Latent Class Analysis of Multiple Parameters of the Individual Socioeconomic Context on Sleepiness

Keywords

sleep - sleepiness - socioeconomic status

Introduction: Recent studies highlight the importance of the social, economic, and demographic environment in which individuals live to sleep health. Sleep health has an unequal distribution in the population, especially considering factors such as workload, race, and socioeconomic status. An important aspect of sleep health is sleepiness, and evidence indicates the association between social parameters and excessive daytime sleepiness, which is prevalent in our society.

Aim: To analyze whether latent variables can be extracted from multiple socioeconomic and demographic variables and check if those constructs have any effects on sleepiness in a representative sample from the population of São Paulo.

Methods: Participants from the São Paulo Epidemiologic Sleep Study (EPISONO) – 3rd edition (2007) answered the following questionnaires: Epworth Sleepiness Scale (ESS), Brazil Economic Classification Criteria (CCEB), marital status, household income, working status (worker, student, none or both), number of jobs, weekly work hours, and self-declared race. Latent variables (or classes) were extracted from the questionnaires results using latent class analysis (LCA). The association between ESS and latent classes was tested utilizing generalized linear models (GLM). The statistical significance criterion was set at p < 0.05.

Results: LCA resulted in 3 factors being extracted, with a 74% entropy value, indicating a reasonable accuracy of the class definition. Class 1 had predominantly whites, those who had high CCEB and income, and thus named “high socioeconomic status.” Class 2 evidenced a high percentage of non-working (70%) women (75%) with diverse races and middle to low income and education, and thus named “non-workers.” Class 3 also presented middle to low income, CCEB, and multiple races, but 93% were workers with 40 or more weekly work hours. Thus, Class 3 was named “workers.” GLM verifying the effect of latent classes on ESS resulted in Class 2 showing a single statistically significant association with a statistically significant lowered ESS score.

Conclusions: Latent class analysis resulted in 3 latent classes, “high socioeconomic status,” “non-workers” and “workers.” “Non-workers” showed a negative significant association with sleepiness, suggesting that absence of work combined with middle to low income was related to lower ESS. A measurement model of socioeconomic variables was built, strengthening the hypothesis that the social context is linked to sleep health.

No conflict of interest has been declared by the author(s).

Publication History

Article published online:
15 June 2023