A biological, latent variable model of health (EarlyBird 68)

Brain Behav Immun. 2014 Aug;40:104-9. doi: 10.1016/j.bbi.2014.02.018. Epub 2014 Mar 6.

Abstract

Objective: To determine whether factor analysis of a set of health-related biomarkers provides evidence of an underlying common dimension of variation, and to explore the relationship between this dimension of variation with positive and negative affect.

Method: Twelve health-related metabolic, immune and body-composition biomarkers at ages 5, 7, 9, 11, 14 and 16years were obtained from the EarlyBird longitudinal cohort of 347 children and supplemented by positive affect (PA) and negative affect (NA) measured at age 16years.

Results: At each age, principal factor analysis revealed that nine of the 12 biomarkers consistently loaded on the first extracted factor, accounting for 25% of the variance at age 5, and 37-44% of the variance at 7-16years. High loading biomarkers included physical indicators of adiposity, insulin resistance, C-reactive protein, triglycerides, and cholesterol. Factor scores at different ages correlated between .48 and .85. Correlations between the first factor scores and mood measured at age 16 were r=-.17 (p=.02) for PA and r=.13 (p=.07) for NA.

Conclusions: There is a latent variable, h, that accounts for about a third of the variance of a set of health related physical and biochemical biomarkers. h is comparatively stable during childhood and is a weak predictor of mood. These data provide a rationale for aggregating biomarkers in psychoneuroimmunological research. The concept of h provides a possible biological rationale for the role of common factors in disease onset and progression, mental illness, and functional disorders.

Keywords: Biomarkers; Factor analysis; Health; Model; Mood.

MeSH terms

  • Adolescent
  • Biomarkers*
  • Child
  • Child, Preschool
  • Factor Analysis, Statistical
  • Female
  • Health Status*
  • Health*
  • Humans
  • Longitudinal Studies
  • Male
  • Models, Statistical*
  • Prospective Studies

Substances

  • Biomarkers