Eduard Vieta, Bipolar Disorders Program, Clinical Institute of Neuroscience, Hospital Clinic, Villarroel170, 08036-Barcelona, Spain. E-mail: email@example.com
Grande I, Goikolea JM, de Dios C, González-Pinto A, Montes JM, Saiz-Ruiz J, Prieto E, Vieta E, for the PREBIS group. Occupational disability in bipolar disorder: analysis of predictors of being on severe disablement benefit PREBIS study data).
Objective: Patients diagnosed with bipolar disorder (BD) are reported to have significant work impairment during interepisode intervals. This study was carried out to assess potential predictors of occupational disability in a longitudinal follow-up of euthymic patients.
Method: We included 327 euthymic patients diagnosed with BD type I or type II, 226 of whom were employed and 101 were receiving a severe disablement benefit (SDB). Sociodemographic data were studied and episode recurrence was assessed along a 1-year follow-up. Logistic regression analysis was applied to determine predictors of receiving SDB. Cox regression was built to study recurrences.
Results: Predictors of receiving SDB were: axis II comorbidity [Odds Ratio (OR) = 2.94, CI: 1.26–6.86, P = 0.013], number of manic episodes (OR = 1.21, CI: 1.10–1.34, P < 0.001), being without stable partner (OR = 2.44, CI: 1.34–4.44, P = 0.004) and older age (OR = 1.08, CI: 1.05–1.12, P < 0.001). Bipolar patients receiving SDB presented more episodic recurrences regardless of polarity than employed bipolar patients (P = 0.002). The time until recurrence in 25% of the bipolar patients receiving SDB was 6.08 months (CI: 4.44–11.77) being 13.08 months (CI: 9.60 to –) in the employed group.
Conclusion: Occupational disability in bipolar patients is associated with axis II comorbidity, more previous manic episodes, not having a stable relationship, older age, and more recurrences at 1-year follow-up.
• Axis II comorbidity, more previous manic episodes, not having a stable relationship, and older age are associated with occupational disability in bipolar patients.
• Bipolar patients receiving a severe disablement benefit present more episodic recurrences than employed bipolar patients in a 1-year follow-up.
• The impact of pharmacological treatments in occupational status was not studied.
• Our findings in a Spanish sample may not be generalizable to other locations.
The weight of evidence suggests that patients diagnosed with bipolar disorder (BD) have functional impairment not only in acute phases of the disorder but also during interepisode intervals (1–3). Previous literature describes difficulties in multiple areas of functioning such as cognition, social relationships, finances, and occupation (4–6). Several variables have been reported to negatively influence functioning such as male gender (7), older age (2), living without a partner (8), number of previous episodes (9), number of previous hospitalizations (2), length of admission (10), rapid cycling (8), psychotic symptoms, (11) and substance use disorder (12).
Occupational adjustment is one functional domain with a major impact on personal and societal costs (13). In a large community sample of 2839 patients diagnosed with BD type I, 65% were unemployed and 40% were receiving disability or public assistance benefits, despite having achieved high educational levels (14). Compared with unipolar depressive patients, bipolar patients lose more workdays and represent around double the annual human capital loss per ill worker (15). Among other common physical and mental disorders, BD was considered to be the second illness with the strongest individual-level impact on this issue in a survey carried out in 24 countries (16).
Although previous studies have been conducted in the field of occupational disability, the measures used to assess it vary greatly across studies and sometimes criteria may substantially overlap (17). Some multicenter studies have assessed occupational outcome using a work subscale of the Longitudinal Interval Follow-up, Evaluation Range of Impaired Functioning Tool consisting of a Likert-type scale (18, 19). Subscales for occupational status from the Functioning Assessment Short Test (FAST) (2), the Sheehan Disability Scale (SDS) (20), the Social and Occupational Functioning Assessment Scale (SOFAS) (21, 22), the World Health Organization Disability Assessment Schedule (WHO-DAS) (23), and the Social Adjustment Scale (SAS) (24) have been employed as well as the Modified Vocational Status (10, 24). Other investigators have measured psychosocial functioning with the Global Assessment of Functioning (9), or considered occupational adaptation (25), absenteeism (26), paid employment (1), turnover in employment status (1), occupational stability (27), semi-structured interviews (28) or self-reported functional impairment (29). In this context, we considered a tangible and socially relevant statement such as receiving a severe disablement benefit (SDB) because of BD, which indicates an absolute disability to work (20, 30).
Aims of the study
Taking into consideration the repercussion of bipolar disorder on occupational status, the aim of this study was to detect predictors of occupational disability as well as to discern if episodic recurrence was more frequent in patients with occupational disability.
Material and methods
Study design and participants
The study entitled Predictors of Recurrence in BD in Spain (PREBIS) is an observational, prospective national, multicenter study of patients diagnosed with BD type I or II. A total of 69 Spanish centers took part in the recruitment. From June 2007 to March 2009, each center planned to enroll 10 consecutive out-patients and follow them during 1 year. Patients were required to have had at least one acute episode within the year prior to recruitment and to be in partial or total remission for a minimum of 2 months. This lapse of time was considered in order not to mistake recurrence with a relapse of the previous episode (31). Patients were 18 years or older and were treated in an out-patient clinic in the Spanish territory. Inability to carry out the follow-up plan at the investigator’s discretion and having taken part in a clinical trial in the 6 months prior to the study were considered as exclusion criteria (Fig. 1). All participants provided written informed consent after receiving a complete description of the study. Procedures were approved by the Ethical and Research Committee of each participating PREBIS Site.
The primary analysis of the PREBIS study has been reported elsewhere (32). A total of 595 bipolar patients were initially recruited and assessed for eligibility in the PREBIS study (Fig. 1). However, two were excluded owing to not meeting inclusion criteria and inability to carry out the follow-up plan at the investigator’s discretion. Of the remaining 593 subjects, occupational status was studied for this specific project: 226 patients were employed, 110 homemaker, four students, 48 unemployed, 70 retired, 31 receiving disability allowance, 101 receiving SDB because of psychiatric illness. Patients receiving a SDB (n = 101) or employed (n = 226) were selected and compared. Nearly three quarters of the sample in both groups were in euthymia (67.3% and 76.4%, respectively) and the rest presented subsyndromal symptomatology.
Procedures and outcome
A baseline assessment was carried out to obtain sociodemographic, clinical, therapeutic, and functional data using structured interviews. The former included age, gender, residency in rural or urban site, education regarding studies up to high school or less and marital status defined as with or without a stable partner. With regard to occupational status, subjects who were employed or receiving SDB were considered for this specific subanalysis. Clinical information included confirmation of BD diagnosis and psychiatric comorbidities, according to Diagnostic and Statistical Manual of Mental Disorders criteria (DSM-IV-TR) and information about medical comorbidities, age at onset of illness, polarity of first episode, time elapsed since first episode, time elapsed since diagnosis, time elapsed since first symptoms, number of depressive, manic, hypomanic and mixed episodes, total number of episodes, family history of psychiatric disorder, predominant polarity, seasonal pattern, number of admissions, attempted suicide, number of attempted suicides, number of attempted suicides within the last year, rapid cycling in the last year (four or more mood episodes during a 12-month period, consistent with DSM-IV-TR criteria), polarity of the most recent episode, length of the most recent episode, severity of the most recent episode assessed by Clinical Global Impression – Bipolar Version (CGI-BP) (33), psychotic symptoms in the most recent episode and mood congruency, admission during and length of the most recent episode. Symptomatology at baseline was assessed with the following scales: the Montgomery–Asberg Rating Scale (MADRS) (34) for depression, the Young Mania Rating Scale (YMRS) (35) for mania, the Hamilton Anxiety Rating Scale (HARS) (34) for anxiety, the CGI-BP (33) for overall clinical judgment and, the Oviedo Sleep Questionnaire (36) for sleep. Pharmacological treatment, adherence, and psychosocial intervention were reported out. Functional status was evaluated by the FAST (37) and the SOFAS (38).
Reassessment was conducted at subsequent visits at months 3, 6, 9, and 12 after baseline. Clinical state was assessed and classified as: euthymia if the patient had two or fewer symptoms corresponding to criteria of a DSM-IV-TR mood episode; subsyndromal state if the patient met criteria for more than two symptoms corresponding to a DSM-IV-TR mood episode criteria without fulfilling criteria for an episode; and recurrence if the patient fulfilled criteria for a DSM-IV-TR mood episode or was admitted because of a mood episode during the 3 months since the last visit. The YMRS, MADRS, CGI-BD, HARS, FAST, SOFAS scales, and the Oviedo Sleep Questionnaire were also administered. Treatment and adherence were registered.
Sociodemographic and clinical data were analyzed with descriptive statistics. Frequencies and percentages were presented for categorical data. Measures of central tendency such as means and medians and of dispersion with standard deviations and ranges were calculated for continuous data. The occupational groups receiving a SDB and being employed were compared for the aforementioned variables by parametric and non-parametric tests, as appropriate.
A logistic regression model was used to identify factors associated with receiving SDB. The variables introduced in the model were based on clinical relevance and on statistical significance (P < 0.05). Receiving a SDB was considered as a dependent variable and independent variables were: age, education (high school or less), marital status (without stable partner), time elapsed since first episode, number of depressive, manic, hypomanic and mixed episodes, rapid cycling, number of admissions, attempted suicide, psychotic symptoms in the most recent episode, axis II comorbidity, and YMRS, MADRS, and HARS basal scores. Odds ratios (OR) with 95% confidence intervals (CI) were obtained. Backward selection method was used to eliminate risk factors not significantly contributing to the prediction of occupational disability.
A Cox regression model was built to compare time to episodic recurrence along 1-year of follow-up between bipolar patients receiving a SDB and employed patients.
Analyses were conducted using the software Statistical Analysis Software version 8.2 (SAS Institute Inc., Cary, NC, USA). All P-values were two-tailed and statistical significance was set at P <0.05.
One-hundred and seventy-two (52.6%) patients of the sample were female with a mean age of 43.5 (SD 10.6) years (Table 1). Two-hundred and forty-eight (77.7%) patients had achieved a high school degree. One-hundred and seventy-seven (54.1%) patients did not have a stable partner and 139 (42.5%) had a family history of psychiatric disorder. Most of the sample (88.1%) was diagnosed with BD type I and only 39 (11.9%) patients had BD type II. Regarding comorbidities, substance use was the most prevalent (32.0%). The most frequent polarities of first episode were depressive and manic (44.8% and 40.8%, respectively), being depressive in the most recent episode (41.1%). The time elapsed from the first episode was 173.8 (SD 114.7) months. Patients required 3.4 (SD 3.2) admissions until recruitment. Only 94 (28.8%) of them presented a seasonal pattern. Manic predominant polarity was nearly two-fold that of depressive predominant polarity (30.3% and 18.4%, respectively). Almost one quarter of the patients (23.0%) had attempted suicide, nearly one-third (29.3%) presented with psychotic symptoms in the most recent episode, and only 5.2% were classified as rapid cycling.
Table 1. Sociodemographic and clinical characteristics of the total sample and the occupational groups
Total (n = 327)
Employed (n = 226)
Disabled (n = 101)
χ2 or t-test or Wilcoxon test
†Values are indicated as n (%).
‡Values are indicated as mean (SD).
No, number; ADHD, Attention-Deficit Hyperactivity Disorder; MARS, Montgomery–Asberg Rating Scale; YMRS, Young Mania Rating Scale; HARS, Hamilton Anxiety Rating Scale; FAST, Functioning Assessment Short Test; SOFAS, Social and Occupational Functioning Assessment Scale.
Age at enrollment (years)‡
High school or less†
Without stable partner†
Family history of psychiatric disorder†
Bipolar disorder type†
Axis II disorder†
Age at onset of illness (years)‡
Polarity of first episode†
Time since first episode (months)‡
Total No. of episodes‡
No. of depressive episodes
No. of manic episodes
No. of hypomanic episodes
No. of mixed episodes
No. of admissions‡
Rapid cycling in last year†
Polarity of last episode†
Psychotic symptoms in last episode†
Basal MARS score‡
Basal YMRS score‡
Basal HARS score‡
Occupational status and predictors
The bipolar patients who received a SDB were older at the time of enrollment [t(325) = −6.63, P < 0.0001], had poorer educational status [χ2(1) = 13.9724, P = 0.0002], were less likely to have a stable partner [χ2(1) = 6.1571, P = 0.0131], and were diagnosed with more axis II comorbidities than employed bipolar patients [χ2(1) = 4.9269, P = 0.0264] (Table 1). From the clinical point of view, the bipolar patients who received a SDB presented with more previous episodes irrespective of mood polarity [t(113) = −4.92, P < 0.0001], had a longer time since the first episode [t(160) = −5.05, P < 0.0001], more attempted suicide [χ2(1) = 15.3423, P = 0.0001], more rapid cycling [χ2(1) = 6.5019, P = 0.0108], and required more admissions [t(113) = −4.93, P < 0.0001] than employed bipolar patients (Table 1). With regard to the most recent episode, bipolar patients who received a SDB presented with more psychotic symptoms than employed bipolar patients [χ2(1) = 4.8813, P = 0.0271]. Baseline clinical and functional assessments using the MADRS, YMRS, HARS, FAST showed significantly higher scores in bipolar patients who received a SDB compared with employed bipolar patients (Wilcoxon statistic = 16247.5, P < 0.0338; Wilcoxon statistic = 16633.5, P < 0.0076; Wilcoxon statistic = 16786.5, P < 0.0026; Wilcoxon statistic = 17285.5, P < 0.0001, respectively), and significantly lower scores in SOFAS (Wilcoxon statistic = 9735.5, P < 0.0001) indicating that these patients were more symptomatic and disabled. In addition, we analyzed the FAST score without considering the occupational domain and a significantly higher impairment was verified in bipolar patients who received a SDB compared with employed bipolar patients (Wilcoxon statistic = 17855.5, P < 0.0001). Significantly, worse performance was also found in each FAST scale domain of autonomy, occupational functioning, cognitive functioning, financial issues, interpersonal relationships, and leisure time in bipolar patients receiving a SDB compared to employed patients (Wilcoxon statistic = 18189.0, P < 0.0001; Wilcoxon statistic = 18259.5, P < 0.0001; Wilcoxon statistic = 19406.0, P < 0.0001; Wilcoxon statistic = 17490.0, P < 0.0004; Wilcoxon statistic = 18243.0, P < 0.0001; Wilcoxon statistic = 19142.0, P < 0.0001, respectively).
The results of the multivariate logistic regression model showed four factors which were significantly associated with receiving SDB (Table 2). Bipolar patients with axis II comorbidity received more SDB than bipolar patients without personality disorder comorbidity (OR = 2.94, CI: 1.26–6.86, P = 0.0129). Bipolar patients with more number of manic episodes received more SDB than bipolar patients with fewer episodes (OR = 1.21, CI: 1.10–1.34, P = 0.0002). Bipolar patients without a stable partner received more SDB than bipolar patients with a partner (OR = 2.44, CI: 1.34–4.44, P = 0.0036) and older bipolar patients received more SDB than younger patients (OR = 1.08, CI: 1.05–1.12, P < 0.0001). Neither education (high school or less), time elapsed since first episode, number of depressive, hypomanic and mixed episodes, rapid cycling, number of admissions, attempted suicide, psychotic symptoms in last episode nor YMRS, MADRS, and HARS scores showed any significant association with receiving a SDB.
Table 2. Predictors of receiving a severe disablement benefit*
OR, odds ratio; CI, confidence interval.
*Variables initially introduced in the model: age, education (high school or less), marital status (without stable partner), time elapsed since first episode, number of depressive, manic, hypomanic and mixed episodes, rapid cycling, number of admissions, attempted suicide, psychotic symptoms in last episode, axis II comorbidity, Young Mania Rating Scale, Montgomery–Asberg Rating Scale, and Hamilton Anxiety Rating Scale at baseline.
Without stable partner
Axis II comorbidity
No. of manic episodes
Time to recurrence
Bipolar patients receiving a SDB had more recurrences than employed patients during the 1-year follow-up [χ2(1) = 9.2710, P = 0.0023] (Table 1). No significant differences were detected regarding the polarity of the recurrence [χ2(1) = 1.7095, P = 0.191]. Time to recurrence for 1-year of follow-up was compared among bipolar patients receiving a SDB with employed bipolar patients. Significant differences were seen between the two groups (HR = 1.91, CI: 1.23–2.97 P = 0.0038) (Fig. 2). The time until recurrence in 25% of the bipolar patients receiving SDB was 6.08 months (CI: 1.05–1.12) being 13.08 (CI: 9.60 to –) months in employed patients.
The present study reports predictors of receiving a SDB because of BD: axis II comorbidity, more previous manic episodes, being without a stable partner, and older age. The granting of a SDB entails a major impairment in occupational status as this allowance is provided to people who are unable to properly perform a job, after careful assessment of government sanitary authorities. Hence, selecting these patients offers a uniform sample with occupational disability for study.
Almost all psychiatric disorders have been associated with functional impairment (39). Therefore, we expected comorbidity to be associated with poorer functioning and, in particular, with lower occupational functioning (28, 40). Axis II disorders were the only comorbidities found to be related to receiving SDB in both unadjusted as well as in adjusted results. This finding is not only justified by the high prevalence of personality disorder in BD (41) but also by the strong evidence of poor functioning in bipolar patients with comorbid personality disorders (28). The difficulties that a bipolar patient may have to adjust to the requirements of a job will be enhanced by the lack of adaptive coping strategies present in individuals with personality disorders. Moreover, another reason which may explain this association is poor treatment adherence in these patients (42). Unoptimal adherence is commonly known to result in more frequent recurrences which, at the same time, lead to poorer cognitive performance and functioning (43). Regarding other comorbidities, Dickerson et al. (24) found substance abuse as a predictor of absenteeism. In the Zimmerman et al. study (40), patients who were unemployed during more than 2 years of the prior 5 years experienced more current panic disorder and lifetime history of alcohol abuse or dependence than patients who had not been unemployed during this time. These comorbidities were not associated with receiving SDB in our study.
Previous episodes have been reported as a predictor of functional impairment (9). However, there is no agreement as yet as to which polarity of the episodes could be more deleterious (44). MacQueen et al. (9) divided the sample into four groups considering predominant polarity. In a subanalysis, the number of previous depressions was reported to be a stronger determinant of outcome than number of previous manias. Nevertheless, our results, which focus on occupational disability are in agreement with the study by Gutiérrez-Rojas et al. (20) who described previous manic episodes to be independently associated with work impairment. This group suggested that previous manic episodes may play a role in occupational disability and in family life whereas previous depressive episodes may be related to social life impairment (20).
In contrast, it is highly consistent that current subsyndromal depressive symptoms have a negative impact on overall functioning (45). In their study on severity of mood symptoms and work productivity among bipolar patients, Simon et al. (1) concluded that depressive symptoms were strongly associated with decreased probability of employment and with more days missed from work because of illness. These findings were not reproduced with manic or hypomanic symptoms. Current symptoms were not identified as a predictor in our study, perhaps because our sample was in partial or total remission rather than in remission, presenting with subthreshold symptoms or with a full episode as in the study by Simon et al.
Regarding demographic variables, marital status is acknowledged to influence the progression of the disorder and functioning (8). Not having a stable partner could be related to limited social support. In previous literature, social support has been reported to be of fundamental importance in the evolution of patients with BD as well as in their quality of life (46, 47). Cohen et al. (48) described poor quality close relationships as a predictor of recurrence in an urban community sample of patients with BD type I. Older age has also been determined as a predictor of functional impairment (2). Age-related cognitive impairment have been reported to impact on middle-aged and elderly adults with BD (49, 50). Moreover, older patients experience more comorbid physical illness and hence, they tend to receive more polypharmacy. On the whole, this may be associated with even more chronicity and impairment (51). However, older age has not been previously established as a predictor of occupational disability. Progression of the disorder advances along the years. Hence, older age could be related to chronicity of the disorder, although time elapsed since first episode or age at onset of illness were not identified as predictors (23), suggesting that other factors associated with impairment with age could be related.
Other variables such as low educational level, rapid cycling, and psychotic symptoms that previously related to functional impairment (8, 11) did not seem to have an association with receiving a SDB in the logistic regression model (12).
Our survival analysis suggested that bipolar patients who were receiving a SDB had more recurrences than employed bipolar patients. In the McLean-Harvard First-Episode Mania Study, Tohen et al. (52) described low prehospitalization occupational status in a 2–4 year follow-up as a predictor of manic recurrence. An increasing number of episodes is linked to a reduction in the likelihood of response to appropriate treatment, whether psychopharmacological as with lithium (53, 54) or psychological as in cognitive behavioural therapy (55). Moreover, recurrence is further associated with poorer social adjustment (56). Social impairment was also evidenced in our study. Bipolar patients receiving a SDB scored higher in the FAST scale and lower in the SOFAS scores than employed bipolar patients. Hence, progression of the disorder in these patients may be more damaging progression than in employed patients.
Neuroprogression may advance swiftly and lead to a broader impairment ranging from social to occupational impairment (57). This progressive trajectory highlights the imperative for early diagnosis and intervention to prevent the development of general disability (56).
Our study has some limitations. Our sample was collected in the context of a prospective, multicenter, observational study across Spain and thus, it may not be possible to generalize some finding to other locations. Furthermore, we could not confirm whether the SDB had granted for BD by Welfare Services. This information was provided by the patients and verified with the family whenever possible. A high number of patients were receiving a SDB compared to bibliography. This fact could be because of the high specialization of some centers involved in the study which provide care to severe bipolar patients. Lastly, we did not study the role that pharmacological treatments might have played in occupational status as has been widely carried out in functional impairment, given that the treatment was naturalistic (58). However, the sample was representative of bipolar patients attending public health services in Spain and all the information was obtained by certified trained psychiatrists.
In conclusion, occupational impairment is commonly encountered among bipolar patients and is not only present in the acute phases but is maintained over long periods of time. This study sheds some light onto occupational disability in bipolar patients, the variables that influence this disability, and the outcome of this group of severe patients, from the social point of view. The results of this study emphasize the importance of early diagnosis and treatment of BD, the clinical relevance of comorbidity with personality disorders, the impact of social support, and the need to closely monitor patients who are occupationally disabled, given their high tendency to recurrence.
This study was sponsored by AstraZeneca. Miguel A González provided statistical support. Dr. I. Grande has received a research grant from Hospital Clínic de Barcelona (HCB). Clinical Trial registration number: NCT00690859.
Declaration of interest
Prof. E. Vieta has received research grants, and served as consultant, advisor or speaker for the following companies: AstraZeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly, Forest Research Institute, Gedeon Richter, GlaxoSmithKline, Janssen-Cilag, Jazz, Lundbeck, Merck Sharp & Dohme, Novartis, Otsuka, Pfizer Inc, Sanofi-Aventis, Servier, Takeda, and UBC.
Dr. E. Prieto: is employed by Astra Zeneca.
Prof. J. Saiz-Ruiz, has been a speaker for and on the advisory boards of Lilly, GlaxoSmithKline, Lundbeck, Janssen, Servier and Pfizer; and has received grant/honoraria from Lilly, Astra-Zeneca, Bristol-Myers and Wyeth.
Dr. J.M. Montes has collaborated with the following companies: Grants from Astra-Zeneca, Pfizer. Servier; Speaker Board: Astra-Zeneca, BMS. Advisory Board: Lundbeck, Pfizer.
Dr. A. González-Pinto has received grant support, acted as consultant, or given presentations for the following pharmaceutical companies: Almirall (Barcelona, Spain), Astra-Zeneca (Madrid, Spain), Eli Lilly (IN, USA), Glaxo-Smith-Kline (Madrid, Spain), Lundbeck (Barcelona, Spain), Pfizer (Madrid, Spain), Sanofi-Aventis (Paris, France), MSD, BMS, Janssen and Boehringer Ingelheim (Barcelona, Spain).
Dr. De Dios has served as speaker for the following companies: AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Janssen-Cilag and Lundbeck.
Dr. Jose Manuel Goikolea has been a speaker o advisory board for Astra-Zeneca, Bristol Myers-Squibb, Eli Lilly, Glaxo-Smith-Kline, Janssen-Cilag, Merck Sharpe and Dohme, Otsuka, Pfizer, Sanofi-Aventis.