The importance of secular changes in lifecourse studies: An example from the North West Adelaide Health Study

Abstract

Background: Lifecourse studies typically span a number of years from the time of the exposure factor, to measurement of the outcome. The lengthy time periods covered will often include secular changes in at least some of the variables being considered, which should be accounted for as much as possible in the analysis strategy. Methods: To demonstrate a method of accounting for secular changes in the analysis strategy, data from the North West Adelaide Health Study were used to examine the effect of Kindergarten Union preschool attendance in South Australia on the risk of ever smoking by 34-67 years old. Secular changes in smoking prevalence were accounted for in the analysis by the addition of a birth cohort variable in the multi-variable model and by analysing the data stratified by birth cohort. Results: Preschool participation was associated with a reduced risk of smoking uptake compared with non-participation. On stratified analysis, consistent with the historic secular trends in tobacco smoking, men who went to preschool had a higher risk of smoking uptake than those who did not go to preschool in the oldest cohort followed by a reduction in risk of smoking uptake in the subsequent two birth cohorts. Women who went to preschool had a reduced risk in each birth cohort of being an ever smoker compared with non-participants, which was also consistent with the secular trends in female tobacco smoking. Conclusions: Considering secular changes in the analysis can contribute to a more defensible analysis strategy, coherent interpretation of the data, and also potentially add validity to the study findings.