Epidemiological and genomic characterization of community-acquired Clostridium difficile infections

Clostridium difficile infections (CDI) are the most common cause of infectious diarrheal infection amongst hospitalized patients in North America and Europe [1]. CDI has recently surpassed methicillin-resistant Staphylococcus aureus as a hospital acquired infection [2, 3]. The incidence, mortality and associated health care costs associated with CDI are significant, with 10–25% of all cases of antibiotic-associated diarrheal onset attributed to CDI [4]. Clinically, nosocomial CDI has been well-studied with several risk factors for acquisition including hospitalization at time of infection, prior hospitalization, older age (> 65), antibiotic therapy (in particular fluoroquinolones, cephalosporins and clindamycin [5, 6]), use of nasogastric tubes, surgical procedures in the gastrointestinal tract, history of inflammatory bowel disease (IBD) and other states of immunosuppression [6, 7]. The pathogenicity of CDI is primarily attributed to the production of two exotoxins, toxin A and B that cause subsequent clinical manifestation including colonic dysregulation and cellular death [8]. A “binary” toxin has also been reported that may further contribute to pathogenesis in certain strains [1]. Additionally, CDI is spore-forming, allowing survival in the physical environment for prolonged periods of time [9], which challenges infection control and prevention in the hospital setting. Several other factors including gut microbiota disruption by antibiotic use and host endogenous antibodies to the toxins have also been implicated in CDI disease progression [6, 7, 10].

Recently, several studies have described the onset of community-acquired CDI (CA-CDI) [6]. The definition of CA-CDI requires the patient to not have been in a hospital or health care facility within the previous 12 weeks or to develop CDI symptoms within 48 h of hospital admission [11]. In contrast, nosocomial CDI requires that symptoms occur greater than 48 h after hospital admission or in less than 4 weeks after discharge from a health care facility [11]. Finally, indeterminate cases occur in the community between 4 and 12 weeks after hospital discharge has occurred [11]. CA-CDI rates are on the rise, with 20–45% [1, 9] of all CDI cases attributed to community onset, and a further 22% of patients having no history of antimicrobial several months prior to CDI onset [12].

CA-CDI rates amongst the pediatric population have also increased over time [13]. The clinical significance of C. difficile detection in infants and young children less than two years old is unclear, as they have been established as asymptomatic carriers [14]. Nevertheless, children may still serve as a means of transmission to other individuals in the household by exposure to spores within the same physical environment [15]. Exposure to C. difficile in the outpatient setting may serve as a means of transmission [16]. Two-thirds of CA-CDI patients had some form of occupational exposure within health care fields that may have preceded their illness [17]. There is also evidence that food-borne exposure to C. difficile may be a means of transmission within the community as spores have been demonstrated to survive normal cooking temperatures [18]. Zoonotic reservoirs exist in several animals including cattle and pigs [15, 19–23].

The aim of this study was to better understand the impact of CA-CDI by identifying risk factors linked to epidemiologically confirmed CA-CDI cases in the metropolitan center of Calgary (approximately 1.3 million people). C. difficile isolates from CA-CDI cases were further characterized by antibiotic susceptibility testing, pulsed field gel electrophoresis and whole genome sequencing.

CA-CDI is associated with significant clinical complications. It is estimated that 40% of patients require hospitalization, 20% suffer severe infection, 4.4% go on to have severe complications, 20% end up with treatment failure, and 28% have recurrent CDI [30]. An economic analysis from Canada found nearly 38,000 cases of CDI in 2012 with a total cost to society of $281 million, in part due to in-hospital investigations, community management and productivity loss [31]. Furthermore, management of CDI relapses alone was over $65 million. This has tremendous impact on the health care system in terms of costs, patient outcomes and burden on already limited resources [15]. We undertook to closely examine CA-CDI using a patient-based telephone questionnaire, to better understand this epidemiological group. We also analyzed the isolates using multiple laboratory methods to determine if unique traits associated with CA-CDI such as antibiotic resistance or genetic features exist.

Several studies have assessed epidemiology and risk factors associated with CA-CDI. One recent study from Japan [32] found the average age to be 58.8 with 50% > 65 years of age, similar to our findings. The gender distribution was also noted to be similar between studies (26.9% male). Interestingly, this study found the majority of patients were more likely to have received antimicrobials (oral fluoroquinolones most notably) and antacids. A recent report from the US found 40% of patients with CA-CDI were not previously exposed to antimicrobials [33]. Our study did assess antimicrobial prescription through the telephone survey but resulted in a large proportion of patients unable to recall the timeframe of antibiotic use. The most striking epidemiological finding in the current study is that all CA-CDI cases had some history of hospitalization if the definition were modified to hospital exposure in the last 12 months rather than the last 12 weeks as currently accepted. This raises the possibility that even CA-CDI has nosocomial origins albeit remote.

Metronidazole and vancomycin are considered first line agents in the clinical treatment of nosocomial C. difficile infection. The susceptibility observed with both antibiotics in our study suggests antibiotic resistance in CA-CDI is not a predisposing factor. There have been some reports [34] and in vitro data [35] suggestive of antibiotic resistance in nosocomial CDI, but limited data exists for CA-CDI. Furthermore, increasing resistance has been reported in toxigenic C. difficile strains. A study from Spain found resistance to metronidazole and intermediate resistance to vancomycin amongst nosocomial CDI, albeit with none resulting in treatment failure [36]. Ultimately, as antibiotic exposure increases in the CA-CDI population, it will be necessary to further monitor resistance patterns and assess clinical utility of standard antibiotic treatment regimens.

Molecular characterization by PFGE analysis and WGS demonstrated the predominance of the NAP1 ribotype in CA-CDI. The NAP1 hypervirulent strain has been associated with increased toxin production within animal models and increased antibiotic resistance, in particular to fluoroquinolones [37]. Reports exist of increased detrimental clinical outcomes including intensive care admission, requirement for colectomy, or death with this strain [38, 39]. Clinical reports of the hypervirulent NAP1 strain have been primarily associated thus far with nosocomial C. difficile infection. The presence of NAP 1 in the community with no recent hospital exposure implies that this strain can enter the nosocomial setting upon admission. Our WGS analysis suggests that CA-CDI isolates are indistinguishable from nosocomial isolates and no unique cluster is especially adapted to the community. CA-CDI essentially mirrors nosocomial isolates at the genomic level in this study. Microbiome analysis has been well elucidated in nosocomial infections, with significant dysbiosis shown in patients following infection [40–42]. Dysbiosis may also be a predisposing factor in the acquisition of CDI not only in the hospital but also in the community.

Limitations of this study include the small sample size and number of patients lost to follow-up when contacted by telephone survey. As well, community clinics are within the vicinity of our hospitals and there is a possibility that these patients had some kind of incidental hospital contact in recent weeks (i.e. less than 12). A study performed on CA-CDI in a rural community where no hospitals are in the vicinity would rule this possibility out. Further limitations include data only acquired from a single centre (at least geographically); limited patient-level data available; observational study design; lack of inclusion of hospitalized patients with CA-CDI; lack of ability to distinguish CDI from C. difficile colonization; and recall or memory with a timespan of months between the occurrence of CDI and the telephone interview. Longitudinal studies may be of value where patients in the community are tested and followed over time in a prospective fashion. Nevertheless, there is a paucity of epidemiological and strain typing data on CA-CDI and these results support a community-nosocomial linkage even in cases traditionally attributed to the community.