Introduction
Community-acquired Clostridioides difficile infection (CA-CDI) was first described in 1980. [Reference Stergachis1] In the past, it was thought that C. difficile was an exclusively hospital-acquired pathogen, but it is now recognized as an important cause of diarrhea in the community setting. CA-CDI can be defined, as per the Center for Disease Control and Prevention (CDC), as an onset of symptoms in the community or ≤ 48 h after admission to a healthcare institution, provided that the time of symptom onset was greater than 12 weeks after the last discharge from a healthcare institution. [Reference McDonald2]
The epidemiology of C. difficile has evolved in the past decades, highlighting an increased transmission of CDI in community settings. [Reference Zanichelli3, Reference Guh4] The infection’s severity ranges from an asymptomatic colonization, mild to severe diarrhea, to life-threatening inflammation to the colon like a fulminant colitis that can lead to death. [Reference Gravel5, Reference Joshi, Macken and Rampton6] Approximately around 40% of patients with CA-CDI require hospitalization, 20% experience treatment failure, and about 28% have recurrent episodes. [Reference Khanna7] Furthermore, CDI has a case-fatality rate of up to 14% within 30 days after diagnosis, with recurrences that can increase illness rates and decrease quality of life; still, morbidity and mortality could be determined by the changing virulence of the pathogen. [Reference Kotila8–Reference Zhang, Prabhu and Marcella10]
CDI not only burdens patients and healthcare workers, but its impact is also noticeable in healthcare costs. CDI may have resulted in as much as $4.8 billion in excess healthcare costs in acute-care facilities during 2008. [Reference Dubberke and Olsen11] Even so, CDI in the community might be underdiagnosed so the true burden of the disease might be greater than the ones reported by studies. [Reference Bloomfield and Riley12] Still, a full appreciation of the burden that CDI has on the healthcare system is necessary for adequate resource allocation.
Even with the burden CA-CDI represents to the healthcare system, there has been considerable heterogeneity in the incidence and prevalence reports. Some studies state that there is a decline in cases of CA-CDI [Reference Du13], while others point towards an increase of cases. [Reference Khanna and Pardi14, Reference Ofori15] Regardless, some studies that report either outcome compare the prevalence of CA-CDI to hospital acquired CDI (HA-CDI), yielding an inaccurate estimate of CA-CDI in the general population. Due to the heterogeneity of the reports and due to the increasing burden that CA-CDI cases are contributing to healthcare, the following systematic review and meta-analysis were developed, with the objective of assessing the epidemiologic profile of community-acquired C. difficile.
Methods
This systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) checklist and the guidelines from the Cochrane Handbook for Systematic Reviews of Interventions (Supplementary material I). This study was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the registration number CRD42023451134.
Databases and search strategy
A comprehensive literature search was performed by an experienced librarian with the collaboration of the research team from inception up to April 2023. The search was conducted in multiple electronic databases including MEDLINE, Scopus, Web of Science, EMBASE, the Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Clinical trials.gov. The search included Medical Subject Headings (MeSH) terms as well as specific keywords related to the research question. A tailored search strategy was done in each database, with a combination of terms such as C. difficile, community acquired, prevalence, incidence, epidemiology. The following is the search strategy used for Web of Science: TS = (((("Clostridioides difficile" OR “Clostridium difficile” OR “Clostridium difficilis” OR “Peptoclostridium difficile” OR “Bacillus difficilis” OR “CA-CDI”) AND ("Community acquired" OR “community acquired infection” OR “community acquired disease” OR “community associated disease” OR “community associated infection” OR “Community-associated” OR “Community-acquired” OR community)) AND (prevalence OR “prevalence study” OR “incidence” OR “incidence rate” OR “rate, incidence” OR “epidemiology”))). The full tailored search for each database can be found in Supplementary material II.
Searching and eligibility of studies
Retrieved articles were exported to EndNote reference software version 9 citation manager where they were deduplicated using the native deduplication function within the software, followed by manual review.
The studies that remained were imported into a systematic review software (Distiller SR), where the studies were screened in two phases: the title and abstract phase and the full-text phase. Articles included in both phases were evaluated independently by two reviewers. Studies included by at least one reviewer in the abstract screening phase were considered for full-text screening; this was done to increase sensibility in the included records.
During the full-text screening, agreement of inclusion between both reviewers was required for the study to be selected. Disagreements at any phase were resolved by consensus. Furthermore, before each phase, a pilot study was conducted to ensure inter-rater agreement by Kappa statistic. A Kappa statistic of >0.70 was set as an appropriate inter-rater agreement. The data extracted included the year of publication, country where the study was conducted, CA-CDI definition used by the authors, the number of samples processed, the diagnostic tool used, age groups included, population used to estimate incidence, CA-CDI cases reported, time frame, and risk factors reported by the authors.
Eligibility criteria
Due to the nature of the outcome (prevalence and incidence), only published and unpublished cross-sectional or observational studies were considered for inclusion. The study population will be any primary study that reports the epidemiologic profile of C. difficile, specifically prevalence or incidence rates. If the studies do not report these indicators explicitly, they can be included if they provide other indicators that can be used to calculate prevalence or incidence rates.
Outcome measurement of the study
The two main quantitative outcomes were the prevalence and incidence of CA-CDI, along with assessing qualitatively the factors associated with CA-CDI. The prevalence of CA-CDI was defined as the percentage of CA-CDI cases from a population of patients presenting diarrhea. Prevalence was extracted either as reported by the authors or the required information was calculated by the research team in the extraction sheet. Incidence was defined as the rate of new cases of CA-CDI over a specified time for the population at risk. Incidence was extracted as reported or calculated by dividing the new cases of CA-CDI reported by the result of multiplying the population at risk and the timeframe of the study in years. [Reference Tenny and Boktor16] Factors associated with CA-CDI were extracted as reported.
Quality assessment
Two authors independently assessed the quality of the studies included. Depending on the study design, AXIS or New-Castle Ottawa (NOS) tools were used for cross-sectional and cohort or case-control studies respectively. [Reference Downes17, Reference Wells18] For studies evaluated with AXIS, a predefined score of 17 of 20 for high-quality studies was set by the research team. On the other hand, for studies evaluated with New-Castle Ottawa, a predefined rating between 0–2, 3–5, and 6–9 was established as poor, fair, and good/high quality, respectively.
Data processing and analysis
All the extracted data were recorded in a Microsoft Excel spreadsheet and cleaned for analysis. Heterogeneity in the data was expected; therefore, a random effects model was established as the primary model for the analysis a priori. We estimated the prevalence of CA-CDI in using a binomial – normal model for meta-analysis of prevalence via the generalized linear mixed model. [Reference Stijnen, Hamza and Özdemir19] CA-CDI prevalence was reported as binomial proportion with 95% confidence intervals (CIs). CA-CDI incidence was also estimated with a generalized linear mixed model with summary findings being reported as CA-CDI cases per 100,000 person-years with 95% CIs. Statistical heterogeneity was tested using Cochran’s Q test and I2 index with its corresponding p-value. A statistical heterogeneity test with a p-value of less than 0.10 was considered significant for heterogeneity. [Reference Cumpston20] The values of I2 defined a priori as low, moderate, and high heterogeneity were 25%, 50%, and 75% respectively. [Reference Higgins and Thompson21] Pooled data are presented with forest plots.
Subgroup analysis, established a priori, by age groups was performed, and furthermore, sensitivity analysis was performed using influential analysis. Publication bias was explored by Egger’s test and visually with funnel plots. All statistical analyses were performed in R 4.3.0 with the meta and dmetar libraries. Factors associated with CA-CDI were qualitatively synthesized.
Results
Characteristics of the studies
A total of 3,642 articles were retrieved on the initial search, from which 1,691 were excluded due to duplication. After title and abstract screening, 349 were included in full-text screening. After screening 49 articles in total, 19 articles were included for the prevalence outcome and 43 for the incidence outcome. A visual representation of the literature screening process can be seen in Figure 1.
Figure 1. PRISMA flow diagram.
Characteristics of the studies and study participants
Approximately 83,105 processed samples (not reported by all studies) for CDI were included in this study. Of the included articles, fifteen were from the USA [Reference Guh4, Reference Khanna22–Reference Abrahamian35]; five from Spain [Reference Alcalá36–Reference Andrés-Lasheras40]; four from Australia [Reference Collins71] [Reference Slimings41–Reference Furuya-Kanamori43]; three from Canada [Reference Zanichelli3, Reference Xia44, Reference Allard45], Scotland [Reference Banks46–Reference Marwick48], Sweden [Reference Malmqvist49–Reference Karlström51]; two from France [Reference Penit52, Reference Lefevre-Tantet-Etchebarne53]; and one from Bailiwick of Jersey [Reference Kumar54], China [Reference Ho55], Finland [Reference Kotila8], Germany [Reference Weil56], Iceland [Reference Vesteinsdottir57], India [Reference Ingle58], Iran [Reference Azimirad59], Ireland [Reference Maisa60], Israel [Reference Na’Amnih61], Japan [Reference Mori and Aoki62], Kuwait [Reference Jamal, Pauline and Rotimi63], Netherlands [Reference van Dorp64], New Zealand [Reference Johnston65], and Slovakia [Reference Garabasova66]. The rest of the extracted characteristics can be seen in Table 1.
Table 1. Baseline characteristics of the included studies
NR: Not reported; CA-CDI: Community acquired Clostridioides difficile infection.
Prevalence of CA-CDI
The overall pooled prevalence of CA-CDI, obtained from a total of 62,148 patients, was 5% (95% CI 3–8; Figure 2). A subgroup analysis by age groups of the included samples from each study was performed, which showed no statistical subgroup difference (p = 0.58, Supplement III).
Figure 2. Pooled prevalence of CA-CDI.
Incidence of CA-CDI
The overall pooled incidence of CA-CDI was 7.53 patients (95% CI 4.45–12.74) per 100,000 person-years (Figure 3). Furthermore, subgroup analysis revealed a statistically significant difference when divided by age group (p < 0.01, Supplement III).
Figure 3. Pooled incidence of CA-CDI.
Heterogeneity and publication bias
This systematic review and meta-analysis detected high heterogeneity for both outcomes (I2 100% 95% CI 100–100, p < 0.001). Preplanned sensitivity analysis was performed via influence analysis. For CA-CDI prevalence, influence analysis showed Maisa et al. [Reference Maisa60] as a potential outlier, influencing the results. Repeating the analysis without Maisa et al. resulted in a pooled prevalence of 4% (95% CI 3–6), with a I2 of 98% (95% CI 98–98). On the other hand, for CA-CDI incidence, influence analysis did not show any potential outliers.
Publication bias was assessed visually and statistically via funnel plots and Egger’s test, respectively. Although the funnel plot for prevalence of CA-CDI visually showed asymmetry, Egger’s test did not indicate the presence of asymmetry (p = 0.1914). Publication bias of CA-CDI incidence showed a different result, with both the funnel plot and Egger’s test showing indication of publication bias (p = 0.0035). Both funnel plots can be seen in Supplementary material IV.
Factors associated with CA-CDI
Sex and gender
Several studies report an increase of CA-CDI cases in female patients. [Reference Lessa28, Reference Gutiérrez, Riddle and Porter29] Furthermore, when compared to HA-CDI cases, CA-CDI patients were more likely to be female. [Reference Khanna30, Reference Banks46, Reference Penit52] Other studies also found a statistically higher incidence of CA-CDI in females when compared to males [Reference Kutty32], where Maisa et al. [Reference Maisa60] reported that CA-CDI cases had lower odds to be male (adjusted odds ratio [aOR] 0.71; 95% CI 0.58–0.87; p < 0.001), but Ingle et al. reported that they did not find a statistically significant difference in gender. [Reference Ingle58]
Antibiotics
Antibiotic use has been identified before as a risk factor for C. difficile associated disease. [Reference Lefevre-Tantet-Etchebarne53, Reference van Dorp64] Several studies reported that CA-CDI was more likely to have received antibiotics in the 2 months prior to developing the disease, with ORs ranging from 8.04–8.12 when compared to controls. [Reference Taori47, Reference Mori and Aoki62] Other authors have reported ORs of 6.09 (95% CI 4.59–8.08) when antibiotics were taken in the previous 6 months [Reference Kuntz31] and an almost 2-fold increased risk of CA-CDI when taking any antibiotic (1.94, 95% CI 1.35–2.77, p = 0.001). [Reference Marwick48]
Some of the most commonly reported antibiotics associated with increased risk for CA-CDI were co-amoxicillin, fluoroquinolone, clindamycin and cephalosporins, fluoroquinolones, beta-lactam/beta-lactamase inhibitors, macrolides, and penicillins. [Reference Dantes26, Reference Miranda-Katz27, Reference Kuntz31, Reference Marwick48] Dantes et al. reported a predicted overall 16.8% (6.0%–26.3%; p = 0.003) decrease in CA-CDI incidence each 10% reduction in the use of all antibiotics. [Reference Dantes26]
On the other hand, Ingle et al. [Reference Ingle58] reported that although antibiotic use was more common in the CA-CDI group as compared to controls (66.7% vs. 38.4%, p = 0.07), the difference was not statistically significant, with other authors reporting similar results. [Reference Vesteinsdottir57] Nevertheless, when compared to HA-CDI, CA-CDI patients are less frequently taking antibiotics (p < 0.001) [Reference Reigadas37].
Gastrointestinal therapy
Gastrointestinal therapy was also commonly evaluated as a risk factor for CA-CDI. Although Kuntz et al. [Reference Kuntz31] reported an aOR of 2.30 (95% CI 1.56–39) of developing CA-CDI when taking gastric acid suppressants 6 months before diagnosis, Jamal et al. reported a no-statistically significant prior exposure to gastrointestinal therapy when compared to control (p = 0.09). [Reference Jamal, Pauline and Rotimi63] Further more Mori et al. reported that prior exposure of antacids in the preceding 2 months was not a risk factor for CA-CDI (OR: 0.59, 95% CI: 0.19–1.85). [Reference Mori and Aoki62]
Additionally, when compared to HA-CDI, CA-CDI patients less frequently received proton pump inhibitors (p < 0.001). [Reference Reigadas37]
Age
Age was commonly reported as related to CA-CDI cases. Some studies have reported older age as a high predictor of CA-CDI cases, with different cut-off points, such as 40, 60, or 65 years [Reference Lessa28, Reference Gutiérrez, Riddle and Porter29, Reference Kutty32, Reference Karlström51, Reference Lefevre-Tantet-Etchebarne53] although not all studies have found the same difference. [Reference Ingle58] Moreover, when compared to HA-CDI, CA-CDI cases were significantly younger. [Reference Khanna30, Reference Reigadas37, Reference Maisa60]
Quality assessment
Out of 35 cross-sectional studies, 29 met the prespecified criteria for high quality study after being assessed using the AXIS tool, while 6 did not. [Reference Khanna22, Reference Suárez-Bode39, Reference Allard45, Reference Banks46, Reference Weil56, Reference Garabasova66] The median score from the AXIS tool was 19. For case-control studies, assessed by the NOS, nine out of ten were of good/high quality, while one study was deemed as fair quality [Reference Kutty32], with a median score of 7.5. All the four cohort studies included were of good/high quality, assessed using NOS, with a median score of 6. While these scores might be seen as abnormally high, it is important to consider that most studies included the total population and did not do any sampling.
Discussion
The systematic review and meta-analysis conducted on 49 studies aimed to provide a comprehensive understanding of the prevalence and incidence of CA-CDI. The investigation involved a meticulous examination of a diverse range of literature, incorporating data from epidemiological studies conducted worldwide, allowing for a more nuanced and representative analysis, enhancing the reliability and generalizability of the results. This comprehensive approach not only contributes to the existing knowledge on CA-CDI but also offers valuable insights for healthcare professionals, researchers, and policymakers involved in the prevention and management of this infectious disease in community settings.
The cumulative prevalence of community-associated C. difficile infection was found to be 5%, significantly lower than the prevalence reported in the surveillance report by the eCDC. [67] According to their findings, 32.7% of cases recorded from 2016–2017 were attributed to community-associated CDI or CDI with an unknown origin of cases. Nevertheless, information regarding prior hospitalization was not consistently gathered for all cases. For those cases where such information was available, the duration of prior hospitalization varied from 4 to 12 weeks, potentially leading to misclassification, a limitation acknowledged by the authors. The same report states that it was twice as common for CA-CDI cases to report prior contact with a long-term care facility in the previous three months than for all CDI cases.
Our qualitative analysis revealed that prior antibiotic exposure emerged as a prominent risk factor for the onset of CA-CDI, aligning with the observations made by Deshpande et al. [Reference Deshpande68] Their study reported an OR of 6.91 (95% CI 4.17–11.44) for any antibiotic use. Notably, with the sole exception of tetracyclines, virtually all other classes of antibiotics exhibited an association with an elevated risk of CA-CDI.
Across all age groups and globally, the incidence of CDI was recorded at 7.5 cases per 100,000 person-years, nearly four times higher than the figure reported by Marwick and colleagues, which stood at 2.0. [Reference Marwick48] However, in the context of adult patients, CA-CDI rates have been documented as reaching as high as 11.16 cases per 100,000 person-years. [Reference Kuntz31] It is worth noting that both referenced studies exclusively analyzed data from adults, unlike the compiled data, which includes a limited number of pediatric cases. Upon sub-analysis of the adult and elderly populations, the incidence escalates to 25 cases per 100,000 patient-years.
While CA-CDI can impact individuals of any gender, certain authors have documented a prevailing occurrence in females with a notable range spanning from 54% to 72.5%. [Reference Bauer69, Reference Thornton70] However, this gender-based trend is not consistently observed across all studies.
There are several limitations of this systematic review and meta-analysis. Firstly, most of the prevalence results had to be calculated, rather than extracted, with data provided by the included studies. This was done because those papers reported CA-CDI proportions with the population being patients with CDI and not patients with diarrhea. Furthermore, although most of the cross-sectional studies met the criteria for high quality studies, most of them were population studies, inflating their AXIS score. Lastly, the funnel plot and Egger’s test showed some indication of publication bias on the incidence outcome; therefore, results should be interpreted cautiously.
It is relevant that most of the reports of CA-CDI include patients with already diagnosed C. difficile, not patients with diarrhea. Including this population directly compares the proportions of CA-CDI versus HA-CDI. If the proportion of CA-CDI increases the proportion of HA-CDI decreases and vice versa. Although this is useful to examine C. difficile behavior when evaluating community and hospital infections and onset, this may limit the understanding of the actual behavior of CDI in the community.
We highly recommend that, if the objective of the scientific community is to examine C. difficile, future studies should include patients with diarrhea as their population, not just patients with CDI. Further recommendations include reporting the exact number used as a denominator when calculating and reporting an incident.
Conclusion
In conclusion, this meta-analysis underscores that among the included studies, the prevalence of CA-CDI stands at 5%, with an incidence rate of 7.3 cases per 100,000 person-years. Noteworthy risk factors identified include sex, prior antibiotic exposure, and age.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0950268825000202.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Author contribution
Conceptualization: A.C., N.A.Á., F.G.R.; Data curation: A.C., A.C.M., J.M.A., N.A.Á., F.G.R.; Funding acquisition: A.C.; Investigation: A.C., A.C.M., J.M.A., N.A.Á., F.G.R.; Methodology: A.C., A.C.M., J.M.A., N.A.Á., F.G.R.; Project administration: A.C., N.A.Á., F.G.R.; Resources: A.C., N.A.Á., F.G.R.; Supervision: A.C., N.A.Á.; Validation: A.C., A.C.M., J.M.A., N.A.Á., F.G.R.; Visualization: A.C., J.M.A., N.A.Á., F.G.R.; Writing – original draft: A.C., A.C.M., N.A.Á., F.G.R.; Writing – review & editing: A.C., A.C.M., J.M.A., N.A.Á., F.G.R.; Formal analysis: N.A.Á., F.G.R.; Software: N.A.Á., F.G.R.
Funding statement
Research was performed with the own research team funding.
Competing interest
The authors declare none.
Open access
