The study population was derived from the Korean National Health Insurance Service (NHIS) database. In South Korea, the NHIS provides mandatory health insurance for all Korean citizens providing nearly all types of health services (16). Furthermore, the NHIS provides health screening examinations for all enrollees aged 40 years or older, which include a self-reported questionnaire on health-related behaviors, anthropometric measurements, and laboratory exams for blood and urine. The NHIS collects health service utilization information on all insured health services including outpatient and inpatient hospitalizations, health screening examinations, diagnostic and treatment-related procedures, and pharmaceutical prescriptions for claims purposes (17). From this database, the NHIS provides a part of their data for research purposes. A number of previous large-scale epidemiological studies using NHIS database has been reported and its validity is described in detail elsewhere (17, 18).
Among 2,338,093 participants without previous CVD in 7 metropolitan areas in South Korea aged 59 years or older in 2011, we excluded 31,540 participants who died before the index date of 1 January 2012. Then, 1 participant who did not have the value of social trust and 149,723 who had missing values for covariates were excluded. The final study population consisted of 2,156,829 participants. Starting from 1 January 2012, all participants were followed-up for CVD or death until 31 December 2016.
Social trust values were derived from the Korean Community Health Survey (KCHS) (19)., which is conducted by the Korean Centers for Disease Control and Prevention in 2011. KCHS is a nationally-representative community-based cross-sectional survey that contains community-level information according to administrative district sites. Among the survey questions, all participants were asked “Do you trust members of your community?” as a measure of social trust (20). The proportion of those who answered ‘yes’ to the social trust question for each administrative district site was calculated and determined as social trust. A total of 253 district sties, with a mean (standard deviation) land area of 55.1 (79.9) km2, covers the entire South Korea land mass (21). Social trust in 2011 were then merged according to each participants’ area of residence, which is also categorized into the same administrative district site. The study population was then divided into quintiles according to social trust values, with the first quintile containing participants residing in areas with the lowest social trust values.
Upon admission, the attending physician is required to insert the International Classification of Diseases, Tenth Revision (ICD-10) codes for the primary disease in which the patient was hospitalized for. CVD was defined when a participant was hospitalized for ICD-10 codes for coronary heart disease (CHD) or stroke for 2 or more days. The ICD-10 codes for CHD and stroke were I20 to I25 and I60 to I69, respectively. Multiple previous studies that used the NHIS database had a similar operational definition for CVD (18). The ICD-10 codes for CVD, CHD, and stroke were adopted from the American Heart Association guidelines (22).
Upon multivariate Cox proportional hazards regression, we adjusted for potential confounding factors including age (continuous, years), sex (categorical, men or women), area of residence (categorical, capital or metropolitan city), household income (categorical, 1st, 2nd, 3rd, and 4th quartiles), and Charlson comorbidity index (continuous). Household income quartile was determined from the insurance premium. Among participants who underwent health examinations within 2 years prior to the index date, we additionally adjusted for lifestyle behaviors and CVD risk factors including smoking (categorical, never, former, and current smokers), alcohol intake (categorical, 0, 0-1, 1-2, 3-4, and 5 times per week), physical activity (categorical, 0, 1-2, 3-4, 5-6, and 7 times per week), body mass index (continuous, kg/m2), systolic blood pressure (continuous, mmHg), fasting serum glucose (continuous, mg/dL), and total cholesterol (continuous, mg/dL). Body mass index was calculated by dividing the weight in kilograms by height in meters squared.
The Chi-squared test for categorical variables and analysis of variance for continuous variables were used to determine the difference in characteristics of the study population according to quintiles of social trust. Multivariate Cox proportional hazards regression was used to calculate the adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) for CVD risk according to quintiles of social trust. Stratified analysis for the association of social trust with CVD risk according to age, sex, household income, and Charlson comorbidity index was conducted. The risk for CVD according to social trust upon additional adjustment for lifestyle behaviors and CVD risk factors among those who underwent health examinations was determined. Finally, subgroup analysis on the association of social trust with CVD risk using those population who did health screening test, was conducted by stratified analysis according to level of smoking, alcohol consumption, and physical activity was also conducted
Statistical significance was determined as a p value of <0.05 in a two-sided manner. All data collection and statistical analysis were conducted using SAS 9.4 (SAS Institute, Cary, NC, USA).