Review of: Jennissen S et al, Am J Psychiatry 2018;175(10):961–969

STUDY TYPE: Systematic review and meta-analysis

What makes psychotherapy work? Contenders abound, including the therapeutic alliance, behavior change, and cognitive restructuring. This meta-analysis looked at the role that insight plays in psychotherapy outcomes.

The authors defined insight as the ability to understand 1) associations between past and present experiences, 2) patterns in relationships, and 3) the relationships between stress, emotions, and psychological symptoms. Their literature review included psychotherapy studies in adults that measured an outcome consistent with that definition. Inclusion criteria were otherwise broad, requiring that the study involved adults who sought therapy for any psychological condition.

From over 13,000 papers initially identified, only 23 met the full criteria. Although the authors’ definition of insight was based on psychodynamic theory, a variety of psychotherapies were included, with psychodynamic, supportive, and cognitive behavioral the most common.

The primary outcome was the effect size of the association between insight and psychotherapy outcomes. After employing a battery of meta-analytic techniques, the authors calculated a correlation coefficient of 0.3 for this association. That’s considered a moderate correlation, about the same as the correlation between age and blood pressure. It is also in the same range reported for other well-established psychotherapy moderators, such as the therapeutic alliance, positive regard, and empathy.

Strengths of the study include a large sample involving 1,112 total subjects and reliance on a standardized meta-analytic methodology. Although broad in scope, a weakness was the lack of a consensus measure of insight (18 instruments used, with a mix of clinician and self-report tools) or of outcome (27 outcome scales, most of them quantifying symptoms) among the studies included. In short, such a wide range of methods and measures allowed in the included studies “may be viewed as a threat to validity,” in the authors’ own words.

Beyond the technical difficulties in definition and measurement, the authors also acknowledge a statistical truism: that correlation does not prove causation. For example, it is possible that insight was not the causative agent, and that the gains in insight were secondary to other changes such as improvements in psychological symptoms or relationships.

TCPR’S TAKE
This study lends some empirical support to the age-old idea that insight is a mechanism of change in psychotherapy. When patients are aware of their own role in the problems they bring to therapy, they are better equipped to find adaptive solutions.