Background Identification has been a problem with the Age-Period-Cohort analysis. Since Age + Cohort = Period, there is no unique solution using generalised linear modelling. To overcome this problem of perfect collinearity we propose to use partial least squares regression (PLSR), a dimension-reduction technique widely used in bioinformatics. Data from a large Taiwanese cohort was used to illustrate our approach.

Methods PLSR is a set of algorithms that aims to maximise the covariance between outcome and successively extracted orthogonal components under the constraint that the sum of squared weights is equal to unity. To assess the impact of age, birth year and year of examination on the levels of metabolic syndrome (MetS) components, we used PLSR to analyse data collected by Mei-Jaw clinics in Taiwan in years 1996 and 2006. Confounders, such as the number of years in formal education, alcohol intake, smoking history status, and betel-nut chewing were adjusted for.

Results As the age of individuals increased, the values of components generally increased. People born after 1970 had lower fasting plasma glucose, lower body mass index, lower diastolic blood pressure, lower triglycerides, lower low-density cholesterol lipids and greater high-density cholesterol lipids. A similar pattern between the trend in levels of metabolic syndrome components against birth year of birth and economic growth in Taiwan were also found.

Conclusions Our study found cohort effects in some MetS components, suggesting associations between the changing environment and health outcomes in later life. PLSR provides a flexible analytical strategy for the Age-Period-Cohort analysis.