assessments of
suicide risk in a time of managed care. Psychosomatics, dangerousness. Psychology, Public Policy, and Law, 7(2),
40(1), 18–27. 409–443.
Hamilton, M. (1960). A rating scale for depression. Journal Luoma, J., Martin, C., & Pearson, J. (2002). Contact with
of Neurology, Neurosurgery and Psychiatry, 2, 56–62 mental health and primary care providers before suicide:
Haney, E. M., O’Neil, M. E., Carson, S., et al. (2012). A review of the evidence. American Journal of Psychiatry,
Suicide risk factors and risk assessment tools: A 159 (6), 909–916.
systematic review. VA-ESP Project #05–225 Maltsberger, J. T. (1992). The psychodynamic formulation: An
(ESP=Evidence-based Synthesis Program). aid in assessing suicide risk. In R. W. Maris, A. L. Berman,
40
J. T. Maltsberger, & R. I. Yufit (Eds.) Assessment and Schulberg, H. C., Bruce, M. L., Lee, P. W.,
Prediction of Suicide (pp.25–49). New York: Guilford. Williams, J. W., & Dietrich, A. J. (2004). Preventing
May, A. M., Klonsky, E. D., & Klein, D. N. (2012). Predicting suicide in primary care patients: The primary care
future suicide attempts among depressed ideators: A 10- physician’s role. General Hospital Psychiatry,
year longitudinal study. Journal of Psychiatric Research, 26, 337–345.
46(7), 946–952. Simon, R. I. (2006). Imminent suicide: The illusion of
Motto, J. A. (1989). Problems in suicide risk assessment. short-term prediction. Suicide and Life Threatening
In D. Jacobs & H. N. Brown (Eds.) Suicide: Behavior, 36, 296–301.
Understanding and Responding (pp. 129–142). Simon, R. I. (2009). Suicide risk assessment forms: Form
Madison, CT: International Universities Press. over substance. The Journal of the American Academy
Motto, J. A. (1992). An integrated approach to estimating of Psychiatry and the Law, 37, 290–293.
suicide risk. In R. W. Maris, A. L. Berman, Simon, R. I. (2012). Suicide risk assessment: gateway to
J. Maltsberger, & I. Yufit, R. (Eds) Assessment and treatment and management. In R. I. Simon & R. E. Hales
Prediction of Suicide (pp. 625–639). New York: Guilford. (Eds.) The American Psychiatric Publishing Textbook of
O’Connor, E., Gaynes, B. N., Burda, B. U., Soh, C., & Suicide Assessment and Management, 2nd edn (pp.
Whitlock, E. P. (2013). Screening for and treatment of 3–28). Washington, DC: American Psychiatric
suicide risk relevant to primary care: A systematic review Publishing.
for the U.S. Preventive Services Task Force. Annals of Steer, R. A., Beck, A. T., Garrison, B., & Lester, D. (1988).
Internal Medicine, 158(10), 741–754. Eventual suicide in interrupted and uninterrupted
Olfson, M., Weissman, M. M., Leon, A. C., Sheehan, D. V., & attempters: A challenge to the cry-for-help hypothesis.
Farber, L. (1996). Suicidal ideation in primary care. Suicide and Life-Threatening Behavior, 18(2), 119–128.
Journal of General and Internal Medicine, 11, 447–453. Swets, A., Dawes, R. M., & Monahan, J. (2000). Psychological
Posner, K., Brown, G. K., Stanley, B., et al. (2011). The science can improve diagnostic decisions. Psychological
Columbia-Suicide Severity Rating Scale: Initial validity Science in the Public Interest, 1, 1–26.
and internal consistency findings from three multisite Szanto, K., Reynolds, C. F., Frank, E., et al. (1996). Suicide in
studies with adolescents and adults. American Journal elderly depressed patients: Is active vs. passive suicidal
of Psychiatry, 168(12), 1266–1277. ideation a clinically valid distinction? American Journal of
Reynolds, W. (1998). Suicidal Ideation Questionnaire. Geriatric Psychiatry, 4(3), 197–207.
Odessa, FL: Psychological Assessment Resources. Tuckman, J., & Youngman, W. (1968). A scale for assessing
Rudd, M. D., Berman, A. L., Joiner, T. E., et al. (2006). suicide risk of attempted suicides. Journal of Clinical
Warning signs for suicide: Theory, research and clinical Psychology, 24(1), 17–19.
applications. Suicide and Life Threatening Behavior, Wenzel, A., Berchick, E. R., Tenhave, T., et al. (2011).
36, 255–262. Predictors of suicide relative to other deaths in
Schmitz, W. M., Allen, M. H., Feldman, B. N., et al. (2012). patients with suicide attempts and suicide ideation:
Preventing suicide through improved training in A 30-year prospective study. Journal of Affective
suicide risk assessment and care: An American Disorders, 132(3), 375–382.
Association of Suicidology Task Force report addressing Wintersteen, M. B., Berman, A. L., & Silverman, M. M.
serious gaps in U.S. mental health training. Suicide and (2013). What is meant by “Increased risk for suicide?”
Life-Threatening Behavior, 42(3), 292–304. JAMA Pediatrics, 167(7), 675–676.

Rethinking Suicide Risk Assessment and Risk Formulation

Uploaded by

Copyright:

Available Formats

Rethinking Suicide Risk Assessment and Risk Formulation

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Rethinking Suicide Risk Assessment and Risk Formulation

Uploaded by

Copyright:

Available Formats

Section 1

4 and risk formulation

A thorough SRA is a necessary precursor to an SRF.

You might also like