Substance Use May Weaken Group Therapy Bonds in Personality Disorder Care

One of the most stubborn challenges in modern psychiatry is the patient who carries two diagnoses at once: a personality disorder and a substance use disorder. Clinicians have long suspected that this combination makes therapy harder, not only because symptoms overlap, but because the very foundation of psychotherapy—the therapeutic alliance—might be eroded by ongoing drug or alcohol use. A new longitudinal, multi-site study from Norway now offers the most detailed picture yet of how comorbid substance use shapes the alliance in specialised personality disorder treatment, and its findings are more nuanced than the field expected.

The research, led by Katharina T. E. Morken of Haukeland University Hospital and the University of Bergen, together with colleagues at Oslo University Hospital, drew on data from the Network of Personality Disorders in Norway, a quality register covering specialised treatment units across the country. In total, 742 patients with personality disorders were followed during treatment delivered between 2017 and 2022. The scale and design of the study matter: rather than a snapshot of a single clinic, the investigators tracked patients repeatedly over time across many treatment centres, allowing them to separate changes within individual patients from stable differences between them.

At baseline, every participant underwent structured diagnostic interviews. Personality disorders were assessed with the Structured Clinical Interview for DSM-5 Personality Disorders, while other mental disorders were captured using the Mini International Neuropsychiatric Interview. Substance use was measured with the first three items of the Drug Use Disorders Identification Test and the three-item consumption version of the Alcohol Use Disorders Identification Test. Crucially, these substance use measures were repeated at six-month intervals throughout treatment, so the researchers could observe how fluctuations in a patient’s own drinking and drug use related to fluctuations in the quality of their therapeutic relationships.

The therapeutic alliance itself was measured from the patient’s perspective using two well-established instruments. For individual therapy, patients completed the Working Alliance Inventory Short Form, which breaks the alliance down into three components: the emotional bond between patient and therapist, agreement on the tasks of therapy, and agreement on its goals. For group therapy, patients completed the Group Questionnaire, which captures not only bond, work, and goal agreement but also the deeper sense of mutuality and collaboration within the group. These alliance ratings were also collected every six months, producing a dense longitudinal dataset.

To analyse this structure, the team used three-level hierarchical linear mixed models, a statistical framework designed for nested data. Repeated alliance assessments were nested within patients, and patients were nested within treatment centres. This approach allowed the investigators to ask a precise question: when a given patient reports higher substance use than their own typical level, does their alliance rating drop? This within-person design is statistically powerful because each patient serves as their own control, filtering out stable traits that might otherwise confound the association.

The headline result concerns group therapy. After accounting for time of assessment, personality disorder diagnosis, other mental disorders, gender, age, and treatment context, the researchers found a significant negative relationship between within-person increases in substance use and the work dimension of the Group Questionnaire. In plain terms, when patients used more drugs than their own average, they reported poorer mutuality in the collaborative work of their therapy groups. This is a specific and clinically meaningful finding: the work component of group alliance reflects how well members engage together in the shared task of treatment, and it is precisely this collaborative core that appears vulnerable to substance use.

Perhaps more striking is what the study did not find. There were no significant associations between substance use and total alliance scores in either group or individual treatment, and no significant links to the emotional bond in either setting. Trends emerged suggesting a negative association between substance use and agreement on tasks and goals in individual therapy, but these did not reach statistical significance. Alcohol consumption, measured by the AUDIT-C, showed no significant associations with any alliance measure. The damaging signal, in other words, was confined to drug use and to one specific facet of the group alliance.

The authors are candid about the implications of this pattern. The negative impact of comorbid substance use on the therapeutic alliance appears to be more limited than previous literature had suggested. Instead, they argue, the preponderance of alliance-related difficulties in patients with personality disorders may reflect the influence of the personality disorder itself, or other patient-level factors that have not yet been characterised. This reframing is important. If clinicians assume that substance use is the primary obstacle to alliance formation in this population, they may misattribute difficulties that actually stem from the interpersonal and emotional features of the personality disorder, such as mistrust, fear of abandonment, or unstable self-image.

For treatment planning, the study points toward pragmatism and responsiveness. The clinical significance statements accompanying the paper emphasise that a sensitive, responsive collaboration between patient and therapist may be critical when tailoring treatment for patients with combined personality disorder and substance use conditions. Because the observed effect was specific to group work and to the mutuality of collaboration, group-based programmes—the backbone of much specialised personality disorder treatment, including mentalization-based approaches delivered at several of the participating units—may need to monitor substance use actively and adjust group interventions when a patient’s use escalates. Rather than excluding these patients from group treatment, the findings suggest vigilance and flexibility.

Methodologically, the study also demonstrates the value of longitudinal, multi-level designs in psychotherapy research. Cross-sectional studies can show that patients with substance use disorders report weaker alliances, but they cannot determine whether substance use drives the alliance or whether some third factor explains both. By modelling repeated measurements within patients across eighteen treatment units, the Norwegian team could isolate the dynamic contribution of substance use with unusual clarity. The remaining uncertainty lies in the association between substance use and alliance being weaker than hypothesised, and in the possibility that missing data or unmeasured patient characteristics still shape the picture. Even so, the central message stands: in specialised personality disorder treatment, comorbid drug use specifically undermines the collaborative work of group therapy, while the broader alliance appears more resilient than feared—a finding that should reassure clinicians and sharpen their attention to the group setting where the vulnerability is real.

Subject of Research: The effect of comorbid substance use on therapeutic alliance in specialised personality disorder treatment

Article Title: The impact of comorbid substance use on individual and group therapeutic alliance in specialised personality disorder treatment: a longitudinal multi-site study

Article References: Morken, K. T. E., Molde, H., Bergwitz, A., Hummelen, B., Pedersen, G., & Kvarstein, E. H. (2026). The impact of comorbid substance use on individual and group therapeutic alliance in specialised personality disorder treatment: a longitudinal multi-site study. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08707-5

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08707-5

Keywords: therapeutic alliance, personality disorder, substance use disorder, group therapy, group cohesion, hierarchical linear models, longitudinal study, Working Alliance Inventory, Group Questionnaire, mental health, psychotherapy, comorbidity