Adverse Events Associated with Advanced Therapy Compared with Conventional Therapy in Patients with Inflammatory Bowel Disease: A Common Data Model Analysis Authors: Hyoung Il Choi, Myoungsuk Kim, Jung Rock Moon, Chang Won Jeong, Young Soo Kim, Sang Youl Rhee, Gyung-Min Park, Yoo Jin Lee, Dong Hoon Baek, Jae Myung Cha Published: 2026 ABSTRACT AbstractBackground/AimsAlthough clinical trials have reported adverse events (AEs) associated with advanced therapies (ATs) in patients with inflammatory bowel disease (IBD), the risks of AEs may be underestimated due to the selective nature of the study populations.MethodsThis retrospective, observational study from 13 hospitals using a common data model compared the risks of AEs between patients receiving ATs and those not receiving ATs. Cox proportional hazards models and Kaplan-Meier analyses were conducted following propensity score matching.ResultsPatients with IBD receiving ATs exhibited a significantly higher risk of tuberculosis (hazard ratio [HR], 2.86; 95% confidence interval [CI], 1.77 to 4.63; p<0.001) and depression/anxiety (HR, 1.58; 95% CI, 1.15 to 2.16; p=0.005) than those not receiving ATs. In subgroup analyses, use of anti-tumor necrosis factor (anti-TNF) agents was associated with increased risks of tuberculosis (HR, 3.74; 95% CI, 2.12 to 6.59; p<0.001) and depression/anxiety (HR, 1.49; 95% CI, 1.07 to 2.09; p=0.019), but a lower risk of osteoporosis (HR, 0.38; 95% CI, 0.15 to 0.99; p=0.040). No significant differences were observed in herpes zoster or malignancy risk between AT and non-AT groups. Small-molecule agents were not associated with increased risks of tuberculosis, osteoporosis, depression/anxiety, or herpes zoster. The malignancy incidence was lower in the small-molecule group (HR, 0.26; p=0.013), although the number of events was small.ConclusionsATs, particularly anti-TNF agents, were associated with an increased risk of tuberculosis and depression/anxiety in patients with IBD. Vigilant monitoring and management of potential AEs are crucial for optimizing treatment outcomes.Keywords: Adverse events, Biological products, Inflammatory bowel diseases INTRODUCTION The incidence and prevalence of inflammatory bowel disease (IBD), encompassing both Crohn’s disease and ulcerative colitis, have increased rapidly in South Korea and other industrialized countries over recent decades.1,2 In parallel, the therapeutic paradigm has shifted markedly toward the early initiation of advanced therapies (ATs), including biologics such as anti-tumor necrosis factor (anti-TNF) agents, anti-integrins, and interleukin-12/23 inhibitors, as well as small molecules like Janus kinase inhibitors, for patients with moderate to severe IBD.3 These ATs offer several advantages over conventional therapies, including targeted immune modulation, faster onset of action, higher rates of clinical and endoscopic remission, and steroid-sparing effects.4,5 Real-world cohort data have demonstrated greater treatment persistence, improved mucosal healing, and reductions in hospitalization and corticosteroid use following AT initiation.6,7 Despite these therapeutic benefits, the expanding use of ATs requires continued vigilance regarding their safety profiles. Reported adverse events (AEs) include opportunistic infections such as tuberculosis (TB) and herpes zoster (HZ), as well as rare but concerning risks of malignancy.8-10 In a systematic review and meta-analysis, the pooled incidence rate of opportunistic infections was approximately 2-fold higher in patients receiving ATs than in those receiving a placebo (0.42 per 100 person-years vs 0.21 per 100 person-years).8 However, limited research has explored the associations between AT use in IBD and potential risks of osteoporosis or psychiatric diseases such as depression and anxiety. As the use of ATs continues to rise, there is a corresponding increase in the occurrence of AEs, underscoring the need for careful risk–benefit assessment, patient selection, and long-term pharmacovigilance. While many clinical trials have reported AEs associated with ATs in IBD, selective inclusion criteria and short follow-up periods may lead to underestimation of AEs, reinforcing the importance of multicenter, real-world studies. Landmark clinical trials of ATs for IBD typically span only a few years; therefore, their duration can be considered relatively short for observing the occurrence of AEs. Moreover, meta-analyses are limited by heterogeneity in AE definitions across studies.8,9 Therefore, this study aimed to quantify the comparative risk of AEs between patients treated with ATs (target cohort) and those not treated with ATs (non-AT group; comparator cohort) in a multicenter, real-world setting using a common data model (CDM), which is a standardized framework that enables harmonized data sharing, analysis, and research across institutions. MATERIALS AND METHODS 1. Data source and ethics The concept and methodology of the CDM have been described in detail in our previous publications.11,12 Electronic health record data from 13 hospitals that had adopted the CDM framework were analyzed (Supplementary Table 1), following propensity score matching (PSM). Detailed information on patient numbers, study periods, target (AT) and comparator (non-AT) groups, and the corresponding target and comparator days after PSM is provided in Supplementary Table 1. The study protocol was reviewed and approved by the Institutional Review Board of Kyung Hee University Hospital (KHNMC IRB number: 2025-10-009). The requirement for written informed consent was waived since the CDM database contained no identifiable data and posed no security risks. In Korea, the “Rare Intractable Disease (RID)” registration program is based on strict and standardized diagnostic criteria for IBD.13 Since its implementation in 2009, patients enrolled in the RID program have been eligible for a co-payment reduction to 10% of medical costs for IBD-related treatment.13 To qualify for inclusion in the RID system, the diagnosis of IBD must meet predefined diagnostic criteria established by the program and be verified by attending physicians and medical institutions. Upon confirmation, patients are registered under unique claim codes, for example, V130 for Crohn’s disease and V131 for ulcerative colitis. All patients with IBD, regardless of disease activity or severity, may be registered in the RID program; therefore, IBD claim data in Korea are verified and reliable. 2. Study design and cohort definition This retrospective, observational, comparative cohort study analyzed the risk of AEs between the AT (target) and non-AT (comparative) cohorts among patients with IBD aged ≥18 years across 13 hospitals. An incident case of IBD was defined as a newly registered case during the study period (January 2012 to December 2022) with no prior IBD claims during the lookback period. Patients with an IBD diagnosis code were excluded if they had received IBD-related medications for fewer than 90 days. Additional exclusion criteria included any AT use in the non-AT group and age <18 years. The AT group was defined as individuals who received any ATs for more than 90 consecutive days, while the non-AT group consisted of individuals treated with conventional medications for the same duration. ATs included anti-TNF biologics (infliximab, adalimumab, golimumab), non-anti-TNF biologics (vedolizumab, ustekinumab), and small-molecule agents (tofacitinib only) (Supplementary Table 2). Other small-molecule agents, such as filgotinib and upadacitinib, were excluded because they were unavailable during the study period. Conventional medications comprised 5-aminosalicylic acid (sulfasalazine and mesalazine), corticosteroids (topical and systemic forms), and immunomodulators (azathioprine, mercaptopurine, and methotrexate) (Supplementary Table 2). Corticosteroids in the form of eye drops, ointments, creams, gels, or lotions were excluded. Patients were considered eligible if they were continuously exposed to the assigned treatment from the index date until loss to follow-up, treatment termination, or death. The time-at-risk period was defined as the interval from the day after the index date to the end of observation or up to 455 days (90 days of prescription and 365 days of follow-up). Patient observation endpoints included discontinuation of follow-up prescriptions, death, or the end of data availability. Medical history and medication use were defined using concept_name (concept_id) within the CDM framework. 3. Outcome cohort definition The outcome cohort (AE) was identified using specific SNOMED-CT codes for patients with AEs among newly diagnosed IBD cases during the study period. In this study, AEs included TB, osteoporosis, malignancy, HZ, and anxiety/depression, recorded in either the principal or secondary diagnosis field. An incident AE case was defined as one newly registered during the study period without any AE-related claims prior to the diagnosis of IBD. To establish the temporal relationship between IBD medications and AEs, patients with AEs diagnosed before the initiation of IBD medications were excluded from this analysis. The follow-up period was defined from the date of IBD diagnosis to the occurrence of an AE or the last follow-up date. Additionally, patients without newly developed AEs were censored at the last day of follow-up or at the date of death. To assess the risk of AEs, 1:1 PSM was performed using a large-scale “Least Absolute Shrinkage and Selection Operator” regression model. This model was trained on thousands of potential confounders automatically extracted from the “Observational Medical Outcomes Partnership-CDM,” encompassing demographic characteristics, comprehensive clinical history, and healthcare utilization data. Using these propensity scores, cohorts were balanced through greedy nearest neighbor matching. Post-matching balance was evaluated using the standardized mean difference (SMD), with values <20% considered acceptable.14 An acceptable SMD indicated that the matched groups were well balanced in their covariates, suggesting that any observed differences in outcomes were more likely attributable to medication effects rather than baseline characteristics. 4. Covariates To enhance the PSM analysis between the target and comparative cohorts, a comprehensive set of 237 covariates was included. These covariates included age, sex, index year, Charlson Comorbidity Index, various comorbidities, and medications prescribed within 365 days prior to the index date. Covariates present in >5% of the total patients before PSM for each AE are presented. The target and comparative cohorts were matched for general medical history, neoplastic history, cardiovascular history, and medication history. The overall comorbidity burden was assessed using the Charlson comorbidity index, CHADS2 score, and CHA2DS2-VASc score. 5. Statistical analysis Data were presented as numbers (percentages) for categorical variables and as means (standard deviations) for normally distributed continuous variables. The ATLAS platform (version 2.12.0) was used for initial CDM analyses. Cox proportional hazards modeling and Kaplan-Meier estimations were performed using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). Cox proportional hazards models were applied to compare the incidence of AEs between matched cohorts, and results were expressed as hazard ratios (HRs) with corresponding 95% confidence intervals (CIs). Kaplan-Meier curves illustrated the proportion of event-free patients over time, and the log-rank test was used to assess the statistical significance of differences between groups. Two-sided p-values <0.05 were considered statistically significant. RESULTS A total of 4,854 patients in the AT group (target cohort) and 2,188 in the non-AT group (comparator cohort) met the eligibility criteria after PSM. Patients in both cohorts who received any IBD medications for more than 90 consecutive days, were at risk for ≤1 day, or did not match the propensity score were excluded. All SMDs were <0.2 after PSM for age group, sex, medical history, cardiovascular disease history, and medication use, indicating good covariate balance. The matched cohorts were therefore considered well balanced, as all SMDs were <0.2 across all analyses (Tables 1-5). In total, 1,602 propensity score-matched pairs for TB (Table 1), 1,620 for osteoporosis (Table 2), 1,598 for anxiety/depression (Table 3), 1,626 for HZ (Table 4), and 1,596 for malignancy (Table 5) were included in the final analysis. Table 1Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for TuberculosisVariableBefore propensity score matchingAfter propensity score matchingAT group (n=4,854)Non-AT group (n=2,188)SMDAT group (n=1,602)Non-AT group(n=1,602)SMDAge group, No. (%), yr<402,529 (52.1)1,501 (68.6) 0.342962 (60.0)1,036 (64.7) 0.09540–49835 (17.2)321 (14.7) 0.069240 (15.0)245 (15.3) 0.00950–59746 (15.4)204 (9.3) 0.185208 (13.0)178 (11.1) 0.05860–69436 (9.0)110 (5.0) 0.155116 (7.2)97 (6.1) 0.04870–79251 (5.2)52 (2.4) 0.14762 (3.9)46 (2.9) 0.055≥8057 (1.2) 0 0.15414 (0.9) 0 0.133Female sex, No. (%)1,715 (35.3)671 (30.7) 0.099535 (33.4)510 (31.8) 0.033Medical history, No. (%)GeneralHypertensive disorder153 (3.2)23 (1.1) 0.14736 (2.2)20 (1.2) 0.076Hyperlipidemia99 (2.0)19 (0.9) 0.09825 (1.6)14 (0.9) 0.063Diabetes mellitus92 (1.9)18 (0.8) 0.09323 (1.4)14 (0.9) 0.053Gastroesophageal reflux disease252 (5.2)101 (4.6) 0.02779 (4.9)65 (4.1) 0.042Osteoarthritis10 (0.2)10 (0.5) 0.0442 (0.1)6 (0.4) 0.050Visual system disorder14 (0.3)3 (0.1) 0.0332 (0.1)3 (0.2) 0.016Neoplasms220 (4.5)35 (1.6) 0.17148 (3.0)32 (2.0) 0.064CardiovascularHeart disease118 (2.4)19 (0.9) 0.12322 (1.4)17 (1.1) 0.028Ischemic heart disease52 (1.1)4 (0.2) 0.1135 (0.3)4 (0.2) 0.012Cerebrovascular disease22 (0.5)2 (0.1) 0.0698 (0.5)2 (0.1) 0.067Medication use, No. (%)Anti-inflammatory/anti-rheumatic products592 (12.2)254 (11.6) 0.018185 (11.5)164 (10.2) 0.042Antithrombotic agents292 (6.0)88 (4.0) 0.09169 (4.3)59 (3.7) 0.032Lipid modifying agents204 (4.2)35 (1.6) 0.15654 (3.4)31 (1.9) 0.089Agents acting on the RAS160 (3.3)35 (1.6) 0.11037 (2.3)32 (2.0) 0.022Opioids1,455 (30.0)618 (28.2) 0.038421 (26.3)445 (27.8) 0.034Calcium channel blockers198 (4.1)37 (1.7) 0.14343 (2.7)32 (2.0) 0.045Beta blocking agents151 (3.1)35 (1.6) 0.10034 (2.1)27 (1.7) 0.032Diuretics180 (3.7)48 (2.2) 0.09054 (3.4)24 (1.5) 0.122Antibacterials for systemic use1,286 (26.5)661 (30.2) 0.083488 (30.5)429 (26.8) 0.082Psycholeptics1,573 (32.4)633 (28.9) 0.075453 (28.3)457 (28.5) 0.006Drugs for obstructive airway diseases507 (10.4)195 (8.9) 0.052150 (9.4)132 (8.2) 0.040Drugs used in diabetes117 (2.4)18 (0.8) 0.12631 (1.9)14 (0.9) 0.090Antiepileptics116 (2.4)29 (1.3) 0.07931 (1.9)18 (1.1) 0.066Antidepressants691 (14.2)456 (20.8) 0.174267 (16.7)253 (15.8) 0.024Antineoplastic agents136 (2.8)72 (3.3) 0.02845 (2.8)41 (2.6) 0.015Psychostimulants61 (1.3)21 (1.0) 0.02820 (1.2)20 (1.2) <0.001AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system.We report covariates over 5% of the total patient cohort prior to propensity score matching. Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Tuberculosis AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system. We report covariates over 5% of the total patient cohort prior to propensity score matching. Table 5Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for MalignancyVariableBefore propensity score matchingAfter propensity score matchingAT group (n=4,854)Non-AT group (n=2,188)SMDAT group (n=1,596)Non-AT group (n=1,596)SMDAge group, No. (%), yr<402,529 (52.1)1,501 (68.6) 0.342981 (61.5)1,049 (65.7) 0.08940–49835 (17.2)321 (14.7) 0.069236 (14.8)242 (15.2) 0.01150–59746 (15.4)204 (9.3) 0.185206 (12.9)171 (10.7) 0.06860–69436 (9.0)110 (5.0) 0.155106 (6.6)90 (5.6) 0.04270–79251 (5.2)52 (2.4) 0.14753 (3.3)44 (2.8) 0.033≥8057 (1.2) 0 0.15414 (0.9) 0 0.133Female sex, No. (%)1,715 (35.3)671 (30.7) 0.099532 (33.3)501 (31.4) 0.042Medical history, No. (%)GeneralHypertensive disorder153 (3.2)23 (1.1) 0.14732 (2.0)20 (1.3) 0.059Hyperlipidemia99 (2.0)19 (0.9) 0.09826 (1.6)13 (0.8) 0.074Diabetes mellitus92 (1.9)18 (0.8) 0.09323 (1.4)14 (0.9) 0.053Gastroesophageal reflux disease252 (5.2)101 (4.6) 0.02783 (5.2)66 (4.1) 0.051Osteoarthritis10 (0.2)10 (0.5) 0.0443 (0.2)6 (0.4) 0.035Visual system disorder14 (0.3)3 (0.1) 0.0333 (0.2)3 (0.2) <0.001Neoplasms220 (4.5)35 (1.6) 0.171 NA NA NACardiovascularHeart disease118 (2.4)19 (0.9) 0.12326 (1.6)16 (1.0) 0.055Ischemic heart disease52 (1.1)4 (0.2) 0.1138 (0.5)3 (0.2) 0.053Cerebrovascular disease22 (0.5)2 (0.1) 0.0696 (0.4)2 (0.1) 0.050Medication use, No. (%)Anti-inflammatory/anti-rheumatic products592 (12.2)254 (11.6) 0.018179 (11.2)159 (10.0)0.041Antithrombotic agents292 (6.0)88 (4.0) 0.09167 (4.2)57 (3.6)0.032Lipid modifying agents204 (4.2)35 (1.6) 0.15652 (3.3)30 (1.9)0.087Agents acting on the RAS160 (3.3)35 (1.6) 0.11034 (2.1)31 (1.9)0.013Opioids1,455 (30.0)618 (28.2) 0.038422 (26.4)442 (27.7)0.028Calcium channel blockers198 (4.1)37 (1.7) 0.14336 (2.3)30 (1.9)0.026Beta blocking agents151 (3.1)35 (1.6) 0.10032 (2.0)27 (1.7)0.023Diuretics180 (3.7)48 (2.2) 0.09048 (3.0)21 (1.3)0.117Antibacterials for systemic use1,286 (26.5)661 (30.2) 0.083489 (30.6)429 (26.9)0.083Psycholeptics1,573 (32.4)633 (28.9) 0.075464 (29.1)458 (28.7)0.008Drugs for obstructive airway diseases507 (10.4)195 (8.9) 0.052140 (8.8)127 (8.0)0.029Drugs used in diabetes117 (2.4)18 (0.8) 0.12629 (1.8)13 (0.8)0.088Antiepileptics116 (2.4)29 (1.3) 0.07927 (1.7)18 (1.1)0.048Antidepressants691 (14.2)456 (20.8) 0.174275 (17.2)258 (16.2)0.029Antineoplastic agents136 (2.8)72 (3.3) 0.02838 (2.4)38 (2.4) <0.001Psychostimulants61 (1.3)21 (1.0) 0.02818 (1.1)20 (1.3)0.012AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system; NA, not available.We report covariates over 5% of the total patient cohort prior to propensity score matching. Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Malignancy AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system; NA, not available. We report covariates over 5% of the total patient cohort prior to propensity score matching. Table 2Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between AT Group (Target Cohort) versus Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for OsteoporosisVariableBefore propensity score matchingAfter propensity score matchingAT group (n=4,854)Non-AT group (n=2,188)SMDAT group (n=1,620)Non-AT group (n=1,620)SMDAge group, No. (%), yr<402,529 (52.1)1,501 (68.6) 0.342980 (60.5)1055 (65.1)0.09640–49835 (17.2)321 (14.7) 0.069248 (15.3)251 (15.5)0.00550–59746 (15.4)204 (9.3) 0.185215 (13.3)174 (10.7)0.07860–69436 (9.0)110 (5.0) 0.155112 (6.9)97 (6.0)0.03870–79251 (5.2)52 (2.4) 0.14754 (3.3)43 (2.7)0.040≥8057 (1.2) 0 0.15411 (0.7) 00.117Female sex, No. (%)1,715 (35.3)671 (30.7) 0.099532 (32.8)503 (31.0)0.038Medical history, No. (%)GeneralHypertensive disorder153 (3.2)23 (1.1) 0.14727 (1.7)17 (1.0)0.053Hyperlipidemia99 (2.0)19 (0.9) 0.09822 (1.4)11 (0.7)0.068Diabetes mellitus92 (1.9)18 (0.8) 0.09321 (1.3)14 (0.9)0.042Gastroesophageal reflux disease252 (5.2)101 (4.6) 0.02778 (4.8)66 (4.1)0.036Osteoarthritis10 (0.2)10 (0.5) 0.0441 (0.1)5 (0.3)0.057Visual system disorder14 (0.3)3 (0.1) 0.0333 (0.2)3 (0.2)<0.001Neoplasms220 (4.5)35 (1.6) 0.17141 (2.5)33 (2.0)0.033CardiovascularHeart disease118 (2.4)19 (0.9) 0.12319 (1.2)18 (1.1)0.006Ischemic heart disease52 (1.1)4 (0.2) 0.1134 (0.2)4 (0.2)<0.001Cerebrovascular disease22 (0.5)2 (0.1) 0.0697 (0.4)2 (0.1)0.059Medication use, No. (%)Anti-inflammatory/anti-rheumatic products592 (12.2)254 (11.6) 0.018180 (11.1)158 (9.8)0.044Antithrombotic agents292 (6.0)88 (4.0) 0.09163 (3.9)60 (3.7)0.010Lipid modifying agents204 (4.2)35 (1.6) 0.15649 (3.0)28 (1.7)0.085Agents acting on the RAS160 (3.3)35 (1.6) 0.11031 (1.9)28 (1.7)0.014Opioids1,455 (30.0)618 (28.2) 0.038422 (26.0)448 (27.7)0.036Calcium channel blockers198 (4.1)37 (1.7) 0.14338 (2.3)31 (1.9)0.030Beta blocking agents151 (3.1)35 (1.6) 0.10035 (2.2)29 (1.8)0.027Diuretics180 (3.7)48 (2.2) 0.09047 (2.9)23 (1.4)0.102Antibacterials for systemic use1,286 (26.5)661 (30.2) 0.083489 (30.2)433 (26.7)0.077Psycholeptics1,573 (32.4)633 (28.9) 0.075458 (28.3)459 (28.3)0.001Drugs for obstructive airway diseases507 (10.4)195 (8.9) 0.052144 (8.9)132 (8.1)0.027Drugs used in diabetes117 (2.4)18 (0.8) 0.12628 (1.7)12 (0.7)0.090Antiepileptics116 (2.4)29 (1.3) 0.07927 (1.7)19 (1.2)0.042Antidepressants691 (14.2)456 (20.8) 0.174269 (16.6)253 (15.6)0.027Antineoplastic agents136 (2.8)72 (3.3) 0.02841 (2.5)40 (2.5)0.004Psychostimulants61 (1.3)21 (1.0) 0.02819 (1.2)20 (1.2)0.006AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system.We report covariates over 5% of the total patient cohort prior to propensity score matching. Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between AT Group (Target Cohort) versus Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Osteoporosis AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system. We report covariates over 5% of the total patient cohort prior to propensity score matching. Table 3Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Anxiety/DepressionVariableBefore propensity score matchingAfter propensity score matchingAT group (n=4,854)Non-AT group (n=2,188)SMDAT group (n=1,598)Non-AT group(n=1,598)SMDAge group, No. (%), yr<402,529 (52.1)1,501 (68.6) 0.342967 (60.5)1,040 (65.1)0.09540–49835 (17.2)321 (14.7) 0.069239 (15.0)243 (15.2)0.00750–59746 (15.4)204 (9.3) 0.185209 (13.1)174 (10.9)0.06760–69436 (9.0)110 (5.0) 0.155111 (6.9)96 (6.0)0.03870–79251 (5.2)52 (2.4) 0.14758 (3.6)45 (2.8)0.046≥8057 (1.2)0 0.15414 (0.9)00.133Female sex, No. (%)1,715 (35.3)671 (30.7) 0.099536 (33.5)502 (31.4)0.045Medical history, No. (%)GeneralHypertensive disorder153 (3.2)23 (1.1) 0.14732 (2.0)16 (1.0)0.082Hyperlipidemia99 (2.0)19 (0.9) 0.09823 (1.4)13 (0.8)0.059Diabetes mellitus92 (1.9)18 (0.8) 0.09323 (1.4)15 (0.9)0.046Gastroesophageal reflux disease252 (5.2)101 (4.6) 0.02779 (4.9)67 (4.2)0.036Osteoarthritis10 (0.2)10 (0.5) 0.044 3 (0.2) 5 (0.3)0.025Visual system disorder14 (0.3)3 (0.1) 0.033 3 (0.2) 2 (0.1)0.016Neoplasms220 (4.5)35 (1.6) 0.17144 (2.8)33 (2.1)0.045CardiovascularHeart disease118 (2.4)19 (0.9) 0.12320 (1.3)17 (1.1)0.018Ischemic heart disease52 (1.1)4 (0.2) 0.113 4 (0.3) 4 (0.3)<0.001Cerebrovascular disease22 (0.5)2 (0.1) 0.069 7 (0.4) 2 (0.1)0.059Medication use, No. (%)Anti-inflammatory/anti-rheumatic products592 (12.2)254 (11.6) 0.018188 (11.8)163 (10.2)0.050Antithrombotic agents292 (6.0)88 (4.0) 0.09167 (4.2)60 (3.8)0.022Lipid modifying agents204 (4.2)35 (1.6) 0.15652 (3.3)29 (1.8)0.092Agents acting on the RAS160 (3.3)35 (1.6) 0.11033 (2.1)28 (1.8)0.023Opioids1,455 (30.0)618 (28.2) 0.038420 (26.3)448 (28.0)0.039Calcium channel blockers198 (4.1)37 (1.7) 0.14341 (2.6)31 (1.9)0.042Beta blocking agents151 (3.1)35 (1.6) 0.10032 (2.0)26 (1.6)0.028Diuretics180 (3.7)48 (2.2) 0.09050 (3.1)20 (1.3)0.129Antibacterials for systemic use1,286 (26.5)661 (30.2) 0.083493 (30.9)435 (27.2)0.080Psycholeptics1,573 (32.4)633 (28.9) 0.075462 (28.9)452 (28.3)0.014Drugs for obstructive airway diseases507 (10.4)195 (8.9) 0.052150 (9.4)131 (8.2)0.042Drugs used in diabetes117 (2.4)18 (0.8) 0.12632 (2.0)13 (0.8)0.101Antiepileptics116 (2.4)29 (1.3) 0.07929 (1.8)17 (1.1)0.063Antidepressants691 (14.2)456 (20.8) 0.174265 (16.6)250 (15.6)0.026Antineoplastic agents136 (2.8)72 (3.3) 0.02844 (2.8)40 (2.5)0.016Psychostimulants61 (1.3)21 (1.0) 0.02815 (0.9)20 (1.3)0.030AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system.We report covariates over 5% of the total patient cohort prior to propensity score matching. Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Anxiety/Depression AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system. We report covariates over 5% of the total patient cohort prior to propensity score matching. Table 4Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Herpes ZosterVariableBefore propensity score matchingAfter propensity score matchingAT group (n=4,854)Non-AT group (n=2,188)SMDAT group (n=1,626)Non-AT group(n=1,626)SMDAge group, No. (%), yr<402,529 (52.1)1,501 (68.6) 0.342977 (60.1)1,050 (64.6)0.09340–49835 (17.2)321 (14.7) 0.069247 (15.2)253 (15.6)0.01050–59746 (15.4)204 (9.3) 0.185213 (13.1)178 (10.9)0.06660–69436 (9.0)110 (5.0) 0.155114 (7.0)98 (6.0)0.04070–79251 (5.2)52 (2.4) 0.14761 (3.8)47 (2.9)0.048≥8057 (1.2) 0 0.15414 (0.9) 00.132Female sex, No. (%)1,715 (35.3)671 (30.7) 0.099531 (32.7)505 (31.1)0.034Medical history, No. (%)GeneralHypertensive disorder153 (3.2)23 (1.1) 0.14733 (2.0)19 (1.2)0.069Hyperlipidemia99 (2.0)19 (0.9) 0.09824 (1.5)14 (0.9)0.057Diabetes mellitus92 (1.9)18 (0.8) 0.09321 (1.3)15 (0.9)0.035Gastroesophageal reflux disease252 (5.2)101 (4.6) 0.02782 (5.0)68 (4.2)0.041Osteoarthritis10 (0.2)10 (0.5) 0.0442 (0.1)6 (0.4)0.050Visual system disorder14 (0.3)3 (0.1) 0.0333 (0.2)3 (0.2)<0.001Neoplasms220 (4.5)35 (1.6) 0.17143 (2.6)32 (2.0)0.045CardiovascularHeart disease118 (2.4)19 (0.9) 0.12323 (1.4)17 (1.0)0.033Ischemic heart disease52 (1.1)4 (0.2) 0.1137 (0.4)4 (0.2)0.032Cerebrovascular disease22 (0.5)2 (0.1) 0.0697 (0.4)2 (0.1)0.059Medication use, No. (%)Anti-inflammatory/anti-rheumatic products592 (12.2)254 (11.6) 0.018181 (11.1)161 (9.9)0.040Antithrombotic agents292 (6.0)88 (4.0) 0.09167 (4.1)61 (3.8)0.019Lipid modifying agents204 (4.2)35 (1.6) 0.15650 (3.1)30 (1.8)0.079Agents acting on the RAS160 (3.3)35 (1.6) 0.11034 (2.1)31 (1.9)0.013Opioids1,455 (30.0)618 (28.2) 0.038429 (26.4)447 (27.5)0.025Calcium channel blockers198 (4.1)37 (1.7) 0.14339 (2.4)34 (2.1)0.021Beta blocking agents151 (3.1)35 (1.6) 0.10035 (2.2)29 (1.8)0.027Diuretics180 (3.7)48 (2.2) 0.09049 (3.0)24 (1.5)0.104Antibacterials for systemic use1,286 (26.5)661 (30.2) 0.083487 (30.0)431 (26.5)0.077Psycholeptics1,573 (32.4)633 (28.9) 0.075468 (28.8)460 (28.3)0.011Drugs for obstructive airway diseases507 (10.4)195 (8.9) 0.052141 (8.7)131 (8.1)0.022Drugs used in diabetes117 (2.4)18 (0.8) 0.12630 (1.8)13 (0.8)0.092Antiepileptics116 (2.4)29 (1.3) 0.07928 (1.7)17 (1.0)0.058Antidepressants691 (14.2)456 (20.8) 0.174273 (16.8)256 (15.7)0.028Antineoplastic agents136 (2.8)72 (3.3) 0.02844 (2.7)40 (2.5)0.016Psychostimulants61 (1.3)21 (1.0) 0.02817 (1.0)20 (1.2)0.017AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system.We report covariates over 5% of the total patient cohort prior to propensity score matching. Distribution of Baseline Characteristics in the Overall Population from the 13 Hospitals between the AT Group (Target Cohort) and the Non-AT Group (Comparator Cohort) before and after Propensity Score Matching for Herpes Zoster AT, advanced therapy; SMD, standardized mean difference; RAS, renin-angiotensin system. We report covariates over 5% of the total patient cohort prior to propensity score matching. 1. Risk of TB between the AT and non-AT groups Cox proportional hazards analyses were conducted to compare the risk of TB between the AT and non-AT groups (Fig. 1). The risk of TB was significantly higher in the AT group than in the non-AT group (HR, 2.86; 95% CI, 1.77 to 4.63; p<0.001). In the subgroup analysis, the risk of TB was significantly higher in the anti-TNF subgroup compared with the non-AT group (HR, 3.74; 95% CI, 2.12 to 6.59; p<0.001). Similarly, the risk of TB was significantly higher in the non-anti-TNF subgroup than in the non-AT group (HR, 3.58; 95% CI, 1.56 to 8.23; p=0.001). In contrast, the small-molecule agent subgroup showed no significant difference in TB risk compared with the non-AT group (HR, 0.89; 95% CI, 0.29 to 2.77; p=0.846). Kaplan-Meier plots comparing the risk of tuberculosis between the advanced therapy and conventional therapy groups (A). Subgroup analyses comparing the risk of tuberculosis in the anti-TNF (B), non-anti-TNF (C), and small-molecule agent groups (D) with that of the conventional therapy group. TNF, tumor necrosis factor. 2. Risk of osteoporosis between the AT and non-AT groups Cox proportional hazards analyses were conducted to compare the risk of osteoporosis between the AT and non-AT groups (Fig. 2). The overall risk of osteoporosis did not differ significantly between the AT and non-AT groups (HR, 0.86; 95% CI, 0.42 to 1.73; p=0.667). In the subgroup analysis, the risk of osteoporosis was significantly lower in the anti-TNF group compared with the non-AT group (HR, 0.38; 95% CI, 0.15 to 0.99; p=0.040). In contrast, no significant difference in osteoporosis risk was observed between the non-anti-TNF subgroup and the non-AT group (HR, 3.13; 95% CI, 0.86 to 11.37; p=0.068), or between the small-molecule agent subgroup and the non-AT group (HR, 1.20; 95% CI, 0.27 to 5.35; p=0.814). Kaplan-Meier plots comparing the risk of osteoporosis between the advanced therapy and conventional therapy groups (A). Subgroup analyses comparing the risk of osteoporosis in the anti-TNF (B), non-anti-TNF (C), and small-molecule agent groups (D) with that of the conventional therapy group. TNF, tumor necrosis factor. 3. Risk of depression/anxiety between the AT and non-AT groups Cox proportional hazards analyses were conducted to compare the risk of depression/anxiety between the AT and non-AT groups (Fig. 3). The risk of depression/anxiety was significantly higher in the AT group than in the non-AT group (HR, 1.58; 95% CI, 1.15 to 2.16; p=0.005). In the subgroup analysis, the risk of depression/anxiety was significantly higher in the anti-TNF subgroup compared with the non-AT group (HR, 1.49; 95% CI, 1.07 to 2.09; p=0.019). In contrast, no significant difference was observed between the non-anti-TNF subgroup and the non-AT group (HR, 1.31; 95% CI, 0.74 to 2.32; p=0.357), or between the small-molecule agent subgroup and the non-AT group (HR, 0.66; 95% CI, 0.32 to 1.35; p=0.248). Kaplan-Meier plots comparing the risk of depression or anxiety between the advanced therapy and conventional therapy groups (A). Subgroup analyses comparing the risk of depression or anxiety in the anti-TNF (B), non-anti-TNF (C), and small-molecule agent groups (D) with that of the conventional therapy group. TNF, tumor necrosis factor. 4. Risk of HZ between the AT and non-AT groups Cox proportional hazards analyses were conducted to compare the risk of HZ between the AT and non-AT groups (Fig. 4). The overall risk of HZ did not differ significantly between the AT and non-AT groups (HR, 1.23; 95% CI, 0.74 to 2.04; p=0.417). In the subgroup analysis, no significant difference in HZ risk was observed in the anti-TNF subgroup (HR, 1.50; 95% CI, 0.85 to 2.65; p=0.158), the non-anti-TNF subgroup (HR, 1.17; 95% CI, 0.51 to 2.68; p=0.702), or the small-molecule agent subgroup (HR, 0.73; 95% CI, 0.25 to 2.19; p=0.579), compared with the non-AT group. Kaplan-Meier plots comparing the risk of herpes zoster between the advanced therapy and conventional therapy groups (A). Subgroup analyses comparing the risk of herpes zoster in the anti-TNF (B), non-anti-TNF (C), and small-molecule agent groups (D) with that of the conventional therapy group. TNF, tumor necrosis factor. 5. Risk of malignancy between the AT and non-AT groups Cox proportional hazards analyses were conducted to compare the risk of malignancy between the AT and non-AT groups (Fig. 5). The overall risk of malignancy did not differ significantly between the AT and non-AT groups (HR, 0.95; 95% CI, 0.64 to 1.42; p=0.798). In the subgroup analysis, no significant difference in malignancy risk was observed between the anti-TNF subgroup and the non-AT group (HR, 1.05; 95% CI, 0.67 to 1.64; p=0.839), or between the non-anti-TNF subgroup and the non-AT group (HR, 1.02; 95% CI, 0.56 to 1.87; p=0.946). However, the risk of malignancy was significantly lower in the small-molecule agent subgroup compared with the non-AT group (HR, 0.26; 95% CI, 0.09 to 0.81; p=0.013). Kaplan-Meier plots comparing the risk of malignancy between the advanced therapy and conventional therapy groups (A). Subgroup analyses comparing the risk of malignancy in the anti-TNF (B), non-anti-TNF (C), and small-molecule agent groups (D) with that of the conventional therapy group. TNF, tumor necrosis factor. DISCUSSION To the best of our knowledge, this is the first distributed network analysis using a CDM database to compare the risk of AEs between the AT and non-AT groups. Our findings demonstrated that patients with IBD receiving ATs had a significantly increased risk of TB and depression/anxiety compared with those in the non-AT group. Among the subgroups, anti-TNF agents were associated with higher risks of TB and depression/anxiety but a lower risk of osteoporosis. No significant differences were observed in the risk of HZ or malignancy between the overall AT and non-AT groups. Small-molecule agents were not significantly associated with increased risk for TB, osteoporosis, depression/anxiety, or HZ but were associated with a lower risk of malignancy. The strengths of this study lie in its large sample size, extensive PSM, and use of real-world data from multiple hospital databases based on the CDM framework, which collectively enhance the reliability and generalizability of the findings to routine clinical practice. The use of Cox proportional hazards models allowed robust quantification of the risk differences, reflecting a rigorous evaluation of outcomes. Additionally, the comprehensive inclusion of diverse patient demographics and the meticulous control of confounding variables underscore the study’s methodological rigor. In the present study, the AT group exhibited a significantly higher risk of TB compared with the non-AT group, consistent with previous evidence indicating that biologic therapy increases the risk of TB reactivation. The observed HR for TB in the overall AT group was 2.86 (p<0.001), with subgroup analyses revealing even higher risks in the anti-TNF (HR, 3.74; p<0.001) and non-anti-TNF (HR, 3.59; p=0.001) subgroups. These findings align with earlier cohort studies demonstrating a 2- to 4-fold increased risk of TB in patients treated with anti-TNF agents, particularly infliximab and adalimumab, compared with conventional therapies.15-17 Conversely, the small-molecule agent group showed no significantly increased TB risk (p=0.846), supporting previous reports suggesting a comparatively lower TB risk profile with tofacitinib.18-20 These results underscore the importance of latent TB screening and prophylaxis before initiating anti-TNF therapies. Although the overall risk of osteoporosis did not differ significantly between the AT and non-AT groups (p=0.667), patients in the anti-TNF group exhibited a significantly lower risk of osteoporosis compared with those in the non-AT group (HR, 0.38; p=0.040), suggesting a potential protective effect on bone metabolism. These findings are consistent with previous studies demonstrating that anti-TNF therapy improves bone mineral density in patients with IBD and rheumatoid arthritis.21-23 The natural balance of bone remodeling is often disrupted by chronic inflammation, as TNF-α is a pro-inflammatory cytokine that significantly contributes to bone loss in inflammatory conditions. Anti-TNF therapy inhibits TNF-α, thereby reducing osteoclast-mediated bone resorption and potentially enhancing osteoblast activity, which together may help prevent bone loss and improve bone mineral density.24 In contrast, no significant differences in osteoporosis risk were observed in the non-anti-TNF group (HR, 3.13; 95% CI, 0.86 to 11.37; p=0.068) or the small-molecule agent group (HR, 1.20; 95% CI, 0.27 to 5.35; p=0.814) compared with the non-AT group, although the wide CIs highlight the need for further investigation. Overall, these findings support the hypothesis that anti-TNF therapy may exert a protective effect against osteoporosis in patients with IBD. In the present study, the risk of depression or anxiety was significantly higher among patients receiving anti-TNF therapy compared with those in the non-AT group (HR, 1.49; 95% CI, 1.07 to 2.09; p=0.019). No significant differences were observed in the non-anti-TNF group (p=0.357) or the small-molecule agent group (p=0.248) compared with the non-AT group. These findings are consistent with previous studies reporting an elevated risk of depression and anxiety in patients with IBD and rheumatoid arthritis treated with biologic agents, particularly anti-TNF therapies.25-28 Given that anxiety and depression are more prevalent in patients with active IBD than in those with inactive disease,28 the observed association between anti-TNF therapy and increased risk of anxiety or depression may partly reflect underlying disease activity rather than a direct pharmacologic effect of the treatment itself. Nonetheless, compensatory immune responses or neuroimmune disruptions following TNF-α inhibition may also contribute to the emergence of mood disorders.29,30 Additionally, the psychological burden of chronic IBD, combined with treatment expectations and potential AEs of anti-TNF agents, may exacerbate psychiatric vulnerability. These findings underscore the importance of proactive mental health assessment and monitoring in patients receiving anti-TNF therapy. In our study, the risk of HZ did not differ significantly between the AT and non-AT groups (p=0.417), including all subgroup analyses. These findings contrast with previous studies that reported an increased risk of HZ in patients with IBD treated with ATs.31,32 Prior network meta-analyses showed that the elevated risk of HZ differed among AT groups, being highest with small-molecule agents.31 Furthermore, a Korean population-based study reported a strong association between corticosteroid use and HZ occurrence.32 In our cohort, the substantial use of corticosteroids in the non-AT group and the limited use of small-molecule agents in the AT group may have influenced the risk of HZ, potentially explaining differences from previous reports.32,33 Although none of the HRs reached statistical significance in our study, most approached or exceeded 1.0, with some above 1.5, and wide 95% CIs, suggesting that a true increased risk may have been undetected due to limited statistical power. In this study, the overall risk of malignancy also did not differ significantly between the AT and non-AT groups (p=0.798). Subgroup analyses similarly revealed no significant increase in malignancy risk in either the anti-TNF or non-anti-TNF subgroups compared with the non-AT group. These findings are broadly consistent with prior studies that have not demonstrated a clear association between anti-TNF therapy and an increased overall risk of malignancy.33,34 In our study, the small-molecule agent group exhibited a significantly lower risk of malignancy relative to the non-AT group. However, this finding requires cautious interpretation, as it may reflect channeling bias arising from reimbursement criteria for small-molecule agents and the very small number of malignancy events in both groups. Current domestic reimbursement guidelines in Korea restrict the use of tofacitinib in patients aged ≥65 years or in those with a potential risk of malignancy. This reverse selection effect likely contributed to the lower observed incidence of malignancy in the small-molecule agent group, reflecting patient selection rather than a true protective effect of the therapy. In addition, there were 13 malignancy events over 1,035.3 person-years in the non-AT group and only 4 events over 1,228.5 person-years in the small-molecule agent group, thus limiting statistical power. Our study has some limitations. First, due to its retrospective observational design, causality cannot be definitively established, and potential confounding cannot be completely excluded despite the study’s extensive scope. For example, key factors such as disease activity, cumulative corticosteroid exposure, and smoking status are not comprehensively captured within the CDM and may contribute to residual confounding, particularly for depression/anxiety and malignancy outcomes. However, no prospective studies on this issue exist, and comprehensive PSM was performed prior to the analysis to minimize potential bias. Additionally, safety profiles observed in real-world data may differ from those reported in clinical trials, as they can be influenced by various regional factors, including disease phenotype, medication accessibility, infection risk, and reimbursement policies. Second, variability in the operational definitions and diagnostic criteria for IBD and AEs may limit the comparability of our findings with previous studies. However, CDM-based studies utilize standardized data structures across multiple medical institutions, thereby improving consistency and reliability. Third, our IBD cohort was defined as individuals who received IBD-related medications for more than 90 consecutive days to ensure diagnostic accuracy. Therefore, the study population may differ from the general population of patients with IBD. Fourth, the relatively small sample sizes in certain subgroup analyses may have reduced the statistical power to detect significant differences. Because this study is likely underpowered to detect modest risk differences for rare outcomes such as malignancy, HZ, and osteoporosis, the absence of statistically significant differences should not be interpreted as evidence of no clinically meaningful effects. For example, overall risk of malignancy should be viewed with caution, as the number of events was very small with wide CIs in subgroups, reflecting the limited precision of these estimates. Finally, our study included only Korean patients, which may limit the generalizability of the findings to other ethnic populations. Future research involving more ethnically and geographically diverse populations would enhance external validity and provide a more comprehensive understanding of treatment-related risks. In conclusion, the use of anti-TNF and other ATs was associated with an increased risk of TB and depression/anxiety but showed no significant association with HZ, malignancy, or osteoporosis in patients with IBD in a real-world setting. Notably, small-molecule agents were associated with a reduced risk of malignancy, possibly due to channeling bias. Appropriate management of potential AEs is essential to ensure optimal clinical outcomes in patients with IBD. ACKNOWLEDGEMENTS We would like to thank Ewha Womans University Medical Center, Ajou University Medical Center, Daegu Catholic University Medical Center, Soonchunhyang University Hospital Bucheon, Soonchunhyang University Hospital Cheonan, Dongsan Hospital, Hallym University Dongtan Sacred Heart Hospital for providing the common data model (CDM) data. SUPPLEMENTARY MATERIALS Supplementary materials can be accessed at https://doi.org/10.5009/gnl250577. Footnotes CONFLICTS OF INTEREST No potential conflict of interest relevant to this article was reported. AUTHOR CONTRIBUTIONS Study concept and design: H.I.C., J.M.C. Data acquisition: H.I.C., M.K., J.R.M., C.W.J., Y.S.K., S.Y.R., G.M.P., Y.J.L., D.H.B., J.M.C. Data analysis and interpretation: H.I.C., M.K., J.R.M., C.W.J., Y.S.K., S.Y.R., G.M.P., Y.J.L., D.H.B., J.M.C. Drafting of the manuscript: H.I.C. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: M.K., J.R.M. Study supervision: J.M.C. Approval of final manuscript: all authors. DATA AVAILABILITY STATEMENT In common data model-based research, access to the original source data is restricted; therefore, the raw data is not available. Researchers have access only to the analyzed results derived from the standardized dataset. However, the R code used for the analysis is available from the corresponding author upon reasonable request. REFERENCES Associated Data Supplementary Materials