You are on page 1of 7

Feasibility of Dialectical Behavior Therapy for Suicidal

Adolescent Inpatients
LAURENCE Y. KATZ, M.D., BRIAN J. COX, PH.D., SHINY GUNASEKARA, M.D., AND ALEC L. MILLER, PSY.D.

ABSTRACT
Objective: To evaluate the feasibility of dialectical behavior therapy (DBT) implementation in a general child and
adolescent psychiatric inpatient unit and to provide preliminary effectiveness data on DBT versus treatment as usual
(TAU). Method: Sixty-two adolescents with suicide attempts or suicidal ideation were admitted to one of two psychiatric
inpatient units. One unit used a DBT protocol and the other unit relied on TAU. Assessments of depressive symptoms,
suicidal ideation, hopelessness, parasuicidal behavior, hospitalizations, emergency room visits, and adherence to fol-
low-up recommendations were conducted before and after treatment and at 1-year follow-up for both groups. In addition,
behavioral incidents on the units were evaluated. Results: DBT significantly reduced behavioral incidents during ad-
mission when compared with TAU. Both groups demonstrated highly significant reductions in parasuicidal behavior,
depressive symptoms, and suicidal ideation at 1 year. Conclusions: DBT can be effectively implemented in acute-care
child and adolescent psychiatric inpatient units. The promising results from this pilot study suggest that further evaluation
of DBT for adolescent inpatients appears warranted. J. Am. Acad. Child Adolesc. Psychiatry, 2004;43(3):276–282. Key
Words: dialectical behavior therapy, suicide, adolescents, inpatients.

Suicide accounts for 23.8% of all deaths among 15- to adolescents. Henggeler et al. (1999) conducted a ran-
24-year-olds in Canada (Health Canada, 2002). It is domized study comparing the effectiveness of hospital-
the third leading cause of death among adolescents in ization to home-based multisystemic therapy for
the United States (National Center for Health Statis- adolescents recommended for admission to hospital;
tics, 1996). Suicidal adolescents are known to make up however, this study included adolescents with varied
44% of adolescent psychiatric admissions (Goldston et indications for hospitalization and did not provide data
al., 1996), and up to 31% of suicidal adolescent inpa- on outcomes related to suicidality. Studies of symptom
tients attempt suicide during the first year following change before and after treatment for suicidal adoles-
discharge from the hospital (Goldston et al., 1999; cent populations have not been conducted. Moreover,
Granboulan et al., 1995). Despite this serious mental studies evaluating specific treatment interventions dur-
health concern, there is a paucity of data on the ef- ing psychiatric hospitalization for this population are
fectiveness of psychiatric hospitalization for suicidal nonexistent. Suicidal behavior cuts across numerous
diagnostic categories and may benefit from treatment
that specifically targets suicidal behavior in addition to
Accepted October 15, 2003.
Drs. Katz and Cox are with the Department of Psychiatry, University of treatment that is directed toward specific diagnoses.
Manitoba, Winnipeg. Dr. Gunasekara is now with the Department of Psychia- Dialectical behavior therapy (DBT) is an empirically
try, McMaster University, Hamilton, Ontario. Dr. Miller is with the Depart- supported treatment originally developed for chroni-
ment of Psychiatry and Behavioral Sciences, Albert Einstein College of Medicine,
Bronx, NY. cally parasuicidal female outpatients diagnosed with
This research was supported by a grant from the Department of Research at borderline personality disorder (BPD) (Linehan et al.,
the Health Sciences Centre in Winnipeg awarded to the first author. The 1991). Parasuicide is defined as any acute, intentional
authors thank Lorraine Cheyne for her efforts in gathering the 1-year follow-up
data.
self-injurious behavior resulting in physical harm, with
Reprint requests to Dr. Katz, PZ-162, PsycHealth Centre, 771 Bannatyne or without intent to die. In Linehan’s (1993a) model,
Avenue, Winnipeg, Manitoba, Canada R3E 3N4; e-mail: LKatz@exchange. patients with BPD have a biologically predisposed dif-
hsc.mb.ca.
ficulty in regulating their emotions. These patients also
0890-8567/04/4303–0276©2004 by the American Academy of Child
and Adolescent Psychiatry. experience pervasive invalidation of their emotional,
DOI: 10.1097/01.chi.0000106854.88132.4f cognitive, and behavioral experiences in their environ-

276 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004


DBT FOR SUICIDAL ADOLESCENTS

ments and develop maladaptive coping strategies (e.g., cidal adolescents from two separate inpatient units
parasuicide) to regulate their emotions (Bezirganian et using different treatment models were evaluated. Both
al., 1993; Dubo et al., 1997; Linehan, 1993a). DBT is units are general child and adolescent psychiatric inpa-
a principle-based psychotherapy that attempts to tient units, from which only adolescents meeting study
change these behaviors by balancing skills-enhancing criteria were recruited. One unit used a 2-week inpa-
change strategies with validation (Linehan, 1993a,b). tient adolescent DBT program modified from the 12-
In the first randomized controlled trial involving week adolescent outpatient program (Miller et al.,
outpatient parasuicidal adult women with BPD, DBT 1997; Rathus and Miller, 2002). The other unit, called
significantly reduced inpatient psychiatric days, para- TAU for research purposes, used a traditional, psycho-
suicide attempts, medical severity of parasuicide and dynamically oriented crisis assessment and treatment
treatment dropouts compared to treatment as usual model. Based on the results of studies of both inpatient
(TAU) (Linehan et al., 1991). In a second randomized (e.g., Bohus et al., 2000) and outpatient adults (e.g.,
controlled trial of female adult outpatient veterans with Linehan et al., 1991) and study of outpatient adoles-
BPD, Koons et al. (2001) found that DBT reduced cents (e.g., Rathus and Miller, 2002), we hypothesized
suicidal ideation, hopelessness, depression, and anger. that DBT could be effectively implemented on a child
Initial empirical studies that attempted to modify DBT and adolescent inpatient unit for the treatment of sui-
for use on adult inpatient units (these modifications cidal adolescents and would produce significantly su-
included features not consistent with DBT principles) perior symptom change from the patient’s baseline
produced equivocal results and raised questions about state compared to TAU. This study presents pretreat-
the effectiveness of this treatment approach for inpa- ment and posttreatment outcome data of these two
tients (Barley et al., 1993; Springer et al., 1996). Some treatment models, as well as data on the 1-year follow-
of these concerns include (1) potential negative rein- up outcomes of these patients.
forcement of parasuicidal behavior by removing the
patient from his or her environment (this removes the
METHOD
patient from an aversive experience and potentially re-
inforces the parasuicidal behavior as the means to es-
Participants
cape the aversive experience), (2) the generalizability of
inpatient behavioral gains to the patient’s natural en- This study was approved by the human ethical review committee
in the Faculty of Medicine at the University of Manitoba. Written
vironment, (3) potential fostering of a contagion effect informed consent was obtained from the patients and their legal
of parasuicidal behavior on inpatient units, and (4) the guardians. The sample comprised 62 inpatients (10 boys, 52 girls)
feasibility of condensing the standard 1-year DBT pro- aged 14 to 17 years (mean = 15.4) who were admitted to a psychi-
atric inpatient unit. Of the patients, 72.6% were white, 1.6% La-
gram for a brief hospital stay. More recently, Bohus et tino, 0% African American, 4.8% Asian/Pacific Islander, 19.4%
al. (2000) corrected some of the difficulties in previous First Nations Populations, and 1.6% other backgrounds. No de-
modifications of inpatient DBT in a pilot study of mographic differences were found between groups (age,
race/ethnicity, urban versus rural) at baseline. Selection of which of
parasuicidal adults with BPD and found that DBT led the two units (DBT or TAU) the patient was admitted to was based
to a reduction in parasuicidal behaviors, depression, predominantly upon bed availability at the time of admission. In-
dissociation, anxiety, and global stress. In a quasiex- clusion criteria for the study consisted of the patient having been
perimental design, Rathus and Miller (2002) found admitted after having made a suicide attempt or having had suicidal
ideation severe enough to warrant admission as determined by a
that DBT reduced psychiatric hospitalizations and im- child and adolescent psychiatrist. In addition, the adolescents
proved retention rates in outpatient therapy among agreed to stay in the hospital for brief treatment. Patients were not
parasuicidal adolescent outpatients compared with eligible for the study if they had mental retardation, psychosis,
bipolar affective disorder, or severe learning disabilities.
TAU. At the time of writing, we are unaware of any Sample size for each group (DBT and TAU) was calculated based
studies evaluating DBT for suicidal adolescent inpa- on a power analysis. Given the finding of large effect sizes in the
tients. Bohus et al., (2000) study, with a p value of <.05 each group would
need to have 26 members for 80% power. Thus, given estimated
The aim of this study was to evaluate whether DBT attrition rates over 1 year, at least 30 members were recruited per
could be effectively implemented in a “real world” gen- group to allow for an adequate final sample to evaluate treatment
eral child and adolescent psychiatric inpatient unit by effectiveness within the groups across time. One-year follow-up
examining a variety of clinical outcomes of suicidal data were available on 26 DBT patients (83% of those initially
enrolled) and 27 TAU patients (90% of those initially enrolled).
adolescent inpatients at a tertiary-care psychiatric hos- Thus there were nine patients who did not complete the 1-year
pital. Pretreatment and posttreatment outcomes of sui- follow-up (six DBT patients, three TAU patients). Eight patients

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004 277


KATZ ET AL.

could not be located and one patient developed bipolar disorder the patients participated in a DBT milieu (with DBT-trained nurs-
during the year and was not included in the study. There were no ing staff) to facilitate skills generalization. The psychiatrist provid-
differences on any measure between those who did not complete ing the DBT treatment was a Canadian board-certified psychiatrist
the study and the final sample. with 2 years of prior adolescent DBT training and supervision by an
intensively trained DBT therapist. The primary DBT treatment
Assessments team consisting of the staff psychiatrist and the full-time nursing
staff met regularly for DBT consultation meetings (a critical mo-
Pretreatment and posttreatment assessment involved a battery of dality of DBT designed to increase adherence to the treatment
self-report measures targeting suicide risk factors. In addition, the model and to increase staff motivation to treat this difficult popu-
number of incident reports filed by nursing staff during the admis- lation). In addition, a DBT consultant was brought in during the
sion was evaluated. Incident reports are mandated reports com- study to evaluate the treatment program.
pleted by hospital staff of clinically significant events on the units TAU consisted of a daily psychodynamic psychotherapy group,
(e.g., these would commonly involve violence toward self or others). individual psychodynamic psychotherapy at least once per week,
The self-report measures consisted of the 13-item Beck Depression and a psychodynamically oriented milieu. Although some aspects of
Inventory (BDI) (Beck and Beck, 1972; Beck et al., 1961), the contingency management are inherent to most inpatient unit mi-
Kazdin Hopelessness Scale for Children (KHS) (Kazdin et al., lieus, no formal behavior therapy was used. The psychiatrist pro-
1986), and the Reynolds’ Suicidal Ideation Questionnaire-Jr. (SIQ) viding the TAU was a Canadian board-certified psychiatrist with 2
(Reynolds, 1988). Subjects completed these measures at admission years of training in adolescent psychodynamic psychiatry and 9
and at discharge. Following completion of the study, the number of years of experience attending to the inpatient ward. The TAU team
incident reports filed by the nursing staff was calculated for both the met regularly for meetings to discuss ward management issues.
DBT and the TAU groups. Ancillary treatment (e.g., pharmacotherapy) was provided to
After 1 year, hospital charts were reviewed for every participant. both groups on an as-needed basis. Staff on the two wards did not
Participants were then followed up through direct telephone con- differ in terms of years of experience or level of training. Family
tact with the participant. In addition, at the 1-year telephone fol- assessments were conducted on both wards by staff social workers,
low-up, parent(s) provided information on emergency room visits, and brief crisis intervention and psychoeducation for families was
hospitalizations, and known parasuicidal behavior in an attempt to used (this did not differ between the two wards). Patients dis-
corroborate the patient and hospital chart reports (this also ensured charged from either of the two units were discharged to commu-
that we captured any service utilization outside the province). The nity-based, outpatient mental health systems. This provided
1-year assessment used the following outcome measures: (1) BDI naturalistic 1-year follow-up.
and SIQ scores at the end of 1 year; (2) number of parasuicidal
behaviors in the year following discharge from hospital as measured Data Analyses
by the Lifetime Parasuicide Count (LPC) (Linehan et al., unpub- Pretreatment to posttreatment effects were evaluated using 2 × 2
lished instrument, 1997); (3) emergency room visits secondary to analyses of variance (ANOVAs) on the BDI, KHS, and SIQ. A
parasuicidal behavior or ideation; (4) number of psychiatric read- Student t test (two-tailed) was performed comparing the total num-
missions to hospital; and (5) adherence to the recommended treat- ber of incident reports per group. Due to concerns that there could
ment (medication and/or psychotherapy) in the year following be nonspecific factors outside of the treatments on each of the
discharge. In preparation for the 1-year follow-up component of wards that influenced the number of incident reports, a post-hoc χ2
the study, patients were administered the LPC at admission. A
(two-sided) was conducted comparing the number of incident re-
1-year version of the LPC was used to record parasuicidal behaviors
ports during a 6-month period before DBT implementation and a
in the year prior to admission and was then administered again at
6-month period after DBT implementation on the DBT ward.
the 1-year follow-up. The LPC is an interview-based measure for
Because eight individuals were not available at the 1-year follow-
recording self-reported parasuicidal behaviors and their medical se-
up analysis, additional 2 × 3 ANOVAs were conducted on the BDI
verity. In addition to information provided by the patients regard-
and SIQ (admission, discharge, and 1-year reports). Medication
ing parasuicide, to increase the validity of the assessment we
and outpatient treatment adherence and rehospitalizations and
included objective measures such as emergency room visits and
emergency room visits were evaluated using χ2 analyses (two-sided).
hospitalizations. Patients in the province of Manitoba, Canada,
Pearson correlations (two-tailed) were calculated for adherence to
younger than 18 who require psychiatric admission must be ad-
medication or outpatient treatment and change in BDI or SIQ
mitted to one of these two units, and this process ensured that the
to control for effects of follow-up treatment on 1-year outcome.
chart review accurately captured readmission rates. In addition, as
Parasuicidal behaviors (LPC scores) were evaluated using 2 × 2
described above, the telephone contact asked about emergency visits
ANOVAs (year before treatment to year after treatment).
to other facilities.
Effect size (Cohen’s d) provides a standardized measure of the
extent of an effect of a treatment. This allows for more direct
Clinical Procedures comparison among different studies. Cohen (1992) has suggested
that d values of 0.20, 0.50, and 0.80 represent small, medium, and
The adolescent inpatient DBT program evaluated in this study
large effect sizes, respectively. Effect sizes (Cohen’s d; Cohen, 1988)
was adapted from the adolescent DBT model developed by Miller
were calculated for the measures with continuous scores (BDI, SIQ,
et al. (1997). Several important modifications have been made to
KHS, and LPC).
the program, and these have been described in detail in a program
description (Katz et al., 2002) and in a clinical case report (Katz and
Cox, 2002). In this study, the DBT program ran for 2 weeks and RESULTS
comprised 10 daily, manualized DBT skills training sessions (Line-
han, 1993b; Miller et al., 1997; skills training handouts are avail- In-Hospital Outcomes
able from the first author). Patients in the DBT program also were
seen twice per week for individual DBT psychotherapy to review The mean length of stay in hospital was 18 days (this
diary cards and conduct behavioral and solution analyses. Finally, was the same in both groups). Table 1 presents mean

278 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004


DBT FOR SUICIDAL ADOLESCENTS

TABLE 1
Mean Scores (and Standard Deviations) at Admission, Discharge, and One-Year Follow-up
Admission Discharge One-Year Effect Size
BDI-13
DBT 20.61 (7.67) 12.97 (8.70) 8.23 (7.16) 1.67
TAU 16.97 (9.18) 12.69 (7.29) 8.07 (7.74) 1.05
SIQ
DBT 52.71 (19.38) 40.90 (24.73) 18.15 (12.52) 2.12
TAU 47.67 (23.51) 37.97 (24.56) 19.35 (17.89) 1.36
KHSa
DBT 9.13 (4.50) 5.87 (4.41) 0.73
TAU 8.27 (5.01) 6.55 (5.34) 0.33
LPC (after log transformation)
DBT 2.26 (1.48) 1.11 (1.22) 0.63
TAU 1.99 (1.40) 0.87 (1.05) 0.73
Note: DBT = dialectical behavior therapy; TAU = treatment as usual; BDI-13 = 13-item Beck Depression Inventory;
SIQ = Suicidal Ideation Questionnaire; KHS = Kazdin Hopelessness Scale for Children; LPC = Lifetime Parasuicide Count.
a
KHS was conducted only at admission and at discharge.

scores on the main outcome measures before and af- of these measures. In addition, Pearson correlations
ter treatment. Results of 2 × 2 repeated-measures were calculated for any possible effect of outpatient
ANOVAs indicated a significant main effect for time pharmacotherapy or outpatient psychosocial treat-
on the BDI (F1,58 = 43.51, p < .001), KHS (F1,58 = ments on the 1-year outcomes on the BDI or SIQ. The
18.37, p < .001), and SIQ (F1,58 = 19.41, p < .001). correlation for pharmacotherapy was not significant on
Both groups showed substantial symptomatic improve- the BDI (r = 0.06, NS) or the SIQ (r = −0.06, NS). In
ment at discharge. There were neither group × time addition, the correlation for psychosocial treatment was
interaction effects nor any between-group differences not significant on the BDI (r = −0.09, NS) or the SIQ
on these measures, indicating that there was no differ- (r = 0.12, NS).
ence between the groups at discharge on any of the Due to a low base rate of rehospitalizations and
measures. emergency room visits, statistical analyses did not de-
With regard to the number of incident reports, DBT tect differences between the groups. However, the ab-
patients had significantly fewer incidents on the ward solute numbers are presented in Table 2. Finally, with
than the TAU group (t1,59 = 1.98, p = .052; Table 2). regard to adherence to follow-up treatment in the year
A significant difference was found in the χ2 that com- following discharge, χ2 analyses were performed re-
pared the 6-month period before DBT implementation
to the 6- month period after DBT implementation on
the DBT ward (χ2 = 43.11, p < .001). DBT had a TABLE 2
100% retention rate for treatment. Objective Outcomes of DBT and TAU Groups
DBT TAU Significance
One-Year Follow-up
Total hospitalizations per group
There were no completed suicides in either group. Year before admission 6 2 NS
Results of 2 × 3 repeated-measures ANOVAs showed a Year after discharge 6 6 NS
continued main effect for time on the BDI (F2,50 = Total ER visits per group
40.26, p < .001) and SIQ (F2,49 = 50.63, p < .001). Year before admission 6 3 NS
The results of the LPC demonstrated a skewed profile Year after discharge 8 14 NS
Total incidents per group in
requiring a log transformation to be performed before hospital 2 10 p = .052
further analysis. Results of a 2 × 2 repeated-measures
ANOVA after the log transformation showed a con- Note: Sample size (n = 53) for total hospitalizations and ER visits
per group consists of patients with 1-year follow-up data available.
tinued main effect for time on the LPC (F1,49 = 30.50,
Sample size (n) for the total incidents per group is equal to the
p < .001). Means and effect sizes for both groups from initial sample (n = 62) minus one patient dropped from the study
admission to 1-year follow-up are presented in Table 1. due to onset of bipolar disorder. NS = not significant; DBT =
There were no group × time interaction effects on any dialectical behavior therapy; TAU = treatment as usual.

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004 279


KATZ ET AL.

garding adherence to medications and psychosocial ment. The symptom reduction began in the hospital
treatments (Table 3). There were no statistically sig- and continued after discharge. Interestingly, this was
nificant differences between the groups in their adher- not affected by adherence to either outpatient pharma-
ence to follow-up. cotherapy or outpatient mental health referrals. Large
effect sizes were seen in both groups on measures of
depression and suicidal ideation. This suggests that
DISCUSSION
brief inpatient treatment may be a more powerful treat-
The present study is the first to evaluate implemen- ment intervention than previously recognized. In the
tation and 1-year outcome of suicidal adolescent inpa- current study with the sample sizes described above,
tients treated with DBT compared to TAU. A number there was not a statistically significant difference be-
of interesting findings were observed. tween DBT and TAU on the BDI, SIQ, or KHS.
Behavior on the ward appeared to vary depending on However, there was an absolute difference in the effect
which treatment was provided. DBT led to a signifi- sizes between the DBT and TAU groups on the BDI
cant reduction in behavioral problems on the ward (1.67 − 1.05 = 0.62), SIQ (2.12 − 1.36 = 0.76), and
compared to TAU. The reduction in incidents was KHS (0.73 − 0.33 = 0.4) (Table 1). This finding sug-
further supported by the finding of a significant reduc- gests that a larger study of inpatient DBT is warranted,
tion in the number of incidents on the DBT ward since as with a larger sample size a statistically significant
the implementation of DBT. Thus there appears to be difference between these two active, effective treat-
something about DBT that led to this change and not ments may be detected. The effect sizes seen are con-
something related to non-DBT ward phenomena. sistent with a study of inpatient DBT for adults with
What aspect of DBT led to this change cannot be BPD (Bohus et al., 2000), where the effect size for
determined from this study, but it may be nonspecific depressive symptoms was similar to the current study.
factors directly targeted by DBT such as increased staff Furthermore, reductions in depression and suicidal ide-
motivation and compassion for these patients, or per- ation were found among suicidal adolescents receiving
haps a specific factor such as staff trained in specific outpatient DBT (Rathus and Miller, 2002), suggesting
behavior therapy and validation principles for parasui- that DBT consistently reduces depression and suicidal
cidal patients. The reduction in behavioral incidents is ideation. In addition, given the current emphasis on
consistent with other studies of DBT in which DBT reduction of inpatient services for all psychiatric popu-
has been shown to improve outcomes on targeted be- lations, further study is warranted of the possible ben-
haviors such as parasuicide, binge eating, and substance efits to suicidal adolescents of inpatient, intensive,
abuse (e.g., Koons et al., 2001; Linehan et al., 1991, short-term treatment. There are no other studies of
2002; Telch et al., 2001). The reduction in incidents inpatient treatment of this kind for adolescents with
may also lead to reduced costs for the consequences of which to compare these results.
behavioral incidents (e.g., injury time for staff, repair Both groups demonstrated a significant reduction in
costs to property) and possibly greater staff satisfaction, the absolute number of parasuicidal behaviors in the
both of which are important to managing inpatient year after discharge. This study found a medium effect
facilities. size for inpatient treatment on future parasuicidal be-
Psychiatric inpatient hospitalization of suicidal ado- havior. Previous studies published in the literature have
lescents using intensive, focused, short-term treatment focused on suicide attempts after discharge and not on
led to significant symptom reduction over a 1-year pe- the broader concept of parasuicide, making it difficult
riod after hospitalization regardless of type of treat- to draw comparisons. The literature on outcomes in

TABLE 3
DBT and TAU Outcomes on Adherence to Pharmacotherapy and Psychosocial Treatment Recommendations
DBT (n = 26) TAU (n = 27)
Prescribed Adhered Prescribed Adhered χ2 Significance
Medications 20 12 15 8 0.36 NS
Psychosocial treatment 25 16 27 11 2.81 .09
Note: DBT = dialectical behavior therapy; TAU = treatment as usual; NS = not significant.

280 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004


DBT FOR SUICIDAL ADOLESCENTS

suicidal adolescents after hospitalization suggests that Limitations


many adolescents go on to attempt suicide after dis- The conclusions of this pilot study should be inter-
charge (Brent et al., 1993; Goldston et al., 1999). As preted in the context of several methodological limita-
much as is possible given the different outcome mea- tions. The relatively small sample size in this study
sure, our findings are consistent with this literature, as requires that the therapeutic gains made in this study
many of our patients did engage in parasuicidal behav- be viewed as preliminary and requires replication.
ior. However, there is also an important difference. The assessments in this study were not blinded. Due
Previous studies have evaluated whether patients en- to the suicidal acuity of this population, a decision was
gage in further attempts but often have not delineated made to have follow-up assessments done by an expe-
the extent of change among the group compared to the rienced nurse with access to the clinical resources to
year before hospitalization. It is also difficult to com- manage a crisis if it occurred. The nurse who was able
pare our results with previous studies, as they do not to meet these criteria could not be blind to the study
delineate the nature of treatment that occurred while in due to her clinical position in the facility.
the hospital. Thus, our finding of significant improve- The patients were not randomized to the treatment
ment in parasuicidal behaviors after hospitalization groups. This was because the wards on which the study
stands in contrast to the more ominous findings of took place serve the emergency room and, as such, bed
previous studies. availability had to determine location of admission of
Finally, DBT appears to engage patients in treat- patients, and true randomization was not possible.
ment and retain them. This has been shown in outpa- However, the patients were admitted predominantly
tient studies of adults and adolescents (Linehan et al., based on bed availability and not the type of treatment
1991, 1999; Rathus and Miller, 2002) and is also that could be provided.
found here, as indicated by the finding that all patients Adherence ratings to DBT were not used. This pilot
who entered DBT completed the inpatient treatment study set out to test the feasibility of implementing
(i.e., there were no treatment dropouts in the experi- DBT on an acute care adolescent inpatient ward. It was
mental treatment). To add to this finding, at the time not clear how to test adherence to DBT in such a
of writing, the unit has had approximately 115 patients modification of the treatment. It is now clear that it can
enter the DBT program, and only 3 patients have not be implemented in a modified form, and future studies
completed the program. Furthermore, at the time of would benefit from using adherence ratings to ensure
writing, no patient has prematurely left a skills training that DBT is being appropriately implemented both in
group session. individual therapy and in skills training.
Thus this study addresses the four concerns identi- The generalizability of this study to many areas of
fied in the introduction regarding the implementation the United States is limited by the absence of African
of DBT on inpatient wards. First, DBT does not ap- American and Latino populations (for information on
pear to negatively reinforce parasuicidal behavior by the use of DBT with these populations in an outpatient
removing the patient from his or her environment, as context, see Rathus and Miller, 2002). In addition, the
we found a reduction in parasuicide in the year follow- use of psychodynamically informed treatments and an
ing hospitalization (this improvement occurred in both average length of stay of 18 days in an acute-care fa-
groups, suggesting that brief hospitalization does not cility in the United States would be unusual.
negatively reinforce parasuicide in adolescents). Sec- Although use of pharmacotherapy was monitored
ond, given the continued improvement in patient func- during the 1-year follow-up, it was not monitored
tioning after discharge in both groups, inpatient gains during the hospitalization. Both units used pharmaco-
appear to generalize to the natural environment (how- therapy on an as-needed basis during the hospitaliza-
ever, this could represent a regression to the mean and tion. Neither unit had any theoretical prohibition to
not generalization of the treatment and thus will re- the use of prn medications; however, both units’ nurs-
quire further study). Third, DBT does not foster a ing staff are trained in deescalation procedures to avoid
contagion effect on inpatient wards; in fact, we found the use of either chemical or physical restraints.
a reduction in behavioral incidents on the ward (this Finally, this study applied DBT to a more hetero-
finding was specific to DBT). Finally, it is feasible to geneous population than has been previously studied.
conduct an abbreviated DBT program on an adolescent BPD was not a criterion for inclusion in the study. This
inpatient unit and to engage patients in this program. may have reduced the likelihood of finding differences

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004 281


KATZ ET AL.

between DBT and TAU. However, this study provides Bohus M, Haaf B, Stiglmayr C, Pohl U, Bohme R, Linehan M (2000),
Evaluation of inpatient dialectical-behavioral therapy for borderline per-
support for a broader application of DBT in suicidal sonality disorder: a prospective study. Behav Res Ther 38:875–887
adolescent inpatients. In addition, many of the out- Brent DA, Kolko DJ, Wartella ME et al. (1993), Adolescent psychiatric
inpatients’ risk of suicide attempt at 6-month follow-up. J Am Acad
come variables examined are phenomena with a low Child Adolesc Psychiatry. 32:95–105
base rate and will require larger sample sizes to ulti- Cohen J (1988), Statistical Power Analysis for the Behavioral Sciences, 2nd ed.
Hillsdale, NJ: Erlbaum
mately detect differences among treatments. Cohen J (1992), A power primer. Psychol Bull 112:155–159
Recommendations for replication thus include larger Dubo ED, Zanarini MC, Lewis RE, Williams AA (1997), Childhood an-
sample size, randomization, blinded assessments, and tecedents of self-destructiveness in borderline personality disorder. Can
J Psychiatry 42:63–69
adherence ratings of DBT therapists. Goldston D, Daniel S, Reboussin D, Kelley A, Ievers C, Brunstetter R
(1996), First- time suicide attempters, repeat attempters, and previous
Clinical Implications attempters on an adolescent psychiatry inpatient unit. J Am Acad Child
Adolesc Psychiatry 35:631–639
This was a study on the feasibility of implementing Goldston D, Daniel S, Reboussin D, Reboussin B, Frazier P, Kelley A
(1999), Suicide attempts among formerly hospitalized adolescents: a
and evaluating DBT for suicidal adolescent inpatients. prospective naturalistic study of risk during the five years after discharge.
Thus implications are more oriented toward future re- J Am Acad Child Adolesc Psychiatry 38:660–671
search efforts than large-scale implementation. Despite Granboulan V, Rabain D, Basquin M (1995), The outcome of adolescent
suicide attempts. Acta Psychiatr Scand 91:265–270
the current climate regarding inpatient psychiatric hos- Health Canada (2002), Suicidal behaviour. A Report on Mental Illnesses in
pitalization, short-term hospitalization of acutely sui- Canada. Ottawa, Canada
Henggeler SW, Rowland MD, Randall J et al. (1999), Home-based mul-
cidal adolescents for focused treatment appears to be tisystemic therapy as an alternative to the hospitalization of youths in
beneficial. One such treatment is DBT, a principle- psychiatric crisis: clinical outcomes. J Am Acad Child Adolesc Psychiatry
38:1331–1339
based, short-term treatment that can be taught and Katz LY, Cox BJ (2002), Dialectical behavior therapy for suicidal adolescent
implemented on inpatient units. Implementation on this inpatients: a case study. Clin Case Stud 1:81–92
Katz LY, Gunasekara S, Miller AL (2002), Dialectical behavior therapy for
ward required a psychiatrist trained in DBT to import inpatient and outpatient parasuicidal adolescents. In: Adolescent Psychia-
this treatment onto the unit and to train the rest of the try: The Annals of the American Society for Adolescent Psychiatry, Flaherty
team, who were completely unfamiliar with this treat- LT, ed. Hillsdale, NJ: Analytic Press, pp 161–178
Kazdin A, Rogers A, Kolbus D (1986), The Hopelessness Scale for Chil-
ment model. The staff received training by the first and dren: psychometric characteristics and concurrent validity. J Consult
fourth authors and the unit began implementation of Clin Psychol 51:241–245
Koons CR, Robins CJ, Tweed JL et al. (2001), Efficacy of dialectical
this treatment with ongoing supervision and consulta- behavior therapy in women veterans with borderline personality disor-
tion by the staff psychiatrist. Training in DBT is avail- der. Behav Ther 32:371–390
Linehan MM (1993a), Cognitive Behavioral Treatment for Borderline Per-
able. The potential benefits of engaging an inpatient sonality Disorder. New York: Guilford
team in such an endeavor are delineated in this study. Linehan MM (1993b), The Skills Training Manual for Treating Borderline
Personality Disorder. New York: Guilford
There are also broader service delivery implications Linehan MM, Armstrong HE, Suarez A, Allmon D, Heard HL (1991),
from this study. The reduction in behavioral incidents Cognitive-behavioral treatment of chronically parasuicidal borderline
on the DBT ward suggests implications for staff safety, patients. Arch Gen Psychiatry 48:1060–1064
Linehan MM, Dimeff LA, Reynolds SK et al. (2002), Dialectical behavior
staff morale, and ward expenditures for the conse- therapy versus comprehensive validation therapy plus 12-step for the
quences of patient behaviors. In summary, the prom- treatment of opioid dependent women meeting criteria for borderline
personality disorder. Drug Alcohol Depend 67:13–26
ising but preliminary results from this investigation Linehan MM, Schmidt H, Dimeff LA, Craft JC, Kanter J, Comtois KA
suggest that further efforts to evaluate DBT modified (1999), Dialectical behavior therapy for patients with borderline per-
sonality disorder and drug dependence. Am J Addict 8:279–292
for suicidal adolescent inpatients appear warranted. Miller AL, Rathus JH, Linehan MM, Wetzler S, Leigh E (1997), Dialectical
behavior therapy adapted for suicidal adolescents. J Pract Psychiatry
REFERENCES Behav Health 3:78–86
National Center for Health Statistics (1996), Advance report of final mor-
Barley WD, Buie S, Peterson E et al. (1993), The development of an tality statistics, 1994. NCHS Monthly Vital Statistics Report 45:63
inpatient cognitive-behavioral treatment program for borderline person- Rathus JH, Miller AL (2002), Dialectical behavior therapy adapted for
ality disorder. J Person Disord 7:232–240 suicidal adolescents. Suicide Life Threat Behav 32:146–157
Beck AT, Beck RW (1972), Screening depressed patients in family practice: Reynolds WM (1988), Suicidal Ideation Questionnaire, Professional Manual.
a rapid technique. Postgrad Med 52:81–85 Odessa, FL: Psychological Assessment Resources
Beck AT, Ward C, Mendelson M, Mock J, Erbaugh J (1961), An inventory Springer T, Lohr NE, Buchtel HA, Silk KR (1996), A preliminary report of
for measuring depression. Arch Gen Psych 4:53–63 short-term cognitive-behavioral group therapy for inpatients with per-
Bezirganian S, Cohen P, Brook JS (1993), The impact of mother–child sonality disorders. J Psychother Pract Res 5:57–71
interaction on the development of borderline personality disorder. Am Telch CF, Agras WS, Linehan MM (2001), Dialectical behavior therapy for
J Psychiatry 150:1836–1842 binge eating disorder. J Consult Clin Psychol 69:1061–1065

282 J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 43:3, MARCH 2004

You might also like