Comparative risk of Clostridioides difficile infection in patients with Crohn’s disease receiving ustekinumab versus anti-TNF therapy: a retrospective cohort study

Abstract

Background:

Antitumor necrosis factor (anti-TNF) agents are associated with an increased risk of Clostridioides difficile (C. difficile) infection (CDI). The risk associated with Ustekinumab (UST) in real-world clinical practice remains poorly defined.

Objectives:

To compare the incidence densities (IDs) and risks of CDI and C. difficile nucleic acid positivity (C. difficile positivity) between patients with Crohn’s disease (CD) initiating UST and those initiating anti-TNF therapy.

Design:

Single-center retrospective cohort study.

Methods:

In a retrospective cohort study, we included 627 CD patients with a negative baseline C. difficile nucleic acid test who initiated UST or anti-TNF therapy between November 2020 and October 2024. IDs and incidence density ratios (IDRs) were calculated. Hazard ratios (HRs) for CDI and C. difficile positivity were estimated using marginal structural models based on Cox regression.

Results:

The cohort included 277 patients (381 person-years, PY) treated with UST and 350 patients (454 PY) treated with anti-TNF agents. CDI developed in 4 of 277 UST-treated participants (1.05/100 PY) and 9 of 350 anti-TNF-treated participants (1.98/100 PY), corresponding to an IDR of 0.53 (95% confidence interval (CI): 0.14–1.99, p = 0.347). C. difficile positivity was recorded in 12 of 277 participants, compared with 25 of 350 participants, corresponding to IDs of 3.15 and 5.50/100 PY, respectively (IDR = 0.57, 95% CI: 0.27–1.19, p = 0.137). After adjustment, no significant differences were observed in the risk of CDI (HR = 0.51, 95% CI: 0.16–1.17, p = 0.274) or C. difficile positivity (HR = 0.59, 95% CI: 0.29–1.21, p = 0.151). An exploratory sensitivity analysis in biologic-naïve patients yielded an adjusted HR of 0.27 (95% CI: 0.05–1.30), which did not reach statistical significance (p = 0.102).

Conclusion:

In this real-world cohort, treatment with UST was not associated with a significantly higher risk of CDI or C. difficile positivity compared to anti-TNF therapy. The consistent direction of point estimates, suggesting a potential risk reduction especially in biologic-naïve patients, remains exploratory and warrants further investigation in larger studies.

Plain language summary

Comparing the risk of a serious gut infection between two different Crohn’s disease treatments

Background & Aim:

Patients with Crohn’s disease, a chronic condition affecting the gut, are often treated with powerful medications called biologics. Some of these, known as anti-TNF drugs, are known to increase the risk of a serious gut infection called Clostridioides difficile (C. diff). A newer biologic, Ustekinumab, is also used, but its real-world risk for this infection was less clear. This study aimed to directly compare the risk of C. diff infection between Crohn’s patients starting Ustekinumab and those starting anti-TNF drugs.

Methods & Results:

The researchers looked back at the medical records of 627 Crohn’s patients who started either Ustekinumab or an anti-TNF drug. All patients were tested and confirmed not to have C. diff at the start of the study. They then tracked how many patients later developed a C. diff infection or tested positive for the bacteria. The results showed that the risk of developing an actual C. diff infection was low in both groups. There was no statistically significant difference in risk between patients taking Ustekinumab and those taking anti-TNF drugs. The same was true for just testing positive for the bacteria. However, in patients who had never taken a biologic drug before, there was a noticeable (though not statistically certain) trend suggesting that Ustekinumab might be associated with a lower risk of C. diff infection.

Conclusion:

In this real-world study of Crohn’s patients, Ustekinumab and anti-TNF therapies were found to have a similar, low risk of causing C. diff infection. While the overall risks are comparable, the possibility that Ustekinumab might offer a safer profile for patients new to biologic drugs is an interesting finding that should be explored in future research.

Keywords

anti-TNF Crohn’s disease infection ustekinumab

Introduction

Clostridioides difficile (C. difficile) infection (CDI) is a significant opportunistic infection in patients with inflammatory bowel disease (IBD).^1,2 Treatment with antitumor necrosis factor (anti-TNF) agents, in particular, has been established as a risk factor for its development.^3–6 Ustekinumab (UST), a monoclonal antibody targeting the shared p40 subunit of interleukin-12 and interleukin-23, is an effective therapy for moderate-to-severe Crohn’s disease (CD).⁷ While its distinct mechanism of action may theoretically influence infection risk differently from anti-TNF agents, the comparative risk of CDI associated with UST in real-world clinical practice remains poorly characterized. Safety data from the UNITI/IM-UNITI clinical trials reported CDI rates of 1.08/100 person-years (PY) with UST versus 2.31/100 PY with placebo.^7,8 However, these findings require validation in routine clinical settings, and a direct comparison with anti-TNF agents is lacking. Therefore, we aimed to compare the incidence densities (IDs) and risks of CDI and C. difficile nucleic acid positivity (C. difficile positivity) between CD patients initiating UST and those initiating anti-TNF therapy in a real-world cohort.

Materials and methods

Study design and data source

We conducted a retrospective, real-world cohort study of CD patients at the Sixth Affiliated Hospital of Sun Yat-sen University. The study period spanned from November 2020 to October 2024, with follow-up until February 1, 2025. All CD patients with CDI or C. difficile positivity were identified from our medical record system. The study was approved by the Institutional Review Board of the hospital (Ethics Approval No: E2024166) and adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline.⁹

Study population

Adult patients (⩾18 years) with a confirmed diagnosis of CD according to standard clinical and endoscopic criteria^10,11 who initiated treatment with UST or an anti-TNF agent (infliximab or adalimumab) were considered for inclusion. All patients were required to have a documented negative nucleic acid test for toxigenic C. difficile within the 2 weeks preceding treatment initiation. Exclusion criteria were applied to minimize misclassification and confounding, as detailed below: (1) compromised outcome assessment: recent use (within 2 weeks prior to baseline) of medications active against C. difficile (e.g., oral vancomycin or metronidazole), or a positive baseline nucleic acid test for toxigenic C. difficile; (2) altered gastrointestinal anatomy or function: a history of colectomy or fecal microbiota transplantation, or the presence of an ileostomy or colostomy; (3) clinical status at high risk for confounding: pregnancy or lactation at enrollment, or the presence of significant concurrent comorbidities, including chronic liver or kidney disease, active tuberculosis, history of malignancy, or a spectrum of uncontrolled cardiometabolic, neurological, or systemic inflammatory conditions; (4) inadequate data or follow-up: incomplete clinical records or a post-baseline follow-up duration of less than 2 months.

Exposure and follow-up

The primary exposure was initiation of either UST or anti-TNF therapy. Patients were considered exposed from the first dose until 90 days after the last recorded dose.¹² The exposure period spanned from baseline until the first occurrence of any of the following: C. difficile positivity, 90 days after termination of treatment, or the end of the follow-up period (February 1, 2025).¹² Treatment was classified as terminated if an end date was documented in the medical records, if more than 90 days had elapsed since the last recorded infusion, or if more than 90 days had passed since the last prescription entry. Transitioning from a reference biologic to its biosimilar was not regarded as treatment termination.¹² Individuals could enter each exposure cohort separately after a 90-day washout, and sequential use of different anti-TNF agents was pooled within the anti-TNF cohort.¹³ Follow-up started at treatment initiation and ended at the earliest of: the outcome event, 90 days after treatment discontinuation, or the end of the study period.

Outcome definitions

The primary outcome was the occurrence of CDI, defined as a positive of C. difficile nucleic acid plus diarrhea (⩾3 times loose stools in 24 h), consistent with IDSA guidelines.¹⁴ The secondary outcome was C. difficile positivity, defined as a positive nucleic acid test for toxigenic C. difficile regardless of symptoms. Testing was performed using the GeneXpert^® C. difficile Assay on stool samples obtained within 48 h of admission.

Covariates

Time-fixed covariates (baseline) included:

(1) Demographic characteristics: gender, age, disease duration, diarrhea times daily, and body mass index (BMI); (2) disease phenotype according to the Montreal classification¹⁰; (3) disease activity measured by the Crohn’s Disease Activity Index (CDAI)¹⁰; (4) pre-enrollment clinical history: number of unplanned hospitalizations, medication history within 60 days (including corticosteroids, immunomodulators, and proton pump inhibitors (PPIs)), and history of antibiotic use within 90 days; (5) number of prior biologic agent exposures.

Time-varying covariates included: Certain covariates, which may be associated with both exposure and outcome and potentially influenced by prior biologic exposure, were considered time-varying based on previous literature.^5,13,15,16 Use of corticosteroids, immunomodulators, PPIs, and antibiotics during follow-up was treated as time-varying covariates, as they may affect the risks of both CDI and C. difficile positivity.

Statistical analysis

Patient characteristics are described as mean and standard deviation or median and interquartile range for continuous variables and as numbers of cases and proportions for categorical variables. The χ² or Fisher’s exact test was used to compare the proportions. Student’s t-test or its nonparametric counterparts were used to compare quantitative variables.

In the original cohort, the IDs and incidence density ratios (IDRs) of CDI and C. difficile positivity were calculated per treatment group. ID was expressed as the number of events per 100 PY with a 95% confidence interval (CI). Time-to-event outcomes were assessed using the Kaplan–Meier method, and between-group differences were compared with the log-rank test. The proportional hazards assumption was verified using Schoenfeld residual tests (all covariates p > 0.05). Kaplan–Meier curves were plotted to visualize survival functions.

To minimize the effect of confounding variables, inverse probability of treatment weighting (IPTW) based on propensity scores (PS) was applied.^17–19 The following potential confounders were included in the PS model based on clinical relevance and prior literature.^6,20,21 We included the following variables: gender, age, disease duration, unexpected hospitalizations, BMI, Montreal classification-defined disease location and behavior, baseline perianal disease, baseline CDAI, concomitant medications at baseline (including immunomodulators, corticosteroids, antibiotics, and PPIs), and the number of prior biologic exposures. First, the PS, defined as the probability of initiating UST treatment, was estimated using a logistic regression model. Second, stabilized weights were calculated as 1/PS for the UST group and 1/(1−PS) for the anti-TNF group. These stabilized weights were trimmed at the 1st and 99th percentiles and applied in subsequent survival analyses.

We conducted a tiered analysis using a marginal structural Cox model (MSM-Cox) to estimate the hazard ratio (HR) of UST versus anti-TNF, addressing confounding at different levels. The proportional hazards assumption was verified using Schoenfeld residuals and was not violated. Crude model (UST-base): A univariable Cox proportional hazards model with treatment group as the sole predictor was fitted to obtain the unadjusted association. The primary MSM analysis (UST + IPTW): To control for baseline confounding, we constructed an MSM using IPTW. The stabilized weights were derived from a logistic regression model to predict treatment assignment. A Cox model was then fitted, which included only the treatment group as a covariate and was weighted by the calculated IPTW. Exploratory analysis (UST + IPTW + timevars): To preliminarily explore the potential role of concomitant medication (such as immunomodulators, corticosteroids, PPIs, or antibiotics) initiated during follow-up between treatment groups and outcomes, we further adjusted for time-dependent concomitant medication as a time-varying covariate in the primary analysis model. Substantial attenuation of the HR in this model would indicate that the difference in CDI risk is mediated through the pathway of reduced requirement for concomitant medications.

All statistical tests were two-sided, and a p < 0.05 or a 95% CI excluding 1 was considered statistically significant. Analyses were performed using R software (version 4.4.3; R Foundation for Statistical Computing, Vienna, Austria).

A prespecified subgroup analysis and sensitivity analysis

Prespecified subgroup analyses were conducted stratified by (1) gender and (2) age at biologic initiation (<40 vs ⩾40 years). A prespecified exploratory sensitivity analysis restricted to biologic-naïve patients was performed to test the robustness of the primary findings.

Results

Patient characteristics

A total of 627 patients were included, of whom 350 received anti-TNF agents, and 277 received UST (Figure 1). The baseline characteristics of the CD patients are summarized in Table 1. Compared to patients treated with anti-TNF, those receiving UST were older (anti-TNF: 28.5 ± 8.60 vs UST: 31.8 ± 11.0 y, p < 0.001), had a longer disease duration (anti-TNF: 41.6 ± 42.4 vs UST: 55.7 ± 57.7 m, p = 0.001), a lower frequency of perianal disease (anti-TNF: 69.4% vs UST: 60.3%, p = 0.021), and a greater numbers of prior biologic therapies (anti-TNF: 0.32 ± 0.49 vs UST: 0.47 ± 0.63, p = 0.001). A higher proportion of patients in the UST group exhibited stricturing (B2) behavior (anti-TNF: 20.9% vs UST: 24.5%) and penetrating (B3) behavior (anti-TNF: 18.0% vs UST: 24.9%), whereas the anti-TNF group had a higher proportion of inflammatory (B1) behavior (anti-TNF: 61.1% vs UST: 50.5%). There was a trend toward more frequent recent immunomodulator use in the anti-TNF group (anti-TNF: 9.43% vs UST: 5.05%, p = 0.056). Gender, unexpected hospitalizations, diarrhea times, BMI, disease location, CDAI, and use of corticosteroids, antibiotics, and PPIs were similar between the two groups. After IPTW adjustment, no statistically significant differences were observed in any baseline variables between the two groups, and the standardized mean differences for all included variables were less than 0.1.

Figure 1.

Patient flowchart of CD patients treated with either anti-TNF or UST who are included in the current study.

Table 1.

Baseline characteristics of Crohn’s disease patients initiating anti-TNF or UST treatment before and after IPTW.

Variable	Before IPTW		p Value	After IPTW^a		p Value	SMD
Variable	Anti-TNF	UST	p Value	Anti-TNF	UST	p Value	SMD
N	350	277		622.29	631.96
Gender—male (N, %)	260 (74.3)	195 (70.4)	0.32	453.5 (72.9)	463.0 (73.3)	0.917	0.009
Age^b (years, mean (SD))	28.5 (8.6)	31.8 (11.0)	<0.001^c	29.61 (9.17)	29.76 (9.95)	0.854	0.015
Disease duration (months, mean (SD))	41.6 (42.4)	55.7 (57.7)	0.001^c	45.94 (45.76)	46.83 (50.34)	0.821	0.018
Unexpected hospitalizations (mean (SD))	0.15 (0.43)	0.18 (0.50)	0.39	0.16 (0.46)	0.16 (0.46)	0.903	0.010
Diarrhea times^b = 1(N, %)	139 (39.7)	99 (35.7)	0.35	237.1 (38.1)	240.1 (38.0)	0.978	0.002
BMI^b ((kg/m²), mean (SD))	19.5 (3.26)	19.4 (3.08)	0.634	19.45 (3.19)	19.50 (3.16)	0.828	0.018
Disease location^b (N, %)			0.052			0.999	0.003
L1 (ileum)	89 (25.4)	94 (33.9)		176.1 (28.3)	179.7 (28.4)
L2 (colon)	13 (3.71)	12 (4.33)		26.5 (4.3)	26.9 (4.3)
L3 (ileocolonic)	248 (70.9)	171 (61.7)		419.8 (67.5)	425.4 (67.3)
Disease behavior^b (N, %)			0.023^c			0.992	0.011
B1 (inflammatory disease)	214 (61.1)	140 (50.5)		348.8 (56.1)	357.0 (56.5)
B2 (stricturing disease)	73 (20.9)	68 (24.5)		141.2 (22.7)	143.2 (22.7)
B3 (penetrating disease)	63 (18.0)	69 (24.9)		132.3 (21.3)	131.8 (20.9)
Perianal disease^b = 1(N, %)	243 (69.4)	167 (60.3)	0.021^c	406.5 (65.3)	413.1 (65.4)	0.992	0.001
CDAI^b (mean (SD))	194 (93.6)	185 (91.6)	0.203	188.43 (91.24)	187.31 (91.40)	0.883	0.012
Concomitant medication^b, N (%)
IM^d = 1	33 (9.43)	14 (5.05)	0.056	46.7 (7.5)	49.7 (7.9)	0.884	0.014
Corticosteroids^d = 1	9 (2.57)	8 (2.89)	1.000	17.0 (2.7)	15.4 (2.4)	0.824	0.019
PPIs^d = 1	11 (3.14)	11 (3.97)	0.733	21.0 (3.4)	22.4 (3.5)	0.911	0.009
Antibiotics^e = 1	31 (8.86)	22 (7.94)	0.791	53.0 (8.5)	55.8 (8.8)	0.892	0.012
Previous types of biologics (mean (SD))	0.32 (0.49)	0.47 (0.63)	0.001^c	0.37 (0.52)	0.38 (0.58)	0.878	0.013

Values in the weighted cohort are presented as weighted counts (or means) and percentages. Weighting was performed using IPTW to balance baseline characteristics.

At baseline.

p < 0.05 indicates statistical significance.

60 days prior to baseline.

90 days prior to baseline.

Anti-TNF, antitumor necrosis factor; BMI, body mass index; CDAI, Crohn’s disease activity index; IM, immunomodulator, including azathioprine, 6-mercaptopurine, thalidomide, and methotrexate; IPTW, inverse probability of treatment weighting; N, number of; PPIs, proton pump inhibitors; SD, standard deviation; SMD, standardized mean differences; UST, ustekinumab.

CDI and C. difficile positivity outcomes

Among the 277 patients treated with UST (381 PY, mean follow-up 1.38 years), CDI and C. difficile positivity occurred in 4 and 12 patients, respectively. The IDs were 1.05/100 PY (95% CI: 0.29–2.69) for CDI and 3.15/100 PY (95% CI: 1.63–5.50) for C. difficile positivity (Table 2). Among the 350 patients receiving anti-TNF therapy (454 PY, mean follow-up 1.30 years), CDI and C. difficile positivity occurred in 9 and 25 patients, respectively. The IDs were 1.98/100 PY (95% CI: 0.91–3.77) for CDI and 5.50/100 PY (95% CI: 3.56–8.13) for C. difficile positivity (Table 2). The IDRs for UST versus anti-TNF were 0.53 (95% CI: 0.14–1.99, p = 0.347) for CDI and 0.57 (95% CI: 0.27–1.19, p = 0.137) for C. difficile positivity (Table 2).

Table 2.

The IDs and IDRs for CDI and C. difficile positivity treated with UST versus anti-TNF among patients with Crohn’s disease.

Statistic	UST (n = 277, 381 PY)		Anti-TNF (n = 350, 454 PY)
Statistic	CDI	C. difficile positivity	CDI	C. difficile positivity
No. of episodes	4	12	9	25
ID/100 PY (95% CI)	1.05 (0.29–2.69)	3.15 (1.63–5.50)	1.98 (0.91–3.77)	5.50 (3.56–8.13)
IDR (95% CI)	0.53 (0.14–1.99)	0.57 (0.27–1.19)	Reference	Reference
p	0.347	0.137	/	/

Anti-TNF, antitumor necrosis factor; C. difficile positivity, Clostridioides difficile nucleic acid positivity; CDI, C. difficile infection; CI, confidence interval; ID, incidence density; IDR, incidence density ratio; PY, person-years; UST, ustekinumab.

Risk analysis

Kaplan–Meier survival analysis

We assessed the cumulative event-free survival for CDI between the UST and anti-TNF groups using the Kaplan–Meier method. Throughout the observation period, the survival curve of the UST group consistently lay above that of the anti-TNF group, suggesting a potential trend toward a reduced risk of CDI with UST. However, the log-rank test indicated no statistically significant difference between the two groups (χ² = 1.329, p = 0.25; Figure 2(a)). Similarly, Kaplan–Meier analysis of C. difficile positivity revealed a comparable distribution in cumulative event-free survival to that of CDI, and the log-rank test also showed no significant difference (χ² = 2.913, p = 0.088; Figure 2(b)).

Figure 2.

Kaplan–Meier curve comparing risk of (a) CDI and (b) C. difficile positivity between UST and anti-TNF treated patients with Crohn’s disease. Y-axis is survival proportion of patients still on therapy, and X-axis is time point of follow-up.

MSMs–Cox analysis

The MSMs–Cox was employed to assess risk factors for CDI. In the unadjusted univariate Cox analysis, no significant difference in CDI risk was observed between UST and anti-TNF (HR = 0.51, 95% CI: 0.16–1.62, p = 0.253; Figure 3(a)). The multivariate MSM-Cox model also showed no significant difference in CDI risk between UST and anti-TNF (adjusted HR = 0.51, 95% CI: 0.16–1.17, p = 0.274; Figure 3(a)). After adjusting for time-dependent concomitant medication, the results remained largely consistent with the primary analysis, yielding an HR of 0.49 (95% CI: 0.15–1.62, p = 0.243; Figure 3(a)). Similarly, an analysis of risk factors for C. difficile positivity was conducted. Univariate analysis indicated no significant difference in C. difficile positivity risk between UST and anti-TNF (HR = 0.55, 95% CI: 0.28–1.10, p = 0.091; Figure 3(b)). The multivariate MSM-Cox model also revealed no significant difference between UST and anti-TNF (adjusted HR = 0.59, 95% CI: 0.29–1.21, p = 0.151; Figure 3(b)). After adjustment for concomitant medications as time-varying covariates, the HR for C. difficile positivity remained largely unchanged at 0.56 (95% CI: 0.27–1.13, p = 0.104; Figure 3(b)). The similarity in the point estimates between the two models suggests that concomitant medication may not be a primary mechanism through which the treatment group influences outcomes in our study population.

Figure 3.

Forest plot showing the HRs (95% CIs) for (a) CDI and (b) C. difficile positivity comparing UST to anti-TNF in the overall Crohn’s disease cohort from the MSM-Cox models.

A prespecified subgroup and sensitivity analysis

Overall, there was no statistically significant heterogeneity in the risks of CDI and C. difficile positivity across subgroups (Table 3). The HR for CDI associated with UST versus anti-TNF was 0.23 (95% CI: 0.05–1.11, p = 0.068) and 3.20 (95% CI: 0.29–34.89, p = 0.340) in males and females, respectively. In patients 40 years or older, the HR was 0.74 (95% CI: 0.05–11.22, p = 0.827). In patients under 40 years old, the HR was 0.48 (95% CI: 0.13–1.82, p = 0.282; Table 3). Similarly, the HR for C. difficile positivity associated with UST versus anti-TNF was 0.42 (95% CI: 0.18–1.03, p = 0.057) and 1.00 (95% CI: 0.29–3.38, p = 0.994) in males and females. In patients 40 years or older, the HR was 0.74 (95% CI: 0.05–11.22, p = 0.827). In patients under 40 years old, the HR was 0.59 (95% CI: 0.28–1.23, p = 0.160; Table 3).

Table 3.

HRs for CDI and C. difficile positivity associated with UST compared with anti-TNF in patients with Crohn’s disease stratified by gender and age.

Subgroup	CDI			C. difficile positivity
Subgroup	HR	95% Cl	p	HR	95% Cl	p
Male	0.23	0.05–1.11	0.068	0.42	0.18–1.03	0.057
Female	3.20	0.30–34.89	0.340	1.00	0.29–3.38	0.994
⩾40	0.74	0.05–11.22	0.827	0.74	0.05–11.22	0.827
<40	0.48	0.13–1.82	0.282	0.59	0.28–1.23	0.160

In a prespecified exploratory sensitivity analysis restricted to biologic-naïve patients, the unadjusted univariate Cox analysis showed that there was no significant difference in CDI risk between UST and anti-TNF (HR = 0.36, 95% CI: 0.08–1.69, p = 0.196; Figure 4(a)). The adjusted HR for CDI comparing UST to anti-TNF was 0.27 (95% CI: 0.05–1.30, p = 0.102; Figure 4(a)), which was consistent in direction with the primary analysis (HR = 0.51). Similarly, an analysis of risk factors for C. difficile positivity was conducted. The unadjusted univariate analysis indicated no significant difference in C. difficile positivity risk between UST and anti-TNF (HR = 0.41, 95% CI: 0.16–1.03, p = 0.058; Figure 4(b)). The adjusted HR was 0.42 (95% CI: 0.16–1.09, p = 0.075), also aligned in direction with the main result (HR = 0.59) and supporting the robustness of the conclusions (Figure 4(b)). After adjustment for concomitant medications as time-dependent covariates, the HRs for CDI and C. difficile positivity remained largely unchanged at 0.28 (95% CI: 0.06–1.37, p = 0.117; Figure 4(a)) and 0.44 (95% CI: 0.17–1.12, p = 0.085), respectively (Figure 4(b)).

Figure 4.

Forest plot showing the HRs (95% CIs) for (a) CDI and (b) C. difficile positivity comparing UST to anti-TNF in the biologic-naïve cohort from the MSM-Cox models.

Discussion

In our cohort study, 350 CD patients treated with anti-TNF and 277 treated with UST were followed for a total of 835 PY. Statistical methods such as IPTW and marginal structural models were applied to minimize confounding bias in this observational study. The key finding was that, compared with anti-TNF agents, UST demonstrated lower IDs of CDI (UST: 1.05/100 PY vs anti-TNF: 1.98/100 PY) and C. difficile positivity (UST: 3.15/100 PY vs anti-TNF: 5.50/100 PY) in CD patients. However, these differences did not reach statistical significance in the adjusted risk analysis (CDI: HR = 0.51, 95% CI: 0.16–1.17, p = 0.274, C. difficile positivity: HR = 0.59, 95% CI: 0.29–1.21, p = 0.151).

The ID of CDI in the UST group observed in our study (1.05/100 PY) was highly consistent with the rate reported in the UNITI/IM-UNITI phase III clinical trials (1.08/100 PY),^7,8 which strengthens the credibility of our findings. A recent study reported a CDI incidence rate of 2.48/100 PY in a UST-treated IBD group; however, it did not distinguish between CD and ulcerative colitis, nor did it clarify the C. difficile nucleic acid status of patients prior to enrollment, thus limiting its direct comparability with our results.²² To our knowledge, this study is the first to provide a head-to-head comparison of CDI risk between UST and anti-TNF agents in CD patients and offers new risk-benefit data to inform clinical decision-making for biologic-naïve patients.

Although the primary analysis did not reach statistical significance, the point estimates for both the overall cohort and the biologic-naïve subgroup consistently favored UST. Crucially, the limited number of CDI events resulted in wide CIs and likely rendered the study underpowered to detect a modest but clinically relevant difference in risk. Therefore, our nonsignificant results must not be interpreted as evidence of equivalence between UST and anti-TNF agents in terms of CDI risk, nor should the observed trends be construed as evidence of superior safety for UST. The directionally consistent signal, particularly in biologic-naïve patients, is best considered hypothesis-generating and justifies further investigation in larger, prospective studies. This exploratory finding may be relevant for patients with high concern for infections, where all safety data require consideration. Furthermore, after adjusting for time-dependent concomitant medication, the effect estimate changed minimally (HR from 0.51 to 0.49), indicating that the observed association is unlikely to be mediated by differences in concomitant drug use.

We acknowledge several important limitations. First, as an observational study, residual confounding cannot be entirely ruled out despite the application of advanced statistical methods such as IPTW and MSMs. Second, and most critically, the limited number of CDI events constrained the statistical power, increasing the risk of a Type II error. Third, the single-center design, combined with higher CDAI in the anti-TNF group, may have introduced surveillance bias due to more frequent healthcare encounters, potentially inflating the CDI detection rate in this group. These limitations underscore that our findings are preliminary and necessitate validation in larger, multicenter, prospective studies with standardized surveillance protocols.

Conclusion

In summary, in this comparative study, UST was not associated with a significantly higher risk of CDI than anti-TNF therapy in patients with CD. Any potential risk difference, including a possible reduction with UST suggested by point estimates, remains uncertain due to the limited number of events and wide CIs. Therefore, larger, multicenter, prospective studies with standardized infection-surveillance protocols are warranted to confirm this potential safety advantage and to provide higher-level evidence for clinical practice.

Footnotes

Acknowledgements

We thank Lishuo Shi for his assistance in the statistical analyses.

Declarations

ORCID iDs

Qi Zhang

Min Zhi

Tao Liu

References

Surawicz

Brandt

Binion

, et al. Guidelines for diagnosis, treatment, and prevention of Clostridium difficile infections. Am J Gastroenterol 2013; 108(4): 478–498; quiz 99.

Paudel

Nair

DVT

Joseph

, et al. Gastrointestinal microbiota-directed nutritional and therapeutic interventions for inflammatory bowel disease: opportunities and challenges. Gastroenterol Rep 2024; 12: goae033.

Sands

Peyrin-Biroulet

Loftus

, et al. Vedolizumab versus adalimumab for moderate-to-severe ulcerative colitis. N Engl J Med 2019; 381(13): 1215–1226.

Balram

Battat

Al-Khoury

, et al. Risk factors associated with Clostridium difficile infection in inflammatory bowel disease: a systematic review and meta-analysis. J Crohns Colitis 2018; 13(1): 27–38.

Dalal

Mitri

Goodrick

, et al. Risk of gastrointestinal infections after initiating vedolizumab and anti-TNFα agents for ulcerative colitis. J Clin Gastroenterol 2023; 57(7): 714–720.

Bohm

Zhang

, et al. Comparative safety and effectiveness of vedolizumab to tumour necrosis factor antagonist therapy for Crohn’s disease. Aliment Pharmacol Ther 2020; 52(4): 669–681.

Feagan

Sandborn

Gasink

, et al. Ustekinumab as induction and maintenance therapy for Crohn’s disease. N Engl J Med 2016; 375(20): 1946–1960.

Sandborn

Feagan

Danese

, et al. Safety of ustekinumab in inflammatory bowel disease: pooled safety analysis of results from phase 2/3 studies. Inflamm Bowel Dis 2021; 27(7): 994–1007.

Von Elm

Altman

Egger

, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008; 61(4): 344–349.

10.

Inflammatory Bowel Disease Group, Chinese Society of Gastroenterology, Chinese Medical Association. Chinese consensus on diagnosis and treatment in inflammatory bowel disease (2018, Beijing). J Dig Dis 2021; 22(6): 298–317.

11.

Gomollón

Dignass

Annese

, et al. 3rd European evidence-based consensus on the diagnosis and management of Crohn’s disease 2016: Part 1: Diagnosis and medical management. J Crohns Colitis 2017; 11(1): 3–25.

12.

Karlqvist

Sachs

Eriksson

, et al. Comparative risk of serious infection with vedolizumab vs anti-tumor necrosis factor in inflammatory bowel disease: results from nationwide Swedish registers. Am J Gastroenterol 2024; 119(12): 2480–2492.

13.

Singh

Heien

Herrin

, et al. Comparative risk of serious infections with tumor necrosis factor α antagonists vs vedolizumab in patients with inflammatory bowel diseases. Clin Gastroenterol Hepatol 2022; 20(2): e74–e88.

14.

McDonald

Gerding

Johnson

, et al. Clinical practice guidelines for Clostridium difficile infection in adults and children: 2017 update by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA). Clin Infect Dis 2018; 66(7): e1–e48.

15.

Robins

A new approach to causal inference in mortality studies with a sustained exposure period—application to control of the healthy worker survivor effect. Math Model 1986; 7(9): 1393–1512.

16.

Robins

Hernán

Brumback

Marginal structural models and causal inference in epidemiology. Epidemiology (Cambridge, MA) 2000; 11(5): 550–560.

17.

Rosenbaum

PR.

Model-based direct adjustment. J Am Stat Assoc 1987; 82(398): 387–394.

18.

Austin

PC.

An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46(3): 399–424.

19.

Austin

Stuart

EA.

Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med 2015; 34(28): 3661–3679.

20.

Lenti

Dolby

Clark

, et al. A propensity score-matched, real-world comparison of ustekinumab vs vedolizumab as a second-line treatment for Crohn’s disease. The Cross Pennine study II. Aliment Pharmacol Ther 2022; 55(7): 856–866.

21.

Yang

Huang

, et al. Comparative effectiveness of ustekinumab vs. Vedolizumab for anti-TNF-naïve or anti-TNF-exposed Crohn’s disease: a multicenter cohort study. EClinicalMedicine 2023; 66: 102337.

22.

Martínez-Lozano

Saralegui-Gonzalez

Reigadas

, et al. Risk factors for Clostridioides difficile infection among patients diagnosed with inflammatory intestinal and rheumatological diseases in the biologic era. BMC Gastroenterol 2025; 25(1): 70.