Data source

We used the 2014 National Health Insurance Database (NHID) of Korea. The National Health Insurance system is a mandatory health insurance program that covers all Korean residents, so the NHID can play a role as infrastructure for epidemiological studies representing the whole population [14]. The data resource profile of NHID has been published elsewhere [15]. The NHID can be used after approval by the review committee of the National Health Insurance Corporation, and it is accessed in a dedicated data analysis room located within the National Health Insurance Corporation building. Details of this process and a guide for accessing the data are available at http://nhiss.nhis.or.kr/bd/ab/bdaba032eng.do.

The prevalence of EIMs in the general population of Korea was measured using the 2014 National Patient Sample data from the Health Insurance Review and Assessment Service (HIRA). The HIRA randomly samples 3% of all Korean healthcare recipients every year, and provides the resulting databases for research purposes [16]. We obtained the National Patient Sample data (HIRA-NPS-2014-0127) after making a request via the website of the Healthcare Big Data Hub (http://opendata.hira.or.kr) operated by the Korean Health Insurance Review and Assessment Service.

The prevalence of EIMs in the general population of Korea was measured using the same International Classification of Diseases, Tenth Edition (ICD-10) codes as those used in IBD patients [17].

Inclusion criteria of patients and the comparison group

In this study, CD and UC were defined according to the corresponding ICD-10 codes (CD with K50, and UC with K51) and registration codes from the rare/intractable diseases (RID) patient-support program (CD: V130, UC: V131). The RID registration program was established in 2009 in order to lessen the burden on patients who are suffering from these diseases in Korea. In order for these IBD patients to be eligible for benefits, their doctor needs to submit a separate form to prove that the condition has been confirmed. Although the validity of these codes has not been evaluated in a separate publication, it was confirmed through a review of the medical records in a study by Kim et al. [10]. As a denominator to investigate the prevalence of EIM, IBD patients from the NHID 2014 data between January 1, 2014 and December 31, 2014 were defined according to the following criteria: 1) a diagnosis with the corresponding ICD-10 codes (CD: K50, UC: K51) in at least 1 claim, and 2) the presence of a registration code from the RID patient-support program (CD: V130, UC: V131) in at least 1 claim. Subjects who were under 19 years of age or older than 75 years of age were excluded. To calculate the prevalence of EIMs in the general population of Korea, the same criteria were applied in the 2014 National Patient Sample data.

Age, sex, and drug utilization patterns were compared between CD and UC patients. The variables considered in the analysis of drug utilization patterns were anti-inflammatory drugs (corticosteroids, mesalazine, and balsalazide), immunosuppressants (methotrexate, azathioprine, sulfasalazine, and 6-mercaptopurine) and tumor necrosis factor (TNF)-alpha inhibitors (adalimumab, infliximab, golimumab, and etanercept). To define the age and sex of patients, we used the National Health Insurance subscriber information. In the NHID, this information was obtained from the 2014 subscriber information database. The NPS did not have such a database, so the age of patients was determined based on their age recorded in the first claim in 2014. Anti-inflammatory drugs, immunosuppressants, and tumor necrosis factor (TNF)-alpha inhibitors were identified using the NHID formulary, and exposure to the drugs was defined as presence of more than 1 the prescription code between January 1, 2014 and December 31, 2014. EIMs in the analysis were categorized into 4 categories: ophthalmologic EIMs (scleritis, episcleritis, and iridocyclitis), hepatopancreaticobiliary EIMs (cholelithiasis, sclerosing cholangitis, and acute pancreatitis), dermatologic EIMs (aphthous stomatitis, psoriasis, erythema nodosum, pyoderma gangrenosum, and Sweet syndrome), and musculoskeletal EIMs (rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, sacroiliitis, osteoporosis, and osteomalacia). The corresponding ICD-10 codes are presented in S1 Table.

Statistical analysis

The baseline characteristics were summarized using the mean and standard deviation for continuous variables and frequency and proportion for categorical variables. The Student t-test was used for the comparison of continuous variables, and the chi-square test or the Fisher exact test was used for categorical variables as appropriate.

The prevalence of an EIM was calculated as the number of patients with that EIM divided by the number of IBD patients. The prevalence of each EIM was presented for CD and UC patients, respectively, with 95% confidence intervals (CIs). We used indirect standardization to compare EIM prevalence between IBD patients and the general population of Korea [18, 19]. The distribution of EIMs in the general population of Korea by age and sex is presented in S2 Table. To estimate the age/sex-standardized prevalence ratio, the number of expected events of each EIM in CD and UC patients was calculated by multiplying the age/sex-specific EIM prevalence in the general population by the number of patients in each age/sex stratum of CD and UC patients, and then summing the total number of expected events. The age/sex-standardized prevalence ratios with 95% CIs were calculated for each EIM as the number of expected events divided by the number of observed events. All statistical tests were 2-sided, and P values < 0.0038 (0.05/13, where 13 equals the number of tests) were considered to indicate statistical significance based on the Bonferroni correction to cope with multiple comparisons. SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA) was used for the statistical analysis.

Ethical considerations

This study protocol was exempted from review by the Institutional Review Board of Seoul National University Hospital and the Seoul National University College of Medicine (IRB number: E-1510-071-711).

Ophthalmologic EIMs

Following joints and the skin, the eye is the site of many immune-related EIMs [22]. According to previous studies conducted in India, Switzerland, and Turkey, ophthalmologic EIMs occur in up to 13% of IBD patients [23–25], and are more frequent in CD patients than in UC patients [26], although our results showed that they tended to be more common in UC patients on a frequency basis. However, this difference was not prominent when the standardized prevalence rate ratio was compared. Rather, the main inconsistency we found with previous studies is that iridocyclitis was more frequently observed than episcleritis. Previous studies have found episcleritis to be the most common ocular EIM, whereas uveitis was relatively rare [7]. In our study, iridocyclitis was more common, but another previous study conducted in Switzerland reported a higher prevalence rate for uveitis than was found in this study [24]. Thus, episcleritis may be less likely to be a comorbidity in Korean IBD patients.

Hepatopancreaticobiliary EIMs

Up to 50% of IBD patients may have hepatobiliary manifestations [7]. PSC has a well-known association with IBD. Seventy-five percent of PSC patients have UC, and 5%-10% of PSC patients have CD [27, 28]. Moreover, 5% of UC patients and 2% of CD patients develop PSC [29]. Ye et al. reported that the prevalence of PSC was 1.1% among 1849 Korean ulcerative colitis patients using medical records (from July 1977 to September 2009) of a tertiary hospital [30]. In our study, the prevalence of sclerosing cholangitis was one-tenth of the reported prevalence, but a tendency for a higher prevalence in UC patients than in CD patients was observed.

Cholelithiasis is a common EIM in IBD patients, and is frequently seen in cases of CD, especially those involving the ileum. It has been reported that up to 13%-34% of cases are caused by obstruction of the enterohepatic circulation, which is known to occur more frequently in IBD patients than in the general population of Sweden [31]. Our data indicated that this condition occurred more frequently in CD patients than in UC patients, and this tendency was more pronounced in comparison to the general population.

Acute pancreatitis is a common side effect of 6-mercaptopurine or azathioprine treatment. However, it is associated with gallstones and is more common in cases of CD. A retrospective multicenter study in Spain reported that acute pancreatitis occurred in 1.6% of IBD patients [32]. A Dutch study reported that acute pancreatitis was four times more common in CD patients and two times more common in UC patients than in the general population [33], which is very similar to our data.

Dermatologic EIMs

It has been reported that 2%-34% of IBD patients experience dermatologic EIMs [34]. In a study performed in Turkey, dermatologic EIMs were present in 9.3% of patients; erythema nodosum was present in 7.4% of patients, and pyoderma gangrenosum in 2.3% [35]. Erythema nodosum has been reported in 15% of CD patients and in 10% of UC patients [36], and it has been reported to be more common in women [36]. Our study also confirmed the tendency for this condition to be more common in CD patients and for there to be a higher prevalence in women. Pyoderma gangrenosum is a much rarer, more severe, and debilitating EIM that is more common in UC patients than in CD patients [37–39]. Its prevalence has been reported to be 1%-10% in UC patients and 0.5%-20% in CD patients [40].

Psoriasis is more common in CD patients, and is known to be more common in CD than in the general population of the United States [41]. Although the prevalence rates identified in our study were lower, the standardized prevalence rate ratio was higher in UC patients than in CD patients.

Oral lesions including aphthous stomatitis are known to be common in IBD patients, and are more common in CD patients [9, 22, 42]. In our data, the prevalence rate was 3%-4% in both CD and UC patients; this rate was significantly higher than that observed in the general population of Korea, but the magnitude of the difference was not large.

Acute febrile neutrophilic dermatosis (Sweet syndrome) is a rare dermatologic EIM [43–45]. The literature on its prevalence is scarce, and only case reports are available. Our data differ from previous reports in that it has been reported to be more common in women, but there were no female patients with this condition in our data. However, there were not enough cases in our data to assess the prevalence rate in a meaningful way.

Musculoskeletal EIMs

Traditionally, the main musculoskeletal EIM is seronegative arthralgia/arthritis [46]. Approximately 5%-10% of UC patients and 10%-20% of CD patients experience these symptoms [47]. Axial arthropathies are less common than peripheral arthralgia/arthritis, and have been reported to be experienced by 3% to 5% of IBD patients [22, 26, 48]. Asymptomatic sacroiliitis can be observed in radiological examinations in up to 52% of CD patients [49]. Although ankylosing spondylitis and isolated sacroiliitis are distinct diseases, sacroiliitis can be reported as ankylosing spondylitis in the health insurance claims data, so it is possible that the prevalence of sacroiliitis was underestimated and that of ankylosing spondylitis was overestimated. When ankylosing spondylitis and sacroiliitis were grouped as axial arthropathy and rheumatoid arthritis and psoriatic arthritis were grouped as peripheral arthralgia/arthritis, the trends reported in the existing literature were reproduced in our study.

Patients with IBD have a high risk of osteoporosis, with consequences including corticosteroid therapy, decreased physical activity, inflammation-related bone resorption, and dietary malabsorption of minerals [50]. As a result, patients with IBD have a greater risk of fractures than the general population of Korea, and this risk is higher in both men and women. The prevalence of osteoporosis has been found to be 7%-35% in CD patients [51–53] and 18% in UC patients [52]. In our data, the prevalence rate was higher in women and the standardized prevalence rate was higher in CD patients.

This study has the advantage of high representativeness because it used health insurance claim data including all Koreans. Due to some limitations, however, some precautions should be taken regarding the correct interpretation of the results. First, the validity of the diagnostic codes for EIMs should be carefully considered. There is a possibility that the prevalence rate was underestimated due to the low claim rate in routine clinical practice for erythema nodosum, pyoderma gangrenosum, and osteomalacia, while overestimation of EIMs could have occurred due to the inclusiveness of the ICD-10 codes for rheumatoid arthritis and iridocyclitis. For the comparison with the general population, however, there is little concern about underestimation and overestimation because the prevalence was measured in the same way in both databases.

Even though we calculated standardized prevalence ratios by adjusting for age and sex, other possible confounding variables were not considered. In particular, obesity has recently been reported to be associated with EIMs [54]; however, it was not possible to evaluate whether patients were obese using the claims database, because this information was not included. Moreover, no further examination was conducted of the relationships between drug utilization patterns and EIM prevalence in this study. Due to the limitation of the cross-sectional study design for evaluating drug effects, we suggest that the associations between EIMs and drug utilization should be carefully interpreted with evidence from studies with higher on the evidence hierarchy. A cohort study with a new-user design would be needed to provide relevant evidence, since the rarity of both IBD and some EIMs will limit the availability of appropriate patients for randomized controlled trials.

This study presents data on the prevalence of each EIM in IBD patients. Considering the possible impact of EIMs on the quality of life of the patients, the high prevalence presented in this study suggests that more attention and appropriate treatment are needed. The study also confirmed that the EIMs are more prevalent in Korean IBD patients than in the general population of Korea, which suggests that more attention should be paid to EIMs in IBD patients in everyday clinical practice, and that our results can be used for patient education.