Clostridioides difficile can cause a severe, often recurrent diarrheal illness, with an estimated 500,000 infections per year in the United States. Reference Ma, Brensinger, Wu and Lewis1 This disease can be debilitating, and 1 in 6 patients will experience a recurrence within 2 months of their diagnosis. Preventing hospital-onset C. difficile infection (HO-CDI) remains a challenge because C. difficile spores can survive on surfaces for months, and intensive cleaning and disinfection are required to eradicate them. 2–Reference Kramer, Schwebke and Kampf6 Many organizations, including the Agency for Healthcare Research and Quality and the US Centers for Disease Control and Prevention, have prioritized efforts to reduce healthcare transmission of C. difficile, and HO-CDI is a nationally reported marker of hospital quality. 2,7
Healthcare exposures increase the risk of CDI. In prior research, C. difficile spores have been detected at multiple locations in hospital rooms, including the bed, floor, sink, and light switch, despite appropriate disinfection measures. Reference Tarrant, Jenkins and Laird8–Reference Samore, Venkataraman, DeGirolami, Arbeit and Karchmer12 These environmental reservoirs for C. difficile pose a risk to hospitalized patients placed in a room or hospital bed that previously held a patient with C. difficile. Prior studies have described an increased odds ratio of 1.1 to 4.5 for HO-CDI when residing in a room where a prior occupant was diagnosed with C. difficile. Reference Shaughnessy, Micielli and DePestel13,Reference Sood, Truelove and Dougherty14 However, to our knowledge, no studies have analyzed the risk of C. difficile transmission associated with being in a hospital bed that previously held someone with C. difficile. The US Food and Drug Administration received >700 reports of bed mattress covers failing to prevent blood or body fluids from leaking into mattresses between 2011 and 2016, making the bed a particularly concerning reservoir for C. difficile. 15
In this retrospective, observational cohort study, we used novel, real-time bed trackers installed in 2 academic hospitals to determine whether residing in a hospital bed that previously held an occupant with C. difficile (ie, “contaminated bed”) increased the risk of HO-CDI. We then used mediation and interaction analyses to determine whether the relationship between a contaminated bed and HO-CDI was explained or modified by a patient being exposed to a contaminated hospital room (eg, a room where the prior occupant had C. difficile).
Methods
Healthcare setting
We used a real-time, radiofrequency- and infrared-light–based location system (AgileTrac, GE Healthcare, Wauwatosa, WI) installed on the subframe of hospital beds to track the movement of beds at 2 academic hospitals within the same healthcare system in Atlanta, Georgia. Hospital A is a 529-bed, academic–community, tertiary-care hospital with women’s health services. Hospital B is a 751-bed, academic, tertiary-care hospital that performs solid organ and hematopoietic stem-cell transplantation. All hospital rooms are single occupancy. At both study hospitals, the environmental hygiene protocols recommend daily cleaning and disinfecting of hospital rooms using a germicidal cleaner and more comprehensive terminal cleaning that involves wiping and disinfecting all surfaces. Rooms with a patient who tested positive for C. difficile were cleaned using either BruTab (Brulin, Indianapolis, IN; hospital A) or Oxycide (EcoLab, St. Paul, MN; hospital B) and received a terminal clean with ultraviolet (UV) light. Hospital beds remain in the room during cleaning and UV application. Cleaning practices are audited via environment-of-care rounds and direct observation. Mattress covers, used in all hospital beds, are semi-impermeable, and mattresses are not removed from the bed frames during cleaning and disinfection. At both hospitals, standard procedure involves inspecting mattresses annually, and all mattresses are rated for 5 years of use. Nearly all mattresses had been replaced in the 2 years prior to study initiation.
Cohort description
We retrospectively identified all patients in tracked hospital beds from April 1, 2018, to August 31, 2019. We excluded beds for which we could not determine the location by the tracking system for >75% of eligible hospital nights. Additionally, we excluded the labor and delivery specialty beds because they rarely move to other units, and neonatal intensive care unit (ICU) beds, which were not tracked by the system. In total, 2.9% of beds were excluded from the analysis. From the electronic medical record, we extracted data on the patients’ hospital room and unit location throughout the admission, demographics, time at risk for C. difficile (defined as the number of hospital days preceding HO-CDI diagnosis or total length of stay for patients without the outcome), and receipt of antibiotics (other than vancomycin or metronidazole) or proton pump inhibitors (PPIs). We calculated the Elixhauser comorbidity score using International Classification of Disease, Tenth Revision (ICD-10) diagnostic codes.
Exposure and outcome variables
A patient was defined as being exposed to a potentially contaminated bed or room (referred to as “contaminated”) if, within the previous 7 days, they had resided in a hospital bed or a room that held a previous occupant with a positive C. difficile test in the previous 90 days (Fig. 1). The primary study outcome was HO-CDI, defined as a positive C. difficile test in a patient hospitalized for >3 days, in accordance with the National Health Safety Network (NHSN) definition. All C. difficile testing was performed using polymer chain reaction (PCR; Xpert C. difficile PCR assay, Cepheid, Sunnyvale, CA). To allow all beds to be tracked (and possibly exposed) using an initial 90-day run-in period, we only included patients as having the outcome of HO-CDI after July 1, 2018.
Statistical analysis
We compared patients with and without the exposure using χ2 tests for categorical variables and the Wilcoxon-Mann-Whitney or Student t test for continuous variables, where appropriate. Using multivariable logistic regression, we evaluated the association between exposure to a contaminated bed and HO-CDI. Model covariates included time at risk for C. difficile and whether the patient required the ICU (assessed prior to HO-CDI or discharge, whichever occurred first). These covariates were chosen based on our a priori hypotheses and were statistically associated with both being in a contaminated bed (exposure) and HO-CDI (outcome). In a sensitivity analysis, we used the same multivariable model but changed the length of time we considered a bed to be contaminated after the prior occupant’s positive C. difficile test from 90 days to 60, 30, 14, and 7 days, respectively.
In a secondary analysis, we repeated the multivariable logistic regression using contaminated bed as our primary exposure and controlling for the same factors as above (ICU and time at risk). We also performed a counterfactual-based mediation analysis with a 4-way decomposition of the total causal effect of mediation and interaction to assess the contribution of a contaminated room to the relationship between a contaminated bed and HO-CDI. 16 Thus, we divided the total causal effect of contaminated room on HO-CDI into 4 components: (1) a controlled direct effect not due to mediation or interaction, (2) a reference effect due to interaction alone, (3) the part of the total effect due to both interaction and mediation (ie, mediated interaction effect), and (4) the part of the total effect due entirely to mediation alone (ie, pure indirect effect). 16 A mediator is an intermediate variable between the exposure and outcome that directly affects their relationship. Interaction occurs when the combined effect of an exposure and a second variable causes an increased effect on the outcome beyond the effect of either variable individually. Both mediation and interaction can occur simultaneously. All P values <.05 were considered statistically significant. The analysis was performed using SAS version 9.4 software (SAS Institute, Cary, NC).
Ethical approval
The Emory University Institutional Review Board approved this study with a waiver of the patients’ informed consent.
Results
We analyzed 25,032 hospital encounters representing 18,860 unique patients. Approximately half the encounters were with female patients (n = 12,938, 51.7%) and Black patients (n = 13,231, 52.9%). The median age was 61 years (IQR, 47–71) (Table 1). In total, 3,155 stool samples were collected and processed for C. difficile, of which 241 (7.6%) were positive. There were 237 (0.9%) hospital encounters with HO-CDI. A higher Elixhauser comorbidity score, longer time at risk, admission to the ICU or hospital B, and receipt of antibiotics or PPIs were all associated with being exposed to a contaminated bed (Table 1).
Note. IQR, interquartile range; ICU, intensive care unit.
a Data are no. (%) unless otherwise specified.
b P value calculated using χ2 tests for categorical variables and the Wilcoxon-Mann-Whitney or Student t test for continuous variables.
c Includes American Indian or Alaskan Native, Asian, Native Hawaiian or other Pacific Islander, and unknown race.
d Prior to diagnosis of HO-CDI or discharge, whichever occurred first.
e Excluded both intravenous and oral metronidazole and vancomycin.
Patients who resided in a contaminated bed were more likely to have HO-CDI in the unadjusted analysis (odds ratio [OR], 1.8; 95% confidence interval [CI], 1.4–2.3) (Table 2). The median time between the prior occupant’s positive C. difficile test and the subsequent occupant’s HO-CDI diagnosis was 38.5 days (interquartile range [IQR], 22–68.5). The median time between the prior positive occupant’s last day in a contaminated bed and the subsequent occupant’s HO-CDI diagnosis was 34.5 days (IQR, 16.5–64.5). Beds moved among an average of 6.7 rooms during the study period.
Note. OR, odds ratio; CI, confidence interval; ICU, intensive care unit.
a Adjusted for ICU admission and time at risk prior to HO-CDI or discharge, whichever occurred first.
b Defined as having an occupant in the previous 90 days with a positive C. difficile test.
In a multivariable analysis, being exposed to a contaminated bed was significantly associated with HO-CDI (adjusted OR, 1.5; 95% CI, 1.2–2.0) after controlling for time at risk and ICU admission prior to HO-CDI or discharge (Table 2). In a sensitivity analysis using the same multivariable model, we adjusted the time we assumed the bed would remain contaminated for after a positive C. difficile test from 90 days to either 60, 30, 14, or 7 days, respectively. For all periods, exposure to a contaminated bed remained a predictor of HO-CDI (Table 3).
Note. OR, odds ratio; CI, confidence interval; ICU, intensive care unit.
a Adjusted for ICU admission and time at risk prior to HO-CDI or discharge, whichever occurred first.
Most patients (68.4%) who were exposed to a contaminated hospital bed were also exposed to a contaminated hospital room. Exposure to a contaminated room was associated with HO-CDI in both an unadjusted analysis (OR, 1.9; 95% CI, 1.5–2.5) and an adjusted analysis, controlling for the same variables used in the bed model (OR, 1.5; 95% CI, 1.1–1.9). In a secondary analysis, 62% (95% CI, 24%–100%) of the relationship between HO-CDI and exposure to a contaminated bed was due to both mediation and interaction between the room and bed status (Fig. 1 and Supplementary Tables online).
Discussion
In this study of over 25,000 hospital encounters, we demonstrated that residing in a hospital bed that previously had an occupant with C. difficile remained a risk factor for HO-CDI in multivariable analysis. This association persisted even after varying the amount of time we assumed the bed would remain contaminated following the prior occupant’s C. difficile diagnosis. Residing in a hospital room that previously had an occupant with C. difficile is also a risk factor for HO-CDI, and the association between contaminated bed and HO-CDI is in part explained by the fact that most patients in a contaminated bed were also exposed to a contaminated room.
These results are consistent with previous research examining the relationship between a prior room occupant with a multidrug-resistant organism (MDRO) and risk of infection or colonization with a MDRO for the subsequent room occupant. Reference Mitchell, Dancer, Anderson and Dehn17 In 2010, Shaughnessy et al Reference Shaughnessy, Micielli and DePestel13 examined the association between hospital room and CDI and found that those with CDI were more likely to have been in a room where the prior occupant had CDI (hazard ratio 2.35, 95% CI 1.21–4.54). This study was limited to critically ill patients at a single hospital, and their analysis only accounted for whether the immediate prior room occupant had CDI. The analysis did not specify whether the same bed was used between patients. Sood et al Reference Sood, Truelove and Dougherty14 recently described similar increased odds of CDI (OR, 1.27; 95% CI, 1.12–1.44) across multiple hospitals when a patient was exposed to a contaminated room, with an increasing risk of CDI with each day exposed to the contaminated room. Lastly, Freedberg et al Reference Freedberg, Salmasian, Cohen, Abrams and Larson9 found that patients are at risk for CDI when the prior bed occupant received antibiotics. Our study builds on these prior investigations and, to our knowledge, is the first to examine the risk of HO-CDI associated with being in a hospital bed that previously had an occupant with C. difficile.
The odds of developing HO-CDI with exposure to a contaminated bed minimally increased in sensitivity analyses in which we decreased the amount of time a bed was considered contaminated, thus decreasing the time between prior occupant’s positive C. difficile test and the subsequent occupant’s diagnosis of HO-CDI. Estimates for the smaller exposure windows were less precise, however, due to the decreasing number of patients considered exposed to a contaminated bed, although the effect estimate (adjusted OR, 1.5) remained consistent. Overall, the sustained the association between contaminated bed and HO-CDI speaks to the robustness of C. difficile spores and their ability to survive on environmental surfaces for an extended period. This research has important implications for understanding transmission dynamics of C. difficile throughout a hospital and indicates the need for improved cleaning and disinfection protocols of the hospital bed and the surrounding healthcare environment. Real-time tracking of hospital beds also has the potential to aid in outbreak investigations of HO-CDI by identifying connections (eg, shared beds) between patients that traditional epidemiologic methods may miss.
Our study had several limitations. Certain beds may reside in the same units consistently, making those beds more likely to have HO-CDI occupants primarily due to the types of patients occupying those units (eg, hematopoietic stem-cell transplant floors or ICUs). We limited our analysis to beds for which we could identify a tracked location in 75% of the eligible nights. However, we do not believe the excluded beds significantly differ from those included. C. difficile testing during the study period was performed using PCR, which may have led to an overdiagnosis of HO-CDI but is likely a more sensitive marker of potentially contaminated hospital beds. We also did not account for clinical severity, which could affect the degree of spore transmission to the next bed occupant. Similarly, our outcome of HO-CDI did not capture asymptomatic acquisition of C. difficile or episodes of C. difficile that occurred after hospital discharge. However, if hospital bed or room does increase odds of C. difficile acquisition, this limitation should bias our results toward the null. The relationships among the hospital bed, hospital room, and HO-CDI are complex. We assumed that being in a contaminated room is on the causal pathway between contaminated bed and HO-CDI; however, it is possible that the contaminated room is a confounder in this relationship. A contaminated bed may also mediate the relationship between contaminated room and HO-CDI, but our mediation analysis remains valid because we also allowed for interaction. Although our analysis included only single-bedded rooms, we believe that our results are applicable to facilities with semiprivate rooms and that having more beds in a room would increase the risk for all occupants of a contaminated room. Lastly, other factors likely exist that we did not control for, including additional components of the healthcare environment and prevalence of C. difficile in the community.
National, regional, and institutional efforts have aimed to reduce HO-CDI through minimizing inappropriate antibiotic prescribing and improving infection prevention practices. Our findings suggest that there may be transmission of C. difficile from the hospital bed to a patient, even up to 90 days after the original patient was diagnosed with C. difficile. Further studies involving genomic sequencing could more directly link possible bed contamination and developing HO-CDI. Human-factor analyses could also elaborate on the interactions that healthcare personnel have with the hospital room, bed, and other surfaces to determine which parts of the healthcare environment contribute the most to transmission of HO-CDI. New technologies or cleaning and disinfection methods that can better eradicate C. difficile spores from a hospital bed and/or the surrounding healthcare environment may lead to significant reductions in healthcare transmission of C. difficile and decrease rates of HO-CDI.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2023.254
Acknowledgments
The authors acknowledge Stuart Tinker for his help in conceptualizing the study and information regarding the real-time bed-tracking system. We also recognize Victoria Walsh, Jill Holdsworth, and Ahmed Babiker for their contributions to this analysis.
Financial support
This research was supported in part by the National Center for Advancing Translational Sciences of the National Institutes of Health (award no. UL1TR002378 and award no. TL1TR002382 to L.S.W.). It was also supported in part by the Centers for Disease Control and Prevention Prevention Epicenter of Emory and Collaborating Healthcare Facilities [PEACH II] (no. U54CK000601 to J.T.J.).
Competing interests
All authors report no conflicts of interest relevant to this article.