اثربخشی درمان رفتاری دیالکتیکی بستری برای اختلال شخصیت مرزی: مطالعه کنترل شده

Abstract Dialectical Behavioral Therapy (DBT) was initially developed and evaluated as an outpatient treatment program for chronically suicidal individuals meeting criteria for borderline personality disorder (BPD). Within the last few years, several adaptations to specific settings have been developed. This study aims to evaluate a three-month DBT inpatient treatment program. Clinical outcomes, including changes on measures of psychopathology and frequency of self-mutilating acts, were assessed for 50 female patients meeting criteria for BPD. Thirty-one patients had participated in a DBT inpatient program, and 19 patients had been placed on a waiting list and received treatment as usual in the community. Post-testing was conducted four months after the initial assessment (i.e. four weeks after discharge for the DBT group). Pre–post-comparison showed significant changes for the DBT group on 10 of 11 psychopathological variables and significant reductions in self-injurious behavior. The waiting list group did not show any significant changes at the four-months point. The DBT group improved significantly more than participants on the waiting list on seven of the nine variables analyzed, including depression, anxiety, interpersonal functioning, social adjustment, global psychopathology and self-mutilation. Analyses based on Jacobson’s criteria for clinically relevant change indicated that 42% of those receiving DBT had clinically recovered on a general measure of psychopathology. The data suggest that three months of inpatient DBT treatment is significantly superior to non-specific outpatient treatment. Within a relatively short time frame, improvement was found across a broad range of psychopathological features. Stability of the recovery after one month following discharge, however, was not evaluated and requires further study.

Introduction Inpatient treatment of patients meeting criteria for borderline personality disorder (BPD) is both widespread (Loranger et al., 1994) and expensive (40% of the highest utilizers of psychiatric services in the United States) (Geller, 1986). Research studies evaluating the effectiveness of inpatient treatment approaches are limited. Currently, psychoanalytic treatment and Dialectical Behavioral Therapy (DBT) are recommended as treatments of choice (American Psychiatric Association, 2001). Bateman & Fonagy, 1999 and Bateman & Fonagy, 2001 compared an 18-month psychoanalytically oriented partial hospitalization program to referral to outpatient standard care (TAU). They found significantly fewer suicide attempts in the partial hospital program compared to TAU after six months of treatment. The number of individuals who were no longer parasuicidal (i.e. no longer attempting suicide or intentionally self-injuring) was significantly lower in the partial hospitalization group by 12 months, as were scores on the global severity scale of the SCL-90-R after 18 months of treatment. Follow-up treatment consisted of 18 months of outpatient psychoanalytically oriented group therapy. Results indicated further significant improvement with continued treatment. Dialectical Behavioral Therapy was developed by Linehan (Linehan, 1993a, Linehan, 1993b and Linehan, Armstrong, Suarez, Allmon and Heard, 1991) as a comprehensive principle-driven outpatient treatment program for individuals meeting criteria for BPD. In a controlled randomized one-year treatment study with chronically suicidal BPD patients, Linehan found that individuals assigned to standard outpatient DBT had more positive outcomes than those assigned to outpatient psychotherapists or mental health treatment centers in the community. Superiority of the outcomes were demonstrated across a number of outcome domains, including reduction of parasuicidal behaviors (i.e. intentional self-injury and suicide attempts), length and frequency of hospitalization, treatment drop out, and improvements in anger regulation and global and interpersonal functioning (Linehan, Armstrong, Suarez, Allmon and Heard, 1991, Linehan, Heard and Armstrong, 1993 and Linehan, Tutek, Heard and Armstrong, 1994). Subsequent reanalysis of the data indicated that superior DBT efficacy could not be accounted for by differences in treatment dose in the two conditions (Linehan et al., 1993). Although these finding have been substantially replicated by Linehan et al. (2002) as well as other investigators of outpatient DBT (Verheul et al., 2003), the number of controlled research studies on DBT is still limited. In addition to the standard outpatient program, DBT has been adapted to various specific settings: family and adolescent treatments (Miller, Ratey, Linehan, Wetzler, & Leigh, 1997), forensic settings, and case management, as well as inpatient and day-treatment settings. In principle, limiting hospitalization is an important part of the DBT philosophy. However, recent data confirmed the clinical evidence that about 80% of BPD patients in Germany experience frequent inpatient treatments on an average of 65 days per year (Jerschke, Meixner, Richter, & Bohus, 1998). Given these data, it is critically important to develop structured and specific inpatient programs for this group of patients. The inpatient DBT treatment program was initially developed by Charles Swenson at New York Hospital, White Plains (Swenson, Sanderson, Dulit, & Linehan, 2001). We have adapted it to European conditions as described in detail previously (Bohus & Bathruff, 2000; Bohus et al., 2000). The three-month treatment has as its goal reductions in four high priority target areas: suicidal behaviors, intentional self-injurious behaviors, treatment interfering behaviors and behaviors that prolong hospitalization. The inpatient DBT staff creates a validating treatment milieu balanced with an emphasis on orienting and educating new patients, identifying and prioritizing their treatment targets and rigorous application of behavior change-oriented treatment strategies. Inpatient DBT treatment procedures include contingency management procedures, skills training including mindfulness training and coaching, behavioral analysis, structured response protocols to suicidal and egregious behaviors on the unit, and consultation team meetings for DBT staff. In addition to the program of Swenson et al., we administered a manualized body-oriented therapy module to improve the acceptance of body experiences and to teach body-oriented distress tolerance skills. Two studies have been conducted to evaluate the efficacy of DBT inpatient treatment. Barley et al. (1993) used a quasi-experimental design to compare the frequency of parasuicidal acts for three time periods: at pre-treatment for 19 months prior to introducing DBT on an inpatient unit, during the 10 months when DBT was being introduced, and over the 14 months while DBT was in full operation. Parasuicidal rates were significantly lower during the third time period than during the other two, and similar rates did not change throughout the entire 43 months on a traditional general psychiatric unit in the same hospital. Bohus, Haaf, Stiglmayr, Pohl, Boehme, and Linehan (2000) have previously published the pre–post-data of 24 female patients who had finished a three-months inpatient DBT treatment. Comparing the month prior to hospitalization and the month after discharge, the authors found significant improvements in ratings of depression, dissociation, anxiety and global stress as well as a highly significant decrease in the number of self-mutilating acts. Despite of these promising results (mean effect sizes at 1.04), the interpretation of the data was hampered by the lack of a control group, the limited number of patients and the heterogeneity of the participants. Due to geographical circumstances, about half of the patients had the opportunity to continue DBT as outpatients after discharge, while the others had to be referred to non-specific treatment as usual. Post-hoc analyses of the data revealed significant differences between these two groups regarding inpatient treatment effectiveness. In this study, we present data from a group of patients none of whom had the opportunity to continue a DBT outpatient program after the discharge and compared treatment outcome with a naturalistic waiting list (WL). In addition, we examined predictors of therapy response

3. Results There were no between-group differences at pre-treatment on any outcome measure. In both groups, 68% reported self-mutilating behavior within the four weeks immediately before study entry. To examine whether difficult cases “weeded out” through self-selection in DBT but not in the WL group, we compared SCL-90-R (GSI) of the dropout group and the completers at admission and found no significant differences (t=−1.294; p=0.204). 3.1. Within-group comparisons Table 1 shows the pre–post-results of the dependent t-tests for the DBT group and the WL group. Table 1. Pre–post-tests (t-tests), n=31 (DBT) and n=19 (WL) Variable Mpre SDpre Mpost SDpost df t p d DES—DBT 26.1 14.6 18.3 15.0 30 3.29 0.003∗∗ 0.53 DES—WL 32.1 14.4 30.1 13.7 16 1.06 0.306 0.14 GAF—DBT 48.5 8.4 59.9 10.3 30 −5.76 <0.001∗∗ 1.21 GAF—WL 48.1 11.1 49.4 9.9 18 −0.46 0.652 0.12 HAMA—DBT 24.0 8.8 18.7 11.5 30 2.91 0.007∗∗ 0.52 HAMA—WL 25.2 9.0 24.6 8.9 18 0.36 0.722 0.007 STAI—DBT 73.1 5.6 64.9 9.9 29 4.23 0.000∗∗∗ 1.02 STAI—WL 74.4 8.0 75.5 6.3 18 −0.93 0.363 0.15 BDI—DBT 31.3 9.4 20.9 13.4 29 4.31 <0.001∗∗ 0.90 BDI—WL1+6h6a5. HAMD—WL 18.6 6.1 19.2 6.1 18 −0.42 0.676 0.10 IIP—DBT 7.61 1.43 6.61 1.87 30 3.02 0.005∗∗ 0.60 IIP—WL 7.89 1.05 7.89 1.00 18 0.00 1.00 0.00 STAXI—DBT 6.43 2.6 6.13 2.6 29 0.69 0.495 0.12 STAXI—WL 7.11 2.2 6.84 2.3 18 0.93 0.367 0.12 SCL-90-R (GSI)—DBT 1.74 0.48 1.18 0.81 29 3.76 <0.001∗∗ 0.84 SCL-90-R (GSI)—WL 1.92 0.68 1.99 0.71 18 −0.72 0.484 0.10 Table options The DBT group showed significant improvement on every outcome measure except anger. To check whether our results would hold up if we had not excluded non-completers, we conducted an intention to treat analysis of the SCL-90 GSI (last observation carried forward for completers plus dropouts, n=40). Within-group differences remain highly significant (t=3.790; p=0.001). Applying a Bonferroni correction requiring p=.0055 for significance, seven of the nine outcome variables remained significant. The WL group did not show significant changes on any outcome measure at the 4-month point. 3.2. Between-group comparisons Analysis of the 68% of individuals with self-mutilating behavior at pre-treatment showed that significantly more patients in the DBT group compared to the WL group (62% and 31% for DBT and WL, respectively) abstained from self-mutilation at post-assessment (χ2=3.11; p=0.039). We used multiple regression analyses to test the influence of group membership on the different continuous outcome variables. The dependent variables were change in scores on the outcome measures between pre- and post-assessment. Group membership and number of DSM-IV criteria met were entered as independent variables. In order to consider the effect of medication on treatment outcome, we entered change of medication as an independent variable when analyzing the DES (controlling for naltrexone) and the BDI and HAMD (controlling for antidepressants). The DBT group improved significantly more than participants without DBT on seven of the nine variables analyzed (see Table 2). Table 2. Multiple regressions with pre-values and group membership Variable n Beta t-value p Rgroup(corr)2 f2 BDI 45 0.501 3.38 0.002∗∗ 0.204 0.26 GAF 50 −0.360 −3.10 0.003∗∗ 0.110 0.12 SCL-GSI 49 0.395 2.95 0.005∗∗ 0.130 0.15 HAMA 45 0.328 2.32 0.025∗ 0.083 0.09 HAMD 45 0.382 2.71 0.010∗ 0.115 0.13 IIP 50 0.302 2.23 0.030∗ 0.069 0.07 STAXI 49 0.032 0.224 n.s. – – DES 46 0.285 1.86 0.071 0.041 0.04 STAI 44 0.474 3.74 0.001∗∗ 0.198 0.25 Table options Applying a Bonferroni correction, the influence of group membership on four of the nine outcome variables remained significant (p<0.0055). Effect sizes f2 for amount of variance due to group membership varied between 0.07 and 0.25, which signifies a medium effect. There were no between-group differences on the DES and the STAXI. In addition, for seven outcome variables (GAF, STAI, DES, GSI, IIP, STAXI and HMD), the degree of change over the 4-month period was predicted by the initial scores of the respective variables; pre-values correlated positively with subsequent improvements. We found no significant influence of the number of DSM-IV criteria or change of medication on any outcome variable. To check whether our between-groups differences would likely hold up if we had not excluded non-completers, an intention-to-treat analysis (ANOVA) of between-group differences on GSI was conducted. The results remained significant (DBT=40; WL=20: F=8.314, p=0.006). 3.3. Clinically relevant change within the DBT group We followed the suggestions of Jacobson et al. (1999) to determine the degree of clinically significant change. We chose the SCL-90-R (GSI) as the main outcome variable. At post-test, correlations between the GSI and the other outcome measures ranged from 0.459 (STAXI) at the lowest to 0.834 (BDI) at the highest. Thus, the GSI as a representation of outcomes appears to be a fair choice. Fig. 1 illustrates the criteria of clinical relevant change and treatment response of each single patient. The first criterion defines clinically relevant change when the client moves from the dysfunctional to the functional range (two standard deviations above the mean for a German female normal population (Franke, 1995). The second criterion (Reliable Change Index, RCI) considers deviation of the tested samples and reliability of the assessment instrument. An RCI larger than ±1.96 is significant (p<0.05). In our sample, 13 patients (41.9%) fulfilled both criteria and could be considered recovered in a clinically relevant way. Those responders showed a mean reduction in the SCL-90-GSI from a pre-treatment score of 1.77 (SD=0.37) to a post-treatment score of 0.45 (SD=0.23), which represents an effect size of 3.03. The non-responders showed no difference in their pre- and post-treatment scores (mean=1.72, SD=0.56 pre and mean=1.74, SD=0.62 post). No individual of the WL-group showed clinically relevant change. White markers: WL individuals; black markers: DBT inpatient treatment. The area ... Fig. 1. White markers: WL individuals; black markers: DBT inpatient treatment. The area between dashed lines is defined by a RCI lower than 1.96. The drawn though lines indicate the cut off-value between functional and dysfunctional SCL-90-R scores. Individuals in the grey area fulfil both criteria of relevant clinical change. Figure options We conducted a series of discriminate analyses to look for predictors of treatment response vs. non-response among patients in the DBT inpatient program. Since these were exploratory analyses, we examined a wide range of variables in three different models with three predictors in each. The three models tested first: social variables (including current employment and age), second: pre-treatment history (frequency and days of lifetime hospitalization, frequency of lifetime suicide attempts, frequency of comorbid axis I disorders), and third: severity of diagnosis (value of the DIB-R, DSM-IV criteria and number of met DSM-IV criteria. We could not find any variable in the pretest scores that discriminated between the responder and the non- responder groups. Since pre-test values of the responder and the non-responder groups on outcome variables did not differ significantly, we did not force them into the equation. A hierarchical non-parametric classification method also did not result in well-separated classes.