Study Design Participant n, gender,
mean age, diagnosis
Intervention and Follow-Up Relevant Measures of Mechanism and Outcome Hypothesized Mechanism(s) of Change and Key Findings Main Conclusions
Bedics et al.[33] RCT n=101
All female
29.3 years
One year follow up
Therapeutic Relationship
DBT participants reported greater self-affirmation, self-love, and self-protect across treatment and follow-up compared to the CTBE group. DBT therapists were experienced as more affirming, protecting, and controlling than CTBE therapists.
TR by itself did not predict outcomes (patient introject and nonsuicidal self-injury), but there was a significant interaction between treatment condition and TR for both outcomes:  DBT participants who experienced their therapist as affirming and protecting reported more positive outcomes. 
Goldman and Gregory [36] RCT n=10
9 female
1 male
27.4 years
BPD plus Alcohol Use Disorder
DDP WAI-observer version
Therapeutic Alliance
Quality of the TA was positively associated with improvements in BPD symptoms and alcohol misuse.
Although there was an association between TA and treatment outcome, the sample size was too small to establish mediation or moderation effects.
Gunderson et al.[42] Exploratory study  n=33
All female
17-35 years
Dynamically-informed and cognitive-behavioural long-term interventions HAQ
Therapeutic Alliance
TA improved over treatment.
TA ratings did not correlate with SCL-90 or GAS outcomes.
Ratings of TA are not reliably associated with change in symptoms.
Lingiardi et al. [43] Exploratory study n=47 patient-therapist dyads
31 female
16 male
29 years
Cluster A (n=12)
Cluster B (n=15)
Cluster C (n=20)
Individual outpatient psychotherapy CALPAS
Therapeutic Alliance
Cluster A patient TA ratings were significantly lower than Cluster B and Cluster C patient ratings. Therapist TA ratings were significantly lower for Cluster B patients than Cluster A and Cluster C patients.
Significant correlations between TA scores and symptom scores (causality not established), but did not measure this at end of treatment. Type of personality disorder impacts upon development of TA.
Marzialiet al.[35] Part of RCT n=18
Gender and average age not reported
Individual Dynamic Psychotherapy
(6-44 sessions; mean = 17.1 sessions)
Therapeutic Alliance
Strong association between early and later alliance scores. TA accounted for a significant proportion of the variance in all measure of treatment outcome at 12 months. TA ratings predicted social adjustment scores only at 24 months.
Early and later patient ratings of the TA can account for treatment outcomes at 12 months.
Spinhoven et al. [32] Part of RCT, exploratory study n=78
72 female
6 male
29.4 years (TFP)
31.7 years (SFT)
(3 years)
Therapeutic Alliance
TA ratings higher in SFT than TFP. Early TA ratings not predictive of clinical improvement. When treatment condition was controlled for, patient ratings of TA predicted later changes on BPDSI.
Treatment outcome can be partly accounted for by the quality of the TA. Quality of TA is affected by type of treatment approach.
Ulvenes et al. [34] Part of RCT n= 46
23 female
23 male
33.5 years (STDP)
34.3 years (CT)
Cluster C personality disorder
(40 sessions)
Psychotherapy Process Q- Short [81]
Therapeutic Alliance
TA positively associated with symptom reduction. Therapist avoidance of affect was positively associated with TA.  Avoidance of affect was associated with better outcomes in CT. A focus on affect was associated with better outcomes in STDP.
The therapeutic bond is associated with treatment outcome, but appears to interact differently with specific components of treatment according to type of treatment.
Muran et al. [50] Clinical trial n=128
68 female
60 male
41.3 years
Cluster C personality disorders
30-session CBT, BRT, and STDP WAI
Single-item measures about rupture intensity and RR
Therapeutic Alliance; Rupture Resolution
TA ratings predictive of change in symptoms. Ruptures rated as lower in intensity associated with higher TA. Greater RR significantly associated with higher TA. No significant correlation between RR and symptom change.
TA associated with treatment outcome. No causality established due to correlational design.
Strauss et al.
Non-randomized clinical trial n=30
34.2 years
28% additional PD
Therapeutic Alliance; Rupture Resolution
Early patient TA ratings were not correlated with early WISPI and SCID-II scores, but were significantly correlated with early BDI change.  RR episodes were associated with substantial (50% or greater) reduction in symptoms. 
Early TA ratings and number of RR episodes predict improvement in personality disorder symptoms and depression.
Daly et al. [48] Adjunct to RCT n=5
4 female
1 male
15.8 years
Rupture Resolution
Significant relationships between: i) Adherence to CAT model of responding to ruptures [46] and RR; ii) Adherence to model of RR and positive outcome; iii) RR and improvement on CGI-I.
RR linked to following the steps in Bennett et al.’s [46] model.
RR associated with better treatment outcome.
Note:n=number; RCT=Randomised Controlled Trial
Personality Disorders: APD: Avoidant Personality Disorder; BPD: Borderline Personality Disorder; OCPD: Obsessive Compulsive Personality Disorder
Interventions: BRT: Brief Relational Therapy; CAT: Cognitive Analytic Therapy; CBT: Cognitive Behavioural Therapy; CCT:Client-centred Therapy; CT: Cognitive Therapy; CTBE: Community Treatment by Experts; DBT: Dialectical Behavioural Therapy; DDP: Dynamic Deconstructive Psychotherapy; SFT: Schema-Focused Therapy; STDP: Short-term Dynamic Psychotherapy; TFP: Transference-Focused Psychotherapy
Mechanisms of Change: RR: Rupture Resolution; TA: Therapeutic Alliance; TR: Therapeutic Relationship
Measures: ARM: Agnew Relationship Measure [82]; BDI: Beck Depression Inventory [41]; BEST: Borderline Evaluation of Severity Index [83]; BPDSI-IV:  Borderline Personality Disorder Severity Index [84]; BPRS: Brief Psychiatric Rating Scale [85]; CALPAS: Californian Psychotherapy Alliance Scale [86]; CGI: Clinical Global Impression Improvement [87]; DDPRQ-10: Difficult Doctor Patient Relationship Questionnaire [88]; GAS: Global Assessment Scale [89]; HAQ: Helping Alliance Questionnaire [90]; IIP: Inventory of Interpersonal Problems; IPO: Inventory of Personality Organisation [73];OBI: Objective Behavioural Index [35]; P-TAS: Patient Therapeutic Alliance Scale [91]; SAS: Social Adjustment Scale [92]; SASBB: Structural Analysis of Social Behaviour [93]; SASII: Suicide Attempt Self Injury Interview [94]; SCID-II: Structured Clinician Interview for DSM-III-R Axis II Disorders [40]; SCL-90: Symptom Checklist [95]; SCL-90-R: Symptom Checklist Revised [96]; SEQ: Session Evaluation Questionnaire [97]; WAI: Working Alliance Inventory [37,98]; WISPI: Wisconsin Personality Disorder Inventory [39]; YSQ: Young Schema Questionnaire [99].
Table 2: A summary of included studies investigating the therapeutic relationship and alliance ruptures.