Study population
This study used the NHIS–NSC database (NHIS-2018-2-290), a population-based cohort database established by the NHIS in South Korea [10]. This is the national representative cohort database for health service use, in which approximately 1,025,340 patients (2.2% of 46,605,433 Korean residents in 2002) were followed up until 2013 by annually updating samples of newborn infants. From this database, all patients aged ≥20 years in 2009 were identified. Patients without health check-up data or those with a history of ischemic heart disease before their enrollment were excluded (Figure 1).
Exposure measures
Community-level SES of study subjects was defined as the local income for the residential area in which they lived in 2009. The gross regional domestic product (GRDP) per capita of 16 regions (seven metropolitan cities, including the Korean capital, and nine provinces) was used to measure the local income of subjects’ residential area and ranked according to the GRDP [11]. Local income was then classified into three categories according to the ranking of GDPR per capita as low (ranks 12-16), moderate (9-11), and high (1-8) (Table 1). Each of the three categories contained a different number of regions as the total population was divided evenly into three categories.
Outcome measures
The outcome measure of interest was IHD, defined as according to the International Classification of Disease, 10th Revision (ICD-10) codes I20, I21, I22, I23, I24, and I25 [12]. Follow-up of all patients began on January 1, 2009, and ended when any of the following occurred: onset of ischemic heart disease, death from any cause, moving to a different region at baseline, and the end of the study period (December 31, 2013).
Confounding variables
Confounding variables evaluated included patients’ age, sex, individual economic status, smoking status, body mass index (BMI) and the incidence of comorbidities including diabetes mellitus (DM), hypertension (HTN), dyslipidemia, peripheral arterial disease (PAD), and stroke at baseline. The people included in the NHIS are ranked into 21 categories on the NHIS-NSC database according to the insurance premiums that they pay. The NHIS calculates individual insurance premiums through consideration of income, assets, standard of living, and other economic factors. In our statistical modelling, individual economic status was evaluated as the average premium value for the insurance premiums in each ranks of NHIS. The history of disease was defined as follows: DM (ICD-10 E11, E12, E13, E14), HTN (I10, I11, I12, I13, I15), dyslipidemia (ICD-10 E78), PAD (ICD-10 I70.0, I70.2, I73.9, I70.8, I70.9, I74.2, I74.3, I74.4, I74.5), and stroke (ICD-10 I60, I61, I63).
Statistical analyses
Data are presented as means (standard deviation, SD) for continuous variables and as numbers (n) and percentages (%) for categorical variables. Demographic and clinical characteristics among the regional income group were compared using the chi-square test or ANOVA, as appropriate. The incidence rate per 1000 person-years and cumulative incidence for IHD were calculated in each group. To evaluate the association between the risk of IHD and regional income level, Cox regression models with mixed effect (“Frailty model”) were used. This model incorporates region-specific random effects to account for within-region homogeneity in outcomes [13]. Hazard ratios and 95% confidence intervals were presented, and the high-income group was considered the reference group. Adjusted hazard ratios were obtained from the model including regions as random effect and age, sex, smoking, BMI, individual economic status, history of DM, HTN, dyslipidemia, PAD, and stroke as covariates. All statistical analyses were performed using SAS version 9.4 software (SAS Institute Inc., Cary, NC, USA), and a two-sided P-value<0.05 was considered statistically significant.