Background: Successful ageing is the term often used for depicting exceptional ageing and can be measured with multidimensional models including physical, psychological and social wellbeing. The aim of this study was to test multidimensional successful ageing models to investigate whether these models can predict successful ageing, and which individual subcomponents included in the models are most significantly associated with successful ageing.

Methods: Successful ageing was defined as the ability to live at home without daily care at the age of 84 years or over. Data on the participants' physical, psychological and social wellbeing were gathered at baseline and the follow-up period was 20 years. Four successful ageing models were constructed. Backward stepwise logistic regression analysis was used to identify the individual subcomponents of the models which best predicted successful ageing.

Results: All successful ageing models were able to predict ageing successfully after the 20-year follow-up period. After the backward stepwise logistic regression analysis, three individual subcomponents of four models remained statistically significant and were included in the new model: having no heart disease, having good self-rated health and feeling useful. As a model, using only these three subcomponents, the association with successful ageing was similar to using the full models.

Conclusions: Multidimensional successful ageing models were able to predict successful ageing after a 20-year follow-up period. However, according to the backward stepwise logistic regression analysis, the three subcomponents (absence of heart disease, good self-rated health and feeling useful) significantly associated with successful ageing performed as well as the multidimensional successful ageing models in predicting ageing successfully.