Data source
We employed the Korea National Health Insurance Service-National Sample Cohort (NHIS-NSC). NHIS-NSC is a representative population-based, prospectively gathered database from the Republic of Korea, spanning from January 1, 2002 to December 31, 201913. The Korean National Health Insurance (NHI) system operates as a single-insurer model that provides universal healthcare coverage with all citizens and healthcare providers. NHIS-NSC contained 1,024,340 participants (2.2% of the Korean population). Systematic stratified random sampling with proportional allocation was applied during participant selection. The study cohort contains several variables: health insurance classifications (including the NHI and Medical Aid Program), sociodemographic variables, mortality records with causes of death, prescription data, and diagnostic information. Participant follow-up continued until December 31, 2019, unless NHI disqualification (such as emigration) or death occurred.
The Institutional Review Board (IRB) of Severance Hospital in Seoul, Republic of Korea (IRB No: [4-2024-0367]) approved this study. The requirement for obtaining written informed consent was waived.
Study design and participants
We conducted a nested case-control study involving 92,444 participants diagnosed with BPH or AGA. To ensure accuracy, we excluded individuals who were potentially misdiagnosed or who obtained prescriptions through illegal proxies, and we included only individuals who had visited outpatient clinics at least three times or were admitted with these diagnoses (Fig. 1). Participants who did not develop BPH or AGA for at least 2 years after cohort enrollment were also excluded to focus on newly diagnosed cases (2-year washout period). Additionally, we excluded individuals diagnosed with BPH or AGA before the age of 40 years to focus our analysis on older adults. Participants covered by the Medical Aid Program were excluded. Furthermore, stringent inclusion criteria were applied, excluding individuals without at least a 30-day prescription of 5-ARI or AB. Eligible participants were followed up from the initial diagnosis date of BPH or AGA until death, disqualification from the NHI, or the end of the observation period on December 31, 2019, whichever occurred first.
Identification of cases and controls
Patients who died were classified as cases. The causes of death were recorded based on the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10). The follow-up period spanned from the diagnosis of BPH or AGA to the date of death. The index date was set as the date of death.
Participants who did not die during the follow-up period were defined as controls. Matching of cases with controls was conducted among patients with BPH or AGA who were alive at the time of matching. Replacement of matched controls was permitted. The case could serve as its own control. Matching was conducted based on several attributes corresponding to the cases, including age (± 1 year), sex (male and female), the duration of follow-up (exact), and date of the first BPH or AGA diagnosis (± 30 days). We excluded individuals with follow-up durations of less than 1 year to focus on the long-term association between 5-ARI intake and mortality. The index date for the controls was the same as that for the matched cases. Participants with missing data were excluded. Following the application of these criteria, controls were randomly chosen among the matched controls at a 1:5 ratio, and any controls lacking corresponding cases were excluded. A total of 3,084 patients and 14,630 controls were included in this study. Among the 3,084 cases, 65 patients were matched at a 1:1 ratio, 85 at a 1:2 ratio, 81 at a 1:3 ratio, and 113 at a 1:4 ratio.
Exposure
The cumulative use of 5-ARIs, specifically dutasteride and finasteride, was the exposure variable. The cumulative defined daily doses (cDDDs) established by the World Health Organization were used to quantify use of 5-ARI. Prescriptions issued during hospital admissions and outpatient visits from the first 5-ARI prescription date to the index date, irrespective of the reasons for 5-ARI intake, were included in the cDDD calculations. Participants who had consumed 5-ARIs beyond the minimal significant prescription threshold (i.e., 30 cDDDs) were classified as ever-users.
The cumulative dose of 5-ARIs was evaluated in two ways: (1) the absolute cumulative dose and (2) the average cumulative dose over the follow-up period. The absolute cumulative dose throughout the follow-up period was classified into six groups: <365 cDDDs, 365–730 cDDDs, 730–1,460 cDDDs, 1,460–2,920 cDDDs, 2,920–5,840 cDDDs, and > 5,840 cDDDs. Dose classification was also conducted based on quartiles.
We further investigated the association between the average cumulative dose per year (calculated by dividing the cumulative dose by the follow-up period and multiplying by 365.25 days) and mortality. The mean absolute cumulative dose among all participants was 2,681 cDDDs with a standard deviation (SD) of 7,923 cDDDs, and the mean average cumulative dose per year was 491 cDDDs/year with a SD of 1,471 cDDDs/year.
Outcomes
All-cause and cause-specific mortality were the primary outcomes. The deaths of the participants and their causes, categorized according to the ICD-10 guidelines, were identified from the Statistics Korea Death Database. Cause-specific mortality was subdivided based on ICD-10 into categories including all types of cancers, CVDs such as stroke and acute myocardial infarction, completed suicides (intentional self-harm), chronic obstructive pulmonary disease (COPD), physical trauma (e.g., car accidents, falls), diabetes mellitus (DM), pneumonia, infections, senility, and other causes not listed above. The ICD-10 codes used for the cause-of-death analysis are provided in Supplementary Table 1.
Covariates
The socioeconomic characteristics at the index date included the economic status (classified by NHI premium amount: low, middle, and high), area of residence (i.e., capital area, metropolitan area, and province), and NHI insurance type (i.e., employee, and dependents of employees, self-employed, dependents of self-employed). Health status indicators at the index date included physical activity level (i.e., inactive; insufficiently active, less than 600 metabolic equivalents of task [MET] minutes per week; sufficiently active, more than 600 MET-min/week)14,15, alcohol consumption (i.e., social drinker, non-drinker and consuming alcohol less than once a week; current drinker, consuming alcohol more than once a week), smoking status (i.e., non-smoker, past smoker, and current smoker), body mass index, annual outpatient visits, and number of hospital admissions. Additionally, previous diagnoses such as chronic kidney disease, hypertension, ischemic heart disease, chronic liver disease, congestive heart failure, DM, cancer, stroke, and any psychiatric disorder were accounted for to adjust for comorbidities. Treatment for BPH included ABs, such as alfuzosin, terazosin, doxazosin, tamsulosin, silodosin, and naftopidil, prescribed for at least 30 days, and any type of surgical intervention for the prostate. All major ingredient codes of medications and ICD-10 diagnostic codes used in the analyses are provided in Supplementary Table 1.
Statistical analysis
A conditional logistic regression model was used to explore the association between 5-ARI intake and mortality. With patients treated as self-referent controls using replacements, unbiased estimations of hazard ratios (HRs) were achieved by using odds ratios derived from conditional logistic regression analyses16–18. The crude HRs with 95% confidence intervals (CIs) and aHRs (adjusted for all covariates) with 95% CIs were also calculated. Trend analyses (P-for-trend) were conducted, considering the cumulative 5-ARI dose categories as ordinal variables, to examine potential dose-response associations within the logistic model19. Subgroup analyses based on age group (below 50s, 60s and 70s, and ≥ 80 years), and type of 5-ARI (i.e., dutasteride or finasteride) were conducted. Although not a matching variable, subgroup analyses were also performed based on the cumulative AB doses during the follow-up period. There was no evidence of multicollinearity as maximum variance inflation factor was 1.28.
Sensitivity analyses were conducted taking into account the alternative minimum cumulative 5-ARI dose criteria (0 cDDDs, 180 cDDDs), the various exclusion criteria regarding the minimum duration between a BPH or AGA diagnosis and death date (no washout period, follow-up duration < 3 years), and different exclusion criteria (any surgical intervention on the prostate or prostate cancer). SAS version 9.4 was utilized for all analyses, and a two-tailed P-value < .05 was deemed statistically significant in all cases.