Prevalence and incidence of systemic lupus erythematosus in South Korea
- First Online:
- Cite this article as:
- Shim, J., Sung, Y., Joo, Y.B. et al. Rheumatol Int (2014) 34: 909. doi:10.1007/s00296-013-2915-9
- 351 Views
The aim of this study was to estimate the nationwide prevalence and incidence of systemic lupus erythematosus (SLE) in South Korea. National Health Insurance claims data covering almost all Koreans (~50 million) during 2006–2010 were analyzed. Individuals with SLE were identified if (1) they had experienced at least one hospitalization for SLE (International Classification of Diseases, 10th revision code M32), (2) they had taken at least one concomitant prescription of immunosuppressant and hydroxychloroquine, or (3) they had taken anti-dsDNA antibody (≥2) or complement tests (≥2) during each calendar year. Incident cases were defined only if they had not been SLE prevalent for the preceding 2 years and had been SLE prevalent for 2 years consecutively thereafter. The annual prevalence (per 100,000) increased slightly from 20.6 [95 % confidence interval (95 % CI) 20.2–21.0] in 2006 to 26.5 (95 % CI 26.0–27.0) in 2010, and the incidence (per 100,000) ranged between 2.5 (95 % CI 2.4–2.6) in 2008 and 2.8 (95 % CI 2.7–2.9) in 2009. The number of SLE-prevalent female patients outnumbered SLE-prevalent male patients by approximately sixfold, with a female-to-male incidence ratio of ~9:1. The prevalence and incidence of SLE increased significantly with age, regardless of sex, to a peak the age of 30–39 years. However, while both of them significantly decreased thereafter in females, this tendency was not observed in males.
KeywordsPrevalenceIncidenceSystemic lupus erythematosusKorea
Systemic lupus erythematosus (SLE) is a chronic autoimmune disorder that may affect diverse organ systems, and for which clinical manifestations tend to accrue gradually over time . There have been tremendous improvements in diagnosis and medical care for SLE since the 1950s, including earlier diagnosis, dialysis, renal replacement, and medication, resulting in an increase in the 5-year survival rate from 50 % to over 90 %, and in the 15- to 20-year survival rate of up to around 80 % [2, 3]. However, SLE still carries with it a significantly higher risk of death than for the general population and, as the patients survive longer, they face a range of complications from the disease itself or from damage consequent to its treatment during their lifetime . SLE also reduces the patient’s health-related quality of life, incurs high medical costs and significant productivity loss, and represents a great burden to both society and the patient [5, 6].
The etiology of SLE has not yet been fully described, but it has been shown that both the magnitude and severity of the disease differ markedly with respect to sex, age, race, and region [3, 7]. In the USA, the prevalence and incidence of SLE are 100–150 (per 100,000) and over 5 (per 100,000), respectively [8–12], and SLE is more predominant among African Americans, Asians, Native Americans, and Hispanics than in Whites [11, 13]. In Norway and the UK, SLE also occurs more frequently in Asians than in Europeans [14, 15]. A recent review of the epidemiology of SLE in the Asia–Pacific region found that the prevalence and incidence ranged from 4.3 to 45.3 (per 100,000) and from 0.9 to 3.1 (per 100,000), respectively . Epidemiologic information regarding SLE in South Korea has yet to be reported.
The aim of the present study was to estimate the nationwide prevalence and incidence of SLE using Korean National Health Insurance (NHI) claims data.
Materials and methods
A mandatory universal health insurance system has been operating in South Korea since 1989, and in 2010, about 50 million people were covered by the NHI program . NHI claims data contain nationwide medical information, including outpatient claims, inpatient claims, and details of both procedures and prescriptions. Claims data are collected for reimbursement. The NHI has healthcare service claims submitted by service providers and thus does not have the results of laboratory examinations. Almost all of these data are collected through an electronic claiming system, and all information is managed by the Health Insurance Research and Assessment Agency (HIRA), which is affiliated to the National Health Insurance Corporation (NHIC).
The present study used all claims records of enrollees aged 1–99 years from the Korean NHI claims database between January 1, 2006 and December 31, 2010. This study was approved by the Care Record Supply Committee of the HIRA. Since personal identification numbers in the claims data were encoded so as to ensure the anonymity of the enrollees, the study protocol did not require approval from an institutional review board.
Identification of SLE cases
Given that the diagnostic accuracy of the claims data may be questionable, SLE cases were identified based on an operational definition rather than solely on the diagnostic codes. Using the 2010 data, algorithms were developed with a view to identifying SLE cases; the best of these was chosen after evaluating their accuracy.
First, several SLE-related characteristics were chosen through rheumatologists’ reviews, including hospitalization due to SLE, prescription (hydroxychloroquine, immunosuppressants, and steroids), and laboratory examinations [anti-dsDNA antibody test and complement test (C3 or C4)]. After confirming whether enrollees had claim issues related to these individual characteristics during the calendar year (January 1–December 31), all information regarding the presence of hospitalization, prescriptions, and laboratory tests was combined to make several algorithms for identifying SLE cases. Their accuracy was evaluated through comparing the number of algorithm-identified cases to that of true cases with SLE. Patients who were registered in the Individual Copayment Beneficiaries Program (ICBP) for rare and intractable disorders were defined as true cases with SLE. The ICBP is an individual copayment beneficiaries program for patients with rare and intractable disease to reduce their burden of medical expenses  Application for registration with this program is made on behalf of patients by their doctor; patients must fulfill at least four of the 1997 revised American College of Rheumatology (ACR) classification criteria . Registration of SLE patients who meet these criteria only began in July 2009. However, since the registration system was in its early days, true SLE patients might have been included as non-SLE cases during that first year if their doctors did not register them with the ICBP due to unawareness of the system. Therefore, only data from 2010 were used to develop and evaluate the accuracy of the algorithms, and all claims issued from medical institutions from which no SLE patients had been registered with the ICBP during the study period were excluded.
We calculated comparability ratio, sensitivity, specificity, accuracy, predictive positive value (PPV), and negative predictive value (NPV) of all algorithms as well as each single SLE-related characteristic. Comparability ratio is a ratio of the number of algorithm-identified SLE cases to that of true SLE cases. Ratio over 1.0 means that algorithm-identified cases were much more than true cases. When ratio is close to 1.0, the number of algorithm-identified cases is similar to that of true SLE case.
Validity of each single SLE-related characteristic and algorithms combined with SLE-related characteristics
(Number of claims)
Each single SLE-related characteristic
Concomitant prescription (≥1)
Immunosuppressant and steroid
Immunosuppressant and HCQ
Steroid and HCQ
Laboratory examination (≥2)
Complement test (C3 or C4)
Anti-dsDNA antibody test
Hospitalization (≥1), or immunosuppressant and HCQ (≥1)
Hospitalization (≥1), or immunosuppressant and HCQ (≥1), or complement (C3 or C4) test (≥2)
Hospitalization (≥1), or immunosuppressant and HCQ (≥1), or anti-dsDNA antibody test (≥2)
Hospitalization (≥1), or immunosuppressant and HCQ (≥1), or complement (C3 or C4) test (≥2), or anti-dsDNA antibody test (≥2)
The best algorithm combined the presence of (1) hospitalization for SLE (International Classification of Diseases, 10th revision, ICD-10, code M32), (2) concomitant prescription of immunosuppressant and hydroxychloroquine, or (3) examination of anti-dsDNA antibody (≥2) or complement test (≥2) during each calendar year. The comparability ratio, accuracy, sensitivity, specificity, and positive and negative predictive values were 1.0, 79, 74, 82, 75, and 81 %, respectively.
Estimation of prevalence and incidence
All claims data during 2006–2010 were used to identify prevalent and incident SLE cases. According to the selected algorithm, subjects were identified as SLE-prevalent cases each year if they had experienced at least one hospitalization with a diagnostic code for SLE (ICD-10 diagnostic code M32), taken concomitant prescription of immunosuppressant and hydroxychloroquine (≥1), or taken anti-dsDNA antibody (≥2) or complement tests (C3 or C4, ≥2) during the calendar year. Incident cases of SLE were identified during 2008–2009. We defined subjects who had no SLE for the preceding 2 years (e.g., 2006–2007) and met the algorithm in the year (e.g., 2008) and also in the following year (e.g., 2009) as incident cases in the year (e.g., 2008).
For annual prevalence, the year-specific numerator was subjects who were prevalent cases in the specific calendar year, and the denominator was the mid-year population from the Korean National Statistical Office of the year. For annual incidence, the year-specific numerator was subjects who were incident cases in the specific calendar year and the denominator was same as for the calculation of prevalence.
Crude rates, sex- and age-specific rates, standardized rates adjusted for sex and age using the 2008 mid-year population, and their 95 % confidence intervals (95 % CIs) were calculated. The crude rates did not differ from the standardized rates, and so they are not presented herein. All analyses were conducted with SAS 9.1.2 Statistical Software (SAS Institute, Cary, NC, USA).
Prevalence and incidence of SLE in South Korea (2006–2010)
No. of cases
Rate per 105 (95 % CI)
No. of cases
Rate per 105 (95 % CI)
No. of cases
Rate per 105 (95 % CI)
The overall incidence (per 100,000) was 2.5 (95 % CI 2.4–2.6) in 2008 and 2.8 (95 % CI 2.7–2.9) in 2009 (Table 2). There were 1,260 incident cases (117 males, 1,143 females) in 2008, and the female-to-male ratio was about 9:1. As with prevalence, the incidence increased significantly with age until 30–39 years and then decreased slowly thereafter; while this trend was observed among females, it was not obvious for males (Fig. 1; Supplement Table 2).
Although numerous studies on SLE prevalence and incidence have been conducted in various regions of the world, SLE burden in South Korea has not yet been reported. This study estimated the nationwide prevalence and incidence of SLE in South Korea using NHI claims data that included almost all Koreans. The prevalence of SLE (per 100,000) was from 20.6 in 2006 to 26.5 in 2010 and tended to slightly increase each year. The incidence of SLE (per 100,000) ranged from 2.5 in 2008 to 2.8 in 2009.
Studies on the prevalence and incidence of SLE in other countries
No. of prevalent cases
Prevalence per 105
Incidence per 105
Hochberg et al. 
Uramoto et al. 
Naleway et al. 
Chakravarty et al. 
USA (CA, PA)
Feldman et al. 
Medicaid claims data
Furst et al. 
MCPs claims data
Barnabe et al. 
Claims, hospitalization data
Peschken and Esdaile 
Molina et al. 
Private health insurance data
Xiang and Dai 
Zeng et al. 
Mok et al. 
Yeh et al. 
Iseki et al. 
Hopkinson et al. 
Johnson et al. 
Lerang et al. 
Stahl-Hallengren et al. 
Alonso et al. 
Laustrup et al. 
Medical records, interview
Govoni et al. 
Hospital discharge data
Alamanos et al. 
The incidence (per 100,000) is over 5 in the USA [9, 11, 12, 21] and 3–5 in Europe [14, 30, 31, 33, 34, 38] and Asian countries [27–29]. Our result was comparable to that reported for Hong Kong and Taiwan. The incidence in the present study was about ninefold higher in females than in males, and in other studies approximately fivefold to eightfold higher [14, 28, 29, 34, 37, 38]. Similar to Taiwan , the incidence of SLE increased continuously with age until its peak, and then declined slowly thereafter (as did the prevalence). However, the incidence for female Norwegians exhibited a bimodal pattern, with first and second peaks occurring at 16–29 and 50–59 years, respectively. This pattern might be related to the use of contraceptive pills during the reproductive age or to the use of hormone therapy after menopause . In other European countries, including the UK, Spain, and Denmark, the incidence of SLE increases until the age of 20–39 years and remains high thereafter to the age of 50–69 years, although there are some differences in the age at which this peaks [30, 34, 35]. In the USA, the incidence of SLE appears to grow steadily with age [11, 12, 21]. In the Caucasian population, late-onset SLE is reported to be more prevalent  and the clinical course is considered to be more benign compared to SLE onset earlier in life . In addition, SLE in Asians is known to be more severe and to have a greater mortality rate . These findings may explain why the peak age in both Europe and the USA appears to be higher and the incidence remains high in older age group.
It is difficult to conclude that SLE is more or less frequent in some countries than in others, because differences of SLE burden between countries might be due to the differences in the study population, design, data source, study period, diagnosis, and diagnostic criteria. However, regional disparities appear to hold true. Differences across countries might be mainly explained by differences of racial makeup in countries. This can be inferred from studies in which differences between ethnic groups who live in the same region were compared: SLE was not only more predominant but also more severe among African Americans, Asians, Native Americans, and Hispanics than among Caucasians [14, 15, 39]. However, even taking into consideration the racial composition, the prevalence of SLE among Asians who live in Europe [14, 15] or the USA [11, 13] appears to be higher than those who reside in their homeland. These findings might also reflect differences in healthcare environmental factors across countries, including the healthcare system, accessibility to medical care, case-finding techniques, and therapeutic environments, together with other risk factors such as industrial pollution and exposure to sunlight [40, 41].
In the present study, we observed a slightly increasing trend in prevalence and incidence. Changes in the incidence and prevalence of SLE have been reported in several studies. The studies concerning trends in the incidence over several decades [9, 37] reported significant increase trends, but other studies, which were recently conducted during relatively short period, have showed almost constant trends [14, 35]. Our increasing trend might be explained by raising social awareness on the disease and expanding coverage of copayment beneficiaries program for rare disease.
Previous studies on SLE have been conducted using medical records from either single centers or specific regions. Even though the use of medical records has an advantage that SLE could be confirmed by ACR classification criteria [1, 14], it has a difficulty to generalize the findings of those studies. Moreover, in case of rare diseases, it is not easy to conduct a nationwide-scale study. For such reasons, health insurance claims data are widely used for researchers. However, there remain some issues that require attention. Even though the Korean NHIC has made an effort to reduce inaccuracies, the diagnostic information in the data is often inaccurate, either due to simple coding errors, misclassification due to a lack of medical knowledge, or the reimbursement system itself [43, 44]. Therefore, cases with a specific disease need to be identified based on a disease-specific algorithm, not simply a diagnostic code. A study from Canada defined the subjects as being patients with SLE based on the presence of ≥1 billing code by a rheumatologist, ≥2 billing codes (≥8 weeks apart but within 2 years) by any physician, or one hospitalization diagnosis . Another study from the USA defined subjects as SLE prevalent if they had at least one inpatient claim with an SLE diagnosis or ≥2 office or emergency room visits at least 30 days apart, within the 12 months following the earliest date of service of a medical claim, with a diagnosis of SLE . These components and their combination may differ between countries. The present study developed several algorithms using information on hospitalization due to SLE, prescription of immunosuppressants, steroids, and hydroxychloroquine, and laboratory examination of anti-dsDNA antibody and complement (C3 or C4). The identification of patients using such algorithms will help when using claims data collected before the beginning of the ICBP for studies on various topics such as healthcare utilization, patterns of care, medical costs, and disease incidence and prevalence.
The present study was subject to several limitations. First, the administrative claims data that were analyzed were obtained only as a result of subjects visiting healthcare institutions. No information was therefore available for SLE patients who did not visit a healthcare institution. That could influence underestimation of the SLE burden. However, it might have had little impact on the findings since NHI claims data cover almost all Koreans and the expanded coverage of the NHI system has made it easy for people to access health institutions. The second limitation relates to the definition of SLE prevalent and incident cases. While SLE patients may have long periods of remission, 2.3 years on average (range 1.0–5.7 years) , all of our algorithm-identified SLE cases were patients who either received treatment such as hospitalization and medications or had SLE-related laboratory examinations each year. Therefore, if their disease activity was in remission and they did not require treatment or undergo any of the aforementioned examinations, they were not regarded as SLE patients in the present study. This might have led to an underestimation of the SLE prevalence. Conversely, the incidence might have been overestimated because we defined SLE cases as incident only if prevalent cases had not had SLE for the preceding 2 years and had SLE for 2 years consecutively thereafter.
In summary, the nationwide prevalence and incidence of SLE in South Korea were estimated using NHI claims data that covered almost all Koreans. The prevalence (per 100,000) of SLE in Korea increased from 20.6 in 2006 to 26.5 in 2010, and the incidence (per 100,000) was about 2.5. Estimating the nationwide burden of disease is an important first step toward understanding the public health impact of this disease. The algorithm selected to identify SLE cases may facilitate various studies that employ claims data.
This study was supported in part by grants from the Korea Healthcare Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (A080588 and A120404).
Conflict of interest