Data source
The database from the National Health Insurance Sharing Service (NHISS) of the National Health Insurance Service (NHIS) of South Korea was used.(36, 37) The NHIS is a public institution responsible for operating mandatory universal health insurance, and nearly 97% of the South Korean population is enrolled in this service, while the remaining 3% is covered by the Medical Aid Program. The NHISS contains medical service claims data such as admissions, emergency room visits, ambulatory care visits, and pharmaceutical services.
The database from the National Cancer Screening Program (NCSP) was also used.(38) The NCSP contains screening data for stomach, liver, colorectal, breast, and cervical cancers, and each screening examination was conducted based on the age of the participants. All South Korean women aged 40 years or older are encouraged to be screened for breast and cervical cancer biennially. Even though the screening program was voluntary, participation rates reached up to 70%.(39)
The data of NHISS and NCSP were anonymized by using individual research numbers instead of social security numbers to protect the privacy of the individuals. The study protocol was approved by the Institutional Review Board of the Samsung Medical Center (IRB No. 2021-03-108).
Case identification
Among 3,109,506 female subjects aged 40 years or older who underwent health examination and screening for breast/cervical cancer on the same day from January 1, 2009, to December 31, 2009, 1,725,502 were identified as postmenopausal without a history of hysterectomy. Those with incomplete information about HRT or OC use (n = 320,845) and those with a previous diagnosis of dementia before the examination (n = 8,003) were excluded to secure the first diagnosis of dementia. In addition, 9,835 subjects who were diagnosed with dementia within one year after the examination were excluded to eliminate the effect of a temporary increase in the diagnosis of dementia by detecting it at the time of examination. After excluding 1,177,231 subjects with no history of depression, 209,588 subjects were deemed eligible for our study, and their medical records were followed until December 31, 2018 (Figure 1). The diagnosis of depression (F32 and F33) was defined based on the International Statistical Classification of Disease and Related Health Problems 10th revision (ICD-10).
Lifetime uses of oral contraceptives
Information on the usage of OC was extracted from self-administered questionnaire data from the cancer screening program. The question was “Are you on or have you ever taken oral contraceptive pills?” Subjects were asked to choose a response among “never”, “use for less than one year”, “use for more than one year”, or “unknown”.
Hormone replacement therapy after menopause
Information on HRT was extracted from the self-administered questionnaire data from a cancer screening program. The question was “Are you on or have you ever taken hormonal agents to relieve postmenopausal symptoms?” Subjects were asked to choose a response among “never”, “use for less than two years”, “use of 2–5 years”, “use for more than five years”, or “unknown”.
Outcomes
The main outcome was the diagnosis of dementia (F00 and F30 for AD; F01 for VD; and F02, F03, and F31 for the other dementia) and the prescription of one or more medications for dementia during the follow-up period. When a subject had more than one code for dementia diagnosis, the subject was classified based on the principal diagnosis. If both AD and VD codes were included in the additional diagnosis of a subject, the subject was classified based on the principal diagnosis of the next hospital visit, and if both AD and VD codes remained as additional diagnosis codes, the subject was classified as another dementia group. Prescribed medications for dementia included donepezil, rivastigmine, galantamine, or memantine.
Covariates
The body mass index (BMI) was calculated using the weight and height of subjects. Lifestyle factors such as smoking, alcohol consumption, and exercise were identified from the NCSP self-questionnaire. Regular exercise was defined as performing a moderate physical activity for more than 30 minutes at least five times a week or vigorous physical activity for more than 20 minutes at least three times a week. Subjects were categorized into levels of income based on the payment of health insurance. Comorbid physical illnesses including hypertension, diabetes mellitus, and dyslipidemia were identified based on ICD-10 codes from past medical records.
Statistical analyses
Continuous variables were displayed as mean ± standard deviation (SD), while categorical variables were displayed as number and percentage. The Student t-test was used to compare differences in individual factors between the groups. Cox proportional hazards regression analyses were conducted to identify the association between hormone therapy and the diagnosis of dementia. The hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) revealed the magnitude of risk of dementia based on duration of hormone therapy. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).