Does perfectionism in bipolar disorder pedigrees mediate associations between anxiety/stress and mood symptoms?

The high rate of comorbidity between anxiety disorders and bipolar disorder is well established (Pavlova et al. 2015). Anxiety disorders have been shown to precede the onset of bipolar disorder (BD) (de Graaf et al. 2003; Perugi et al. 2001) and there are high rates of anxiety disorders in families affected by BD (Nurnberger et al. 2011; Perich et al. 2015; Merikangas et al. 2014). However, the mechanisms that underlie the comorbidity between anxiety disorders and BD are not fully understood (Mitchell 2015; Provencher et al. 2012). To move beyond purely descriptive studies, we propose that these disorders co-occur because they share common maintaining processes (Harvey et al. 2004). Cognitive styles are important maintaining factors in cognitive models of psychopathology (Harvey et al. 2004), and we propose that the cognitive style of perfectionism may be an explanatory factor in the high co-occurrence of BD and AD’s. Here, we specifically examined perfectionism as a potential mediator of known associations between anxiety and stress symptoms and mood symptomatology (Alloy et al. 2006; Boylan et al. 2004; Corry et al. 2013), in a large family study.

The development of BD is influenced strongly by genetics, accounting for as much as 85% of the variance in who develops mania (McGuffin et al. 2003). In families, a summary of 5 studies estimated that the risk of an affective disorder in first degree relatives of those with BD ranges from 24 to 31% and the risk of BD ranges from 7 to 22% (Merikangas et al. 2002). Given this high genetic contribution to the development of BD, there remains the task of identifying how and when this biological vulnerability is expressed with more proximal biological, psychological and environmental triggers being important factors (Johnson 2005; Jones and Bentall 2008) BD pedigrees have higher rates of Schizoaffective Disorder (BD type), BD-I, BD-II and Major Depressive Disorder that control families (Gershon et al. 1982). There is also emerging evidence that the family environment plays an important role in influencing the development of beliefs and attitudes related to achievement and attainment of goals (Johnson 2005; Chen and Johnson 2012). Hence, studies of extended families with some members diagnosed with BD may be useful to determine the contribution of perfectionistic cognitions and beliefs relating to goal attainment to anxiety, stress, and other core mood symptoms such as depression and hypomania.

Cognitive models of the development of psychopathology propose that cognitive, mood and behavioural symptoms arise when maladaptive beliefs and cognitions are triggered by congruent life events (Beck et al. 1979). Recently, cognitive models of BD symptoms and mood dysregulation have been proposed that attempt to explain both the depressive and (hypo)manic symptoms characteristic of BD (Holmes et al. 2008; Mansell et al. 2007). Mansell et al. (2007) have proposed a transdiagnostic model of mood dysregulation and bipolar disorder, within which it is proposed that self-critical or shaming beliefs may be important in driving anxious thoughts and cognitive appraisals of affect, as well as bodily sensations that ultimately influence the ascent into (hypo)mania or descent into depression (Mansell et al. 2007). In parallel, a separate cognitive model of BD emphasizes the role that anxiety plays in the development of BD symptoms (Holmes et al. 2008). The model of Mansell et al. (2007) in particular draws on findings that perfectionism is a core cognitive style of BD, along with high self-criticism and an emphasis on goal attainment and avoidance of failure (Alloy et al. 2009a; Lam et al. 2004; Mansell and Pedley 2008; Scott et al. 2000). For example, in a prospective study, self-criticism, performance focus and high self-standards interacted with congruent life events to predict both depressive and (hypo)manic symptoms in those with BD-II and cyclothymia (Francis-Raniere et al. 2006). Lam et al. (2004) found that beliefs related to goal attainment distinguished those with BD from normal controls and were correlated with number of hospitalizations for mania and BD episodes in general (Lam et al. 2004). Hewitt et al. (1998) examined perfectionism as a multidimensional construct in a mixed sample of unipolar and BD patients. Hewitt found that self-oriented perfectionism (whereby the individual has high standards for themselves) was uniquely associated with chronic depression symptoms, while socially-prescribed perfectionism (whereby individuals perceive that others have high standards for them) was been uniquely associated with chronic (hypo)manic symptoms (Hewitt et al. 1998). A perfectionistic cognitive style has also been associated with anxiety and stress symptoms (Hewitt and Flett 2002; Frost and DiBartolo 2002; Wheeler et al. 2011). Previous associations with perfectionism extend to variations in state and trait anxiety (Flett et al. 1995), social anxiety disorder (Juster et al. 1996), obsessive–compulsive disorder (Antony et al. 1998), panic disorder (Antony et al. 1998) and a tendency to worry (Chang 2000). Consistent with this, greater levels of perfectionism have also been associated with greater cortisol responses to stress in situations designed to induce a fear of negative evaluation (Wirtz et al. 2007) a core cognitive style in social anxiety disorder. These findings, which confirm perfectionism as a key cognitive style in both BD and the anxiety disorders, provided the impetus to examine perfectionism as a potential mediator of known associations between anxiety and stress symptoms and mood symptomatology.

Consistent with the psychological models of BD symptoms (Holmes et al. 2008; Mansell et al. 2007), we have previously reported that anxiety and stress symptoms mediated the effects of self-critical perfectionism and goal attainment beliefs on current bipolar depressive symptoms after controlling for current hypomanic symptoms (Corry et al. 2013). For hypomanic symptoms, stress symptoms were a significant mediator of the relationship between self-critical perfectionism and current hypomanic symptoms; however, these mediation models were no longer significant after controlling current depressive symptoms. Similar findings were reported by O’Garro-Moore et al. (2015) using prospective data from those with bipolar spectrum disorders. Of the six cognitive styles examined as mediators, only perfectionism was a significant mediator of the relationship between the presence of an anxiety disorder and depressive symptoms. No cognitive styles significantly mediated the relationship between anxiety disorder and hypomanic symptoms.

This study thus examined the hypotheses that perfectionism and goal attainment values would mediate the associations between (subclinical) anxiety/stress symptoms and hypomanic and depressive symptomatology, in BD patients and their unaffected relatives. We chose to extend the work of Hewitt et al. by examining perfectionism as a multidimensional construct in a family sample. We hypothesized that: (1) self-oriented perfectionism would mediate the relationship between anxiety and stress symptoms and chronic depressive symptoms; and (2) socially-prescribed perfectionism and goal attainment values would mediate the relationship between stress and anxiety symptoms and chronic (hypo)manic symptoms. In line with Hewitt et al. (1998) findings, these predictions reflected our expectation that: (1) self-oriented perfectionism would be uniquely related to chronic depression depressive symptoms; and (2) socially-prescribed perfectionism and goal attainment values would be uniquely related to chronic (hypo)manic symptoms.

The results highlight the potential role of perfectionism as a mechanism explaining the co-occurrence between BD and anxiety. The results further suggest that different dimensions of perfectionism are uniquely associated with chronic depressive and (hypo)manic symptoms. Univariate associations between self- and socially-prescribed perfectionism and BD symptomatology were consistent with previous reports (Lam et al. 2004; Hewitt et al. 1998). Similarly, we confirmed significant relationships between both self- and socially-prescribed perfectionism and anxiety and stress symptoms (Wheeler et al. 2011). Like Hewitt (Hewitt et al. 1998), we found that self-oriented perfectionism was uniquely associated with chronic depressive symptoms after controlling for chronic (hypo)manic symptoms; this suggests a preoccupation with high standards for oneself and a punitive response when mistakes are made, or when perfectionistic standards are not met. Previous studies have found self-oriented perfectionism to be associated with unipolar depressive symptoms (Hewitt and Flett 1993; Hewitt et al. 1996) and there are similarities between the cognitive styles of unipolar and bipolar depression (Scott and Pope 2003). The particular relevance of self-oriented perfectionism for depression (versus socially-prescribed perfectionism) is consistent with research highlighting the role of ruminative, self-focused attention in the onset and maintenance of depression (Nolen-Hoeksema 1991, 2000; Kuehner and Weber 1999). Self-analytical self-focus has been found to influence global negative self-judgments (e.g. “I am a failure”) which increases depressive symptoms (Rimes and Watkins 2005).

For hypomanic symptoms, our findings suggest that an interpersonal focus is more important given that socially-prescribed perfectionism is defined as perceiving that others hold excessively high standards for oneself. We found that socially-prescribed perfectionism was uniquely associated with chronic (hypo)manic symptoms after controlling for chronic depressive symptoms. This is consistent with Shepero et al’s (Shapero et al. 2015) finding that higher public self-consciousness (reflecting a person’s awareness of other’s view of themselves) distinguished BD patients from those with MDD (along with higher rumination on positive affect) (Shapero et al. 2015). Socially-prescribed perfectionism is also considered to be more maladaptive than self-oriented perfectionism, with self-criticism being the most problematic component (Blankstein and Dunkley 2002). In a sample of individuals with BD-II and cyclothymia, Alloy et al. found that higher autonomy (a construct related to perfectionism) and self-criticism predicted a greater likelihood of (hypo)manic episodes over 3.2 year follow-up (Alloy et al. 2009b). Our findings strengthen the hypothesis that a cognitive style characterized by a concern about the negative perceptions of others is associated with hypomanic symptoms.

The second aim of our study was to examine whether perfectionism mediates the relationship between stress and anxiety symptoms and BD symptoms in an attempt to understand mechanisms underlying the high comorbidity between anxiety symptoms and BD. For chronic bipolar depressive symptoms, the mediation models were significant with self-orientated perfectionism found to significantly mediate the relationship between anxiety and stress symptoms and bipolar depressive symptoms. These findings remained after controlling for the presence of chronic (hypo)manic symptoms. These findings are consistent with our previous finding in a clinical sample (Corry et al. 2013). Our hypothesis that socially-prescribed perfectionism and goal attainment values would mediate the relationship between anxiety and stress symptoms and chronic hypomanic symptoms was partially supported. Only socially-prescribed perfectionism (not goal attainment values) was uniquely associated with chronic hypomanic symptoms in the linear regression models in the current study; therefore, we considered only socially-prescribed perfectionism in the mediation models. In mediation models, socially-prescribed perfectionism significantly mediated associations between both anxiety and stress, and chronic hypomanic symptoms after controlling for the presence of chronic depressive symptoms. This finding is inconsistent with our previous study and likely represents methodological differences in the measurement of (hypo)mania (chronic in the present study versus current in our previous study) (Corry et al. 2013). It is important to note that the present findings do not imply that perfectionism is the only mediator between anxiety and stress symptoms and bipolar disorder symptoms. Perfectionism is only a partial mediator in the mediation models, leaving room for other constructs to play a role. For example, the perfectionism literature also highlights the importance of coping styles, emotional regulation strategies and congruent life events to explain the relationship between perfectionism and psychopathology [e.g. (Dunkley and Blankstein 2000; Dunkley et al. 2003)].

Our findings are consistent with current cognitive, transdiagnostic models of BD. Specifically, that self-critical, perfectionistic beliefs are related to both depressive and (hypo)manic symptoms (see Mansell et al. 2007; Mansell 2007) and that anxiety is likely a precursor to depressive symptoms in BD (Holmes et al. 2008). Consistent with Mansell et al.’s and Holmes’ model it is proposed that depressive and/or (hypo)manic symptoms arise from the use of unhelpful emotional regulation strategies (ascent and descent behaviours) in response to anxiety and stress symptoms driven by perfectionistic and/or goal attainment beliefs [see (Mansell 2007) for an illustrative case formulation], (Holmes et al. 2008; Mansell et al. 2007). A key component of Mansell’s model, the conflicted appraisals of an individual’s internal state, was not examined here, but would be crucial in further understanding the transition from stress and anxiety symptoms to depressive and (hypo)manic symptoms. The role that perfectionism may play in these appraisals also warrants further investigation. It may be that anxiety/stress and depression are not merely co-occurring emotional states (or comorbid disorders) but that there is a dynamic, iterative process that occurs within an individual that sometimes produces a predominantly anxious presentation (i.e. meeting criteria for an anxiety disorder) and at other times produces symptoms consistent with a formal mood episode. Indeed, (Mansell 2007) proposes that the manner in which mood fluctuations express themselves over time should determine the diagnosis and that the nature of current symptoms will be a complex result of: (1) experiencing previous vicious cycles; (2) the nature of the underlying beliefs; and (3) the current environment.

The presence of these significant relationships in the current sample, comprising both affected and unaffected relatives within affected families, suggests that these processes are operating outside of formal episodes and may represent intermediate phenotypes in those at genetic risk of developing either depressive or (hypo)manic episodes. Perfectionism has been proposed to have developmental origins (Flett et al. 2002), with maternal criticism when standards are not met (Bleys et al. 2016; Greblo and Bratko 2014) and parental psychological control found to be particularly important (Soenens et al. 2005). Our findings are also consistent with perfectionism being a transdiagnostic psychological risk factor for psychopathology generally (Bieling et al. 2004; Egan et al. 2011). In BD, previous research suggests that family socialization plays a role in shaping beliefs towards goal attainment and achievement that conveys a vulnerability to developing the cognitions that are characteristic of (hypo)mania (Chen and Johnson 2012). Given that the presence of perfectionistic cognitions and anxiety/stress symptoms may be necessary but not the whole picture of risk for experiencing BD symptoms, future research is necessary to better understand what additional risk factors are present that give rise to BD symptomatology against a background of familial risk.

A number of limitations in this study are important to acknowledge. Firstly, the cross-sectional nature of the study means that caution must be exercised when inferring causality. Ultimately longitudinal data will be required to more definitively investigate relationships between perfectionism, anxiety/stress, and mood symptom development over time. The cross-sectional nature of the study also makes it impossible to be certain about the directionality of the relationship between the variables of interest. However, our hypotheses are proposed on the basis of other studies that have used longitudinal designs and have attained similar results (O’Garro-Moore et al. 2015; Alloy et al. 2009b). Secondly, the present study examined anxiety (and stress) symptoms dimensionally rather than using formal categorical diagnostic criteria; to better understand issues of comorbidity with BD, it will be useful to examine these relationships among people with specific categorical anxiety diagnoses, such as those defined by DSM-5. Thirdly, anxiety has been found to influence the course of bipolar disorder in other ways. Higher likelihood of relapse (Otto et al. 2006) increased suicidal behaviour (Dilsaver et al. 2006; Young et al. 1993) and greater number of episodes (Bauer et al. 2005) have all been found to be increased in those with comorbid BD and anxiety disorders compared to those with BD only. While beyond the scope of the current study it would be useful to examine the role of perfectionism mediating the relationship between anxiety and other markers of BD severity. For example, people with BD who exhibit a cognitive style characterized by rigidity and pessimism are less likely to recover from their BD depressive episode or, if able to recover, take longer to do so (Stange et al. 2013). Additionally, we did not measure the occurrence of life events in the current study. The cognitive-diathesis stress model of psychopathology proposes that negative cognitive styles or schemas interact with life events to trigger symptoms of psychopathology (Beck 1967, 1976); specific life events associated with achievement may be particularly relevant to the development of mood episodes in BD (Alloy et al. 2005; Koenders et al. 2014). Lastly, the variables of interest were assessed using retrospective, self-report measures and it therefore it could be argued that responses are unduly influenced by the current mood state of the participants. Future research using experience-sampling methodology would enable the relationships between anxiety and stress symptoms, perfectionism and BD symptoms to be more clearly elucidated.

In sum, this study provides support for the role of perfectionism in the co-occurrence of anxiety, stress, depression and (hypo)manic symptoms in BD. There are promising treatments for clinical perfectionism (Egan et al. 2014) and also transdiagnostic approaches to targeting both anxiety and BD symptoms (Ellard et al. 2012) that may be useful in psychological treatment approaches. These findings further suggest that BD patients and/or those considered at risk of BD should be routinely screened for the presence of maladaptive perfectionism and considered for the need for psychological intervention to address anxiety, depressive and (hypo)manic symptoms, in addition to appropriate pharmacotherapy.