Impulsivity can be conceptualised as a predisposition towards rapid, unplanned actions without regard to consequences [Reference Moeller, Barratt, Dougherty, Schmitz and Swann20]. As impulsive behaviour is one of the diagnostic criteria for mania  the concept of impulsivity is particularly relevant to bipolar disorder (BD). It is unclear, however, whether increased impulsivity occurs in depression as well and whether it persists as a trait characteristic during periods of remission.
Studies of patients with unipolar or bipolar depression suggest that increased impulsivity is strongly related to symptom severity [Reference Corruble, Benyamina, Franck, Falisssard and Hardy5,Reference Liebowitz, Stallone, Dunner and Fieve17]. In addition, tryptophan depletion in unaffected first degree relatives of BD patients has been associated both with increased impulsivity but also with lowering of mood [Reference Quintin, Benkelfat, Launay, Arnulf, Pointereau-Bellenger and Barbault26].
Two recent studies reported that BD patients judged “euthymic” scored similarly to manic patients on the Barratt Impulsiveness Scale (BIS-11), suggesting that impulsivity is a trait feature of this disorder [Reference Swann, Bjork, Moeller and Dougherty30,Reference Young, Biggs, Zieglar and Meyer31]. However, patients were described as having “a wide range of depressive symptoms” which raises doubt about their “euthymic” status.
The present study had two main objectives:
• to revisit the question of impulsivity as a trait characteristic of BD. To examine this we compared BIS-11 scores between healthy volunteers and BD patients with variable levels of psychopathology, ranging from remitted to syndromal. Furthermore, we explored the relationship of impulsivity to personality dimensions, which are considered trait features, using the Eysenck Personality Questionnaire (EPQ). Psychoticism incorporates the idea of poor impulse control and we hypothesised that, if impulsivity is indeed a trait feature of BD, then both the Psychoticism and BIS-11 scores should be higher in remitted patients;
• to examine the effect of manic and depressive symptoms on the level of impulsivity in BD patients. Our initial hypothesis was that impulsivity would be increased in patients regardless of symptom polarity.
Patients with bipolar disorder type 1 (BD1) or II (BDII), aged between 18 and 70 years, were recruited from secondary care services by advertisement. The diagnosis of BD was established using the Structured Clinical Interview for Axis 1 DSM-IV disorders [Reference First, Spitzer, Gibbon and Williams11]. Patients with current substance dependence or concomitant CNS disorders were excluded. Comparison subjects were also recruited through advertisement and were considered for inclusion if they had no personal history of physical or psychiatric disorders and had never received treatment for psychological problems. The study was approved by our local ethics committee. Written informed consent was obtained from all participants after complete description of the protocol.
Demographic variables assessed were age, gender and education. Participants completed the EPQ [Reference Eysenck and Eysenck7]. This 90-item questionnaire assesses the orthogonal global personality traits of Psychoticism (scored 0 to 25), Extraversion (scored 0 to 21) and Neuroticism/Stability (score 0 to 23). In the EPQ, unlike its precursors the Maudsley Personality Inventory [Reference Eysenck6] and the Eysenck Personality Inventory, items relating to impulsivity are included in Psychoticism and not Extraversion, which mainly focuses on sociability [Reference Eysenck and Nyborg9]. Apart from impulsivity, Psychoticism also incorporates non-conformity and emotional coldness while the Neuroticism/Stability dimension measures levels of emotional stability. Those with high scores are seen as generally nervous and prone to sadness, anxiety and emotional fluctuations. The EPQ also includes a “Lies” dimension (scored 0 to 21), which reflects a tendency to dissimulation (“faking good”).
Impulsivity was assessed with the BIS-11 based on the principal-component analysis of the scale by Patton et al. [Reference Patton, Standford and Barratt23]. BIS-11 is a 30-item questionnaire of statements about one's self rated on a scale of 1 (rarely/never) to 4 (almost always/always). It consists of Attentional (rapid shifts in attention/impatience with complexity), Motor (acting impetuously) and Non-Planning (absence of weighing up long-term consequences of actions) subscales.
In all participants, symptomatology was rated on the Montgomery-Asberg Depression Rating Scale (MADRS) [Reference Montgomery and Asberg21] and the Young Mania Rating Scale (YMRS) [Reference Young, Biggs, Zieglar and Meyer32]. BD patients were further assessed on:
• clinical history, including length of illness, number and type of episodes in the preceding year and current medication;
• the Clinical Global Impressions Scale (CGI) [Reference Guy, Rush, Pincus and First12].
BD patients were categorised in three groups based on CGI scores to reflect current levels of illness (syndromal, subsyndromal and remitted). The “syndromal” group comprised patients with a CGI score of less than or equal to 3 (“mildly unwell” and above). The “subsyndromal” group included patients with a CGI score of 2 and YMRS and MADRS scores of less than 13. Remission was defined as a score of 1 on the CGI and a score of less than or equal to 9 on the MADRS [Reference Zimmerman, Chelminski and Posternak33] and less than or equal to 6 on the YMRS [Reference Martinez-Aran, Vieta, Colom, Reinares, Benabarre and Torrent19].
2.3. Statistical analysis
Comparisons of personality and impulsivity measures between groups (volunteers, remitted, subsyndromal and syndromal patients) were performed using analysis of covariance with total MADRS and YMRS scores are covariates. Significant results were followed-up with Bonferroni post-hoc comparisons. Pearson's or Spearman's correlations were used to examine relationships between variables; the level of statistical significance was set at p ≤ 0.01 to compensate for multiple testing.
Predictors of impulsivity in the BD sample were investigated using hierarchical stepwise multiple regression analyses with the BIS-11 subscales as dependent variables. Each analysis included four sub-sets of variables (blocks) decided a priori as follows: block 1 = demographic details (age, gender, education); block 2 = clinical history (length of illness, number of admissions); block 3 = MADRS and YMRS total scores; block 4 = EPQ Neuroticism, Extraversion and Psychoticism scores. Threshold for entry into the model was set at p ≤ 0.01. Symptom scores were entered before personality measures, as even low levels of depression are known to affect personality ratings [Reference Coppen and Metcalf4,Reference Hirschfeld, Klerman, Keller, Anderson and Clayton14,Reference Lejoyeux, Arbaretaz, McLoughlin and Ades16,Reference Martinez-Aran, Vieta, Colom, Reinares, Benabarre and Torrent18]. Diagnostic tests on the final models showed the residuals to be normally distributed and homoscedastic.
A hundred and six BD patients (78 women, 28 men) and 30 healthy volunteers participated in this study. Only 10 patients had BDII and they were equally distributed across the three patient subgroups. The mean age ± S.D. of the patients and controls were 50.0 ± 10.0 and 48.6 ± 10.5 years respectively. Educational levels were high, with 42.5% of patients and 46.7% of controls having a college or other higher degree. The clinical characteristics of the BD sample are presented in Table 1.
There were no significant differences between the total sample of BD patients and controls on age (t134 = 0.68; p = 0.49), gender (χ2 = 0.15; p = 0.43) or education status (χ2 = 0.17; p = 0.42). Similarly, none of the three BD subgroups differed from each other or from the controls on any of these demographic variables.
We avoided categorising episodes into depressive, manic or mixed since most patients present with a mixture of symptoms [Reference Cassidy, Forest, Murray and Carroll3] and opted for a dimensional approach instead.
3.2. EPQ personality and BIS-11 impulsivity scores
The mean scores of the remitted, subsyndromal, syndromal and comparison groups on the EPQ and BIS-11 are given in Table 2. There were no significant differences in dissimulation (Lie) or Extraversion scores between controls and any BD sub-group. Remitted patients did not differ from controls on any EPQ dimension while Psychoticism and Neuroticism scores increased incrementally the higher the levels of current symptomatology.
There were no differences in Attentional, Motor, Non-Planning Impulsivity or BIS-11 total scores between remitted patients and controls but we observed a significant incremental increase across all subscale scores between remitted and symptomatic BD subgroups.
3.3. Correlates of impulsivity in BD
In the total BD sample (n = 106), the BIS-11 total score correlated positively with Psychoticism (Spearman's rho = 0.40; p = <0.001) and Neuroticism (Spearman's rho = 0.50; p < 0.001) but not Extraversion (Spearman's rho = –0.15; p = 0.13). Of the demographic variables, only education appeared relevant being negatively correlated with the Attentional and Non-Planning, but not the Motor subscale scores. Of the clinical variables only current manic and depressive symptom scores showed (positive) correlations with impulsivity.
3.4. Predictors of impulsivity in BD patients
The results of the regression analyses are shown in Table 3. Current depressive symptom ratings were the strongest predictor of impulsivity in the BD sample. MADRS scores explained around a quarter of the variance in total BIS-11 scores. Psychoticism added between 7 and 10% to the variance explained in the models predicting Motor, Non-Planning and BIS-11 total scores but contributed little to Attentional Impulsivity. Educational attainment, by contrast, contributed just over 10% to the amount of variance explained for Attentional, Non-Planning and total BIS-11 scores, but had no impact on Motor Impulsivity. Neuroticism explained 6% of the variance in Attentional Impulsivity and 4% in the BIS-11 total score.
4.1. Impulsivity as a trait feature of BD
We found no differences between remitted BD patients and healthy volunteers in any subscale or in the total BIS-11 score. Both subsyndromal and syndromal BD groups had higher impulsivity scores on all BIS-11 subscales compared to remitted patients and healthy volunteers. This contrasts with studies by Swann et al. [Reference Swann, Bjork, Moeller and Dougherty30,Reference Young, Biggs, Zieglar and Meyer31] where the BIS-11 scores of “euthymic” patients were found to be similar to those of manic inpatients. The mean subscale and total BIS-11 scores of our controls and that of Swan et al. [Reference Swann, Bjork, Moeller and Dougherty31] are nearly identical (mean ± S.D., BIS-11 total score in Swann et al. = 59.9 ± 9.3; this study = 60.8 ± 10.0). It is therefore unlikely that our contrasting results are due to differences in the administration of the scale but are probably attributable to the clinical features of the patient samples. Our criteria for remission were a CGI score of 1 and MADRS and YMRS scores that were equal or below the mean scores of these scales in studies of healthy controls [Reference Zimmerman, Chelminski and Posternak33]. Swann et al. [Reference Swann, Bjork, Moeller and Dougherty31] defined “euthymia” as the absence of an acute mood episode meeting DSM-IV criteria in the preceding 6 months, a definition that would allow the inclusion of patients with subsyndromal symptoms. Indeed the authors report “a wide range of depressive symptoms” in their sample but give no details. The total BIS-11 score of the “euthymic” patients in the study of Swan et al. was higher (77.1 ± 13.8) than that obtained in our syndromal group (73.0 ± 12). This is perhaps a reflection of the different sampling frame of the two studies; patients in the Swan et al. study came from specialist clinics that may attract more complex and co-morbid cases than those found in secondary care settings such as ours. Peluso et al. [Reference Peluso, Hatch, Glahn, Monkul, Sanches and Najt24] also reported increased BIS-11 scores in “euthymic” BD patients compares to healthy controls; in their study the mean total score for healthy controls was 56.1 ± 8.2 which is again remarkably similar to the score for healthy participants in the present study. Their “euthymic” group however has a mean Hamilton Depression Rating Scale of 8.8, which is above the conventional cut-off for euthymia.
Since even residual mood symptoms are associated with higher BIS-11 scores we suggest that in studies of impulsivity in BD wide definitions of remission should be avoided. Our results also suggest that in BD impulsivity scores are higher in depression, as well as mania, a finding consistent with that of Corruble et al. [Reference Corruble, Benyamina, Franck, Falisssard and Hardy5] who also found high impulsivity scores in depression. Impulsivity ratings appeared to reduce as depressive symptoms remitted with successful treatment [Reference Corruble, Benyamina, Franck, Falisssard and Hardy5].
4.2. EPQ measures and clinical phases in BD
We found no differences between remitted BD patients and controls in any personality dimension. Apart from Extraversion, which showed no association with clinical status, both Psychoticism and Neuroticism scores were increased in subsyndromal and syndromal patients.
Our results regarding Extraversion are consistent with other studies which also reported no significant differences between remitted BD patients and controls [Reference Bech, Shapiro, Sihm, Nielsen, Sorensen and Rafaelsen2,Reference Hirschfeld, Klerman, Clayton, Keller, McDonald-Scott and Larkin13,Reference Martinez-Aran, Vieta, Colom, Reinares, Benabarre and Torrent18,Reference Quintin, Benkelfat, Launay, Arnulf, Pointereau-Bellenger and Barbault25,Reference Spitzer, Endicott and Robins28] and the absence of an effect of symptom levels also supports earlier views that Extraversion may not be significantly affected by mood [Reference Liebowitz, Stallone, Dunner and Fieve18].
Neuroticism scores were increased in subsyndromal and syndromal but in remitted patients. In contrast, two studies from the National Institute of Mental Health (NIMH) collaborative programme on the psychobiology of depression reported increased scores in the Neuroticism scale of the Maudsley Personality Inventory (MPI) between remitted BD patients and controls [Reference Kerr, Schapira, Roth and Garside15,Reference Spitzer, Endicott and Robins28]. In both studies remission was defined by the schedule for affective disorders and schizophrenia [Reference Spitzer, Endicott and Robins29] which allows for the presence of symptoms below the cut-off of a full episode. Therefore it is likely that subsyndromal patients were included and this may explain the higher levels of neuroticism observed. Since symptoms at the time of the personality assessments were not quantified in either study, it is not possible to know. Finally, this is the first study to report on psychoticism in BD. Psychoticism scores were generally low and although they tended to increase with higher levels of symptomatology this did not reach statistical significance.
4.3. Impulsivity, personality and symptoms
Current symptoms were by far the best predictors of Attentional, Motor and Non-Planning Impulsivity in our BD sample. MADRS scores explained about 17 to 26% of the variance. Manic symptoms did not make an independent contribution over and above that of depressive symptoms. Scores however were low (mean of 5.9 ± 7.6) and co-varied with depressive symptoms. Of the EPQ dimensions, Neuroticism and Psychoticism explained between 4 to 6% and 7 to 10% of the variance respectively.
Psychoticism predicted impulsivity scores across all the subscales making its largest contribution to motor and BIS-11 total scores indicating that impulsivity, as argued by Eysenck, is a component of Psychoticism [Reference Eysenck8].
4.4. Methodological issues
Although the EPQ and the BIS-11 are widely used in psychiatric research their application in BD may require special consideration.
Close examination of the EPQ items suggests the possibility that the experience of repeated and chronic mood symptoms may influence the way BD patients respond to some of the questions. We highlight two such questions here to illustrate our point. In item 84 of the EPQ subjects are asked: “Are you sometimes bubbling over with energy and sometimes very sluggish?”. Item 68 asks: “Have you ever wished you were dead?”. BD patients may have difficulties separating aspects of their behaviour can be directly attributed to their illness or are part of their “normal” personality.
The BIS-11 Attentional Impulsivity scale was designed to measure rapid shifts in attention and impatience with complexity, a pattern commonly seen in manic patients [Reference Patton, Standford and Barratt22,Reference Solomon, Tacie Shea, Leon, Mueller, Coryell and Maser27]. In contrast, depressed patients have problems shifting attentional focus [Reference Patton, Standford and Barratt22,Reference Solomon, Tacie Shea, Leon, Mueller, Coryell and Maser27]. The observed correlation of Attentional Impulsivity scores with depressive symptoms suggests that this subscale of the BIS-11 detects general attentional dysfunction without being informative about its nature.
An obvious drawback of the BIS-11 data is that they are based on self-report. It is therefore important to examine the relationship of this scale with objective measures obtained by formal cognitive testing and we plan to focus on this in future studies. Finally, the study sample comprised primarily of female BD patients; although male and female patients in our study did not differ in any measure of clinical status, personality trait or impulsivity scores the inclusion of male patients in further studies will be helpful in ensuring the generalisability of our findings.
In summary we found that impulsivity may not be a trait feature of BD but may fluctuate depending on symptoms. Remitted BD patients did not show abnormalities in any personality dimension measured by the EPQ whereas Neuroticism and Psychoticism scores were increased in syndromal and sub-syndromal patients. Therefore the clinical status of patients needs to be clearly and rigorously defined in all future studies in this field. We are currently engaged in a longitudinal follow-up of these patients that will allow us to describe better the relationship between impulsivity, personality and changes in symptom severity or polarity.