Clostridioides difficile Infection Deceleration: What Is Next?

In this issue of Clinical Infectious Diseases, Kar and Freedberg [1] describe 3 phases of Clostridioides difficile infection (CDI): emergence, acceleration, and deceleration. The emergence phase (1935–1993) is characterized by the coming together of 3 critical components: pseudomembranous colitis (PMC), the organism C. difficile, and antibiotic use. The discovery of C. difficile harbored asymptomatically in infants was indicative that infants are the only natural human hosts of C. difficile under the One Health concept. The association of PMC with clindamycin led to the discovery of C. difficile and its toxins as the causative agent. Not only was the cause of CDI found, but an effective therapy, oral vancomycin, was discovered almost simultaneously, as were the diagnostic cell cytotoxicity assay and anaerobic culture of C. difficile. Thus, by 1979 all of the diagnostic and treatment resources were presumably discovered and in retrospect may have led to underestimation of the importance of CDI in hospitals. Undiscovered at the time was the hospital epidemiology of CDI and the importance of antibiotic resistance in C. difficile to the transmission of CDI in the hospital. In retrospect, we now have evidence from molecular typing of the epidemiology of CDI in one hospital over 10 years from 1982 through 1991 [2]. Restriction endonuclease analysis (REA) typing on 696 CDI isolates identified 174 unique REA types over 10 years. CDI rates ranged from 3.2 to 9.9 cases per 1000 discharges and were significantly higher in 1982, 1983, 1985, and 1991 than in other years (mean, 122 vs 70 cases; P = .002). Large clusters (≥10 CDI cases in consecutive months) were caused by REA types B1 and B2 (RT053) in 1982 and 1983, and B1 in 1985, and accounted for 35%–71% of all isolates in those years. B1 and B2 differed by only a single REA band but evolved independently. Both were clindamycin resistant and illustrate how antibiotic-resistant strains transmit in hospital and produce outbreaks layered on a background of multiple different C. difficile strains. We do not know how generalizable this single-hospital experience was, but it was a harbinger of multihospital outbreaks of clindamycin-resistant REA J group (RT001) of CDI in 1989–1992 [3]. Parenthetically, the subsequent epidemic BI/NAP1/027 strain was identified in 1984, 1988, and 1990 but did not transmit and was not fluoroquinolone resistant. The acceleration phase (1993–2010) is well described by Kar and Freedberg [1] and was first documented in the United States by McDonald et al [4] on isolates obtained from outbreaks in 2000–2003 in 6 states and designated BI/NAP1. High-level fluoroquinolone resistance, absent in earlier isolates, was now documented. The remarkable and unexplained ability of this strain (also designated RT027) to spread internationally was documented in 2 closely related lineages: FQR1 is 99% probable to have originated in the United States; and FQR2, which accounts for the majority of isolates and multiple transmissions to the United Kingdom and European Union, is 59% probable of US origin and 33% probable of Canadian origin [5]. The current deceleration phase (2010–present) reflects not simply a reduction in CDI, but the uncovering of endemic strains that filled the vacuum left by the epidemic, RT027 lineage. The decline of RT027 mirrors what occurred when clindamycin use declined and the clindamycin-resistant REA group J (RT001) strain prevalence decreased [3]. Multiple studies revealed that the incidence of REA group BI (RT027) decreased in response to a decrease in fluoroquinolone usage, whereas the emerging endemic groups showed no such association. Critically, the lineages filling that vacuum share no dominant antimicrobial resistance signatures and are not new arrivals. Endemic REA groups Y, G, and L were never the dominant strain, yet were present consistently throughout 10 years in one hospital [2]. We postulate that this may be the CDI epidemiology prototype during the current deceleration phase. Whole genome sequencing of isolates from 482 European hospitals confirmed that RT027 and RT001/072 revealed tight within-country clustering consistent with hospital-to-hospital transmission, whereas endemic RT014, RT020, RT015, RT002, and RT078 showed no clustering, more consistent with community dissemination [6]. This does not mean that endemic strains cannot develop significant resistance. RT106, currently fluoroquinolone susceptible in the United States, acquired high-level fluoroquinolone resistance in England, as did RT078 in the Netherlands in humans and pigs. This serves as a reminder that endemic strains carry epidemic potential. The changing patient profile reveals the scale of the colonization reservoir. The contemporary patient described by Kar and Freedberg is younger, community-dwelling, and often minimally symptomatic—a sharp contrast with the elderly, hospitalized, antibiotic-exposed prototype of the acceleration phase. Community-acquired CDI (CA-CDI) now accounts for 53% of cases, and approximately 40% of those patients lacked any antibiotic exposure in the prior 12 weeks. Asymptomatic colonization has become a largely unmeasured phenotype. Longitudinal genomic surveillance in a US medical intensive care unit revealed 5.9% admission prevalence of toxigenic carriage but only 1% cross-transmission among admission-negative patients; however, carriers faced a 24-fold greater risk of healthcare-associated CDI (HA-CDI), implicating an unrecognized community reservoir rather than hospital transmission [7]. Nucleic acid amplification testing (NAAT) has made potential colonization readily visible. The epidemiological response to this newly visible colonization reservoir matters. Treating asymptomatic carriers is not indicated and can cause measurable harm. Oral vancomycin does not produce long-term clearance; colonization rates at 2 months are not significantly lower than placebo [8]. More fundamentally, a randomized controlled trial in NAAT-positive, toxin-negative patients demonstrated that the microbiome disruption produced by oral vancomycin is comparable in magnitude to that produced by ceftriaxone [9]. The shift toward NAAT-based testing simultaneously expanded our epidemiological understanding of C. difficile but also increased incidence trends during the acceleration phase. The introduction of NAAT screening followed by toxin testing algorithm revealed that asymptomatic colonization or clinically milder CDI (as measured by NAAT-positive, toxin-negative test results) constitute the majority of positive tests, which are currently unreported as National Healthcare Safety Network HA-CDI. Facilities transitioning to multistep NAAT screening algorithms showed immediate rate declines that reflect algorithmic reclassification as much as epidemiological change. The true deceleration in HA-CDI remains unclear. The boundaries between HA-CDI and CA-CDI are increasingly difficult to sustain epidemiologically. Outpatient settings such as infusion centers, dialysis units, and urgent care centers manage patients whose CDI risk profiles compare to those of inpatients, yet without equivalent infection prevention infrastructure or stewardship oversight. Per the Centers for Disease Control and Prevention report on antibiotic use in the United States, since COVID-19, outpatient antibiotic prescriptions have surged from 613 to 852 per 1000 persons. Proton pump inhibitors (PPIs) disrupt the gut microbiome, and prescription PPI use was 8.6% of US adults as of 2018, exceeding 15% in older adults [10]. Preparing for a different kind of CDI requires community-based surveillance, stewardship that reaches outpatient prescribers, and diagnostic methods that distinguish infection from the colonization reservoir. Four decades of CDI epidemiology have been organized around HA-CDI and antimicrobial resistance, a framework appropriate for the acceleration phase but increasingly insufficient for the deceleration phase that Kar and Freedberg describe [1]. We postulate that contemporary endemic isolates lack antibiotic resistance (although they could become resistant) and colonize readily but do not induce high-level hospital transmission. Meeting this challenge demands surveillance capable of capturing community acquisition and asymptomatic colonization and epidemiological systems that reflect where C. difficile currently circulates, not simply where it has historically been easiest to measure. Defining the factors beyond antimicrobial resistance that enable persistent colonization will be essential for developing interventions targeting these contemporary C. difficile endemic lineages. Financial support. This work was supported by the US Department of Veterans Affairs Career Development Award (grant number IK2BX005609) to A. M. S.