Longitudinal genomic surveillance of carriage and transmission of Clostridioides difficile in an intensive care unit

Abstract

Despite enhanced infection prevention efforts, Clostridioides difficile remains the leading cause of healthcare-associated infections in the United States. Current prevention strategies are limited by their failure to account for patients who carry C. difficile asymptomatically, who may act as hidden reservoirs transmitting infections to other patients. To improve the understanding of asymptomatic carriers' contribution to C. difficile spread, we conducted admission and daily longitudinal culture-based screening for C. difficile in a US-based intensive care unit over nine months and performed whole-genome sequencing on all recovered isolates. Despite a high burden of carriage, with 9.3% of admissions having toxigenic C. difficile detected in at least one sample, only 1% of patients culturing negative on admission to the unit acquired C. difficile via cross-transmission. While patients who carried toxigenic C. difficile on admission posed minimal risk to others, they themselves had a 24-times greater risk for developing a healthcare-onset C. difficile infection than noncarriers. Together, these findings suggest that current infection prevention practices can be effective in preventing nosocomial cross-transmission of C. difficile, and that decreasing C. difficile infections in hospitals further will require interventions targeting the transition from asymptomatic carriage to infection.

Fig. 1. Flow diagram of study enrollment and inclusion criteria.

a, Overall study. FMD, fibromuscular dysplasia. b, Admission-level analyses. c, Strain-level analyses.

Fig. 2. C. difficile strain diversity in the overall patient population and between epidemiologically linked carriers.

a,b, ST diversity shown cumulatively during the study period (a), colored according toxigenic (red) and non-toxigenic strains, and longitudinally by month restricted to toxigenic strains only (b), with each ST indicated by color; STs are ordered from most common (bottom) to least common (top) and the proportion of samples from which toxigenic C. difficile was detected each month is depicted in the top line chart. For pairs of isolates of the same ST, the number of SNVs was calculated and evaluated in the context of epidemiological data to identify thresholds for identifying high-confidence transmission linkages. c, Genomic distance between isolates within patients (red) compared to between patients (blue); 95% of within-patient pairs were related within two SNVs. d, Genomic distance between isolates from patients with overlapping ICU admissions (red) versus none (blue); isolates from overlapping admissions were enriched at 0–1 SNVs. e, Genomic distance between isolates from patients ever hospitalized in the same single-patient room (red) versus never (blue); no association was observed between hospitalization in the same room and SNV distance.

Fig. 3. Interrogation of genomic linkages involving patients with culture-based acquisition of toxigenic C. difficile.

a, Genetic distance between pairs of isolates from unique patients within the study (red) versus between study isolates and isolates from a geographically distinct collection (blue) among STs implicated in genetically linked acquisitions; ST8 shows patients commonly linked within two SNVs across geographical sites. b, Culture-based acquisitions of toxigenic C. difficile and their genomic linkages to other patient isolates. The gray boxes indicate ICU admission. The connecting lines indicate genomic linkages within two SNVs. ST8 isolates were linked across temporally distant hospitalizations. c, Culture-based acquisitions of toxigenic C. difficile. The empty circles indicate negative screens and the filled circles indicate positive screens; points are colored according to ST. Culture-based acquisitions genomically linked to another patient with high confidence are indicated by the diamonds; unlinked acquisitions are shown as circles. CDI, defined as a positive clinical PCR test for the toxin B gene, is indicated by a plus sign.

Fig. 4. Potential undetected importations of C. difficile into the ICU.

a, Detection of C. difficile-associated OTU from admission stool samples via 16S rRNA gene sequence analysis, stratified according to whether the patient subsequently acquired toxigenic C. difficile during their admission; C. difficile-associated OTUs were enriched among stool or rectal swab samples of patients who subsequently cultured positive for toxigenic C. difficile during their admission compared to those who did not (P = 4.5 × 10⁻⁶ via a two-sided chi-squared test). b, Dot plot depicting the 14 instances of intermittent toxigenic C. difficile detection among 38 admissions with at least three samples and with toxigenic C. difficile recovered from at least two samples.

Fig. 5. Risk of hospital-onset CDI based on admission colonization status.

a,b, Cumulative hazard curves depicting the unadjusted risk of developing a healthcare-associated CDI during hospitalization among patients who carried toxigenic C. difficile on admission to the ICU (red) (a) compared to those who did not carry any (gray), and patients who carried non-toxigenic C. difficile on admission to the ICU (blue) compared to those who did not carry any (gray) (b). Patients carrying toxigenic strains on admission were at significantly increased risk of hospital-onset CDI compared to admission-negative patients (log-rank test, P = 3 × 10⁻¹³), while those carrying non-toxigenic strains showed no significant difference (log-rank test, P = 0.6). The thick lines indicate the estimated cumulative hazard in the designated group. The vertical lines indicate censored patients. The shaded areas indicate the 95% confidence bands.

Extended Data Fig. 1. Whole-genome phylogeny of sequenced C. difficile isolates.

Species-wide phylogenetic tree with Sequence Type (ST) and toxin status mapped onto the tips. Only the first isolate from a particular ST collected from a single patient is included.

Extended Data Fig. 2. Sequence type identification over time.

Distribution of sequence types of recovered C. difficile isolates over the study period, ordered from most common type (top) to least common type (bottom).

Extended Data Fig. 3. Culture pattern for patients with multiple sequence types detected over time.

Dot plot depicting the 13 ICU admissions where more than one Sequence Type was recovered within a single admission.

Extended Data Fig. 4. Daily C. difficile colonization burden during study period.

Daily frequency of C. difficile detection in the ICU colored by toxigenic (red) vs. non-toxigenic (blue). Toxigenic C. difficile was recovered from at least one patient on 186/276 (67.4%) of sampled study days, while non-toxigenic C. difficile was recovered from at least one patient on 119/276 (43.1%) of sampled study days.