Using Gamma and Quantile Regressions to Explore the Association between Job Strain and Adiposity in the ELSA-Brasil Study: Does Gender Matter?

Int J Environ Res Public Health. 2017 Nov 17;14(11):1404. doi: 10.3390/ijerph14111404.


This paper explores the association between job strain and adiposity, using two statistical analysis approaches and considering the role of gender. The research evaluated 11,960 active baseline participants (2008-2010) in the ELSA-Brasil study. Job strain was evaluated through a demand-control questionnaire, while body mass index (BMI) and waist circumference (WC) were evaluated in continuous form. The associations were estimated using gamma regression models with an identity link function. Quantile regression models were also estimated from the final set of co-variables established by gamma regression. The relationship that was found varied by analytical approach and gender. Among the women, no association was observed between job strain and adiposity in the fitted gamma models. In the quantile models, a pattern of increasing effects of high strain was observed at higher BMI and WC distribution quantiles. Among the men, high strain was associated with adiposity in the gamma regression models. However, when quantile regression was used, that association was found not to be homogeneous across outcome distributions. In addition, in the quantile models an association was observed between active jobs and BMI. Our results point to an association between job strain and adiposity, which follows a heterogeneous pattern. Modelling strategies can produce different results and should, accordingly, be used to complement one another.

Keywords: adiposity; body mass index; job strain; quantile regression models; waist circumference.

MeSH terms

  • Adiposity / physiology*
  • Adult
  • Body Mass Index
  • Brazil
  • Female
  • Humans
  • Male
  • Middle Aged
  • Obesity / etiology*
  • Obesity / physiopathology*
  • Occupational Stress / complications*
  • Occupations / statistics & numerical data*
  • Regression Analysis
  • Sex Factors
  • Surveys and Questionnaires
  • Waist Circumference