Study Population

This retrospective cohort study was approved by the Catholic University of Korean Institutional Review Board (IRB number: VC16EISI0022). The requirement for informed consent was waived, as the NHIS anonymized all participants according to strict confidentiality guidelines prior to distribution. This study was performed in accordance with the principles of the Declaration of Helsinki and all relevant laws in South Korea.

The study population was derived from the Korean National Health Insurance Service–Health Screening Cohort (NHIS-HEALS). In South Korea, nearly all citizens are covered for health insurance by the NHIS, resulting in an enrollment rate of 97%. The NHIS provides mandatory health insurance to all citizens covering nearly all forms of healthcare services. Furthermore, all citizens aged 40 years or older can undergo a national health screening examination provided by the NHIS biannually. The NHIS collects and maintains information on all forms of healthcare services including sociodemographic information on all enrollees, inpatient and outpatient hospital visits, pharmaceutical prescriptions, and results from national health screening examinations. From this nationwide claims data, the NHIS provides a part of their database for research purposes called the NHIS-HEALS. Multiple epidemiologic studies have used the NHIS database and its validity is described in detail elsewhere. ^,

After excluding participants diagnosed with AMD during 2002, we determined whether each study participant was diagnosed with AMD during 2003-2005. Starting from the index date of January 1, 2006, all participants were followed up for development of AD or PD according to AMD diagnosis until December 31, 2013. Among 335,109 participants aged 50 years or older, we excluded 1,202 AMD patients diagnosed during 2002. We also excluded 20,687 and 2,504 participants with missing values on covariates and participants diagnosed with AD or PD prior to the index date, respectively. Finally, we excluded 2,376 participants who died before the index date, resulting in a final study population of 308,340 participants.

Key Variables

Operational definitions for AMD, AD, and PD were adopted from previous studies. Using the hospital visit records, AMD was defined as patients having visited the ophthalmologic outpatient department at least twice or having been hospitalized under the diagnosis code for AMD. The International Classification of Diseases, Tenth Revision (ICD-10) codes were used, in which AMD was defined as H35.3. AD was defined as being prescribed antidementia drugs under the ICD-10 code for AD (F00, G30). The considered antidementia drugs included donepezil, galantamine, rivastigmine, and memantine. Finally, PD was defined as patients having visited the outpatient department at least twice or having been hospitalized under the ICD-10 code for PD (G20).

Upon multivariate Cox proportional hazards regression, we included variables that could be potential confounders. The considered covariates included age (continuous, years), sex (categorical, men and women), household income (categorical, first, second, third, and fourth quartiles), smoking (categorical, never, past, and current smokers), alcohol consumption (categorical, none, 0-1, 1-2, 3-4, and 5 or more times per week), physical activity (categorical, none, 1-2, 3-4, 5-6, and 7 times per week), body mass index (continuous, kg/m ² ), systolic blood pressure (continuous, mm Hg), fasting serum glucose (continuous, mg/dL), total cholesterol (continuous, mg/dL), and Charlson comorbidity index (continuous). Household income was derived from the insurance premium and body mass index was calculated by the weight in kilograms divided by the height in meters squared.

Statistical Analysis

In order to determine the differences in descriptive characteristics between participants with and without AMD, we used the χ ² test for categorical variables and t test for continuous variables. We used Cox proportional hazards regression to obtain adjusted hazards ratios (aHRs) and 95% confidence intervals (CIs) for AD and PD risk according to diagnosis of AMD. We conducted a sensitivity analysis on the association of AMD with PD and PD among participants with lifestyle behaviors. Participants with lifestyle behaviors included never-smokers, non–alcohol drinkers, participants who underwent regular exercise, and those with all of the above. We also conducted a sensitivity analysis by excluding participants who were diagnosed with AD or PD within the first 1, 2, and 3 years of follow-up. We conducted a stratified analysis on the association of AMD with AD and PD according to age groups of less 70 and 70 years or more. Finally, the risk for AD or PD according to AMD with additional adjustments for smoking, alcohol intake, and physical activity measured within the first 2 years of follow-up among those who underwent health examinations during that period (n = 196,854) was determined.

Statistical significance was considered at a P value of <.05 in a 2-sided manner. All data collection and statistical analyses were conducted using SAS 9.4 (SAS Institute Inc, Cary, North Carolina, USA).