By creating subgroups of statin users and non-users with comparable cardiovascular risk profile and subgroups with comparable propensity score for statin use we aimed to minimize this bias. We do not assume any bias induced by selective drop-out of statin users who refused BAY 43-9006 customer reviews cognitive testing, as statin users and non-users were equally represented among participants with complete and incomplete cognitive data. Third, the primary outcome measure was based upon a single cognitive test, mainly investigating executive functions controlled by the frontal lobe. We have to acknowledge that our cognitive tests, like other measurements of cognition, Trail-Making Test or Modified Telephone Interview for Cognitive Status are relatively rough measurements of cognition which may not be sensitive enough to detect changes in cognition apparent to the patient. However, or floor effect and thereby more sensitive to subtle changes in cognitive performance in both young and old persons as compared to the MMSE, TMT or TICS-M. Moreover, the main findings were confirmed using performance on the VAT. Fourth, the use of a computer database is an imperfect measure of adherence to statin therapy. Nevertheless, we think that pharmacy-based data are more reliable than self-reported statin therapy as used in previous studies. Finally, the PREVEND cohort is enriched for elevated albuminuria which could induce selection bias, as albuminuria is a risk factor for CVD.20 However, a sensitivity analysis in a subsample representative for the general population did not change results. Therefore, the generalizability of our data is probably well preserved. Notable other strengths of our study are that participants were well phenotyped with respect to cardiovascular risk and cognitive performance, the long follow-up of statin use and the detailed data on statin use obtained from a computerized pharmacy database. Previous studies used dichotomized or self-reported statin use as main determinant and had shorter durations of follow-up, although it could be argued that follow-up should be even longer than in our study as clinically relevant cognitive dysfunction might not take several years but several decades to develop. Finally, to our knowledge, we are the first to report on an in-depth analysis of statin use and cognitive function in a large population-based cohort. In conclusion, this large population-based cohort, statin use was not independently associated with better cognitive function. Statin users with long duration of use or high doses of statins had a similar cognitive performance as non-users. There was no difference in results in persons with either low or high cardiovascular risk, or in older versus younger subjects. Our findings add to the current knowledge that neither early-life nor long exposure to statins is associated with preserved cognitive function. In our opinion, there is no support for a relevant therapeutic benefit of statin use on cognitive function. Cardiovascular disease is the leading cause of death in Western countries.