In this study, we used big data from a national health insurance database to find trend changes in PCOS and its relationship with EH. Due to the peculiarities of the Korean medical system, big data suitable for use in time-series analyses of disease diagnosis and treatment are available through a unified system, and many studies using those data are being published.
According to our study, the prevalence rates of PCOS in all participants, those younger than 30 years, and those older than 30-years in 2016 were 332.7, 458.4, and 400.4 per 100,00 people, respectively. These are all less than 1%, which is lower than the percentages reported in other studies. The prevalence (estimated) of PCOS among Korean women was 5.8% according to a 2011 study including about 8,000 women of reproductive age . According to a Chinese study that analyzed data from medical examination centers, the prevalence rate of PCOS is 2.2% . Estimates of prevalence from community-based or hospital-based studies are likely to be overestimated in comparison to those based on national data. The national PCOS prevalence in Turkey is 258.5 per 100,000 (0.26%), which is lower than the 6.1% estimated by community-based study [20; 21]. The average national PCOS prevalence in Europe was 276.4 per 100,000 people (0.28%) in 2016, which was lower than our results . We estimated the prevalence of EH in 2016 to be 158.3 per 100,000 people. The prevalence of EH in premenopausal women with abnormal uterine bleeding in Japan is 6.2%, which is higher than the prevalence reported in our study (< 0.2%) and that in a Chinese study of infertile women (3.0%) [22; 23]. The prevalence rate varies depending on race, time of year, and participants.
PCOS has a variety of phenotypes, and the diagnostic criteria are controversial. The National Institute of Health (NIH), the European Society of Human Reproduction and Embryology (ESHRE), and the Androgen Excess Society (AES) presented diagnostic criteria for PCOS in 1990, 2003, and 2006, respectively. These variable diagnostic criteria have an impact on PCOS prevalence. The prevalence of PCOS in a community sample of the Iranian population was 7.1% using the NIH, 11.7% according to the AES, and 14.6% according to the ESHRE criteria . In a study conducted in a government-based laboratory with the largest number of female staff (n = 527) employed by a single laboratory in Ankara, Turkey, the prevalence of PCOS according to NIH, AES, and ESHRE criteria was 6.1%, 15.3%, and 19.9%, respectively . In a Chinese meta-analysis, the prevalence of PCOS was high in the order of those aged between 21 and 30, 10 to 20, 31 to 40, and above 40 years (17.23%, 10.26%, 9.13%, and 2.22%, respectively). Furthermore, the majority of PCOS research included only people younger than 50 years .
We found that when the incidence of PCOS increased sharply, a non-significant increase occurred in the incidence of EH. EH was thought to occur at a relatively constant rate in people diagnosed with PCOS, but our data did not confirm the expected increase in EH when PCOS diagnoses increased rapidly. A rapid increase in PCOS incidence was seen between 2012 and 2016, but the incidence of EH in those years increased only slightly, by 3.4% annually. In 2009 and 2016, the incidence rates (CRs) of PCOS among people in their 30s were 193.9 and 400.4 per 100,000, respectively. That is relatively high compared with the 71.15 and 73.88 cases per 100,000 of PCOS reported among people of reproductive age in East Asia in 2007 and 2017 .
When we examined trend changes by age distribution, we found that those younger than 30 had the highest rate, and those older than 50 had the lowest rate. Those results indicate that PCOS tends to be diagnosed and persist mainly in people in their 20s and 30s.
We found that between 2009 and 2016, 15–30% of patients diagnosed with PCOS received medication to treat the disease, with the highest rate of medication usage reported in people younger than 30 years. Prescription rates for medications that offer endometrial protection increased by 21.4% between 2009 and 2016, whereas those for medications to treat infertility decreased by 21%. Medications to treat excess androgen did not change much, from 0.7–0.3% over seven years. Whereas oral contraceptives play a role in controlling ovulatory amenorrhea, which affects the occurrence of EH, medications for infertility can be used to treat hyperandrogenism.
Our factor analysis examining EH in PCOS patients produced some interesting findings. Previous studies reported a positive association between PCOS and EH, on the one hand, and T1D and T2D, on the other, after adjusting for BMI [27–34]. That association has been reported to be specific to obese women [35–37], and excessive insulin circulating through that mechanism can stimulate hypertrophy of endometrial cells, causing EH [38; 39]. A history of HT has also been suggested as a risk factor for EH in several case-control and cohort studies [30; 40–45]. Similar results were found in our study, with T2D, obesity, HT, HL, and infertility increasing the risk of EH in people who already have PCOS. Additionally, the cumulative duration of oral contraceptive & progestin use for PCOS correlated highly with the development of EH.
A major strength of our study is the inclusion of nationwide population data, which provided evidence of an increase in PCOS prevalence and incidence. Additionally, we found several factors that correlate with the presence of EH in people with PCOS. Nonetheless, this study has a few limitations. First, many people in their 40s and 50s with high disease rates were included in the later years of our study period because we built our dataset by extracting data from people with PCOS and EH disease codes, not from all Korean females. Additionally, our data did not allow us to determine whether PCOS is a risk factor for EH. Second, the follow-up period for the PCOS patient group diagnosed later during the data period was not long enough to adequately observe the progression of the disease. When analyzing the factors influencing the occurrence of EH in PCOS patients, only those diagnosed with PCOS within about 10 years prior to the first occurrence of EH were considered. We did not determine the duration from first PCOS diagnosis to first EH diagnosis. Third, medical practices and behaviors not included in the data might have affected the trend changes we found for PCOS and EH. In addition, we did not fully control for confounding effects because of non-measurable potential confounding factors such as laboratory data or patient-reported outcomes. Fourth, databases containing diagnosis codes for diseases might not record information on the severity of the disease, and changes in diagnostic criteria over time might not be accurately distinguished. In particular, PCOS is a heterogeneous disorder that was mapped to a single ICD code despite its various diagnosis criteria. Fifth, since the disease is defined even if there is only one ICD-10 diagnostic code which may be given for the purpose of examination, the disease classification can be ambiguous. There is no indication as to the outcomes following diagnostic labelling. We considered endometrial hyperplasia and endometrial adenoma hyperplasia as endometrial hyperplasia regardless of severity. Sixth, while we evaluated PCOS treatment based on the prescribed drug code following diagnosis of PCOS, we could not rule out that the drug was treatment of other diseases. Therefore, the results must be interpreted with caution, and more accumulated data are needed to further analyze the relationship between PCOS and EH.