Study subjects
The source population for this study was women who had been prescribed CGD-CHPs between 1997 and 2007 in a randomly sampled cohort from all NHI female beneficiaries in Taiwan. This research included females from the randomly sampled cohort aged 1 to 99 years as the study subjects, and age was calculated by subtracting the subject’s birthday from the 1st of July of each year. The NHI system provides universal health insurance coverage, and covered more than 97% of the Taiwanese population in 2002[9]. Consequently, data from the NHI system are widely used by researchers in various fields[10]. Both CM and Western medicine doctors must follow a standard diagnostic procedure using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) coding system for claiming reimbursement. Because the NHI system of Taiwan does not reimburse the use of CM for inpatient services, we only investigated the use of CM for outpatient services. The NHI reimbursement data files were transformed into a research database entitled the National Health Insurance Research Database (NHIRD) that was maintained by the National Health Research Institutes. In the NHIRD, all the identification information is encrypted but the medical records files are retained, including information from 1997 to 2007 on medical care facilities and specialties, drugs and/or treatment regimens (dosages, dosage frequency and prescription duration), patient sex and date of birth, date of visit, and three major diagnoses coded in the ICD-9-CM format. The NHIRD contains a one-million cohort systematically and randomly selected from the 23 million people in the NHI database who were ever insured under the NHI system in 2005. For each subject, we collected all records of CHPs during 1997 to 2007 as a female cohort. We considered the first diagnosis to be the major diagnosis on the records for outpatient department visits, which were coded in ICD-9-CM and then lumped together into different broader disease categories, e.g., lists of ICD-9-CM codes 710–739 were classified as diseases of the musculoskeletal system and connective tissue, while ICD-9-CM codes 460–519 were classified as diseases of the respiratory system.
List of licensed CHPs
The list of reimbursed CHPs was obtained from the Bureau of the NHI website. The corresponding drug information about specific mixtures or names was obtained from the Committee on Chinese Medicine and Pharmacy (CCMP) website, and included the proportions of each constituent, date, approval period as a drug, code and manufacturers’ names. The CCMP list shows that 10,413 CHPs were licensed during the study period, of which 10,309 CHPs were covered by the NHI system. All CHPs with the same CCMP standard formula were classified under the same category, regardless of slight variations among products of different pharmaceutical companies[11]. For example, there are 62 approved licenses for the formula Ge gen tang containing coumestrol, genistein and/or daidzein.
Selection of herbs and estimation of cumulative doses
The CGD-CHPs examined in this study included Psoraleae Fructus, Puerariae Radix, Sojae Semen Praeparatum, Sophorae Flavescentis Radix and Sophorae Flos. We examined all of the CHPs licensed by the CCMP between 1997 and 2007, including single herbs and herbal formulae, to determine whether they included CGD-CHPs. The original weight of each herb was then calculated according to the contents manufactured by the individual pharmaceutical company registered for each licence and approved by the CCMP.
Statistical analysis
The data were analyzed by descriptive statistics, including the decomposition of the coumestrol, genistein and daidzein contents of the licensed and prescribed CHPs, CGD-CHP prescription rates stratified by age, the medians (with 10th and 90th percentiles) of the cumulative doses of coumestrol, genistein and daidzein, population distribution of patients who had been potentially exposed to coumestrol, genistein and daidzein under various dosages, the frequencies of the disease categories prescribed with CGD-CHPs, the most frequently prescribed herbal formulae containing CGD-CHPs, and so on. All of these analyses were performed using the SAS software package (version 9.1; SAS Inc., Cary, NC, USA).