Introduction
Despite the fact that clinical staging is widely accepted in many medical areas, psychiatry seems to be the exception to this rule (Reference Fava and Kellner1). Neglecting staging models may be especially unwise in an illness such as bipolar disorder, where several clinical outcome studies clearly show how the prognosis worsens dramatically over time and with number of episodes.
Staging is particularly useful in order to differentiate pharmacological and non-pharmacological treatment strategies according to it. To this end, it is essential to determine whether the number of previous episodes – which might be subsequently categorised into a staging model – may influence treatment response. Regarding the psychological treatments for bipolar disorder, there is some evidence that the number of previous episodes may affect the response to cognitive-behavioural therapy (Reference Scott, Paykel and Morriss2). To date, no data is available regarding other existing psychotherapies (Reference Vieta3). Hereby, we present a 5-year post hoc study on the efficacy of group psychoeducation for bipolar disorders according to the number of previous episodes.
Materials and methods
This is a post hoc, exploratory subanalysis of data obtained from a larger study on the 5-year efficacy of group psychoeducation in the prophylaxis of recurrences in patients with bipolar disorder (Reference Colom, Vieta and Sanchez-Moreno4). In the primary study, bipolar I and II subjects were randomised either to a manual-based group psychoeducative intervention (Reference Colom and Vieta5) or to a non-structured group intervention (control group). All patients received pharmacological treatment as well. Further information on study procedures is available in the original article (Reference Colom, Vieta and Martinez-Aran6).
For this subanalysis, we have compared the outcome of psychoeducated versus non-psychoeducated patients according to the number of previous episodes at study entry. Data regarding previous episodes at study entry was recorded prospectively in the majority of cases (up to 70% of the registered episodes). All the patients included in the original study were recruited from the naturalistic prospective follow-up of the Bipolar Disorders Program of the Hospital Clinic database.
Retention in the primary follow-up study for at least 2 years was considered as inclusion criteria. The retrospective assessment of previous episodes, when necessary, was performed by consulting medical registers, as well as patient and family reports.
Assessments
All subjects were assessed monthly by trained psychiatrists and every 2 weeks by research assistants who had instructions to contact the psychiatrist if a recurrence was suspected. Both evaluators were blind to treatment, and patients were requested not to reveal significant details.
Recurrence and symptom assessment, as well as treatment registration, was performed every 2 months. Diagnostic and psychometric baseline evaluation included the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I), the Structured Clinical Interview for DSM-IV Axis II Disorders (SCID-II), Young Mania Rating Scale (YMRS), Hamilton Depression Rating Scale (HDRS-17) and Holmes and Rahe Inventory for stressful life events, which were repeated every 2 months or whenever a new episode was suspected by the psychiatrist in charge.
Main outcome measures
Time to recurrence
The primary outcome measure was time to recurrence, defined as the time (days) elapsed between the baseline and the emergence of a new acute episode according to DSM-IV criteria (YMRS ≥ 20 for manic recurrence; YMRS ≥ 12 for hypomanic recurrence; HDRS-17 ≥ 17 for depressive recurrence; HDRS-17 ≥ 12 and YMRS ≥ 20 for mixed recurrence).
Time spent acutely ill
A secondary outcome measure was time spent ill. We obtained this data by prospectively registering the number of days in which patients fulfilled criteria for a specific episode.
Results
Considering time to recurrence as outcome measure, we found that patients with seven or more episodes at study entry did not show any significant improvement with psychoeducation according to Kaplan-Meier survival curve analysis (control group, n = 25; psychoeducation group, n = 28; log-rank = 3.375, significance = 0.066). Patients with six or less episodes showed a statistically significant improvement on the same test (control group, n = 32; psychoeducation group, n = 32; log-rank = 4.295, significance = 0.038).
When we considered time spent ill as outcome measure, patients with more than 14 episodes did not benefit from psychoeducation. Patients with up to 7 or 8 episodes showed a benefit in terms of fewer days spent in hypomania, depression, mixed episodes or any episodes but not mania, while patients with up to 9–14 episodes showed a benefit in terms of fewer days spent in hypomania and depression but not in mixed states or mania. Only patients who presented up to 6 episodes showed reduction in time spent in any episode polarity. Table 1 shows the differences between groups regarding time spent acutely ill during the study.
Comparison of psychoeducated and non-psychoeducated patients on time spent on an episode at 5 years according to the number of previous episodes
CG, control group; NS, non-significant; PG, psychoeducation group; SD, standard deviation; Ts, time spent (days).
Discussion
This post hoc subanalysis attempts to determine a possible relationship between poor response to psychotherapy and number of previous affective episodes, an issue already proposed by other studies (Reference Scott, Paykel and Morriss2). Despite all the methodological and statistical limitations of this preliminary proof-of-concept test study, the number of previous episodes clearly appears to reduce the response to psychoeducation, perhaps in a more subtle way than that observed in cognitive-behavioural therapy, where the therapy was not effective for patients with more than 11 episodes (Reference Scott, Paykel and Morriss2). In our study, the treatment efficacy cut point differs according to the outcome variable used. Thus, if time to recurrence is considered, seven previous episodes draw the line between responders and non-responders to group psychoeducation. On the other hand, when time spent ill is considered, this limit raises to 14 previous episodes. Therefore, psychoeducation may have some effect on patients between 7 and 14 previous episodes, although more subtle.
The efficacy of group psychoeducation in reducing time spent in depressive episodes is significant and, in fact, it vanishes only in patients who present more than 14 episodes. Even then, a type II error may be present, as the sample size is considerably reduced. The efficacy of psychoeducation on depression should be highlighted, particularly as depression represents the predominant polarity in bipolar disorder (Reference Colom, Vieta, Daban, Pacchiarotti and Sánchez-Moreno7,Reference Rosa, Andreazza and Kunz8) and leads to a poor functional and cognitive outcome (Reference Malhi, Ivanovski, Hadzi-Pavlovic, Mitchell, Vieta and Sachdev9).
Several hypotheses may explain the potential lack of efficacy of psychoeducation with more veteran patients. First, the likelihood of suffering from a consistent cognitive impairment is largely increased in this subgroup, as reported by Martinez-Aran et al. (Reference Martinez-Aran, Vieta and Reinares10). Second, patients with a higher number of previous episodes may have more difficulties in changing habits and, hence, may benefit less from psychoeducation. The more optimistic view would point at the lack of difference between psychoeducated patients and controls as a result of longer experience that veteran patients have in the management of bipolar disorder, thereby receiving less benefit from psychoeducation.
Our study clearly cannot test the staging model for bipolar disorders as originally described by Berk et al. (Reference Berk, Conus and Lucas11,Reference Berk, Hallam and McGorry12). According to this proposal, patients could be classified into five different stages: ‘stage 0’ (asymptomatic at risk stage), ‘stage 1a and 1b’ (prodromes presentation), ‘stage 2’ (first-episode mood disorder), ‘stage 3’ (recurrence) and ‘stage 4’ (treatment resistance). Inclusion criteria of our original study (Reference Colom, Vieta and Martinez-Aran6) consisted on a reliable diagnosis of bipolar disorder which, by definition, excludes patients on stages 0 and 1 and most patients on stage 3, unless they presented a single episode of mania or hypomania (at onset), which was the case only for four patients (3%) of our original sample (n = 120). As patients had to be euthymic for at least 6 months before the study entry, none of them would fulfill criteria for stage 4.
Our study has many limitations; namely, it was not designed or powered to test the present hypothesis; some data on prior episodes was collected retrospectively and, finally, the sample size decreased every time we considered patients with a higher number of previous episodes that may lead to a type II error for this subgroup of patients. The above-mentioned limitations may affect the interpretation of our results, thus this article should be merely seen as a preliminary proof-of-concept data
On the other hand, the present data may have clinical relevance, pointing at the need to start psychoeducation promptly, when the number of episodes is still low, supporting the idea of early intervention (Reference Scott, Colom and Vieta13). However, we think that present results should not lead to therapeutic nihilism or pessimism with our more veteran patients. Perhaps, group psychoeducation may not be as effective for this subgroup because it was not tailored for these difficult-to-treat patients. Thus, we would need a reformulation of psychoeducation for veteran patients that may include other strategies such as neurocognitive rehabilitation. Further research is needed in this area.
Acknowledgments
The authors of this study would like to thank the support and funding of the Spanish Ministry of Health, Instituto de Salud Carlos III, CIBERSAM. Dr Colom is funded by the Spanish Ministry of Science and Innovation, Instituto Carlos III, through a ‘Miguel Servet' postdoctoral contract (CP08/00140). Dr Martínez-Arán is funded by the Spanish Ministry of Science and Innovation, Instituto Carlos III, through a ‘Miguel Servet' postdoctoral contract (CP07/00144). The present study was performed at the Bipolar Disorders Program, CIBERSAM-IDIBAPS, Hospital Clinic of Barcelona. Dr Francesc Colom has served as advisory or speaker for the following companies: Astra Zeneca, Pfizer Inc, Glaxo-Smith-Kline, Eli-Lilly, Sanofi-Aventis, Tecnifar & Shire. Dr Eduard Vieta has served as consultant, advisor or speaker for the following companies: AstraZeneca, Bial, Bristol-Myers, Eli-Lilly, Glaxo-Smith-Kline, Jannssen-Cilag, Lundbeck, Merck-Sharp and Dohme, Novartis, Organon, Otsuka, Pfizer Inc, Sanofi-Aventis, Servier & UCB Pharmaceuticals. Drs Martínez-Arán, Torrent and Reinares, Mrs Palomino-Otiniano, Bonnin, Rosa and Franco and Mr Mazzarini report no conflicts of interest.
