Using Generalized Additive Models to Compare Age Trends in Actigraphy and Self-Report Sleep Measures Across the Life Span in Individuals with and without Psychiatric Disorders

As of 2016, over 1 billion people worldwide, or about 16% of the world’s population, were affected by mental and addictive disorders, causing a considerable burden of disease (Rehm & Shield, 2019). Sleep disturbances are present across mental disorders (Baglioni et al., 2016) and improving sleep quality significantly improves mental health (Scott et al., 2021). This is an area of public health interest due to the health impact of sleep and mental disorders as well as the fact that they show a strong relationship and are potentially modifiable factors for use in interventions. A study examining age trends in actigraphy and self-report sleep across the lifespan using generalized additive models, found associations between sleep and age (Wallace et al., 2022). While these findings are interesting, their methods focused only on “healthy” people, excluding those with comorbid mental and sleep disorders. Given the association between sleep disturbances and mental disorders, I decided to repeat the analysis, but to include those with psychiatric conditions in the models, examining the association between sleep and mental disorders across the lifespan. Using the Pittsburgh Lifespan Sleep Databank, a comprehensive databank which incorporates both actigraphy and self-report sleep data over most of the lifespan, I ran generalized additive models predicting selected actigraphy and self-report sleep outcomes from a smoothed age/psychiatric condition interaction and relevant covariates. Within the healthy control, self-report sleep duration, sleep onset and offset decreased through middle age. Within those presenting with a psychiatric condition, self-report sleep duration decreased through middle age and then showed a large increase into old age. Actigraphy sleep onset showed a marginally significant linear decrease across the lifespan; self-report sleep onset showed a similar decrease, but did not present in a linear pattern. Sleep offset showed the most nonlinear pattern, oscillating over the lifespan with an overall neutral trend. From these results, we can see that the association between age and sleep differs based on if an individual presents with mental illness, offering an opportunity for future study into why these differences exist and how treating sleep disturbances could lead to better mental health in the long run.