Week 7 Discussion Response to Classmates

profilesalel.rgpl3
Week7Gutierrez2006.pdf

129Suicide and Life-Threatening Behavior 36(2) April 2006  2006 The American Association of Suicidology

2005 SHNEIDMAN AWARD ADDRESS

Integratively Assessing Risk and Protective Factors for Adolescent Suicide Peter M. Gutierrez, PhD

This article briefly reviews key issues in adolescent suicide risk assessment and proposes that assessing risk and protective factors in combination has the best probability of informing the field’s understanding of this complex problem. Sev- eral newer measures are described along with summaries of their psychometric properties. A recommended protocol utilizing 4 developmentally appropriate, valid, and reliable self-report measures is suggested as a parsimonious way of gath- ering information on the range of risk and protective factors. Finally, a novel approach of employing short-term longitudinal studies to test models of adoles- cent suicide attempts is described along with a rationale for not focusing research efforts on completed suicide. Implications of this approach for prevention efforts conclude the article.

It is widely accepted that suicide and suicide- propriate interventions for at-risk teens should improve their overall quality of life and in-related behaviors (e.g., suicide attempts, threats,

and ideation) are significant public health crease the chances they will lead long, healthy, and productive lives. The challenge is deter-problems, and that adolescents in particular

are at high risk of engaging in non-lethal sui- mining how best to identify at-risk teens. Multiple approaches can be taken tocide-related behaviors (Grunbaum et al., 2002).

It is also well established that the best predic- assess for adolescent suicide risk. For exam- ple, it has been recommended by the Ameri-tor of future risk behaviors is a history of past

behaviors (Pinto, Whisman, & McCoy, 1997). can Medical Association that physicians regu- larly ask about suicide during appointmentsTherefore, it follows that identifying adoles-

cents who are at greatest risk of engaging in with their young patients (Kuchar & DiGui- seppi, 2003). Many schools have annual screen-suicide-related behaviors before those behav-

iors become serious should reduce the inci- ing days where students are asked about a variety of health risk related problems, in-dence of suicide. Additionally, providing ap- cluding suicide (Reynolds, 1991). In addition, school personnel are often trained to note suicide risk markers in their students and toPeter Gutierrez is Associate Professor in

the Psychology Department at Northern Illinois follow-up as necessary (Gould & Kramer, University. 2001). Mental health professionals working

Address correspondence to Peter Gutier- with adolescents should know how to deter- rez, Associate Professor, Northern Illinois Uni-

mine the level of suicide risk and should beversity, Department of Psychology, 1425 West attuned to indications of change in risk.Lincoln Highway, DeKalb, IL 60115-2892; E-

mail: [email protected] The goal of this paper is to present an

130 Assessing Risk and Protective Factors

overview of a theoretically based and empiri- impossibility. A frequently made statement in the aftermath of a youth suicide is “I don’tcally supported approach to adolescent sui-

cide risk assessment which acknowledges the know why he did it, he had so much to live for.” Conversely, clinicians are sometimesdevelopmental differences between adoles-

cents and other age groups. This approach is amazed that their clients are not suicidal, given the magnitude of the problems withfounded in the belief that the combination

of risk and protective factors affecting each which they grapple (Linehan, Goodstein, Nielsen, & Chiles, 1983). The same underly-adolescent best determines the probability of

engaging in suicide-related behaviors at any ing process can explain both situations. It is neither the number of things one has to livegiven time (Gutierrez, Osman, Kopper, &

Barrios, 2000). Most established assessment for, nor the number of problems one faces which determines who will or will not die bytools (e.g., self-report measures, structured

clinical interviews) focus solely on risk fac- suicide; rather, it is the combined effect of both (Jobes & Drozd, 2004; Jobes & Mann,tors, which I believe leaves the researcher or

clinician with only half the picture. 1999). The ambivalence often observed in suicidal individuals can be explained by theA great many tools have been devel-

oped over the years to assess thoughts related interaction of risk and protective factors. At times when the risk factors are stronger, theto suicide (e.g., Beck Scale for Suicidal Ide-

ation; Beck & Steer, 1991; Suicidal Ideation individual is more attracted to death and ap- pears more suicidal. An increase in protectiveQuestionnaire; Reynolds, 1988), probability

of engaging in suicide-related behaviors (e.g., factors (e.g., receiving positive social feed- back) or a decrease in risk factors (e.g., effec-Suicide Probability Scale; Cull & Gill, 1982),

and other known risk factors (e.g., Beck De- tive treatment for depression) may fairly rap- idly tip the scales toward more attraction forpression Inventory; Beck, Steer, & Brown,

1996; Hopelessness Scale for Children; Kaz- life and the appearance of being less suicidal (Orbach, 1989, 1997).din, Rodgers, & Colbus, 1986). Most of these

scales were originally intended for use with Numerous studies support these basic premises (e.g., Gutierrez, King, & Ghaziud-adults. Unfortunately, researchers and clini-

cians regularly use these scales when assess- din, 1996; Orbach et al., 1991; Osman et al., 1994). At a general level, suicidal adolescentsing adolescents, often with little to no empir-

ical evidence that this approach is appropriate. tend to be more attracted to death and re- pulsed by life while also being more weaklyKnowing that a scale is valid and reliable

when administered to adults is no guarantee attracted to life and repulsed by death than nonsuicidal adolescents (Gutierrez, 1999).the same is true when it is completed by ado-

lescents. Young people have more limited life Linehan and colleagues (1983) concluded that assessing protective factors against sui-experience and are less emotionally and cog-

nitively developed than adults. These differ- cide is both possible and practical. They de- termined that individuals can report bothences may influence how they interpret and

respond to questions about their thoughts, strong reasons to want to die and strong rea- sons to stay alive. Although that work has fo-feelings, and behaviors. Therefore, unless

firm evidence exists that a scale is develop- cused on adults, I believe the general theory applies to adolescents as well, and has empiri-mentally appropriate, valid, and reliable when

used with adolescents it should not be used cal support (e.g., Gutierrez et al., 2000). Translating theory into practice isfor something as serious as suicide risk as-

sessment. never easy, but it is a challenge I have been working on for many years. My early re-Suicide risk is a fluid construct, af-

fected by many factors which may change search focused more on trying to assess well established risk factors in different ways thandramatically from moment to moment, mak-

ing accurate assessment a significant chal- had been studied in the past (Gutierrez, 1999; Gutierrez et al., 1996). Two things frustratedlenge and prediction of future behavior a near

Gutierrez 131

me about that work. The first was the strug- based on the Reasons for Living Inventory (Linehan et al., 1983) but is developmentallygle to find developmentally appropriate self-

report measures for the constructs of interest appropriate for adolescents. For example, the responsibility to children subscale became re-and the second was finding evidence of the

reliability and validity of the scales which did sponsibility to friends on the RFL-A. This measure was specifically designed to assessexist. Focusing on self-report measures was

important because the size of the samples be- protective factors which may help keep ado- lescents’ suicidal impulses in check. It wasing studied for the purposes of model testing

can be quite large. determined that not only is the RFL-A a valid and reliable measure (Gutierrez et al.,Self-report measures are the most eco-

nomical, in terms of cost and administration 2000), but it has better predictive power than the Beck Hopelessness Scale (Beck, Weiss-time, research tools available. They are useful

for assessing many risk factors, including fre- man, Lester, & Trexler, 1974). Due to its length and ease of administration and scor-quency of ideation, previous attempts, and

threats of future attempts (Kaplan et al., 1994). ing, the RFL-A is well suited to use for re- search, screening, and clinical purposes. HighWhen the variables of interest are primarily

internal affective states and cognitions, as was subscale and total scores indicate strong rea- sons for living (i.e., protective factors) andthe case in much of this work, one must rely

on the individuals’ subjective evaluations of low scores suggest weaker reasons for living (i.e., greater risk).these states. While there are reasons to ques-

tion self-report data, especially from adoles- As was previously stated, both risk and protective factors should be assessed at thecents, there are few viable alternatives. There

are no physiological measures of suicidal ide- same time, so the RFL-A alone does not fully accomplish that goal. Finding an efficientation, for example. Youth are capable of pro-

viding valid self-report data, but they tend to way to gather information about individuals’ history of suicide-related behaviors seemedendorse fewer symptoms than do either par-

ents or clinicians (Kendall, Cantwell, & Kaz- to be a potentially valuable endeavor. To ac- complish this goal, a semistructured inter-din, 1989). In addition, participants may be

more comfortable divulging information view form (Gutierrez, 1999) was adapted and pilot tested as a self-report measure. Eventu-about topics related to suicide through the

relative anonymity of a self-report scale than ally, a scoring system for the Self-Harm Be- havior Questionnaire (SHBQ), which allowsby speaking face-to-face with an interviewer

(Erdman, Greist, Gustafson, Taves, & Klein, for differential weighting and combining of several different types of response formats,1987; Greist et al., 1973). Self-report mea-

sures are not a replacement for structured was validated (Gutierrez, Osman, Barrios, & Kopper, 2001). Four subscale (i.e., inten-and semistructured clinical interviews, but

they are a crucial component of a compre- tional self-injury, suicide attempts, threaten- ing suicide, and suicidal ideation) scores andhensive assessment and are invaluable for

purposes of on-going case management. a total score can be calculated. Significant differences were found between suicidal andIn addition, any large scale screening

effort must rely on self-report measures as a nonsuicidal individuals on the total and all subscale scores. In addition, suicidal ideationfirst step in identifying at-risk youth. Recog-

nizing the utility and importance of self- and suicide threat significantly contributed to prediction of scores on an established mea-report measures, developing and evaluating

measures of risk and protective factors has sure of suicide risk. A unique feature of the SHBQ is thatbecome a major focus of the work I am doing

with my collaborators and students. Initially, it assesses a broad spectrum of suicide-related behaviors in addition to nonsuicidal self-injury.the focus of the work was on the Reasons for

Living Inventory for Adolescents (RFL-A; In addition, specific time frames are estab- lished for the various behaviors along withOsman et al., 1998), which is theoretically

132 Assessing Risk and Protective Factors

both the hoped for and actual outcomes of concerns which inhibit conduct of this type of research. First, the behavior of interestthe behaviors (e.g., to communicate distress).

These elements result in a measure which has (i.e., completed suicide) is statistically of low occurrence (Kochanek, Murphy, Anderson,the flexibility of a semi-structured interview,

but maintains the ease of administration one & Scott, 2004). Extremely large samples of youth would need to be followed over manyexpects from a traditional self-report instru-

ment. Finally, it is flexible enough to be ap- years in order to have adequate data for the types of analyses which must be conducted.propriate for both research and clinical appli-

cations across a wide range of settings. Work Second, ethical considerations raise the ques- tion of whether it is appropriate for research-is underway to validate the SHBQ for use

with adolescents—the preliminary study uti- ers to be in possession of information, which if acted upon, could prevent loss of life. Thelized a college student sample—and appears

promising (Muehlenkamp & Gutierrez, 2004). purpose of predictive validity research is to determine how well future behaviors can beThe factor structure and reliability estimates

from the adolescent data were quite similar forecast based on prior responses to a mea- sure. If researchers intervene by referringto those derived from young adults.

It is now possible to consider what high risk individuals to treatment, they inval- idate their protocol and lose that data set. Incombination of psychometrically sound and

developmentally appropriate measures can be other words, good clinical practice and ethi- cal behavior is at odds with good researchused to assess adolescent suicide risk. Utiliz-

ing the RFL-A, SHBQ, Reynolds Adolescent methodology. Even if the first two concerns can be addressed, the importance of the be-Depression Scale-2 (RADS-2; Reynolds, 2002),

and Suicidal Ideation Questionnaire (SIQ; havior being predicted requires that very strong evidence of validity exist before it canReynolds, 1988) as a screening packet or part

of a comprehensive clinical assessment of at- be recommended that the measures be used for that purpose. Both false negatives andrisk youth provides a wealth of useful data

in a relatively brief amount of time. These false positives carry high costs, although they are significantly lower for false positives.measures form the core of a long-running

project studying risk and protective factors in Replication, ideally by independent investi- gators, of previous findings is the strongesthigh school students (Gutierrez, Watkins, &

Collura, 2004). Students identified as high evidence in support of any research outcome. That means that several large, time consum-risk based on their scores on these measures

are frequently found to be suffering from sig- ing, expensive, and high risk studies will need to be conducted before the issue of predictivenificant adjustment problems, affective disor-

ders, social problems, and family stressors upon validity can be resolved. It seems that a reasonable compromisefollow-up interview. Some are not found to

be depressed or suicidal, but to be experienc- could be made which would still allow for the rigorous conduct of scientific investigationsing disruptive levels of distress due to undi-

agnosed attentional or learning disabilities. to predict completed suicide, but would make the process much more manageable. PerhapsBecause these four measures have well estab-

lished psychometric properties they are also the criterion behavior should not be com- pleted suicide, but rather attempted suicide,ideal for use when constructing and validat-

ing new measures or examining the psycho- where the potential lethality of the attempt is very carefully quantified. Non-lethal suicide-metrics of existing scales.

In the future, studies of the predictive related behaviors are unfortunately not low probability events during adolescence (Grun-validity of the RFL-A and SHBQ must be

conducted. As was noted by Goldston (2003) baum et al., 2002). Longitudinal studies as- sessing the predictive validity of measures,in his comprehensive review of child and ad-

olescent risk measures, very little information while still cumbersome, could track suicide attempts with fewer ethical and practical con-exists in this area. There are several practical

Gutierrez 133

siderations. The issue of justifying the risk of will attempt. In turn, reductions in attempts should be highly correlated with reductionsa participant attempting suicide still exists,

but I believe can be handled with the appro- in completed suicides. Research efforts in scale developmentpriate methodology. Short-term (i.e., 2 to 3

month) longitudinal studies conducted in in- and validation should be focused on the accu- racy of predicting the range of non-lethalpatient settings can provide large enough

samples of the behaviors of interest to con- suicide-related behaviors. We have been grappling for too long with the myriad issuesduct the necessary analyses, while providing

the protection which comes from maintain- related to trying to predict low occurrence deaths. I believe it is time we give ourselvesing participants in a carefully controlled envi-

ronment where the risk of death is quite low a break and move forward with work that can be practically and ethically conducted. Once(Gutierrez & Osman, 2004). In fact, in this

preliminary study there was no subject attri- a solid group of research tools exist, with es- tablished reliability and the full range of va-tion for any reason. Although the data did

not answer the question of how well com- lidity, complex models of adolescent suicide can be developed and tested. Ideally thepleted suicide was predicted, the scales per-

formed quite well in predicting which adoles- short-term longitudinal research methodol- ogy described above will be the primarycents made subsequent attempts.

Specifically, it was found that the odds method of testing these models. It is the re- sults of research along those lines which hasratio attributed to hopelessness was 1.46, de-

pressive symptoms 1.14, and low reasons for the greatest chance of informing efforts to prevent adolescent suicide. Let us just hopeliving .24 (Gutierrez & Osman, 2004). Slightly

over 39% of the participants made at least that the researchers, prevention specialists, clinicians, crisis workers, and all other con-one suicide attempt during the course of the

study, so it was not a low probability occur- cerned parties working on the various pieces of this puzzle communicate with and learnrence in this sample. In addition, hopeless-

ness contributed to 58% and reasons for living from each other. Addressing the problem of adolescent suicide requires a coordinated and56% of the explained variance in continued

suicidal ideation across the study period. integrated effort. We have the skills, energy, and increasingly the resources to bring toThese results are based on data from only

103, mostly Caucasian, adolescent inpatients bear on the problem, but we need to tighten the focus. I began by arguing for the inte-but they appear promising. These findings

should be replicated by independent re- grated assessment of risk and protective fac- tors, which I still believe is vitally important.searchers before this methodology is widely

adopted. Screening and assessment are early stages in a long process which requires coordinated ef-Assessment of suicide risk and treat-

ment of suicidal adolescents have not been forts throughout. In the same way as the causes of adolescent suicide are complex andthe only areas to which suicidologists have

made contributions. Our field has invested an multifaceted, so too are the solutions. I have proposed what I believe is a decent model forenormous amount of time, energy, and money

in suicide prevention—i.e., prevention of better studying the causes and I hope that the information I am helping to generate can bedeath by suicide. By it’s very nature, suicide

requires that some action be taken on the used by others to more effectively prevent deaths.part of the individual which ultimately results

in their death. If no action is taken there can A frequently made statement in my lab is that we would like to some day work our-be no death. Therefore, identifying which

youth are most likely to attempt suicide and selves out of a job. I would gladly turn my attention to studying other problems if theproviding them with the necessary services to

address the underlying factors contributing combined efforts of all the dedicated profes- sionals in our field result in eliminating ado-to that probability should mean that fewer

134 Assessing Risk and Protective Factors

lescent suicide. That is an idealistic goal, but improvements in clinical practice can bring us closer to our goal. I challenge everyoneone that is more than worth the effort. Im-

provements in research methodology, inno- involved in these efforts to think about the problem of adolescent suicide integrativelyvative applications of the knowledge gener-

ated by basic research, wide-scale screening and collaboratively to keep the forward mo- mentum going.and prevention efforts, expansion of the ex-

isting crisis center network, and continued

REFERENCES

Beck, A., & Steer, R. (1991). Manual for Gutierrez, P. M., Osman, A., Barrios, the Beck Scale for Suicidal Ideation. San Antonio, F. X., & Kopper, B. A. (2001). Development and TX: Psychological Corporation. initial validation of the Self-Harm Behavior Ques-

Beck, A. T., Steer, R. A., & Brown, G. K. tionnaire. Journal of Personality Assessment, 77(3), (1996). Manual for the Beck Depression Inventory-II. 475–490. San Antonio, TX: Psychological Corporation. Gutierrez, P. M., Osman, A., Kopper,

Beck, A. T., Weissman, A., Lester, D., & B. A., & Barrios, F. X. (2000). Why young people Trexler, M. (1974). The measurement of pessi- do not kill themselves: The Reasons for Living In- mism: The Hopelessness Scale. Journal of Consult- ventory for Adolescents. Journal of Clinical Child ing and Clinical Psychology, 42, 861–865. Psychology, 29, 177–187.

Cull, J. G., & Gill, W. S. (1982). Suicide Gutierrez, P. M., Watkins, R., & Col- Probability Scale. Los Angeles, CA: Western Psy- lura, D. (2004). Suicide risk screening in an urban chological Services. high school. Suicide and Life-Threatening Behavior,

Erdman, H. P., Greist, J. H., Gustafson, 34, 421–428. D. H., Taves, J. E., & Klein, M. H. (1987). Sui- Jobes, D. A., & Drozd, J. F. (2004). The cide risk prediction by computer interview: A pro- CAMS approach to working with suicidal patients. spective study. Journal of Clinical Psychiatry, 48, Journal of Contemporary Psychotherapy, 34, 73–85. 464–467. Jobes, D. A., & Mann, R. E. (1999). Rea-

Goldston, D. B. (2003). Measuring sui- sons for living versus reasons for dying: Examin- cidal behavior and risk in children and adoles- ing the internal debate of suicide. Suicide and Life- cents. Washington, DC: American Psychological Threatening Behavior, 29, 97–104. Association. Kaplan, M. L., Asnis, G. M., Sanderson,

Gould, M. S., & Kramer, R. A. (2001). W. C., Keswani, L., DeLecuona, J. M., & Youth suicide prevention. Suicide and Life-Threat- Joseph, S. (1994). Suicide assessment: Clinical in- ening Behavior, 31 (Supplement), 6–31. terview vs. self-report. Journal of Clinical Psychol-

Greist, J. H., Gustafson, D. H., Stauss, ogy, 50, 294–298. F. F., Rowse, G. L., Laughren, T. P., & Chiles, Kazdin, A., Rodgers, A., & Colbus, D. J. A. (1973). A computer interview for suicide-risk (1986). The Hopelessness Scale for Children: Psy- prediction. American Journal of Psychiatry, 130, chometric characteristics and concurrent validity. 1327–1332.

Journal of Consulting and Clinical Psychology, 54,Grunbaum, J. A., Kann, L., Kinchen, 241–245.S. A., Williams, B., Ross, J. G., Lowry, R., et al.

Kendall, P. C., Cantwell, D. P., & Kaz-(2002). Youth risk behavior surveillance—United din, A. E. (1989). Depression in children and ado-States, 2001. Morbidity and Mortality Weekly Re- lescents: Assessment issues and recommendations.ports, 54, 1–64. Cognitive Therapy and Research, 13, 109–146.Gutierrez, P. M. (1999). Suicidality in pa-

Kochanek, K. D., Murphy, S. L., Ander-rentally bereaved adolescents. Death Studies, 23, son, R. N., & Scott, C. (2004). Deaths: Final359–379. data for 2002. National Vital Statistics Reports, 53.Gutierrez, P., King, C. A., & Ghaziud- Hyattsville, MD: National Center for Health Sta-din, N. (1996). Adolescent attitudes about death tistics.in relation to suicidality. Suicide and Life-Threaten-

Kuchar, P. S., & DiGuiseppi, C. (2003).ing Behavior, 26, 8–18. Screening for suicide risk. In Guide to clinical pre-Gutierrez, P. M., & Osman, A. (2004, ventive services, second edition: Mental disorders andOctober). Prediction of adolescent suicide reattempts. substance abuse. U. S. Preventive Services Task Force,Paper presented at the Kansas Conference in http://cpmcnet.columbia.edu/texts/gcps/gcps0060.Clinical Child and Adolescent Psychology, Law-

rence, KS. html, accessed 4-15-03.

Gutierrez 135

Linehan, M. M., Goodstein, L. J., Niel- Scale in adolescent samples. Journal of Clinical Psy- chology, 50, 847–855.sen, S. L., & Chiles, J. A. (1983). Reasons for

Osman, A., Downs, W. R., Kopper, B. A.,staying alive when you are thinking of killing Barrios, F. X., Besett, T. M., Linehan, M. M.,yourself: The Reasons for Living Inventory. Jour- et al. (1998). The Reasons for Living Inventorynal of Consulting and Clinical Psychology, 51, 276– for Adolescents (RFL-A): Development and psy-286. chometric properties. Journal of Clinical Psychology,Muehlenkamp, J. J., & Gutierrez, P. M. 54, 1063–1078.(2004, November). Validation of the Self-Harm Pinto, A., Whisman, M. A., & McCoy,

Behavior Questionnaire in adolescents. Poster ses- K.J.M. (1997). Suicidal ideation in adolescents: sion presented at the annual meeting of the Asso- Psychometric properties of the Suicidal Ideation ciation for the Advancement of Behavior Therapy, Questionnaire in a clinical sample. Psychological New Orleans, LA. Assessment, 9, 63–66.

Orbach, I. (1989). Children who don’t want Reynolds, W. M. (1988). Suicidal Ideation to live. San Francisco: Jossey-Bass Publishers. Questionnaire: Professional manual. Odessa, FL:

Psychological Assessment Resources.Orbach, I. (1997). A taxonomy of factors Reynolds, W. M. (1991). A school-basedrelated to suicidal behavior. Clinical Psychology: Sci-

procedure for the identification of adolescents atence and Practice, 4, 208–224. risk for suicidal behaviors. Family CommunityOrbach, I., Milstein, E., Har-Even, D., Health, 14, 64–75.Apter, A., Tiano, S., & Elizur, A. (1991). A

Reynolds, W. M. (2002). Reynolds Adoles-Multi-Attitude Suicide Tendency scale for adoles- cent Depression Scale: Professional Manual, 2nd Edi-cents. Psychological Assessment: A Journal of Consult- tion. Odessa, FL: Psychological Assessment Re-

ing and Clinical Psychology, 3, 398–404. sources, Inc. Osman, A., Barrios, F. X., Osman, J. R.,

Hoffman, J., Hammer, R. & Panak, W. F. (1994). Manuscript Received: July 1, 2005 Revision Accepted: August 8, 2005Validation of the Multi-attitude Suicide Tendency