Clostridioides difficile infection in inflammatory bowel disease: an observational study in flares of ulcerative colitis and a South Asian region-specific systematic review and meta-analysis

Abstract

Background:

Clostridioides difficile infection (CDI) is well-recognised as a cause of flare in inflammatory bowel disease (IBD).

Objectives:

To prospectively evaluate CDI as a cause of flare in moderate-to-severe ulcerative colitis (UC) and perform a region-specific systematic review to evaluate the role of CDI in IBD patients in South Asia.

Design:

A single-centre prospective observational study followed by a region-specific systematic review.

Data sources and methods:

The observational study was conducted between December 2024 and December 2025 and included patients with moderate-to-severe UC flares. Triple testing using Stool glutamate dehydrogenase (GDH), an enzyme immunoassay (EIA) for toxin A/B, and polymerase chain reaction (PCR) testing was performed. A literature search in PubMed, Embase, and Scopus was conducted on 5th December 2025 to identify relevant studies from South Asia. Information on the study population (IBD type, age, gender, disease activity), the testing method, and the outcomes of CDI testing were extracted. The pooled prevalence of CDI was estimated using a random-effects model. The risk of Bias was assessed using Joanna Briggs’ tool.

Results:

Of the 101 patients with active UC, 59 had acute severe UC. Six patients tested positive for GDH, whereas only one tested positive for EIA and PCR. A total of 13 studies reporting on 1267 patients were included in the systematic review. Pooled prevalence of CDI was 0.06 (0.03–0.11, I² = 80.8%). Subgroup analyses were performed by testing method, study type, and region. However, there was persistent statistical heterogeneity. Funnel plot and Egger’s test suggested the presence of publication bias.

Conclusion:

Our observational study shows a low prevalence of CDI as a cause of UC flares. The findings of the systematic review suggest high variability in the CDI positivity across South Asia. These differences were not fully explained by the method of testing, the type of study, or the geographic location of the study.

Plain language summary

Clostridioides difficile infection in South Asian IBD

People with Inflammatory Bowel Disease (IBD) often experience worsening of symptoms, or “flares,” which can sometimes be caused by the bacteria Clostridioides difficile. This study investigated the frequency of this infection among patients with ulcerative colitis in India. We observed 101 patients using highly accurate triple-testing and found the infection was uncommon (only 1%). In addition, a review of previous studies in the region suggested a higher rate (7%), possibly due to inconsistent testing methods used in the past. The study concludes that while the infection is more common in severe cases, standardized testing is crucial to avoid misdiagnosis

Introduction

Clostridioides difficile, a gram-positive, spore-forming, anaerobic bacillus, can cause a range of gastrointestinal symptoms, from mild diarrhoea to fulminant colitis. Among several factors that are known to be associated with a high risk of C. difficile infection (CDI), inflammatory bowel disease (IBD), especially ulcerative colitis (UC), is an important one. Compared to the general population, UC patients are at a significantly higher risk of CDI (about eight times). ¹ These patients often present with flares of underlying IBD that are attributed to CDI. Over the last few decades, the incidence of CDI has been increasing (3.9% in 2004 from 2.4% in 1998) among UC patients in the United States. Similarly, increasing rates are also noted among hospitalised IBD patients compared to non-IBD hospitalised patients. ¹ Several studies have confirmed that CDI is associated with adverse IBD-related outcomes, including prolonged hospital stay, failure of IBD therapy, need for surgical intervention, and mortality.^{2 –5} Despite this, the diagnosis of CDI in the setting of IBD is challenging. This is due to an atypical clinical presentation, differences in sensitivity and specificity among available methods for detecting C. difficile (enzyme immunoassay (EIA), nucleic acid amplification test (NAAT), culture), and difficulty in distinguishing symptomless carriage of the organism from the toxigenic causative strain. EIA for C. difficile toxin (A and B) and glutamate dehydrogenase (GDH) are commonly used in couples to alleviate this dilemma. GDH is used as an initial test due to its higher sensitivity. In case of discrepancies, NAAT is done as a confirmatory test. Culture of toxigenic strains and next-generation sequencing (NGS) are not routinely used.

Individuals of South Asian origin have been reported at a higher risk of CDI compared to the White population (19% vs 9%, p < 0.05). ⁶ The data from South Asia, a new frontier of IBD, with respect to CDI as a cause of flare, is rather limited. Therefore, we conducted an observational study to evaluate the role of CDI in flares of ulcerative colitis at a tertiary care centre. In addition, we conducted a region-specific systematic review and meta-analysis to examine the role of CDI in flares of IBD in this region.

Methods

Observational study

Patients and setting

A prospective observational study was conducted from December 2024 to December 2025 in a large tertiary care referral centre in North India (Chandigarh). The study population consisted of adult patients with an established diagnosis of UC who presented with moderate-to-severe disease activity. We included the patients with a diagnosis of ulcerative colitis who presented with moderate-to-severe flare of disease as defined by the Complete Mayo Score. ⁷ The diagnosis of acute severe ulcerative colitis (ASUC) was based on modified Trulove Witts criteria. ⁸ We excluded patients with age <18 years, an uncertain diagnosis, a diagnosis of Crohn’s disease, indeterminate colitis, or other non-UC colitides, or if stool samples for CDI testing could not be obtained or processed as per protocol (e.g., logistical constraints, inadequate sample, or sample not submitted).

Diagnosis of CDI

CDI was defined based on laboratory confirmation of toxigenic C. difficile in stool in the context of a compatible clinical presentation (diarrhoea and/or worsening of UC symptoms). All patients underwent three tests – GDH antigen using a commercial EIA, EIA for toxin A/B and –polymerase chain reaction (PCR) for C. difficile or toxin genes. The methodological details for these tests are provided in the Supplemental File. Only patients with evidence of a positive GDH with confirmatory toxin or PCR positivity were classified as having CDI. This definition was used because positive GDH is non-specific and may be present in organisms other than C. difficile.

Systematic review and meta-analysis

Search strategy and selection

We conducted the systematic review in accordance with the PRISMA extension for systematic reviews and meta-analyses. ⁹ The review was not registered in PROSPERO, although this would have been desirable. The search was performed on 5th December 2025 from three electronic biomedical databases (PubMed, Scopus and Embase) to search available literature on CDI in IBD in South Asia. We used a pre-developed search strategy, including relevant index terms and keywords such as ‘Inflammatory Bowel Disease’, ‘IBD’, ‘Ulcerative colitis’, ‘Crohn’s disease’, and ‘Clostridioides difficile’ to capture the underlying condition, combined with country terms, to identify studies from South Asia. A detailed search strategy has been provided in Supplemental Table 3. To ensure completeness of the search, we also searched the reference lists of all eligible studies.

Inclusion and exclusion criteria

We included all studies (including abstracts) reporting CDI in patients with inflammatory bowel disease (UC, CD or both) from South Asia. We included patients regardless of IBD subtype or disease severity. We included both full papers and abstracts if they provided the relevant information. We excluded mechanistic studies, animal studies, editorials, comments, narrative and systematic reviews, guidelines, and meta-analyses. We also excluded case reports or series reporting on <10 patients. In addition, studies reporting on CDI outside the IBD setting or where data on the IBD population could not be extracted were excluded. Each of these steps, including study screening, selection, and data extraction, was completed independently by two reviewers (DB, SR, AC), and any discrepancies were resolved through discussion with another reviewer (VS).

Data extraction

We extracted data from studies evaluating CDI in patients with inflammatory bowel disease that reported the proportion of positivity, regardless of the diagnostic test used. We extracted region-specific data from both prospective and retrospective studies. We also extracted data about the study location, the study period, and the year of publication. We extracted details of IBD patients, including numbers, gender, and severity of underlying disease. All patients/studies who did not meet the criteria for IBD were excluded from the analysis. We investigated the details of the underlying IBD, including type (UC/CD), extent, and severity of the disease. The definition of disease severity used in each study was also extracted. In addition, details of the method(s) used for CDI diagnosis were also extracted.

Risk of bias

The risk of bias assessment was conducted using the Joanna Briggs Institute Case Series Tool. ¹⁰ The choice of the tool was made for multiple reasons: JBI tools are flexible and not limited by strict scoring, and since most studies were patient series reporting CDI infections, we used the tool for case series. Publication bias was assessed using Egger’s test, and a funnel plot was made for further evaluation. Potential small-study effects and funnel plot asymmetry were evaluated using a linear regression test.

Data synthesis

Studies were organised using a web-based systematic review management platform, Rayyan (Rayyan Systems Inc., Boston, USA), which automatically removes duplicates. ¹¹ Statistical analysis was performed in R (Version 4.5.1; R Core Team, 2025) using the metafor and meta packages.^12,13 Pooled estimates were obtained using a random-intercept logistic regression model, accounting for between-study heterogeneity. Between-study variance (τ²) was estimated using the maximum-likelihood estimator.

Heterogeneity assessment

Statistical heterogeneity was assessed using Cochran’s Q statistic, and the magnitude of heterogeneity was quantified using the I² statistic, calculated from Q and interpreted according to conventional thresholds. An I² value> 50% was considered significant heterogeneity and was further evaluated using subgroup analysis. Subgroups were analysed for CDI based on disease severity, testing method, and study design. A Baujat plot was generated to identify studies that contributed most to heterogeneity. Subsequently, a leave-one-out analysis was performed to identify potential heterogeneity arising from a single study.

Results

Observational study

Population characteristics

We screened 152 patients with UC flare for our study, out of which 51 were excluded. Of 51 patients, CD was diagnosed in 20; the severity of UC was mild in 22; the stool sample was not suitable for sampling in 4; and five patients did not give consent. Among the 101 patients included, 49 were females, and the most common age group of presentation was less than 30 years (28.7%, n = 29). Proportions of E3, E2 and E1 disease were 57.4% (n = 58), 41.6% (n = 42) and 1% (n = 1). Of the 101 patients, 59 had ASUC, while the remaining 42 had moderate disease activity.

Results of testing for CDI

Finally, 101 UC patients with moderate or severe disease underwent testing for CDI with the index test (GDH plus toxin EIA and PCR). Out of all samples tested, 6 out of 101 (5.9%) samples were positive for GDH. Among these, one sample (1%) was additionally positive for CDTA (both toxin assay and PCR). This patient was hospitalised for ASUC and needed second-line therapy in view of failure of intravenous steroids in addition to oral vancomycin for CDI. The patient responded to cyclosporine and was eventually discharged. No equivocal toxin A/B results were observed in the study population. Additional results are provided in the Supplemental File.

Systematic review and meta-analysis

Study selection

A total of 265 citations were identified, and after removing duplicates, 220 were screened for titles and abstracts. In addition to these studies, two others were included (one identified through citation tracking and the current observational study). Two reviewers performed the screening independently (DB, AC). After initial screening, 202 citations were excluded, and 18 manuscripts underwent full-text screening. A third researcher (VS) resolved the disagreements after reaching a consensus with the other reviewers. Eventually, a total of 13 studies were included in the systematic review, and 7 studies were excluded for various reasons (Supplemental Table 3). Figure 1 depicts the flow of study selection and screening (PRISMA flow chart).

Figure 1.

PRISMA flow diagram of study selection and screening.

Characteristics of the included studies

A total of 13 studies were included for our meta-analysis.^{6,14 –24} Most of them (n = 12) were from India, and one study was from Bangladesh. Ten studies were prospective, two were retrospective, and the design of one was unclear from the available abstract. Five studies included patients with only active UC; four included patients with ASUC; three included all UC patients, irrespective of severity; and one included all IBD (both UC and CD), irrespective of severity. Table 1 summarises the key characteristics of included studies.^{6,14 –24}

Table 1.

Characteristics of the studies included in the systematic review.

S. no.	Study	Centre	Study period	Design	Study population	N_IBD (UC, CD numbers if available)	UC_Flare	UC_Remission	ASUC	ActiveUC	Population characteristics (mean age, gender)	Disease characteristics
1	Kochhar R 1993	Chandigarh	NA	Prospective	All ulcerative colitis	50	25	25	NA	25	Mean Age 35.7 ± 15.1, Gender: M 24 (48%)	Duration: 4.5 (1.5–11.4), E3: 13 (26%), E2: 14 (28%), E1: 23 (46%)
2	Vaishnavi C 2003	Chandigarh	NA	Prospective	All ulcerative colitis	94	NA	NA	NA	NA	Age: 17–72 years, M:52 (55.3%)	NA
3	Balamurugan R 2008	Vellore	Oct 2004–July 2005	Prospective	All ulcerative colitis	37	29	8	5	29	Age: 41.3 + 12.7 years, M: 26 (70.27%), Urban: 29 (78.4%)	Duration: Mean 4 years, E3: 16 (43.2 %), E2: 14 (37.8%), E1: 7 (19%), quiescent: 8 (21.6%), mild: 13 (35.1%), moderate: 11 (29.7%), severe in 5 (13.5%)
4	Iyer VH 2013	Vellore	Jan 2009–Jan 2011	Prospective	Active ulcerative colitis	87	87	0	39	48	Age: 35.2 (12.6) years, M: 51 (58.62%)	Duration: NA E3: 50 (57.4%), E2: 21 (24.1%), E3: 16 (18.3%), Severe disease, 39 (44.8%); Moderately severe disease, 11 (12.6%); Mildly active disease, 37 (42.6%)
5	Sethi SS 2018	Delhi	2017–2018	Unclear (Abstract)	ASUC	43	43	0	43	43	NA	NA
6	Mundhra S 2023	Delhi	May 2016–December 2021	Retrospective	ASUC	153	153	0	153	153	Age: 34.92 ± 12.24, M: 86 (56.2%)	Duration: 36 (IQR 16–55.5), E3-67.3%, E2-32.7%
7	Ghosh CK 2024	Dhaka	NA	Prospective	Active ulcerative colitis	75	75	0	NA	75	NA	Duration: 6–18 months in 40%, E3: 65.3%, E2: 29.4%, E1: 5.3%
8	Ghoshal U 2024	Lucknow	November 2017–September 2019	Prospective	All IBD	160	NA	NA	NA	NA	Age: 37.19 ± 16, M: 88 (55%)	UC: 136 (85%), CD: 24 (15%)
9	Singh AK 2025	Chandigarh	Jan 2022–Jul 2024	Retrospective	ASUC	117	117	0	117	117	Age: 35 years, M: 67 (57.26)%	Duration: Median 24 months (range 3–80), E2-37.6%, E3: 62.4%
10	Singh A 2025	Ludhiana	Jan 2020–June 2024	Prospective	ASUC	145					Age: 36 years IQR 26–48.5 years, M: 82 (56.56%)	Duration: 1 year (IQR, 0.5–3 years), E2 (n = 77, 53.10%), E3 (n = 52, 35.86%), E1 (n = 16, 11.03%)
11	Kutar M 2025	Mumbai	May 1, 2021–October 31, 2022	Prospective	Active IBD	65	NA	NA	NA	NA	34 (52%)	58 in UC and 7 in CD; E1: 3 (5.2%); E2: 19 (32.8%); E3: 36 (62%), L3: 5 (71.5%), L2: 2 (28.5%), B2 disease, 1 (14.3%); B1 disease, 6 (85.7%)
12	Panchal H 2026	Gurugram	May 2023–May 2024	Prospective	Active ulcerative colitis	140	140	0	69	140	Age: 41.7 years (95% CI 41.0–42.3), M: 77 (55%)	Duration: 30 months (Q1: 16.3 months, Q3: 32.25 months; IQR:16 months)., E3: 60.7%, E2: 39.2%
13	Bansal D 2026	Chandigarh	July 2024–December 2025	Prospective	Active ulcerative colitis	101	101	0	59	101	Age: 39.7 ± 12.5; M: 52 (51.5%)	Duration: 24 months, E1: 1 (1%); E2: 42 (41.6%); E3: 58 (57.4%)

ASUC, acute severe UC; CD, Clostridioides difficile; IBD, inflammatory bowel disease; UC, ulcerative colitis.

CDI prevalence in IBD

A total of 13 studies, including 1267 patients, reported the prevalence of CDI in IBD. Overall, the prevalence of CDI in IBD was 0.06 (95% CI 0.03–0.11), with I² = 80.8% (Figure 2). Due to significant statistical heterogeneity, we performed multiple subgroup analyses

Figure 2.

Overall prevalence of CDI.

CDI, Clostridioides difficile infection.

Subgroup analysis based on the method of testing

Six studies used the EIA method, three used dual testing (GDH and EIA), two used non-standard methods, and two used triple testing (GDH, EIA for toxins, and PCR). On sub-group analysis based on methodology, the prevalence of CDI was 3% (95% CI 0.01–0.07), 5% (95% CI 0.02–0.14), 9% (95% CI 0.04–0.21), and 13% (95% CI 0.09–0.20) in the triple testing, EIA, dual testing, and non-standard methods, respectively (Figure 3).

Figure 3.

Subgroup analysis of prevalence of CDI based on study method.

CDI, Clostridioides difficile infection.

Subgroup analysis based on study design

Among nine prospective studies, the overall prevalence was detected 6% (95% CI 0.03–0.11). Among the retrospective studies, prevalence was 4% (95% CI 0.00–0.43). However, heterogeneity remained high (I² = 73% for prospective and 92% for retrospective studies; Supplemental Figure 3).

CDI prevalence and disease severity

Three studies provided information on the prevalence of CDI in relation to disease severity. Of the 167 participants with ASUC, eight had CDI. Whereas none of the 161 patients with moderate UC had CDI (pooled OR 6.16, 95% CI 1.06–35.76, I² = 0%; Figure 4).

Figure 4.

Subgroup analysis of CDI according to disease severity.

CDI, Clostridioides difficile infection.

Sensitivity analysis

We noted significant heterogeneity in the meta-analysis, even in the sub-groups. To further investigate this, we did a Baujat plot to identify studies of critical importance (Supplemental Figure 4). The Baujat plot shows two such studies (Singh et al. and Mundra et al.) that contributed most to the heterogeneity. Consequently, we conducted a Leave-One-Out analysis by systematically omitting each study. Omitting the study by Singh et al, the maximum reduction in the heterogeneity was noted. However, heterogeneity remained high (I² = 76.5%). Details of Leave-one-Out analysis are provided in Supplemental Table 4.

Risk of bias

Publication bias was assessed using Egger’s test and represented in the funnel plot (Figure 5(a)). Egger’s test suggested the presence of publication bias with a T-statistic of −4.6 (p = 0.008). A trimfill analysis was performed, which added five hypothetical studies and provided a revised estimate of 0.13 (0.07–0.24) with an I² of 84.7% (Figure 5(b)). The asymmetric funnel plot can, however, also be due to high heterogeneity in the included studies.

Figure 5.

(a) Funnel plot suggestive of publication bias and (b) Trimfill plot with hypothetical studies.

The risk of bias assessment, depicted in Supplemental Table 5, shows a high risk of bias in 4 of 13 studies and a low risk of bias in seven studies.

Discussion

In this work, we present data from our observational study and South Asian region-specific systematic review on the frequency of CDI in IBD flares. The observational study suggested that C. difficile was an infrequent cause of flares of ulcerative colitis in our setting. Although compromised by the relatively low numbers and single-centre nature of the study, which limit generalisability, the study used robust methodology for testing (including all three diagnostic tests, i.e. GDH, Toxin assay and PCR), to ensure no cases were missed. However, because only one patient tested positive, we could not evaluate any clinically meaningful association of C. difficile infection with clinical presentation or outcomes. In the systematic review and meta-analysis, we evaluated the prevalence of C. difficile infection among IBD patients from the South Asia region. The overall prevalence of CDI in this region is 7%, which is consistent with previous studies and Western literature. We also observed six times the odds of infection in severe colitis compared to moderately severe colitis, although these findings should be interpreted cautiously, given the low overall numbers and remain exploratory. However, these findings do underscore the role of CDI in severe disease flares and re-emphasise the importance of testing for CDI.

Diagnostic tests for C. difficile have variable sensitivity and specificity. Commonly used tests include EIA for the detection of GDH, toxin A, toxin B, and NAATs. Other methods, like culture of toxigenic strains, and cell cytotoxicity neutralisation assays (CCNA), are not commonly used. GDH and NAAT have similarly high sensitivity, but GDH is relatively less specific. EIAs for toxin A and B have higher specificity for toxigenic strains; their sensitivity is less compared to GDH. It is important to distinguish colonisation with C. difficile from C. difficile infection in the setting of IBD flares. Unfortunately, the presence of pseudo-membranes is infrequent in the setting of IBD. While the positivity of the toxin assay is more likely to suggest CDI, NAATs may be extremely sensitive and may pick up colonisation. Therefore, international guidelines recommend a two-step algorithm that includes a sensitive assay (GDH) and a specific assay (EIA for toxin A/B) to obviate overdiagnosis and unnecessary treatment.^{25 –27} This relatively recent standardisation of the detection method led to a variable estimation of the prevalence of C. difficile infection in the setting of IBD in our systematic review, as different studies used different methods.

Most of the studies in the systematic review used EIA as the sole diagnostic method. We observed that five studies were published before these guidelines and only four studies used the dual testing (for GDH and EIA for toxin A/B). The prevalence estimates in these studies varied widely, ranging from 1% to 22%. However, prevalence varied from 1% to 19% with the dual testing as well (1%, 3%, 13% and 19% in Bansal, Panchal, Singh and Sethi et al, respectively). This might be related to differences in the underlying populations, geographic regions, or the use of different kits for assessment. While non-standardised methods suggested high prevalence, it is interesting to note that the use of triple testing did not seem to increase the positivity rates, and both studies using triple testing seem to provide a relatively lower positivity rate. The data from studies utilising EIA alone is extremely heterogeneous, with prevalence ranging from 1% to 22%. This heterogeneity is difficult to explain in the wake of the expected lower sensitivity of EIA in general. We also did a region-specific subgroup analysis. However, even this did not count for heterogeneity. In our prospective study of 101 subjects with active UC, triple testing identified only one confirmed C. difficile infection using GDH and toxin/PCR assays.

We noted significant heterogeneity arising from several factors. The method of detection was non-uniform, varying from dual tests to non-standardised tests. Subgroup analysis based on study design, disease severity, and study population did not resolve the heterogeneity. However, high heterogeneity has been reported in previous systematic reviews on C. difficile infection in setting of IBD. ²⁸ Possible reasons could be variations in study population (geographic, disease severity, antibiotic use), testing methodology etc. In addition, the I² values get inflated when the prevalence is towards extremes (very high or very low), as in the present case. In addition, the pooled data should be interpreted cautiously due to potential publication bias. Another limitation is the relatively small numbers in the severity-based analysis, which should be considered as exploratory. In addition, we have combined information from prospective and retrospective studies because of various reasons: limited number of available studies, to capture real-world data, which often is retrospective. In addition, while systematic reviews of treatments can include biased data from retrospective studies due to selection bias for a particular therapy, this might be a lesser concern in observational prevalence data. Another limitation is the lack of registration of the protocol and the lack of a certainty assessment. Our analysis suggests a clear need for a multicentre study to evaluate the actual risk of CDI using standardised, homogeneous assessment. This will help determine whether the difference in CDI prevalence estimates is due to methodological issues or reflects a real difference across regions. Further, most of the included studies are from India, and the issue remains under-evaluated in other countries of the region. Our study has several important strengths. It is the first region-specific meta-analysis of the prevalence of CDI in IBD in South Asian countries, and that most of the included studies were prospective.

Conclusion

Our meta-analysis shows that the prevalence of C. difficile infection in ulcerative colitis is around 7%, with higher odds in severe flares (based on limited data). The findings from our observational study and the systematic review suggest considering C. difficile testing in severe flares. In addition, it would be prudent to use a two-step testing suggested by the guidelines to ensure a secure diagnosis. Due to significant heterogeneity and methodological inconsistencies, large studies using standardised dual testing are needed. Dual testing helps avoid overdiagnosis and should be used to detect C. difficile infection in IBD.