Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
338 Suicide and Life-Threatening Behavior 37(3) June 2007
2007 The American Association of Suicidology
An Evaluation of Crisis Hotline Outcomes
Part 2: Suicidal Callers
Madelyn S. Gould, PhD, MPH, John Kalafat, PhD,
Jimmie Lou HarrisMunfakh, BA, and Marjorie Kleinman, MS
In this study we evaluated the effectiveness of telephone crisis services/hot-
lines, examining proximal outcomes as measured by changes in callers’ suicide
state from the beginning to the end of their calls to eight centers in the U.S. and
again within 3 weeks of their calls. Between March 2003 and July 2004, 1,085
suicide callers were assessed during their calls and 380 (35.0%) participated in the
follow-up assessment. Several key findings emerged. Seriously suicidal individuals
reached out to telephone crisis services. Significant decreases in suicidality were
found during the course of the telephone session, with continuing decreases in
hopelessness and psychological pain in the following weeks. A caller’s intent to
die at the end of the call was the most potent predictor of subsequent suicidality.
The need to heighten outreach strategies and improve referrals is highlighted.
Crisis hotlines are one of the oldest suicide events, such as interpersonal losses and legal
or disciplinary problems, with suicide (Brentprevention resources in the United States
(Litman, Farberow, Shneidman, Heilig, & et al., 1993; Marttunen, Aro, & Lonnqvist,
1993; Rich, Fowler, Fogarty, & Young, 1988;Kramer, 1965; Shneidman & Farberow,
1957) and United Kingdom (Day, 1974), and Gould, Fisher, Parides, Flory, & Shaffer,
1996; Runeson, 1990). Furthermore, suicideare now ubiquitous sources of help world-
wide. One rationale for crisis hotlines is usually contemplated with psychological
ambivalence, as evidenced by surviving sui-(Mishara & Daigle, 2000; Shaffer, Garland,
Gould, Fisher, & Trautman, 1988) is that sui- cide attempters who often report that the
wish to die coexisted with wishes to be res-cidal behavior is often associated with a crisis.
The psychological autopsy research generally cued and saved (Shaffer et al., 1988). This
wish sometimes results in a “cry for help,”supports the association of stressful life
Dr. Gould is a Professor at Columbia University in the Division of Child and Adolescent Psychi-
atry (College of Physicians & Surgeons) and Department of Epidemiology (School of Public Health),
and a Research Scientist at the New York State Psychiatric Institute; Dr. Kalafat is a faculty member
of Rutgers Graduate School of Applied and Professional Psychology; Ms. Munfakh and Ms. Kleinman
are in the Division of Child and Adolescent Psychiatry and New York State Psychiatric Institute.
Supported by a cooperative agreement from the Substance Abuse and Mental Health Services
Administra tion (SAM HSA) , U79 SM5 4128. We gratefully acknowledge Dr. Rachel Strohl and Ms. Re-
becca Brent Weinberg for their assistance in data collection and analysis, respectively. While they must
remain anonymous, we want to thank the participating centers for their dedication and hard work on the
project.
Address correspondence to Madelyn S. Gould, PhD, MPH, Division of Child & Adolescent Psy-
chiatry, NYSPI, 1051 Riverside Drive, Unit 72, New York, NY 10032; E-mail: gouldm@childpsych.
columbia.edu
Gould et al. 339
which can be addressed by those with special et al. are consistent with surveys of hotline
users that indicate that young White femalestraining (Litman et al., 1965). Lastly, crisis
services may provide relief to an individual are the most frequent callers to hotline ser-
vices (King, 1977; Litman et al., 1965; Slemwho is in the “final common pathway to sui-
cide” (Shaffer et al., 1988) by providing the & Cotler, 1973). More recently, Lester
(1997) conducted a meta-analysis of 14 stud-opportunity for immediate support at these
critical times through services that are conve- ies on the relationship of suicide prevention
centers on suicide rates. While the results ofnient, accessible, and available outside of
usual office hours. individual studies did not always reach statis-
tical significance, Lester found a significantDespite strong theoretical and practi-
cal justification as a suicide prevention strat- overall preventive effect. Finally, Leenaars
and Lester (2004) reported two studies onegy, hotlines’ empirical effectiveness has yet
to be demonstrated unequivocally. One mea- the number of suicide prevention centers in
ten Canadian provinces and two territories.sure of the effectiveness of telephone crisis
services has been the assessment of suicide The first assessed the relationship between
the density of centers in 1985 and age-rates in communities served by the centers.
Studies examining the impact of crisis hot- adjusted rates for 1985–1989 and found no
significant preventive impact. The second as-lines on mortality have largely employed eco-
logical designs. These studies have compared sessed the relationship between the density
of centers in 1994 and age-adjusted rates forthe suicide rates in areas with and without a
crisis program or in areas before and after the 1994–1998 and found negative correlations
between presence of centers and change inintroduction of a crisis program. Several stu d-
ies (Barraclough & Jennings, 1977; Bridge, the suicide rates for 8 of the 12 correlations.
That is, the more centers, the lower the sui-Potkin, Zung, & Soldo, 1977; Jennings, Bar-
raclough, & Moss, 1978; Lester, 1973, 1974; cide rates. When the Yukon and Northwest
territories were excluded, the correlation co-Riehl, Marchner, & Moller, 1988; Wiener,
1969), including a meta-analysis (Dew, Bro- efficients “approached or reached statistical
significance” (p. 67). They concluded thatmet, Brent, & Greenhouse, 1987), found no
significant effects of hotlines on suicide rates. this indicated “a preventive impact, though
weak, of suicide prevention centers on sui-A significant effect of Samaritan suicide pre-
vention programs in England was found by cide in Canada” (p. 67). However, caution is
advised against the use of the term impact asBagley (1968), but the results were not repli-
cated by other researchers using more elabo- the authors correctly note that the study was
correlational and did not take into accountrate and accurate statistical techniques (Bar-
raclough & Jennings, 1977; Jennings et al., changes in other social variables over the pe-
riod.1978). These broad measures of community
suicide rates did not, however, consider the It is difficult to draw conclusions about
the effectiveness of crisis centers from studiespopulations reached by crisis services. Miller,
Coombs, Leeper, and Barton (1984) exam- of the relationship between the presence of
suicide prevention/crisis centers and commu-ined race-sex-age–specific suicide rates in
U.S. counties with and without, and before nity suicide rates without a consideration of
a complementary evaluation of proximal out-and after the introduction of, a suicide pre-
vention program. A significant reduction in comes among crisis center users. One means
to evaluate proximal outcomes is through si-the suicide rate in young White females was
found, but no evidence of an impact in other lent monitoring of calls (Mishara & Daigle,
1997). Mishara and Daigle listened to 617population groups emerged. In their paper,
the authors also reported a replication of telephone calls from suicidal callers to two
Canadian suicide centers. Immediate ortheir findings on a second set of counties for
a different time span. The findings of Miller proximal effects on the reduction of depres-
340 Suicidal Crisis Caller Outcomes
sive mood and in suicidal urgency were given. The project was approved by the Insti-
tutional Review Boards of New York Statelinked to specific intervention styles, most
notably an empathetic Rogerian style, which Psychiatric Institute/Columbia University and
Rutgers Graduat e School of Applied and Pro-also included directive components. King,
Nurcombe, Bickman, Hides, and Reid (2003) fessional Psychology. A confidentiality certifi-
cate was obtained from the Department ofrated 100 taped suicide calls to Kids Help
Line in Australia. Significant decreases in sui- Health and Human Service through the Sub-
stance Abuse and Mental Health Servicescidality and significant improvements in the
mental state of youth were observed during Administration (SAMHSA).
the course of the call (King et al., 2003).
The present study employed the call- Sample
ers’ own ratings of their mental state and sui-
cidality, in response to a standardized set of Adult suicidal individuals calling eight
telephone crisis services/hotlines across theinquiries by the crisis counselors, at the be-
ginning and end of the call to assess the im- United States were the targeted population
for this study. Between March 2003 and Julymediate proximal effect of the crisis interven-
tion. Research findings have indicated that 2004 telephone crisis counselors conducted
assessm en ts with 1,085 suicidal callers (39.4%individuals’ self ratings of their own suicidal
states are more predictive of their subsequent male and 60.6% female). Individuals who
called a center more than once during thesuicidality than clinicians’ ratings ( Joiner,
Rudd, & Rajab, 1999). A follow-up assess- data collection period were only assessed
during their first contact with the center. Thement, 2 to 4 weeks later, was also conducted
in the present study to assess the duration of majority (72.0%) of assessed suicide callers
called the center’s local crisis hotline tele-an effect and the telephone intervention’s im-
pact on future suicidal risk and behavior. To phone number, the remaining called 1-800-
SUICIDE, a national network of crisis cen-our knowledge this is the first evaluation of
telephone crisis services to employ such a fol- ters. Of the 426 calls received on the 1-800-
SUICIDE line, 277 (65%) were suicide calls.low-up assessment, despite a follow-up being
considered a critical evaluation strategy (King There were 654 nonparticipants who were
not assessed because crisis counselors, usinget al., 2003; Mishara & Daigle, 2000).
The aims of the present study are to their own clinical criteria, considered the
callers’ risk status to be “too high.” Thesedetermine (1) the extent to which callers to
telephone crisis services are seriously sui- callers were in an acute suicidal state, and as
such, efforts to moderate their suicidalitycidal; (2) whether significant decreases in
suicidality occur during the call; (3) the ex- and/or initiate rescue procedures took prece-
dence over the administration of our stan-tent and predictors of suicidality after the
call; (4) the callers’ perceptions of the utility dard risk assessment (described in the mea-
sures section below). As noted in Kalafat etof the intervention; and (5) the types of refer-
rals given during the calls, and the extent to al. (this issue), other callers were not assessed
because call volume was too high, the callerwhich callers follow through with them.
refused /h ung up, the coun se lor thought it not
appropriate to assess, or phone problems ex-
isted. Among these non-assessed callers, weMETHODS
could not differentiate suicidal from nonsui-
cidal crisis callers. Thus, we do not have aA detailed description of the methods
of this study has been provided in the accom- precise estimate of the total number of sui-
cidal callers; the lower bound of the estimatepanying article by Kalafat and colleagues
(this issue). With the exception of the vari- is 1,739 (1,085 + 654), yielding a 62.4% par-
ticipation rate (upper bound).ables and sample that are unique to this arti-
cle, only a brief description of the methods is Between April 2003 and August 2004
Gould et al. 341
follow-up assessments were conducted with ment, and case management of suicidal pa-
tients, and the chapter on psychiatric and380 of the 1,085 suicide callers who com-
pleted the baseline assessment (35.3%). Fol- psychological factors in a report by the Insti-
tute of Medicine (Goldsmith, Pellmar, Klein-low-up assessments were conducted between
1 and 52 days from the baseline assessment man, & Bunney, 2002), which showed evi-
dence supportin g Shneidm an ’s (19 93) conce ptdate, with the average being 13.5 days. For
the 380 suicide callers who were followed, of psychological pain as a contributing factor
to suicidal behavior. The assessment was also30.3% were male and 69.7% female; their
age ranged from 18–72, and the mean was influenced by the empirical risk factors re-
viewed by Joiner, Walker, Rudd, and Jobes36.1 years. The ethnic distribution was
66.3% White, 15.2% African American, (1999) and the factor-analytic study of the
Modified Scale for Suicidal Ideation ( Joiner,10.2% Hispanic, 3.5% Native American,
3.2% Asian, and 1.6% Other. Ethnicity was Rudd, & Rajab, 1997). Practical considera-
tions as to the feasibility of conducting a risknot coded for six callers.
The reasons for no follow-up assess- assessment within the context of a telephone
intervention also shaped the suicide risk as-ment for 705 suicidal callers were: 311
(44.1%) callers at baseline refused re-contact; sessment. This was based on input from the
crisis center directors on our advisory board273 (38.7%) callers at baseline were not
asked by the counselors if they wanted to re- and crisis center counselors who piloted the
assessments (described in Kalafat et al., thisceive a follow-up call; 63 (9.0%) callers gave
consent at baseline for follow-up contact but issue).
Questions assessing callers’ risk statusthe follow-up interviewers received passive
or active refusals at follow up; and 58 (8.2%) included suicidal ideation and behavior, in-
tent to die, hopelessness, and psychologicalcallers gave the crisis counselors invalid con-
tact information. Common reasons for coun- pain. Three questions were asked about the
caller’s thoughts of suicide (any thoughts,selors not asking for consent for the follow-
up call were that the caller had to quickly persistence of thoughts, and control over
thoughts); one question assessed whether theterminate the call or hung up before the
counselor could ask. A significantly greater caller considered suicide the only possible
option to solve problems; one question askedproportion of suicidal callers (38.7%) com-
pared to crisis callers (8.5%) were not asked about current plans (plus narrative of “how,”
“when,” and “where”); one question askedfor consent at baseline. Suicide callers who
did not complete a follow-up assessment whether the caller had taken any action or
preparatory behavior to kill or harm him/were significantly more intent on dying (F =
15.3, p < .001), more hopeless (F = 14.2, p < herself immediately prior to the call; and
three questions assessed past attempts (life-.001), more likely to be rescued (χ
2
= 19.9,
p < .001), and less likely to be given a referral time occurrence, number of attempts, and
whether treatment was required). These(χ
2
= 24.9, p < .001) at baseline compared to
suicide callers who completed the follow-up. questions were asked at the beginning of the
call. Suicidal thoughts, plans, and attemptsHowever, changes in suicide state (intent to
die, hopelessness, and psychological pain) since the call to the center were reassessed
at the follow-up assessment. Three a priorifrom the beginning to the end of the baseline
call did not vary as a function of follow-up scales—intent to die, hopelessness, and psycholog-
ical pain—were the three major outcomes ofparticipation status.
the study, and were asked at the beginning of
the call to the center and repeated at the endMeasures
of the call and at the follow-up. These out-
comes were chosen in collaboration with ourSuicide Risk Status. The suicide risk
assessment was shaped by Chiles and Stro- advisory board, with particular input from
the crisis center directors (see details in Kala-sahl’s (1995) book on the assessment, treat-
342 Suicidal Crisis Caller Outcomes
fat et al., this issue). These outcomes were Service Utilization and Compliance.
These questions included the type of referralconsidered to be appropriate targets for an
intervention plan and their attenuation dur- (emergency services, mental health services,
social services, and information and referraling a crisis call was deemed to be critical.
The items within the intent to die, hopeless- services) and the extent of follow through.
Information on the type of referral was ob-ness, and psychological pain domains were
each rated on a 5-point scale and averaged to tained from the crisis counselors at baseline
and the referral follow through questionsderive each scale score. Higher scores indi-
cated more of the particular domain. Intent to were asked of the callers at the follow-up as-
sessment.die was assessed by two questions, “How
much do you really want to die?” and “How
likely are you to carry out your thoughts/ Procedures
plans to kill yourself?” The correlation of the
items was 0.43. Hopelessness was comprised of Baseline assessments (Time 1) were
conducted by center counselors near the be-two questions; callers were asked how hope-
ful they felt about the future and whether ginning of calls, prior to providing interven-
tion services to callers. The suicide risk as-they felt they could go on (correlation =
0.32). Psychological pain consisted of two items sessment was conducted with callers if they
had any thoughts about killing themselves.assessing current hurt, anguish, and misery
(not physical pain) and whether callers could The suicidal crisis was either self-defined by
the caller or identified by the crisis workertolerate the way they felt if their current situ-
ation did not change (corr el ati on = 0.47). The after an assessment of risk. Not all counselors
felt comfortable initiating a suicide risk as-correlations of the scales at the beginning of
the call were 0.52 (intent to die and hopeless- sessment without some clinical indicator,
such as depression, or some veiled threat. Be-ness), 0.38 (intent to die and psychological
pain), and 0.43 (hopelessness and psychologi- cause we tried to minimize interference with
the usual interactions between the counselorscal pain). (The remaining measures are also
described in Kalafat et al., this issue). and the callers, we did not require the cen-
ters’ counselors to routinely initiate the riskClient Feedback on Call. The client
feedback questions were asked at the follow- assessment. Upon completing the interven-
tion, counselors then conducted another as-up assessment. Two open-ended questions
about what was or was not helpful about the sessment at the end of the call (Time 2),
which included a subset of the initial ques-call initiated the assessment: “Thinking back
to the call you placed to the crisis line, can tions to determine whether the intervention
reduced callers’ suicidal status. Local data co-you tell me how the call was helpful to you?”
“Can you tell me what was not helpful about ordinators reviewed the centers’ call records
on an ongoing basis and compared them tothe call?” Twenty-one close-ended questions
followed the open-ended assessment and completed assessments to assure that all eligi-
ble callers were being assessed. If assessmentsprovided ratings in three areas: helper inter-
ventions, emotion regulation, and overall ef- were not conducted with potentially eligible
callers, the coordinators reviewed the call re-fectiveness, but the responses to the close-
ended questions will be the focus of a future cords for these callers with the crisis counse-
lors. Immediately preceding the end of thepaper.
Plan of Action and Comp lia nc e. This set calls, counselors used a standardized script to
ask callers if the research team could contactof questions assessed whether callers remem-
bered, agreed with, and followed through them in 1 to 2 weeks to see if they were inter-
ested in participating in the follow-up assess-with plans of actions developed by the crisis
counselors with the callers. These questions ment. The follow-up assessments were con-
ducted by independent research interviewerswere asked at the follow-up assessment.
Gould et al. 343
who had prior training and experience as chological pain, and hopelessness) as indicated
by the intraclass correlation coefficients,telephone crisis counselors. The training,
quality control procedures, and consent pro- which were all close to zero (ranging from
.004 to .05). Therefore, the use of mixed-cedures are described in detail in the article
by Kalafat and colleagues (this issue); only effects linear models to account for the clus-
tering variable of center was unnecessary.safety procedures, specific to suicide callers,
will be described here. In the beginning of Center was included as a covariate in the
analyses.the risk assessment during the call to the cen-
ter, suicide callers were asked if they had A repeated measures design was em-
ployed to examine changes over time, alwaysdone anything, including preparatory behav-
ior, to hurt or kill themselves before they employing center as the between subjects
factor. The measures were assessed at threecalled the crisis center. If a caller was in im-
minent danger, the crisis center stopped the points: near the beginning of the call (Time
1), at the end of the call (Time 2), and atinterview and initiated their standard rescue
procedures. The assessment was only contin- follow-up (Time 3). The repeated measures
for the suicide callers were intent to die, psy-ued if it was helpful to keep the caller en-
gaged while waiting for emergency rescue chological pain, and hopelessness. These re-
peated measures were also examined as aservices to arrive.
The follow-up assessment included function of the suicide risk elements (i.e.,
whether the caller had a suicide plan, hadcriteria to be used by our interviewers to de-
termine whether callers needed intervention made a preparatory or actual action to harm/
kill self prior to the call, or had an attemptat follow-up. The need for intervention was
defined by a past plan or actual attempt at history).
A series of logistic regression analysesself-injury since speaking with the center, or
a serious intent to die at the time of the fol- were conducted to determine the baseline
predictors of any suicidality (thoughts, plans,low-up interview. The method for getting
help to callers consisted of follow-up inter- or attempts) following the crisis call. The in-
depende nt variables includ ed in sepa ra te mod -viewers re-connecting callers back to the
center they had initially phoned. If callers els were intent to die, psychological pain, and
hopelessness (each at the beginning and endwere unable to participate in a call-back to
the center immediately after completing of the baseline call), persistence of suicide
thoughts, control over thoughts, consideringtheir interviews, follow-up interviewers ob-
tained callers’ consent for the center to con- suicide as the only solution to problems,
plans to kill self, actions or preparatory be-tact the callers. In this last instance, the fol-
low-up interviewer contacted the center and havior before the call, and a history of an at-
tempt. Age and gender were included in allgave them the caller’s contact information
and details as to why the caller needed inter- models. All significant predictors in the ini-
tial models were entered simultaneously asvention.
independent variables in a final multivariate
analysis.Analytic Strategy
Those callers followed up were com-
pared to those who were not followed up onThe primary sampling unit of the
study was crisis center, and the secondary baseline measures at the beginning of the call
(as previously described) by means of univari-sampling unit was caller within center. Thus,
we examined the extent of within-center ate analyses of variance. Interactions between
follow-up status and changes from Time 1 toclustering in order to determine whether this
clustering variable warranted inclusion in the Time 2 were examined using two-way analy-
ses of variance.analyses. The sample clusters (center) had lit-
tle impact on outcomes (intent to die, psy- The statistical analyses were con-
344 Suicidal Crisis Caller Outcomes
ducted with SPSS statistical software (version 49.3%) (χ
2
= 24.5, p < .001). There was no
significant difference in the risk profile of12.0). Given the number of comparisons, re-
sults were considered significant at α<.001, callers to the centers’ regular line and to 1-
800-SUICIDE.but results at α<.01 are presented in the
tables.
Rescues
Counselors reported initiating rescueRESULTS
procedures with 136 (12.6%) of the callers
who participated in the baseline assessment.Presenting Problems
Rescue procedures were significantly more
likely to be initiated for the callers who hadSuicide callers contacted the centers
with a variety of problems ranging from abuse/ engaged in preparatory behavior or had done
something to hurt/kill themselves (37.9%)violence (10.0%), physical health problems
(16.1%), work problems (12.8%), addictions than for callers who had not taken these ac-
tions (10.8%) (χ
2
= 49.2, p < .001). Of the(17.9%), base needs (25.9%), mental health
problems (54.7%), and interpersonal prob- suicidal callers who had taken some action to
hurt/kill themselves and had not initiatedlems (58.4%), along with their suicidal crises.
Gender differences were significantly related rescue (n = 54), eight had been unable to
have a rescue initiated because the center wasto only one type of problem: males (24.8%)
had significantly more addiction problems unable to identify the caller’s telephone num-
ber or the caller refused or hung up prema-than females (13.5%) (χ
2
= 21.4, p < .001).
turely. Rescues were initiated significantly
more often for callers who had a current planRisk Profile
to hurt/kill themselves (19.2%) than for
those without a plan (4.9%) (χ
2
= 45.3, p <Of all the suicide callers who com-
pleted the baseline assessment (1,085 callers), .001). Rescues were also initiated more often
for callers who had a history of previous sui-over half (585 callers) had a suicide plan
when they called the crisis center and 8.1% cide attempts (15.2%) than for those with no
such history (8.5%) (χ
2
= 10.0, p < .01).(88 callers) had taken some action to harm or
kill themselves immediately before calling
the center. More than half (57.5%, 624 call- Referrals
ers) had made prior suicide attempts, of
which 53.2% (332 callers) had made multiple Out of the 1,085 callers who partici-
pated in the baseline assessment, 506 (46.6%)attempts and 44.1% (275 callers) had made
single attempts. There were 17 callers (2.7% were given a new referral, of which 284
(56.1%) were to mental health resources. Anof those who had prior attempts) for whom
the number of prior suicide attempts was not additional 116 (10.7%) callers were referred
back to their current therapist or services. Ofcoded. Only 22.2% of the suicidal callers had
no current plans, actions, or a history of sui- the 380 callers who participated in the fol-
low-up, 221 (58.2%) were given a new refer-cidal behavior; 5.7% had all three suicide
risks. Of those with current suicide plans, ral at baseline, of which 151 (68.3%) were
to mental health resources. An additional 52366 (62.6%) had a history of past attempts.
Of those who had taken some action to (13.7%) callers were referred back to their
current therapist or services. The overall re-harm/ki ll themselves immed iat el y befo re the ir
call, 63 (71.6%) had a history of past at- ferral rate for those who participated in the
baseline was 57.3% and the rate of referraltempts. The suicide risk profile of males and
females was similar with the exception of a for those who participated in the follow-up
was 71.8%.significantly higher rate of previous suicide
attempts among the females (64.8% versus Overall, the rate of referrals was some-
Gould et al. 345
what lower for callers with more serious sui- had a suicide plan, who had taken actions to
hurt/kill themselves, or who had a history ofcide risk profiles compared to other callers.
Callers who had current plans to hurt/kill suicide attempts had higher scores on psy-
chological pain and were significantly morethemselves received fewer referrals (44.2%)
than callers who had no current plans (53.0%) hopeless and intent on dying, there were no
significant interactions between the suicide(χ
2
= 7.4, p < .01). Callers who had taken ac-
tion to hurt/kill themselves also received risk profile factors and time. In other words,
changes from Time 1 to Time 2 were notfewer referrals (34.5%) than callers who had
not taken any action (49.3%) (χ
2
= 6.5, p = modified by the suicide risk profile.
.01). Callers who had at least one previous
suicide attempt were given the same rate of Intermediate Outcomes
referrals (46.7%) as callers who did not have
at least one previous attempt (51.2%) (χ
2
= There were significant reductions in
callers’ psychological pain (F = 13.1, p < .001)1.8, p > .05). This referral pattern may reflect
the significantly greater propensity of coun- and hopelessness (F = 17.0, p < .001) from
the end of the call (Time 2) to follow-upselors to initiate rescues among callers with
higher risk profiles, thus precluding any (Time 3) among the 380 suicide callers who
completed a follow-up assessment (Table 3).other follow-up recommendations.
However, there was no significant reduction
in callers’ intent to die during this periodImmediate Outcomes
(F = 0.19, p > 05). At follow-up, 43.2% (164/
380) of callers reported any suicidality (ide-For the 1,085 callers who completed
the baseline assessment, there was a signifi- ation, plan, or attempt) since their call to the
center. Of these, 17.1% (28/164; 7.4% of to-cant reduction in suicide status from the be-
ginning of the call (Time 1), to the end of tal sample) had made a suicide plan, and
6.7% (11/164; 2.9% of total sample) hadthe call (Time 2) on intent to die (F = 130.8,
p < .001), hopelessness (F = 112.8, p < .001), made a suicide attempt. Of those who made
a suicide attempt after their call to the center,and psychological pain (F = 181.4, p < .001)
(Table 1). The extent to which the immediate 63.6% (7/11) had made a prior attempt some
time before their call. Intent to die at the endoutcomes were modified by the suicide risk
profile factors (plans, actions, and prior at- of the baseline call (OR = 1.7, 95% CI = 1.2,
2.3, p < .001), having made any specific plantempts) was examined (Table 2). Despite the
considerable overlap among the risk factors, to hurt or kill self prior to the call (OR = 1.6,
95% CI = 1.02, 2.4, p < .04), and persistentas previously noted, each was examined sepa-
rately as a potential modifier. This analytic suicidal thoughts at baseline (OR = 1.6, 95%
CI = 1.03, 2.4, p < .04) were statistically sig-strategy allowed the clinical import of each
factor to be highlighted. While callers who nificant predictors of any suicidality (ide-
TABLE 1
Immediate Outcomes from Beginning (Time 1) to End (Time 2)
of Call
Main Effect
Time 1 Time 2 of Time
Outcomes Mean (SD) Mean (SD) Fp
Intent to Die 2.81 (1.07) 2.31 (1.04) 130.84 .001
Hopelessness 3.41 (0.99) 2.87 (0.97) 112.79 .001
Psych Pain 4.09 (0.92) 3.47 (1.08) 181.37 .001
346 Suicidal Crisis Caller Outcomes
TABLE 2
Immediate Outcomes by Suicide Risk Profile
Intent to Die Hopelessness Psychological Pain
Main Interaction Main Interaction Main Interaction
Effect Effect of Effect Effect of Effect Effect of
Time 1 Time 2 of Risk Time by Risk Time 1 Time 2 of Risk Time by Risk Time 1 Time 2 of Risk Time by Risk
Mean Mean Mean Mean Mean Mean
Outcome (SD)(SD) Fp Fp(SD)(SD) Fp F p(SD)(SD) Fp F p
Risk Profile
Plan 3.15 2.59 3.62 3.03 4.32 3.68
(n = 585) (1.04) (1.10) 109.9 0.001 4.13 ns (0.97) (0.97) 40.26 .001 3.72 ns (0.80) (1.05) 50.05 .001 0.71 ns
No Plan 2.42 1.98 3.16 2.69
(n = 468) (0.96) (0.85) (0.96) (0.92)
Action 3.28 2.85 3.72 3.28 4.37 3.79
(n = 88) (1.15) (1.30) 19.40 .001 1.14 ns (1.04) (1.10) 11.34 .001 1.40 ns (0.89) (1.09) 8.04 .01 0.70 ns
No Action 2.77 2.27 3.38 2.84 4.07 3.45
(n = 980) (1.05) (1.01) (0.98) (0.95) (0.92) (1.08)
Multiple Attempts 3.06 2.50 3.62 2.98 4.22 3.54
(n = 332) (1.03) (1.07) 10.97 .001 1.22 ns (0.98) (1.00) 6.43 .01 3.83 ns (0.83) (1.07) 1.90 ns 1.48 ns
Single Attempts 2.76 2.28 3.28 2.82 4.08 3.45
(n = 275) (1.01) (0.99) (0.97) (0.92) (0.88) (1.09)
No Attempts 2.67 2.20 3.33 2.83 4.02 3.34
(n = 440) (1.09) (1.03) (0.98) (0.96) (0.98) (1.10)
Gould et al. 347
TABLE 3
Intermediate (Follow-up) Outcomes
Main Effect T2–T3
Time 1 Time 2 Time 3 of Time Contrast
Mean (SD) Mean (SD) Mean (SD) FpFp
Intent to Die 2.80 (0.90) 2.35 (0.90) 2.25 (0.95) 7.57 .01 0.19 ns
Hopelessness 3.27 (0.93) 2.72 (0.87) 2.24 (1.09) 47.84 .001 17.03 .001
Psych Pain 4.07 (0.89) 3.42 (1.06) 2.85 (1.22) 52.84 .001 14.13 .001
ation, plan, or attempt) at follow-up (43.2% two open-ended questions about what was or
was not helpful about the call. There were sixof the callers) (Table 4). When these three
items were entered simultaneously in the lo- positive categories most frequently men-
tioned by suicide callers: listen and let talkgistic regression model, only intent to die at
the end of the baseline call remained a signif- (23.2% of responses; 40.8% of callers), warm
and caring etc. (9.7%; 17.1%), options foricant predictor (OR = 1.7, 95% CI = 1.2, 2.3,
p < .002). dealing with concerns (7.5%; 13.2%), avail-
able and patient (7.3%; 12.9%), calm downCaller Feedback. At follow-up, 380
suicide callers provided a total of 668 positive (6.9%; 12.1%), and think clearly/new per-
spective (6.9%; 12.1%). Notably, 11.6% (n =responses and 83 negative responses to the
TABLE 4
Predictors of Suicidality (Thoughts, Plans, or Attempts) Following Telephone Intervention
Model 1
a
Model 2
b
Suicide Risks Odds Ratio (CI) p Odds Ratio (CI) p
Persistent thoughts† 1.6 (1.03–2.4) .03 1.3 (0.8–2.0) .30
Control over thoughts† 1.4 (0.9 –2.2) .16 – –
Suicide as only possible option† 0.8 (0.5 –1.3) .29 – –
Plans† 1.6 (1.02–2.4) .04 1.4 (0.8–2.0) .35
Actions/preparatory behavior† 1.1 (0.5 –2.8) .80 – –
Prior attempts† 1.4 (0.9 –2.2) .11 – –
Intent to die‡—beginning of call 1.0 (0.8 –1.3) .96 0.9 (0.7–1.2) .62
—end of call 1.7 (1.2 –2.3) .001 1.7 (1.2–2.3) .002
Hopelessness‡—beginning of call 1.1 (0.9 –1.5) .41 – –
—end of call 1.3 (0.9 –1.7) .15 – –
Psychological pain‡—beginning of call 1.0 (0.8 –1.4) .87 – –
—end of call 1.1 (0.9 –1.4) .52 – –
Note. Age and gender were included in all models.
†Dichotomous item
‡5-point scale
a
Each suicide risk variable was entered into separate logistic models, with exception of
intent to die, psychological pain, and hopelessness for which the same measure at the beginning
and end of call were entered simultaneously.
b
Suicide risk variables that were statistically significant in model 1 were entered simulta-
neously in model 2. Intent to die (beginning of call) was entered into model 2 despite not being
statistically significant in model 1 in order to account for it when assessing intent to die (end
of call).
348 Suicidal Crisis Caller Outcomes
44) of suicide callers said that the call pre- number of times. Fifty-two percent (n = 56)
of the 107 callers had received a new referralvented them from killing or harming them-
selves. or referral back to a mental health resource,
yet only 15.8% (17) had either completed orThe most frequent negative feedback
concerned problems with the referral (10.8% set up an appointment.
of responses; 23.7% of callers). Other con-
cerns were raised about unhelpful interven-
tions; such as counselors being condescend- DISCUSSION
ing, not concerned, or abrupt (16.9% of
responses; 3.7% of callers); counselors pro- Several studies have suggested that tele-
phone crisis services do not reach individualsviding unhelpful solutio ns /su gg est io ns (12.1% ;
2.6%); and counselors not identifying the at high risk for suicide but instead attract
lower-risk suicidal individuals who are moreproblem (8.4%; 1.8%). Six re spo nd ent s stated
that the call was too short (7.2%; 1.6%) and likely to attempt than complete suicide (Clum,
Patsiokas, & Luscomb, 1979; Greaves, 1973;six stated that the helper asked too many
questions (7.2%; 1.6%). Lester, 1972; Maris, 1969; Sawyer, Sudak, &
Hall, 1972). The higher proportion of fe-Action Plan Compliance. Of the 380
suicide callers who participated in the follow- males who call telephone crisis services is
consistent with this conjecture (Miller et al.,up, counselors developed plans of action with
278 (73.2%) callers. Examples of action plans 1984; Mishara & Diagle, 2000). Although
our study also found that females were moreincluded having a friend come over to stay
with caller; and calling friends and family likely than males to call crisis services, the
profile of the suicide callers indicated sub-members. At follow up, 60 (21.6%) of the
278 callers did not recall the plan. Of those stantial levels of risk. Over half of the suicidal
callers had current plans to harm themselvesrecalling the plan, 102 (46.8%) callers com-
pleted “all” of the plan, 34 (15.6%) callers when they called the crisis service and nearly
10 percent had taken some action to hurt orcompleted “most,” 28 (12.8%) callers com-
pleted “some” of the plan, 24 (11.0%) callers kill themselves immediately prior to their
call. Furthermore, nearly 60 percent of thesaid the plan was still “in process,” and 26
(11.9%) callers had not carried out any of the suicidal callers had made previous suicide at-
tempts, one of the strongest predictors ofplan. The extent of follow through was not
coded for four callers (1.8%). completed suicide (Gould, Greenberg, Velt-
ing, & Shaffer, 2003; Groholt, Ekeberg,Follow Through with Referral. Of the
151 follow-up suicidal callers who were given Wickstrom, & Hadorsen, 1997; Reinherz et
al., 1995). Notably, the suicide risk exhibiteda new mental health referral, 35% had kept
or made an appointment with a mental by our sample of suicide callers is probably
underestimated, given the substantial pro-health service in the period between the ini-
tial call to the center and the follow-up as- portion of callers who were not assessed as
part of our research protocol at baseline (n =sessment.
Re-Contact with the Center. Of the 380 654) because they were deemed at too high a
risk of suicide by the telephone counselors.suicide callers who participated in the follow-
up, 107 (28.2%) callers had another contact Thus, our study empirically supports an ear-
lier impression that seriously suicidal individ-with the crisis center after their initial call.
Of these callers, 59 (55.1%) callers had one uals are reaching out to telephone crisis ser-
vices (Dew et al., 1987).additional contact, 19 (17.8%) callers had
two contacts, 9 (8.4%) callers had three con- The clinical effectiveness of the crisis
intervention is consistent with the significanttacts, 4 (3.7%) callers had four contacts, 10
(9.3%) callers had between 5 and 30 contacts, decreases in suicidality, specifically, intent to
die, hopelessness and psychological pain, foundand 6 (5.6%) callers did not remember the
Gould et al. 349
during the course of the telephone session, A sizable minority, nearly 30 percent,
of suicidal callers had another contact withsimilar to a recent evaluation of telephone
counseling services (King et al., 2003). The the crisis center after their initial call. This is
consistent with reported rates of repeated useimmediate suicidality outcomes were not
modified by the suicide risk status of the call- of telephone crisis services (Apsler & Hoople,
1976; Mishara & Daigle, 2000; Murphy,ers. This suggests that the reductions in sui-
cidality were not simply a function of “re- Wetzel, Swallow, & McClure, 1969; Speer,
1971; Wold, 1973). This finding is difficult togression to the mean,” which would have
been more consistent with greater decreases interpret; it may indicate that the caller
found the initial intervention to be useful oramong higher risk individuals. In light of
these positive proximal outcomes, the rela- may merely indicate that the callers are inap-
propriately relying on the crisis hotlinetively weak, albeit positive, preventive impact
of suicide prevention centers on community rather than getting the mental health services
they need. The latter is suggested by oursuicide rates (Leenaars & Lester, 2004; Les-
ter, 1997) suggests that greater efforts are finding that only 16 percent of the repeat
callers followed through with a mental healthneeded to attract a greater proportion of sui-
cidal individuals in the community. referral after their initial call to the centers.
The need to improve referrals to mentalIn the weeks following the crisis inter-
vention, callers’ hopelessness and psychologi- health services by telephone crisis services is
also highlighted by several findings in thecal pain continued to lessen but the intensity
of their intent to die did not continue to di- present study: over half of suicidal callers
presented with mental health problems at theminish. Moreover, a substantial proportion
(43.2%) of the callers continued to express time of the call; only about a third of the sui-
cidal callers were given a new referral to asuicidal ideation a few weeks after the initial
call and nearly 3 percent had made a suicide mental health resource or a referral back to
such a service; only a third of suicide callersattempt after their call. The callers’ intent to
die score at the end of the crisis intervention had followed through with the referral; and,
the most frequent negative feedback by sui-was the only significant independent predic-
tor of suicidality following the call; although cidal callers was about problems with refer-
rals. While callers’ follow through with refer-having made any specific plan to hurt or kill
self prior to the call and persistent suicidal rals is a function of many factors, including
caller motivation (Stein & Lambert, 1984), itthoughts at baseline were also significant, al-
beit not independent, predictors of any sui- appears that steps need to be taken by crisis
centers and counselors to address the factorscidality (ideation, plan, or attempt). Our
findings suggest that outreach strategies, such under their control; for example, increasing
their knowledge of community resources,as follow-up calls, may need to be height-
ened, particularly for suicidal callers with a matching caller needs with appropriate ser-
vices, and fostering connectedness to supporthigh level of intent to die and for callers with
a history of suicide attempts, who were sig- services (De Leo, Buono, & Dwyer, 2002).
nificantly overrepresented among those who
reattempted shortly after their call to the Limitations
center. Moreover, outreach efforts during the
course of the call may also need to be ex- The study has important limitations, as
describ ed in Kalafa t et al. (this issue), whichpanded in light of our findings that a rescue
procedure was initiated for only 40 percent also apply to the current article. A particu-
larly important limitation is that the studyof suicidal callers who had engaged in either
preparatory behavior or an actual action to was uncontrolled, and the lack of a control
condition makes it difficult to definitively at-hurt or kill themselves immediately prior to
calling the center. tribute the reduction in suicidality to the cri-
350 Suicidal Crisis Caller Outcomes
sis intervention. However, ethical concerns Conclusions
about compromising the clinical services pro-
vided to callers in crisis precluded the inclu- Our study provides empirical evidence
that seriously suicidal individuals are reach-sion of a control condition. Another limita-
tion specific to this article was the low ing out to telephone crisis services. The clin-
ical effectiveness of the crisis intervention isparticipation rate at follow-up, reflecting the
difficulty of implementing outreach proce- consistent with the significant decreases in
suicidality found during the course of thedures with suicidal callers. One major obsta-
cle was the crisis counselor’s reluctance to ask telephone session, and the continuing de-
crease in callers’ hopelessness and psycholog-for the caller’s consent for re-contact. This is
an area that needs to be addressed in the ical pain in the weeks following the crisis
intervention. Without a control group, how-training of crisis counselors. The substantial
differences observed between the suicidal ever, these effects cannot be definitively at-
tributed to the crisis intervention. Our find-callers who were followed and those who
were lost to follow-up are problematic. ings also suggest that follow-up outreach
strategies may need to be heightened, partic-Those who participated in the follow-up
were significantly less suicidal than the non- ularly for suicidal callers with a history of sui-
cide attempts, who were significantly over-participants; however, changes in suicide
state from the beginning to the end of the represented among those who reattempted
shortly after their call to the center. Thecall did not vary as a function of follow-up
participation status; thus, we are reassured need to improve referrals to mental health
services by telephone crisis services is alsothat the findings generally apply to most call-
ers in a suicidal crisis. The results may indeed highlighted. Lastly, any suicide risk assess-
ment should include a re-evaluation of theunderestimate the impact of the intervention
on suicidality because rescue procedures caller’s intent to die at the end of the call, in
light of its predictiveness of subsequent sui-were initiated significantly more often for the
suicidal callers who were not followed and cidality.
were most likely initiated for a substantial
proportion of the high risk individuals who
were not assessed at baseline.
REFERENCES
Apsler, R., & Hoople, H. (1976). Evalua-
Chiles, J. A., & Strosahl, K. D. (1995).
tion of crisis intervention services with anony-
The suicidal patient: Principals of assessment, treat-
mous clients. American Journal of Community Psy-
ment, and case management. Washington, DC:
chology, 4, 293–302.
American Psychiatric Press.
Bagley, C. (1968). The evaluation of a sui-
Clum, G. A., Patsiokas, A. T., & Lus-
cide prevention scheme by an ecological method.
comb, R. L. (1979). Empirically-based compre-
Social Science and Medicine, 2, 1–14.
hensive treatment program for parasuicide. Jour-
Barraclough, B. M., Jennings, C., &
nal of Consulting and Clinical Psychology, 47,
Moss, J. R. (1977). Suicide prevention by the Sa-
937–945.
maritans: A controlled study of effectiveness. The
Day, G. (1974). The Samaritan movement
Lancet, 1, 237–238.
in Great Britain. Perspectives in Biology and Medi-
Brent, D. A., Perper, J. A., Moritz, G.,
cine, 17, 507–512.
Baugher, M., Roth, C., Balach, L., et al.
De Leo, D., Buono, M. D., & Dwyer, J.
(1993). Stressful life events, psychopathology and
(2002). Suicide among the elderly: The long-term
adolescent suicide: A case control study. Suicide
impact of a telephone support and assessment in-
and Life-Threatening Behavior, 23, 179–187.
tervention in northern Italy. British Journal of Psy-
Bridge, T. P., Potkin, S. G., Zung,
chiatry, 181, 2226–2229.
W. W., & Soldo, B. J. (1977). Suicide prevention
Dew, A. M., Bromet, E. J., Brent, D., &
centers: Ecological study of effectiveness. Journal
Greenhouse, J. B. (1987). A quantitative litera-
of Nervous and Mental Disease, 164, 18–24.
Gould et al. 351
ture review of the effectiveness of suicide preven- impact of suicide prevention centers on the sui-
cide rate in the Canadian provinces. Crisis, 25,tion centers. Journal of Consulting and Clinical Psy-
chology, 55, 239–244. 65–68.
Lester, D. (1972). The myth of suicideGoldsmith, S. K., Pellmar, T. C., Klein-
man, A. M., & Bunney, W. E.(Eds.). (2002). Re- prevention. Comprehensive Psychiatry, 13, 555–560.
Lester, D. (1973). Prevention of suicide.ducing suicide: A national imperative. Washington,
DC: The National Academies Press. Journal of the American Medical Association, 225,
992.Gould, M. S., Fisher, P., Parides, M.,
Flory, M., & Shaffer, D. (1996). Psychosocial Lester, D. (1974). Effect of suicide pre-
vention centers on suicide rates in the Unitedrisk factors of child and adolescent completed sui-
cide. Archives of General Psychiatry, 53, 1155–1162. States. Health Services Reports, 89, 37–39.
Lester, D. (1997). The effectiveness ofGould, M. S., Greenberg, T., Velting,
D., & Shaffer, D. (2003). Youth suicide risk and suicide prevention centers. Suicide and Life-Threat-
ening Behavior, 27, 304–310.preventive interventions: A review of the past 10
years. Journal of the American Academy of Child and Litman, R. E., Farberow, N. L., Shneid-
man, E. S., Heilig, S. M., & Kramer, J. A. (1965).Adolescent Psychiatry, 42, 386–405.
Greaves, G. (1973). An evaluation of a 24- Suicide-prevention telephone service. Journal of
the American Medical Association, 192, 107–111.hour crisis intervention clinic. Canada’s Mental
Health, 21, 13–15. Maris, R. W. (1969). The sociology of sui-
cide prevention: Policy implications of differencesGroholt, B., Ekeberg, O., Wichstrom,
L., & Hadorsen, T. (1997). Youth suicide in Nor- between suicidal patients and completed suicides.
Social Problems, 17, 132–149.way, 1990–1992: A comparison between children
and adolescents completing suicide and age- and Marttunen, M. J., Aro, H. M., & Lonn-
qvist, J. K. (1993). Precipitant stressors in adoles-gender-matched controls. Suicide and Life-Threat-
ening Behavior, 27, 250–263. cent suicide. Journal of the American Academy of
Child and Adolescent Psychiatry, 32, 1178–1183.Jennings, C., Barraclough, B. M., &
Moss, J. R. (1978). Have the Samaritans lowered Miller, H. L., Coombs, D. W., Leeper,
J. D., & Barton, S. N. (1984). An analysis of thethe suicide rate? A controlled study. Psychological
Medicine, 8, 413–422. effects of suicide prevention facilities on suicide
rates in the United States. American Journal ofJoiner, T., Rudd, M. D., & Rajab, M. H.
(1997). The Modified Scale for Suicidal Ideation Public Health, 74, 340–343.
Mishara, B. L., & Daigle, M. (1997). Ef-among suicidal adults: Factors of suicidality and
their relation to clinical and diagnostic indicators. fects of different telephone intervention styles
with suicidal callers at two suicide prevention cen-Journal of Abnormal Psychology, 106, 260–265.
Joiner, T. E., Rudd, M. D., & Rajab, ters: An empirical investigation. American Journal
of Community Psychology, 25, 861–885.M. H. (1999). Agreement between self- and clini-
cian-rated suicidal symptoms in a clinical sample Mishara, B. L., & Daigle, M. (2000).
Helplines and crisis intervention services: Chal-of young adults: Explaining discrepancies. Journal
of Consulting and Clinical Psychology, 67, 171–176. lenges for the future. In D. Lester (Ed.), Suicide
prevention: Resources for the millennium (pp. 153–Joiner, T. E., Walker, R. L., Rudd,
M. D., & Jobes, D. A. (1999). Scientizing and 171). Philadelphia: Brunner-Routledge.
Murphy, G. E., Wetzel, R. D., Swallow,routinizing the assessment of suicidality in outpa-
tient practice. Professional Psychology: Research and C. S., & McClure, J. N. (1969). Who calls the
suicide prevention center: A study of 55 personsPractice, 30, 447–453.
Kalafat, J., Gould, M. S., Harris- calling on their own behalf. American Journal of
Psychiatry, 126, 314–324.Munfakh, J. L., & Kleinman, M. (this issue). An
evaluation of crisis hotline outcomes. Part 1: Non- Reinherz, H. A., Giaconia, R. M., Silver-
man, A. B., Friedman, A., Pakiz, B., Frost, A. K.,suicidal crisis callers. Suicide and Life-Threatening
Behavior, 37, 000–000. et al. (1995). Early psychosocial risks for adoles-
cent suicidal ideation and attempts. Journal of theKing, G. D. (1977). An evaluation of the
effectiveness of a telephone counseling center. American Academy of Child and Adolescent Psychia-
try, 34, 599–611.American Journal of Community Psychology, 5,
75–83. Rich, C. L., Fowler, R. C., Fogarty,
L. A., & Young, D. (1988). San Diego suicideKing, R., Nurcombe, R., Bickman, L.,
Hides, L., & Reid, W. (2003). Telephone coun- study: III. Relationships between diagnoses and
stressors. Archives of General Psychiatry, 45, 589–seling for adolescent suicide prevention: Changes
in suicidality and mental state from beginning to 592.
Riehl, T., Marchner, E., & Moller,end of a counseling session. Suicide and Life-
Threatening Behavior, 33, 400–411. H.-J. (1988). Influence of crisis intervention tele-
phone services (“Crisis Hotlines”) on the suicideLeenaars, A. A., & Lester, D. (2004). The
352 Suicidal Crisis Caller Outcomes
rate in 25 German cities. In H.-J. Moller, A. Slem, C. M., & Cotler, S. (1973). Crisis
phone services: Evaluation of a hotline program.Schmidtke, & R. Welz (Eds.), Current Issues of Sui-
cidology (pp. 431–436). Berlin: Springer-Verlag. American Journal of Community Psychology, 1, 219–
227.Runeson, B. (1990). Psychoactive sub-
stance use disorder in youth suicide. Alcohol, 25, Speer, D. C. (1971). Rate of caller re-use
of a telephone crisis service. Crisis Intervention, 3,561–568.
Sawyer, J. B., Sudak, H. S., & Hall, S. R. 83–86.
Stein, D. M., & Lambert, M. J. (1984).(1972). A follow-up study of the 53 suicides
known to a suicide prevention center. Suicide and Telephone counseling and crisis intervention: A
review. American Journal of Community Psychology,Life-Threatening Behavior, 2, 227–238.
Shaffer, D., Garland, A., Gould, M., 12, 101–126.
Wiener, I. W. (1969). The effectiveness ofFisher, P., & Trautman, P. (1988). Preventing
teenage suicide: A critical review. Journal of the a suicide prevention program. Mental Hygiene, 53,
357–363.American Academy of Child and Adolescent Psychia-
try, 27, 675–687. Wold, C. I. (1973). A two-year follow-up
of suicide prevention center patients. Suicide andShneidman, E. S. (1993). Suicide as psych-
ache. Journal of Nervous and Mental Disease, 181, Life-Threatening Behavior, 3, 171–183.
147–149.
Shneidman, E. S., & Farberow, N. L. Manuscript Received: October 1, 2006
Revision Accepted: January 23, 2007(1957). Clues to suicide. New York: Blakison.
