Actigraph-derived variables as predictors of bipolar disorder traits and states: theoretical and empirical considerations

Bipolar disorder (BD) is a chronic and potentially disabling mood disorder that is characterised by recurrent episodes of mania and depression. Signature features of episodes include changes in sleep and psychomotor activity patterns. Research has also proposed that instability of the circadian system may be part of the neurobiological diathesis to the disorder. The aim of the current project was to investigate all of these features for associations with vulnerability to BD at both the trait and state level using actigraphy, a non-invasive measure of activity rhythms and sleep. The outcomes of the project were also expected to inform clinical practice on the usefulness of actigraphic monitoring for the purposes of episode prodrome identification. Three studies were conducted in pursuit of these aims. Study 1 investigated the trait vulnerability concept by conducting factor analysis of two prominent measures of vulnerability to BD - the 2-dimension GBI and the 5-dimension TEMPS-A - using a mixed sample of well participants and those with a self-reported history of depression and/or mania (N = 484). The results showed that two dimensions, characterized by vulnerability to depression and vulnerability to mania, provided the most parsimonious solution in this sample. Items from the GBI measure dominated the explained variance in the solution. In Study 2, the GBI measure was used to separate a sample of well participants into groups of higher and lower levels of trait vulnerability to BD (n = 35 for both groups). The groups were then compared across 7 days on a series of sleep and circadian activity variables derived from actigraphy. The higher vulnerability group had a significantly lower mean amplitude in the circadian activity rhythm compared to the lower vulnerability group, an indication of reduced circadian stability. There were no differences however, on other features of activity or sleep. Study 3 investigated the prospective association between sleep, activity, and mood deterioration on a daily basis (state vulnerability to BD) in a sample of outpatients with BD (N = 11). Actigraphy was again used to measure the sleep and circadian activity variables. Mood and additional sleep variables were measured using self-report. Multilevel regression analyses showed that self-reported sleep, but not actigraph-derived sleep, was a significant predictor of mood deterioration the following day. The results also showed that daytime activity was a significant predictor of mood on the same day. There were no significant associations between instability of the circadian activity rhythm and mood deterioration. The following conclusions were made based on the findings of the three-study design of the current project; (i) a 2-dimension structure appears to provide the most theoretically consistent and parsimonious model of trait vulnerability to BD, (ii) the circadian instability hypothesis of trait vulnerability to BD is tentatively supported, (iii) daytime activity is the only measure derived from actigraphy to be prospectively associated with daily mood change in BD. The latter conclusion in particular provides impetus for suggestions of how actigraphic monitoring may be applied in the clinical setting, including limitations associated with such monitoring. Accurate identification of manic and depressive prodromes is essential for successful relapse prevention. The current project identified a limited number of suitable candidates for actigraphic monitoring of vulnerability to BD states, as well as a potential biomarker of trait vulnerability to BD (reduced stability of the 24-hour activity rhythm).