Evidence of chaotic mood variation in bipolar disorder

Using long-term daily mood records obtained from patients with bipolar disorder and normal subjects, we sought to determine the temporal pattern of mood in bipolar disorder.

Time series of 1.0 to 2.5 years from seven rapid-cycling patients with bipolar disorder and 28 normal controls were obtained. These were evaluated with several techniques to identify whether the temporal pattern of mood originates from a periodic, a random, or a deterministic source.

True cyclicity was not apparent in the power spectra of either the normal subjects or the patients with bipolar disorder. Instead, spectra with a broadband "l/f" shape were observed in both groups, and these spectra were significantly flatter in normal subjects (P = .02). Correlation dimension estimates are a measure of nonlinear deterministic structure, and convergent estimates could be obtained for six of the seven patients with bipolar disorder and none of the normal subjects (P < .001). Additional findings are consistent with these results.

These studies indicate that mood in patients with bipolar disorder is not truly cyclic for extended periods. Nonetheless, self-rated mood in bipolar disorder is significantly more organized than self-rated mood in normal subjects and can be characterized as a low-dimensional chaotic process. This characterization of the dynamics of bipolar disorder provides a unitary theoretical framework that can accommodate neurobiologic and psychosocial data and can reconcile existing models for the pathogenesis of the disorder. Furthermore, consideration of the dynamical structure of bipolar disorder may lead to new methods for predicting and controlling pathologic mood.