اثربخشی پایدار رفتار درمانی دیالکتیکی برای اختلال شخصیت مرزی

Abstract Dialectical Behaviour Therapy (DBT) is considered one of the most promising treatments for borderline personality disorder (BPD). Recently, we reported significantly positive effects of 12 months DBT on parasuicidal behaviour and impulsivity in a mixed group of female BPD patients with and without substance abuse. Fifty-eight women with BPD were randomly assigned to either 52 weeks of DBT or treatment as usual (TAU). Follow-up assessment took place at 78 weeks, i.e., 6 months after discontinuation of DBT. Participants were clinical referrals from addiction treatment and psychiatric services. Outcome measures included parasuicidal behaviour, impulsivity and substance abuse. Six months after treatment discontinuation, the benefits of DBT over TAU in terms of lower levels of parasuicidal and impulsive behaviours, and in alcohol use, sustained. No differences between the treatment conditions were found for drug abuse. In conclusion, DBT seems to have a sustained effect on some of the core symptoms of BPD and on alcohol problems in a mixed population of female borderline patients with and without substance abuse problems.

. Introduction Borderline Personality Disorder (BPD) is a persistent and severe mental disorder (American Psychiatric Association, 1994). It is, therefore, extremely important to develop and test treatment interventions with beneficial effects that sustain beyond the actual treatment period and that significantly improve the course of BPD. In a recent 12 months randomized controlled trial, we replicated the positive short-term effects of Dialectical Behaviour Therapy (DBT) originally reported by the developer of DBT (Verheul et al., 2003; Linehan, Armstrong, Suarez, Allmon, & Heard, 1991). With the exception of some data from a small, naturalistic follow-up study (n=39: DBT=19; TAU=20), currently no data are available on the long-term efficacy of DBT in the treatment of BPD ( Linehan, Heard, & Armstrong, 1993). This study examines whether the treatment results, observed at the end of treatment, sustain over 6 months follow-up.

3. Results Table 1 shows the observed means based on the number of subjects with valid data at a given measurement time. Baseline data include all patients of the Intention-to-treat sample (N=58). It is important to keep in mind that differences in mean values between tx and tX+1 in this table do not necessarily reflect the true difference over time because of possible selective missing values. Note that the number of missing values and the patients who contribute missing values varies over time. The last two columns in Table 1 give the back transformed means of the transformed values log(X+1) as modelled by the GLMM approach described in the statistical analysis paragraph. These values can be interpreted as estimated medians and more accurately reflect the true differences over time. Table 1. Descriptive data DBT Control GLMM–IIa N b M c Sdd Exp (log)e N M Sd Exp(log) DBT Control Impulsive behaviour Baselinef 27 1.76 1.44 1.45 31 1.48 1.12 1.28 Start treatment (wk 0) 25 1.49 1.01 1.28 27 1.12 0.91 0.94 1.36 0.98 End treatment (wk 52) 23 0.92 0.74 0.79 25 1.06 0.84 0.88 0.70 0.95 Follow-up (wk 78) 20 1.08 0.93 0.89 24 0.85 0.57 0.76 0.74 0.81 Parasuicidal behaviour Baseline 27 0.55 0.63 0.44 31 0.66 0.83 0.51 Start treatment (wk 0) 25 0.36 0.54 0.28 27 0.40 0.48 0.33 0.30 0.33 End treatment (wk 52) 23 0.19 0.38 0.15 25 0.48 0.63 0.38 0.16 0.33 Follow-up (wk 78) 20 0.23 0.45 0.17 24 0.43 0.73 0.30 0.12 0.30 Self mutilation Baseline 27 48.9 117.1 14.02 28 78.2 257.5 17.09 Start treatment (wk 0) 25 1.28 3.33 0.47 27 2.65 4.72 1.09 0.51 1.11 End treatment (wk 52) 22 3.31 13.15 0.55 24 41.6 78.76 5.55 0.33 5.00 Follow-up (wk 78) 18 10.9 34.28 0.90 22 33.9 99.39 2.99 0.76 3.32 Alcohol Baseline 27 3.78 3.79 2.16 31 2.87 3.44 1.51 Start treatment (wk 0) 25 3.28 3.60 1.73 27 2.15 3.06 1.09 1.98 0.96 End treatment (wk 52) 23 1.61 2.71 0.78 25 2.56 3.35 1.34 0.97 1.32 Follow-up (wk 78) 20 2.55 3.40 1.28 24 2.00 3.03 0.99 1.18 1.01 Soft drugs Baseline 27 2.00 3.52 0.77 31 0.65 1.64 0.32 Start treatment (wk 0) 25 1.04 2.82 0.34 27 0.56 2.04 0.17 0.47 0.20 End treatment (wk 52) 23 0.57 2.15 0.18 25 0.36 1.44 0.14 0.19 0.14 Follow-up (wk 78) 20 1.55 3.44 0.52 24 0.58 1.98 0.19 0.48 0.24 Hard drugs Baseline 27 1.96 3.08 0.87 31 1.29 2.85 0.50 Start treatment (wk 0) 25 1.40 2.74 0.61 27 0.63 1.67 0.28 0.69 0.54 End treatment (wk 52) 23 0.52 1.65 0.22 25 1.04 2.89 0.31 0.30 0.19 Follow-up (wk 78) 20 0.90 2.36 0.33 24 0.42 2.04 0.11 0.48 0.26 a Estimates of means at ti adjusted for missing observations based on GLMM approach with time continuous in the time by treatment interaction with separate interaction terms for treatment and follow-up period. b N=number of subjects with valid data at a given measurement time. c M=arithmetic mean. d Sd=standard deviation. e Exp (log)=Exp (mean of log transformed variabele+1)−1=comparable with geometric means and to be interpreted as medians. f Baseline data: including the patients of the intention-to-treat sample. Table options Table 2 shows the main results of the GLMM model. It shows that 12 months of DBT had a significant effect on the extent of impulsive behaviour, self-mutilating behaviour and alcohol consumption. Between start and end (week 0 through 52), all these behaviours decreased more in the DBT group compared to the control group (p values β9⩽.05). Moreover, these treatment effects were sustained in the first six months following treatment discontinuation (p values β10⩾.05). No initial and sustained effects are found for parasuicidal behaviour, soft drugs and hard drugs. It should be noted here that, although the β10 values are not significantly different from 0, most β10 estimates are positive, indicating a possible tendency towards diminishing effects between week 52 and week 78. Table 2. Initial (52 weeks) and sustained (78 weeks) effect of DBT compared to TAU Outcome Coefficienta F value P BPDS-2: impulsive behaviour Treatment (β1) b −0.20750 F(1,54)=5.12 0.03 Timec F(4,248)=0.11 0.98 Time by treatment I (β9)d −0.00641 F(1,248)=11.93 0.00 Time by treatment II (β10)e +0.00451 F(1,248)=1.30 0.26 BPDSI-5:parasuicidal behaviour Treatment (β1) b −0.10780 F(1,54)=2.10 0.15 Timec F(4,248)=1.45 0.22 Time by treatment I (β9)d −0.00225 F(1,248)=1.99 0.16 Time by treatment II (β10)e −0.00001 F(1,248)=0.00 0.99 LPC: self mutilating behaviour* Treatment (β1) b −1.5246 F(1,51)=11.12 0.00 Timec F(1,51)=4.90 0.03 Time by treatment I (β6)*,d −0.0236 F(1,51)=11.85 0.00 Time by treatment II (β7)*,e +0.0249 F(1,51)=2.00 0.16 BPDSI: Alcohol Treatment (β1) b −0.33180 F(1,54)=2.33 0.13 Timec F(4,54)=2.19 0.08 Time by treatment I (β9) d −0.01092 F(1,54)=5.33 0.02 Time by treatment II (β10) e +0.00914 F(1,54)=0.78 0.38 BPDSI: soft drugs Treatment (β1) b −0.19950 F(1,54)=1.77 0.19 Timec F(4,54)=1.36 0.26 Time by treatment I (β9) d −0.00371 F(1,54)=0.85 0.36 Time by treatment II (β10) e +0.00569 F(1,54)=1.23 0.27 BPDSI: hard drugs Treatment (β1) b +0.08724 F(1,54)=3.27 0.08 Timec F(4,54)=0.67 0.61 Time by treatment I (β9) d −0.00163 F(1,54)=1.05 0.31 Time by treatment II (β10) e +0.00521 F(1,54)=1.94 0.17 *LPC has slightly different model β6 and β7 in LPC model are equivalent to β9 and β10 in BPDSI model (see Section 2.5). a SAS Proc Mixed version 8.2, dependent variable ln(X+1) with X is outcome measure in question; for model see statistical analysis section article; structure covariance matrix: unstructured for LPC, Alcohol, Drugs-I and Drugs-II; compound symmetry for BPDSI-2 and Toeplitz for BPDSI-5; β′s in table refer to model described in Section 2.5. b Difference between DBT and TAU at week 52, difference at start of treatment (t0) can be found by β9−52*β10 which is 0.12582 for BPDSI-2, 0.00816 for BPDSI-5, −0.2974 for LPC, 0.23604 for alcohol, −0.00608 for drugs-I and 0.172 for drugs-II. c The time main effect, with time as a categorical variable, comprises 7 parameters (β2 thru β8); since they are not relevant for this article, only the overall test of the time effect is presented in this table. d Change in difference between DBT and TAU per week between start and end of treatment (week 0 and week 52). e Change in difference per week between DBT and TAU between end of treatment and follow-up (week 52 and week 78). Table options In order to get insight in the clinical relevance of the results we looked at the estimated model based median scores for all outcome measures at the start of treatment (week 0), the end of treatment (week 52), and at 26-weeks follow-up (week 78). Although all initial effects seem to sustain over the follow-up period, the psychopathology scores (impulsive behaviour, self-mutilating behaviour, alcohol abuse) increase in the DBT group and decrease in the TAU group, resulting in smaller differences between DBT and TAU during the 26-weeks follow-up period. Finally, in the 12-months study it was shown that, although less patients in the DBT group (7%) than control subjects (26%) attempted suicide during the year, this difference was not statistically significant (χ21=3.24; P=0.064). During the 26 weeks follow-up period only one patient of the DBT group (4%) attempted suicide, compared to six participants of the control condition (19%). This difference again showed no statistical significance (χ21=3.1; P=0.08).