Objectives: To test whether deferred retirement is associated with delayed onset of Alzheimer's disease (AD), and, if so, to determine whether retirement age still predicts the age at onset of AD when two potential biases are considered.
Methods: The study sample was gathered from the Impact of Cholinergic Treatment Use/Data Sharing Alzheimer cohort (ICTUS/DSA), a European study of 1,380 AD patients. Information regarding retirement age, onset of symptoms and covariates was collected at baseline whereas age at diagnosis was gathered from the patient's medical record prior to study entry. Linear mixed models, adjusted for gender, education, occupation, center, country, household income, depression and cardiovascular risk factors were conducted on 815 patients.
Results: (1) The global analyses (n = 815) revealed that later age at retirement was associated with later age at diagnosis (β = 0.31, p < 0.0001); (2) once the selection bias was considered (n = 637), results showed that this association was weaker but remained significant (β = 0.15, p = 0.004); (3) once the bias of the reverse causality (i.e., the possibility that subjects may have left the workforce due to prior cognitive impairment) was considered (n = 447), the effect was no longer significant (β = 0.06, p = 0.18).
Conclusion: The present study supports that there is an association between retirement age and age at onset of AD. However, the strength of this association appears to be overestimated due to the selection bias. Moreover, the causality issue remains unresolved. Further prospective investigations are mandatory in order to correctly address this question.