- Top of page
- Material and methods
- Declaration of interests
Bipolar I disorder is an episodic illness characterized by extreme shifts in mood, energy, and functioning. It is associated with high rates of suicide  and is the sixth leading cause of disability worldwide . Frequent episode recurrence remains common even with the best available medications , underscoring the need for more effective adjunctive psychosocial treatments. A better understanding of the psychological risk factors and concomitants of this illness is essential to developing such treatments and improving wellbeing.
Impulsivity is a feature of many psychiatric disorders and influences a number of important outcomes, including social adjustment , occupational functioning , and quality of life . There are several reasons to think that impulsivity is especially relevant to bipolar I disorder. First, impulsivity is one of the criteria for diagnosing a manic episode (i.e., ‘excessive involvement in pleasurable activities that have a high potential for painful consequences’ (, p. 362)) and gives rise to some of the most disruptive behaviors that can occur during manic episodes, such as unrestrained spending, sexual indiscretions, and embarking on risky financial ventures. Second, impulsivity is strongly related to manic symptom severity [8-10]. Finally, research suggests that elevated impulsivity predicts the onset of bipolar disorder [11, 12] and is associated with a more severe illness course [13, 14].
Several authors have highlighted that impulsivity is not a unitary construct, but subsumes multiple, statistically distinguishable facets [15, 16]. For example, the three higher-order subscales of the widely used Barrett Impulsiveness Scales assess the overlapping but separable facets of Attentional, Motor, and Non-planning impulsivity . Similarly, the Fun-seeking subscale of the Behavioral Activation Scale  is highly correlated with other measures of impulsivity, but has specificity for measuring trait-like tendencies to pursue novelty and reward with little regard for potentially painful consequences [19-22]. More recently, two factor analytically derived scales were developed to measure trait-like tendencies to act impulsively when experiencing strong emotion: the Positive Urgency  and Negative Urgency  scales assess tendencies to act impulsively when experiencing strong positive and strong negative emotions, respectively. These forms of impulsivity also have differential predictive validity for a range of important behavioral and psychiatric outcomes [23-26].
A considerable number of studies have assessed impulsivity in bipolar disorder during both symptomatic and euthymic phases of the illness [9, 10, 27, 28]. Taken together, these studies suggest that impulsivity is consistently elevated during mania. Findings during inter-episode periods are more mixed, however, leaving open the possibility that impulsivity is not relegated to symptomatic periods, but remains elevated during euthymia.
Bipolar disorder has been robustly linked to reward sensitivity [29-32] and to intense emotional experiences [33, 34], as well as difficulties regulating emotion [35, 36]. Early results from studies in analog samples suggest reward and strong emotion may also represent important preconditions for impulsivity in bipolar disorder. Specifically, elevated scores on the Behavioral Activation System (BAS) Fun-seeking scale have been reported among people at putative risk for bipolar disorder  and among those diagnosed with bipolar spectrum disorder [11, 31, 38]. People at high risk for mania also endorse experiencing heightened impulsivity when in the throes of strong emotion . Based on these preliminary findings, we hypothesized that individuals diagnosed with bipolar I disorder would show elevations in facets of impulsivity associated with reward and strong emotion, and that these forms of impulsivity would most strongly impair psychosocial functioning within this diagnostic group.
Developing a more refined understanding of the forms of impulsivity that remain stably elevated across symptomatic and euthymic phases of bipolar I disorder and are most strongly associated with psychosocial functioning represents an important first step toward developing targeted interventions aimed at preventing some of the most disruptive behaviors characterizing this illness.
Aims of the study
The present study had three aims: first, to compare specific facets of impulsivity hypothesized to be elevated among individuals diagnosed with bipolar I disorder compared with controls during the inter-episode period; second, to test whether psychiatric comorbidity explains any observed elevations in impulsivity within the bipolar group; and third, to assess which specific facets of impulsivity are most strongly associated with psychosocial functioning in the bipolar group.
- Top of page
- Material and methods
- Declaration of interests
Preliminary analyses indicated that all dependent variables were normally distributed (skewness and kurtosis estimates<∣2∣). As shown in Table 1, participants in the bipolar and control groups were well matched on age, gender, and years of education. Analyses also indicated that procedures for following participants with bipolar disorder until remission were effective: there were no differences between the bipolar and control groups on manic or depressive symptoms, and mean scores for both groups were well below the clinical cutoffs for mania and depression. As shown in Table 1, the bipolar I group reported a fairly severe illness history. Compared with controls, participants in the bipolar group were less likely to be employed and more likely to meet criteria for current anxiety, impulse control, and lifetime substance use disorder. They also had significantly lower GAF scores.
Table 1. Descriptive data by diagnostic group
|Demographic and clinical variables||Bipolar I (N = 91) Mean (SD)||Control (N = 80) Mean (SD)|
|Age||37.8 (11.6)||35.0 (12.1)|
|Years education||14.8 (2.0)||14.5 (2.1)|
|Percent with anxiety disorder**||52||26|
|Percent with past substance use disorder**||54||26|
|Percent with impulse control disorder**||36||9|
|Manic symptom level (BRMS)||2.7 (3.0)||1.9 (2.8)|
|Depressive symptom level (MHRSD)||3.3 (4.3)||1.9 (3.1)|
|Global assessment of functioning**||67.2 (12.0)||83.3 (10.4)|
|Age of first manic episode||22.0 (9.1)|| |
|Number of previous manic episodes||9.4 (10.2)|| |
|Previous hospitalizations for mania||1.7 (3.1)|| |
|Age of first depressive episode||18.2 (8.7)|| |
|Number of previous MDEs||12.0 (12.0)|| |
|Previous hospitalizations for MDE||1.2 (2.3)|| |
The median correlation among impulsivity measures was 0.46, suggesting that the scales cover separable aspects of impulsivity. Negative Urgency and BIS-11 Attentional Impulsiveness were correlated most strongly, r = 0.64, P < 0.001. Positive Urgency and BAS Fun-seeking were correlated most weakly, r = 0.12, ns. Gender was not significantly correlated with any of the impulsivity measures in the full sample: all rs < |0.15|, all ps > 0.06. Within the bipolar group, impulsivity scores were not associated with mood symptoms (BRMS, MHRSD scores) or medications (Somatotherapy Index scores): all rs < |0.17|, all ps > 0.11.
Which facets of impulsivity differentiate the bipolar group from the control group?
Because impulsivity measures were moderately intercorrelated, a multivariate analytic strategy was used. To examine group differences in the set of self-reported impulsivity measures, a one-way multivariate analysis of variance (manova) was conducted with group (bipolar vs. control) as the sole predictor . The bipolar and control groups differed significantly in impulsivity, Wilks' λ = 0.64, F(6,165) = 15.58, P < 0.001, partial n2 = 0.362.
To determine which specific facets of impulsivity contributed to the omnibus group difference, six parallel one-way anovas (Bonferroni corrected, P < 0.008) were conducted with diagnosis (bipolar, control) as the independent variable and each of the self-reported impulsivity variables as the dependent variable. As shown in Table 2, these one-way anovas yielded significant group differences for five of the six impulsivity variables (all except BAS Fun-seeking), with Positive Urgency yielding the largest effect size among the impulsivity variables. Effect sizes for the Positive Urgency and Negative Urgency scales were statistically indistinguishable, but only Positive Urgency yielded a significantly larger effect size than the remaining non emotion-related impulsivity measures.
Table 2. Univariate anova tests of group differences in impulsivity (df = 1, 170)
|Impulsivity scale||Bipolar mean (SD)||Control mean (SD)|| F || P ||Partial η2|
|Positive Urgency||33.74 (10.89)||21.00 (7.29)||78.69||<0.001||0.316|
|Negative Urgency||3.64 (0.65)||2.92 (0.69)||49.67||<0.001||0.226|
|BIS-11 Attentional||18.51 (4.25)||15.44 (4.07)||23.25||<0.001||0.120|
|BIS-11 Motor||20.61 (4.63)||17.49 (4.38)||20.46||<0.001||0.107|
|BIS-11 Non-planning||26.85 (5.05)||22.81 (5.02)||27.50||<0.001||0.139|
|BAS Fun-seeking||12.79 (2.49)||12.13 (2.36)||3.24||0.074||0.019|
Are the elevated impulsivity scores observed in bipolar disorder explained by psychiatric comorbidity?
To determine whether group differences in comorbid anxiety, impulse control, and lifetime substance use disorders drove the group differences observed for Positive Urgency, Negative Urgency, and BIS-11 Attentional, Motor, and Non-planning Impulsiveness, a parallel series of forward selection multiple regressions was conducted to predict each of the impulsivity measures, with dichotomous variables for anxiety disorders; impulse control disorders; and lifetime history of alcohol abuse, alcohol dependence, substance abuse, and substance dependence entered in block 1 and diagnostic group (bipolar vs. control) entered in block 2 as independent variables. As shown in Table 3, the mean differences observed between the bipolar and control groups on five of the impulsivity measures were not explained by any of the aforementioned forms of psychiatric comorbidity. Moreover, Positive Urgency was the only impulsivity variable that was not significantly related to comorbid anxiety, impulse control, or lifetime substance use disorder status in the full model.
Table 3. Forward selection regression analyses examining effects of bipolar diagnosis on impulsivity, controlling for comorbid conditions
| ||Positive Urgency||Negative Urgency||BIS-11 Attention||BIS-11 Motor||BIS-11 Non-planning|
| ΔR 2 || b || ΔR 2 || b || ΔR 2 || b || ΔR 2 || b || ΔR 2 || b |
|Block 1||0.15**|| ||0.30**|| ||0.19**|| ||0.14**|| ||0.26**|| |
|Anxiety disorder|| ||–|| ||0.24*|| ||0.18*|| ||–|| ||–|
|Impulse control disorder|| ||0.13|| ||0.20*|| ||0.22*|| ||0.19*|| ||0.26**|
|Lifetime alcohol abuse|| ||–|| ||–||–|| || ||–|| ||0.17*|
|Lifetime alcohol dependence|| ||–|| ||–||–|| || ||–|| ||–|
|Lifetime substance abuse|| ||–|| ||–||–|| || ||0.19*|| ||–|
|Lifetime substance dependence|| ||0.09|| ||0.14*|| ||0.09|| ||–|| ||0.18*|
|Block 2||0.20**|| ||0.08**|| ||0.04*|| ||0.04*|| ||0.03*|| |
|Bipolar diagnosis|| ||0.50**|| ||0.31**|| ||0.22*|| ||0.22*|| ||0.20*|
Which facet(s) of impulsivity are the strongest correlates of psychosocial functioning?
To determine whether study site or any demographic, clinical, or treatment variables were potential confounds for GAF, a forward selection multiple regression analysis was conducted using site, demographic variables (age, gender, years of education), mood symptoms (BRMS and MHRSD), and medication (dose equivalency levels) as independent variables, and GAF as the criterion variable. In the final model, only site remained significant (b = 12.67, t = 2.12, P = 0.04); thus, site was controlled for in the subsequent regression.
To examine whether the self-rated impulsivity scales predicted GAF after controlling for study site and comorbid diagnoses, a hierarchical linear regression analysis was conducted with site in block 1, comorbid diagnoses (anxiety disorders, impulse control disorders, and lifetime alcohol abuse, alcohol dependence, substance use, and substance dependence) in block 2, the six impulsivity measures in block 3, and GAF as the dependent variable.
Site accounted for 14.4% of the variance in GAF, ΔF(1,87) = 14.63, P < 0.001. After controlling for site, comorbid diagnoses accounted for an additional 3.9% of the variance in GAF, ΔF(6,81) = 0.64, P = 0.70. Finally, after controlling for site and comorbid diagnoses, Positive Urgency accounted for 24.2% of the variance in GAF scores, b = −0.45, t(78) = −4.60, P < 0.001. No other impulsivity variable accounted for significant variance in GAF. The full model was significant, F(6,75) = 5.27, P < 0.001, accounting for 42.5% of the variance in GAF scores.
- Top of page
- Material and methods
- Declaration of interests
This is the first study to conjointly examine multiple facets of impulsivity, including those related to key motivational and emotional correlates of bipolar disorder, in a sample of individuals diagnosed with the illness. Our results indicate that during inter-episode periods, individuals with bipolar I disorder report pronounced elevations on all of the impulsivity measures examined, except BAS Fun-seeking, and these elevations are not explained by comorbid anxiety, impulse control, or lifetime substance use disorder status.
The largest group difference (in terms of effect size) was observed for Positive Urgency, a facet of impulsivity that assesses the tendency to behave impulsively when experiencing strong positive emotions. This finding withstood control for comorbid anxiety disorders, impulse control disorders, and lifetime history of substance use disorders. In fact, Positive Urgency accounted for one-third of the variance in diagnosis of bipolar disorder and explained considerable variance in psychosocial functioning within the bipolar group.
Overall, these results suggest that strong positive emotions may represent an important precondition for impulsivity among individuals with bipolar I disorder. This finding is particularly notable when considered alongside a burgeoning literature suggesting that bipolar disorder is associated with context-insensitive elevations in positive emotion (cf. [34, 35]): the type of emotion that people with bipolar disorder are most susceptible to experiencing also confers the greatest vulnerability to their behaving in rash and ill-considered ways. This profile suggests two potential targets for therapeutic intervention in bipolar I disorder: i) developing effective emotion regulation strategies to maintain healthy levels of positive emotion and ii) implementing plans for preventing impulsive behavior when strong positive emotions do unfold.
At least one existing line of basic research shows early promise for an intervention that addresses these targets: setting implementation intentions in advance of strong emotional states . An implementation intention is a self-regulatory strategy in the form of a concrete if–then plan (i.e., ‘If situation X arises, then I will do Y’) that specifies when, where, and how the goal (in this case, keeping impulsive behavior at bay) will be achieved. Setting implementation intensions leads to a keener awareness of high-risk situations when they arise, enabling the chosen behavior to be performed more automatically. This strategy has been found to be effective both for regulating emotions [52, 53] and for overcoming the impact of emotions on various behaviors, including risk-taking [54-57].
The present study has several notable strengths. We recruited a large and well-characterized sample of people with bipolar I disorder during the inter-episode period and a demographically matched group of controls. Participants with bipolar disorder were followed longitudinally until they achieved symptom levels comparable to those of the control participants, reducing the likelihood that mood-state-dependent effects influenced the impulsivity findings. Finally, targeted sampling of the bipolar group allowed us to consider the potential influence of psychiatric comorbidity on impulsivity.
Several limitations are also apparent. First, the cross-sectional design of this study precludes us from being able to draw conclusions about the causal nature or underlying mechanisms of the observed relations among impulsivity, bipolar disorder, and psychosocial functioning. Second, this study relied on self-report measures of impulsivity, which are susceptible to response and self-presentational biases, as well as to shared method variance. Although the impulsivity measures we used have strong psychometric properties, and it has been argued that the degree of bias in self-ratings is relatively small , future studies would benefit from including behavioral measures of impulsivity in conjunction with experimental mood inductions or naturalistically occurring mood fluctuations to examine emotion-based impulsivity as it unfolds. Experience sampling methodology could also provide more direct information regarding the temporal dynamics of impulsive reactions to strong emotions in real-world contexts. Finally, the GAF scale yields a single score and thus only bluntly assesses psychosocial functioning. Future studies would benefit from using more refined indices that assess functioning in a range of life domains.
Although we took care to measure emotion-based impulsivity during the inter-episode period and explicitly instructed participants with bipolar disorder to reflect on periods of wellness when considering their impulsive tendencies, the current study does not help to decipher whether emotion-relevant impulsivity is more accurately conceptualized as a vulnerability factor or a ‘scar’ of bipolar disorder. Recent findings suggest that emotion-based impulsivity is elevated among individuals at putative risk for developing mania  and is related to a polymorphism marking variation in serotonergic function . Future research would thus benefit from integrating emotion-based measures of impulsivity with biological ones, including those related to serotonergic function.
In sum, our findings provide an important first step toward developing a more refined understanding of impulsivity in bipolar disorder and underscore the pivotal influence of strong positive emotion on impulsive behavior and functioning in this illness. Research corroborating these early self-report findings with multiple methods, including the kinds of laboratory-based and ecological measures outlined above, could lead to more fine-tuned interventions that help to stem self-destructive impulsive behaviors and improve the wellbeing of persons with bipolar disorder.
Declaration of interests
- Top of page
- Material and methods
- Declaration of interests
The statements below reflect the interests of all coauthors covering the past 2 years: Luma Muhtadie, Sheri L. Johnson, Charles S. Carver and Ian H. Gotlib have no interests to declare generally or in relation to this report. Terrence Ketter has a financial interest/arrangement or affiliation with one or more organizations that could be perceived as a real or apparent conflict of interest.
Grant/Research Support: Agency for Healthcare Research and Quality, AstraZeneca Pharmaceuticals LP, Cephalon Inc.Eli Lilly and Company, Pfizer Inc., Sunovion Pharmaceuticals.
Consultant: Allergan, Inc., Avanir Pharmaceuticals, Forest Pharmaceuticals, Janssen Pharmaceuticals, Sunovion Pharmaceuticals, Teva Pharmaceuticals.
Lecture Honoraria: Abbott Laboratories, Inc, GlaxoSmithKline, Otsuka Pharmaceuticals.
Royalties: American Psychiatric Publishing, Inc.