Investigating the relationship between Helicobacter pylori and intestinal Clostridioides difficile infection: evidence from a cross-sectional study

Abstract

Background

Helicobacter pylori (H. pylori) infection in the stomach can lead to alterations in the gut microbiota. However, the association between H. pylori infection and Clostridioides difficile infection (CDI) remains unclear.

Methods

We conducted a retrospective cross-sectional study to evaluate the association between H. pylori infection and CDI. We included patients hospitalized for diarrhea at our center between 2019 and 2024 who underwent both H. pylori and CDI testing during their hospital stay. H. pylori infection was diagnosed using either the carbon-13 urea breath test or the rapid urease test. CDI was diagnosed using a two-step testing algorithm. Patients were categorized into H. pylori-positive and -negative groups. We compared the prevalence of CDI between groups and assessed the association using logistic regression. To adjust for potential confounders, propensity score matching (PSM) was performed, followed by further analysis.

Results

We included 1,624 patients: 732 H. pylori-positive and 892 H. pylori-negative. The prevalence of CDI was significantly higher in the H. pylori-positive group (13.7%, 95% CI: 11.2%-16.2%) compared to the H. pylori-negative group (9.0%, 95% CI: 7.1%-10.8%) (P = 0.003). Univariate logistic regression showed that H. pylori infection was associated with CDI (OR: 1.606, 95% CI: 1.176–2.194; P = 0.003). Multivariate analysis also suggested that H. pylori infection was independently associated with CDI (OR: 1.817, 95% CI: 1.320–2.501; P < 0.001). This association remained significant after PSM.

Conclusions

Our findings suggest that current H. pylori infection may be associated with Clostridioides difficile infection.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-025-04227-w.

Introduction

H. pylori is a Gram-negative bacterium causing globally distributed gastric infections [1, 2]. The World Health Organization (WHO) has classified H. pylori as a Group 1 carcinogen, recognizing its established role in the development of chronic gastritis, peptic ulcer disease, and gastric carcinoma [3–5]. Recent research suggests that H. pylori affects more than just gastric diseases; it can influence the gut microbiome by modifying gastric acid secretion and disrupting intestinal microbial balance [6–8].

As a leading nosocomial pathogen, Clostridioides difficile represents a predominant cause of infectious diarrhea, with CDI being a significant consequence of antimicrobial therapy. According to Centers for Disease Control and Prevention (CDC) epidemiological reports, Clostridioides difficile causes over 450,000 infections annually in the United States. The disease is responsible for nearly 29,000 deaths each year [9]. Additionally, cases of community-acquired Clostridioides difficile infection (CA-CDI) have been increasing in recent years, particularly among healthy adults who have not received antibiotic treatment [10, 11]. The occurrence of Clostridioides difficile infection is influenced by multiple factors, including antibiotic use, exposure to the hospital environment, host immune status, and gut microbiome disruption [12–14].

Recent studies have mainly investigated how antibiotics and acid-suppressing drugs used to treat H. pylori affect the risk of subsequent CDI [15–18]. Few studies have examined the potential association between gastric H. pylori infection and subsequent intestinal Clostridioides difficile infection risk. Therefore, we conducted the following study to elucidate the potential association between Helicobacter pylori infection and CDI.

Materials and methods

Patient selection

We conducted a retrospective cross-sectional of diarrheal patients referred to the Department of Gastroenterology at Nanjing Second Hospital between 2019 and 2024. This study analyzed existing medical records to investigate the association between Helicobacter p.infection status and CDI. We applied the following exclusion criteria: (1) Hospitalized patients without concurrent testing for Clostridioides difficile infection; (2) Patients lacking testing for Helicobacter pylori; (3) Patients aged < 18 years. We extracted the following variables from patients’ medical records: age (< 60 and ≥ 60 years), sex (male and female), marital status (married and unmarried), smoking, alcohol consumption, type of medical insurance (Resident Medical Insurance and Employee Medical Insurance), place of residence (urban and rural), length of hospital stay (< 7 days and ≥ 7 days), patient source (initially presenting and referred patients), history of H. pylori infection, history of CDI, comorbidities (inflammatory bowel disease [IBD], liver cirrhosis, malignancies, diabetes, cardiovascular diseases, malnutrition, peptic ulcer, and atrophic gastritis), prior medication use (antibiotics, proton pump inhibitors [PPIs], and probiotics), H. pylori detection methods (carbon-13 urea breath test and endoscopic biopsy), and CDI.

The length of hospital stay refers to the duration of hospitalization for the current diarrhea-related episode at our center. Histories of H. pylori infection and CDI were obtained from past medical records and indicate prior episodes of infection. The use of antibiotics, PPIs, and probiotics was defined as administration within the four weeks prior to admission. Missing data were uniformly coded as ‘unknown’ and included as a separate category in the statistical analyses. All data were collected during hospitalization. Data extraction was primarily conducted by two physicians (YZ and HFT), and any discrepancies were resolved through adjudication by a third physician (LLY).

Laboratory methods

H. pylori infection was diagnosed using either the carbon-13 urea breath test (Sensitivity: 96–98%, Specificity: 95–97%) or the rapid urease test (Sensitivity: 80–99%, Specificity: 92–100%) on biopsy specimens obtained during endoscopy [19, 20]. The carbon-13 urea breath test was performed using a diagnostic kit (Beijing Huaken Anbang Technology Co., Ltd.) and a carbon-13 urea breath test analyzer (Model HY-IREXA, Guangzhou Huayou Mingkang Optoelectronic Technology Co., Ltd.). The rapid urease test was performed on biopsy specimens obtained during gastroscopy using an Olympus CV-290 endoscope, and analyzed with a rapid urease test kit (Suzhou Rongshi Medical Technology Co., Ltd.). Clostridioides difficile infection was detected using a two-step method. Step 1: Glutamate dehydrogenase (GDH) (Sensitivity: 82%, Specificity: 91%) and toxin A/B enzyme immunoassays (EIA) (Sensitivity: 75%, Specificity: 95%) were performed simultaneously using a commercial assay kit (TECHLAB, USA) [21]. Interpretation of results: specimens were considered CDI-positive if both tests were positive, and CDI-negative if both were negative. In cases of discrepant results, a second-step test was conducted. Step 2: Real-time quantitative PCR (qPCR) targeting (Sensitivity: >92%, Specificity: >99%) the Clostridioides difficile toxin genes (tcdA and tcdB) was performed using a real-time fluorescence-based qPCR kit (Meizheng Bio-Tech Co., Ltd., Shandong, China) on a CFX96 Real-Time PCR Detection System (Bio-Rad, USA) [22]. If the qPCR result indicated the presence of both toxin A and B genes, the specimen was considered CDI-positive.

Grouping and outcomes

Based on diagnostic confirmation of H. pylori infection, participants were divided into two comparison groups: infected (H. pylori-positive) and non-infected (H. pylori-negative). The primary outcomes were the prevalence of CDI and the association with CDI.

Sample size calculation and power analysis

We estimated the required sample size for comparing two proportions using the Fleiss formula. To detect an absolute difference of 4.7% in CDI prevalence between the H. pylori-positive (13.7%) and H. pylori-negative (9.0%) groups, with a two-sided significance level of 5% and a statistical power of 80%, a minimum of 704 participants per group was required. As the actual sample sizes exceeded this threshold (732 in the H. pylori-positive group and 892 in the H. pylori-negative group), the study was adequately powered to detect a meaningful difference in the primary outcome. Post hoc power analysis confirmed that, despite the relatively low prevalence of CDI, the large sample size ensured sufficient statistical power (power = 0.85).

Statistical analysis

Comparative analysis of demographic and clinical characteristics between the two cohorts was performed using Pearson’s chi-square test. First, we compared the prevalence of CDI between H. pylori-positive and H. pylori-negative patients. Subsequently, univariate logistic regression was performed to identify variables associated with CDI (p < 0.05). Variables identified as associated with CDI in the univariate logistic regression analysis were subsequently included in a multivariate logistic regression model to further evaluate independent associations. For the multivariate model, the Hosmer-Lemeshow test was used to assess goodness-of-fit, with a p-value > 0.05 indicating an adequate model fit. Variance inflation factors (VIFs) were calculated to evaluate multicollinearity, with all variables showing VIFs < 5, suggesting no significant multicollinearity. Cook’s distance (Cook’s D) was employed to identify influential observations, with Cook’s D > 0.5 indicating the presence of potential outliers. Additionally, to address baseline imbalances between the two groups, PSM was applied to balance covariates, followed by further comparison of CDI prevalence and risk between the groups. Subgroup analyses were conducted to explore differences in outcomes across various subgroups. Interaction effects in the subgroup analyses were assessed using multiplicative interaction terms, with significance evaluated by the Wald test. An interaction p-value < 0.05 indicated a statistically significant difference. We also applied the Benjamini-Hochberg false discovery rate (FDR) correction to the p-values from each subgroup analysis. An FDR-adjusted p-value < 0.05 was considered statistically significant. To evaluate the potential impact of unmeasured confounding, we conducted a series of worst-case scenario sensitivity analyses. The adjusted ORs in the sensitivity analyses were calculated using the Cornfield approximation method. Chi-square tests, logistic regression analyses, the Hosmer-Lemeshow goodness-of-fit test, variance inflation factor calculations, Cook’s distance calculations, and interaction analyses were all performed using SPSS version 25.0. Benjamini-Hochberg FDR correction, and propensity score matching were conducted using R version 4.1.2.

Results

Patient selection and clinical characteristics

We initially identified 6,872 hospitalized patients with diarrhea. After multiple rounds of screening, the final cohort comprised 1,624 eligible participants, including 732 H. pylori-positive and 892 H. pylori-negative individuals. The detailed screening process is illustrated in Fig. 1.

Fig. 1.

Flowchart of patient selection

Comparative analysis showed no significant differences between the two groups in terms of age, sex, marital status, smoking, alcohol consumption, medical insurance, place of residence, duration of Hospitalization, patient source, H. pylori detection method, history of H. pylori infection, history of CDI, comorbidities (including IBD, liver cirrhosis, malignancy, diabetes, and cardiovascular diseases), prior antibiotic use, and PPI use.(Table 1) However, H. pylori-positive patients exhibited higher rates of malnutrition, peptic ulcer, atrophic gastritis, and prior probiotic use compared to H. pylori-negative patients (Table 1).

Table 1.

Patients characteristics between the H. pylori-positive and the H. pylori-negative group

Variables	H. pylori-negative	H. pylori-positive	P
	n = 892	n = 732
Sex, %
Male	547(61.3%)	480 (65.6%)	0.077
Female	345(38.7%)	252(34.4%)
Age, %
< 60	637(71.4%)	549(75.0%)	0.105
≥ 60	255(28.6%)	183(25.0%)
Marital status, %
Unmarried	321(36.0%)	284(38.8%)	0.244
Married	571(64.0%)	448(61.2%)
Smoking, %
No	647(72.5%)	535(73.1%)	0.803
Yes	245(27.5%)	197(26.9%)
Drinking, %
No	685(76.8%)	554(75.7%)	0.601
Yes	207(23.2%)	178(24.3%)
Medical Insurance, %
Resident Medical Insurance	553(62.0%)	459(62.7%)	0.769
Employee Medical Insurance	339(38.0%)	273(37.3%)
Place of Residence, %
Urban Residents	495(55.5%)	394(53.8%)	0.502
Rural Residents	397(44.5%)	338(46.2%)
Duration of Hospitalization, %
< 7 days	682(76.5%)	529(72.3%)	0.054
≥ 7 days	210(23.5%)	203(27.7%)
Patient source, %
Initially presenting patients	744(84.3%)	602(82.2%)	0.534
Referred patients	148(16.7%)	130(17.8%)
Method of H. pylori diagnosis, %
Rapid urease test	557(62.4%)	437(59.7%)	0.362
Urea Breath Test	335(37.6%)	295(40.3%)
Previous H. pylori infection, %
No	486(54.5%)	393(53.7%)	0.816
Yes	258(28.9%)	222(30.3%)
Unknown	148(16.6%)	117(16.0%)
Previous Clostridioides difficile infection, %
No	754(84.5%)	655(89.5%)	0.051
Yes	71(8.0%)	36(4.9%)
Unknown	67(7.5%)	41(5.6%)
Clostridioides difficile infection, %
No	812(91.0%)	632(86.3%)	0.003
Yes	80(9.0%)	100(13.7%)
Inflammatory bowel disease, %
No	608(68.2%)	509(69.5%)	0.552
Yes	284(31.8%)	223(30.5%)
Liver cirrhosis, %
No	761(85.3%)	601(82.1%)	0.08
Yes	131(14.7%)	131(17.9%)
Malignant tumors, %
No	754(84.5%)	626(85.5%)	0.579
Yes	138(15.5%)	106(14.5%)
Diabetes, %
No	778(87.2%)	645(88.1%)	0.586
Yes	114(12.8%)	87(11.9%)
Cardiovascular disease, %
No	690(77.4%)	559(76.4%)	0.638
Yes	202(22.6%)	173(23.6%)
Malnutrition, %
No	833(93.4%)	660(90.2%)	0.018
Yes	59(6.6%)	72(9.8%)
Peptic ulcer, %
No	835(93.6%)	637(87.0%)	< 0.001
Yes	57(6.4%)	95(13.0%)
Atrophic gastritis, %
No	756(84.8%)	500(68.3%)	< 0.001
Yes	136(15.2%)	232(31.7%)
Prior probiotics use, %
No	483(54.1%)	346(47.3%)	0.006
Yes	409(45.9%)	386(52.7%)
Prior antibiotic use, %
No	672(75.3%)	571(78.0%)	0.207
Yes	220(24.7%)	161(22.0%)
Prior proton-pump inhibitors use, %
No	611(68.5%)	527(72.0%)	0.126
Yes	281(31.5%)	205(28.0%)

Notably, the prevalence of CDI differed between groups. The H. pylori-positive group (13.7%, 95% CI: 11.2–16.2%) demonstrated significantly higher rates of Clostridioides difficile infection than the negative group (9.0%, 95% CI: 7.1%−10.8%) (P = 0.003) (Fig. 2).

Fig. 2.

Comparison of the prevalence of CDI between the H. pylori-positive and H. pylori-negative groups

Univariate and multivariate logistic regression analyses

Univariate logistic regression analysis revealed that H. pylori-positive patients had a significant positive association with CDI (OR: 1.606, 95% CI: 1.176–2.194, p = 0.003) (Table 2). Additionally, we found that patients aged ≥ 60 years, with a history of CDI, IBD, prior antibiotic use, and prior PPI use were positively associated with CDI, whereas prior probiotic use was negatively associated with CDI (Table 2).

Table 2.

Univariate logistic regression analysis of the association for CDI

Variables	Univariate logistic regression	P
Sex
Male	Reference
Female	1.023(0.742–1.409)	0.892
Age
< 60	Reference
≥ 60	1.451(1.043–2.018)	0.027
Marital status
Unmarried	Reference
Married	0.925(0.673–1.271)	0.63
Smoking
No	Reference
Yes	1.166(0.831–1.638)	0.374
Drinking
No	Reference
Yes	1.119(0.784–1.598)	0.536
Medical Insurance
Resident Medical Insurance	Reference
Employee Medical Insurance	1.116(0.813–1.532)	0.497
Place of Residence
Urban Residents	Reference
Rural Residents	0.827(0.604–1.133)	0.236
Duration of Hospitalization
< 7 days	Reference
≥ 7 days	1.257(0.893–1.769)	0.190
Patient source
Initially presenting patients	Reference
Referred patients	1.053(0.702–1.580)	0.803
Previous H. pylori infection
No	Reference
Yes	1.216(0.862–1.714)	0.265
Unknown	0.870(0.547–1.384)	0.556
Previous Clostridioides difficile infection
No	Reference
Yes	2.274(1.380–3.747)	0.001
Unknown	1.308(0.727–2.354)	0.37
Inflammatory bowel disease
No	Reference
Yes	1.466(1.065–2.019)	0.019
Liver cirrhosis
No	Reference
Yes	0.998(0.655–1.521)	0.993
Malignant tumors
No	Reference
Yes	1.098(0.719–1.676)	0.665
Diabetes
No	Reference
Yes	0.984(0.613–1.579)	0.947
Cardiovascular disease
No	Reference
Yes	1.283(0.903–1.821)	0.164
Malnutrition
No	Reference
Yes	1.501(0.907–2.484)	0.114
Peptic ulcer
No	Reference
Yes	1.242(0.755–2.045)	0.393
Atrophic gastritis
No	Reference
Yes	1.196(0.837–1.71)	0.326
Prior probiotics use
No	Reference
Yes	0.631(0.460–0.866)	0.004
Prior antibiotic use
No	Reference
Yes	1.699(1.215–2.375)	0.002
Prior proton-pump inhibitors use
No	Reference
Yes	1.573(1.141–2.167)	0.006
H. pylori infection
No	Reference
Yes	1.606(1.176–2.194)	0.003

We further included H. pylori infection, age, history of CDI, IBD, prior probiotic use, prior antibiotic use, and prior PPI use in a multivariate logistic regression model. Multivariate logistic regression analysis demonstrated that H. pylori infection remained significantly associated with CDI (OR: 1.817, 95% CI: 1.320–2.501, p < 0.001) (Table 3). A Hosmer-Lemeshow test p-value > 0.05 indicated good model fit for the multivariate model. VIFs for all variables in the multivariate model were < 5, suggesting no significant multicollinearity. The maximum Cook’s D among patients included in the multivariate model was 0.2 (< 0.5), indicating the absence of significant influential outliers.

Table 3.

Multivariate logistic regression analysis of the association for CDI

Variables	Multivariate logistic regression	P
Age
< 60	Reference
≥ 60	1.584(1.127–2.227)	0.008
Previous Clostridioides difficile infection
No	Reference
Yes	1.998(1.148–3.479)	0.014
Unknown	1.273(0.694–2.337)	0.435
Inflammatory bowel disease
No	Reference
Yes	1.121(0.580–2.168)	0.734
Prior probiotics use
No	Reference
Yes	0.471(0.334–0.664)	< 0.001
Prior antibiotic use
No	Reference
Yes	1.814(1.227–2.682)	0.003
Prior proton-pump inhibitors use
No	Reference
Yes	1.334(0.631–2.821)	0.451
H. pylori infection
No	Reference
Yes	1.817(1.320–2.501)	< 0.001

Propensity score matching analysis

We performed PSM at a 1:1 ratio between the H. pylori-positive and H. pylori-negative groups, matching all covariates using a caliper of 0.02. After PSM, 602 patients were included in both the H. pylori-positive and H. pylori-negative groups. Before matching, the two groups differed in the distribution of variables such as malnutrition, peptic ulcer, atrophic gastritis, and prior probiotic use. After PSM, no significant differences were observed between the H. pylori-positive and H. pylori-negative groups across all covariates (Supplementary Table 1). The histograms of propensity scores showed that the two cohorts were well matched (Supplementary Fig. 1).

After PSM, the prevalence of CDI in the H. pylori-positive group (14.6%, 95% CI: 11.8%−17.4%) remained significantly higher than that in the H. pylori-negative group (8.5%, 95% CI: 6.2%−10.7%) (p < 0.001) (Supplementary Fig. 2). Similarly, after PSM, univariate logistic regression analysis still showed that H. pylori-positive patients had a significantly positive association with CDI (OR: 1.850, 95% CI: 1.284–2.665, p < 0.001).

Subgroup analyses

To further analyze whether there were differences in outcomes across different subgroups, we conducted subgroup analyses based on variables such as age, history of CDI, presence of IBD, prior probiotic use, prior antibiotic use, and prior PPI use (Fig. 3). Among patients aged ≥ 60 years, H. pylori positivity was significantly associated with CDI (OR = 2.083, 95% CI: 1.204–3.603, P = 0.009, FDR = 0.040), whereas the association in patients aged < 60 years did not reach statistical significance (OR = 1.450, 95% CI: 0.991–2.121, P = 0.056, FDR = 0.056). However, the interaction test between age and H. pylori infection was not statistically significant (P for interaction = 0.297). Similarly, among patients not using PPIs, H. pylori positivity was significantly associated with CDI (OR = 1.629, 95% CI: 1.095–2.424, P = 0.016, FDR = 0.046), whereas the association did not reach statistical significance among PPI users (OR = 1.655, 95% CI: 0.995–2.752, P = 0.052, FDR = 0.056). Again, the interaction test was not significant (P for interaction = 0.962). H. pylori infection was significantly associated with CDI across all remaining subgroups (history of CDI, presence of IBD, prior probiotic use, and prior antibiotic use). These findings suggest that, although variations in effect estimates were observed across certain subgroups, no statistically significant interaction was identified, indicating the absence of a consistent subgroup-specific effect.

Fig. 3.

The association of CDI in H. pylori-infected patients across different subgroups

Sensitivity analyses

To evaluate the potential impact of unmeasured confounding, we conducted a series of worst-case scenario sensitivity analyses. For each potential confounder not directly measured in our dataset—such as detailed antibiotic exposure(type and duration), comorbidity severity, healthcare exposure(intensity and duration)—we assumed an implausibly extreme distribution of risk factors between H. pylori-positive and H. pylori-negative groups. Specifically, we modeled a situation where all individuals with the high-risk condition (e.g., long-duration or high-risk antibiotics, advanced malignancy, or poorly controlled comorbidities) were exclusively assigned to the H. pylori-positive group, while all individuals without the high-risk condition were assigned to the H. pylori-negative group. We then calculated the adjusted ORs under each scenario, assuming that each high-risk exposure confers a five-fold increase in the risk of CDI (risk ratio = 5). The sensitivity analyses yielded the following adjusted ORs under the extreme assumptions (Supplementary Table 2) Under these highly implausible and extreme distributions, the adjusted ORs mostly remained close to the primary estimate, varying between approximately 0.98 and 1.37, except for healthcare exposure, which showed a markedly lower OR of 0.321 when all H. pylori-positive patients were assumed to have high-intensity, long-duration healthcare exposure while none in the negative group had such exposure.This finding suggests that healthcare exposure as an unmeasured confounder may introduce substantial bias in the estimated association between H. pylori infection and CDI risk. For other factors, the impact appeared more modest, indicating relative robustness of the primary findings to unmeasured confounding under these extreme assumptions.

Discussion

Nearly 4.4 billion people worldwide are infected with H. pylori, making it the most common chronic bacterial infection globally [23]. Recent global data indicate that the prevalence of H. pylori infection has declined from 52.6 to 43.9% over the past three decades [1]. With growing awareness of the role of H. pylori infection in the development of chronic active gastritis, peptic ulcer, peptic ulcer bleeding, and gastric cancer, increasing attention has been given to the detection, treatment, and prevention of H. pylori infection [24]. In recent years, the incidence and severity of CDI have been increasing globally, driven by the rising use of antibiotics, immunosuppressants, and corticosteroids, as well as the emergence and widespread transmission of hypervirulent strains such as RT027 [25, 26]. Our study demonstrated an association between H. pylori infection and CDI. H. pylori infection notably was notably associated with CDI, and this positive association was not influenced by underlying diseases, prior probiotic, antibiotic, and PPI use.

Previous studies have suggested that H. pylori infection may be associated with changes in the intestinal microbiome. H.pylori colonization has been reported to be correlate with measurable alterations in gut microbial diversity. However, findings from current studies remain inconsistent. Some studies indicated that patients with H. pylori infection exhibited greater gut microbiota diversity than those who were H. pylori-negative [7, 27–29]. In contrast, a study by Martín-Núñez et al. found that H. pylori-positive patients had significantly lower α-diversity in their gut microbiota compared to H. pylori-negative individuals [30]. Similarly, a recent study found that patients infected with CagA-positive H. pylori strains showed reduced microbial α-diversity [31]. Consistently, a study by Wang et al. also indicated that the fecal microbiota α-diversity in H.pylori-negative control patients was significantly higher than that in H.pylori-positive patients [32]. In addition to diversity differences, H. pylori infection has been linked to variation in the relative abundances of specific gut microbial taxa [8]. For instance, Iino et al. observed higher abundances of Haemophilus, Gemella, Streptococcus, and Actinobacteria in infected individuals [27], while Frost et al. reported increased levels of Prevotella, Parasutterella, Betaproteobacteria, Holdemanella, Alisonella, and Howardella, alongside reduced levels of Pseudoflavonifractor and Bacteroidetes [7]. Sun et al. found notable shifts in gut microbial composition, including decreased Verrucomicrobia and Proteobacteria and increased Collinsella, Dorea, and Parabacteroides in infected patients [29]. A recent study also reported elevated levels of Clostridioides difficile and Salmonella, along with decreased abundances of Firmicutes, Bacteroides, Lactobacillus, Escherichia coli, and Methanobrevibacter smithii in H. pylori-positive individuals [6]. These findings raise the possibility that H. pylori-associated disruptions of the gut microbiota may allow pathogenic bacteria, such as Clostridioides difficile, to expand their ecological niche.

H. pylori infection has been associated with altered gut metabolic profiles, based on the imbalance of the gut microbiota. Sun et al. indicated that H. pylori infection was strongly linked to the metabolism of intestinal short-chain fatty acids (SCFAs) [29]. SCFAs, primarily including acetate (AA), propionate (PA), and butyrate (BA), are generated by gut bacteria through the fermentation of indigestible carbohydrates. They play several important roles, including supporting intestinal barrier integrity, regulating immune responses, and reducing inflammation [33]. Sun et al. observed notable declines in AA and BA concentrations among patients with confirmed H. pylori infection [29]. Previous studies suggested that short-chain fatty acids, especially BA, may significantly inhibit the growth of Clostridioides difficile within the intestinal ecosystem [34–38]. Regarding acetate, a study by Fachi et al. demonstrated that although acetate does not inhibit the growth or toxin production of Clostridioides difficile in vitro or in vivo, it may enhance host resistance to CDI through coordinated actions on neutrophils and group 3 innate lymphoid cells (ILC3s) [39]. Similarly, a recent study based on metagenome-assembled genomes (MAGs) suggested that acetate may play a potentially protective role within the disrupted gut microbiota caused by Clostridioides difficile [40]. Taken together, these previously published findings suggest a possible association between H. pylori infection, changes in microbiota and metabolites, and CDI susceptibility, which warrants further investigation in longitudinal studies.

Although H.pylori infection may be associated with CDI through alterations in intestinal microbiome and disruption of short-chain fatty acid metabolism, few studies have directly examined the potential association between H. pylori infection and CDI. Most research has focused on the impact of drug therapy—particularly high-dose antibiotic use—during H. pylori eradication on the development of CDI. However, the impact of anti-Helicobacter pylori therapy on CDI remains controversial. Early case reports suggested that anti-Helicobacter pylori therapy, particularly the use of antibiotics, may contribute to the development of Clostridioides difficile infection [41–45]. However, a recent large-scale retrospective cohort study found no association between H. pylori treatment and subsequent CDI (OR = 1.49; 95% CI: 0.67–3.29) [18]. Our study primarily focused on the association between Helicobacter pylori infection itself and Clostridioides difficile infection. Muhsen et al. explored the link between serological markers of H. pylori and CDI, supporting our findings [46]. Their results showed that compared to H. pylori seronegative patients, individuals who were seropositive for H. pylori IgG but lacked CagA IgG antibodies had a threefold increased risk of Clostridioides difficile infection. Meanwhile, patients with the H. pylori CagA phenotype had a ninefold higher likelihood of developing CDI [46]. However, H. pylori seropositivity does not necessarily indicate current infection, and thus, serological markers cannot fully assess the relationship between active H. pylori infection and CDI. In our study, H. pylori infection was primarily confirmed using the rapid urease test and urea breath test, minimizing the confounding effect of previous H. pylori infection. Our study suggests that current H. pylori infection, rather than past infection, is an independent association with intestinal CDI. Although we found that current H.pylori infection may be associated with intestinal CDI, whether eradication of H.pylori can reduce the risk of CDI remains to be further investigated.

Moreover, our subgroup analysis indicated that the association between CDI and H. pylori infection was not influenced by the presence of IBD, history of CDI, prior antibiotic, and probiotic use. However, we found that in patients aged less than 60 years, H. pylori infection was not significantly associated with CDI, whereas in the subgroup of patients aged over 60 years, H. pylori infection remained significantly positively associated with CDI. Studies have shown that with increasing age, the human gut microbiome becomes imbalanced, characterized by an increase in pro-inflammatory symbiotic bacteria and a decrease in beneficial symbiotic microorganisms [47–49]. Previous studies have also indicated that advanced age is a significant risk factor for Clostridioides difficile infection [50, 51]. Therefore, the gut microbiome imbalance in elderly patients may increase the susceptibility of individuals with H. pylori infection to Clostridioides difficile infection. Similarly, we also found that in patients not using PPIs, H. pylori infection was significantly associated with CDI, whereas in the PPI user subgroup, the association between H. pylori infection and CDI was not statistically significant. We speculate that this may be due to a certain false-negative rate of H.pylori testing in patients who have recently used PPIs [52, 53], which may have resulted in some H.pylori-positive patients being misclassified into the H.pylori-negative group, thereby attenuating the observed association between H.pylori and CDI in the PPI user subgroup. Although differences were observed in some subgroups during the subgroup analysis, all interaction tests were not statistically significant; therefore, no clear subgroup effect was identified overall.

Multiple studies have demonstrated that the use of broad-spectrum antibiotics is associated with an increased risk of CDI [14, 54, 55]. Different antibiotics may confer varying risks of CDI, with carbapenems (imipenem, meropenem, ertapenem), fluoroquinolones (levofloxacin, ciprofloxacin), clindamycin, amoxicillin-clavulanate, and third/fourth-generation cephalosporins being associated with a higher risk of CDI [14]. Similarly, compared to a 7-day antibiotic course, a 14-day course is associated with a 27% increased risk of CDI [56]. For treatment-naïve H.pylori–infected patients, the current ACG guidelines recommend bismuth quadruple therapy (BQT) for 14 days as the first-line regimen when antibiotic susceptibility is unknown [57]. The use of high-dose, long-duration broad-spectrum antibiotics raises significant concerns about the increased risk of CDI. For patients with an initial, non-severe episode of CDI, the latest IDSA/SHEA and ESCMID guidelines recommend fidaxomicin as the first-line treatment, with vancomycin and metronidazole as alternative options [58]. Given the potential association between H.pylori infection and intestinal CDI, clinicians remain concerned that eradicating H. pylori in patients with concurrent CDI may exacerbate the course of the infection. Currently, the ACG guidelines on H.pylori infection, as well as the latest IDSA/SHEA and ESCMID guidelines for CDI, did not address treatment strategies for patients with concurrent H. pylori and Clostridioides difficile infections. Only Abboud et al. have reported a case of concurrent Helicobacter pylori and Clostridioides difficile infection and shared their clinical experience regarding the management of such co-infections [59]. Given the observed association between H. pylori infection and CDI in this cross-sectional study, and the complexity of managing patients with concurrent H. pylori and CDI, we recommend considering screening for CDI in H. pylori-infected patients presenting with diarrhea. However, considering the exploratory nature of our study and the absence of validated risk stratification frameworks or evidence-based guidelines for managing co-infections of H. pylori and Clostridioides difficile, the clinical implications of our findings should be considered preliminary. While our results suggest a potential association that may warrant heightened clinical awareness, they are not intended to support specific screening protocols or treatment algorithms at this stage. In addition, our findings only suggest a potential association between H. pylori and CDI. Investigating the risk-benefit ratio of H. pylori therapy in the context of potential CDI risk is beyond the scope of this study. Such an analysis would require longitudinal data stratified by treatment regimens, comorbidities, and patient-specific risk factors, which were not available in our cross-sectional design. Future research incorporating detailed risk modeling and cost-effectiveness analysis is needed to quantify these clinical trade-offs and inform personalized, actionable treatment strategies.

This is the first systematic analysis to evaluate the association between H. pylori infection and intestinal CDI. However, our study may still have the following limitations. First, it is important to note that the present study employed a cross-sectional design, which inherently limits the ability to infer any temporal or causal relationship between H.pylori infection and CDI. As such, it remains unclear whether H. pylori infection precedes, coincides with, or follows CDI onset. This limitation should be considered when interpreting our findings. Additionally, short-term fatal cases may have been missed (prevalence-incidence bias), and it was difficult to distinguish transient colonization from persistent infection. To better understand the potential causal mechanisms and temporal dynamics between H. pylori infection and CDI, future studies should adopt prospective cohort designs. Longitudinal follow-up of patients with confirmed H. pylori status prior to CDI onset would be essential to clarify the directionality and potential causality of this association. Second, as a retrospective study, although we employed multivariable regression analysis and PSM to reduce selection bias and retrospective bias, inherent biases may still exist and influence the results. We were unable to obtain detailed information on the specific types, duration, and timing of antibiotic and PPI use, which are closely associated with the risk of CDI. Certain antibiotics, such as clindamycin and fluoroquinolones, as well as prolonged use of these medications, are known to carry a higher risk of CDI. If systematic differences in these exposures existed between the H. pylori-positive and H. pylori-negative groups, substantial confounding bias could have been introduced into the observed association. Similarly, recent use of antibiotics and PPIs may influence the results of H.pylori testing, leading to a higher rate of false-negative results and, consequently, an underestimation of the true prevalence of H.pylori infection in our study population. In addition, we were unable to assess the intensity and duration of recent healthcare exposure, a key risk factor for the development of CDI. In our extreme-case sensitivity analyses, healthcare exposure emerged as a potentially critical confounding variable. When assuming that all H. pylori-positive individuals experienced high-intensity, long-duration healthcare exposure, while none in the negative group had such exposure, the adjusted OR decreased to 0.321. This finding suggests that healthcare exposure may have the potential to introduce substantial and directional bias into the observed association. Although we adjusted for the presence of comorbidities, we did not further assess their severity or specific impact on host immune function. The severity of comorbid conditions may influence individual susceptibility to CDI and could be unequally distributed between H. pylori infection groups, thereby acting as a potential confounder. Collectively, the absence of evaluation for these important clinical variables not only limits our ability to further explore the underlying mechanisms but also poses a challenge to the reliability of our results. The lack of adjustment for these potential confounders represents a major limitation of this study and constrains our capacity to interpret the association between H. pylori infection and CDI risk. Future studies should be based on more comprehensive clinical data, with detailed documentation and control of factors such as antibiotic use, comorbidity severity, and healthcare exposure, in order to further validate our findings. Meanwhile, our study included only diarrheal patients who underwent testing for both H.pylori and CDI. Since the incidence of CDI may be higher among patients with diarrhea than in the general population, this could introduce a degree of selection bias. In addition, some patients presented directly to our center, while others were referred from other hospitals. Although there was no significant difference in patient source between the two groups, we acknowledge that referral patterns may be influenced by factors such as disease severity, hospital tier, or the clinical judgment of the referring physician. Therefore, the potential for referral bias cannot be entirely excluded. Third, our study was conducted at a single center and did not include data from other centers, regions, or ethnic groups. Although the majority of patients in our center were from the same province, regional differences in factors such as diet, living environment, and socioeconomic status may be relatively minimal. Similarly, both groups of patients were enrolled through the same recruitment channels, with largely overlapping enrollment periods. The diagnosis of Clostridioides difficile infection (CDI) was conducted using identical criteria for both groups. We also compared baseline demographic characteristics—including age, sex, urban versus rural residence, and insurance type—and found no significant differences. However, we were unable to obtain data on certain potentially important socioeconomic variables, such as income level and educational attainment. These factors may influence patients’ disease awareness and access to healthcare services, thereby introducing potential bias. Therefore, we acknowledge that the study population may not fully represent a truly comparable source population. Furthermore, different regions or hospitals may have varying antibiotic prescribing patterns and infection control protocols. Relevant studies have shown that strengthening hospital-based antibiotic stewardship can reduce CDI rates and improve antimicrobial resistance patterns [54, 60]. Similarly, enhancing environmental disinfection, hand hygiene among healthcare workers, and implementing appropriate visitor contact precautions can reduce the overall risk of CDI in hospital settings [61, 62]. Moreover, H.pylori strains and their effects on the microbiome may vary geographically. The conclusions of our study may be influenced by the specific characteristics of H.pylori strains prevalent in China. Therefore, whether these findings are generalizable to other centers, populations, or regions requires further investigation. Fourth, some of our data contained missing values. Although classifying missing data as “unknown” allowed us to retain the sample size, non-random missingness may have introduced residual confounding. Fifth, although we proposed potential mechanisms underlying the association between Helicobacter pylori infection and intestinal CDI, these hypotheses are based on inferences from prior studies. Our research did not include direct microbiome or metabolomic analyses to support these assumptions. Moreover, we did not examine the interactions between gastric and intestinal microbiota, nor investigate how such interactions might mediate the observed association. Therefore, these mechanistic links should be interpreted with caution, and future studies integrating multi-omics approaches are needed to validate these potential pathways.

Conclusion

This study provides new evidence that current H. pylori infection, rather than past infection, is independently associated with intestinal Clostridium difficile infection. Given the study’s limitations, multicenter prospective randomized trials are needed to validate these findings. Further research should also explore the complex interactions between gastric and intestinal microbiomes and the underlying mechanisms through multi-omics approaches.

Abbreviations

H. Pylori

Helicobacter pylori

CDI

Clostridioides difficile infection

PSM

Propensity score matching

WHO

World Health Organization

CDC

Centers for Disease Control and Prevention

CA-CDI

Community-acquired Clostridioides difficile infection

PPIs

Proton pump inhibitors

IBD

Inflammatory bowel disease

GDH

Glutamate dehydrogenase

EIA

Enzyme immunoassays

qPCR

Real-time quantitative PCR

VIFs

Variance inflation factors

Cook’s D

Cook’s distance

FDR

False discovery rate

SCFAs

Short chain fatty acids

AA

Acetate

PA

Propionate

BA

Butyrate

ILC3s

Group 3 innate lymphoid cells

MAGs

Metagenome-assembled genomes

BQT

Bismuth quadruple therapy

Authors’ contributions

LLY designed the study. Data analysis was done by LLY, and HFT. YZ and CLD wrote the draft of the manuscript. JJQ, YY and CW wrote the revised version of the manuscrip. All authors edited the manuscript, interpreted study findings, and approved the final draft of the manuscript.

Funding

None.

Data availability

The data used to support the findings of this study are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the guidelines of the Declaration of Helsinki. The requirement for ethics approval and informed consent was waived by the Medical Ethics Committee of the Second Hospital of Nanjing, Affiliated to Nanjing University of Chinese Medicine, due to its retrospective design and use of deidentified data. Patients were assigned serial numbers according to the order of data entry. After collecting the required variables, all direct identifiers (such as names, identification numbers, and medical record numbers) were removed. Additionally, no patient-identifiable information is presented anywhere in the manuscript, ensuring the protection of patient privacy.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.