Rheumatology
Rheumatol Int DOI 10.1007/s00296-014-3147-3
INTERNATIONAL
ORIGINAL ARTICLE - OBSERVATIONAL RESEARCH
Prevalence, incidence, and associated factors of avascular necrosis in Korean patients with systemic lupus erythematosus: a nationwide epidemiologic study Young Bin Joo · Yoon-Kyoung Sung · Jee-Seon Shim · Jae-Hoon Kim · Eui-Kyung Lee · Hye-Soon Lee · Sang-Cheol Bae
Received: 19 June 2014 / Accepted: 30 September 2014 © Springer-Verlag Berlin Heidelberg 2014
Abstract Avascular necrosis (AVN) is one of the most frequent types of organ damage in systemic lupus erythematosus (SLE). However, little is currently known about the epidemiology of AVN in SLE patients. The aim of this study was to estimate the prevalence and incidence of AVN in Korean patients with SLE based on National Health Insurance (NHI) claims data and to determine the risk factors for AVN among SLE patients. This study was conducted using the 2006–2010 data of 25,358 SLE patients from the NHI program. AVN cases were defined as those with at least one diagnosis of AVN. The prevalence was calculated by dividing the number of AVN cases by the number of SLE cases in the same year. The annual incidence was calculated by dividing the number of incident AVN cases by the number of SLE-prevalent cases not previously
Electronic supplementary material The online version of this article (doi:10.1007/s00296-014-3147-3) contains supplementary material, which is available to authorized users. Y. B. Joo · Y.-K. Sung · H.-S. Lee . S.-C. Bae (*) Department of Rheumatology, Hanyang University Hospital for Rheumatic Diseases, 17 Haengdang-Dong, Seongdong-Gu, Seoul 133-792, Republic of Korea e-mail: [email protected] J.-S. Shim Clinical Research Center for Rheumatoid Arthritis, Seoul, Republic of Korea J.-H. Kim Division of Rheumatology, Department of Internal Medicine, Korean University College of Medicine, Seoul, Republic of Korea E.-K. Lee School of Pharmacy, Sungkyunkwan University, Suwon, Republic of Korea
diagnosed with AVN. Patients who developed AVN in 2008–2010 were compared with SLE patients who did not develop AVN to identify any risk factors. The prevalence of AVN among SLE patients (2006–2010) was 31.5–34.2 per 1,000 persons and was similar in all the years studied. The incidence per 1,000 persons of AVN among SLE patients was 8.6 [95 % confidence interval (95 % CI) 6.9–10.3] in 2008, 9.8 (95 % CI 8.0–11.6) in 2009, and 8.4 (95 % CI 6.8–10.0) in 2010. Regression analysis indicated that taking an oral corticosteroid [odds ratio (OR) 2.12, 95 % CI 1.39–3.23] or an intravenous corticosteroid (OR 1.5, 95 % CI 1.2–1.89) was significantly associated with AVN. In addition, AVN was associated with use of immunosuppressive agents (OR 2.12, 95 % CI 1.66–2.72), hydroxychloroquine (OR 1.4, 95 % CI 1.09–1.81), and lipid-lowering agents (OR 1.78, 95 % CI 1.24–2.57) among the prescribed medications, and with hypertension (OR 1.39, 95 % CI 1.08–1.79) among the comorbidities. The prevalence and incidence of AVN among SLE patients, which were 31.5– 34.2 and 8.4–9.8 per 1,000 persons, respectively, may be representative of the entire population of symptomatic AVN patients with SLE in Korea. AVN is associated with the use of corticosteroids, immunosuppressants, hydroxychloroquine, lipid-lowering agents, and with hypertension. Studies of large, prospective cohorts are needed to confirm these results. Keywords Avascular necrosis · Systemic lupus erythematosus · Prevalence · Incidence · Risk factor
Introduction Avascular necrosis (AVN) is a well-known manifestation of systemic lupus erythematosus (SLE), which causes
13
significant joint pain and limitation of motion, eventually decreasing the patient’s quality of life, particularly in young patients [1]. The pathogenesis of AVN can be partially explained by subchondral bone necrosis due to interruption of the blood supply from the end artery to the bone [2]. SLE itself and the drugs used to treat it, such as corticosteroids, may accelerate this impairment of the blood supply [3, 4]. The reported prevalence of AVN in SLE patients is 3–30 % if only symptomatic patients are included, and up to 40 % when asymptomatic patients are also included [1, 3, 5–10]. This wide variation in previous estimates is attributable to differences in the definition of AVN, the diagnostic methodologies, and the patient populations studied. Since most of the studies involved only a small number of SLE patients and were carried out in a single center, the results may not accurately represent the characteristics of the entire SLE patient population. Therefore, we evaluated the prevalence and incidence of AVN among SLE patients in 2006–2010, based on the large Korean National Health Insurance (NHI) claims database. The possible risk factors for AVN were also investigated.
Rheumatol Int
four aforementioned characteristics was defined as the index date for patients with SLE. This algorithm was developed previously to estimate the prevalence and incidence of SLE using the claims database, and its comparability ratio and positive and negative predictive values were 1.0, 75, and 81 %, respectively. Prevalence of AVN AVN cases, among the SLE-prevalent cases, were defined as having at least one diagnosis of AVN for any reason (ICD-10 codes M870, M870x, M871, M871x, M872, M872x, M873, M873x, M878, M878x, M879, M879x, M905, and M905x). The accuracy of such a diagnosis was validated by Kang et al. [15]. The annual prevalence of AVN in 2006–2010 was calculated by dividing the number of AVN-prevalent cases with at least one diagnostic claim by the number of SLE-prevalent cases during the same year. That is, the year-specific numerator was subjects who were prevalent cases in the specific calendar year, and the denominator was prevalent SLE patients in the specific calendar year, as described previously [14]. Incidence of AVN
Patients and methods Data source This study was conducted using the Korean NHI claims database. This is a centralized claims database managed by the Health Insurance Research and Assessment Agency (HIRA), which is affiliated with the National Health Insurance Corporation. It contains medical claims data for the entire Korean population [11–13]. Both the inpatient and outpatient claims from the 2006–2010 NIH database were used. Demographic information (sex, age) and enrollee medical claims information, including comorbidities and prescribed medications, were collected. This study was approved by the Care Record Supply Committee of the HIRA. Since personal identification numbers in claims data were encoded so as to guarantee anonymity, the study protocol did not require approval from an institutional review board. Identification of SLE cases SLE-prevalent cases were defined as (1) hospitalization for SLE [code M32 in the International Classification of Diseases, 10th revision (ICD-10)], or (2) concomitant prescription of an immunosuppressant and antimalarial drugs, or (3) examination by anti-dsDNA antibody (≥2) or complement tests (≥2) during the applicable calendar year [14]. The earliest date of a medical claim based on any of the
13
The annual incidence of AVN in 2008–2010 was calculated by dividing the number of incident AVN cases by the number of SLE-prevalent cases that had not been diagnosed with AVN previously. The accuracy of the annual incidence of AVN was improved by excluding cases with an AVN diagnostic code during 2006–2007. That is, we defined subjects who had no AVN for the preceding 2 years (e.g., 2006–2007) and met the code in the year (e.g., 2008) as incident cases in the year (e.g., 2008). With the exception of those for which the AVN diagnostic code was assigned prior to the index date of SLE, all subjects meeting the AVN case definition during 2008–2010 were identified as AVN incident cases. The earliest date of a medical claim with an AVN diagnostic code was defined as the index date of AVN, and each subject was included in the calculation of annual incidence. Possible risk factors for AVN The baseline characteristics of SLE cases were compared in those with and without newly developed AVN in 2008– 2010. These characteristics were obtained from the medical claim on the AVN index date for cases with AVN and from the latest medical claim for those without AVN. Comorbid conditions and prescribed medications were also assessed during the 1-year pre-index period in the cases with AVN, and during the 1-year period prior to the latest claim in the cases without AVN. Comorbid conditions such as diabetes
Rheumatol Int Table 1 Crude and standardized prevalences of avascular necrosis (AVN) in patients with systemic lupus erythematosus (SLE) SLE, n
AVN, n
Crude prevalence per 1,000 persons (95 % CI)
Standardized prevalencea (95 % CI)
2006 2007 2008 2009
10,080 10,771 11,609 12,351
335 347 366 409
33.2 (29.6, 36.8) 32.2 (28.8, 35.6) 31.5 (28.3, 34.7) 33.1 (29.9, 36.3)
32.5 (29.2, 35.8) 31.9 (28.6, 35.2) 31.5 (28.3, 34.7) 33.2 (29.9, 36.5)
2010
13,316
456
34.2 (31.1, 37.3)
34.9 (31.5, 38.3)
Data are expressed as number (95 % confidence intervals) 95 % CI 95 % confidence interval a
All data are standardized for age and sex using the 2008 SLE population
mellitus, hypertension, osteoporosis, and lupus nephritis were included as potential risk factors or variables of interest to the study, as were prescribed medications, including intravenous and oral corticosteroids, immunosuppressive agents (cyclophosphamide, mycophenolate mofetil, cyclosporine, and azathioprine), hydroxychloroquine, and lipid-lowering agents. Among the intravenous corticosteroids, we included only methylprednisolone as we wished to focus on the impact of high-dose steroid injections on the development of AVN. Statistical analysis Crude and standardized prevalence and incidence data were adjusted for sex and age using the 2008 SLE population, and their 95 % confidence intervals (CIs) were calculated. Differences between cases with and without AVN were analyzed using t tests for continuous variables and chi-squared tests or Fisher’s exact tests for categorical variables. Logistic regression was used to develop a multivariate model with AVN as the outcome and age, sex, hypertension, lupus nephritis, intravenous corticosteroid, oral corticosteroid, hydroxychloroquine, immunosuppressive agent, and lipid-lowering agent as potential associated factors. However, other possible clinical and laboratory variables (Raynaud’s phenomenon, smoking, pregnancy, thrombosis, or antiphospholipid antibodies, etc.) were not considered as co-variables since the necessary information was not available in our claims database. Sensitivity analyses of prevalence and incidence were conducted since false positives were possible in cases where only diagnostic codes were used. Sensitivity analyses were limited to those patients with further evidence of SLE, based on the following additional medical claims: (1) hospitalization on the index date, (2) the presence of an orthopedist visit on the index date or during the year following the index date, or (3) the presence of a radiologic study [X-ray or magnetic resonance imaging (MRI)] on the index date, or (4) the presence of any one of the above. All tests were two sided, and the level of statistical significance
Fig. 1 Prevalence of AVN in 2009, stratified according to sex and age
was set at p
Rheumatol Int DOI 10.1007/s00296-014-3147-3
INTERNATIONAL
ORIGINAL ARTICLE - OBSERVATIONAL RESEARCH
Prevalence, incidence, and associated factors of avascular necrosis in Korean patients with systemic lupus erythematosus: a nationwide epidemiologic study Young Bin Joo · Yoon-Kyoung Sung · Jee-Seon Shim · Jae-Hoon Kim · Eui-Kyung Lee · Hye-Soon Lee · Sang-Cheol Bae
Received: 19 June 2014 / Accepted: 30 September 2014 © Springer-Verlag Berlin Heidelberg 2014
Abstract Avascular necrosis (AVN) is one of the most frequent types of organ damage in systemic lupus erythematosus (SLE). However, little is currently known about the epidemiology of AVN in SLE patients. The aim of this study was to estimate the prevalence and incidence of AVN in Korean patients with SLE based on National Health Insurance (NHI) claims data and to determine the risk factors for AVN among SLE patients. This study was conducted using the 2006–2010 data of 25,358 SLE patients from the NHI program. AVN cases were defined as those with at least one diagnosis of AVN. The prevalence was calculated by dividing the number of AVN cases by the number of SLE cases in the same year. The annual incidence was calculated by dividing the number of incident AVN cases by the number of SLE-prevalent cases not previously
Electronic supplementary material The online version of this article (doi:10.1007/s00296-014-3147-3) contains supplementary material, which is available to authorized users. Y. B. Joo · Y.-K. Sung · H.-S. Lee . S.-C. Bae (*) Department of Rheumatology, Hanyang University Hospital for Rheumatic Diseases, 17 Haengdang-Dong, Seongdong-Gu, Seoul 133-792, Republic of Korea e-mail: [email protected] J.-S. Shim Clinical Research Center for Rheumatoid Arthritis, Seoul, Republic of Korea J.-H. Kim Division of Rheumatology, Department of Internal Medicine, Korean University College of Medicine, Seoul, Republic of Korea E.-K. Lee School of Pharmacy, Sungkyunkwan University, Suwon, Republic of Korea
diagnosed with AVN. Patients who developed AVN in 2008–2010 were compared with SLE patients who did not develop AVN to identify any risk factors. The prevalence of AVN among SLE patients (2006–2010) was 31.5–34.2 per 1,000 persons and was similar in all the years studied. The incidence per 1,000 persons of AVN among SLE patients was 8.6 [95 % confidence interval (95 % CI) 6.9–10.3] in 2008, 9.8 (95 % CI 8.0–11.6) in 2009, and 8.4 (95 % CI 6.8–10.0) in 2010. Regression analysis indicated that taking an oral corticosteroid [odds ratio (OR) 2.12, 95 % CI 1.39–3.23] or an intravenous corticosteroid (OR 1.5, 95 % CI 1.2–1.89) was significantly associated with AVN. In addition, AVN was associated with use of immunosuppressive agents (OR 2.12, 95 % CI 1.66–2.72), hydroxychloroquine (OR 1.4, 95 % CI 1.09–1.81), and lipid-lowering agents (OR 1.78, 95 % CI 1.24–2.57) among the prescribed medications, and with hypertension (OR 1.39, 95 % CI 1.08–1.79) among the comorbidities. The prevalence and incidence of AVN among SLE patients, which were 31.5– 34.2 and 8.4–9.8 per 1,000 persons, respectively, may be representative of the entire population of symptomatic AVN patients with SLE in Korea. AVN is associated with the use of corticosteroids, immunosuppressants, hydroxychloroquine, lipid-lowering agents, and with hypertension. Studies of large, prospective cohorts are needed to confirm these results. Keywords Avascular necrosis · Systemic lupus erythematosus · Prevalence · Incidence · Risk factor
Introduction Avascular necrosis (AVN) is a well-known manifestation of systemic lupus erythematosus (SLE), which causes
13
significant joint pain and limitation of motion, eventually decreasing the patient’s quality of life, particularly in young patients [1]. The pathogenesis of AVN can be partially explained by subchondral bone necrosis due to interruption of the blood supply from the end artery to the bone [2]. SLE itself and the drugs used to treat it, such as corticosteroids, may accelerate this impairment of the blood supply [3, 4]. The reported prevalence of AVN in SLE patients is 3–30 % if only symptomatic patients are included, and up to 40 % when asymptomatic patients are also included [1, 3, 5–10]. This wide variation in previous estimates is attributable to differences in the definition of AVN, the diagnostic methodologies, and the patient populations studied. Since most of the studies involved only a small number of SLE patients and were carried out in a single center, the results may not accurately represent the characteristics of the entire SLE patient population. Therefore, we evaluated the prevalence and incidence of AVN among SLE patients in 2006–2010, based on the large Korean National Health Insurance (NHI) claims database. The possible risk factors for AVN were also investigated.
Rheumatol Int
four aforementioned characteristics was defined as the index date for patients with SLE. This algorithm was developed previously to estimate the prevalence and incidence of SLE using the claims database, and its comparability ratio and positive and negative predictive values were 1.0, 75, and 81 %, respectively. Prevalence of AVN AVN cases, among the SLE-prevalent cases, were defined as having at least one diagnosis of AVN for any reason (ICD-10 codes M870, M870x, M871, M871x, M872, M872x, M873, M873x, M878, M878x, M879, M879x, M905, and M905x). The accuracy of such a diagnosis was validated by Kang et al. [15]. The annual prevalence of AVN in 2006–2010 was calculated by dividing the number of AVN-prevalent cases with at least one diagnostic claim by the number of SLE-prevalent cases during the same year. That is, the year-specific numerator was subjects who were prevalent cases in the specific calendar year, and the denominator was prevalent SLE patients in the specific calendar year, as described previously [14]. Incidence of AVN
Patients and methods Data source This study was conducted using the Korean NHI claims database. This is a centralized claims database managed by the Health Insurance Research and Assessment Agency (HIRA), which is affiliated with the National Health Insurance Corporation. It contains medical claims data for the entire Korean population [11–13]. Both the inpatient and outpatient claims from the 2006–2010 NIH database were used. Demographic information (sex, age) and enrollee medical claims information, including comorbidities and prescribed medications, were collected. This study was approved by the Care Record Supply Committee of the HIRA. Since personal identification numbers in claims data were encoded so as to guarantee anonymity, the study protocol did not require approval from an institutional review board. Identification of SLE cases SLE-prevalent cases were defined as (1) hospitalization for SLE [code M32 in the International Classification of Diseases, 10th revision (ICD-10)], or (2) concomitant prescription of an immunosuppressant and antimalarial drugs, or (3) examination by anti-dsDNA antibody (≥2) or complement tests (≥2) during the applicable calendar year [14]. The earliest date of a medical claim based on any of the
13
The annual incidence of AVN in 2008–2010 was calculated by dividing the number of incident AVN cases by the number of SLE-prevalent cases that had not been diagnosed with AVN previously. The accuracy of the annual incidence of AVN was improved by excluding cases with an AVN diagnostic code during 2006–2007. That is, we defined subjects who had no AVN for the preceding 2 years (e.g., 2006–2007) and met the code in the year (e.g., 2008) as incident cases in the year (e.g., 2008). With the exception of those for which the AVN diagnostic code was assigned prior to the index date of SLE, all subjects meeting the AVN case definition during 2008–2010 were identified as AVN incident cases. The earliest date of a medical claim with an AVN diagnostic code was defined as the index date of AVN, and each subject was included in the calculation of annual incidence. Possible risk factors for AVN The baseline characteristics of SLE cases were compared in those with and without newly developed AVN in 2008– 2010. These characteristics were obtained from the medical claim on the AVN index date for cases with AVN and from the latest medical claim for those without AVN. Comorbid conditions and prescribed medications were also assessed during the 1-year pre-index period in the cases with AVN, and during the 1-year period prior to the latest claim in the cases without AVN. Comorbid conditions such as diabetes
Rheumatol Int Table 1 Crude and standardized prevalences of avascular necrosis (AVN) in patients with systemic lupus erythematosus (SLE) SLE, n
AVN, n
Crude prevalence per 1,000 persons (95 % CI)
Standardized prevalencea (95 % CI)
2006 2007 2008 2009
10,080 10,771 11,609 12,351
335 347 366 409
33.2 (29.6, 36.8) 32.2 (28.8, 35.6) 31.5 (28.3, 34.7) 33.1 (29.9, 36.3)
32.5 (29.2, 35.8) 31.9 (28.6, 35.2) 31.5 (28.3, 34.7) 33.2 (29.9, 36.5)
2010
13,316
456
34.2 (31.1, 37.3)
34.9 (31.5, 38.3)
Data are expressed as number (95 % confidence intervals) 95 % CI 95 % confidence interval a
All data are standardized for age and sex using the 2008 SLE population
mellitus, hypertension, osteoporosis, and lupus nephritis were included as potential risk factors or variables of interest to the study, as were prescribed medications, including intravenous and oral corticosteroids, immunosuppressive agents (cyclophosphamide, mycophenolate mofetil, cyclosporine, and azathioprine), hydroxychloroquine, and lipid-lowering agents. Among the intravenous corticosteroids, we included only methylprednisolone as we wished to focus on the impact of high-dose steroid injections on the development of AVN. Statistical analysis Crude and standardized prevalence and incidence data were adjusted for sex and age using the 2008 SLE population, and their 95 % confidence intervals (CIs) were calculated. Differences between cases with and without AVN were analyzed using t tests for continuous variables and chi-squared tests or Fisher’s exact tests for categorical variables. Logistic regression was used to develop a multivariate model with AVN as the outcome and age, sex, hypertension, lupus nephritis, intravenous corticosteroid, oral corticosteroid, hydroxychloroquine, immunosuppressive agent, and lipid-lowering agent as potential associated factors. However, other possible clinical and laboratory variables (Raynaud’s phenomenon, smoking, pregnancy, thrombosis, or antiphospholipid antibodies, etc.) were not considered as co-variables since the necessary information was not available in our claims database. Sensitivity analyses of prevalence and incidence were conducted since false positives were possible in cases where only diagnostic codes were used. Sensitivity analyses were limited to those patients with further evidence of SLE, based on the following additional medical claims: (1) hospitalization on the index date, (2) the presence of an orthopedist visit on the index date or during the year following the index date, or (3) the presence of a radiologic study [X-ray or magnetic resonance imaging (MRI)] on the index date, or (4) the presence of any one of the above. All tests were two sided, and the level of statistical significance
Fig. 1 Prevalence of AVN in 2009, stratified according to sex and age
was set at p
