Predictors and Moderators of Post-Treatment Depression Severity After Hypnotherapy and Cognitive Behavioral Therapy in an RCT for Mild-to-Moderate Depression · Brockmann AL, Pfeiffer L, Batra A et al.

4. Discussion

In this secondary analysis, we wanted to identify overall and treatment-specific predictors and moderators for treatment outcome with HT and CBT. Overall, we found that in addition to the negative influence of depressive symptoms at baseline on treatment outcome, self-efficacy was associated with better outcomes independent of the type of treatment and also in each treatment-specific regression analysis. However, we did not find support for treatment-related predictors such as cognitive and behavioral variables for CBT and hypnotic susceptibility or expectations for HT.

The negative effect of depressive symptoms at baseline on treatment outcome is already known (Hamilton & Dobson, 2002; Catarino et al., 2018). In severe depression in particular, the severity of depressive symptoms at baseline seems to be a consistent predictor for negative treatment outcome (Lyons & Delgadillo, 2025). Nevertheless, the small variance explained by depressive symptoms at baseline in our study underlines the relevance of other predictors.

The results of this study indicate that self-efficacy can be seen as a global predictor for outcome of psychotherapy for depression. This would be in line with Maddux and Meier (1995). Self-efficacy was found to enhance the probability of a treatment response in CBT for depressed patients (Stiles-Shields et al., 2015). In line with Alladin (2009) and Yapko and Criswell (2023), self-efficacy was also predictive for patients undergoing HT. This result indicates that HT techniques such as ego-strengthening (for example, by creating positive future images and integrating negative and positive experiences of the past; see also Alladin, 2009) could improve the course of depression during treatment for patients with higher self-efficacy at baseline. CBT is also characterized by providing a treatment rationale for the patient, as well as a transparent session structure, goal-oriented agenda setting, clear tasks and homework (A. T. Beck et al., 1979; J. S. Beck, 2020), and the idea of making the patient an expert on the disorder through psychoeducation and self-management (King & Boswell, 2019; Kanfer & Hagerman, 1981). One explanation for self-efficacy as a predictor for post-treatment severity could be that CBT as well as HT in our study contributed to the improvement in patients with high self-efficacy.

There was no association of rumination or behavioral activation with post-treatment outcome, either in the whole sample or in CBT. Rumination was considered crucial regarding relapse rates and recovery from depression (e.g., Michalak et al., 2011). However, in contrast to previous studies (Michalak et al., 2011; Upton et al., 2025), CBT treatment did not involve specific rumination-targeted techniques such as mindfulness or others, which could explain the absence of an effect related to cognitive factors. Nevertheless, behavioral activation was a substantial part of CBT treatment in this study, which is in line with the evidence (e.g., Ekers et al., 2014). We measured behavioral activation only via self-report and did not assess or track the actual “real” engagement of patients in activities. Therefore, we might not have been able to assess real changes in behavior with BADS.

The absence of a relationship between hypnotic suggestibility and treatment outcome could be explained by the type of HT used in the RCT. In the treatment with HT, a broad range of HT techniques was allowed. Therefore, HT not only included hypnotic techniques such as trances and imaginative work but also indirect techniques, seeding or finding metaphors, and externalizing the symptoms. The model behind the treatment manual was used mainly for the selection of treatment modules rather than the therapeutic rationale and background. The effect of the influence of suggestibility on the outcome of HT might have been influenced by the type of treatment. In Alladin and Alibhai (2007), for example, hypnosis was added to CBT, and some of the differential effects in their RCT might be explained by the suggestibility. Since HT is not financially covered by the health care system in Germany and is therefore considered a more “alternative” treatment for patients, the expectation of efficacy may have positively influenced the effect of the treatment with HT (Butler et al., 2006; Doering et al., 2018), which would be also in line with the social cognitive theory for hypnosis put forward by Lynn et al. (2008). In contrast, efficacy expectations were overall significantly higher for treatment with CBT than with HT in our study but were not significantly associated with treatment outcome. Nevertheless, there was a small correlation with outcome (r = −0.22) in patients with CBT, but expectations were not selected for the treatment-specific predictors or moderators, potentially due to the small power. HT is a treatment which focuses on reinstalling positive resources via imagination, time progression and association techniques (Yapko, 2006; Lankton & Matthews, 2010). Furthermore, depression is interpreted as a strategy that was helpful during the past to cope with interpersonal problems or difficult life events but is now no longer functional or in the patient’s volitional control (Yapko, 2006; Alladin, 2010). When integrated with CBT, HT can shape social and problem-solving skills and can contribute to better self-regulation strategies (see, for example, Yapko & Criswell, 2023). Future studies should also investigate global predictors for treatment outcome when both CBT and HT are integrated.

The lack of the support for the theory-driven variables in this study might also indicate that there are common factors (such as self-efficacy, expectations, or the therapeutic alliance) rather than therapy-specific ingredients that contribute to the outcome of psychotherapy independently of the type of treatment (see also Wampold, 2015). Despite assessing treatment fidelity and relying on treatment manuals, it could be that the some of the contextual factors (such as the therapeutic alliance, empathy, and treatment expectations) as well as the severity of the depression at baseline contributed more on the post-treatment severity than the expected treatment-specific variables. Nevertheless, previous research also failed to find consistent treatment-specific predictors for outcome, which could also be due to the different nature of study designs, sample sizes, and the complexity of the predictors (Eilertsen & Eilertsen, 2023).

As exploratory results, concomitant antidepressant medication was associated with poorer post-treatment outcome in the HT group in this exploratory secondary moderator analysis. This result is in contrast to most studies and meta-analyses, which found that severely depressed patients would benefit from psychotherapy with additional ADM (Cuijpers et al., 2014; Cuijpers et al., 2020). Similar to our results, some studies found the same for CBT without additional ADM (Lopez-Gomez et al., 2019; Whiston et al., 2019). Our result regarding ADM use, though, should be interpreted with care. ADM use was reduced to a binary variable despite different types of ADM being taken. Differences in depressive symptoms, educational level, prior treatment history, treatment resistance, chronicity, relapse risk, biological severity, comorbidity, patient preference, treatment expectations, and clinician judgment in patients taking ADM compared to those without ADM use might explain this.

While we found that patients with a lower level of education (lower than high school) showed better outcomes when receiving CBT instead of HT, other studies found that a higher level predicted a better treatment outcome (Hirschfeld et al., 1998; van Beljouw et al., 2010) and is considered to be a protective factor regarding mental health conditions such as depression and anxiety (Bjelland et al., 2008). In our RCT, the sample showed an overall high level of education (75% had a high-school or university degree) and cannot be considered as a representative sample. The results should be interpreted with care.

Limitations

Overall, it is important to note that this study is a secondary exploratory analysis of an RCT comparing HT with CBT in patients with mild-to-moderate depression. The results can only be interpreted with care and cannot be generalized to other populations beyond mild-to-moderate depression or to patients with severe or chronic depression and should be considered preliminary. With regard to sample size, our total sample with N = 141 fulfilled the rules of Green (1991), whereas the sample size should be N = 104 + m (number of predictors) for multiple linear regression. Regarding the subgroup analyses, the sample size was limited to 74 CBT and 68 HT patients. Another rule of thumb for regression analyses stated that the minimum size should be N = 20 + 5m. This could also be fulfilled by the subgroups because five variables were tested for their predictive value. Nevertheless, it is possible that smaller effects and correlations were not detected due to the lower power. Particularly with regard to the influence of expectation but also of behavioral activation on outcome, it can be assumed that the effect remained undetected due to the sample size. Furthermore, moderating effects were found for variables where the amount of information was artificially reduced to a binary variable and sample sizes in subgroups were small. Given that we select variables not only based on theoretical assumptions and treatment rationales but also based on correlation analyses in an explorative nature, there is also a risk of multiple testing, which can result in false-positive findings. We did not correct for multiple predictor and moderator analyses. Due to the stepwise regression, there could also have been an effect of overfitting. Finally, the influence of the predictors and moderators on treatment outcome is only of a correlative but not causal nature. To account for causal relationships or mediating effects, the predictive variables should have been additionally assessed during the treatment period. It is also important to note that it cannot be concluded that patients undergoing treatment with HT in a current depressive episode without ADM would show better effects when receiving HT instead of CBT. There could have been a confounding effect of ADM use, for example, with the severity of symptoms, the treatment history, and other clinical variables. The study was initially designed to compare HT with CBT in a non-inferiority RCT and not to measure explicit predictors and moderators. The study also lacks external validation.