Reading focus
Syllabus: What Explains Mood Episodes in Bipolar Disorder?
Follow how the analysis separates usual levels from changes around those levels.
Why you’re reading it. Evaluate part of the feedback proposal without equating temporal prediction with proof of causation.
A question to consider. If insomnia predicts later depression but not increased activation on average, which claims from the framework remain plausible but untested by these measures?
Why this reading
People who usually sleep poorly may differ from other people, while a change in one person’s sleep poses a different question. Lewis and colleagues separate these comparisons using repeated observations. The study shows why a model of an individual’s changing mood cannot simply be read from a correlation across people.
Before you read
A within-person association concerns departures from that person’s usual level; a between-person association compares usual levels across people. A lagged relationship connects measurements separated in time. Inertia describes how strongly a deviation carries over to the next observation. The model allows several relationships to differ between participants; understanding these distinctions does not require reproducing the estimation.
Read for
- Follow how Figure 1 separates stable differences, changing states, and associations over time. Which parameter answers each substantive question?
- Inspect the insomnia and mood measures, including removal of sleep items from the mood totals. Why does this reduce one form of overlap without making the measures independent?
- Compare the average insomnia–depression and insomnia–activation relationships with variation across people. Consider what weekly measurement, missing observations, and medication information leave unresolved.
Consider afterwards
- Could people with more insomnia have more depressive symptoms overall even when a change in insomnia does not predict a change for a particular person? Explain how.
- What might daily sleep-duration data reveal that weekly insomnia ratings cannot?
Editorial context
The analysis used 21 months of weekly self-reports from 649 UK participants with bipolar I or II. It lacked medication data and selected people able and willing to provide sustained observations. Insomnia predicted subsequent depressive symptoms on average; the average prospective association with increased manic symptoms was not established, although relationships varied. Insomnia was not a measure of reduced need for sleep or circadian phase. Temporal modeling does not eliminate unmeasured causes, and symptom fluctuations are not equivalent to diagnosed episodes.