Inflammatory bowel disease (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), is caused by idiopathic chronic inflammation in the intestinal tract and is characterized by a waxing and waning clinical course.^1,2 Currently, several immunomodulators and biologic agents that suppress the immune response are prescribed as the mainstream treatment for IBD to control the chronic inflammation in the intestinal tract and achieve a remission state.^3-7 The introduction of immunosuppressive agents can improve intestinal symptoms and significantly reduce IBD-related complications in many patients with IBD. However, frequent, long-term, and combined use of immunosuppressive agents can lead to critical clinical issues, such as opportunistic infections, in IBD patients.^3,8-12

Pneumocystis jirovecii pneumonia (PJP) is a fungal infection that most commonly occurs in immunocompromised hosts, such as those with cancer or human immunodeficiency virus (HIV) infection, transplant recipients, or those taking immunosuppressive medications.^13,14 PJP is rare but may be fatal once contracted.^15,16 Previous studies have reported that the incidence of PJP in IBD patients is higher compared to that in the general population.^15,17 However, there is still a lack of consensus on the issue of PJP prophylaxis in IBD patients.^18-20 European Crohn’s Colitis Organisation guidelines recommend primary PJP prophylaxis for patients on triple immunosuppressive therapy and recommend its consideration in patients on dual immunosuppressive therapy, especially if one is a calcineurin inhibitor.²¹ However, the level of evidence for this statement is low, and it does not reflect the demographic or other underlying diseases of IBD patients. In addition, to the best of our knowledge, only a few studies have investigated the risk factors and clinical prognosis for PJP in IBD patients.^8,22-24

Therefore, in the present study, we aimed to investigate the risk factors for PJP in IBD patients and their clinical prognosis using claims data from Korea.

1. Data source and study design

This was a cohort study using claims data from the National Health Insurance Service (NHIS) in South Korea. The NHIS is a government-operated health insurance service that covers approximately 97% of the Korean population.²⁵ This service provides all medical services, from diagnosis to treatment, to its subscribers. The NHIS claims database collects records of medical services, including patient demographics, diagnosis, prescription drugs, and medical institutions. The present study was conducted using data from 2010 to 2017 NHIS claims. This information was anonymized, and as a retrospective study, patients' informed consent was waived. The Institutional Review Board at Seoul National University Bundang Hospital approved this research (IRB number: X-1711-435-905).

2. Definitions

The definition of IBD was based on the International Classification of Disease, the tenth revision (ICD-10) codes and the rare intractable disease (RID) registration program in Korea, which was designed to provide medical cost benefits for patients with certain rare diseases including IBD. For registration of RID (V code), specific diagnostic criteria should be fulfilled, and a physician’s certification is required. Previous studies have demonstrated the accuracy of incident IBD provided by both the V and ICD-10 codes in Korea.^26,27 CD patients were defined as those with the ICD-10 code K50 and RID code V130, while UC patients were defined as those with the ICD-10 code K51 and RID code V131.²⁸

PJP was defined in patients with the ICD-10 code B59 or who were prescribed a preferred or alternative therapy aimed at treating PJP for at least 7 days following chest computed tomography. The preferred regimen was 15–20 mg/kg/day trimethoprim plus 75–100 mg/kg/day sulfamethoxazole, and the alternative regimens were 600 mg every 8 hours or 300–460 mg every 6 hours clindamycin plus 30 mg base/day primaquine and 4 mg/kg/day pentamidine.²⁹ To eliminate unintentional lag-time bias between the diagnosis of IBD and PJP events, patients diagnosed with PJP who had passed at least 1 year from the date of diagnosis of IBD were included. Meanwhile, patients with a history of HIV infection (n=16) and patients with a previous history of malignancy (n=28) were excluded from the study.

We extracted the demographic information, underlying disease, and medication history for the included patients using the ICD-10 code and prescription code from the database. The age at diagnosis of IBD was classified into A1 (<16 years), A2 (17–40 years), and A3 (>40 years), based on the Montreal classification of CD. Previously identified risk factors for PJP infection in patients without HIV infection include leukopenia, hypoalbuminemia, advanced age, and comorbidities.^23,30-33 Based on a review of the literature, we paid attention to comorbidities that may be risk factors for PJP and extracted the information on comorbidities using the ICD-10 codes; diabetes mellitus (DM; E11-14), hypertension (I10-13), dyslipidemia (E78), and chronic obstructive pulmonary disease (COPD; J43-44). The ever use of each medication was defined as a case in which the drug was prescribed at least once following a diagnosis of IBD. When evaluating the association between the drug and the incidence of PJP, drug use was defined in cases where the drug was prescribed within 3 months before the onset of PJP. High-dose corticosteroid was defined as 20 mg or more of steroid prescribed for 2 weeks or more based on prednisolone.²¹ For cases of PJP occurrence, dual or triple immunosuppressive therapy was defined as the concurrent prescription of two or three of the following medications within 3 months prior to the onset of PJP; steroid, calcineurin inhibitor (cyclosporine, tacrolimus), immunomodulator (azathioprine, 6-mercaptopurine, and methotrexate), and biologics (infliximab, adalimumab, golimumab, and vedolizumab). For cases without PJP, the definition was based on the concurrent prescription of two or three of the mentioned medications without any specific time constraint. When defining dual or triple immunosuppressive therapy, the inclusion of steroids was determined based on their usage, regardless of the dosage. PJP prophylaxis was defined as cases in which a prophylactic regimen of sulfamethoxazole/trimethoprim was administered (one-strength tablet orally daily or one double-strength tablet orally three times weekly).²⁹

3. Outcomes

The outcome of the study was to evaluate the risk factors related to the incidence of PJP in IBD patients and their clinical prognosis. We extracted cases of PJP from newly diagnosed IBD patients between 2010 and 2017, according to the definitions mentioned above. We further assessed covariates, including demographic data, IBD-related medications, and underlying diseases, in IBD patients that could be attributed to the occurrence of PJP. In addition, prognostic factors such as intensive care unit (ICU) admission, length of ICU hospitalization, and 30-day mortality were investigated.

4. Statistical analysis

Categorical variables were presented as numbers and percentages, while continuous variables were presented as the mean standard deviation or median (interquartile range). Analyzing the differences in characteristics with the chi-square test for categorical variables and the t-test or Mann-Whitney U test for continuous variables. Risk factors for the occurrence of PJP in IBD patients were evaluated using the Cox proportional hazard model, and the results were presented as a hazard ratio (HR) with a 95% confidence interval (CI). The adjusted models were applied to adjust for covariates and to evaluate the effect of dual and triple immunosuppressive therapy on PJP incidence. The models were adjusted for age, sex, and underlying diseases. A p-value less than 0.05 was considered statistically significant. SAS version 9.4 was used for statistical analysis (SAS Institute, Cary, NC, USA).

1. Characteristics of patients with IBD

A total of 39,462 IBD patients newly diagnosed from 2010 to 2017 (12,623 CD and 26,839 UC) were included in the study. The mean follow-up duration was 4.6 years. The mean age at the time of diagnosis of IBD was 38.6±17.6 years, and 14,535 patients (36.8%) were female. IBD patients who had ever been treated with systemic corticosteroids, immunomodulators, and anti-tumor necrosis factor agents accounted for 84.5%, 37.3%, and 17.9% of all patients, respectively. Table 1 presents the baseline characteristics of the included patients.

Table 1 . Baseline Characteristics of Patients with IBD.

Variable	IBD (n=39,462)	CD (n=12,623)	UC (n=26,839)
Demographic characteristics
Age at diagnosis	38.6±17.6	29.9±16.3	42.7±16.7
A1 (16 yr)	3,120 (7.9)	2,222 (17.6)	898 (3.4)
A2 (17–40 yr)	19,230 (48.7)	7,627 (60.4)	11,603 (43.2)
A3 (>40 yr)	17,112 (43.4)	2,774 (21.9)	14,338 (53.4)
Female sex	14,535 (36.8)	3,790 (30.0)	10,745 (40.0)
Comorbidities
Diabetes mellitus	1,731 (4.3)	375 (2.9)	1,356 (5.1)
Hypertension	5,107 (12.9)	942 (7.5)	4,165 (15.5)
Dyslipidemia	3,408 (8.6)	590 (4.7)	2,818 (10.5)
COPD	2,316 (5.9)	690 (5.5)	1,626 (6.1)
Ever use of medication
5-Aminosalicylic acid	36,675 (92.9)	11,106 (88.0)	25,569 (95.3)
Systemic corticosteroids	33,346 (84.5)	10,551 (83.6)	22,795 (84.9)
High-dose steroids	8,666 (22.0)	3,972 (31.5)	4,694 (17.5)
Calcineurin inhibitor	228 (0.6)	95 (0.8)	133 (0.5)
Immunomodulator	14,717 (37.3)	8,866 (70.2)	5,851 (21.8)
Anti-TNF agent	7,081 (17.9)	4,361 (34.6)	2,720 (10.1)
Vedolizumab	472 (1.2)	163 (1.3)	309 (1.2)
Dual therapy	9,716 (24.6)	5,547 (43.9)	4,169 (15.5)
Triple therapy	1,842 (4.7)	920 (7.3)	922 (3.4)

Data are presented as mean±SD or number (%)..

IBD, inflammatory bowel disease; CD, Crohn’s disease; UC, ulcerative colitis; COPD, chronic obstructive pulmonary disease; TNF, tumor necrosis factor..

2. Clinical features of patients with PJP

Of the 39,462 patients with IBD, 84 (31 CD and 53 UC) patients were diagnosed with PJP. The incidence of PJP in UC was 0.43 per 1,000 person-years and in CD it was 0.53 per 1,000 person-years. The mean age of patients at the time of diagnosis of PJP was 59.0±19.5 years. At the time of the diagnosis of PJP, the median duration of IBD was 31.5 months (interquartile range, 21.7 to 53.9 months). Of the 84 patients, 31 (36.9%) received primary PJP prophylaxis. The demographic and clinical characteristics of IBD patients with PJP are presented in Table 2.

Table 2 . Clinical Features of IBD Patients with PJP.

Variable	IBD (n=84)	CD (n=31)	UC (n=53)
Demographic characteristics
Age at diagnosis of PJP, yr	59.0±19.5	53.4±20.4	62.2±18.5
Duration of IBD at diagnosis of PJP, mo	31.5 (21.7–53.9)	30.2 (17.3–44.5)	34.2 (22.9–56.1)
Female sex	33 (39.3)	14 (45.2)	19 (35.8)
Comorbidities
Diabetes mellitus	16 (19.1)	5 (16.1)	11 (20.8)
Hypertension	30 (35.7)	11 (35.5)	19 (35.9)
Dyslipidemia	18 (21.4)	6 (19.4)	12 (22.6)
COPD	17 (20.2)	4 (13.0)	13 (24.5)
Medication used within 3 mo of PJP diagnosis
5-Aminosalicylic acid	70 (83.3)	23 (74.2)	47 (88.7)
Systemic corticosteroids	72 (85.7)	25 (80.7)	47 (88.7)
High-dose steroid s	40 (47.6)	16 (51.6)	24 (45.3)
Calcineurin inhibitor	6 (7.1)	3 (9.7)	3 (5.7)
Immunomodulator	34 (40.5)	16 (51.6)	18 (34.0)
Anti-TNF agent	18 (21.4)	8 (25.8)	10 (18.9)
Vedolizumab	2 (2.4)	0	2 (3.8)
Dual therapy	21 (25.0)	7 (22.6)	14 (26.4)
Triple therapy	11 (13.1)	6 (19.3)	5 (9.4)
PJP prophylaxis	31 (36.9)	9 (29.0)	22 (41.5)

Data are presented as mean±SD, median (interquartile range), or number (%)..

IBD, inflammatory bowel disease; PJP, Pneumocystis jirovecii pneumonia; CD, Crohn’s disease; UC, ulcerative colitis; COPD, chronic obstructive pulmonary disease; TNF, tumor necrosis factor..

3. Risk factors for PJP in IBD patients

Multivariable Cox regression analysis revealed that, in patients with CD, an A3 (age at diagnosis >40 years) classification (HR, 6.12; 95% CI, 1.58 to 23.80) and ever use of high-dose steroids (HR, 3.95; 95% CI, 1.78 to 8.77) and calcineurin inhibitor (HR, 4.89; 95% CI, 1.36 to 17.61) were independent risk factors for PJP. Conversely, DM (HR, 2.51; 95% CI, 1.19 to 5.31), COPD (HR, 3.41; 95% CI, 1.78 to 6.52), and ever use of high-dose steroids (HR, 3.41; 95% CI, 1.80 to 6.48) and calcineurin inhibitor (HR, 4.34; 95% CI, 1.29 to 14.68) were significantly associated with the incidence of PJP in UC patients (Table 3).

Table 3 . Multivariable Analysis to Identify Risk Factors for PJP in IBD Patients.

Variable	CD						UC
	Univariable			Multivariable			Univariable			Multivariable
	HR (95% CI)	p-value		HR (95% CI)	p-value		HR (95% CI)	p-value		HR (95% CI)	p-value
Age at diagnosis
A1 (≤16 yr)	1 (reference)			1 (reference)			1 (reference)			1 (reference)
A2 (17–40 yr)	0.59 (0.15–2.36)	0.46		0.73 (0.18–3.00)	0.67		0.24 (0.05–1.18)	0.08		0.36 (0.07–1.81)	0.21
A3 (>40 yr)	6.28 (1.88–21.00)	<0.01		6.12 (1.58–23.80)	<0.01		1.46 (0.36–6.03)	0.60		1.60 (0.37–6.93)	0.53
Female sex	1.93 (0.95–3.91)	0.07		1.14 (0.55–2.36)	0.73		0.83 (0.47–1.45)	0.51		0.90 (0.51–1.59)	0.71
Comorbidities
Diabetes mellitus	6.73 (2.58–17.52)	<0.01		1.61 (0.54–4754)	0.39		5.10 (2.63–9.91)	<0.01		2.51 (1.19–5.31)	0.02
Hypertension	7.39 (3.54–15.42)	<0.01		1.55 (0.60–4.02)	0.36		3.14 (1.79–5.50)	<0.01		1.24 (0.64–2.40)	0.52
Dyslipidemia	5.49 (2.25–13.40)	<0.01		1.17 (0.42–3.31)	0.77		2.74 (1.44–5.22)	<0.01		1.10 (0.53–2.32)	0.80
COPD	2.67 (0.94–7.64)	0.07		1.24 (0.42–3.63)	0.70		5.30 (2.83–9.90)	<0.01		3.41 (1.78–6.52)	<0.01
Ever-use of medication
5-Aminosalicylic acid	0.36 (0.16–0.81)	0.01		0.57 (0.23–1.40)	0.21		0.37 (0.16–0.86)	0.02		0.58 (0.23–1.44)	0.24
High-dose corticosteroids	2.18 (1.07–4.40)	0.03		3.95 (1.78–8.77)	<0.01		3.67 (2.14–6.30)	<0.01		3.41 (1.80–6.48)	<0.01
Immunomodulator	0.44 (0.22–0.88)	0.02		0.95 (0.39–2.32)	0.91		1.74 (0.98–3.07)	0.06		1.14 (0.56–2.36)	0.72
Calcineurin inhibitor	13.48 (4.09–17.61)	<0.01		4.89 (1.36–17.61)	0.02		11.05 (3.45–35.44)	<0.01		4.34 (1.29–14.68)	0.02
Anti-TNF agent	0.64 (0.28–1.42)	0.27		0.91 (0.35–2.40)	0.86		1.98 (1.00–3.94)	0.05		1.04 (0.43–2.55)	0.37

PJP, Pneumocystis jirovecii pneumonia; IBD, inflammatory bowel disease; CD, Crohn’s disease; UC ulcerative colitis; HR, hazard ratio; CI, confidence interval; COPD, chronic obstructive pulmonary disease; TNF, tumor necrosis factor..

4. Risk analysis of dual or triple immunosuppressive therapy for the incidence of PJP in IBD patients

For CD patients, we identified no association between dual therapy and the incidence of PJP, whereas triple therapy showed a significant association with the incidence of PJP in the adjusted model (HR, 5.69; 95% CI, 2.32 to 14.48). In UC patients, both dual therapy (HR, 2.53; 95% CI, 1.36 to 4.70) and triple therapy (HR, 3.90; 95% CI, 1.54 to 9.88) were significantly associated with the occurrence of PJP (Table 4).

Table 4 . Adjusted Risk Analysis of Dual and Triple Immunosuppressive Therapy for IBD Patients with Pneumocystis jirovecii Pneumonia.

Adjusted model	CD			UC
	HR (95% CI)*	p-value		HR (95% CI)*	p-value
Dual therapy†	0.69 (0.29–1.68)	0.42		2.53 (1.36–4.70)	<0.01
Triple therapy†	5.69 (2.32–14.48)	<0.01		3.90 (1.54–9.88)	<0.01

IBD, inflammatory bowel disease; CD, Crohn’s disease; UC ulcerative colitis; HR, hazard ratio; CI, confidence interval..

*Cox proportional hazards regression was used to estimate the HR and 95% CI with adjustment for age, sex, diabetes mellitus, hypertension, dyslipidemia, and chronic obstructive pulmonary disease; ^†Dual and triple therapies were predefined in the manuscript as the simultaneous prescription of two or three of the following drugs, respectively: steroids, calcineurin inhibitors, immunomodulators, and biologics..

5. Clinical course of PJP in IBD patients

During the follow-up period, 31 of 12,632 patients with CD developed PJP. Among them, 27 patients (87.1%) were hospitalized, and the mean duration of hospitalization was 21.3 days. Eleven patients (35.5%) were treated in the ICU, and the mean ICU length of stay was 27.4 days. Four patients (12.9%) died within 30 days following the diagnosis of PJP. The mean duration between PJP diagnosis and death was 5.0±6.2 days. Meanwhile, 53 out of 26,839 patients with UC developed PJP. Overall, 43 patients (81.1%) were hospitalized, and the mean duration of hospitalization was 10.16 days. Twelve patients (22.6%) were treated in the ICU, and the mean ICU length of stay was 24.0 days. Six patients (11.3%) died within 30 days following the diagnosis of PJP. The mean duration between PJP diagnosis and death was 1.5±2.8 days. Table 5 provides a summary of the PJP prognosis in IBD patients.

Table 5 . Clinical Prognosis of Pneumocystis jirovecii Pneumonia in Inflammatory Bowel Disease Patients.

Variable	CD (n=31)	UC (n=53)
Hospitalization
Yes	27 (87.1)	43 (81.1)
Duration, day	21.3±19.9	10.2±10.5
ICU care
Yes	11 (35.5)	12 (22.6)
Duration, day	27.4±18.5	24.0±17.6
30-day mortality
Yes	4 (12.9)	6 (11.3)
Duration from diagnosis of PJP to death, day	5.0±6.2	1.5±2.8

Data are presented as number (%) or mean±SD..

CD, Crohn’s disease; UC, ulcerative colitis; ICU, intensive care unit..

This large-scale population-based cohort study in South Korea comprehensively investigated the risk factors for the development of PJP in patients with IBD. In CD patients, A3 (age of diagnosis >40 years) classification and ever use of high-dose steroids and a calcineurin inhibitor were associated with the occurrence of PJP, while in UC patients, DM, COPD, and ever use of high-dose steroids and a calcineurin inhibitor were independent risk factors for PJP occurrence. The adjusted model for covariate revealed that triple immunosuppressive therapy in both CD and UC was significantly associated with an increased risk of PJP, while dual therapy was associated with only UC. In addition, among the 84 PJP patients identified in this study, 23 (27%) were treated in ICU, and 10 (12%) died within 30 days, suggesting that PJP is a life-threatening infectious disease.

We investigated the risk factors for PJP in patients with IBD to better define the subgroups that may derive the greatest benefit from PJP prophylaxis. Based on an analysis of the existing literature, advanced age, lymphopenia, high-dose steroid use, triple immunosuppression, and certain comorbidities, such as COPD, have been shown to influence the incidence of PJP.^{21-23,30,31,34} The risk factors for PJP in this study were comparable to those in previous studies.^8,23,24 Our present study demonstrated that COPD and DM were associated with an increased risk of PJP in patients with UC. Patients with pulmonary comorbidities, including COPD, were more likely to develop PJP.^22,31 In addition, several previous studies have shown an association between IBD and COPD that acts via the gut-lung axis.^34-38 Schuijt et al.³⁵ further reported that mice with depleted gut microbiota have an increased risk of bacterial infection in the lung. Collectively, it can be assumed that dynamic changes in gut and lung microorganisms in patients with IBD and COPD are associated with an increased risk of PJP. DM was also found to be a risk factor for PJP, and it is well-known that chronic comorbidities such as DM are associated with an increased risk of infection.^33,39 It is unclear whether impaired immunity related to underlying comorbidities may increase PJP risk. Further studies are required to confirm these findings.

Immunosuppressive drugs, including biologics, are used more frequently in patients with IBD.^3,4 Our study is noteworthy in that we demonstrated that dual or triple immunosuppression therapy for UC and triple immunosuppression therapy for CD were significantly associated with an increased risk of PJP episodes. In particular, the risk of developing PJP in patients with IBD was approximately 5-fold higher in patients treated with triple therapy than in those who were not. Moreover, a history of high-dose steroid and calcineurin inhibitor use was significantly associated with the incidence of PJP in patients with IBD. High-dose steroids suppress CD4⁺ T cell responses, a major risk factor for PJP in non-HIV immunosuppressed populations.^23,40 Calcineurin inhibitors can contribute to the occurrence of PJP by inhibiting interleukin-2 production by helper T cells.^17,41,42 These findings are mostly consistent with the indications for PJP prophylaxis recommended by the European Crohn’s Colitis Organisation guideline.²¹ However, unlike the European Crohn’s Colitis Organisation guideline, the results of this study suggest that even during dual immunosuppressive therapy, PJP prophylaxis may be considered more vigorous in patients with UC. According to the findings of this study, PJP prophylaxis may be considered for patients who are on or planning to be treated with three immunosuppressive agents for CD and two or more for UC. A possible explanation of this difference between UC and CD in our study was challenging. Even after searching various literature, it was difficult to find plausible explanation. Therefore, further research based on a larger number of cases are needed to confirm our findings. Meanwhile, as the aim of the study was not to evaluate the risk-benefit of PJP prophylaxis for high-risk patients with IBD, prospective studies are warranted to suggest this strongly.

PJP is a known problem, with a high mortality rate (17.9% to 30.0%) in patients with IBD.^15,43,44 Although our study demonstrated that 30-day mortality occurred in only 10 patients (12.9%), given the results of previous studies, it can be assumed that these infectious complications cause significant mortality. In HIV-negative patients, 31% to 60% of the patients were admitted to the ICU, and the overall mortality was 19% to 47%.^40,45 Tachypnea, tachycardia, a high C-reactive protein level, and the requirement for mechanical ventilation were all linked to a poor prognosis.^45,46 Further investigation is needed on risk factors related to poor prognosis in IBD patients with PJP.

The strength of the present study is that it is the first population-based cohort study investigating the risk of PJP in Asia. In addition, we analyzed a large sample size. We also accounted for comorbidities through administrative data. Identifying IBD patients at risk for PJP infections may provide better patient outcomes through prophylactic strategies. Moreover, our study was the first to examine clinical prognosis, including hospitalization and ICU care, in patients with IBD who developed PJP. On the other hand, there are some limitations to our study. First, we were unable to include the microbiological diagnosis of P. jirovecii in the definition of PJP in the study due to the inherent limitation of the study design. However, in real clinical practice, there are cases where it is difficult to obtain respiratory specimens or establish a microbiological diagnosis. In such cases, a presumptive treatment for PJP may be initiated based on a combination of clinical symptoms and radiologic findings. Therefore, our proposed definition of PJP may be more realistic. Second, the severity of CD and UC could not be demonstrated as the database contained no information regarding the disease severity. Third, we could not evaluate the effects of newer biologics, such as ustekinumab and JAK inhibitors, which have recently been approved. Lastly, our study primarily focused on identifying risk factors for the development of PJP, and as such, we are unable to provide conclusive evidence regarding the optimal timing of PJP monitoring or the impact of prophylaxis for PJP.

In conclusion, this large-scale, nationwide study revealed certain risk factors for PJP in patients with IBD. There were differences in risk factors for PJP between patients with UC and CD, but a history of high-dose steroid and calcineurin inhibitor use was associated with an increased risk of PJP in both conditions. In addition, the risk of PJP infection was much higher in patients treated with triple immunosuppressive therapy than in those not being treated. Given the potentially high morbidity and mortality associated with PJP, we would recommend continuing to assess the need for primary PJP prophylaxis on a case-by-case basis, particularly in older patients with COPD and diabetes and in patients taking triple immunosuppressive agents.

This work was supported by funding from the Seoul National University Bundang Hospital Research Fund (grant number 09-2018-0005).

We would like to thank Sang-Hyeon Park (College of Medicine, The Catholic University of Korea, Seoul, Korea) for the assistance with statistical analysis.

No potential conflict of interest relevant to this article was reported.

Study concept and design: S.W.H., H.Y. Data acquisition and analysis: K.D.H., S.W.L. Data interpretation: J.Y., S.W.H., H.Y. Drafting of the manuscript: J.Y., S.W.H. Critical revision of the manuscript: H.Y. Administrative, technical, or material support: C.M.S., Y.S.P., N.K., D.H.L., J.S.K. Approval of final manuscript: all authors.

The data underlying this article were provided with permission from the Korean National Health Insurance Services. All data, except the results, cannot be shared publicly due to the policy of the national health authorities.