The authors of this paper do not have any commercial associations that might pose a conflict of interest in connection with this manuscript.
Professor Nick Craddock, Department of Psychological Medicine, Henry Wellcome Building for Biomedical Research, Wales School of Medicine, Cardiff University, Heath Park, Cardiff, CF14 4XN, UK. Fax: +44 (0)2920 687068; e-mail: email@example.com
Objectives: Studies have suggested that episode polarity at illness onset in bipolar disorder may be predictive of some aspects of lifetime clinical characteristics. We here examine this possibility in a large, well-characterized sample of patients with bipolar I disorder.
Methods: We assessed polarity at onset in patients with bipolar I disorder (N = 553) recruited as part of our ongoing studies of affective disorders. Lifetime clinical characteristics of illness were compared in patients who had a depressive episode at first illness onset (n = 343) and patients who had a manic episode at first illness onset (n = 210).
Results: Several lifetime clinical features differed between patients according to the polarity of their onset episode of illness. A logistic regression analysis showed that the lifetime clinical features significantly associated with a depressive episode at illness onset in our sample were: an earlier age at illness onset; a predominantly depressive polarity during the lifetime; more frequent and more severe depressive episodes; and less prominent lifetime psychotic features.
Conclusions: Knowledge of pole of onset may help the clinician in providing prognostic information and management advice to an individual with bipolar disorder.
Bipolar disorders are heterogeneous, and the identification of more homogeneous subgroups of patients has the potential to facilitate more targeted clinical advice and interventions, as well as being useful in studies investigating the aetiology of affective disorders. One clinical feature that may identify such subgroups is the type of episode, depressive or manic, that occurs first in the bipolar illness. This ‘polarity at onset’ has been shown to be familial (1) and to distinguish groups of bipolar individuals who differ in lifetime clinical features of illness (1–4).
The aim of the present study was to examine lifetime clinical course characteristics in a large sample of patients with bipolar disorder according to the episode polarity at illness onset. Unlike previous studies, we have also examined measures of episode severity and frequency using the Bipolar Affective Disorders Dimension Scales (BADDS) (5). These scales were developed by our group as an adjunct to conventional categorical diagnosis in order to provide a richer characterization of the individual’s lifetime experience of illness (6, 7), and we have found it useful in our research (8–11). We have examined a large, well-characterized sample of unrelated patients with narrowly defined bipolar I disorder recruited within the UK.
Based on the findings of previous studies (1–4), we hypothesised that patients whose onset episode of illness was depression would have a predominantly depressive course of illness and that there may be differences between mania-onset and depression-onset patients on clinical characteristics such as age at onset, predominant pole of illness, rapid cycling, and lifetime experience of psychosis.
Patients and methods
Subjects were recruited as part of our ongoing studies of affective disorders using both systematic and nonsystematic recruitment methods. Systematic recruitment involved the identification of potentially suitable participants through the systematic screening of community mental health teams. Subjects were recruited nonsystematically via advertisements in local general practitioner surgeries and the local media and through patient support organisations. Individuals were included if they fulfilled criteria for bipolar I disorder (BPI) according to DSM-IV (12) and ICD-10 (13) and had experienced at least one episode of major depression. All participants were aged 18 years or over. As participants were recruited as part of our ongoing molecular genetic studies, they were required to be of UK/Eire white ethnicity. When studying the genetic risk factors involved in complex diseases, restricting ascertainment to a single ethnic/racial group is important in order to reduce heterogeneity and thereby minimize the likelihood of false positives caused by differences in the genetic background of participants.
Voluntary written informed consent was obtained from all participants who were interviewed using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) (14). Psychiatric/general practice case notes were also reviewed. All available information was collated into a structured, written vignette, and best-estimate lifetime diagnoses (according to DSM-IV and ICD-10) were made by at least two members of the research team blind to each other’s ratings.
Based on the information contained in the vignettes, patients were rated for key clinical variables (such as age at onset and number of episode of illness). They were also rated on the Global Assessment Scale (GAS) (15) for the worst ever episode of depression and the worst ever episode of mania, and on the BADDS. The GAS is a rating scale for measuring the overall functioning of a patient during a specified time frame on a continuum from psychological or psychiatric illness to health. The scale values range from ‘1’, which represents the hypothetically most unwell individual, to ‘100’, the hypothetically healthiest. The BADDS comprises four dimensions that provide a quantitative measure (on a 0–100 scale) of lifetime experience of psychopathology in each of four domains (mania, depression, psychosis, and congruence of psychosis). The mania dimension is a measure of the frequency and severity of manic-like episodes, and the depression dimension a measure of the severity and frequency of depressive-like episodes. Higher ratings on the scale indicate higher frequency and severity. The psychosis dimension is a measure of the prominence of lifetime psychotic features. The scale takes into account both the number and duration of episodes with and without psychotic features. Higher ratings indicate an increased prominence of psychosis as a lifetime feature of the illness. The fourth dimension is an estimate of how congruent psychotic features have been with mood state during episodes, with 0 representing complete congruence (i.e., psychotic symptoms are completely congruent with affective state and only occur during affective episodes) and 100 representing complete incongruence (i.e., psychotic symptoms predominate the illness and occur chronically outside, or in absence of, affective episodes).
Each participant was diagnosed and had key clinical variables rated by at least two members of the research team blind to each other’s ratings. All interviews and consensus ratings were conducted by trained psychiatrists or psychologists. Inter-rater reliability was formally assessed using joint ratings of 20 cases with a range of mood disorder diagnoses. Mean overall kappa statistics of 0.85 and 0.83 were obtained for DSM-IV and ICD-10 diagnoses, respectively. Mean kappa statistics and intraclass correlation coefficients (ICCs) for other key clinical variables ranged from 0.81–0.99 and 0.85–0.97, respectively.
Polarity of the onset episode (first ever episode of mood illness) was determined by comparing the reported age at onset for the first ever major depressive episode and the first ever manic episode (episodes displaying predominantly manic features, including mixed episodes and hypomanic episodes). Where patients reported the same age at onset for both episode poles, vignettes were examined to establish which pole occurred first. Only patients for whom the polarity of the onset episode could clearly be established were included in our analyses (N = 553). Of these patients, 210 (38%) were included in the mania-onset group (predominant polarity coded as ‘1’) and 343 (62%) were included in the depression-onset group (predominant polarity coded as ‘0’).
The demographic and lifetime clinical characteristics of the two groups were compared using chi-square tests for categorical data and the Mann-Whitney U-test for nonparametric continuous data.
Binary logistic regression using forward stepwise likelihood ratio for variable selection was performed to assess which variables were most associated with the polarity of the onset episode of illness. Odds ratios (ORs) of >1 indicate that a higher score is associated with greater likelihood of a manic episode at illness onset, whereas ORs of <1 indicate that a higher score is associated with a greater likelihood of a depressive episode at illness onset.
There was no statistically significant difference between the two groups for age at interview or method of recruitment. There were significantly fewer males (28.3%) in the depression-onset group compared to the mania-onset group (37.6%) (p = 0.022). The mean illness duration for the depression-onset group (22 years) was significantly longer than for the mania-onset group (15 years) (p < 0.001).
Family history of affective disorder was not significantly different between the two groups.
Table 1 shows the clinical characteristics that differed significantly between the two groups. Patients in the depression-onset group experienced significantly more illness episodes (p = 0.031), more depressive episodes (p < 0.001), and fewer manic episodes (p = 0.004) when compared to the mania-onset group (see Fig. 1). The lifetime predominant polarity of episodes of illness experienced by the depression-onset group was depression (57.7% of episodes), whereas for the mania-onset group the lifetime predominant polarity was mania (57.1% of episodes). Patients in the depression-onset group had a significantly longer duration of the longest ever depressive episode (p > 0.001) compared to patients in the mania-onset group, whereas there was no significant difference in the mean duration of the longest ever manic episode between the two groups.
Table 1. Lifetime clinical characteristics of bipolar I patients (N = 553) according to the polarity of the onset (first ever) episode of illness
Age at first episode of mood illness (mania or depression) (years)
Age at first episode of depression (years)
Age at first episode of mania (years)
No suicide attempt
Rapid cycling (4 or more episodes in a year)
Global Assessment Scale
As can be seen in Table 1, the mean age at illness onset (age at first mood episode) and age at onset of depression for patients in the depression-onset group was significantly lower (both p < 0.001), and the age at onset of mania higher (p < 0.001), compared to the mania-onset group. There were no significant differences between the two groups in the age at first contact with psychiatric services or in the number of psychiatric admissions. There was significantly more suicidal behaviour in the depression-onset group (p < 0.001) and more rapid cycling (four or more episodes in a year) (p = 0.017) compared to the mania-onset group.
Patients in the depression-onset group were significantly less functionally impaired according to the GAS during their worst ever manic episode (p < 0.001), but more impaired during their worst ever depressive episode (p < 0.001), than patients in the mania-onset group. Similarly, patients in the depression-onset group scored significantly higher on the BADDS depression dimension (p < 0.001), but lower on the BADDS mania dimension (p < 0.001), reflecting the increased occurrence and severity of depressive episodes, and the decreased occurrence and severity of manic episodes in the depression-onset group, compared to the mania-onset group (see Table 1).
When comparing the presence of psychotic features in the two groups as a categorical variable (present or absent on a lifetime ever basis), there was no statistically significant difference. However, significantly lower scores on the BADDS psychosis dimension for the depression-onset group (p < 0.001) reflect a lower prominence of psychotic features in episodes of illness during the lifetime of patients in the depression-onset group compared to patients in the mania-onset group.
Logistic regression was carried out to determine which combination of lifetime clinical features was best associated with the polarity of the onset episode of illness. Variables that were significant at the 0.05% level in the univariate comparisons were entered into the regression, including sex and illness duration. The significant variables in the best fitting model were: predominant polarity [OR = 4.4, 95% confidence interval (CI): 2.3–8.4], age at onset of illness (OR = 1.06, 95% CI: 1.03–1.09), BADDS depression (OR = 0.96, 95% CI: 0.94–0.98), and BADDS psychosis (OR = 1.026, 95% CI: 1.01–1.04).
We found supporting evidence for significant differences in the clinical features of patients with bipolar I disorder according to the polarity of their onset episode of illness. Consistent with our hypothesis, several measures indicating a more ‘depressive’ lifetime course were significantly more common in those whose illness started with a depressive episode. Thus, in agreement with previous studies, we found that a depressive onset was associated with a greater number of depressive episodes (2–4), an increased risk of suicide attempt during the lifetime (2–4), and more rapid cycling (four or more episodes in a year) (3) compared to patients with an onset episode of mania.
Previous studies examining clinical characteristics in relation to polarity at onset have not included measures of functional impairment in episodes of depression or mania. In addition to assessing the frequency of depressive episodes, as has been done in previous studies, our use of the BADDS and the GAS also enabled us to take into account the severity of episodes. We found more severe impairment in functioning during the worst ever episode of depression in patients whose onset episode was depression, compared to those whose onset episode was mania. Similarly, we found that patients whose onset episode was mania experienced more severe impairment in functioning during their worst ever manic episode compared to patients with an onset episode of depression.
Some previous studies have suggested that there is no difference in the presence of psychotic features according to the polarity of the onset episode of illness (1, 2), although others have found that patients with a depressive episode at illness onset are less likely to experience psychosis during their life course (3, 4). When we compared the definite presence or absence of lifetime psychotic features as a binary variable in our sample, we found no difference between those with depressive and those with manic poles of onset. This lack of difference is consistent with two of four previous studies (1, 2). However, using the BADDS psychosis scale, we have found that the lifetime prominence of psychotic features during mood episodes is significantly greater in patients who experienced a manic episode at illness onset compared to those who experienced a depressive episode at onset. This is consistent with the other two of four previous studies (3, 4). Thus, although there is no difference between the mania-onset and depression-onset groups in terms of presence or absence of lifetime psychotic features, there is a difference in the predominance of lifetime psychotic features between the two groups. This illustrates the importance of a more detailed consideration of clinical features beyond a simple ‘presence/ absence’ dichotomy.
Our finding of a depressive onset being more common in females contrasts with Kassem et al. (1), but is in agreement with Perlis et al. (2). Our finding of depressive onset being associated with an earlier age at onset is consistent with two previous studies (1, 2).
Our findings are of potential clinical importance. Once a patient has experienced his or her first manic episode, the clinician can use knowledge of polarity of the illness onset as an indicator of the likely predominant pole of illness. This may be helpful in providing information and advice to the patient. Our observation here applies to the situation in which a patient has experienced his or her first manic episode and a diagnosis of bipolar disorder is made. We note that studies have suggested that there are differences in the clinical characteristics of depressive episodes in patients with unipolar and bipolar disorder (16–18). Such differences could potentially be of use in predicting the risk of bipolar disorder in patients who experience a depressive episode without having experienced a prior manic episode.
Strengths of our study include a large, well-characterized sample and the use of measures of episode frequency, severity, and impairment. Nonetheless, a number of limitations should be considered when interpreting our findings. First, retrospective assessment measures were used to obtain clinical course information and to establish the polarity of the onset episode of illness. However, we found good agreement between the information obtained from participants during the in-depth semistructured interviews and the information obtained from the case note reviews. Another limitation relating to the retrospective nature of this study was that we were unable to assess the effects of medication on illness course. We further note that it is possible that there may have been differences in treatment regimes between the two groups that could potentially impact on illness course/severity.
We focused our analyses on a narrowly defined sample of patients with bipolar I disorder; hence, our findings relate to this group of bipolar patients. It will be important for future studies to examine polarity at onset in patients also with bipolar II disorder in order to establish the generalizability of our findings across the bipolar spectrum.
Our study focused on the first clinically significant mood episodes. Some individuals in the mania-onset group may have experienced subclinical depressive symptoms prior to the first clinically significant episode. It is possible that these patients may represent a clinically significant subgroup; however, assessing this was beyond the scope of our study.
In summary, we found, in agreement with previous studies, that the polarity of the first episode of illness in bipolar disorder is depressive for at least two thirds of patients (3, 4, 19). We have also found that patients with a depressive pole of onset tend to have a course of illness characterised by more frequent depressive episodes and more lifetime depressive morbidity. The increase in severity of depressive episodes and the associated impairment in functioning in patients with an onset episode of depression illustrate a need for treatment regimes targeted at preventing and reducing the burden associated with depressive episodes in patients for whom depressive episodes are associated with greater morbidity. Similarly, our findings regarding increased severity and associated impairment during manic episodes in patients with a manic episode at onset point to the need for improved prevention and treatment regimes in order to reduce the burden of manic episodes in patients for whom these episodes result in the greatest impairment. The division of patients with bipolar I disorder according to polarity at onset and predominant polarity during illness course (i.e., ‘depression prone’ and ‘manic prone’ subtypes) may not only have implications in clinical practice, but may also provide more homogenous subgroups of patients for the purpose of research.
We would like to express our gratitude to all of the participants who so generously gave their time to participate in the study. We would also like to thank Dr. George Pelios for database programming. We are grateful to the Wellcome Trust for financial support.