This study was the first to incorporate both direct and indirect measures into the assessment of anger and aggressiveness self-concepts in patients with OCD. Furthermore, it assessed possible mediators that could explain the relationship between OCD and anger or anger suppression, namely an inflated sense of responsibility, non-acceptance of emotions, and social desirability. Based on psychodynamic  and cognitive theories , as well as previous studies (e.g., [8, 11, 17]), we assumed differences between patients with OCD and healthy controls regarding trait anger and anger suppression. Furthermore, we expected that an inflated sense of responsibility would mediate the relationship between OCD and anger or anger suppression.
Trait anger and suppressed anger according to a direct measure
As expected, patients with OCD reported higher trait anger and anger suppression in the direct self-report (STAXI-2) compared to healthy controls. Both results are in line with the majority of previous studies [8,9,10,11]. Higher self-reported anger suppression supports the assumptions of Rachman  and Freud , who suggested that anger is suppressed rather than expressed in OCD. According to Rachman , individuals with OCD would experience feelings of guilt and a tendency to blame themselves rather than others, because they feel responsible for preventing harm and failing to fully do so.
Aggressiveness self-concept as measured by an indirect measure (IAT)
Contrary to our expectations, we found no difference in the aggressiveness self-concept as measured by an indirect measure (Agg-IAT) between patients with OCD and healthy controls. The exploratory analysis that compared patients with checking-related compulsions (according to the OCI-R) and healthy controls revealed similar results. In the first study that used an Agg-IAT in patients with OCD , the whole sample of patients with OCD and healthy controls did not differ, however a subsample of patients with checking-related compulsions (according to the OCI-R) showed a more peaceful self-concept compared to the control group. Even though scores of patients with OCD and patients with checking-related symptoms of OCD did not statistically differ in the previous study, they showed a nominal difference (checking: D2-score = − 0.67; whole OCD sample: D2-score = − 0.54). Yet, the D2-scores in this study were very similar between patients with checking-related compulsions and the whole OCD sample (checking: D2-score = − 0.52; whole OCD sample: D2-score = − 0.51). Several reasons could explain the difference between the aggressiveness self-concept of patients with checking-related symptoms in the previous study and this study. One reason could be that patients in the previous study had more severe symptoms of OCD, especially checking-related symptoms. The patients with checking-related symptoms in this study had a descriptively slightly higher Y-BOCS score (Cludius et al. 2017, M = 24.81; this study, M = 27.10, t(45) = 1.93, p = 0.06). However, the disorder duration was descriptively slightly longer in the previous study (Cludius et al. 2017, M = 16.94; this study, M = 11.55, t(45) = 1.77, p = 0.08). The score for checking-related symptoms on the OCI-R was similar (Cludius et al. 2017, M = 9.14; this study, M = 9.00, t(45) = 0.24, p = 0.81). Therefore, differences in disorder severity do not seem likely to explain differences in the aggressiveness self-concept in this study compared to the previous study. Another reason could be that the sample in this study was too small to replicate the results of the previous study. Pooling the Agg-IAT effect sizes from the present and the former studies  in a mini meta-analysis (fixed effects; Hedges g; 95% CI) also resulted in non-significant group differences [healthy controls vs. all patients g = 0.05 (− 0.25; 0.36); healthy controls vs. patients with checking-related compulsions g = 0.28 (− 0.10; 0.66)]. Taken together, it could be possible that the more peaceful self-concept in the previous study was a mere false positive finding and that the modest sample sizes in both studies render replications of previous results difficult. Another explanation could be that the Agg-IAT may not be specific enough to assess implicit aggressiveness in OCD. The stimulus words used in the Agg-IAT refer to overt aggression (e.g., fight, revenge) which may not appropriately depict anger and aggressiveness related to OCD. Future studies could establish and use indirect measures which tap more specifically into the concept of suppressed anger.
Possible reasons for elevated anger and anger suppression scores in OCD
First, social desirability, non-acceptance of negative emotions, and an inflated sense of responsibility were tested as possible mediators between OCD and anger or anger suppression. Social desirability did not mediate the relationship between group and anger or anger suppression scores.Patients and healthy controls did not differ on the social desirability scale (SES-17). Thus, it is very unlikely that patients report anger differently because of differences in how they present themselves, for example, due to high moral standards.
Second, an inflated sense of responsibility did not mediate the relationship between group and anger, but mediated the relationship between group and anger suppression. This is in line with the theory by Rachman , who assumed that anger would be suppressed rather than expressed in patients with OCD, because they take full responsibility. According to this, anger suppression would be a consequence of OCD. However, as this study did not measure temporal relationships, the causality remains to be tested.
Third, non-acceptance of negative emotions mediated the relationship between group and trait anger as well as anger suppression. To the best of our knowledge, no previous study has assessed that link in OCD. Regarding depression, a study showed that emotional suppression was only associated with symptoms of depression when moderated by non-acceptance of emotions . In an ecological momentary assessment study in patients with borderline-personality disorder, non-acceptance of negative emotions increased the number of negative complex emotions prior to symptoms of self-injury . This may indicate that lower acceptance of negative emotions could lead to an increase in anger, which leads to a greater number of OCD symptoms. However, laboratory or ecological momentary assessment studies are needed to assess this causal link. Furthermore, next to OCD, MDD, and borderline personality disorder, non-acceptance of negative emotions has been found to be associated with other mental disorders, such as eating disorders  and social anxiety disorder . This might indicates a transdiagnostic factor which likely refers to all kinds of internalizing disorders . Future research is necessary to assess whether the link between non-acceptance of negative emotions and anger or anger suppression is specific to OCD or whether it is associated with other (groups) of disorders.
Fourth, our analyses regarding the subgroups of patients (with or without a current comorbid MDD or current comorbid anxiety disorder, intake or no intake of antidepressants) give an indication that group differences to healthy controls, regarding trait anger, are not attributable to comorbid MDD, anxiety disorder or to the intake of antidepressant medication. However, it is possible that comorbidity explains some of the difference regarding anger suppression. Only patients with current comorbid MDD reported higher anger suppression compared to healthy controls, whereas patients with no current comorbid depression did not show significantly higher anger suppression scores. Similar results were found when dividing the group of patients into those with and without anxiety disorders and those who use antidepressants and those who do not. Notably, the sample sizes of the subgroups were quite small and the non-significant effects showed, in some cases, moderate effect size estimates. Nevertheless, this study cannot fully differentiate whether higher anger suppression in OCD is specifically attributable to OCD or rather to MDD, anxiety disorders, or the intake of antidepressants.
This study shows a number of limitations. First, most patients with OCD reported a comorbid disorder according to the M.I.N.I., mostly anxiety disorders or MDD. Furthermore, patients on average showed moderate depressive symptoms on the PHQ-9. Some studies indicated that anger and anger suppression could be moderated by depressive symptoms, as relationships between OCD and anger were no longer significant after controlling for depression [11, 13]. However, by controlling for depressive symptoms it is likely that relevant variance in the relationship between OCD and anger is removed (see ). Also, many patients were medicated, mostly with antidepressants (SSRIs). The serotonergic system has been linked to aggression, whereas antidepressants have been found to reduce aggression in patients with a personality disorder . However, excluding patients without comorbid disorders or antidepressant medication would have reduced the ecological validity of our study as comorbidity rates, especially with MDD and anxiety disorders, are generally high in OCD . Future studies should be designed to test those possible mediators in the relationship between OCD and anger or anger suppression, for example, by including clinical control groups or recruiting larger samples. Second, an inflated sense of responsibility was measured in one dimension with overestimation of threat, as those load on the same factor . Our results related to the inflated sense of responsibility are limited insofar, as we cannot draw conclusions about each single dysfunctional belief but only the respective bias dimension. Third, as stated above, this study gives an indication as to whether certain mediators are predictive of anger or anger suppression. However, it does not prove causal links. Future studies should use laboratory or longitudinal designs to test if an inflated sense of responsibility and non-acceptance of emotions are risk factors for the development of anger or anger suppression.
Cognitive behavioral therapy with exposure and response prevention (ERP) is the gold standard intervention in OCD. However, about 30% of those who undergo ERP do not profit sufficiently. Additionally, about a quarter of patients with OCD refuse to engage in ERP at all, and 30% discontinue the treatment prematurely . Our results may give some indication that training adaptive emotion regulation strategies may be beneficial as an add-on to ERP, which could also reduce drop-out rates and enhance remission. First, our preliminary evidence shows that targeting anger suppression may be important. According to Rachman , anger can arise in patients with OCD due to an inflated sense of responsibility, when additional demands are made on the individual. Thus, the person may feel overwhelmed by tasks that he/she should fulfill in therapy, especially when engaging in ERP. As a consequence, anger may be provoked during therapy and may even be directed towards the therapist. It might be helpful for patients to know that it has not been empirically found that patients show a more aggressive self-concept, neither in this study nor in a previous study . Furthermore, it could be beneficial if patients learned to reduce anger suppression, for example, using more adaptive emotion regulation strategies. Similarly, as non-acceptance of emotions mediated the relationship between OCD and anger or anger suppression, patients could also profit from a training to enhance emotional acceptance. For example, emotion regulation therapy  or acceptance and commitment therapy approaches  could help patients with OCD learn to identify and accept instead of avoid their emotions and to use the emotional information to identify needs or guide behavior and thinking rather than suppressing emotions. In a single-case study, an emotional awareness training was combined with exposure to OCD-specific and non-specific emotional cues . This study provided first evidence that learning acceptance of emotions could improve exposure therapy, at least for some patients. Acceptance and commitment therapy, which aims to teach patients to accept negative emotional states, has shown some efficacy in OCD  but might not increase efficacy when added to ERP . However, as the acceptance and commitment approach is not inferior , it might be an alternative in cases, in which the standard approach does not work (or is not feasible). Third, meta-cognitive interventions, such as meta-cognitive training , could help to reduce dysfunctional beliefs, such as an inflated sense of responsibility, which may lead to a further reduction of anger and anger suppression.