Introduction
Clostridioides difficile infection (CDI) is a major concern among inpatients, as it poses huge burdens on health facilities, and is considered one of the most common hospital-acquired infections (HAI). Reference Liu, Monaghan, Yadegar, Louie and Kao1 In recent decades, the epidemiology of CDI has evolved due to the worldwide spread of hyper-virulent strains. Reference Yakob, Riley and Paterson2 This, in addition to data suggesting superiority of oral vancomycin over metronidazole in moderate to severe CDI, Reference Stevens, Nelson and Schwab-Daugherty3 coupled with potential rare neurological toxicities associated with metronidazole use, Reference Daneman, Cheng and Gomes4 resulted in the Infectious Diseases Society of America (IDSA) as well as the European Society of Clinical Microbiology and Infectious Diseases to change their recommendations from metronidazole to oral vancomycin as the initial therapy, even for primary mild CDI. Reference McDonald, Gerding and Johnson5,Reference van Prehn, Reigadas and Vogelzang6
Given the expected rise in usage of oral vancomycin worldwide due to the evolving epidemiology and the revisions in treatment guidelines, Reference Dubberke, Puckett and Obi7 it is crucial to examine all potential repercussions, including the risk of promoting vancomycin-resistant enterococci (VRE) emergence. Vancomycin administered orally is almost exclusively not absorbed, reaching high concentrations in the gut, which is infested with enterococci, specifically among hospitalized high-risk prone patients. Reference Fishbein, Hink and Reske8,Reference Tomas, Mana and Wilson9 VRE is one of the worldwide most significant, high-priority pathogens, as declared by the World Health Organization. 10 The IDSA guidelines hypothesized that the risk of VRE emergence among patients treated with oral (or via enema) vancomycin is negligible. Reference McDonald, Gerding and Johnson5 This was based primarily on one Canadian study published ∼15 years ago, which found that on an “institutional-level,” the increased usage of oral vancomycin (due to an epidemic spread of hyper-virulent strains at that specific facility in the early 2000’s), did not increase the incidence of VRE acquisitions at the hospital. Reference Miller, Bernard, Thompson, Grima and Pepin11 In a “patient-level” study, conducted at the US Department of Veterans Affairs (VA) health system, from 2006-–2016 (mainly pre-guidelines revisions), metronidazole and vancomycin equally impacted the risk for VRE acquisitions, although VRE data was reviewed for only 3 months following the CDI, and patients were not actively screened for VRE. Reference Stevens, Khader and Echevarria12 In a trial conducted at Shamir Medical Center (SMC) (Israel) from 2010 to 2015 (pre guidelines revisions), among 349 CDI patients, oral (or enema) vancomycin treatment was independently associated with later (up to 6 m) VRE acquisitions (aOR = 6.7; 95% CI = 1.1–39.5), despite the study including only 13 post-CDI VRE patients. Reference Zilberman-Itskovich, Youngster and Lazarovitch13 Our primary aim of this study, was to investigate, at a patient level, the possible independent association between the use of oral (w/o enema) vancomycin for CDI, and later (up to 6 mo) acquisitions of VRE, in settings subjected to various proactive VRE screening policies (eg, hospital, long-term care institutions) during the years following guidelines revisions and implementation.
Methods
A retrospective cohort study (2016–2020) was conducted at SMC, among adults (>18 years) with acute primary CDI (per established definition Reference McDonald, Gerding and Johnson5 ), with no prior VRE carriage documentation (ie, prior to discontinuation of current anti-CDI therapy at the index event), admitted either to the Medicine, general Surgery, or adult intensive-care units (ICU) services, ie, the only services who manage adult CDI patients at the hospital. Adults admitted during the study years to these services at SMC were rectally screened for VRE upon admission, and weekly thereafter, according to institutional policy. Patients who died during the index hospitalization were excluded from the primary aim analysis (ie, predictors for VRE acquisition in 6 mo), but were included in the descriptive analysis. Only the first single CDI episode per patient was captured (ie, relapses/recurrences were captured as outcomes). Data was retrieved from all SMC records. In addition, out-of-hospital data pertaining to microbiological VRE results (from long-term care or non-SMC acute care facilities, who have various VRE screening and prevention policies implemented at each institution) were available. Out-of-hospital mortality data (all cause) was extracted from a registry governed by the Ministry of Interior. The study was approved by the institutional ethic committee prior to its initiation.
Testing algorithm for CDI consisted of eligible stool samples Reference Lewis II14 and rapid membrane enzyme initial immunoassay (C. diff quick check complete®; Techlab; Radford, VA, USA), followed by confirmatory genotypic testing whenever the initial stool test was inconclusive (Xpert® C. difficile; Cepheid). Reference Crobach, Planche and Eckert15 Toxin-positive samples were mandatory to define CDI. 16 VRE surveillance samples were processed in accordance with Clinical and Laboratory Standards Institute criteria. Reference Lewis II14 VRE was considered an acquisition if it was isolated from either a clinical or a surveillance sample according to established criteria. Reference Crobach, Planche and Eckert15 The CDI acquisition location, ie, healthcare facility-onset (HO); community-onset, healthcare facility–associated (CO-HCFA); and community-associated (CA), were determined according to IDSA classification. Reference McDonald, Gerding and Johnson5 The CDI severity was analyzed according to the Zar criteria, Reference Zar, Bakkanagari, Moorthi and Davis17 as well as the Horn index. Reference Horn, Sharkey and Bertram18
Statistical analyses were conducted with IBM SPSS 29.0 (2024). The independent predictors for VRE acquisition were queried in univariable analyses (Chi-square, Fisher exact, Mann-Whitney, and student’s t-tests). Parameters with significant association per univariable analyses (p-value≤0.05), were modeled by using logistic regression (backward selection). Variables were tested for confounding (parameters with ≥10% were retained in the model) and collinearity (at each step). A propensity score (PS)-matched analysis, pertaining to the “risk” to receiving treatment with oral (w/o enema) vancomycin as the anti-CDI therapy, was conducted (multivariable forward likelihood regression) to further control for confounders associated with the primary aim analysis. The PS was incorporated into the regression model, and the statistical association between oral (w/o enema) vancomycin treatment and later VRE acquisition was also analyzed by PS-matched pairs regression.
Results
Six hundred and six (606) patients with primary CDI who met the inclusion criteria as depicted above were included during the 5-year study period. The median age was 75 years (IQR = 65–85), 54% were females, and the mean Charlson’s comorbidity index was 6.7±3.2. Reference Charlson, Pompei, Ales and MacKenzie19 There were 452 (75%) patients with HO CDI, 93 (15%) with CO-HCFA CDI, and 61 (10%) with CA CDI. Overall, 354 (58%) patients experienced severe disease, Reference Zar, Bakkanagari, Moorthi and Davis17 and the mean Horn index was 2.4 ± 1. Reference Horn, Sharkey and Bertram18 By day 14, 426 patients (70%) were cured, while 106 patients (23%) experienced recurrence or relapse within six months. Reference McDonald, Gerding and Johnson5 During the index hospitalization, 153 patients (25%) had died, and 227 (38%) had died within 90 days. Ten patients (2%) underwent urgent colectomy. The main anti-CDI regimen included oral (w/o enema) vancomycin in 450 of the patients (77%), while oral and intravenous metronidazole were administered to 170 (29%) and 29 (5%) of the patients, respectively. Fidaxomicin and vancomycin were administered concurrently to two patients. Within 6 months of the index CDI event, 79 patients had acquired VRE.
Four-hundred and fifty-three patients (453, 75%) had survived the index hospitalization, and 65 (14%) of them had acquired VRE in the following 6 months. Predictors for VRE acquisition among survivors of the index hospitalization are depicted in Table 1. Demographics (age, gender) did not differ between groups. Patients who acquired the CDI in the community (CA), had a significantly lower risk of acquiring VRE (Table 1). The morbidity outcomes were significantly worse among patients who later acquired VRE: ie, their median length of stay following the CDI was 12 vs 7 days (P < 0.001), only 77% vs 91% of patients were cured by day 14 (OR = 0.3, CI-95% = 0.2–0.6), and their risk of relapse or recurrence, readmissions in the following 3 months, functional status deterioration, and long-term care facility (LTCF) discharge after being initially admitted from home, were all significantly increased (bottom Table 1). The severity of the acute CDI was also significantly worse among patients who later acquired VRE, as per the Horn Reference Horn, Sharkey and Bertram18 and Zar Reference Zar, Bakkanagari, Moorthi and Davis17 indexes, along with additional established CDI acute illness indices (eg, acute kidney injury, reduced cognition, high fever and high peripheral leucocytes count, reduced serum albumin, and longer duration of diarrhea). Patients who later acquired VRE, were significantly more often on systemic antibiotics on their CDI diagnosis and systemic antibiotics were prescribed significantly more often during the course of their illness (up to day 14 of CDI initiation). Patients with CDI who later acquired VRE, were more often already carriers of additional non-VRE MDROs, were more functionally dependent, and possessed significant additional burden of comorbidities as reflected by elevated Charlson’s scores. Reference Charlson, Pompei, Ales and MacKenzie19 Moreover, this group of patients had recent significant exposure to multiple health settings (ie, LTCF residency, previous hospitalizations in acute-care hospitals, and days in the hospital prior to the CDI initiation), procedures (feeding tubes), and medications (proton-pump inhibitors, carbapenems, systemic vancomycin, and aminoglycosides).
Risk factors for VRE acquisition in 6 months, among patients with acute Clostridioides difficile infection who survived the index hospitalization, Shamir Medical Center, 2016–2020
Note. CDI – Clostridiodes difficile Infection; VRE – Vancomycin-resistant enterococci; MDRO – Multi Drug Resistant Organisms; OR – Odds ratio; CI – Confidence interval; IQR – interquartile range; SD – standard deviation; LTFC – Long term care facility ACH – acute care hospital; PPI – proton pump inhibitors; ICU – intensive care unit; WBC – White blood cell.
a Valid percent: percent after removing missing values from the denominator.
b If the diarrhea and/or other symptoms of Clostridium difficile infection and the stool sample was sent after 71 hours of hospitalization (ie, ≥3 d), or if the patient was hospitalized in an acute care facility (ie hospital) in the past month.
c If the diarrhea and/or other symptoms of Clostridium difficile infection or the stool sample were sent before 72 hours of hospitalization (ie, <3 d), but the patient lives in any long-term care facility, and/or visits regularly a hemodialysis unit, or regularly visit a day-care clinic.
d If the diarrhea and/or other symptoms of Clostridium difficile infection or the stool sample was sent before 72 hours of hospitalization (ie, <3 d), and the patient does not live in a long-term care facility, and/or does not visit regularly a hemodialysis unit, or regularly visit a day-care clinic, and/or was not hospitalized in the past month.
e Staphylococcus aureus resistant to oxacillin (MRSA), vancomycin-resistant enterococci (VRE), Enterobacterales which is resistant to any 3rd or 4th generation cephalosporin (ESBL), Enterobacterales with meropenem MIC > 1 μg/dL (CRE), Acinetobacter baumannii, or Pseudomonas aeruginosa in the past 24 months.
f Any assistance in any activity of daily living (ie, activities of daily living, eg eating/feeding, dressing, ambulation, bathing and showering, bowel and/or bladder management, toilet hygiene)
g Decreased kidney function: (serum creatinine ≥ 1.7mg%) for three or more months.
h Examples include: Bronchiectasis, COPD, asthma, interstitial lung disease.
i Examples include: Rheumatoid arthritis, systemic lupus erythematosus, scleroderma, and vasculitis.
j Patient has or had neurologic disease of any sort including MS (multiple sclerosis), tumor, Parkinson’s, vascular cerebral disease.
k Any of the following: Neutropenia (<500 neutrophils/mm3), steroid use for > 48 hours in the past 30 days, chemotherapy and/or radiotherapy in the past 3 months, immunomodulating agents (eg, anti TNF-α inhibitors, additional monoclonal antibodies agents) in the past 3 months, people living with human immunodeficiency virus/acquired immunodeficiency syndrome, bone marrow or solid organ transplantation.
l If patient was admitted with acute CDI to the index hospitalization, then the time at risk was captured as 0 (zero) days.
m Patient regularly attended an outpatient clinic prior (3 mo) to current hospitalization, Examples: hemodialysis, unit, sub-specialty clinic at least weekly visits.
n Any type of surgery (minor to major) or invasive procedure in the past 6 months: eg, permanent central line insertions, percutaneous endoscopic gastrostomy insertion, ascites paracentesis, percutaneous coronary intervention, abscess drainage.
o Permanent devices include any of the following: chronic urinary catheters (eg, silicon-based catheters), tracheotomies, chronic tunneled central lines (eg, PICC line), orthopedic external fixators, implanted defibrillators, pacemakers, drains of any sort (eg, genitourinary stoma). Prosthetic heart valves, prosthetic joints and urologic or coronary internal stents were not considered a permanent device.
p Noradrenaline, dopamine, or adrenaline were prescribed on day of CDI diagnosis.
q Increase in serum creatinine of ≥0.3 mg/dL or ≥50% of baseline creatinine within 48 hours or urine output of <0.5 mL/kg/hour for >6 hours.
r Patient received systemic non-anti-CDI antibiotics from day 3 to day 14 from CDI diagnosis date.
s Hours to initiation of anti-CDI therapy are defined as the time from obtaining the CDI test from the patient, to the time of administration of the first dose of anti-CDI treatment to the patient.
t Empiric therapy - a regimen that was given from 3 days prior to index sample collection, and up to 3 days (ie, < 3 d) from sample collection date.
u Main therapy - a regimen that was given from 3 days (ie, ≥3) to 14 days following sample collection date.
There were few consolidative main anti-CDI therapeutic regimens (administered from day 3 to 14), which were significantly associated with later VRE acquisition per bivariable analysis, namely oral (w/o enema) vancomycin, fidaxomicin, and oral metronidazole (which was a protecting factor). None of the empiric regimens (administered from -3 to <3 days of positive sample collection) were associated with VRE emergence.
Per multivariable analysis, the only independent predictors for later VRE acquisitions that remained significant in the regression model were 1) main anti-CDI therapy with oral (w/o enema) vancomycin (aOR = 4.3, CI-95% = 1.8-10.4, P < 0.001), 2) recent exposure to systemic vancomycin (aOR = 3.2, CI-95% = 1.7–5.9, P < 0.001), 3) dependent functional status in background (aOR = 2.3, CI-95% = 1.3–4.2, P = 0.006), and 4) diabetes (aOR=1.8, CI-95% = 1.1–3.1, P = 0.04). As described in Methods, a PS was developed in order to control for the multiple potential confounders that might be associated with managing CDI patients on main oral (w/o enema) vancomycin regimen, with the final model depicted in Table 2. After enforcing the PS to the model, the main oral (w/o enema) vancomycin anti-CDI therapy remained an independent predictor for later VRE acquisitions (aOR = 3.6, CI-95%=1.55–8.3, P < 0.01). In matched analysis, which reduced the cohort size to 160 PS matched pairs (n = 320), main oral (w/o enema) vancomycin therapy remained independently associated with later VRE acquisitions (aOR = 4.43, CI-95% = 1.64–120, p < 0.01).
Propensity score of the “risk” to receiving oral (w/o enema) vancomycin for acute Clostridioides difficile infection, Shamir Medical Center, 2016–2020
Note. OR – odds ratio; CI – confidence intervals; CDI – Clostridiodes difficile Infection.
a Massive diarrhea implies over 3 recorded episodes on the date of CDI diagnosis.
Discussion
The recent evolving epidemiology of CDI had resulted in revisions of the treatment guidelines, as issued by professional societies worldwide. Reference McDonald, Gerding and Johnson5,Reference van Prehn, Reigadas and Vogelzang6 In turn, this resulted in an increased usage of oral vancomycin, even for primary mild CDI episodes, Reference Dubberke, Puckett and Obi7 without exploring, on individual “patient-level” controlled analyses, the possible (and theoretically logical) hypothesis that it could promote VRE acquisitions among treated patients.
It is perceived, although never scientifically quantified, that the majority of VRE acquisitions at hospitals follow the mode of horizontal patient-to-patient spread. Reference Mitchell, Dancer, Anderson and Dehn20 This may have led to ecological studies which have tried to explore the association between oral (w/o enema) vancomycin and later VRE emergence at an “institutional-level.” However, the potential selective pressure which could result from the direct application of oral (w/o enema) non-absorbable vancomycin to the patients’ gut, leading to VRE emergence, should have also been trialed among individual patients. In a previous study conducted at SMC during 2010-2015 (349 patients) prior to the revisions in treatment guidelines, while the majority of patients were still managed on metronidazole (61%), we already found a marker for this phenomenon, ie, a significant independent association (albeit with wide confidence intervals), between patients managed with oral (w/o enema) vancomycin as their main anti-CDI regimen, and later VRE acquisitions (aOR = 6.3, CI-95% 1.9–21.6) Reference Zilberman-Itskovich, Youngster and Lazarovitch13 . However, only 13 cases of later VRE acquisitions were reported, and proactive surveillance programs for VRE detection among asymptomatic carriers were not yet recommended and conducted during those years (neither at acute-care hospitals, nor at LTCFs), Reference Zilberman-Itskovich, Youngster and Lazarovitch13 as it was in other studies conducted globally thus far. Reference Stevens, Khader and Echevarria12
This current study was executed at SMC among a larger cohort of inpatients (n = 606), during the following five years (2016–2020), at which treatment guidelines were revised and gradually implemented. The portion of patients managed with vancomycin at SMC had increased substantially (77%), in line with studies performed in other countries. Reference Stevens, Khader and Echevarria12 In order to avoid potential biases a priori with regards to the association between oral vancomycin administered for CDI and later VRE acquisitions, the study was conducted in settings (ie, acute-care hospitals and LTCFs), and at a time frame where active VRE surveillance programs were recommended (although nonuniformly implemented), and more importantly, out-of-SMC data pertaining to VRE acquisitions were available. In addition, this aforementioned association was determined only among adult patients (excluding pediatrics) with unique episodes of primary CDI (avoiding the misclassification of possible recurrence or relapse episodes), and only among survivors of the index hospitalization (avoiding the misclassification of those acutely sickest group of patients who eventually died, to the “favorable outcome group,” ie, patients who did not acquire VRE). As depicted in Table 1, multiple parameters were significantly associated with later VRE acquisition. This outcome was most common among patient groups who were considered to be the sickest, with the worst and most complex background conditions, who had experienced excessive recent healthcare exposures, and who had the most severe CDI acute illness indices. It is important to note that empiric oral (w/o enema) vancomycin administration was not associated with later VRE acquisition and was similar in both groups (17% vs 18%, P = 0.9), while oral (w/o enema) vancomycin as main therapy had a significant association with the primary outcome. Empiric oral vancomycin use is frequently early discontinued, upon receipt of a negative result and/or termination of symptoms. The fact that empiric oral vancomycin treatment was not associated with later VRE acquisitions, while consolidative treatment with oral (w/o enema) vancomycin was independently associated with VRE emergence, could better direct future preventive interventions.
By factoring all the parameters with significant associations to later VRE acquisitions (Table 1), into a multivariable regression model, oral (w/o enema) vancomycin main anti-CDI treatment remained the strongest predictor independently associated with later VRE acquisitions (aOR = 4.3, CI-95% = 1.8–10.4). Moreover, this independent association remained significant after controlling for PS (reflecting the “risk” of receiving anti-CDI oral [w/o enema] vancomycin), both in paired-matched PS analysis (aOR = 4.43, CI-95% = 1.64–120), or following enforcement of the PS parameter to the regression model (aOR = 3.6, CI-95% = 1.55–8.3). This implies that patients who were administered oral (w/o enema) vancomycin as their main treatment for their primary CDI at SMC from 2016 to 2020 and who had survived the index hospitalization, had an at least 1.55-fold increased risk of acquiring VRE in the following six months. This is a significant modifiable threat, which should be analyzed and addressed by other health institutions.
Patients with CDI who acquired VRE had significant increments in acute-illness indices (eg, 68% vs 48% had severe disease per Zar score, Reference Zar, Bakkanagari, Moorthi and Davis17 acute kidney injury, duration [days] of diarrhea), as well as worse morbidity outcomes such as longer hospital stays, lower cure rates, and greater functional status deterioration (Table 1). Reference Katz, Ford, Moskowitz, Jackson and Jaffe21 While those poor outcomes may have been partially attributed to the patients’ statuses and conditions prior to admission (as evidenced by higher Charlson comorbidity indices Reference Charlson, Pompei, Ales and MacKenzie19 , functional dependency Reference Katz, Ford, Moskowitz, Jackson and Jaffe21 , and LTCF residency), it also underscores the serious implications and consequences of VRE acquisition. Reference Sigakis, Posluszny, Maile, Jewell and Engoren22,Reference Fujitani, George, Morgan, Nichols and Murthy23 The commonality of these devastating outcomes among this group of patients further highlights the critical importance of preventing VRE transmission. This study provides an additional measure to potentially prevent VRE acquisitions, by promoting the implementation of stewardship initiatives with regard to usages of oral (w/o enema) vancomycin for primary CDI.
Our study has several limitations, primarily since this was an observational, retrospective, chart-review-based investigation from a single center. This implies that not all eligible patients were screened for VRE (ie, the outcome of interest) as per the institutional policy, at all potential opportunities, and that some of the documentation, for certain parameters and certain patients, were missing or inaccurate. However, there are no reasons to assume that these potential biases negatively impacted either group. It is also important to trial this hypothesis at other centers who use fidaxomicin more often for primary acute CDI, compared to SMC. The long study period (5 years), as well as the lack of association of VRE acquisition to any specific hospitalization division (Table 1), imply that local VRE clusters or outbreaks (which naturally occurred in the hospital during the study’s years), had not seriously impacted the interpretation of the observations for the primary aim analyses.
In conclusion, oral (w/o enema) vancomycin for CDI had a strong significant independent association with later acquisitions of VRE. This could promote local stewardship reassessments and interventions to reduce the usage of oral vancomycin for certain CDI indications, which could then both improve the management for certain acute CDI patients, while also curb the continued emergence and spread of VRE at the hospital.
Acknowledgments
All authors have read and agreed to the submitted version of the manuscript.
Financial support
No financial support was provided relevant to this article.
Competing interests
All authors had declared no conflicts of interest relevant to this article.
