Exploring negative experiences in psychotherapy using an NLP approach on online forum data

Area of Science:

• Psychology
• Computational Linguistics
• Health Services Research

Background:

• Negative psychotherapy experiences affect 3-25% of patients, significantly impacting outcomes.
• The causes of these negative experiences are underexplored in traditional research.
• Online patient forums offer anonymous insights into dissatisfaction.

Purpose of the Study:

• To identify and analyze the causes and themes of psychotherapy dissatisfaction using online forum data.
• To expand existing frameworks for understanding negative therapy experiences.
• To highlight overlooked aspects of patient dissatisfaction.

Main Methods:

• Collected and anonymized online forum posts discussing psychotherapy.
• Utilized natural language processing (NLP) and large language models to identify dissatisfaction.
• Applied clustering, topic modeling, sentiment analysis, and classification, validated by human raters.

Main Results:

• Extracted 28,079 passages reflecting psychotherapy dissatisfaction.
• Identified 55 subthemes, including therapist misbehavior, negative effects, poor alliance, treatment mismatch, and healthcare frustrations.
• Uncovered themes such as structural barriers and lasting psychological consequences.

Conclusions:

• Psychotherapy dissatisfaction is frequent and multifaceted, extending beyond established categories.
• NLP and mixed-methods approaches can reveal overlooked themes in patient experiences.
• There is a critical need to better recognize and address negative therapy experiences and their consequences.