Long-Term Statin Use and the Risk of Parkinson’s Disease | Page 2
Published Online: August 12, 2013
Bitya Friedman, MD; Amnon Lahad, MD, MPH; Yizchak Dresner, MD, MS; and Shlomo Vinker, MD
Exposure was defined as statin purchase as recorded in the computerized database between the years of 2001 and 2005. Statins purchased were simvastatin (93%), pravastatin, atorvastatin, and rosuvastatin. The level of exposure was calculated as the number of months of statin use before the diagnosis of PD. Statin use was considered chronic if at least 6 monthly prescriptions were dispensed over a period of 9 consecutive months. Purchase of fewer than 6 monthly prescriptions was not considered as exposure.
Outcome was defined as incident of PD as indicated by a record of at least 2 monthly prescriptions for anti-parkinsonian drugs. These drugs are used specifically for parkinsonian symptoms, primary or secondary, while the medical diagnoses in computerized patient files are not always as reliable as the central register.
In order to validate the diagnosis, we reviewed manually 500 medical files of patients identified from the computerized database as having newly diagnosed PD according to the criteria mentioned above. Diagnosis of PD was validated if a clinical diagnosis of PD was recorded in the file at the time that anti-parkinsonian medication was instituted. The diagnosis was validated in more than 90% of the files.
A number of independent confounding variables were included in the statistical analysis, based on previous studies linking them to PD: (1) patient sex; (2) low socioeconomic status (defined as exempted by social security from copayments); (3) history of diabetes, ischemic heart disease, hypertension, or previous CVA; and (4) smoking status as regularly recorded in the electronic files. Due to a high rate of missing data, body mass index was not included in the statistical analysis.
The analysis was conducted in 2 stages. In the first stage, we used the Adult Treatment Panel III classification of cholesterol levels to divide the whole cohort into 4 groups according to baseline LDL-C levels.29 We determined the incidence of PD in the different groups and compared relative risks, adjusting for the potential confounders. In the second stage, we compared the PD incidence among statin users with that among nonusers, adjusting for confounders and for baseline LDL-C levels. The association between the different variables was analyzed by using Cox proportional hazards models.30
The cohort consisted of 94,308 subjects over the age of 45 years, of whom 52.8% were female and 22.3% were defined as having low socioeconomic status (Table 3). The mean age at the end of the study period was 66.90 years (standard deviation, 10.85 years).
A total of 29,714 participants (31.5% of the cohort) started use of statins for a minimum of 6 months during the study period. During the study period, 1035 patients were diagnosed with PD. The incidence was 1.1% (1035/94,308), or 15.7 cases per 100,000 person-years. The mean age of individuals with newly diagnosed PD was 73.69 years (standard deviation, 9.0 years) at end of the study period.We compared PD incidence at different baseline LDL-C levels. The cohort at this stage included 87,971 subjects (Table 4), after omitting 6205 cases (6.6% of the cohort) where LDL-C was first recorded after initiation of statin treatment. The number of subjects with newly diagnosed PD was 824.
There was no statistically significant difference in PD incidence among the different LDL-C quartiles (Table 5). A number of variables were found to be associated with an increase in PD incidence: Male sex (odds ratio [OR],1.57; 95% confidence interval [CI], 1.36-1.81; P <.001), previous CVA (OR, 1.97; 95% CI, 1.68-2.29; P <.001) and low socioeconomic status (OR, 1.33; 95% CI, 1.14-1.54; P <.001). Smoking was not associated with a change in PD incidence (OR, 0.93; 95% CI, 0.75-1.14; P = .4).
Next, we added statin use as a variable in the regression model (Table 5). Statin use was found to be associated with a significant decrease in the incidence of PD (OR, 0.73; 95% CI, 0.60-0.88; P = .001). The association of the other variables with PD incidence was not affected.
After finding a significant association between the exposure (statin use) and the outcome (PD incidence), we looked for a dose-related effect.
We defined 4 categories of statin use according to number of treatment years: none, up to 2.5 years, 2.5 to 4 years, and more than 4 years. After adding these variables to the regression model, we observed a continuing trend of decreased PD incidence associated with statin use. However, the association was not statistically significant beyond 2.5 years of treatment (Table 5).
The wide use of statins in prevention of cardiovascular morbidity warrants a continuous search for possible adverse effects associated with this group of medications.
A few studies over the past years have examined the possible association between statin use and PD. However, their findings vary widely, as discussed in the introduction. Our historical prospective cohort study was designed to overcome some of the weaknesses of previous studies. One advantage was that it enabled us to show a temporal association between the exposure (cholesterol level and statin use) and the outcome (a new diagnosis of PD). Another advantage of choosing a cohort design was reduction of selection bias. A population-based study based on data from unselected medical files from primary care clinics reflects the epidemiology of the disease more accurately than casecontrol studies.
One of the main advantages of this study is its use of objective data, without reliance on patient-reported information or even physician-reported diagnosis. All information— laboratory tests, medical diagnoses, and drug purchase—was obtained from a computerized database. In this way we minimized potential information bias.
The statistical power of this study is high due to the large size of the cohort and the number of patients with incident PD, which is the largest number reported in any prospective study to date. There have been studies with larger cohorts and longer periods of observation (eg, Simon and colleaues14); however, the number of incident cases of PD was smaller, probably due to the younger ages of subjects included and the larger proportion of women.
In our study we were able to show that long-term statin use was associated strongly with a significant decrease in PD incidence (OR, 0.73; P = .001). No association was found between baseline LDL-C levels and risk of PD.
PDF is available on the last page.