Table 3.

Psychodynamic and MBT studies

Study design and number of studies (N) with references	Sample size, date, and country of publication	Cohort diagnoses and demographics	Main findings
MBT vs inactive/non-specialist comparators
RCT (n = 4) [43, 154–156]	Sample size: 20–100 (n = 4). Date: 1990–1999 (n = 1); 2000–2009 (n = 2); 2010–2019 (n = 1). Country: Asia (n = 1); UK (n = 3).	Diagnoses: “BPD” diagnosis (n = 4). Demographics: no data reported.	RCTs with primary outcomes: In the primary outcomes of 2 RCTs, compared to controls, participants receiving MBT showed improvement in the proportion of patients making suicide attempts (1/1) and in “BPD” symptoms (1/1). Compared to controls, participants receiving MBT showed improvement in all non-primary outcomes.
Non-randomised experiments, observational studies, quasi experiment, and natural experiment with pre-post comparison (n = 6) [157–162]	Sample size: < 20 (n = 2); 20–100 (n = 3); > 100 (n = 1). Date: 2010–2019 (n = 6). Country: Europe (n = 6).	Diagnoses: “BPD” (n = 4) or other “personality disorder” diagnosis (n = 1); “personality disorder” diagnosis and poor functioning (n = 1). Demographics: 100% female (n = 1).	In 1 non-randomised experiments, compared to controls, participants improved on some non-primary outcomes. In studies with comparisons over time only, participants showed improvements on all primary and non-primary outcomes.
MBT vs specialist comparators
RCT (n = 4) [163–166]	Sample size: 20–100 (n = 1); > 100 (n = 3). Date: 2000–2009 (n = 1); 2010–2019 (n = 3). Country: Europe (n = 3); UK (n = 1).	Diagnoses: “BPD” diagnosis (n = 3); “BPD” and suicide attempt or life-threatening self-harm (n = 1). Demographics: 50–79% White (n = 1).	RCTs with primary outcomes: In the primary outcomes of RCTs, compared to specialist controls, participants receiving MBT showed improvement in suicidal behaviours (1/1) and number of hospitalisations (1/1), but not in “borderline symptoms” (0/1). Compared to specialist comparators, participants receiving MBT did not show improvements in most non-primary outcomes.
Non-randomised experiments, observational studies, quasi experiment, and natural experiment with pre-post comparison (n = 5) [70, 167–170]	Sample size: 20–100 (n = 4); > 100 (n = 1). Date: 2010–2019 (n = 5). Country: Europe (n = 1), UK (n = 4).	Diagnoses: “BPD” (n = 2) or “personality disorder” diagnosis (n = 3). Demographics: 50–79% White (n = 1); 80–99% White (n = 3).	In non-randomised studies, compared to an alternative psychoanalytic model, the MBT group did not significantly improve on the primary outcome of bed use (0/1). Compared to specialist treatments, participants receiving MBT did not show improvements in more than half of non-primary outcomes. In 2 studies with comparisons over time only, participants improved on less than half of the outcomes.
Studies of modified MBT
RCT (n = 1) [171]	Sample size: > 100 (n = 1). Date: 2020- (n = 1). Country: Europe (n = 1).	Diagnoses: “Personality disorder” diagnosis (n = 1). Demographics: no data reported.	RCTs with primary outcomes: Compared to lower intensity outpatient MBT, higher intensity day hospital MBT showed no difference in the primary outcome of symptom severity and non-primary outcomes.
Psychodynamic treatment vs inactive/non-specialist comparators
RCT (n = 6) [109, 172–176]	Sample size: 20–100 (n = 4); > 100 (n = 2). Date: 1990–1999 (n = 2); 2000–2009 (n = 3); 2010–2019 (n = 1). Country: Europe (n = 3); North America (n = 3).	Diagnoses: “BPD” (n = 1) or other “personality disorder” diagnosis (n = 4); long term psychiatric difficulties disrupting functioning (n = 1). Demographics: no data reported.	RCTs with primary outcomes: In the primary outcomes of RCTs, compared to controls, participants receiving psychodynamic therapy showed improvement in symptom severity (2/2), social functioning (1/2), and interpersonal functioning (1/1), but not dysfunctional “borderline beliefs” (0/1), anxiety symptoms (0/1), or the number of participants meeting diagnostic criteria for a “personality disorder” diagnosis (0/1). Compared to controls, participants receiving psychodynamic therapy improved on most non-primary outcomes.
Non-randomised experiments, observational studies, quasi experiment, and natural experiment with pre-post comparison (n = 26) [48, 177–201]	Sample size: < 20 (n = 1); 20–100 (n = 18); > 100 (n = 7). Date: 1990–1999 (n = 6); 2000–2009 (n = 12); 2010–2019 (n = 7); 2020- (n = 1). Country: Australia (n = 7); Europe (n = 10); North America (n = 6); UK (n = 3).	Diagnoses: “Personality disorder” (n = 11) or “BPD” diagnosis/criteria (n = 8); “personality disorder” diagnosis and comorbid Axis I mental health problems (n = 3); treatment resistant depression with comorbid “personality disorder” and childhood trauma (n = 1); “personality disorder” diagnosis and poor interpersonal functioning (n = 2); poor interpersonal functioning (n = 1). Demographics: 100% female (n = 1); 80–99% White (n = 3); 100% White (n = 1).	In non-randomised experiments, participants showed improvements compared to controls on the following primary measures: reflective functioning (2/2), “personality disorder” symptoms (1/1), social functioning (1/1), and depressive symptoms (1/1). Compared to controls, participants improved on almost all non-primary outcomes. In studies with comparisons over time only, participants improved in all primary outcomes in interpersonal functioning (3/3) and symptom severity (1/1) and close to all non-primary outcomes.
Uncontrolled intervention development studies and single case study with multiple measures (n = 1) [202]	Sample size: 20–100 (n = 1). Date: 2000–2009 (n = 1). Country: North America (n = 1).	Diagnoses: “BPD” symptoms and suicidal or self-injurious behaviour (n = 1). Demographics: 100% female (n = 1); > 50% White (n = 1).	Studies with comparisons over time only: One uncontrolled feasibility trial found that patients given psychodynamic therapy improved over time on outcomes (1/1).
Psychodynamic treatment vs specialist comparators
RCT (n = 8) [67, 138, 139, 203–207]	Sample size: 20–100 (n = 8) Date: 1990–1999 (n = 1); 2000–2009 (n = 2); 2010–2019 (n = 5). Country: Europe (n = 5); Europe and North America (n = 1); North America (n = 2).	Diagnoses: “BPD” (n = 5) or other “personality disorder” diagnosis (n = 3). Demographics: 50–79% White (n = 1); 80–99% White (n = 2); 100% White (n = 1).	RCTs with primary outcomes: In primary outcomes of RCTs, compared to cognitive therapy, participants receiving psychodynamic therapy did not significantly improve in symptom severity (0/1). In 1/3 RCTs, compared to General Psychiatric Management, participants receiving psychodynamic therapy made significantly more overall progress in therapy overall. Compared to specialist controls, participants receiving psychodynamic therapy did not show improvements on almost any non-primary outcomes. One RCT did not make direct comparisons between groups.
Non-randomised experiments, observational studies, quasi experiment, and natural experiment with pre-post comparison (n = 4) [208–211]	Sample size: 20–100 (n = 3); > 100 (n = 1). Date: 1990–1999 (n = 1); 2010–2019 (n = 3). Country: Europe (n = 3); North America (n = 1).	Diagnoses: “BPD” (n = 2) or other “personality disorder” diagnosis (n = 1); “personality disorder” diagnosis with or without comorbid substance misuse (n = 1). Demographics: no data reported.	In 1 non-randomised experiment, compared to DBT, participants given Dynamic Deconstructive Psychotherapy had significantly greater improvement in the primary outcome of symptom severity (1/1). Compared to controls, participants improved on all or most non-primary outcomes. In 1 study with comparisons over time only, patients with and without comorbid substance misuse improved on outcomes.
Comparisons of psychodynamic treatment settings
Non-randomised experiments, observational studies, quasi experiment, and natural experiment with pre-post comparison (n = 6) [141, 212–216]	Sample size: > 100 (n = 6). Date: 2000–2009 (n = 2); 2010–2019 (n = 4). Country: Europe (n = 3), UK (n = 2); Europe and UK (n = 1).	Diagnoses: “Personality disorder” diagnosis (n = 5); severe “personality disorder” diagnosis (n = 1). Demographics: no data reported.	Six non-randomised experiments compared psychodynamic treatment in varying contexts. There were no significant differences between day hospital, outpatient, and inpatient services on the primary outcome (symptom severity) or non-primary outcomes. Community or step-down services resulted in significantly improved non-primary outcomes compared to residential services.
Studies of adapted psychodynamic treatment
RCT (n = 2) [217, 218]	Sample size: 20–100 (n = 2). Date: 2000–2009 (n = 1); 2010–2019 (n = 1). Country: North America (n = 2).	Diagnoses: “BPD” diagnosis and alcohol use or substance dependence (n = 2). Demographics: no data reported.	RCTs with primary outcomes: In the primary outcomes of RCTs, comparing Dynamic Deconstructive Psychotherapy combined with alcohol rehabilitation to TAU with alcohol rehabilitation for patients with co-occurring substance use disorders, Dynamic Deconstructive Psychotherapy patients showed significantly greater clinically meaningful improvement (1/1) and improved in alcohol misuse (1/1) and use of institutional care (1/1). Participants receiving Dynamic Deconstructive Psychotherapy showed significant improvements in more than half of non-primary outcomes compared to TAU.
Non-randomised experiments, observational studies, quasi experiment, and natural experiment with pre-post comparison (n = 1) [219]	Sample size: 20–100 (n = 1). Date: 2011–2019 (n = 1). Country: Europe (n = 1).	Diagnoses: “BPD” diagnosis (n = 1). Demographics: relatively low socio-economic status (n = 1).	In 1 non-randomised experiment, a brief psychoeducational program based on General Psychiatric Management was more effective than generic outpatient treatment (1/1).