Cohort profile: the National Health Insurance Service-National Health Screening Cohort (NHIS-HEALS) in Korea.

PURPOSE: The National Health Insurance Service-Health Screening Cohort (NHIS-HEALS) is a cohort of participants who participated in health screening programmes provided by the NHIS in the Republic of Korea. The NHIS constructed the NHIS-HEALS cohort database in 2015. The purpose of this cohort is to offer relevant and useful data for health researchers, especially in the field of non-communicable diseases and health risk factors, and policy-maker. PARTICIPANTS: To construct the NHIS-HEALS database, a sample cohort was first selected from the 2002 and 2003 health screening participants, who were aged between 40 and 79 in 2002 and followed up through 2013. This cohort included 514 866 health screening participants who comprised a random selection of 10% of all health screening participants in 2002 and 2003. FINDINGS TO DATE: The age-standardised prevalence of anaemia, diabetes mellitus, hypertension, obesity, hypercholesterolaemia and abnormal urine protein were 9.8%, 8.2%, 35.6%, 2.7%, 14.2% and 2.0%, respectively. The age-standardised mortality rate for the first 2 years (through 2004) was 442.0 per 100 000 person-years, while the rate for 10 years (through 2012) was 865.9 per 100 000 person-years. The most common cause of death was malignant neoplasm in both sexes (364.1 per 100 000 person-years for men, 128.3 per 100 000 person-years for women). FUTURE PLANS: This database can be used to study the risk factors of non-communicable diseases and dental health problems, which are important health issues that have not yet been fully investigated. The cohort will be maintained and continuously updated by the NHIS. © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2017. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

It is a cohort with a large sample size, with a relatively low rate of attrition over more than 10 years.

It contains the date and cause of death, which were determined using the national database and extensive information on healthcare usage regarding inpatient and outpatient visits to healthcare institutions and medication histories.

Variables on health behaviours are limited since those data were obtained from self-reporting. In addition, the disease diagnoses in the claim data might not accurately reflect patients’ medical conditions.

Introduction

The National Health Insurance Service-Health Screening Cohort (NHIS-HEALS) is a cohort of participants who participated in health screening programmes provided by the NHIS in the Republic of Korea (hereafter ‘Korea’). The purpose of this cohort is to offer relevant and useful data for a wide range of health researchers.

NHIS-HEALS is based on information obtained through the national health screening programmes of Korea. Since 1995, the NHIS has provided general national health screening programmes, including an oral health screening programme, to improve the health status of Koreans through the prevention and early detection of diseases.1 2 In 2007, a health screening programme for transitional ages, aimed at those aged 40 and 66 years, was also launched.3 NHIS-HEALS incorporates information from these three major health screening programmes for the adult Korean population (see online supplementary figure 1). All insured adults are eligible for a general health screening programme that is biennially conducted (annually for manual workers). The participation rate in the general health screening programme among the eligible population was 74.8% in 2014.4 The general health screening programme can be applied at least once every 2 years for the entire population of Korean adults aged 40 years or older. The healthcare institutions for screening are designated according to the Framework Act on Health Examinations, and must meet the standards of manpower, facilities and equipment.

The NHIS established the National Health Information Database (NHID) in 2011, which incorporates all data from the NHIS and consists of five databases:5 an eligibility database, a national health screening database, a healthcare usage database, a long-term care insurance database and a healthcare provider database. The NHID covers the entire population of Korea (50 million) and thus has proven unwieldy for researchers. The NHIS constructed a representative 2% sample cohort database, the NHIS-National Sample Cohort (NHIS-NSC),6 but the NHIS-NSC did not meet the high demand for research requiring both health screening data and long-term health outcomes. The NHIS therefore constructed the NHIS-HEALS cohort database in 2015 to support a wide range of public research. The NHIS-HEALS has been made publicly available to facilitate wider use of the health screening database, and includes a larger sample of health screening participants than the NHIS-NSC.

Cohort description

The participants of the cohort

The eligibility criteria for the general health screening programme provided by the NHIS varied according to the insurance type of beneficiaries. Employed individuals were eligible at all ages, while the self-employed were eligible if they were the head of household of a family. The dependents of the employed and family members of the self-employed heads of household were eligible only for those aged 40 years or older. Among the beneficiaries of the medical aid programme, which is a tax-based governmental programme for low-income families that covers approximately 3% of all Koreans, heads of household 19–64 years of age and family members 41–64 years of age were eligible for the general health screening programme. Medical aid beneficiaries have been included in the general health screening programme since 2012.

To construct the NHIS-HEALS database, a sample cohort was first selected from the 2002 and 2003 health screening participants, who were aged between 40 and 79 in 2002 and followed up through 2013. This cohort included 514 866 health screening participants who comprised a 10% simple random sample of all health screening participants in 2002 and 2003. Since only a small proportion of people aged less than 40 participated in the health screening programme, and the response rate was very low among people aged 80 years or older, the NHIS-HEALS was limited to adults aged 40 to 79 years. Gender-specific and age-specific distributions of the cohort population, the source population (all health screening participants) and the overall Korean population are presented in online supplementary table 1.7 Under the current National Health Insurance Act, the data can only be used for research purposes without patients’ individual consent. Nevertheless, identification is difficult because the sample was drawn from the entire population and the data use deidentified individual keys that were created for the NHIS-HEALS.

The general characteristics of the cohort population at baseline are presented in table 1. A total of 54.2% of the participants were men. The number of participants aged 40–44 years was highest among all age groups, accounting for a quarter of the sample (25.2%). A total of 55.3% of the participants lived in non-metropolitan areas, which covers some urban areas and all rural areas. The most common insurance type was health insurance for the employed. A total of 0.6% of the participants had any disabilities. The biennial screening participant rates ranged from 65.1% to 70.9% during the 2004–2013 period. Of the sample population, 31.6% participated six times in the health screening programmes during the follow-up period. A total of 42.3% of the men and 96.2% of the women were non-smokers. Nearly half of the men (45.7%) drank alcohol more than once per week, while most of the women (82.5%) rarely drank. Of the men, 49.7% never engaged in exercise at least once per week, compared with 67.0% of the women.

Table 1

General characteristics of the National Health Insurance Service-Health Screening Cohort subjects at baseline (2002–2003)

Variables		n	%
Sex	Men	279 125	54.2
	Women	235 741	45.8
Age	40–44	129 979	25.2
Mean: 52.641	45–49	107 002	20.8
SD: 9.635	50–54	80 080	15.6
	55–59	64 952	12.6
	60–64	59 328	11.5
	65–69	41 828	8.1
	70–74	21 615	4.2
	75–79	10 082	2.0
Region	Seoul metropolitan city	89 344	17.4
	Other metropolitan cities	141 055	27.4
	Non-metropolitan area	284 467	55.3
Insurance type	Self-employed insured	197 992	38.5
	Employed insured	316 359	61.4
	Medical aid beneficiary	515	0.0
Disability	No	511 964	99.4
	Yes	2902	0.6
No of participants (biennial) in 2002–2013	2002–2003	514 866	100.0
	2004–2005	334 966	65.1
	2006–2007	352 158	68.4
	2008–2009	361 043	70.1
	2010–2011	364 757	70.9
	2012–2013	345 693	67.1
The frequency of biennial screening participation in 2002–2013	6	162 782	31.6
	5	129 786	25.2
	4	88 755	17.2
	3	58 628	11.4
	2	42 042	8.2
	1	32 873	6.4
Risk factors in 2002–2003 (baseline)		Men/Women	Men %/Women%
Cigarette smoking	Non-smoker	112 577/218 147	42.3/96.2
	Ex-smoker	41 519/2170	15.6/1.0
	Current smoker	112 143/6476	42.1/2.9
Smoking duration	<10 years	18 724/3108	12.2/36.0
	10–29 years	93 620/3646	60.9/42.2
	≥30 years	41 318/1892	26.9/21.9
Alcohol drinking	Rarely	96 441/189 721	35.1/82.5
	2–3 times per month	52 995/24 104	19.3/10.5
	More than once per week	125 688/16 134	45.7/7.0
Exercise	None	134 524/153 342	49.7/67.0
	1–2 times per week	80 104/37 738	29.6/16.5
	More than three times per week	55 916/37 669	20.7/16.5

Follow-up interval

The cohort was followed up through 2013 annually for the eligibility information including death information and healthcare usage (all participants), and not annually for the health screening information (only those who meet the eligibility criteria, biennially, for the screening programme and those who participated in the screening programme). Information on death (date and cause of death) from Statistics Korea was individually linked using unique personal identification numbers. By law, all deaths must be reported to Statistics Korea. Personal information regarding insurance contribution (a proxy for income), residential area and disability status was tracked every year from the eligibility database. The eligibility information was collected from the Public Information Sharing System, National Tax Service and Ministry of Health and Welfare of Korea, and managed by the NHIS, which has 178 regional branches and approximately 13 000 employees across Korea. As the NHIS covers the entire population of Korea, the healthcare usage information included all visits (inpatient, outpatient and pharmacy visits) to healthcare facilities that occurred in Korea. Information about the healthcare facilities was also monitored annually. Regarding the health screening follow-ups, 31.6% of the participants were monitored biennially until 2013, and 93.6% of the participants were examined at least once after a baseline screening. The cohort will be maintained and continuously updated by the NHIS.

The key variables

The key variables of the NHIS-HEALS, which were mainly constructed from the variables of the NHID, are presented in table 2 and online supplementary table 2. The eligibility database included information about income-based insurance contributions (a proxy for income), demographic variables, and date and cause of death. Variables for specific health problems and risk factors from questionnaires (cigarette smoking status/dose/duration, frequency per week and amount per day of alcohol drinking—regardless of the type of alcohol, type and days per week of physical activity, medical history and family history) and bioclinical laboratory results (blood pressure, fasting glucose, lipid profile, haemoglobin, urine stick test, creatinine, liver enzyme, body mass index and waist circumference) were included in the health screening database. Some variables changed during the follow-up period. The healthcare usage database was based on data collected during the process of claiming healthcare services and included information on records of inpatient and outpatient usage (diagnosis, length of stay, treatment costs and services received) and prescription records (drug code, days prescribed and daily dosage). The healthcare provider database included information on types of healthcare institutions, healthcare human resources and equipment.

Table 2

Major variables in the National Health Insurance Service-National Health Screening Cohort database

Domain		Variables	Year
			2002	2003	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013
Target health problems	Hypertension	Systolic blood pressure	√	√	√	√	√	√	√	√	√	√	√	√
		Diastolic blood pressure	√	√	√	√	√	√	√	√	√	√	√	√
	Diabetes mellitus	Fasting blood glucose	√	√	√	√	√	√	√	√	√	√	√	√
	Dyslipidaemia	Total cholesterol	√	√	√	√	√	√	√	√	√	√	√	√
		Triglyceride								√	√	√	√	√
		HDL (high densitiy lipoprotein) cholesterol								√	√	√	√	√
		LDL (low density lipoprotein) cholesterol								√	√	√	√	√
	Anaemia	Haemoglobin	√	√	√	√	√	√	√	√	√	√	√	√
	Kidney/urinary disease	Urine glucose	√	√	√	√	√	√	√
		Urine blood	√	√	√	√	√	√	√
		Urine pH	√	√	√	√	√	√	√
		Urine protein	√	√	√	√	√	√	√	√	√	√	√	√
	Chronic kidney disease	Creatinine								√	√	√	√	√
	Liver disease	AST (aspartate transaminase)[SGOT (serum glutamic-oxaloacetic transaminase)]	√	√	√	√	√	√	√	√	√	√	√	√
		ALT (alanine transaminase)[SGPT (serum glutamic-pyruvic transaminase)}	√	√	√	√	√	√	√	√	√	√	√	√
		r-GTP (gamma-glutamyl transpeptidase)	√	√	√	√	√	√	√	√	√	√	√	√
	Frailty/lower leg weakness	Neurological examination for lower leg for subjects at age 40 or 66						√	√	√	√	√	√	√
	Osteoporosis	Bone density for subjects at age 40 or 66						√	√	√	√	√	√	√
	Periodontal diseases	Dental examination	√	√	√	√	√	√	√	√	√	√	√	√
	Cognitive impairment, depression	Mental health screening						√	√	√	√	√	√	√
	Common and uncommon diseases	Disease diagnosis per ICD-10 codes; Operation and procedure history, Medication history (generic name code, dose, duration of prescription, and material codes)	√	√	√	√	√	√	√	√	√	√	√	√
	All-cause and cause- specific deaths	Vital statistics including dates and causes of deaths	√	√	√	√	√	√	√	√	√	√	√	√
Risk factors	Cigarette smoking	Cigarette smoking status	√	√	√	√	√	√	√	√	√	√	√	√
		Daily smoking dose	√	√	√	√	√	√	√
		Past daily smoking dose								√	√	√	√	√
		Current daily smoking dose								√	√	√	√	√
		Smoking duration	√	√	√	√	√	√	√
		Smoking duration (ex-smoker)								√	√	√	√	√
		Smoking duration (current smoker)								√	√	√	√	√
	Alcohol	Drinking frequency	√	√	√	√	√	√	√
		Days of drinking per week								√	√	√	√	√
		Amount of drinking per count	√	√	√	√	√	√	√
		Amount of drinking per day								√	√	√	√	√
	Obesity	Body mass index	√	√	√	√	√	√	√	√	√	√	√	√
		Waist circumference							√	√	√	√	√	√
	Physical activity	Days of activity per week	√	√	√	√	√	√	√
		Days of vigorous activity per week								√	√	√	√	√
		Days of moderate activity per week								√	√	√	√	√
		Days of mild activity per week								√	√	√	√	√
	Dental caries, etc	Dental examination	√	√	√	√	√	√	√	√	√	√	√	√
Medical history and family history	Medical history	Hypertension, diabetes mellitus, dyslipidaemia, pulmonary tuberculosis, stroke, ischaemic heart disease, etc	√	√	√	√	√	√	√	√	√	√	√	√
	Family history	Hypertension, diabetes mellitus, stroke, ischaemic heart disease etc	√	√	√	√	√	√	√	√	√	√	√	√
Healthcare usage		Date of visit, types of medical institutions (clinics/hospitals/tertiary hospitals/public health centres), types of visit (inpatient/outpatient/emergency/intensive care), length of stay, medical cost (insurer/patient)	√	√	√	√	√	√	√	√	√	√	√	√
Healthcare provider		Location, type of hospitals, no of beds, medical equipment, human resources, specialties of physicians	√	√	√	√	√	√	√	√	√	√	√	√
Socioeconomic and demographic factors		Age, sex, age, residential area, insurance type (the employee insured, the self-employed insured, dependents, medical aid), monthly insurance contributions (a proxy for income), types and grades of disabilities	√	√	√	√	√	√	√	√	√	√	√	√

Findings to date

As the NHIS-HEALS was launched in December 2015, no noteworthy studies have yet been published. However, several studies using the health screening and healthcare usage database of the NHID have been published. Studies have examined the associations of body mass index with cancer risk8 and mortality,9 glucose levels with cancer risk10 and hospitalisation,11 smoking with cancer12 13 and diabetes mellitus,14 physical activities with body mass index15 and cholesterol levels with cancer risk.16 These research results have had positive impacts on health promotion by raising awareness of various public health issues, with an example being the lawsuit against the tobacco industry by the NHIS.17 The NHIS-HEALS will provide additional strong evidence regarding the issues that were assessed in previous studies using the NHID by including the cause of death, unlike the NHID.

We herein present the basic statistics of NHIS-HEALS for future data users. We calculated the prevalence rates of various conditions, the incidence density of those conditions, healthcare usage rates and mortality. The rates were age-standardised using the census population of Statistics Korea in 2005 and the world standard population.18 The rates that were standardised using the world standard are presented below.

Prevalence rates for specific health problems identified from the health screening database at baseline (2002–2003) are presented in table 3. The age-standardised prevalence of anaemia in the NHIS-HEALS was 9.8%, with a higher rate in women (15.5%) than men (5.9%) (p<0.001). The age-standardised prevalence of diabetes mellitus was 8.4%, while the age-standardised prevalence of hypertension in the NHIS-HEALS was 36.1%. The prevalence of diabetes and hypertension was higher in men than women (p<0.001). The age-standardised prevalence of obesity (body mass index of 30 kg/m2 or greater) in NHIS-HEALS was 2.7%, while the prevalence of overweight (body mass index of 25 kg/m2 or greater, but less than 30 kg/m2) was 31.0%. The age-standardised prevalence of hypercholesterolaemia in the NHIS-HEALS was 14.3%; the rate was higher in women (16.0%) than men (12.4%) (p<0.001). The age-standardised prevalence of abnormal urine protein tests was 2.0%, and the rate was the same (2.0%) in both sexes. When we compared these results with those of the Korean National Health and Nutrition Examination Survey for participants aged 40 or over,19 generally similar levels of prevalence of anaemia, diabetes, hypertension, obesity and hypercholesterolaemia were found.

Table 3

Crude and age-standardised (with the 2005 Korean census and world standard populations as references) prevalence rates (%) for specific health problems in the health screening database of the National Health Insurance Service-National Health Screening Cohort database at baseline, 2002–2003

	All				Men				Women
	Cohort n	Crude rates	Age-standardised rates		Cohort n	Crude rates	Age-standardised rates		Cohort n	Crude rates	Age-standardised rates
			Census	WHO			Census	WHO			Census	WHO
Anaemia*	514 256	9.2	9.8	9.8	278 866	4.4	5.6	5.9	235 390	15.6	15.9	15.5
Diabetes mellitus	514 190	7.9	8.2	8.4	278 826	9.0	9.4	9.6	235 364	6.2	6.4	6.6
Prediabetes†		23.9	23.8	23.9		26.0	25.9	25.9		20.8	20.7	20.9
Hypertension	514 581	34.4	35.6	36.1	278 989	38.0	39.1	39.4	235 592	29.8	30.3	31.0
High-normal‡		17.0	16.9	16.8		17.8	17.8	17.8		15.7	15.5	15.6
Abnormal liver function test§	514 286	6.2	5.8	5.8	278 866	5.7	5.3	5.1	235 420	6.5	6.3	6.4
Obesity	514 350	2.9	2.7	2.7	278 901	2.1	1.9	1.9	235 449	3.7	3.6	3.6
Overweight¶		32.1	31.2	31.0		33.1	31.4	31.0		30.6	29.8	30.1
Hypercholesterolaemia**	513 887	14.3	14.2	14.3	278 681	13.0	12.5	12.4	235 206	15.8	15.6	16.0
Abnormal urine blood††	513 046	5.9	6.0	6.1	278 228	3.2	3.4	3.4	234 818	9.3	9.3	9.4
Abnormal urine protein‡‡	513 095	2.0	2.0	2.0	278 252	1.9	2.0	2.0	234 843	1.9	2.0	2.0

*Hb <13 g/dL (men), <12 g/dL (women).

†Diabetes mellitus: fasting glucose of 126 mg/dL; prediabetes, 100–126 mg/dL.

‡Hypertension: systolic blood pressure of 140 mm Hg or diastolic blood pressure of 90 mm Hg; high-normal, systolic blood pressure of 130 mm Hg or diastolic blood pressure of 85 mm Hg.

§More than two times than upper limit of normal (ULN); ULN for alanine aminotransferase: men, 30 IU/L; women, 19 IU/L.

¶Obesity: body mass index ≥30 kg/m2; overweight: body mass index 25–29.9 kg/m2.

**Total cholesterol ≥240 mg/dL (6.2 mmol/L).

††Urine dip-stick test for occult blood: +1, +2, +3, +4.

‡‡Urine dip-stick test for protein: +1, +2, +3, +4.

The incidence density for specific health problems based on information from the health screening database in 2005–2013 is presented in table 4. To identify incident cases, we excluded patients who were previously diagnosed in the first 3 years (2002–2004) of the study period, because the data did not include the baseline information (participants’ screening and healthcare usage records before 2002). With reference to previous studies,20–22 the exclusion period was set as the first 2 years, starting in 2002 (2002–2003) or 2003 (2003–2004). The incidence density was highest for hypertension (4.7%), followed by anaemia (2.9%), hypercholesterolaemia (2.6%), abnormal urine blood (2.3%) and diabetes mellitus (1.7%).

Table 4

Crude and age-standardised (with the 2005 Korean census and world standard populations as references) incidence density (per 100 person-years) for specific health problems in the health screening database of the National Health Insurance Service-National Health Screening Cohort database, 2005–2013

	All				Men				Women
	Cohort n	Crude rates	Age-standardised rates		Cohort n	Crude rates	Age-standardised rates		Cohort n	Crude rates	Age-standardised rates
			Census	WHO			Census	WHO			Census	WHO
Incidence density (2005–2013)
Anaemia*	450 207	2.6	2.9	2.9	259 787	1.9	2.3	2.4	190 420	3.7	3.8	3.8
Diabetes mellitus†	461 760	1.7	1.7	1.7	243 937	2.1	2.1	2.1	217 823	1.2	1.3	1.3
Hypertension‡	299 857	4.1	4.6	4.7	146 050	4.4	4.8	4.9	153 807	3.8	4.4	4.5
Abnormal liver function test§	471 917	1.4	1.4	1.3	255 924	1.1	1.0	1.0	215 993	1.8	1.8	1.8
Obesity¶	496 720	0.3	0.3	0.3	270 970	0.2	0.2	0.2	225 750	0.3	0.3	0.3
Hypercholesterolaemia **	420 838	2.7	2.6	2.6	230 116	2.0	1.9	1.9	190 722	3.4	3.4	3.4
Abnormal urine blood ††	471 226	2.4	2.3	2.3	264 380	1.4	1.4	1.4	206 846	3.7	3.6	3.6
Abnormal urine protein‡‡	499 428	0.7	0.7	0.8	269 992	0.8	0.8	0.8	229 436	0.7	0.7	0.7

Incident cases were defined as those cases newly diagnosed in 2005–2013 who did not meet the diagnostic criteria in 2002–2004.

*Hb <13 g/dL (men), <12 g/dL (women).

†Fasting glucose of 126 mg/dL.

‡Systolic blood pressure of 140 mm Hg or diastolic blood pressure 90 mm Hg.

§More than two times than upper limit of normal (ULN); ULN for alanine aminotransferase: men, 30 IU/L; women, 19 IU/L.

¶Body mass index ≥30 kg/m2.

**Total cholesterol ≥240 mg/dL (6.2 mmol/L).

††Urine dip-stick test for occult blood: +1, +2, +3, +4.

‡‡Urine dip-stick test for protein: +1, +2, +3, +4.

The healthcare usage rates of 10 major diseases at baseline based on the healthcare usage database are presented in online supplementary table 3. The rates were highest for acute upper respiratory infections and influenza (46.5%), followed by dyspepsia and other diseases of the stomach and duodenum (29.7%) and other diseases of the eye and adnexa (22.3%).

The mortality rates of the cohort population are presented in table 5, and survival curve of participants is presented in figure 1. We calculated mortality rates using the entire sample data of NHIS-HEALS from 2003 to 2013. The age-standardised (defined with reference to the Korean census population) mortality rate for the first 2 years (through 2004) was 463.6 per 100 000 person-years, while the rate for 5 years (through 2007) was 678.3 per 100 000 person-years and the rate for 10 years (through 2012) was 910.2 per 100 000 person-years. In men, the mortality rate was higher than in women (2-year mortality rates of 680.4 per 100 000 person-years for men and 250.8 per 100 000 person-years for women) (p<0.001).

Table 5

Number of all-cause deaths through 2012 (10 years after baseline) and crude and age-standardised (with the 2005 Korean census and world standard populations as references) mortality rates (per 100 000 person-years) in the National Health Insurance Service-National Health Screening Cohort database

All-cause	No of cohort population*	All				Men				Women
		No of deaths	Crude rate	Age-standardised mortality rates		No of deaths	Crude rate	Age-standardised mortality rates		No of deaths	Crude rate	Age-standardised mortality rates
				Census	WHO			Census	WHO			Census	WHO
Mortality rates (2003–2012)†
2 year (2004)	512 802	3705	360.2	442.0	463.6	2748	493.1	648.9	680.4	957	203.1	238.3	250.8
5 year (2007)	503 007	13 097	513.0	646.0	678.3	9 336	676.5	917.8	963.0	3761	320.6	383.9	404.8
10 year (2012)	483 421	33 058	657.9	865.9	910.2	22 684	839.0	1199.9	1260.5	10 374	447.0	556.2	586.4

*No of cohort population at the end of the year.

†Death cases were defined as those cases who died in 2003–2012.

Figure 1

Survival curve of participants by sex in the National Health Insurance Service-National Health Screening Cohort database.

The major causes of death by sex during the follow-up period (2003–2013) are presented in table 6 and figure 2. Causes of death were classified using the list of 56 causes of death of Statistics Korea, which was derived from the list of 80 causes of death for the tabulation of mortality statistics recommended by WHO. The most common cause of death was malignant neoplasm in both sexes (406.6 per 100 000 person-years for men, 140.5 per 100 000 person-years for women). Heart disease was the second most common cause in men (91.0 per 100 000 person-years) and the third most common cause in women (50.8 per 100 000 person-years). Cerebrovascular diseases were the third most common cause in men (89.0 per 100 000 person-years) and the second most common cause in women (64.5 per 100 000 person-years). Suicide was the fourth most common cause overall (31.6 per 100 000 person-years), the fourth most common cause in men (45.6 per 100 000 person-years) and the fifth most common cause in women (16.9 per 100 000 person-years).

Table 6

Cause-specific death rates for leading causes of death (2003–2013) and age-standardised (with the 2005 Korean census and world standard populations as references) mortality rates (per 100 000 person-years) in the National Health Insurance Service-National Health Screening Cohort database

Rank	All				Men				Women
	Cause of death	Crude rates	Age-standardised rates		Cause of death	Crude rates	Age-standardised rates		Cause of death	Crude rates	Age-standardised rates
			Census	WHO			Census	WHO			Census	WHO
1	Malignant neoplasms	269.5	239.5	264.4	Malignant neoplasms	365.8	364.1	406.6	Malignant neoplasms	157.6	128.3	140.5
2	Cerebrovascular diseases	68.8	66.9	76.5	Heart diseases	71.7	79.2	91.0	Cerebrovascular diseases	63.5	55.4	64.5
3	Heart diseases	59.7	60.8	70.8	Cerebrovascular diseases	73.3	78.9	89.0	Heart diseases	45.7	42.0	50.8
4	Suicide	32.1	30.0	31.6	Suicide	44.3	43.3	45.6	Diabetes mellitus	22.8	20.0	23.5
5	Diabetes mellitus	25.6	24.5	28.1	Chronic lower respiratory infections	26.8	32.2	38.1	Suicide	18.0	15.8	16.9
6	Transport accidents	23.8	21.6	22.6	Transport accidents	33.2	31.3	32.6	Transport accidents	12.8	11.2	12.0
7	Chronic lower respiratory infections	19.0	19.8	23.7	Pneumonia	19.7	30.8	39.7	Pneumonia	9.8	11.1	14.5
8	Pneumonia	15.1	19.5	25.1	Diabetes mellitus	28.1	28.9	32.3	Hypertensive diseases	11.2	9.9	11.8
9	Diseases of liver	18.0	15.6	16.3	Diseases of liver	28.2	25.1	26.0	Chronic lower respiratory infections	10.0	9.8	12.3
10	Hypertensive diseases	10.0	11.0	13.5	Hypertensive diseases	8.9	12.5	15.8	Alzheimer’s disease	4.6	7.0	9.9

The cause of death was classified using the selection list of 56 causes of death provided by Statistics Korea, which originated from the list of 80 causes of death for the tabulation of mortality statistics recommended by WHO.

Figure 2

The major 10 causes of death by sex in the cohort sample of the National Health Insurance Service-National Health Screening Cohort database.

Strengths and limitations

The NHIS-HEALS has several strengths. First, it is a cohort with a large sample size (n=514 866), with a relatively low rate of attrition over more than 10 years of follow-up due to the nature of the national administration data. Second, a questionnaire survey, physical examination, dental health screening and clinical laboratory tests were performed for all cohort members. This database can be used to study the risk factors of non-communicable diseases and dental health problems, which are an important health issue that has not yet been fully investigated. Third, the NHIS-HEALS contains the date and cause of death, which were determined using the national database for cause of death produced by Statistics Korea, which allows investigations such as burden-of-disease studies. Statistics Korea annually reports cause of death statistics, and a previous study reported the accuracy of the cause of death to be 92%.23 Fourth, the NHIS-HEALS contains extensive information on healthcare usage regarding inpatient and outpatient visits to healthcare institutions and medication histories.

The NHIS-HEALS also has weaknesses. The study subjects are slightly younger than the general population of Korea. Variables on health behaviours are limited since those data were obtained from self-reporting in nationwide health screenings. In addition, the disease diagnosis variables in the healthcare claim data might not accurately reflect patients’ medical conditions, but only healthcare usage sensitive to the Korean fee-for-service payment and reimbursement system.