Do self-ratings of the Pittsburgh Sleep Quality Index reflect actigraphy recordings of sleep quality or variability? An exploratory study of bipolar disorders versus healthy controls

Sleep disturbances are typical symptoms of acute episodes of bipolar disorder (BD) and differentiate euthymic BD cases from healthy controls (HC). Researchers often employ objective recordings to evaluate sleep patterns, such as actigraphy, whilst clinicians often use subjective ratings, such as the Pittsburgh Sleep Quality Index (PSQI). As evidence suggests the measures may disagree, we decided to compare subjective (PSQI) and objective (3 weeks of actigraphy) sleep profiles in BD cases and HC (n = 154). We examined whether a dimensional approach helps to illustrate different patterns of sleep disturbances and whether the concordance between subjective and objective recordings varies according to clinical status (BD versus HC). Principal component analysis (PCA) extracted two factors from the PSQI, and separate PCAs of actigraphy recordings extracted two factors for mean values of sleep parameters and one factor for intra-individual variability. Correlational and linear regression analyses of PCA-derived dimensions demonstrated that, in both BD and HC, a PSQI "Sleep duration-efficiency" factor was significantly correlated with an actigraphy "Sleep initiation-duration" factor. Furthermore, in BD cases only, the PSQI total score and a PSQI "Sleep Impairments" factor were each significantly correlated with an actigraphy "Sleep Variability" factor. Overall, we found that subjective experiences of sleep may be modulated by different components of objectively recorded sleep in BD compared with HC. Also, the use of PCA enabled us to consider the multi-dimensional nature of subjective sleep, whilst the inclusion of intra-individual sleep variability afforded a more subtle evaluation of objective sleep.