Application of Robust Linear Mixed Model to Study the Relationship between Shift Work and Longitudinal Changes of Cholesterol and Controlling of Influence of Baseline Weight

The association between shift of work and chronic diseases such as cardiovascular diseases and their related mechanisms has been investigated in several studies. The aim of this study is to investigate the relationship between longitudinal change in total cholesterol as a main CVD risk factor and the shift of work، with controlling of the effect of baseline weight at time of recruitment. This retrospective cohort study consists of 674 employees of Polyacryl Iran Corporation (PIC) of Isfahan from 1992 to 2009. The relationship between shift of work and cholesterol based on weight status at baseline of recruitment with the effect of confounding factors including age، BMI، pre-employment cholesterol، glucose، triglyceride، urea، work types، education، and marital status was studied. A linear mixed model with skew-elliptical errors was used for analysing the data. Parameter estimation was done using Bayesian approach using WinBUGS statistical software. Bayesian confidence interval was used for testing regression coefficients. Linear mixed model with skew-normal error was better compared to other models. In this model the relationship between shift of work and cholesterol changes with controlling of confounding factors was significant in those who were not overweight at baseline (β = 2. 25، P < 0. 001، 95% CI = 0. 67، 3. 88)، but was not significant for those who were overweight at baseline of employment. By increasing of age، the mean rate of cholesterol changes was significantly higher in shift workers who had normal weight at baseline of recruitment compared with shift workers who were over-weight at baseline of recruitment. Furthermore، linear mixed model with heavy tailed and skewed error indicate better fitting than the standard linear mixed model.