Abstract
In this retrospective cohort from three Missouri hospitals from January 2017 to August 2020, hospital-onset Clostridioides difficile infections (CDI) were more common during the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic at the tertiary care hospital. Risk factors associated with hospital-onset CDI included year of hospitalization, age, high-risk antibiotic use, acid-reducing medications, chronic comorbidities, and SARS-CoV-2 infection.
Keywords: SARS-CoV-2, COVID-19, Clostridoides difficile, hospital-acquired infections, hospital-onset CDI
Introduction
Data on rates of hospital-onset Clostridoides difficile infections (HO-CDIs) during the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic are mixed.1, 2 This study’s objective was to determine the relationship between SARS-CoV-2 infection and HO-CDI development utilizing a historical and a SARS-CoV-2-era cohort at three hospitals.
Methods
We assembled a retrospective cohort of patients aged ≥18 years admitted between 1/1/2017-8/31/2020 for ≥48 hours at three Missouri hospitals (one large tertiary care referral and two community hospitals). Patients had to be discharged by 9/30/2020 for complete capture of admission information. We excluded psychiatric, rehabilitation, obstetrics/gynecology, or hospice admitting services due to low HO-CDI risk. Data were collected from the BJC HealthCare Informatics database. Washington University School of Medicine Institutional Review Board approved this study with a waiver of informed consent.
HO-CDI was defined as: a positive laboratory test from a specimen collected ≥48 hours after admission, and HO-CDI date was defined as first positive test result collection date during admission. The TOX A/B II (TechLab, Blacksburg, VA) test was used at all hospitals. Admissions in which a positive HO-CDI was identified <48 hours into the admission were excluded.
Demographics collected included sex, race, age, and payor. Comorbidities were defined using ICD-10-CM (International Classification of Diseases, Tenth Revision, Clinical Modification) diagnosis codes based on Elixhauser classification;3, 4 diagnoses indicating leukemia and bone marrow/peripheral stem cell transplant were captured. Procedures/surgeries per ICD-10-PCS codes included bone marrow/peripheral stem cell transplant, colorectal/small bowel surgery, dialysis, mechanical ventilation, and indwelling urinary catheterization. Feeding tube was defined using procedure and diagnosis codes. All inpatient medications orders were captured. Antibiotic risk for HO-CDI was defined as: high-risk (cephalosporins, clindamycin); low-risk (aminoglycosides, aztreonam, beta-lactamases, carbapenems, colistin, daptomycin, fosfomycin, macrolides, nitrofurantoin, oxazolidinones, penicillins, piperacillin-tazobactam, quinupristin-dalfopristin, rifampin, sulfamethoxazole-trimethoprim, tetracyclines, tigecycline).5 Fluoroquinolones, high-risk antibiotics, and low-risk antibiotics were categorized hierarchically: high-risk antibiotic, else low-risk antibiotic, else fluoroquinolone.
Modified Acute Physiology and Chronic Health Evaluation (APACHE) II score (all items except Glasgow coma scale)6 was calculated. SARS-CoV-2 testing during admission and/or diagnosis code (ICD-10-CM: U07.1) captured SARS-CoV-2 diagnosis.
Procedures, non-antibiotic medications, and SARS-CoV-2 status were captured for the entire admission for those without HO-CDI, and captured through calendar day before HO-CDI for admissions with HO-CDI. Antibiotics were captured throughout admission for those without HO-CDI, and captured through two calendar days before HO-CDI for admissions with HO-CDI.
Due to infrequent coincident SARS-CoV-2 and HO-CDI at the community hospitals, univariate/multivariable analyses were limited to hospital 1. For univariate analysis, chi-square tests examined risk factors for HO-CDI. Variables selected for the multivariable model were those identified in the literature as risk factors for CDI or with clinical/biological plausibility and p <0.05 in univariate analysis. We assessed multicollinearity by examining tolerance values to endure independence of explanatory variables. A generalized estimating equations model was created to estimate adjusted odds ratios accounting for admissions clustering within a patient; backward selection was used with cutoff of p<0.05 for model retention. Analyses were performed in SAS v9.4 (Cary, NC).
Results
The cohort included 220,444 patients ≥18 years of age after excluding patients admitted to low-risk HO-CDI services (n=33,796) and admissions with CDI within 48 hours (n=552). Patients were predominantly white (64.4%), and female (50.5%); median modified APACHE II score was 6 (interquartile range 4–9). Table 1 shows the cohort’s descriptive statistics.
Table 1.
Characteristics of the Cohort of Admissions (2017-2020) Among Patients at Three St. Louis Hospitals Admitted for ≥ 48 Hours and Eligible for Hospital-onsetClostridioides difficileInfection.
| Variable | Value |
Hospital 1 |
Hospital 2 |
Hospital 3 |
|||
|---|---|---|---|---|---|---|---|
| Total | CDI n (%) |
No CDI n (%) |
CDI n (%) |
No CDI n (%) |
CDI n (%) |
No CDI n (%) |
|
| Total number of admissions | 656 | 132,595 | 83 | 38,804 | 91 | 48,215 | |
| Age, median (IQR) | 64 (55, 71) | 61 (48, 70) | 68 (59, 79) | 63 (51, 74) | 76 (66, 85) | 70 (58, 80) | |
| Male | 397 (60.5) | 69,508 (52.4) | 37 (44.6) | 17,796 (45.9) | 50 (54.9) | 21,408 (44.4) | |
| Modified APACHE II score | 1st quartile (lowest values) | 87 (13.3) | 27,448 (20.7) | 6 (7.2) | 6,744 (17.4) | 12 (13.2) | 11,201 (23.2) |
| 2nd quartile | 113 (17.2) | 30,958 (23.3) | 9 (10.8) | 9,317 (24.0) | 17 (18.7) | 13,132 (27.2) | |
| 3rd quartile | 165 (25.2) | 38,801 (29.3) | 19 (22.9) | 11,657 (30.0) | 30 (33.0) | 13,730 (28.5) | |
| 4th quartile (highest values) | 291 (44.4) | 35,388 (26.7) | 49 (59.0) | 11,086 (28.6) | 32 (35.2) | 10,152 (21.1) | |
| Race | White | 494 (75.3) | 88,586 (66.8) | 27 (32.5) | 12,411 (32.0) | 77 (84.6) | 40,473 (83.9) |
| Black | 138 (21.0) | 39,793 (30.0) | 52 (62.7) | 25,695 (66.2) | 13 (14.3) | 6,659 (13.8) | |
| Other/ missing | 24 (3.7) | 4,216 (3.2) | 4 (4.8) | 698 (1.8) | 1 (1.1) | 1,083 (2.2) | |
| Medicaid/self-pay | 174 (26.5) | 43,413 (32.7) | 28 (33.7) | 15,839 (40.8) | 15 (16.5) | 7,237 (15.0) | |
| Year of admissiona | 2017 | 141 (21.5) | 34,314 (25.9) | 29 (34.9) | 9,955 (25.7) | 19 (20.9) | 12,554 (26.0) |
| 2018 | 192 (29.3) | 36,295 (27.4) | 28 (33.7) | 10,917 (28.1) | 38 (41.8) | 13,088 (27.1) | |
| 2019 | 199 (30.3) | 38,496 (29.0) | 19 (22.9) | 11,149 (28.7) | 30 (33.0) | 14,431 (29.9) | |
| 2020 | 124 (18.9) | 23,490 (17.7) | 7 (8.4) | 6,783 (17.5) | 4 (4.4) | 8,142 (16.9) | |
| Admission with COVID-19b | 8 (8.7) | 745 (4.7) | 2 (33.3) | 483 (10.6) | 0 (0.0) | 286 (5.5) | |
Abbreviations: COVID-19, coronavirus disease 2019; CDI, C. difficile infection; SD, standard deviation.
Admissions in 2020 were only through an admit date of August 31, 2020.
Defined as during the admission, or the day before hospital-onset C. difficile infection or earlier, if applicable. Percentage of COVID admissions is reported using admissions during the COVID-19 era (admission date March 12, 2020 to August 31, 2020).
In 16,976 admissions, there was ≥1 C. difficile test, with 830 positive assays (4.9% positivity): 656 at hospital 1, 83 at hospital 2, and 91 at hospital 3 (Table 1). During the pandemic, there were no patients with SARS-CoV-2 and HO-CDI at hospital 3, 2 at hospital 2, and 8 at hospital 1.
Due to infrequent HO-CDIs at the community hospitals, we applied our multivariable model only to hospital 1. Univariate risk factors for HO-CDI are in eTable 1. Independent risk factors associated with HO-CDI in the multivariable model are in Table 2. Year of hospitalization was associated with increased risk of HO-CDI. Compared to 2017, year 2020 admissions with SARS-CoV-2 had an odds ratio (OR) for HO-CDI of 2.19 (95% CI 1.07, 4.48), while year 2020 admissions without SARS-CoV-2 had an OR of 1.20 (95% CI 0.93, 1.55).
Table 2.
Multivariable Model of Significant Risk Factors for Hospital-onsetClostridioides difficileInfection at Hospital 1.
| Variable | Value | OR (95% CI) |
|---|---|---|
| Year and COVID-19a admission status | 2017 | 1.00 |
| 2018 | 1.34 (1.07, 1.68) | |
| 2019 | 1.28 (1.02, 1.60) | |
| 2020, admission without COVID-19 | 1.20 (0.93, 1.55) | |
| 2020, admission with COVID-19 | 2.19 (1.07, 4.48) | |
| Age | 18–45 years | 1.00 |
| 46–55 years | 1.38 (1.01, 1.88) | |
| 56–65 years | 1.70 (1.30, 2.22) | |
| 66–75 years | 1.61 (1.21, 2.13) | |
| 76+ years | 1.74 (1.27, 2.38) | |
| Male | 1.22 (1.04, 1.45) | |
| Modified APACHE II score | 1st quartile (lowest values) | 1.00 |
| 2nd quartile | 1.19 (0.89, 1.59) | |
| 3rd quartile | 1.32 (1.01, 1.74) | |
| 4th quartile (highest values) | 1.83 (1.40, 2.40) | |
| Antibiotic useb | No use of these antibiotics | 1.00 |
| Fluoroquinolone | 0.39 (0.13, 1.18) | |
| Low risk antibiotic | 1.16 (0.85, 1.59) | |
| High risk antibiotic | 1.60 (1.26, 2.03) | |
| Proton pump inhibitor or H2 antagonista | 1.68 (1.34, 2.10) | |
| Congestive heart failure | 1.31 (1.09, 1.58) | |
| Depression | 1.25 (1.04, 1.52) | |
| Leukemia | 4.64 (3.68, 5.85) | |
| Liver disease | 1.56 (1.24, 1.98) | |
| Lymphoma | 2.61 (2.00, 3.39) | |
| Paralysis | 1.68 (1.30, 2.16) | |
| Pulmonary circulation disease | 1.66 (1.18, 2.34) | |
| Feeding tube / gastrostomya | 2.01 (1.51, 2.67) | |
| Bone marrow/stem cell transplanta | 8.10 (5.94, 11.04) | |
| Mechanical ventilationa | 2.54 (2.03, 3.19) |
b Antibiotic use was categorized in hierarchical fashion based on use of a high-risk antibiotic, else a low-risk antibiotic, else a fluoroquinolone. Antibiotic use defined as during the admission, or two days before hospital-onset C. difficile infection or earlier, if applicable. Antibiotics at high risk for CDI were cephalosporins and clindamycin; antibiotics at low risk for CDI were aminoglycosides, aztreonam, beta-lactamases, carbapenems, colistin, daptomycin, fosfomycin, macrolides, nitrofurantoin, oxazolidinones, penicillins, piperacillin-tazobactam, quinupristin-dalfopristin, rifampin, sulfamethoxazole-trimethoprim, tetracyclines, and tigecycline.
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; CI, confidence interval; COVID-19, coronavirus disease of 2019; OR, odds ratio.
Defined as during the admission, or the day before hospital-onset C. difficile infection or earlier, if applicable.
Increasing age and modified APACHE-II score were risk factors for HO-CDI. High-risk antibiotic use and acid-blocking agent use during hospitalization were associated with HO-CDI, but low-risk antibiotics and fluoroquinolones were not (Table 2). Comorbidities were also risk factors for HO-CDI (Table 2).
Discussion
The current study included adult patients hospitalized for ≥48 hours at three Missouri hospitals to identify risk factors for HO-CDI during the early pandemic. In contrast to larger studies,1, 2 we found that at a large, tertiary referral hospital, SARS-CoV-2 was associated with greater risk of HO-CDI. For the two smaller community hospitals in our study, only 2 hospital-onset CDI cases were identified during the pandemic, and therefore we were unable to include them in the model. In addition, the patient populations at the community hospitals were significantly different than those admitted to hospital 1, such as more patients at hospital 1 having leukemia, lymphoma, or a bone marrow transplant also precluding combining data from all three hospitals for analyses (data not shown).
Multiple other risk factors (besides SARS-CoV-2) for HO-CDI were elucidated, many previously described, including age, high-risk antibiotic use, and acid-blocking agent use.7, 8, 9 Particular comorbidities such as leukemia and lymphoma, were associated with increased risk of HO-CDI, as previously demonstrated.7, 10
Our data suggests some patients may have increased risk for HO-CDI as a result of the SARS-CoV-2 pandemic, which may be affected by hospitals’ patient case-mix.10
Our study is limited by the small number of HO-CDI cases in the early pandemic. It is possible a larger dataset might find differences in risk factors for HO-CDI at the community hospitals. In addition, it is possible that risk factors for hospital-onset CDI are different between community and tertiary-care hospitals, something we were unable to capture due to infrequent HO-CDI at the community hospitals.
In conclusion, during the early pandemic, SARS-CoV-2 was associated with an increased risk of HO-CDI in a tertiary care referral hospital.
Potential Conflicts of Interest
All authors report no potential conflicts of interest.
Acknowledgments
None
Financial disclosure
This study was funded by CDC grant 75D30120C09598 to JPB. The content is solely the responsibility of the authors and does not necessarily represent the official view of the CDC. JHK is supported by the National Institute of Allergy and Infectious Diseases, National Institutes of Health (grant 1K23AI137321 to J. H. K.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Key points
-
–
Using generalized estimating equations models, Clostridioides difficile infections were more common among patients with SARS-CoV-2 infections at a tertiary care hospital
Footnotes
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.ajic.2023.05.015.
Appendix A. Supplementary material
Supplementary material
References
- 1.Evans M.E., Simbartl L.A., Kralovic S.M., et al. Healthcare-Associated Infections in Veterans Affairs Acute and Long-Term Healthcare Facilities During the Coronavirus Disease 2019 (COVID-19) Pandemic. Infect Control Hosp Epidemiol. 2022:1–24. doi: 10.1017/ice.2022.93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Granata G., Petrosillo N., Al Moghazi S., et al. The burden of Clostridioides difficile infection in COVID-19 patients: A systematic review and meta-analysis. Anaerobe. 2021 doi: 10.1016/j.anaerobe.2021.102484. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Elixhauser A., Steiner C., Harris D.R., Coffey R.M. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27. doi: 10.1097/00005650-199801000-00004. [DOI] [PubMed] [Google Scholar]
- 4.HCUP Elixhauser Comorbidity Software. Healthcare Cost and Utilization Project (HCUP). us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp. Accessed December 21, 2020.
- 5.Dubberke E.R., Yan Y., Reske K.A., et al. Development and validation of a Clostridium difficile infection risk prediction model. Infect Control Hosp Epidemiol. 2011;32(4):360–366. doi: 10.1086/658944. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Knaus W.A., Draper E.A., Wagner D.P., Zimmerman J.E. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13(10):818–829. [PubMed] [Google Scholar]
- 7.Dubberke E.R., Reske K.A., Yan Y., Olsen M.A., McDonald L.C., Fraser V.J. Clostridium difficile--associated disease in a setting of endemicity: identification of novel risk factors. Clin Infect Dis. 2007;45(12):1543–1549. doi: 10.1086/523582. [DOI] [PubMed] [Google Scholar]
- 8.Thomas C., Stevenson M., Riley T.V. Antibiotics and hospital-acquired Clostridium difficile-associated diarrhoea: a systematic review. J Antimicrob Chemother. 2003;51(6):1339–1350. doi: 10.1093/jac/dkg254. [DOI] [PubMed] [Google Scholar]
- 9.D'Silva K.M., Mehta R., Mitchell M., et al. Proton pump inhibitor use and risk for recurrent Clostridioides difficile infection: a systematic review and meta-analysis. Clin Microbiol Infect. 2021 doi: 10.1016/j.cmi.2021.01.008. [DOI] [PubMed] [Google Scholar]
- 10.Dubberke E.R., Reske K.A., Olsen M.A., et al. Risk for Clostridium difficile Infection After Allogeneic Hematopoietic Cell Transplant Remains Elevated in the Postengraftment Period. Transplant Direct. 2017;3(4) doi: 10.1097/TXD.0000000000000662. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary material