Multivariate analysis of bipolar mania: Retrospectively assessed structure of bipolar I manic and mixed episodes in randomized clinical trial participants

Abstract

Background

Manic episodes are heterogeneous. Mixed states may differ in important clinical characteristics from other manic episodes. However, it has not been established whether mixed states are a distinct type of episodes, or a common basic structure exists across manic episodes.

Methods

Using 2179 well-characterized subjects in the pretreatment phase of six randomized, clinical trials, we conducted rotated factor analysis followed by cluster analysis, using all items from the Young Mania Rating Scale and the Montgomery–Åsberg Depression Scale. Analyses were conducted for all subjects (n=2179) and for those in Diagnostic and Statistical Manual of Mental Disorders – Fourth Edition (DSM-IV) mixed (n=644) and non-mixed (n=1535) episodes separately.

Results

There were five factors characterized (in order of variance accounted for) as depression, mania, sleep disturbance, judgment/impulsivity and irritability/hostility. Cluster analysis identified five clusters. Three were predominately manic, with depression scores below average for the overall group. Two had high average depression scores; these clusters differed in irritability/hostility. Judgment/impulsivity scores were similar across factors. Essentially identical factors and clusters existed whether analyses were done in all subjects or only in subjects classified by DSM-IV as mixed or non-mixed.

Limitations

Exclusion criteria of studies may limit generalizability of findings.

Discussion

All manic episodes, whether mixed or non-mixed, shared a similar structure according to factor/cluster analysis. Patients with high depression factor scores were heterogeneous with respect to irritability. These data suggest that depressive symptoms should be considered a dimensional property across manic episodes, rather than as defining a specific type of episode.

Introduction

Bipolar disorder has a complex presentation where episodes of affective dysregulation recur over time. While depressive and manic episodes have been described for centuries, there is still surprisingly little consensus over the structure of an episode of bipolar disorder. What are the essential building blocks of an episode and how can they combine?

While episodes of bipolar disorder are considered to be depressive or manic (hence ‘bipolar’), it has long been recognized that components of depressive and manic states can combine to produce so-called mixed states. Definitions of mixed states are controversial:

• It is not established whether depression is a dimensional characteristic of manic episodes, or whether mixed states are a distinct category of depressive-manic episodes.

• Mixed states may themselves be heterogeneous, with different forms having different clinical features and outcomes.

• Both the Diagnostic and Statistical Manual of Mental Disorders – Fourth Edition (DSM-IV) and the International Statistical Classification of Diseases and Related Health Problems – Tenth Revision (ICD-10) diagnostic criteria used a categoric approach for mixed mania that essentially requires presence of full syndromal depression and mania; it has been suggested that this is too restrictive, and that meaningful clinical features emerge at subsyndromal levels of depressive symptoms (Swann et al., 2009).

• Mixed states or mixed characteristics could be validated by establishing relationships to response to treatments, course of illness or other clinical characteristics (Swann et al., 2002).

The symptomatic structure of manic episodes has been explored by applying multivariate statistics to symptom rating scales (Azorin et al., 2008, Bertschy et al., 2007, Cassidy et al., 1998, Dilsaver et al., 1999, Harvey et al., 2008, Lipkovich et al., 2008, Rossi et al., 2001, Sato et al., 2002, Swann et al., 2001). These studies have confirmed the heterogeneity of manic and mixed episodes, but had limitations (Cassidy et al., 1998, Dilsaver et al., 1999, Rossi et al., 2001, Sato et al., 2002), including relatively small samples and non-standardized symptom measures. Better understanding of the clinical structure of manic episodes would be useful in predicting treatment response and clinical course of illness, and in answering basic questions about episodes of bipolar disorder. Perhaps most important among these is whether there is a common basic structure underlying the apparent heterogeneity of bipolar episodes.

Using a large database from clinical trials in manic and mixed episodes (El Mallakh et al., 2010, Keck et al., 2003, Keck et al., 2009, Sachs et al., 2006, Vieta et al., 2005, Young et al., 2009), we have investigated the structure of manic and mixed episodes.

We conducted a factor/cluster analysis of Young Mania Rating Scale (YMRS) and Montgomery–Åsberg Depression Rating Scale (MADRS) line item scores from a pooled population of patients with manic or mixed episodes of bipolar I disorder enrolled in six clinical trials with aripiprazole in order to:

• identify the factor structure of psychiatric symptoms of the patients included in these studies;

• determine whether there are distinct clusters of episode types based on these factors;

• determine whether the same factor structure exists for episodes classified as being mixed or non-mixed for the sample as a whole.