National Health Insurance Service-Health Screening Research Articles

Association of sudden sensorineural hearing loss with asthma: a longitudinal follow-up study using a national sample cohort

ObjectiveTo investigate the risk of sudden sensorineural hearing loss (SSNHL) in asthma patients.DesignA longitudinal follow-up study using a retrospective cohortSettingThe 2002–2013 Korean National Health Insurance Service-Health Screening CohortParticipants and interventionsThe...

Association between Thyroid Cancer and Breast Cancer: Two Longitudinal Follow-Up Studies Using a National Health Screening Cohort.

Background: The purpose of this study was to evaluate the association between thyroid cancer and breast cancer. Methods: Data from the Korean National Health Insurance Service-Health Screening Cohort were collected from 2002 to 2013. In study I, 3949 thyroid cancer participants were 1:4 matched with 15,796 control I participants, and hazard ratios (HRs) with 95% confidence intervals (CIs) for breast cancer were evaluated using a stratified Cox proportional hazard model. In study II, 3308 breast cancer participants were 1:4 matched with 13,232 control II participants, and HRs with 95% CIs for thyroid cancer were assessed in the same way as in study I. In the subgroup analyses, associations were analyzed according to radioactive iodine (RAI) treatment and age (<60 years old and ≥60 years old). Results: The adjusted HR for breast cancer in the thyroid cancer group was 1.64 (95% CI = 1.13–2.39, p = 0.010). The adjusted HR for thyroid cancer in the breast cancer group was 1.91 (95% CI = 1.47–2.49, p < 0.001). In the subgroup analyses, the groups that were older and not treated with RAI treatment showed consistent results in study I, and the younger and older groups showed consistent results in study II. Conclusions: Based on this cohort study, breast and thyroid cancer have a reciprocal positive association.

The Occurrence of Alzheimer's Disease and Parkinson's Disease in Individuals With Osteoporosis: A Longitudinal Follow-Up Study Using a National Health Screening Database in Korea.

Background: Public health concerns regarding the potential link between osteoporosis and the increased occurrence of Alzheimer’s disease (AD) and Parkinson’s disease (PD) have been raised, but the results remain inconsistent and require further validation. Here, we investigated the long-term relationship of osteoporosis with the occurrence of AD/PD using data from a large-scale nationwide cohort.Methods: This longitudinal follow-up study included 78,994 patients with osteoporosis and 78,994 controls from the Korean National Health Insurance Service-Health Screening Cohort database (2002–2015) who were matched using propensity score matching at a 1:1 ratio based on age, sex, income, and residential area. A Cox proportional hazard model was used to assess the association between osteoporosis and the occurrence of AD/PD after adjusting for multiple covariates.Results: During the follow-up period, AD occurred in 5,856 patients with osteoporosis and 3,761 controls (incidence rates: 10.4 and 6.8 per 1,000 person-years, respectively), and PD occurred in 1,397 patients and 790 controls (incidence rates: 2.4 and 1.4 per 1,000 person-years, respectively). The incidences of AD and PD were significantly higher in the osteoporosis group than in the matched control group. After adjustment, the osteoporosis group exhibited 1.27-fold and 1.49-fold higher occurrences of AD (95% confidence interval (CI) = 1.22–1.32) and PD (95% CI = 1.36–1.63) than the controls, respectively. The results of subgroup analyses supported the increased occurrence of AD and PD in patients with osteoporosis, independent of income, residential area, obesity, smoking, alcohol consumption, hyperlipidemia, hypertension, or blood glucose level.Conclusion: Our results indicate that the presence of osteoporosis may increase the likelihood of developing two common neurodegenerative diseases in adults aged ≥40 years.

Association between Rheumatoid Arthritis and Meniere's Disease: A Longitudinal Follow-Up Study Using a National Health Screening Cohort.

This study aimed to evaluate the impact of pre-existing rheumatoid arthritis (RA) on the occurrence of Meniere’s disease (MD). The 2002–2015 Korean National Health Insurance Service—Health Screening Cohort data were retrospectively analyzed. A total of 3038 participants with RA were matched with 12,152 control participants for demographic factors. The occurrence of MD was evaluated in both the RA and control groups. The hazard ratios (HRs) of RA for participants with MD were calculated using a stratified Cox proportional hazard model. Additionally, subgroup analyses were conducted. The rate of MD was not different between the RA and control groups (1.5% vs. 1.3%, standardized difference = 0.01). The HR was not higher in the RA group than in the MD group (adjusted HR = 1.03, 95% confidence interval = 0.73–1.44, p = 0.885). A higher HR of RA for participants with MD was found in the ≥60-year-old subgroup in the crude model but not in the adjusted model. An association between RA and MD was not found in any of the other subgroups. A previous history of RA was not related to an increased risk of MD.

Blood Pressure Levels and Risks of Dementia: a Nationwide Study of 4.5 Million People

There are inconsistent results on the impacts of controlling blood pressure (BP) on the risk of dementia. We investigated the association between BP and risk of dementia subtypes by antihypertensive treatment and comorbidities. Using the Korean National Health Insurance Service-Health Screening Database from 2009 to 2012, a total of 4 522 447 adults aged 60+ years without a history of dementia were analyzed and followed up for a mean of 5.4 years. Individuals were classified according to their baseline systolic BP (SBP) and diastolic BP; SBP 130 to <140 mm Hg and diastolic BP 80 to <90 mm Hg were used as reference groups. The risk of overall dementia and probable Alzheimer disease was significantly higher in the SBP≥160 and lower SBP groups. These U-shaped associations were consistent regardless of antihypertensive use or comorbidities. The risk of probable vascular dementia (VaD) was not higher among lower SBP groups and increased gradually as SBP increased. Although there was a linear association between SBP and the risk of probable VaD in individuals not taking antihypertensives or without comorbidities, there was a U-shaped association in individuals taking antihypertensives or with comorbidities. Patterns of association between diastolic BP and risk of probable Alzheimer disease or probable VaD were similar to those with SBP, except for the risk of probable VaD in individuals taking antihypertensives. In conclusion, risks of probable Alzheimer disease and probable VaD were different among lower BP groups. Although the risk of dementia appears higher in people with lower BP receiving antihypertensives, this finding may be affected by comorbidities.

Association between Temporomandibular Joint Disorder and Weight Changes: A Longitudinal Follow-Up Study Using a National Health Screening Cohort.

This study aimed to investigate BMI changes following a temporomandibular joint disorder (TMJD) diagnosis. The Korean National Health Insurance Service-Health Screening Cohort from 2002 to 2015 was used. In Study I, 1808 patients with TMJD (TMJD I) were matched with 7232 participants in comparison group I. The change in BMI was compared between the TMJD I and comparison I groups for 1 year. In study II, 1621 patients with TMJD (TMJD II) were matched with 6484 participants in comparison group II participants. The change in BMI was compared between the TMJD II and comparison II groups for 2 years. In Study I, the BMI change was not associated with TMJD. In Study II, the BMI change was associated with TMJD in the interaction of the linear mixed model (p = 0.003). The estimated value (EV) of the linear mixed model was −0.082. The interaction was significant in women < 60 years old, women ≥ 60 years old, and the obese I category. TMJD was not associated with BMI changes after 1–2 years in the overall population. In women and obese patients, TMJD was associated with a decrease in BMI after 2 years.

Association between antibiotics use and diabetes incidence in a nationally representative retrospective cohort among Koreans

Numerous studies have reported that antibiotics could lead to diabetes, even after adjusting for confounding variables. This study aimed to determine the causal relationship between antibiotics use and diabetes in a nationally representative cohort. This retrospective cohort study included adults aged 40 years or older who were enrolled in the Korean National Health Insurance Service-Health Screening Cohort. Antibiotic exposure was assessed from 2002 to 2005 and newly diagnosed diabetes mellitus was determined based on diagnostic codes and history of antidiabetic medication use from 2006 to 2015. Multivariate Cox proportional hazards model was used to assess the association between antibiotic use and diabetes incidence. The mean age of the 201,459 study subjects was 53.2 years. People who used antibiotics for 90 or more days had a higher risk of diabetes (adjusted hazard ratio [aHR] 1.16, 95% confidence interval [CI] 1.07–1.26) compared to non-users. Those who used five or more classes of antibiotics had a higher risk of diabetes than those who used one antibiotic class (aHR 1.14; 95% CI 1.06–1.23). The clear dose-dependent association between antibiotics and diabetes incidence supports the judicious use of antibiotics in the future.

National cohort and meteorological data based nested case\u2013control study on the association between air pollution exposure and thyroid cancer

The objective of this study was to evaluate the influence of exposure to meteorological conditions, including air pollution, on thyroid cancer. A nested case–control study was conducted utilizing 4632 patients with thyroid cancer and 18,528 control subjects who were matched at a 1:4 ratio by age group, sex, income, and region of residence. Korean National Health Insurance Service-Health Screening Cohort data from 2002 to 2015 were used. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated for thyroid cancer correlated with meteorological and air pollution exposure over a moving average of 3 years before the index dates. For all participants, the adjusted ORs associated with relative humidity (1.01, 95% CI 1.00–1.03, P value = 0.023), ambient atmospheric pressure (1.02, 95% CI 1.01–1.03, P value < 0.001), and sunshine duration (1.17, 95% CI 1.04–1.31, P value = 0.007) indicated correlations with the occurrence of thyroid cancer; however, these results were inconsistent in the subgroup analyses. Overall, exposure to nitrogen dioxide (NO2) (1.33, 95% CI 1.24–1.43, P value < 0.001) and particulate matter (PM10) (0.64, 95% CI 0.60–0.69, P value < 0.001) were related to thyroid cancer. These relationships persisted in the subgroup analyses. In conclusion, thyroid cancer occurrence was positively associated with NO2 exposure and negatively associated with PM10 exposure.

Association Between Asthma and Meniere's Disease: A Nested Case-Control Study.

This study aimed to investigate the association between Meniere's disease and prior history of asthma. A nested case-control study. Among the patients aged ≥40 years from the Korean National Health Insurance Service-Health Screening Cohort 2002-2015, 7,734 diagnosed with Meniere's disease between 2004 and 2015 were enrolled in this study. From the 505,834 participants without a prior history of Meniere's disease, 30,936 control participants were selected based on age, sex, income, and region of residence. The presence of a prior history of asthma before the diagnosis of Meniere's disease was compared between the Meniere's disease group and control group using conditional logistic regression with odds ratios (ORs) and 95% confidence intervals (95% CIs). Prior history of asthma was associated with 1.30 (1.21-1.39) times higher odds of Meniere's disease (E-value=1.92 [1.71]). Both allergic asthma and nonallergic asthma patients demonstrated higher odds of Meniere's disease than the control group (adjusted OR=1.21 [1.08-1.36] and E-value=1.72 [1.37] for allergic asthma; adjusted OR=1.26 [1.17-1.36] and E-value=1.83 [1.60] for nonallergic asthma). This study is the first to demonstrate a positive relationship between Meniere's disease and prior asthma history in adults. The association between prior asthma history and Meniere's disease was noted in both allergic and nonallergic asthma cases. The potential development of Meniere's disease should be considered when managing the asthma patients with dizziness or vertigo. 3 Laryngoscope, 132:864-872, 2022.

Fasting Blood Glucose Variability and Unfavorable Trajectory Patterns Are Associated with the Risk of Colorectal Cancer.

Background/AimsThe relationship between fasting blood glucose (FBG) variability and colorectal cancer (CRC) remains ill-defined. This study aimed to evaluate the association of FBG variability with CRC risk in the healthy population without overt diabetes.MethodsIn the data from the Korean National Health Insurance Service-Health Screening Cohort, we included individuals examined by FBG testing at least 3 times between 2002 and 2007. FBG variability was calculated using standard deviation (SD) and coefficient of variation (CV).ResultsRegarding FBG variability, an increase in the quintile of SD or CV was independently associated with CRC risk (all p for trend <0.01). When the change in FBG was classified into six trajectory patterns, unfavorable trajectory patterns (high stable and upward) were significantly associated with increased CRC risk (hazard ratio [HR] 2.30, p=0.003; HR 1.19, p=0.007, respectively). In subgroup analyses according to the sex, a significant association between FBG variability (SD or CV) and CRC risk was observed in men but not in women. The high stable and upward pattern were also associated with CRC risk in men (HR 2.47, p=0.002; HR 1.21, p=0.012) but not in women.ConclusionsThis study identified that FBG variability and unfavorable trajectory patterns were significantly associated with increased CRC risk in the healthy population without overt diabetes. Our findings suggest that FBG variability as well as FBG itself may be a predictive factor for the development of CRC.

Association between air pollution and chronic rhinosinusitis: a nested case-control study using meteorological data and national health screening cohort data.

Inconsistent results about the effect of air pollution on chronic rhinosinusitis (CRS) have been reported. This study aimed to evaluate the impact of meteorological conditions/air pollution on the prevalence of CRS in adult Koreans. The data from the Korean National Health Insurance Service-Health Screening Cohort from 2002 through 2015 were used. A CRS group (defined as ICD-10 codes J32, n=6159) was matched with a control group (n=24,636) in 1:4 ratios by age, sex, income, and region of residence. The meteorological conditions and air pollution data included the daily mean, highest, and lowest temperature (°C), daily temperature range (°C), relative humidity (%), ambient atmospheric pressure (hPa), sunshine duration (hr), and the rainfall (mm), SO2 (ppm), NO2 (ppm), O3 (ppm), CO (ppm), and PM10 (μg/m3) levels before the CRS diagnosis. Crude and adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for CRS were analyzed using logistic regression analyses. When the NO2 level increased by 0.1 ppm, the odds for CRS increased 5.40 times, and when the CO level increased by 1 ppm and PM10 increased by 10 μg/m3, the odds for CRS decreased 0.75 times and 0.93 times, respectively. Other meteorological conditions, such as the mean/highest/lowest temperature, temperature range, rainfall and other air pollution, such as SO2 and O3, were not statistically significant. NO2 for 90 days before the index date increased the risk of CRS in all subgroups, except for the nasal polyp and older age subgroups. CRS is related to high concentrations of NO2.

Short- and long-term exposure to air pollution increases the risk of stroke.

Many epidemiological studies have observed the association of air pollutant exposure with the onset, progression, and mortality of stroke. The aim of this study was to investigate the associations of air pollutants, including SO2, NO2, O3, CO, and PM10, with stroke according to exposure duration. Data from the Korean National Health Insurance Service-Health Screening Cohort from 2002 to 2015 were obtained. The 21,240 patients who were admitted for or died due to stroke were 1:4 matched for age, sex, income, and region of residence with 84,960 control participants. The meteorological factors of mean, highest, and lowest temperatures; relative humidity; ambient atmospheric pressure; and air pollutant concentrations (SO2, NO2, O3, CO, and PM10) were analyzed to determine their associations with stroke. The odds ratios for stroke after exposure to each meteorological factor and air pollutant at 7 and 30 days were calculated in the stroke and control groups. Subgroup analyses were conducted according to age, sex, income, and region of residence. The odds ratio associated with seven days of exposure to CO was 1.16 (95% CI = 1.04-1.31) in stroke patients. For 30 days of exposure, the odds ratio associated with CO was 1.16 (95% CI = 1.02-1.32) in stroke patients. Seven and 30 days of NO2 exposure were inversely associated with stroke. The odds ratio associated with seven days of exposure to O3 was 1.16 (95% CI = 1.01-1.32) in ischemic stroke patients. Both ischemic and hemorrhagic stroke had negative associations with 7 and 30 days of NO2 exposure. Both short- and long-term exposure to CO were related to stroke.

Association between Statin Use and Meniere's Disease: Results from a National Health Screening Cohort.

The protective effects of statins against inner ear diseases have been suggested. This study investigated the relationship between previous statin use and the occurrence of Meniere’s disease (MD). Participants ≥40 years old in the Korean National Health Insurance Service-Health Screening Cohort 2002–2015 were enrolled. A total of 7734 MD participants were matched with 38,670 comparison participants. The dates of statin prescriptions for the 2 years before the onset of MD were examined. A conditional logistic regression analysis was performed to estimate the odds ratios (ORs) of statin use for MD. Regarding the different types of statins, lipophilic statins, but not hydrophilic statins, were associated with lower odds of MD in the <65 year-old group (adjusted OR = 0.81, 95% CI = 0.68–0.97, p = 0.023). Prior statin use did not show association with MD in the adult population. Regarding the different types of statins, lipophilic statin use was related to a lower rate of MD in a middle-aged population.

Effects of short- and long-term exposure to air pollution and meteorological factors on Meniere\u2019s disease

The association between air pollutants and Meniere’s disease has not been explored. The present study investigated the relationship between meteorological factors and air pollutants on Meniere’s disease. Participants, aged ≥ 40 years, of the Korean National Health Insurance Service-Health Screening Cohort were included in this study. The 7725 patients with Meniere’s disease were matched with 30,900 control participants. The moving average meteorological and air pollution data of the previous 7 days, 1 month, 3 months, and 6 months before the onset of Meniere’s disease were compared between the Meniere’s disease and control groups using conditional logistic regression analyses. Additional analyses were conducted according to age, sex, income, and residential area. Temperature range; ambient atmospheric pressure; sunshine duration; and levels of SO2, NO2, O3, CO, and PM10 for 1 month and 6 months were associated with Meniere’s disease. Adjusted ORs (odds ratios with 95% confidence interval [CI]) for 1 and 6 months of O3 concentration were 1.29 (95% CI 1.23–1.35) and 1.31 (95% CI 1.22–1.42), respectively; that for the 1 and 6 months of CO concentration were 3.34 (95% CI 2.39–4.68) and 4.19 (95% CI 2.79–6.30), respectively. Subgroup analyses indicated a steady relationship of O3 and CO concentrations with Meniere’s disease. Meteorological factors and air pollutants were associated with the rate of Meniere’s disease. In particular, CO and O3 concentrations were positively related to the occurrence of Meniere’s disease.

Incidence and risk factors analysis for mortality after total knee arthroplasty based on a large national database in Korea

This study aimed to analyze the rates and risk factors of postoperative mortality among 560,954 patients who underwent total knee arthroplasty (TKA) in Korea. The National Health Insurance Service-Health Screening database was used to analyze 560,954 patients who underwent TKA between 2005 and 2018. In-hospital, ninety-day, and one-year postoperative mortality, and their association with patient’s demographic factors and various comorbidities (ie., cerebrovascular disease, congestive heart failure, and myocardial infarction) were assessed. In-hospital, ninety-day and one-year mortality rates after TKA were similar from 2005 to 2018. The risk of in-hospital mortality increased with comorbidities like cerebrovascular disease (hazard ratio [HR] = 1.401; 95% confidence interval [CI] = 1.064–1.844), congestive heart failure (HR = 2.004; 95% CI = 1.394 to 2.881), myocardial infarction (HR = 2.111; 95% CI = 1.115 to 3.998), and renal disease (HR = 2.641; 95% CI = 1.348–5.173). These co-morbidities were also independent predictors of ninety-day and one-year mortality. Male sex and old age were independent predictors for ninety-day and one-year mortality. And malignancy was risk factor for one-year mortality. The common preoperative risk factors for mortality in all periods were male sex, old age, cerebrovascular disease, congestive heart failure, myocardial infarction, and renal disease. Malignancy was identified as risk factor for one-year mortality. Patients with these comorbidities should be provided better perioperative care.

Risk of Neuropsychiatric Diseases According to the Use of a Leukotriene Receptor Antagonist in Middle-Aged and Older Adults with Asthma: A Nationwide Population-Based Study Using Health Claims Data in Korea

Although a leukotriene receptor antagonist (LTRA) is an effective drug for asthma, there has been increasing concern regarding neuropsychiatric (NP) adverse reactions. However, evidence for this association is not sufficient, especially in adults. To investigate the association between the use of an LTRA and the risk of developing NP diseases in adults with asthma. We performed a nationwide, retrospective, cohort study using data from the National Health Insurance Service-Health Screening Cohort (NHIS-HEALS). We selected asthma patients with no previous use of an LTRA, and NP outcomes were defined by the registration of certain International Classification of Diseases, 10th Revision codes (F00-F59) during follow-up. We obtained the hazard ratio (HR) for NP diseases according to the use of an LTRA. Overall, 61,571 asthma patients without LTRA experience were enrolled, and 12,168 of them took an LTRA during the follow-up period. In the adjusted model, the HR for newly diagnosed NP diseases showed no significant difference according to use of an LTRA (HR 1.01; 95% confidence interval 0.83-1.23; P= .952). Subgroup analysis for associations between duration of LTRA use and risk of NP disease indicated no significance for all groups (<6, 6 to <12, 12 to <24, and ≥24 months). Common NP diseases included dementia (75.4% vs 76.1%), mood disorders (12.68% vs 12.80%), and panic disorders (5.63% vs 3.53%) in LTRA users and non-LTRA users, respectively, and there was no significant difference in the prevalence of each NP disease in either group. The current study showed the use or duration of LTRA exposure was not associated with the occurrence of NP diseases in Korean adult asthmatics.

Associations of hypertension burden on subsequent dementia: a population-based cohort study

In this nationwide cohort study, we assessed the effects of hypertension burden and blood pressure (BP) control on dementia in different age subgroups. From the Korean National Health Insurance Service-Health Screening cohort from January 1, 2005 to December 31, 2013, we enrolled 428,976 subjects aged 40–79 years without previous diagnosis of dementia or stroke. During a mean follow-up of 7.3 ± 1.5 years, 9435 (2.2%) were diagnosed with dementia. Per 10 mmHg increase in systolic BP (SBP), risk of dementia was increased by 22% (95% confidence interval [CI] 1.15–1.30) in subjects aged 40–59 years and 8% (95% CI 1.04–1.11) in subjects aged 60–69 years. No significant associations were observed in subjects aged ≥ 70 years. Among subjects aged 40–59 years, both vascular and Alzheimer’s dementia risks were increased with increasing SBP. Increasing hypertension burden (proportion of days with increased BP) was associated with higher dementia risk (hazard ratio [HR] 1.09 per 10% increase, 95% CI 1.08–1.10). Among patients with baseline SBP ≥ 140 mmHg, optimal follow-up SBP (120–139 mmHg) was associated with decreased dementia risk (HR 0.69, 95% CI 0.50–0.95). Hypertension burden was associated with higher risks of dementia. Adequate BP control was associated with lower risk of dementia in individuals aged < 70 years.

Body weight variability and cancer incidence in men aged 40\xa0years and older-Korean National Insurance Service Cohort

Repeated weight fluctuation has been proposed as a potential risk factor for increasing morbidity and mortality including cancer. We aimed to investigate the association between body weight variability (BWV) and all cancer and site-specific cancer incidence and the impact of smoking on these associations. A total of 1,759,848 cancer-free male subjects who had their weight measured at least 5 times from the National Health Insurance Service-Health Screening Cohort from 2002 to 2011 were included and followed up until 2015. BWV was defined as the average absolute difference between successive values (ASV). The risk of cancer and site-specific cancer from BWV was identified using Cox proportional hazards regression analysis using hazard ratios (HRs) and 95% confidence intervals (CIs) adjusted for potential confounders including weight, and stratified analysis was also conducted according to smoking status. During the 7,015,413 person-years of follow-up, 11,494 patients (0.65%) developed new-onset cancers. BWV was associated with a higher risk of all cancers after adjustment for confounders. The highest BWV quintile group compared to the lowest had greater risks of all cancers and site-specific cancers including lung, liver, and prostate cancer (HR 1.22, 95% CI 1.15–1.30; HR 1.22, 95% CI 1.07–1.39; HR 1.46, 95% CI 1.19–1.81; HR 1.36, 95% CI 1.15–1.62, in all cancers, lung, liver and prostate cancer, respectively). Due to small number of cancer occurrence, the risk of kidney cancer was increased, but statistically insignificant (HR 1.38, 95% CI 0.91–2.10). Similar results were observed in noncurrent smokers. However, in current smokers, the risks of all cancers and only prostate cancer were significantly increased in the highest BWV quintile group (HR 1.19, 95% CI 1.09–1.31; HR 1.51, 95% CI 1.08–2.11). The risk of kidney cancer also increased in this group, although the finding was not statistically significant (HR 1.77, 95% CI 0.87–3.63) This study suggested BWV is an independent risk factor for cancer in men, especially in lung, liver, and prostate cancer, but evidence was weaker in kidney cancer. This association remained significant only in prostate cancer in current smokers.

Bidirectional association between migraine and rheumatoid arthritis: two longitudinal follow-up studies with a national sample cohort

ObjectiveTo investigate the bidirectional association between migraine and rheumatoid arthritis (RA).DesignTwo longitudinal follow-up studies.SettingData collected from a national cohort between 2002 and 2013 by the Korean National Health Insurance Service-Health...

Machine Learning-Based Cardiovascular Disease Prediction Model: A Cohort Study on the Korean National Health Insurance Service Health Screening Database.

Background: This study proposes a cardiovascular diseases (CVD) prediction model using machine learning (ML) algorithms based on the National Health Insurance Service-Health Screening datasets. Methods: We extracted 4699 patients aged over 45 as the CVD group, diagnosed according to the international classification of diseases system (I20–I25). In addition, 4699 random subjects without CVD diagnosis were enrolled as a non-CVD group. Both groups were matched by age and gender. Various ML algorithms were applied to perform CVD prediction; then, the performances of all the prediction models were compared. Results: The extreme gradient boosting, gradient boosting, and random forest algorithms exhibited the best average prediction accuracy (area under receiver operating characteristic curve (AUROC): 0.812, 0.812, and 0.811, respectively) among all algorithms validated in this study. Based on AUROC, the ML algorithms improved the CVD prediction performance, compared to previously proposed prediction models. Preexisting CVD history was the most important factor contributing to the accuracy of the prediction model, followed by total cholesterol, low-density lipoprotein cholesterol, waist-height ratio, and body mass index. Conclusions: Our results indicate that the proposed health screening dataset-based CVD prediction model using ML algorithms is readily applicable, produces validated results and outperforms the previous CVD prediction models.