Inequities in cardiovascular risk and lifestyle factors among individuals with developmental disabilities: evidence from Korea’s national health screening program

Data sources

We derived the study population from the National Health Insurance Service (NHIS) database. The NHIS provides mandatory health care coverage for nearly all Koreans (approximately 50 million people), including NHI enrollees (97%) and Medical Aid beneficiaries (3%), including emergency, inpatient, and outpatient care. The NHIS database contains socio-demographic data such as sex, age, residence, health coverage type and NHI contributions and clinical information regarding medical claims. In addition, the database includes disability information from a national disability registration system, which recognizes 15 types of disabilities graded from 1 (most severe) to 6 (least severe) or as moderate (grades 4 to 6) to severe (grades 1 to 3).

NHIS also delivers the National General Health Screening Program (GHSP). Employees and self-employed are eligible every two years for the program (employees involved in manual labor every year) regardless of age, as are dependents aged 20 years or older and Medical Aid beneficiaries aged 20 to 64 years³⁶. The GHSP includes anthropometric data, blood pressure (BP), laboratory examinations including fasting glucose and total cholesterol, and a standardized self-reporting questionnaire regarding medical history and health-related lifestyle factors, including smoking, alcohol consumption, and physical exercise.

This study sampled 10,413,089 participants, 20% of the 2012 population, considering sex, age, and regional distribution from the NHIS database. Enrollee information and GHSP data were collected through 2020 (data number: NHIS-2022-1-629).

Ethical considerations

All study process and methods were performed in accordance with the relevant guidelines and regulations of NHIS and Korea Institute for Health and Social Affairs. Since this study used de-identified data provided by the NHIS after anonymization according to strict confidentiality guidelines, the Institutional Review Board (IRB) of Korea Institute for Health and Social Affairs (IRB number: 2022-004) exempted an ethics review. The IRB recognizes that informed consent is not applicable for the analysis of de-identified data.

Study design and population

We collected 2019 and 2020 GHSP information for individuals aged 20 years or older with and without developmental and intellectual disabilities. Among 17,171,277 individuals (8,554,053 in 2019 and 8,617,224 in 2020), 5,641,005 (5,623,993 without and 17,012 with disabilities) participated in GHSP. We matched participants with developmental or intellectual disabilities controls by age (± 1-year band) and sex. We then matched participants in a 1:1 ratio using propensity score.

Definition of the study population

In this study, we defined the target population based on the Welfare Law for Persons with Disabilities as those certificated as having developmental disabilities (intellectual disability, ASD) in the registration system³⁷. Intellectual disability denotes impaired intellectual development without ASD, typically defined by an intelligence quotient (IQ) of 70 or lower, presenting clear limitations in intellectual and cognitive abilities. Autism, often referred to as ASD, is a developmental disorder characterized by challenges in interpersonal communication³⁸.

Outcomes

We measured overweight based on anthropometric data and high blood pressure, abnormal fasting blood glucose, and high blood cholesterol based on laboratory examination. We categorized an individual as overweight if they had a body mass index over 25.0 kg/m²and a waist circumference more than 85 cm for females and 90 cm for males based on the Korean Society for the Study of Obesity³⁹. High blood pressure was defined as a diastolic BP ≥ 90 mm Hg or a systolic BP ≥ 140 mm Hg. We defined diabetes as fasting blood glucose levels of ≥ 126 mg/dl. Abnormal cholesterol levels were defined as ≥ 240 mg/dl total cholesterol, a HDL level of < 40 mg/dl, and ≥ 160 mg/dl LDL levels.

Next, we assessed physical activity, smoking, and drinking habits based on the self-reported questionnaires. We verified if individuals performed vigorous and moderate physical activity at least once a week. Vigorous physical activities included running, aerobic exercise, and high-speed bicycling, while moderate activities included power walking, doubles tennis, and normal-speed bicycling. Vigorous and moderate physical activity were defined as binary variables, with a value of 1 if individuals engaged in the respective activity at least once a week and 0 otherwise. We also measured current smoking and drinking status over the past year. Current smoking status was defined as a binary variable, with a value of 1 for individuals who were current smokers and 0 for non-smokers over the past year. Drinking status was defined as a binary variable, with a value of 1 for individuals who consumed alcohol in the past year and 0 for non-drinkers.

Potential confounders

We included the following potential confounding variables as covariates: sex, age, health coverage type, residence (metropolitan, city, and rural), income quintile, and examination year (2019, 2020). We employed age as a continuous variable with age square. Healthcare coverage included employment-based NHI enrollees, other NHI enrollees, and Medical Aid, which is a subsidy program for individuals with a low income. We categorized income level into five groups using contribution quintiles: medical Aid and first contribution quintile, second contribution quintile, third contribution quintile, fourth contribution quintile, and fifth contribution quintile.

Statistical analyses

To minimize any effects from measured confounders and obtain comparability between groups, we matched (1:1 ratio) the cohort based on the propensity score (PS). We estimated PS using a multivariable logistic regression model within the sex-age exact-matched sample with all predefined covariates. Baseline characteristics for both unmatched and matched samples were compared using chi-square tests for categorical variables and t-statistics for continuous variables. We first examined differences in each outcome between those with and without disabilities using both unmatched and matched samples, employing t-statistics for both. Then, in the age and sex exact- and PS-matched samples, we carried out a multivariable logistic regression using disability groups and other covariates for each outcome measure to estimate the incidence and 95% confidence intervals (CI) across groups. We performed all statistical analyses using SAS Enterprise Guide 7.1 for Windows (SAS Institute, Cary, USA).