Confounding of the association between statins and Parkinson disease: systematic review and meta-analysis.

PURPOSE Both statins and higher serum cholesterol have been reported to reduce to risk of Parkinson Disease (PD). Given the importance of adjusting for cholesterol levels when evaluating the effect of statins, we assessed whether the protective effect of statins would remain when adjustment for cholesterol is performed. METHODS We conducted a systematic review of epidemiologic studies that reported quantitative estimates of the association between statins and incident PD and were published through February 2016. Random-effects meta-analysis was used to assess the effect of statins on PD separately among the studies that adjusted for either cholesterol or hyperlipidemia and studies that did not. RESULTS Ten eligible studies that evaluated the use of statins and incident PD were identified. Among the six studies that did not adjust for cholesterol, a protective effect of statins was observed (relative risk 0.75; 95% confidence intervals (CI) 0.60 to 0.92). Excluding studies with possible bias because of reverse causation or stratifying on study design did not affect the results. No protective effect was observed among the four studies that adjusted for either cholesterol of hyperlipidemia (relative risk 0.91; 95%CI 0.68 to 1.22). The effect estimate for studies that adjusted for cholesterol was 1.04 (95%CI 0.68 to 1.59) when a study with immortal time bias was excluded. CONCLUSIONS The apparent protective effect of statins on risk of PD is at least partially explained by confounding by statin indication. Immortal time bias and healthy user effects also could have contributed to biased results. Copyright © 2016 John Wiley & Sons, Ltd.