Two-Year Randomized Controlled Trial and Follow-up of Dialectical Behavior Therapy vs Therapy by Experts for Suicidal Behaviors and Borderline Personality Disorder

The researchers compared two treatments for women with borderline personality disorder (BPD) who had recent suicidal or self-injurious behavior:

**Dialectical Behavior Therapy (DBT):** A structured, manualized treatment combining individual therapy, group skills training, phone coaching, and a therapist consultation team. DBT focuses on teaching emotion regulation, distress tolerance, interpersonal effectiveness, and mindfulness skills. Treatment lasted one year, with weekly individual sessions (1 hour) and weekly group skills training (2.5 hours).

**Community Treatment by Experts (CTBE):** A control condition designed to be the best possible comparison. Therapists were nominated by community mental health leaders as experts in treating suicidal/BPD patients. They used their own non-behavioral approaches (psychodynamic, supportive, eclectic, etc.) and were not trained in DBT. They provided whatever treatment they believed was best, with no restrictions on frequency or duration.

The primary outcomes were suicide attempts, self-injurious behaviors (e.g., cutting, burning), emergency services use (hospitalizations, emergency department visits), and treatment dropout. Secondary outcomes included general psychological functioning, depression, and anger.

**101 women** aged 18–45 (mean age ~30 years)

All met DSM-IV criteria for borderline personality disorder

All had at least two suicide attempts or self-injury episodes in the past 5 years, with at least one in the past 8 weeks

Recruited from a university outpatient clinic and community referrals in Seattle, Washington

Exclusion criteria: bipolar I disorder, schizophrenia, other psychotic disorders, seizure disorder requiring medication, or a primary diagnosis of substance dependence requiring detoxification

Participants were matched on age, number of past suicide attempts, negative prognostic indicators (e.g., prior hospitalizations), and number of lifetime self-injuries and psychiatric hospitalizations before randomization

**Suicide Attempt Self-Injury Interview (SASII):** A structured interview assessing the frequency, method, medical severity, and intent of suicidal and self-injurious acts. Administered every 4 months (trimesters) by blinded assessors.

**Medical Risk Scale:** Rated the medical lethality of each suicide attempt/self-injury act on a 0–6 scale (0 = no risk, 6 = death). Higher scores = more dangerous behavior.

**Treatment History Interview (THI):** Tracked psychiatric hospitalizations, emergency department visits, and outpatient treatment use.

**Hamilton Rating Scale for Depression (HAM-D):** Clinician-rated depression severity.

**State-Trait Anger Expression Inventory (STAXI):** Self-reported anger.

**Global Assessment of Functioning (GAF):** Clinician-rated overall psychological, social, and occupational functioning (0–100 scale).

**Treatment dropout:** Defined as missing 4 consecutive weeks of scheduled sessions without returning.

All assessors were blind to treatment condition. Inter-rater reliability was checked and maintained above 0.80 for all key measures.

**Design:** Two-year randomized controlled trial (RCT) with 1 year of active treatment and 1 year of post-treatment follow-up. Participants were randomized to DBT or CTBE using an adaptive randomization procedure that balanced groups on four matching variables (age, suicide attempt history, negative prognostic indication, and lifetime self-injuries/hospitalizations).

**Randomization:** Yes, adaptive randomization (not simple coin-flip). This means the computer algorithm adjusted assignments during the trial to keep the two groups balanced on key prognostic variables. This is stronger than simple randomization because it reduces the chance that one group ends up sicker by luck.

**Blinding:** Outcome assessors were blind to treatment condition. Participants and therapists obviously could not be blind (you know whether you're learning DBT skills or doing psychodynamic therapy). This is a limitation, but the blinding of assessors reduces bias in outcome measurement.

**Duration:** 12 months of treatment, then 12 months of follow-up with no study-provided treatment (participants could seek any treatment they wanted during follow-up). Assessments occurred every 4 months during both phases.

**Statistical approach:** Intent-to-treat analysis (all participants analyzed in the group they were randomized to, regardless of whether they completed treatment). Survival analysis (Cox proportional hazards models) for time-to-event outcomes (suicide attempts, dropout). Mixed-effects models for repeated measures (hospitalizations, emergency visits). This is appropriate for handling missing data and correlated observations over time.

**What this design can prove:** The RCT design with blinded assessment can establish causality—that DBT caused the observed differences in outcomes compared to CTBE. The adaptive randomization and matching reduce the chance that pre-existing differences explain the results. The 1-year follow-up tests whether effects persist after treatment ends.

**What this design cannot prove:** It cannot tell us which specific component of DBT is responsible for the effects (the "active ingredient"). It cannot rule out that DBT's superiority is due to its structured, manualized nature rather than its specific content (any structured treatment might beat unstructured expert therapy). It cannot generalize to men, adolescents, or people without BPD. The non-blinded therapists mean that therapist allegiance effects (therapists believing in their approach) could contribute to outcomes.

**Major methodological weaknesses:**

Small sample size (101 participants) limits statistical power for subgroup analyses

No placebo or "treatment as usual" control—only compared to expert therapy

High rate of psychiatric medication use in both groups (about 70% on antidepressants, 40% on anxiolytics), which was not controlled

The CTBE condition, while designed to be strong, may have been less structured than DBT, making it a comparison of structured vs. unstructured treatment rather than DBT vs. non-behavioral therapy specifically

Follow-up period allowed naturalistic treatment, so differences during follow-up could reflect ongoing treatment differences rather than lasting effects of DBT

**Primary outcomes (suicide attempts and self-injury):**

**Suicide attempts:** Over 2 years, 23.5% of DBT participants made at least one suicide attempt vs. 46.9% of CTBE participants. DBT participants were half as likely to attempt suicide (hazard ratio = 2.66, 95% CI not reported, p = 0.005). This means the risk of attempting suicide was 2.66 times higher in CTBE than DBT.

**Self-injurious acts (non-suicidal):** No significant difference between groups in the number of self-injury episodes (p = 0.17). However, when combining suicide attempts and self-injury, DBT participants had lower medical risk scores (F(1,50) = 3.2, p = 0.04), meaning their self-harm acts were less medically dangerous.

**Suicide ideation hospitalizations:** DBT participants required fewer hospitalizations for suicide ideation (F(1,92) = 7.3, p = 0.004).

**Secondary outcomes (treatment utilization and dropout):**

**Treatment dropout:** 19.6% of DBT participants dropped out vs. 46.9% of CTBE participants over 1 year. DBT participants were 3.2 times less likely to drop out (hazard ratio = 3.2, p < 0.001).

**Psychiatric hospitalizations:** DBT participants had fewer psychiatric hospitalizations during the treatment year (F(1,92) = 6.0, p = 0.007). The effect persisted during follow-up but was smaller.

**Emergency department visits:** DBT participants had fewer psychiatric emergency department visits (F(1,92) = 2.9, p = 0.04).

**Other outcomes:**

**Depression (HAM-D):** Both groups improved, but DBT showed significantly greater reduction in depression scores at the end of treatment (p = 0.03). By the end of follow-up, the difference was no longer significant.

**Anger (STAXI):** DBT participants showed greater reductions in anger expression (p = 0.05) at the end of treatment, but not at follow-up.

**Global functioning (GAF):** Both groups improved, with no significant difference between groups.

**Suicide attempt risk reduction:** DBT reduced the risk of attempting suicide by about 53% relative to CTBE. In absolute terms, about 23 out of 100 DBT participants attempted suicide over 2 years, compared to 47 out of 100 CTBE participants. This means you would need to treat about 4 people with DBT instead of expert therapy to prevent one additional suicide attempt (number needed to treat = 4.3).

**Dropout reduction:** DBT kept 80% of participants in treatment for the full year, compared to only 53% in CTBE. This is a 27 percentage point improvement in treatment retention.

**Hospitalization reduction:** The exact number of hospitalizations saved is not reported as a simple count, but the statistical tests show a clear reduction. Given that psychiatric hospitalizations cost $1,000–$2,000 per day in 2006, this represents substantial cost savings.

**Medical risk reduction:** Among those who did self-harm, DBT participants' acts were rated as less medically dangerous. This is clinically meaningful because it reduces the chance of accidental death or permanent injury.

**Acknowledged by authors:**

Small sample size (101 participants) limits generalizability and statistical power for detecting differences in rare events

All participants were female, so results may not apply to men with BPD

The CTBE condition, while designed to be expert treatment, may not represent the best possible non-DBT treatment

High rates of medication use in both groups could confound results

The study could not blind participants or therapists to treatment condition

**Critical reader observations:**

The study was funded by the National Institute of Mental Health, but the lead author (Marsha Linehan) developed DBT, creating a potential conflict of interest. However, the use of blinded assessors and independent statistical analysis mitigates this concern.

The CTBE therapists were nominated by community leaders, but their actual expertise in treating suicidal BPD patients was not independently verified. Some may have had limited experience with this specific population.

The study excluded people with bipolar I disorder, schizophrenia, and primary substance dependence, so results do not apply to these groups.

The follow-up period allowed participants to seek any treatment, so differences during year 2 could reflect ongoing treatment disparities rather than lasting effects of DBT.

The study did not measure quality of life, social functioning, or employment outcomes, which are important for understanding real-world impact.

The statistical analysis used multiple comparisons without explicit correction for familywise error rate, though the primary outcomes were pre-specified.

For someone running their own n=1 experiment (or working with a therapist to test DBT-like approaches):

**What to test:**

The core DBT skills package: mindfulness, distress tolerance, emotion regulation, and interpersonal effectiveness. You don't need a full DBT program to test these skills individually.

Specific skills to try: "STOP" skill (Stop, Take a step back, Observe, Proceed mindfully) for crisis moments; "Opposite Action" for emotion regulation; "DEAR MAN" for interpersonal conflict.

**Minimum meaningful duration:**

The study used 12 months of treatment, but measurable improvements in suicide attempts appeared within the first 4 months. For a self-experiment, commit to at least 8–12 weeks of daily skills practice before evaluating effects.

Skills training groups met weekly for 2.5 hours. For self-experimenters, aim for 30–60 minutes of daily skills practice (reading, exercises, journaling).

**What to measure (specific metrics):**

**Primary metric:** Frequency of suicidal thoughts or self-harm urges (rate per week). Use a 0–10 scale for urge intensity.

**Secondary metrics:** Number of actual self-harm episodes; number of crisis calls or emergency room visits; days between episodes (survival time).

**Process metrics:** Daily skills practice time (minutes); number of skills used per week; self-rated distress tolerance (0–10 scale before and after using a skill).

**Mood metrics:** Daily mood rating (0–10); anger episodes per week; sleep quality (hours, disruptions).

**Key confounds to control for:**

**Medication changes:** If you're on psychiatric medication, keep doses stable during the test period. Any change could confound results.

**Therapist contact:** If you're also in therapy, note the frequency and type. The study compared DBT to expert therapy—if you're doing both, you can't isolate DBT's effect.

**Life stressors:** Major life events (job loss, relationship breakup, death) can trigger suicidal crises. Track these separately and note whether skills helped you cope.

**Substance use:** Alcohol and drugs can increase impulsivity and reduce skill effectiveness. Track use separately.

**Sleep and exercise:** Both affect mood regulation. Keep these stable if possible, or track them as covariates.

**What a positive result would look like:**

A 50% or greater reduction in suicidal thoughts or self-harm urges within 8–12 weeks

At least 2 consecutive weeks without any self-harm behavior (compared to baseline frequency)

A decrease in the medical severity of any self-harm that does occur (e.g., moving from cutting to less dangerous behaviors)

Improved ability to tolerate distress without acting on urges (e.g., urge intensity drops from 8/10 to 4/10 within 30 minutes of using a skill)

Fewer crisis calls or emergency visits (e.g., from 2 per month to 0 per month)

Reduced treatment dropout: if you're in therapy, staying in treatment for the full test period is itself a positive outcome

**Important caveat:** This study tested DBT as a complete package delivered by trained professionals. A self-experiment with individual skills is not equivalent. If you have active suicidal thoughts or self-harm urges, work with a mental health professional. Do not attempt to treat serious suicidal behavior on your own.