This population-based, retrospective claim database study utilized the Longitudinal Health and Welfare Database (LHWD) provided by the Health and Welfare Data Science Center (HWDC) of Taiwan. This sampled database is derived from Taiwan’s National Health Insurance database, which covers 99.9% of Taiwan’s population as of 201411. The HWDC established and verified the sampled database.
Study design and patient selection criteria
The study period was from 2001 to 2014, and our patient enrollment period was from 2001 to 2009. The study flowchart is depicted in Figure 1. The International Classification of Diseases, ninth revision (ICD-9) code 571.X was used for the confirmation of CHD diagnoses, and codes 362.50, 362.51, and 362.52 were used for the diagnosis of AMD. Patients diagnosed as having CHDs twice from January 1, 2001 to December 31, 2009 were enrolled. Those aged < 40 years were excluded. The diagnosis date was defined as the date of the first diagnosis of CHDs, and the index date was defined as the date of the first recorded use of silymarin for Study Group I.
Among the patients diagnosed as having CHDs, Study Group I included patients with silymarin use and excluded patients who had used silymarin before the diagnosis date, those who had used silymarin for fewer than 90 days within 1 year after index date, those who had AMD before the diagnosis date, and those who were diagnosed as having AMD within 1 year after the index date. Among patients diagnosed as having CHDs who did not use silymarin, those diagnosed as having AMD before the diagnosis date were excluded, and those remaining were grouped into Study Group II.
From patients without CHDs or silymarin use, we randomly selected a group of patients and assigned them to the Comparison Group, which was 1:1 matched with Study Group II by age, sex, and year of diagnosis date. Because the patients in the Comparison Group were not diagnosed as having CHDs, the date used for matching was called the reference date in the Comparison Group. Patients from the Comparison Group who were diagnosed as having AMD before the reference date were excluded.
This study was approved by Ethics Committee of Taipei Medical University (TMU-JIRB, No.: N201610012). Because all NHIRD data are delinked and deidentified, the requirement for obtaining informed consent was waived. We confirm that all methods were performed in accordance with the relevant guidelines and regulations and the study had been performed in accordance with the Declaration of Helsinki.
Risk factors and confounding factors
The potential risk factors for AMD, including hypertension, diabetes mellitus, hyperlipidemia, cerebrovascular diseases, coronary artery diseases, heart failure, peripheral vascular diseases, gastrointestinal disorders, dementia, obesity, tobacco use disorder, and alcohol abuse were identified in the patients. The potential confounding factors of medication use, including that of lipid-lowering agents (statin, fibrate), antihypertensive agents (angiotensin-converting-enzyme inhibitors, angiotensin II receptor blockers, calcium channel blockers), and aspirin, were also evaluated.
Because we aimed to investigate the long-term use of silymarin in patients with CHDs, only those who had used silymarin for more than 90 days per year were included. We evaluated the occurrence of AMD in each group as the primary outcome. For the evaluation of the dose-response as the secondary outcome, the Anatomical Therapeutic Chemical (ATC)/Defined Daily Dose (DDD) system was used12. Owing to lack of data regarding silymarin DDD in World Health Organization suggestions, we set its DDD to 420 mg according to the package insert13.
Among the different patient groups, means with standard deviations were used to describe age, and proportions were used to describe all other variables (including sex, comorbidities, and co-medication). Student’s t test and Pearson’s chi-squared test were performed to evaluate the differences in all the other baseline characteristics. The Cox proportional hazards regression model was used to analyze the risk of AMD between the groups. The potential risk factors mentioned in the previous section were included in the model.
Two-tailed hypothesis tests were used in all analyses, and p < 0.05 indicated statistical significance. SAS (version 9.4; SAS Institute, Cary, NC, USA) was used for all statistical analyses.