HCW indicates health care worker.
ICER indicates incremental cost-effectiveness ratio; and QALY, quality-adjusted life-year.
eAppendix. Agent-Based Simulation Model
eTable 1. Transmission Parameters Estimates
eTable 2. Discounted QALY Analysis
eTable 3. Probabilistic Sensitivity Analysis Results of Incremental Cost-effectiveness Ratios for 100 000 Runs
eFigure 1. Schematic of the 200-Bed Model Hospital and Possible Agent Movement
eFigure 2. One-Way Sensitivity Analysis of Cost-Saving Interventions
eFigure 3. One-Way Sensitivity Analysis of Cost-effective Interventions
eFigure 4. Incremental Cost-effectiveness of 100 000 Runs of the Probabilistic Sensitivity Analysis
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Barker AK, Scaria E, Safdar N, Alagoz O. Evaluation of the Cost-effectiveness of Infection Control Strategies to Reduce Hospital-Onset Clostridioides difficile Infection. JAMA Netw Open. 2020;3(8):e2012522. doi:10.1001/jamanetworkopen.2020.12522
What is the most cost-effective infection control strategy for reducing hospital-onset Clostridioides difficile infection?
In this economic evaluation study, an agent-based simulation of C difficile transmission at a 200-bed model hospital found 5 dominant interventions that reduced costs and improved outcomes compared with baseline practices, as follows: daily cleaning (the most cost-effective, saving $358 268 and 36.8 quality-adjusted life-years annually), terminal cleaning, health care worker hand hygiene, patient hand hygiene, and reduced intrahospital patient transfers. The incremental cost-effectiveness of implementing multiple intervention strategies quickly decreased beyond a 2-pronged bundle.
The findings of this study suggest that institutions should streamline infection control bundles, prioritizing a small number of highly cost-effective interventions.
Clostridioides difficile infection is the most common hospital-acquired infection in the United States, yet few studies have evaluated the cost-effectiveness of infection control initiatives targeting C difficile.
To compare the cost-effectiveness of 9 C difficile single intervention strategies and 8 multi-intervention bundles.
Design, Setting, and Participants
This economic evaluation was conducted in a simulated 200-bed tertiary, acute care, adult hospital. The study relied on clinical outcomes from a published agent-based simulation model of C difficile transmission. The model included 4 agent types (ie, patients, nurses, physicians, and visitors). Cost and utility estimates were derived from the literature.
Daily sporicidal cleaning, terminal sporicidal cleaning, health care worker hand hygiene, patient hand hygiene, visitor hand hygiene, health care worker contact precautions, visitor contact precautions, C difficile screening at admission, and reduced intrahospital patient transfers.
Main Outcomes and Measures
Cost-effectiveness was evaluated from the hospital perspective and defined by 2 measures: cost per hospital-onset C difficile infection averted and cost per quality-adjusted life-year (QALY).
In this agent-based model of a simulated 200-bed tertiary, acute care, adult hospital, 5 of 9 single intervention strategies were dominant, reducing cost, increasing QALYs, and averting hospital-onset C difficile infection compared with baseline standard hospital practices. They were daily cleaning (most cost-effective, saving $358 268 and 36.8 QALYs annually), health care worker hand hygiene, patient hand hygiene, terminal cleaning, and reducing intrahospital patient transfers. Screening at admission cost $1283/QALY, while health care worker contact precautions and visitor hand hygiene interventions cost $123 264/QALY and $5 730 987/QALY, respectively. Visitor contact precautions was dominated, with increased cost and decreased QALYs. Adding screening, health care worker hand hygiene, and patient hand hygiene sequentially to the daily cleaning intervention formed 2-pronged, 3-pronged, and 4-pronged multi-intervention bundles that cost an additional $29 616/QALY, $50 196/QALY, and $146 792/QALY, respectively.
Conclusions and Relevance
The findings of this study suggest that institutions should seek to streamline their infection control initiatives and prioritize a smaller number of highly cost-effective interventions. Daily sporicidal cleaning was among several cost-saving strategies that could be prioritized over minimally effective, costly strategies, such as visitor contact precautions.
Clostridioides difficile is the most common hospital-acquired infection in the United States, responsible for more than 15 000 deaths and $5 billion in direct health care costs annually.1 Health care facilities are a major source of new infections, and in-hospital prevention is critical to decreasing its overall incidence. Efforts to control C difficile infection (CDI) have intensified in recent years, with the addition of CDI to Medicare’s Hospital-Acquired Condition Reduction Program.2 However, the results of targeted infection control initiatives have been variable, and CDI incidence continues to rise.1,3,4
Nationwide, interventions are typically implemented simultaneously in multi-intervention bundles.3 This strategy makes it impossible to identify the isolated effects of single interventions using traditional epidemiologic methods.5 However, by developing an agent-based simulation model of C difficile transmission, our group was previously able to evaluate the clinical effectiveness of 9 interventions and 8 multi-intervention bundles in a simulated general, 200-bed, adult hospital.6 All hospitals operate in a setting of constrained resources. Thus, evaluating the cost-effectiveness of common infection control interventions is essential to providing evidence-based recommendations regarding which strategies to prioritize and implement.
While several C difficile cost-effectiveness studies have been published, the overwhelming majority focus on comparing treatment or diagnostic testing modalities.7 Among those that assess infection control initiatives, most evaluate a single intervention or single bundle. To our knowledge, only 2 other studies8,9 have investigated the comparative cost-effectiveness of multiple C difficile interventions. Neither evaluated emerging patient-centered interventions, such as screening at admission or patient hand hygiene. Furthermore, both studied environmental cleaning only as a bundled strategy and did not distinguish between daily and terminal cleaning8 or daily cleaning, terminal cleaning, and hand hygiene.9 Daily cleaning and screening are highly effective in their own right,6,10,11 and an evaluation of the cost-effectiveness of single-intervention strategies such as these is essential. Thus, we aimed to evaluate the cost-effectiveness of 9 infection control interventions and 8 multi-intervention bundles using an agent-based model of adult C difficile transmission.
We previously published an agent-based model of C difficile transmission in a simulated general, 200-bed, tertiary, acute care adult hospital.6 Output from this model was used to evaluate the cost-effectiveness of infection control strategies in terms of 2 primary outcomes: the cost per quality-adjusted life-year (QALY) saved and cost per hospital-onset CDI (HO-CDI) averted. The study was reviewed and approved by the University of Wisconsin–Madison institutional review board. This study follows the recommendations of the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guideline.12
For additional modeling details, see the eAppendix in the Supplement. Briefly, the model simulated a dynamic hospital environment and 4 agent types (ie, patients, visitors, nurses, and physicians), during a 1-year time period (eFigure 1 in the Supplement).6 Patients were categorized into 1 of 9 clinical states representing their CDI-related status. These clinical states were updated every 6 hours by a discrete-time Markov chain. Patients in the colonized, infected, recolonized, or recurrent infection states were contagious and could transmit C difficile to other agents and the environment. Once contaminated, visitors, nurses, physicians, and the environment could transmit C difficile to susceptible patients and the environment. The probability of transmission occurring during a given interaction was dependent on the agent types involved and the duration of the interaction (eTable 1 in the Supplement). Key model parameter estimates are shown in Table 1.6,10,13-94 The model was developed and run in NetLogo software version 188.8.131.52 We used synchronized random numbers, which allowed us to directly compare runs under different intervention scenarios, while minimizing variability owing to chance.96
We simulated the effects of 9 interventions, as follows: daily cleaning with sporicidal products; terminal cleaning with sporicidal products; patient hand hygiene; visitor hand hygiene; health care worker hand hygiene; visitor contact precautions; health care worker contact precautions; reduced intrahospital patient transfers; and screening for asymptomatic C difficile colonization at admission. Each intervention was modeled individually at an enhanced and ideal implementation level that reflected typical and optimal implementation contexts, respectively. We also simulated 8 infection control bundles that included between 2 and 5 enhanced-level interventions. Ideal-level interventions were not included in the bundle strategies because in general they did not result in considerable improvement compared with enhanced-level strategies. Thus, they were not deemed a high priority for bundle inclusion.
All strategies were compared with a baseline state, in which no interventions were enacted but standard hospital practices, such as hand hygiene, occurred at rates expected in a nonintervention context (Table 1). Ideal-level single interventions were also compared with the enhanced-level of each intervention, and bundles were compared among themselves. Each single intervention and bundle was simulated 5000 times. One replication of the simulation took approximately 115 seconds on a single core of a 1.80 GHz Intel Core i5-5350U processor with 8 GB of RAM running macOS Mojave version 10.14.3.
This study was conducted from the hospital perspective. Cost estimates (Table 21,14,62,97-140) were derived from the literature and converted into 2018 US dollars using the Personal Consumption Expenditure Health Index.141 Fixed and variable costs were considered. Both were higher for corresponding ideal-level vs enhanced-level interventions. Fixed costs included the cost of additional infection control staffing to implement, support, and serially evaluate compliance with an intervention (eAppendix in the Supplement). Ideal-level interventions had increased intervention compliance. Thus, the variable costs inherent in each successful intervention event (ie, alcohol-based hand rub product, labor related to alcohol-based hand rub hygiene time) also increased. We assumed that all costs occurred in the same year as the patient’s hospital visit; therefore, costs were not discounted. The excess cost attributable to a single CDI was estimated at $12 313 (range, $6156-$18 469).100,102,142
The number of HO-CDIs per year was output directly from the model for each run.6 We defined HO-CDI based on the Centers for Disease Control and Prevention’s guidelines as symptomatic diarrhea plus a positive laboratory test result on a specimen collected more than 3 days after hospital admission.143 We calculated QALYs using model output and the utility values shown in Table 2. To determine the QALYs lost because of CDI-associated mortality, the age distribution for CDI cases was used in conjunction with age-specific utility values from healthy adults. Mean life expectancies were derived from the Centers for Disease Control and Prevention life tables, accounting for a mean Charlson Comorbidity Index for in-hospital CDI patients of 2.57.102 The total number of deaths output from the model was multiplied by 0.48 to account for C difficile–associated mortality.1,135 Discounting future QALYs is controversial144; thus, they were not discounted in the primary analysis, similar to costs. Results of a supplemental analysis in which future QALYs were discounted at 3% is included in eTable 2 in the Supplement.
The minor loss in QALYs due to CDI symptoms was calculated from a mean symptomatic period of 4.2 days and utility value for symptomatic CDI of 0.81.132,133 Since there is no established utility measure of CDI in the United States, this followed a standard practice of basing it on that of noninfectious diarrhea.123-127 A loss in QALYs owing to time spent in a hospital admission was accounted for with a 0.63 utility value for hospitalized patients, derived using the EuroQol-5D instrument.134 Thus, it was possible to have a net negative QALY, despite a minimally net positive CDI averted.
Incremental cost-effectiveness ratios (ICERs) for HO-CDIs averted and QALYs gained were calculated using 2 methods. In the first approach, we found means for each intervention’s costs, HO-CDIs, and QALYs across all runs. We then calculated ICERs using these means for compared interventions. In the second method, an ICER was calculated based on the costs, HO-CDIs, and QALYs of 2 interventions for each run. These ICERs were then used to calculate the proportion of runs that met 21 willingness-to-pay thresholds. We assumed that any run resulting in negative incremental QALYs was not cost-effective. Analysis was conducted in R version 3.4.3 (R Project for Statistical Computing). No statistical testing was performed, so no prespecified level of significance was set.
A probabilistic sensitivity analysis was conducted varying cost and QALY parameter estimates simultaneously. Estimates were varied using the triangular distribution, with the minimum, mean, and maximum values reported in Table 2. Each single intervention and bundle simulation was run 100 000 times. One-way sensitivity analyses were also performed using the minimum and maximum reported values (Table 2).
In this agent-based model of a simulated 200-bed tertiary, acute care, adult hospital, 5 of 9 enhanced-level interventions were dominant compared with baseline hospital practices, resulting in cost savings, increased QALYs, and averted infections, as follows: daily cleaning (the most cost-effective, saving $358 268, 25.9 infections, and 36.8 QALYs annually), terminal cleaning, health care worker hand hygiene, patient hand hygiene, and reduced patient transfers (Table 3 and Figure 1). The clinical consequences of these interventions ranged considerably, with daily cleaning preventing more than 16 times as many infections as the patient transfer intervention (25.9 vs 1.6). Screening at admission cost $1283 per QALY, while health care worker contact precautions and visitor hand hygiene interventions cost $123 264 and $5 730 987 per QALY, respectively. The visitor contact precautions intervention was dominated, with increased costs and decreased QALYs.
Improving from enhanced to ideal intervention levels offered only small clinical benefits for most interventions (Table 3). It was cost saving and most effective for ideal health care worker and patient hand hygiene, averting an additional 7.1 and 4.0 HO-CDIs a year, respectively, compared with enhanced interventions. The ideal level was cost-effective for daily cleaning ($18 399/QALY), terminal cleaning ($5275/QALY), and patient transfer ($6194/QALY) at a willingness-to-pay threshold of $50 000/QALY.
Cost-effectiveness of the bundle strategies varied based on a bundle’s intervention components (Table 3). Adding patient hand hygiene to the health care worker hand hygiene intervention was cost saving, saving a mean of $32 588 and 4.2 QALYs annually in the model 200-bed hospital compared with the health care worker hand hygiene intervention alone. When screening, health care worker hand hygiene, and patient hand hygiene interventions were sequentially added to daily cleaning to form 2-, 3-, and 4-pronged bundles, the ICERs for these additions were $29 616, $50 196, and $146 792 per QALY, respectively.
We also evaluated the percentage of times each intervention was cost-effective at 21 willingness-to-pay thresholds. These results are presented as an acceptability curve (Figure 2). Daily cleaning consistently had the greatest proportion of runs that were cost-effective, with 99% of runs cost-effective at a willingness-to-pay threshold of $5000 per QALY.
Detailed results of the 1-way sensitivity analyses and probabilistic sensitivity analysis are included in eFigure 2, eFigure 3, eFigure 4, and eTable 3 in the Supplement. The trends in comparative cost-effectiveness were stable across most variations in cost and utility parameters. The 5 cost-saving interventions were most sensitive to hospitalization costs (eFigure 2 in the Supplement). Screening at admission was most sensitive to increased costs of polymerase chain reaction testing. Visitor hand hygiene and health care worker contact precautions were most sensitive to changes in age-related utility values (eFigure 3 in the Supplement). Most notably, in the probabilistic sensitivity analysis (eFigure 4 in the Supplement), the patient-centered intervention bundle (comprised of screening at admission, patient hand hygiene, and patient transfer) changed from cost-saving to a cost of $245/QALY, and the visitor hand hygiene intervention became dominated (compared with $5 730 987/QALY) (eTable 3 in the Supplement).
In this model-based economic evaluation, daily cleaning, health care worker hand hygiene, patient hand hygiene, terminal cleaning, and reduced patient transfers were all found to be cost saving. Daily cleaning was the most clinically effective and cost-effective intervention by far, saving $358 268, 25.9 infections, and 36.8 QALYs annually in the 200-bed model hospital. In comparison with the other existing C difficile simulation models, Brain et al9 found that a cleaning and hand hygiene bundle had the greatest increase in QALYs and was the most cost-saving of 9 bundle strategies. Nelson et al8 reported that increasing environmental cleaning within the context of multi-intervention bundles resulted in minimal gains in effectiveness. However, their bundle strategies included up to 6 interventions simultaneously and are not comparable with an isolated daily cleaning intervention. Similarly, a recent multicenter trial by Ray et al145 found that reduction of C difficile environmental cultures did not correlate with reduced infection rates. However, this study is also not comparable, given that it targeted sporicidal daily cleaning only in known CDI rooms and did not change practices for non-CDI patient rooms and hospital common rooms. Thus, it appears that blocking asymptomatic transmission by using sporicidal products hospitalwide may be essential to obtaining a reduction in HO-CDI rates.
Among all the interventions we modeled, health care worker hand hygiene is the most well studied and has been shown to be cost saving in several prior contexts. Chen et al146 reported that every dollar spent on their hospital’s 4-year hand hygiene program resulted in a $32.73 return on investment (2018 USD). Likewise, Pittet et al147 found that hand hygiene needed to account for less than 1% of the concurrent decline in hospital-associated infections at their institution to be cost saving. Our results are also in line with the prior modeling studies. Nelson et al8 reported that adding health care worker hand hygiene to existing bundles increased total QALYs with few additional costs, and health care worker hand hygiene was a key component of the most cost-saving cleaning and hygiene bundle in the study by Brain et al.9
C difficile screening has also recently been shown to be highly effective at reducing HO-CDI in real-world and modeling contexts.6,10,11,148,149 This intervention was highly cost-effective in our model, at a cost of $1283/QALY and is similar to the results of the study by Bartsch et al,124 in which screening cost less than $310/QALY (2018 USD).124 Both are likely conservative estimates because the cost-effectiveness of screening is expected to increase if the intervention is targeted to high-risk populations. In fact, when Saab et al149 modeled a C difficile screening and treatment intervention exclusively for patients with cirrhosis, costs were found to be 3.54 times lower than under baseline conditions.
The Veterans Affairs methicillin-resistant Staphylococcus aureus (MRSA) screening bundle, instituted at Veterans Affairs hospitals nationwide in 2007, provides a precedent for large-scale screening implementation. It ultimately had a 96% participation rate and reduced MRSA by 45% among patients not in the intensive care unit patients and 62% among patients in the intensive care unit.90 The cost-effectiveness of this intervention was calculated at between $31 979 and $64 926 per life-year saved (2018 USD).97 Given the evidence from our study and others,124,149 we expect that screening for C difficile would be even more cost-effective than the Veteran Affairs MRSA initiative. However, additional work is needed to identify which populations to target before widespread implementation.
While screening is not yet standard practice, contact precautions are a mainstay of C difficile infection prevention programs.3 They are recommended by the Society for Healthcare Epidemiology of America for both health care workers and visitors of patients with CDI.150,151 However, evidence for these guidelines is based primarily on studies of other pathogens and theoretical transmission concerns,108,152 given that C difficile–targeted studies are lacking. In our study, we found neither health care worker nor visitor contact precautions to be cost-effective. The enhanced-level health care worker contact precautions intervention cost $123 264 per QALY, with another $136 135 per QALY for the ideal-level implementation. The results were even worse for visitor contact precaution interventions, with the enhanced level being dominated and the ideal level costing $1 669 089 per QALY. Thus, it is likely that the screening intervention, which, as modeled, prompts the use of visitor and health care worker contact precautions for asymptomatic colonized patients, would be even more cost-effective if contact precautions were not used for asymptomatic patients who test positive.
Recognizing that all hospitals operate in an environment of constrained resources, support must be shifted from minimally effective, high-cost interventions, such as visitor contact precautions, to more innovative, cost-effective solutions. For example, patient hand hygiene, which is rarely incorporated into C difficile bundles,3 was 1 of only 2 interventions to be cost saving at both the enhanced and ideal level. It was also cost saving compared with health care worker hand hygiene alone. In fact, all 2-pronged intervention bundles investigated in this study were cost saving. However, incremental intervention cost-effectiveness decreased beyond 2-intervention bundles. Adding subsequent interventions to the 2-pronged daily cleaning and screening at admission bundle came at an ICER of $50 196/QALY for the third strategy, $146 792/QALY for the fourth strategy, and $758 618/QALY for the fifth strategy.
The recommendation to implement a smaller number of highly effective interventions runs contrary to the current infection control climate. A recent review of CDI bundles found that more than half of bundles include 6 or more components, with a minimum of 3 and maximum of 8 interventions.3 Given the lack of evidence and guidelines surrounding bundle composition, it is not surprising that institutions seek to maximize CDI reduction by implementing increasingly larger bundled strategies. However, our results provide evidence that continuing to increase bundles without accounting for the cost and effectiveness of individual components may be counterproductive, depending on institutional priorities and cost constraints. Instead, institutions should consider streamlining their infection control initiatives and may opt to focus on a smaller number of highly cost-effective interventions.
It is important to note that while many of the interventions in this study were cost saving, they are not without upfront costs. Even at the enhanced level, each intervention required the employment of additional infection control nursing staff. These individuals have the critical responsibility of coordinating implementation, assessing compliance, providing direct frontline feedback, and iteratively evaluating intervention effectiveness. Hospital administrative buy-in and financial support is key to both the initial implementation of an intervention and sustaining its long-term success.
This study has limitations. The cost-effectiveness results presented in this study are inherently dependent on the quality of our agent-based model, which underwent rigorous verification and validation processes.6 It suffers from limitations of the original model, such as assuming transmission of a generic C difficile strain and the lack of an antibiotic stewardship intervention. Particularly relevant to this study, we did not stratify CDI by severity or include complications such as colitis or toxic megacolon. By evaluating all cases using a utility value that corresponds to mild to moderate CDI, we likely underestimate the true cost-effectiveness of these interventions.
To our knowledge, this was the first C difficile cost-effectiveness analysis to compare standard infection control strategies and emerging patient-centered interventions. In a field that lacks specific guidance regarding the cost-effectiveness of interventions targeting C difficile, this study provides critical evidence regarding where to allocate limited resources for the greatest potential success. Daily sporicidal cleaning is among several promising, cost-saving strategies that should be prioritized over minimally effective, costly strategies, such as visitor contact precautions. Maintaining the status quo, focused on large, multipronged bundles with variable efficacy, will continue to shift limited resources away from more productive, cost-saving strategies that have greater potential to improve patient outcomes.
Accepted for Publication: May 25, 2020.
Published: August 13, 2020. doi:10.1001/jamanetworkopen.2020.12522
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Barker AK et al. JAMA Network Open.
Corresponding Author: Anna K. Barker, MD, PhD, Department of Internal Medicine, University of Michigan, 3116 Taubman Center, SPC 5368, 1500 E Medical Center Dr, Ann Arbor, MI, 48109 (email@example.com).
Author Contributions: Drs Barker and Alagoz had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Barker, Safdar, Alagoz.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Barker, Alagoz.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Barker, Alagoz.
Obtained funding: Barker, Safdar.
Administrative, technical, or material support: Scaria, Safdar.
Supervision: Safdar, Alagoz.
Conflict of Interest Disclosures: Dr Alagoz reporting having previously served as a paid consultant to Biovector Inc, a company active in the area of infection control, outside the submitted work. No other disclosures were reported.
Funding/Support: This work was supported by a predoctoral traineeship from the National Institutes of Health (grant number, TL1TR000429) to Dr Barker. The traineeship is administered by the University of Wisconsin–Madison, Institute for Clinical and Translational Research, funded by National Institutes of Health (grant number, UL1TR000427). It is also supported by the Veterans Health Administration National Center for Patient Safety Center of Inquiry in the United States Department of Veterans Affairs to Dr Safdar. This research was also supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health Office of the Director (award number, DP2AI144244).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the US Department of Veterans Affairs of the US government.