Association between Clostridioides difficile Infection and Multidrug-Resistant Organism Colonization or Infection among Hospitalized Adults: A Case-Control Study

Introduction

Clostridioides (formerly Clostridium) difficile is a major cause of healthcare-associated infections [1,2]. Previous studies have investigated co-infection between Clostridioides difficile infection (CDI) and multidrug-resistant organism (MDRO) exposures, identified common risk factors including antibiotic use and dysbiosis, and compared measures of frequency [3–9]. However, none of the previous studies estimated risk or controlled for predisposing factors. The primary objective of our study was to assess ambidirectional association between MDRO colonization/infection and CDI in hospitalized adult patients through a case-control design. We calculated odds ratios and investigated three predisposing factors – age, antibiotic use, and comorbidities – in the association. A secondary objective was to assess directionality of the association. This study may inform current preventive practice against CDI.

Methods

Study Design

Upon approval by the University of Virginia (UVA) Institutional Review Board, we conducted an ambidirectional case-control study. Each CDI case was matched 1:1 to a control whose negative CDI test date was within five days of the positive CDI test date. The first positive test date within the study period and the first negative test date of each hospitalization within the study period were eligible for case and control matching, respectively. Each patient was eligible to be matched once. We assessed MDRO exposure status from one year before to six months after the CDI. We included adult (18 years or older) inpatients who had tcdB PCR assay tests between May 1 and July 10, 2015. UVA Health System only allowed testing on patients with diarrhoea, defined as three loose stools within 24 hours, and with reasonable suspicion for CDI, including having associated signs and symptoms such as fever, abdominal discomfort, increased white blood cell counts, or age >60 years. Only symptomatic patients were included to minimize detection of colonization. Patients with inconclusive or undocumented tcdB PCR results and controls with a history of CDI were deemed ineligible.

We used Epic, an electronic medical record, as our source of patient data. Both CDI and exposure status were determined by analyzing microbiology lab results, including blood, urine, and other sites for screening, and cultures for methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococci (VRE), carbapenem-resistant Enterobacteriaceae (CRE), and extended spectrum beta-lactamase (ESBL) producing organisms. Infection status was determined based on infectious disease physician notes and antibiotics given. It was assumed that infection occurrence was also indicative of colonization, but not vice versa. Antimicrobial resistance was defined according to Magiorakos et al.’s 2012 study [10] – resistance to at least one antibiotic in three different classes. Intermediate susceptibility did not meet our qualification for resistance.

Statistical Methods

We assessed distribution frequencies of the patient population and used binomial testing with the Z approximation to estimate differences between case and control groups. We then conducted simple and multiple regression, with two multiple regression models controlling for age, antibiotic use, and comorbidity status – one for MDRO colonization and one for infection. We categorised MDRO exposures by timeframe relative to CDI test date and conducted single and multiple regression analyses on each. Timeframes included 1-12 months before CDI testing, within 1 month before or after CDI testing, and 1-6 months after. All analyses were performed using SAS University Edition. We calculated 95% confidence intervals (CI) for each odds ratio (OR). P-values less than 0.05 were deemed statistically significant.

Results

Association between MDRO Exposure and CDI

780 distinct patients had diarrhetic stool submitted for CDI testing. Of the 383 eligible patients, 100 subjects were randomly selected for the study, 50 cases and 50 controls. MDRO exposures, demographic characteristics, and predisposing factors of cases and controls are presented in Table I.

Table I:

Univariate Associations of Clostridioides difficile Infection (CDI) by Risk Factors

Factor	Cases N=50 n (%)	Controls N=50 n (%)	Total N=100 n (%)	P-valuec
Multidrug-Resistant Organism (MDRO)
Colonization
Methicillin-resistant Staphylococcus aureus (MRSA)	17 (34)	9 (18)	26 (26)
Vancomycin resistant enterococcus (VRE)	17 (34)	13 (26)	30 (30)
Carbapenem-resistant Enterobacteriaceae (CRE)	1 (2)	1 (2)	2 (2)
Extended spectrum β-lactamase (ESBLs) producing Gram-negative bacteria	3 (6)	4 (8)	7 (7)
Other	8 (16)	1 (2)	9 (9)	0.01*
At least one of these	32 (64)	18 (36)	50 (50)	0.01*
MDRO Infection
MRSA	7 (14)	8 (16)	15 (15)
VRE	5 (10)	4 (8)	9 (9)
CRE	0 (0)	0 (0)	0 (0)
ESBL	3 (6)	3 (6)	6 (6)
Other	8 (16)	1 (2)	9 (9)	0.01*
At least one of these	20 (40)	12 (24)	32 (32)
Gender
Male	25 (50)	27 (54)	52 (52)	0.69
Female	25 (50)	23 (46)	48 (48)	0.69
Age Group
Less than 65 yr	28 (56)	30 (60)	58 (58)	0.69
65 yr or More	22 (44)	20 (40)	42 (42)	0.69
Median Age	63.5	60.0	60.0	0.68
Antibiotic Use within 6 weeks prior to CDI result
None	6 (12)	8 (16)	14 (14)	0.56
1 antibiotic class	7 (14)	9 (18)	16 (16)	0.59
2 or more antibiotic classes	37 (74)	33 (66)	70 (70)	0.38
Antibiotic Exposure within 6 weeks prior to CDI result
Penicillin	6 (12)	8 (16)	14 (14)	0.56
Beta-lactam/Beta-lactamase Inhibitor	8 (16)	10 (20)	18 (18)	0.60
Carbapenem	7 (14)	10 (20)	17 (17)	0.42
Cephalosporin	34 (68)	28 (56)	62 (62)	0.22
Macrolide	7 (14)	12 (24)	19 (19)	0.20
Fluoroquinolone	12 (24)	10 (20)	22 (22)	0.63
Metronidazole	16 (32)	20 (40)	36 (36)	0.40
Intravenous Vancomycin	21 (42)	28 (56)	49 (49)	0.16
Hospitalization within 12 weeks prior to CDI test
None	12 (24)	17 (34)	29 (29)	0.27
At least one	38 (76)	33 (66)	71 (71)	0.27
ICU Admission within 12 weeks prior to CDI testa
None	29 (58)	29 (58)	58 (58)	1.00
At least one	21 (42)	21 (42)	42 (42)	1.00
Operation within 12 weeks prior to CDI test
None	31 (62)	30 (60)	61 (61)	0.84
Surgery	17 (34)	17 (34)	34 (34)	1.00
Any Invasive Operation (surgery or procedure)	19 (38)	20 (40)	39 (39)	0.84
Concurrent Infection Status
None	25 (50)	23 (46)	48 (48)	0.69
Any	25 (48)	27 (54)	52 (52)	0.69
Charlson Comorbidity Index (CCI)b
0	13 (26)	12 (24)	25 (25)	0.82
1-2 (low)	11 (22)	19 (38)	30 (30)	0.81
3-4 (moderate)	18 (36)	7 (14)	25 (25)	0.01*
>=5 (high)	8 (16)	12 (24)	20 (20)	0.32
Median CCI	3	2	2	0.31

^aWithin 12 weeks before, but more than 24 hours before CDI test date

^bCalculated using Mary E. Charlson’s CCI method and defined categorised indices (1987)

^cPearson Chi-Square test and Mann-Whitney U test, as appropriate; Values of p<0.05 were considered statistically significant

^*p<0.05

The odds of a CDI diagnosis were above three times as high for subjects with at least one MDRO colonization 12 months before to 6 months after the CDI test date (OR 3.16 CI 1.4, 7.15), and approximately two times as high for subjects with at least one MDRO infection during that timeframe (OR 2.11 CI 0.89, 4.99). Odds ratios by type of MDRO were insignificant, likely due to small sample size. MDROs that did not qualify as MRSA, VRE, CRE, or ESBL, but were determined to be resistant to at least one antibiotic in three different classes. Associations between CDI and both MDRO colonization and infection remained positive when adjusting for age over 65, exposure to multiple antibiotic classes, and comorbidity status, through multiple logistic regression. The odds ratio for the former increased to 3.38 (CI 1.38, 8.24) with maintained statistical significance, and that for the latter remained near two (OR 2.03 CI 0.85, 4.85). These associations were stronger than those with age, antibiotic use, and comorbidity status.

Association Directionality

To assess directionality of the association between MDRO colonization/infection and CDI, we conducted regression analysis for each exposure timeframe relative to the CDI event: 1-12 months before CDI testing, within 1 month before or after testing, and 1-6 months after testing (Table II). Simple regression analysis identified that 1-12 months before CDI testing was the only exposure period with significant association with CDI, for both MDRO colonization and infection. Adjusted analyses confirmed statistically significant associations between CDI and both MDRO colonization and infection 1-12 months before CDI, with ORs near five and above five, respectively (Table II).

Table II:

Analysis of Risk Factors Associated with Clostridioides difficile Infection (CDI) by Exposure Timeframe Relative to CDI Diagnosis

Risk Factor	Simple Regression Model OR (95% C.I.)	Multiple Regression Model OR (95% C.I.)
Multidrug-Resistant Organism (MDRO)
Colonization
At least one before CDI	4.95 (1.30, 18.81)*	4.71 (1.22, 18.12)*
At least one during CDI	1.56 (0.68, 3.56)
At least one after CDI	2.55 (0.62, 10.49)
Age 65 years or More		1.04 (0.45, 2.41)
Two or more antibiotic classes		1.34 (0.49, 3.66)
Charlson Comorbidity Index (CCI)		0.97 (0.80, 1.18)
MDRO Infection
At least one before CDI	5.27 (1.08, 25.78)*	5.03 (1.02, 24.82)*
At least one during CDI	1.00 (0.36, 2.77)
At least one after CDI	1.53 (0.25, 9.59)
Age 65 years or More		1.04 (0.45, 2.40)
Two or more antibiotic classes		1.44 (0.53, 3.92)
CCI		0.97 (0.80, 1.18)

^*p<0.05

Discussion

Although past research of the association between MDRO exposure and CDI assessed correlations and frequencies without estimating risk through measures of association and not controlling for common covariates, their results support those found in this study. Similar to previous studies [8, 9], we found a positive association between MRSA or VRE and CDI. Interestingly, although only detected in 7 subjects, ESBL was more common in controls than cases in our study, which may suggest differential propensity of MDROs to increase risk for CDI. Unique to our study, we found that patients with CDI, independent of age, antibiotic exposure and comorbidity, are likely to also experience colonization and/or infection with MDROs in the timeframe surrounding CDI diagnosis, especially with MDRO colonization preceding CDI. These results indicate that the MDRO risk factor for CDI may be stronger than the covariates that CDI prevention efforts commonly target. Thus, enhanced management of MDRO colonization or infection may have secondary preventive benefits against CDI.

The strengths of our study exist in the rigor of multiple regression analysis and the similarity of case and control groups before adjustment, with regards to common CDI risk factors. The limitations of our study include that the data were retrospective and limited to one health system and the inpatient setting.