Strengths and Limitations
The main strength of the study was the sizable long-term population-based patient group included in our study. The advantages of using the NHIRD such as long-term comprehensive follow-up and universal coverage has been described in a previous study.(29) We excluded patients with previous thyroid diseases due to possible interplay of a history of thyroid diseases that might have caused Graves’ diseases. We also qualified the definition of PCOS events in our study to more than twice outpatient or one hospitalized record. Generally, our database size ensures similar distributions due to well-balanced matching, and reduces the heterogeneity and selection bias.
There are some limitations to our study. Since PCOS and autoimmune thyroid diseases, including Graves’ diseases, have a familial clustering phenomenon, an alternative explanation may be suggested to a common genetic pathway.(30) However, the genetic background of individuals was not available in our NHIRD scheme. The development of Graves’ diseases may depend on different ethnic groups. Lab data such as hormone level was also not available in NHIRD, posing limitations in further addressing the association of the role of hormones in the association of PCOS and Graves’ diseases. Although the Bureau of NHI uses strict auditing mechanisms to reimburse insurance claims, the ICD-9 codes claimed from the NHIRD might be inaccurate due to diagnostic uncertainty and misclassification in reality. The symptoms of PCOS and Graves’ diseases may sometimes be too mild to be recognized by patients. Thus, the underrepresentation of the identified population may occur and eclipse the accuracy of prevalence.