Comparative Real‐World Effectiveness of Anti‐TNF, Ustekinumab and Vedolizumab in Crohn's Disease Over 24 Months: A Pooled Analysis From the Prospective UMBRELLA‐IBD Registry in Germany Authors: Arne Bokemeyer, Florian Tran, Sandra Plachta‐Danielzik, Elena Gilman, Thomas Wenske, Matthew Gaskins, Stefan Schreiber, Bernd Bokemeyer Published: 2026 ABSTRACT ABSTRACTBackgroundReal‐world evidence is needed to understand the comparative effectiveness of biologics in Crohn's disease (CD) under routine‐care conditions. We conducted a pooled analysis of individual patient data from prospective German CD registries (BioCrohn, RUN‐CD, VEDO‐IBD) in the UMBRELLA‐IBD data warehouse.MethodsWe included 1567 adults with CD who initiated anti‐TNF therapy (adalimumab or infliximab), ustekinumab or vedolizumab. The main outcome was clinical remission (CR) at 24 months (HBI ≤ 4). Additional outcomes included HBI, EQ‐VAS and treatment persistence over time. Between‐group comparisons used propensity‐score‐based inverse probability of treatment weighting.ResultsAfter weighting, CR following induction was observed in a slightly higher proportion of patients initiating anti‐TNF therapy (74.5%) than ustekinumab (69.4%) or vedolizumab (65.3%) (p = 0.035). Clinical response was 77.9%, 73.5% and 67.8%, respectively (p = 0.015). At 24 months, CR rates were similar in magnitude (anti‐TNF, 53.4%; ustekinumab, 63.5%; vedolizumab, 58.7%) but were higher with ustekinumab than anti‐TNF (p = 0.004). Over 24 months, treatment persistence differed (log‐rank p < 0.001) and was highest for adalimumab (81.4%), lowest for infliximab (66.6%), and intermediate for ustekinumab (78.6%) and vedolizumab (76.9%). HBI decreased and remained low, and EQ‐VAS improved, with no relevant between‐group differences.ConclusionOverall effectiveness in CD was high, with only small differences between biologic therapies. At the end of induction, CR was slightly higher with anti‐TNF therapy than with vedolizumab, whereas at 24 months it was slightly higher with ustekinumab than with anti‐TNF. Marked differences in treatment persistence among the evaluated biologics suggest that persistence as an endpoint is not fully explained by CR alone. Introduction Crohn's disease (CD) and ulcerative colitis (UC) are chronic inflammatory bowel diseases (IBD) marked by alternating phases of spontaneous relapse, treatment‐induced remission and variable levels of chronic activity [1]. Ongoing inflammatory activity increases the risk of intestinal complications such as strictures and fistulas, extraintestinal manifestations and psychosocial impairment [2, 3]. This contributes to substantial unmet medical need, particularly following the failure of conventional therapies or biologics [4, 5]. Given that many patients experience treatment failure after only a few years, evidence from real‐world practice is needed to inform the use of available biologics [6]. Anti‐tumour necrosis factor (anti‐TNF) agents such as infliximab and adalimumab inhibit TNF, a key pro‐inflammatory cytokine, and are mainstay treatments for CD and other immune‐mediated inflammatory diseases [7, 8]. Ustekinumab, a monoclonal antibody targeting the p40 subunit of interleukins (IL)‐12/23, was approved by the European Medicines Agency (EMA) in 2016 for treating CD patients who have not responded to conventional therapies and has demonstrated high efficacy in randomised controlled trials (RCTs) [9]. Vedolizumab, a gut‐selective anti‐α4β7 integrin antibody, was approved by the EMA in 2014 for adults with moderately to severely active CD or UC who have failed conventional therapy [10]. In clinical trials and real‐world evidence (RWE) studies after approval, vedolizumab has demonstrated effectiveness and, in some settings, a more favourable safety profile than TNF inhibitors [11, 12, 13, 14, 15, 16]. As multiple biologic options are now available for CD, treatment decisions in routine clinical practice increasingly involve choosing between TNF inhibitors and newer agents such as vedolizumab or ustekinumab despite the scarcity of comparative evidence. Data from head‐to‐head (H2H) randomised trials that might inform such decisions in CD is sparse. In the SEAVUE study, a recent RCT presenting an H2H comparison of UST and ADA in CD patients reported no significant differences in efficacy between the two drugs during the induction and maintenance phases of therapy [17]. In the STARDUST study, an open‐label randomised trial evaluating a treat‐to‐target strategy with ustekinumab, outcomes at 52 weeks were not clearly superior to standard care [18]. However, while RCTs are essential for demonstrating treatment efficacy under controlled conditions, their generalisability to routine clinical practice is limited by restrictive inclusion and exclusion criteria. In particular, older patients, individuals with relevant comorbidities or complications, those with extraintestinal manifestations, and patients with prior intestinal surgery are often excluded from pivotal trials. As a result, more real‐world studies are needed to evaluate the effectiveness and durability of biologic therapies in broader, more heterogeneous patient populations that better reflect daily clinical practice [19]. A small but growing body of RWE studies has therefore examined the comparative effectiveness of different biologic classes in CD. Earlier observational studies indicate that ustekinumab is an effective treatment option in routine clinical practice [20, 21, 22, 23, 24, 25, 26, 27, 28, 29], including a comparison with adalimumab during induction therapy, in which a trend was observed favouring ustekinumab in TNF‐experienced patients [30]. Many of these studies, however, were retrospective and based on relatively small cohorts. More recently, real‐world evidence on vedolizumab has expanded substantially. In a recent systematic literature review and meta‐analysis, vedolizumab was associated with favourable long‐term effectiveness and treatment persistence in patients with CD, with better outcomes in biologic‐naïve than in biologic‐experienced patients [31]. Nevertheless, head‐to‐head evidence is still limited, and there remains a need for more prospective real‐world studies with larger sample sizes that better reflect routine clinical practice. The UMBRELLA‐IBD data warehouse helps address this gap by enabling pooled analyses of individual patient data (IPD) from prospective German CD registries that use harmonised documentation protocols. Here, we present a pooled UMBRELLA‐IBD analysis comparing the real‐world effectiveness of anti‐TNF therapy (adalimumab or infliximab), ustekinumab and vedolizumab in adults with CD over up to 24 months of follow‐up. Materials and Methods Study Design and Data Sources We conducted a pooled, retrospective–prospective cohort analysis using primary data from three investigator‐initiated, prospective, non‐interventional registries contributing CD data to the UMBRELLA‐IBD data warehouse: BioCrohn, RUN‐CD and VEDO‐IBD. These registries enrolled patients prospectively from multiple sites throughout Germany (n = 82) and collected data under routine clinical care conditions using harmonised documentation protocols and predefined quality‐assured procedures, including remote and on‐site monitoring. Each contributing registry obtained approval from the relevant ethics committees, and all patients provided written informed consent. The study was conducted in accordance with the principles of the Declaration of Helsinki and its subsequent amendments. BioCrohn contains clinical data from 1530 patients with CD enrolled between 2008 and 2012 (registry duration: 2008–22). The registry includes information on the date of diagnosis and treatment initiation, which allows identification of patients with early disease (≤ 12 months' duration) and longitudinal follow‐up [32]. RUN‐CD consecutively enrolled 901 patients with CD initiating a new biologic therapy between 2017 and 2020 and followed them longitudinally for up to 3 years (registry duration: 2017–24) [33]. VEDO‐IBD was established to assess vedolizumab versus biologics as induction and maintenance therapy in the routine care of UC and CD patients (enrolment 2017–21; registry duration: 2017–22) and includes a total of 1284 patients, of whom 557 have CD [34]. The other registries that contribute data to the UMBRELLA‐IBD data warehouse, including data on ulcerative colitis, are shown in Table S1. Patient Cohort and Homogeneity Analysis For the present analysis, we included all patients who (a) had a confirmed diagnosis of CD according to the criteria of the German Society for Gastroenterology, Digestive and Metabolic Diseases (DGVS) [35] and the European Crohn's and Colitis Organisation (ECCO) [36], (b) were ≥ 18 years of age and (c) initiated a new biologic therapy. Patients with prior exposure to the index biologic or incomplete baseline data were excluded. A total of 1567 patients met the inclusion criteria. To assess the appropriateness of pooling data from the three contributing registries, we performed a homogeneity analysis based on registry‐specific 24‐month clinical remission rates defined using the Harvey–Bradshaw Index (HBI ≤ 4), as shown in Figure S1. Cochran's Q test indicated no statistically significant heterogeneity (Q = 2.94, df = 2, p = 0.230), and the I 2 statistic (32%) suggested only moderate heterogeneity, supporting the statistical appropriateness of pooling the data sets for joint analysis. This approach for pooling IPD is further supported by simulation studies showing that analyses based on IPD can provide more valid and precise estimates than aggregate‐data methods such as network meta‐analyses, especially when treatment effect modifiers are relevant [37]. Baseline Data and Treatment Regimens The following patient characteristics were recorded at baseline: age, sex, body mass index, disease duration, smoking status, family history of CD, prior biologic exposure and concomitant or recent medication use (glucocorticoids and immunosuppressants) before initiation of the index biologic. Clinical disease activity was assessed using the HBI. Patients initiated treatment with an anti‐TNF agent (adalimumab or infliximab), ustekinumab or vedolizumab in routine clinical care. Because this was a non‐interventional study, treatment selection, dosing and any dose adjustments were undertaken in accordance with the approved product labels and at the discretion of the treating physician. Outcomes and Definitions The main outcome was clinical remission (CR) at month 24, defined as an HBI score ≤ 4. Additional outcomes were CR at month 12; induction‐phase outcomes assessed at weeks 14–16 after treatment initiation, namely CR, steroid‐free CR (SFCR) and clinical response; SFCR at month 24; treatment persistence over 24 months; clinical disease activity over time assessed using the HBI; and health‐related quality of life assessed using the EQ‐VAS. SFCR was defined as CR (HBI ≤ 4) at the respective assessment time point in the absence of systemic corticosteroid use or oral budesonide during the 8 weeks preceding the assessment. Clinical response was defined as an improvement in HBI of ≥ 3 points from baseline and/or achievement of CR (HBI ≤ 4). Treatment persistence was defined as time on the index biologic therapy without discontinuation or switch. Statistical Analysis We performed descriptive analyses to characterise the overall population and treatment groups. Categorical variables are reported as absolute and relative frequencies, and continuous variables as mean with standard deviation or median with interquartile range (IQR), as appropriate. To reduce treatment selection bias and confounding, propensity scores (PS) were used to derive inverse probability of treatment weights (IPTW). For analyses comparing three treatment groups (anti‐TNF therapy [adalimumab or infliximab], ustekinumab and vedolizumab), the PS was estimated using a multinomial logistic regression model in which treatment group was regressed on the following baseline covariates: sex, age, smoking status, disease duration, family history of CD, body mass index, HBI at baseline, prior biologic exposure and prior use of glucocorticoids and immunosuppressants. For drug‐specific analyses comparing four treatment groups (adalimumab, infliximab, ustekinumab and vedolizumab), IPTW was derived analogously using a multinomial PS model with four treatment levels. Weights were stabilised and truncated at the extreme percentiles (1st and 99th percentiles). Covariate balance before and after weighting was assessed using standardised mean differences (SMDs) accounting for the weights. SMDs < 0.1 were considered indicative of adequate balance for the most important baseline parameters. For less prognostically important parameters, an SMD of up to 0.2 was considered acceptable [38, 39, 40, 41, 42]. Balance before and after IPTW is shown in Table 1 and Figure S2. TABLE 1Baseline characteristics of patients with Crohn's disease (a) before and (b) after IPTW.(a) Baseline characteristics of patients with Crohn's disease before IPTWBaseline characteristicsAnti‐TNF (n = 935)UST (n = 364)VDZ (n = 268)Global p valueSex0.032* Female50.4%56.9%57.5%Male49.6%43.1%42.5%Prior biologic exposure21.3%84.9%46.3%< 0.001** Smoker31.7%29.1%24.6%0.080Family history of CD11.6%15.9%14.2%0.089Previous steroid use88.9%95.3%90.7%0.002** Previous immunomodulator use68.2%79.4%64.6%< 0.001** Baseline HBI4.66 ± 4.385.89 ± 4.834.41 ± 3.92< 0.001** Age (mean ± SD)38.27 ± 13.0241.46 ± 14.1143.15 ± 15.23< 0.001** Disease duration, yrs9.85 ± 9.4012.84 ± 10.5111.61 ± 11.77< 0.001** BMI, kg/m2 (mean ± SD)24.73 ± 5.5825.35 ± 5.3325.69 ± 5.480.021* (b) Baseline characteristics of patients with Crohn's disease after IPTWBaseline characteristicsAnti‐TNF (n = 944)UST (n = 358)VDZ (n = 266)Global p valueSex0.962Female53.1%52.2%52.6%Male46.9%47.8%47.4%Prior biologic exposure40.9%40.9%42.9%0.843Smoker29.7%31.6%27.2%0.494Family history of CD12.5%14.2%13.2%0.706Previous steroid use90.5%93.0%89.4%0.242Previous immunomodulator use68.8%69.6%69.1%0.968Baseline HBI4.87 ± 4.515.15 ± 4.404.93 ± 4.470.605Age (mean ± SD)40.03 ± 13.1940.89 ± 14.1939.39 ± 14.670.381Disease duration, yrs10.97 ± 9.7211.75 ± 11.0510.70 ± 10.940.370BMI, kg/m2 (mean ± SD)25.26 ± 5.9325.24 ± 5.4325.18 ± 5.190.979Abbreviations: anti‐TNF, anti‐tumour necrosis factor; BMI, body mass index; CD, Crohn's disease; HBI, Harvey–Bradshaw Index; IPTW, inverse probability of treatment weighting; SD, standard deviation; UST, ustekinumab; VDZ, vedolizumab; yrs, years.* p < 0.05.** p < 0.001. Baseline characteristics of patients with Crohn's disease (a) before and (b) after IPTW. Abbreviations: anti‐TNF, anti‐tumour necrosis factor; BMI, body mass index; CD, Crohn's disease; HBI, Harvey–Bradshaw Index; IPTW, inverse probability of treatment weighting; SD, standard deviation; UST, ustekinumab; VDZ, vedolizumab; yrs, years. p < 0.05. p < 0.001. Analyses were conducted using modified intention‐to‐treat (mITT) and as‐observed approaches. The mITT population included all patients who initiated the index biologic therapy and had an HBI assessment at the respective endpoint. Patients who discontinued or switched from the index therapy before endpoint assessment were classified as outcome failures. Complete information was required for medication and remission status, whereas partial missingness in other variables was tolerated under the assumption that data were missing at random. The as‐observed analysis included only patients remaining on the index biologic at each time point and used observed data without imputation. Baseline differences between treatment groups were assessed using chi‐squared tests or Fisher's exact test for categorical variables, as appropriate. Changes from baseline to each visit in ordinal or continuous variables were analysed using paired Wilcoxon tests or paired t‐tests, depending on the distribution of the data. A multivariable logistic regression analysis was performed to identify baseline factors associated with CR at 24 months. Candidate variables included treatment group, age, sex, body mass index, smoking status, disease duration, family history of IBD, prior biologic exposure, prior glucocorticoid and immunosuppressant use, and baseline HBI. Anti‐TNF therapy was used as the reference category for treatment‐group comparisons. Variables were selected using backward stepwise selection based on Wald statistics, and results are reported as odds ratios (ORs) with 95% confidence intervals. Treatment persistence over 24 months was evaluated using Kaplan–Meier methods and compared between groups using log‐rank tests. EQ‐VAS outcomes over follow‐up were analysed using a weighted general linear model. Bonferroni correction was applied to all post hoc tests to control for type I error. A two‐sided p value < 0.05 was considered statistically significant. All statistical analyses were conducted using IBM SPSS Software Version 29. Results Baseline Characteristics and Covariate Balance Baseline characteristics of the study population before and after IPTW are shown in Table 1. Prior to weighting, relevant imbalances between treatment groups were observed for several baseline characteristics. After IPTW, covariate balance in the adjusted pseudo‐cohort improved substantially, with SMDs less than 0.1 for all important baseline parameters, indicating successful balancing (Figure S2). Because of the non‐interventional design of the contributing registries, baseline CRP (n = 1388), faecal calprotectin (n = 417) and endoscopic findings (n = 290) were not available for all patients. Mean baseline CRP differed slightly between the treatment groups: 34.9 mg/L in the anti‐TNF group, 71.5 mg/L in the ustekinumab group and 37.9 mg/L in the vedolizumab group. CRP was significantly higher with ustekinumab than with anti‐TNF therapy (p < 0.001) or vedolizumab (p = 0.010), whereas the anti‐TNF and vedolizumab groups did not differ significantly. Mean faecal calprotectin did not differ significantly between the three treatment groups (anti‐TNF: 611 μg/g stool; ustekinumab: 752 μg/g stool; vedolizumab: 476 μg/g stool; p = 0.247). Similarly, the proportion of patients with deep ulcers on ileocolonoscopy did not differ significantly between groups (anti‐TNF: 63.1%; ustekinumab: 82.6%; vedolizumab: 70.9%; p = 0.094). In the overall cohort, disease location according to the Montreal classification was L1 in 29.8%, L2 in 21.0%, L3 in 48.6% and L4 in 0.5% of patients. Disease behaviour was B1 in 54.8%, B2 in 23.8% and B3 in 21.5%. The distribution of Montreal location and behaviour was balanced across the three treatment groups, with no statistically significant differences. At baseline, concomitant azathioprine was used by 186 of 935 patients (19.9%) in the anti‐TNF group, 30 of 364 patients (8.2%) in the ustekinumab group and 28 of 268 patients (10.4%) in the vedolizumab group. Perianal fistulas at baseline were present in 11.8% of patients in the anti‐TNF group, in 13.2% of patients in the ustekinumab group and in 12.2% of patients in the vedolizumab group. There were no statistically significant differences between the groups. Treatment Persistence Treatment persistence over 24 months differed significantly between biologic therapies (Figure 1; log‐rank p < 0.001). Adalimumab showed the highest persistence (81.4%) and infliximab the lowest (66.6%), with ustekinumab and vedolizumab in between (78.6% and 76.9%). Treatment persistence over 24 months following initiation of biologic therapy in patients with Crohn's disease from the BioCrohn, VEDO‐IBD and RUN‐CD registries. Clinical Remission and Response During Induction After IPTW adjustment, induction outcomes (weeks 14–16) differed between treatment groups (Table 2). CR rates were highest among patients treated with anti‐TNF therapy (74.5%) compared with ustekinumab (69.4%) and vedolizumab (65.3%; global p = 0.035). SFCR was also more frequently achieved with anti‐TNF therapy (66.1%) than with ustekinumab (57.4%) or vedolizumab (58.6%; p = 0.028). Similarly, clinical response rates were highest in the anti‐TNF group (77.9%), followed by ustekinumab (73.5%) and vedolizumab (67.8%; p = 0.015). In pairwise comparisons, anti‐TNF therapy was associated with significantly higher rates of CR, SFCR and clinical response compared with vedolizumab, whereas differences between anti‐TNF therapy and ustekinumab were significant only for SFCR. No statistically significant differences in the various induction outcomes were observed between ustekinumab and vedolizumab. TABLE 2IPTW‐adjusted remission (clinical and steroid‐free) and clinical response following induction (weeks 14–16) in patients with Crohn's disease treated with anti‐TNF, ustekinumab or vedolizumab (VEDO‐IBD and RUN‐CD).OutcomeAnti‐TNF (n = 450)UST (n = 316)VDZ (n = 235)Global p valueClinical remission74.5%69.4%65.30.035Steroid‐free clinical remission66.1%57.4%58.60.028Clinical response77.9%73.5%67.80.015 Note: Anti‐TNF includes adalimumab and infliximab. Induction‐period outcomes were available for VEDO‐IBD and RUN‐CD. Pairwise comparisons: anti‐TNF vs. VDZ was significant for clinical remission (p = 0.011), steroid‐free clinical remission (p = 0.047) and clinical response (p = 0.005); anti‐TNF vs. UST was significant for steroid‐free clinical remission (p = 0.028). Other pairwise comparisons were not significant.Abbreviations: Anti‐TNF, anti‐tumour necrosis factor; IPTW, inverse probability of treatment weighting; UST, ustekinumab; VDZ, vedolizumab. IPTW‐adjusted remission (clinical and steroid‐free) and clinical response following induction (weeks 14–16) in patients with Crohn's disease treated with anti‐TNF, ustekinumab or vedolizumab (VEDO‐IBD and RUN‐CD). Note: Anti‐TNF includes adalimumab and infliximab. Induction‐period outcomes were available for VEDO‐IBD and RUN‐CD. Pairwise comparisons: anti‐TNF vs. VDZ was significant for clinical remission (p = 0.011), steroid‐free clinical remission (p = 0.047) and clinical response (p = 0.005); anti‐TNF vs. UST was significant for steroid‐free clinical remission (p = 0.028). Other pairwise comparisons were not significant. Abbreviations: Anti‐TNF, anti‐tumour necrosis factor; IPTW, inverse probability of treatment weighting; UST, ustekinumab; VDZ, vedolizumab. Maintenance Outcomes at 6, 12 and 24 Months After IPTW adjustment, the proportions of patients who achieved CR and SFCR by month 6 did not differ significantly between the anti‐TNF, ustekinumab and vedolizumab groups (Table S2). CR rates at 12 and 24 months after IPTW adjustment are shown in Figure 2. At 12 months (anti‐TNF, n = 935; UST, n = 364; VDZ, n = 268), CR was observed in 62.0% of patients treated with anti‐TNF therapy, 67.9% of those treated with ustekinumab, and 56.2% of those treated with vedolizumab (p = 0.011). Pairwise comparisons indicated that the difference between ustekinumab and vedolizumab at this time point was statistically significant (p = 0.003). At 24 months (anti‐TNF, n = 719; UST, n = 286; VDZ, n = 187), the corresponding CR rates were 53.4% (anti‐TNF), 63.5% (UST) and 58.7% (VDZ). Here, pairwise comparisons indicated that the difference between ustekinumab and anti‐TNF at this time point was statistically significant (p = 0.004) (Figure 2). Comparing point estimates at 12 and 24 months within each treatment group, we observed a slight decrease in CR with anti‐TNF therapy and ustekinumab, whereas the estimate for vedolizumab was stable with a minimal increase (Figure 2). These within‐group differences were not statistically significant. IPTW‐adjusted clinical remission at 12 and 24 months in patients with Crohn's disease treated with anti‐TNF therapy, ustekinumab or vedolizumab. Anti‐TNF, anti‐tumour necrosis factor; IPTW, inverse probability of treatment weighting; UST, ustekinumab; VDZ, vedolizumab. In additional IPTW‐adjusted analyses, we compared 12‐month CR rates by baseline Montreal disease location (L1 vs. L2/3) and behaviour (B1 vs. B2/3). CR did not differ by disease location (L1: 61.6%; L2/3: 62.1%; p = 0.864). In contrast, CR was slightly but significantly more frequent in patients with non‐stricturing, non‐penetrating disease behaviour (B1) than in those with stricturing or penetrating disease behaviour (B2/3) (64.7% vs. 59.7%; p = 0.047). In the multivariable logistic regression analysis (Table 3), after adjusting for potential confounding variables, ustekinumab and vedolizumab were associated with higher odds of CR at 24 months compared with anti‐TNF therapy (ustekinumab: odds ratio [OR] 1.48; p = 0.031; vedolizumab: OR 1.43; p = 0.049). Prior immunomodulator use was also associated with higher odds of CR (OR 1.36; p = 0.033). In contrast, prior biologic exposure (OR 0.696; p = 0.023), family history of IBD (OR 0.659; p = 0.018) and higher baseline HBI (OR per 1‐point increase 0.880; p = 0.001) were associated with lower odds of CR. BMI and smoking status were included in the model but were not significantly associated with CR at 24 months. Among the variables included in the model, baseline HBI showed the strongest statistical association with CR (p = 0.001). Each one‐point increase in HBI was associated with a 12% reduction in the odds of achieving CR at 24 months (OR = 0.88). TABLE 3Multivariable logistic regression analysis of baseline factors associated with clinical remission at 24 months.Baseline characteristicsOR95% CI p UST vs. anti‐TNF 1.483 1.038–2.120 0.031* VDZ vs. anti‐TNF 1.426 1.001–2.031 0.049* Prior biologic exposure 0.696 0.509–0.952 0.023* Age0.9990.989–1.0090.825Disease duration0.9970.984–1.0110.720Sex1.1850.931–1.5080.168BMI0.9870.965–1.0090.243Current smoker0.8480.652–1.1040.221Family history of IBD 0.659 0.466–0.932 0.018* Previous steroid use0.8830.567–1.3740.580Previous immunomodulator use 1.363 1.026–1.811 0.033* Baseline HBI 0.880 0.853–0.909 0.001* Note: For treatment comparisons, anti‐TNF therapy was the reference group. ORs > 1 indicate higher odds of clinical remission at 24 months. *p‐value < 0.05 was statistically significant.Abbreviations: anti‐TNF, anti‐tumour necrosis factor; BMI, body mass index; CI, confidence interval; HBI, Harvey–Bradshaw Index; IBD, inflammatory bowel disease; OR, odds ratio; UST, ustekinumab; VDZ, vedolizumab. Multivariable logistic regression analysis of baseline factors associated with clinical remission at 24 months. Note: For treatment comparisons, anti‐TNF therapy was the reference group. ORs > 1 indicate higher odds of clinical remission at 24 months. *p‐value < 0.05 was statistically significant. Abbreviations: anti‐TNF, anti‐tumour necrosis factor; BMI, body mass index; CI, confidence interval; HBI, Harvey–Bradshaw Index; IBD, inflammatory bowel disease; OR, odds ratio; UST, ustekinumab; VDZ, vedolizumab. In a subgroup analysis restricted to bio‐naïve patients, CR at 24 months was achieved by 56.5% of those treated with anti‐TNF therapy, 65.7% of those treated with ustekinumab and 71.6% of those treated with vedolizumab. In pairwise comparisons, CR rates were higher with vedolizumab than with anti‐TNF therapy (p = 0.004), whereas differences between anti‐TNF therapy and ustekinumab and between ustekinumab and vedolizumab were not statistically significant (Figure 3). IPTW‐adjusted clinical remission at 24 months in bio‐naïve patients with Crohn's disease treated with anti‐TNF therapy, ustekinumab or vedolizumab. At month 24, IPTW‐adjusted CR and SFCR rates differed among the individual therapies (adalimumab, infliximab, ustekinumab and vedolizumab), with the anti‐TNF group analysed by agent rather than as a pooled class (Table 4). Similar findings for CR at months 6 and 12 are shown in Tables S3 and S4. CR at month 24 was most frequent among patients treated with ustekinumab (63.3%), followed by adalimumab (59.1%) and vedolizumab (57.6%), and was lowest among infliximab‐treated patients (42.6%; global p < 0.001). A similar pattern was observed for SFCR at month 24, which was achieved by 58.4% of ustekinumab‐treated patients, 55.0% of vedolizumab‐treated patients and 53.6% of adalimumab‐treated patients, compared with 41.6% of infliximab‐treated patients (global p < 0.001). In pairwise comparisons, CR and SFCR rates were significantly higher with ustekinumab, adalimumab and vedolizumab than with infliximab (Table 4). TABLE 4IPTW‐adjusted clinical and steroid‐free clinical remission at month 24 in patients with Crohn's disease treated with adalimumab, infliximab, ustekinumab or vedolizumab (BioCrohn, RUN‐CD and VEDO‐IBD).OutcomeADA (n = 456)IFX (n = 264)UST (n = 287)VDZ (n = 190)Global p valueClinical remission59.1%42.6%63.3%57.6%< 0.001Steroid‐free clinical remission53.6%41.6%58.4%55.0%< 0.001 Note: Pairwise comparisons vs. IFX: clinical remission: UST p < 0.001, ADA p < 0.001, VDZ p = 0.002; steroid‐free clinical remission: UST p < 0.001, ADA p = 0.002, VDZ p = 0.006.Abbreviations: ADA, adalimumab; IFX, infliximab; IPTW, inverse probability of treatment weighting; UST, ustekinumab; VDZ, vedolizumab. IPTW‐adjusted clinical and steroid‐free clinical remission at month 24 in patients with Crohn's disease treated with adalimumab, infliximab, ustekinumab or vedolizumab (BioCrohn, RUN‐CD and VEDO‐IBD). Note: Pairwise comparisons vs. IFX: clinical remission: UST p < 0.001, ADA p < 0.001, VDZ p = 0.002; steroid‐free clinical remission: UST p < 0.001, ADA p = 0.002, VDZ p = 0.006. Abbreviations: ADA, adalimumab; IFX, infliximab; IPTW, inverse probability of treatment weighting; UST, ustekinumab; VDZ, vedolizumab. Clinical Remission at 24 Months Among Induction Responders Among patients who achieved a clinical response during induction, IPTW‐adjusted CR rates at 24 months differed between treatment groups (Figure 4). CR at 24 months was achieved by 56.0% of induction responders treated with anti‐TNF therapy, 69.1% of those treated with ustekinumab and 73.7% of those treated with vedolizumab. In pairwise comparisons, CR rates at 24 months were higher with ustekinumab than with anti‐TNF therapy (p = 0.005) and with vedolizumab than with anti‐TNF therapy (p < 0.001). No statistically significant difference was observed between ustekinumab and vedolizumab. Similar findings were observed at 12 months, although the absolute differences were smaller: CR rates among induction responders were 67.0% with anti‐TNF therapy, 80.2% with ustekinumab and 70.8% with vedolizumab. Pairwise comparisons showed significantly higher CR rates with ustekinumab and vedolizumab than with anti‐TNF therapy (p < 0.05 for both comparisons), whereas ustekinumab and vedolizumab did not differ significantly. 24‐month clinical remission in induction responders after IPTW. Clinical Disease Activity Over 24 Months Clinical disease activity over time is shown in Figure 5 for the unweighted (non‐IPTW) cohort. At baseline, HBI was significantly higher in patients initiating ustekinumab than in those initiating anti‐TNF therapy or vedolizumab (p < 0.001). In all treatment groups, HBI decreased significantly from baseline to the end of induction and remained lower at all subsequent follow‐up visits, including at 6, 12 and 24 months (p < 0.001 for each therapy). In contrast to the IPTW‐adjusted remission analyses described above, no statistically significant between‐group differences in HBI were observed between treatment groups at follow‐up visits after baseline in this unweighted cohort. Overall, HBI decreased over 24 months in all treatment groups, with comparable levels between the groups after induction (Figure 5). HBI over 24 months in patients with Crohn's disease initiating a new biologic therapy (anti‐TNF therapy, ustekinumab or vedolizumab). Anti‐TNF, anti‐tumour necrosis factor; HBI, Harvey–Bradshaw Index; IPTW, inverse probability of treatment weighting; UST, ustekinumab; VDZ, vedolizumab. Health‐Related Quality of Life Over 24 Months Health‐related quality of life, assessed after IPTW adjustment using the EQ‐VAS, improved substantially over time in all treatment groups (Figure 6). At baseline, mean EQ‐VAS values were lowest among ustekinumab‐treated patients and highest among patients treated with anti‐TNF therapy and vedolizumab. By 6 months, EQ‐VAS scores increased markedly in all groups and continued to improve or stabilise through 12 and 24 months. From baseline to month 24, the increase in EQ‐VAS was statistically significant within each group (p < 0.001 for all therapies). After IPTW adjustment, no statistically significant between‐group differences in EQ‐VAS were observed over time (p = 0.736). These findings indicate sustained improvements in patient‐reported health status over 24 months that were comparable among biologic therapies in patients who remained on treatment without a switch or stoma. EQ‐VAS over 24 months in patients with Crohn's disease without treatment switch or stoma after IPTW adjustment. Anti‐TNF, anti‐tumour necrosis factor; EQ‐VAS, EuroQol Visual Analogue Scale; UST, ustekinumab; VDZ, vedolizumab. Discussion Head‐to‐head evidence from RCTs in CD is still limited. In the SEAVUE randomised trial in biologic‐naïve patients, ustekinumab and adalimumab showed similar effectiveness after induction and at 12 months, with high clinical remission rates in both arms [17]. In the open‐label STARDUST trial, a treat‐to‐target strategy with ustekinumab did not yield clearly superior outcomes compared with standard care at 52 weeks [18]. Real‐world comparative analyses complement these trial findings by assessing treatment outcomes under routine‐care conditions in broader patient populations and over longer follow‐up. In this pooled analysis of prospective real‐world registry data from the UMBRELLA‐IBD data warehouse, we compared the effectiveness of anti‐TNF therapy, ustekinumab and vedolizumab in CD over up to 24 months. Using propensity‐score‐based IPTW to address baseline differences between treatment groups, we assessed CR and clinical response after induction and at prespecified follow‐up assessments during maintenance therapy, as well as disease activity and health‐related quality of life. Overall, CR rates were high among all evaluated biologics, and between‐therapy differences were small at each assessment time point. After induction, CR and clinical response rates were slightly higher with anti‐TNF therapy, whereas at later maintenance assessments, small, time‐point‐specific differences were observed. These included significantly higher CR rates for ustekinumab in the relevant comparisons (vs. vedolizumab at 12 months and vs. anti‐TNF at 24 months) using propensity‐score‐based IPTW adjustment. In subgroup analyses, 24‐month CR was higher with vedolizumab and ustekinumab than with anti‐TNF among induction responders, and higher with vedolizumab than with anti‐TNF therapy in biologic‐naïve patients. The results of our adjusted multivariable analyses support the clinical plausibility of the CR findings, with higher odds of 24‐month CR associated with ustekinumab and vedolizumab use (vs. anti‐TNF therapy), lower baseline disease activity and fewer markers of more difficult‐to‐treat disease. Among the variables included in the model, baseline HBI showed the strongest statistical association with 24‐month CR (p = 0.001). Each 1‐point increase in HBI was associated with a 12% reduction in the odds of achieving CR at this time point (OR = 0.88). Altogether, these findings suggest that all of the evaluated biologics can achieve meaningful benefit in specialised routine care, and that comparative differences may vary by assessment time point and patient subgroup. This interpretation is consistent with other recently published real‐world evidence. A systematic review and meta‐analysis of real‐world studies reported favourable long‐term effectiveness and treatment persistence for vedolizumab compared with anti‐TNF therapy, including during longer follow‐up and in biologic‐naïve populations [31]. In addition, in a propensity‐score‐matched real‐world comparison among patients with CD previously exposed to anti‐TNF therapy, ustekinumab showed higher remission rates than vedolizumab at 14 weeks, whereas differences at 52 weeks were not statistically significant [43]. Other RWE studies also support ustekinumab as an effective option in routine care [19, 22, 23, 24, 25, 26, 27, 28, 29]. In a real‐world induction cohort comparing adalimumab and ustekinumab, adalimumab showed higher overall response and remission rates, but subgroup analyses suggested broadly similar outcomes in anti‐TNF–experienced patients, with a numerical signal favouring ustekinumab for clinical response [30]. We observed marked differences in treatment persistence between individual biologics, with adalimumab showing the highest and infliximab the lowest persistence over 24 months. Persistence is commonly used as a surrogate endpoint in observational research because it incorporates multiple dimensions of real‐world treatment success, including perceived effectiveness and satisfaction among both patients and physicians, and it may also be influenced by tolerability or discontinuation for any reason, including adverse events [44, 45]. In routine clinical care, persistence may be further affected by practical considerations such as treatment burden, route and logistics of administration, monitoring requirements, and patient or physician preferences. Accordingly, it is not unexpected that persistence and remission outcomes did not correspond fully in our cohort. This underscores that treatment persistence in routine practice captures factors beyond clinical remission. Additionally, health‐related quality of life assessed using the EQ‐VAS improved over time, and after adjustment, we did not observe relevant between‐group differences among patients who remained on therapy without a switch or stoma. Overall, these findings suggest that remission status alone may not fully capture clinically meaningful changes within the range of low disease activity, and that patients who respond and remain on their index biologic can maintain improved symptoms and quality of life over longer‐term follow‐up. Strengths and Limitations A strength of the present study is the pooling of individual patient data (IPD) from multiple prospective German IBD registries in the UMBRELLA‐IBD data warehouse, resulting in a large cohort of more than 1500 patients. This enabled harmonised outcome definitions and consistent analytic approaches among the registries. Treatment comparisons were adjusted for baseline differences using IPTW based on propensity scores. Using patient‐level data also allowed the same covariates to be applied consistently across all registries and outcomes rather than relying on study‐level summaries and indirect comparisons. Compared with network meta‐analyses based on aggregated data, pooled IPD approaches can reduce the heterogeneity introduced by differing outcome definitions and adjustment strategies and can support more consistent subgroup analyses when treatment effect modifiers are relevant [37]. Several limitations should be considered when interpreting our findings. First, as this was an observational study, treatment allocation was not randomised, and residual confounding cannot be excluded even after IPTW adjustment. The use of standardised prospective data collection across the registry design and standardised documentation supports data quality. However, as with all observational analyses using inverse probability of treatment weighting (IPTW), the validity of the adjusted comparisons depends on the assumption that all relevant confounders were measured and adequately modelled. Although weighting was performed based on a comprehensive set of available baseline characteristics, unmeasured confounding cannot be completely ruled out, and the inferential weight of these adjusted comparisons remains lower than that of randomised controlled trial data. Second, the pooled analysis combined data from independent registries. Although documentation procedures and the definitions of variables were harmonised through UMBRELLA‐IBD, differences in site‐level practice, follow‐up intensity, and data completeness may have influenced the observed outcomes. However, heterogeneity in CR at 24 months among the contributing registries was moderate and not statistically significant (Cochran's Q, p = 0.230; I 2 = 32%), supporting the appropriateness of pooling the data for the main endpoint. Third, because of the non‐interventional design of the contributing registries, some outcomes were available only in subsets of patients or registries (e.g., induction‐phase endpoints, therapeutic drug monitoring and immunogenicity data). This reduced the sample size for these analyses and may limit comparability between time points. Fourth, the study reflects routine care in Germany, which may limit generalisability to other health systems. At the same time, the multicentre design and inclusion of patients treated in everyday practice throughout the country support the external validity of the findings for similar routine‐care settings. Conclusions To our knowledge, this is the largest prospective real‐world comparison of biologic effectiveness in Crohn's disease based on pooled individual patient data from registries, including more than 1500 patients. In this analysis from the UMBRELLA‐IBD data warehouse, anti‐TNF therapy, ustekinumab and vedolizumab were each associated with improved clinical and patient‐reported outcomes over up to 24 months. Although some pairwise comparisons reached statistical significance, the absolute differences in clinical remission rates were small: anti‐TNF therapy versus vedolizumab at the end of induction, ustekinumab versus vedolizumab at 12 months, and ustekinumab versus anti‐TNF therapy at 24 months. In subgroup analyses, 24‐month CR rates were slightly higher with vedolizumab than with anti‐TNF therapy in biologic‐naïve patients, and higher with vedolizumab and ustekinumab than with anti‐TNF therapy among induction responders, but, again, the absolute differences were small. Treatment persistence varied markedly across biologics, indicating that long‐term persistence in routine practice is not fully explained by remission outcomes alone. Together with sustained improvements in disease activity and health‐related quality of life among patients remaining on therapy, these findings suggest that durable clinical benefit can be achieved with various biologic options in specialised routine care. These real‐world data complement evidence from randomised trials by reflecting comparative effectiveness and treatment persistence under routine‐care conditions in broader, more heterogeneous patient populations and over longer follow‐up. In summary, the overall effectiveness of all therapies was high in this real‐world cohort, regardless of the biologic used. Further research should evaluate newer therapies and incorporate endpoints beyond clinical remission to better characterise long‐term treatment benefit. Author Contributions Arne Bokemeyer: conceptualization, methodology, investigation, validation, formal analysis, supervision, visualization, project administration, writing – original draft. Bernd Bokemeyer: conceptualization, methodology, data curation, investigation, validation, formal analysis, supervision, funding acquisition, visualization, project administration, resources, writing – original draft. Stefan Schreiber: conceptualization, methodology, investigation, validation, formal analysis, supervision, visualization, project administration, writing – review and editing, funding acquisition. Florian Tran: conceptualization, methodology, investigation, validation, formal analysis, supervision, visualization, project administration, writing – original draft. Thomas Wenske: software, data curation, investigation, validation, formal analysis, writing – review and editing. Sandra Plachta‐Danielzik: conceptualization, methodology, data curation, investigation, validation, formal analysis, funding acquisition, project administration, writing – review and editing. Elena Gilman: data curation, investigation, software, formal analysis, validation, writing – review and editing. Matthew Gaskins: writing – review and editing, formal analysis, visualization. Funding This work was supported by the Competence Network IBD in Germany. Disclosure Guarantor of the article: Priv.‐Doz. Dr. Arne Bokemeyer. Ethics Statement Study procedures were approved (BioCrohn: AZ A161/08; Run‐CD: AZ D575/16; VEDO‐IBD: AZ D461/(17)) by the Ethics Committee of the Faculty of Medicine of Kiel University, Germany, in compliance with the Professional Code of Conduct of the Schleswig‐Holstein Chamber of Physicians. Consent All study participants provided written informed consent before study enrolment. Conflicts of Interest A.B. has received consulting fees and lectures from AbbVie, Alfasigma, Amgen, BMS, CED Service GmbH, Falk, Ferring, Galapagos, Janssen, Pfizer, Sanofi and Takeda outside the submitted work. F.T. has received speaker's fees from AbbVie, Bristol‐Myers‐Squibb, Celltrion Healthcare, Dr. Falk Pharma, Eli Lilly, Ferring Pharmaceuticals, J&J, Sanofi, Takeda, consultancy fees from AbbVie, J&J and Takeda, and scientific funding from Sanofi/Regeneron. T.W. does not have any relevant financial or non‐financial interests to disclose. S.P.‐D. does not have any relevant financial or non‐financial interests to disclose. E.G. does not have any relevant financial or non‐financial interests to disclose. M.G. have received fees from Ferring and from various German and Swiss universities for medical writing and scientific language editing, outside the submitted work. S.S. has received personal fees from AbbVie, Arena, BMS, Biogen, Celltrion, Celgene, IMAB, Gilead, HIMA, IMAB‐Biopharma, MSD, Mylan, Pfizer, Fresenius, Janssen, Takeda, Theravance, Provention Bio, Protagonist and Falk outside the submitted work. B.B. has received consulting fees from: AbbVie, MSD, Shire, Ferring, Hospira, Takeda, Galapagos, Movetis, Shield Therapeutics, Pfizer, Biogen, Janssen, Hexal, Lilly, Hexal and Celgene, outside the submitted work. Financial support for lectures and teaching from: AbbVie, Ferring, Galapagos, MSD, Merckle, Falk, HLR, Shield Therapeutics, Pfizer, Celltrion, Takeda, Lilly, Janssen and Mundipharma, outside the submitted work. Grant and financial support for research from AbbVie, Pfizer, Lilly, Janssen, Takeda and Ferring, outside the submitted work. Supporting information Figure S1: 24‐month clinical remission rates with 95% confidence intervals among 1567 biologic‐treated patients (anti‐TNF, n = 935; UST, n = 364; VDZ, n = 268) from the three contributing registries (RUN‐CD, VEDO‐IBD and BioCrohn study) included in the pooled analysis. Figure S2: Covariate balance after weighting: standardised mean differences for baseline characteristics included in the primary IPTW model. Table S1: Prospective Competence Network IBD studies and UMBRELLA‐IBD database (status 08/2025). Table S2: IPTW‐adjusted clinical remission and steroid‐free remission at month 6 in patients with Crohn's disease treated with anti‐TNF therapy, ustekinumab or vedolizumab. Table S3: IPTW‐adjusted clinical and steroid‐free clinical remission at month 6 in patients with Crohn's disease treated with adalimumab, infliximab, ustekinumab or vedolizumab (BioCrohn, RUN‐CD and VEDO‐IBD). Table S4: IPTW‐adjusted clinical and steroid‐free clinical remission at month 12 in patients with Crohn's disease treated with adalimumab, infliximab, ustekinumab or vedolizumab (BioCrohn, RUN‐CD and VEDO‐IBD).