Background: Recorded serum 25(OH)D in survey data varies with observed and unobserved respondent characteristics. The aim of this study was to expose latent population sub-groups and examine variation across groups regarding relationships between serum 25(OH)D and observable characteristics. Methods: This study explored the role of unobserved heterogeneity on associations between surveyed 25(OH)D and various factors using a sample (n = 2,641) extracted from the Saudi Health Interview Survey (2013). Linear regression and finite mixture models (FMM) were estimated and compared. The number of latent classes in the FMM was chosen based on BIC score. Result: Three latent classes were identified. Class I (39.82%), class II (41.03%), and class III (19.15%) with mean 25(OH)D levels of 22.79, 34.88, and 57.45 ng/ml respectively. Distinct patterns of associations with nutrition, behaviour and socio-demographic variables were recorded across classes that were not revealed in pooled linear regression. Conclusion: FMM has the potential to provide additional insights on the relationship between 25(OH)D levels and observable characteristics. It should be more widely considered as a method of investigation in this area.
Bibliographical notePublisher Copyright:
© 2021 Al-Sumaih et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
- Research Article
- Physical sciences
- Biology and life sciences
- Medicine and health sciences
- Research and analysis methods
ASJC Scopus subject areas