Vicki R. Lewis, Lindsey Clark, Raj Ratwani
Approximately 5 percent of hospitalized patients suffer a health care-associated infection (HAI), and ambulatory chronic care patients, such as those on hemodialysis for treatment of end-stage renal disease, have seen a 31 percent increase in HAIs. Research in other industries has shown that safe systems are achieved by considering the entire system to enable performance specifications to be met. The sociotechnical systems (STS) model was applied to identify and evaluate factors that lead to HAIs among patients in an ambulatory dialysis unit (ADU). This information was used to recommend and test an intervention for purposively aligning the STS factors to reduce HAIs in an ADU. The framework to apply the STS model was the macroergonomic analysis and design (MEAD) method, an approach used in analyzing complex work systems. The framework is used to understand relationships and interactions within a system that can affect issues such as HAI incidence and intervention efforts. The MEAD method identifies "variances" (nonoptimal situations or misalignments of work-system factors) that occur in the STS domains (external environment, physical environment, technology, people, and organization). Variances within and between domains are then analyzed and appropriate interventions designed. In this paper, we discuss the STS model, the MEAD framework, and the application of the model and framework to identify misaligned work-system factors and the development of an intervention to reduce variances that potentially lead to HAIs in an ADU. In addition, we discuss the challenges encountered and lessons learned in conducting this study.
Five percent of hospitalized patients in the United States develop a health care-associated infection (HAI),1 and ambulatory chronic care patients, such as those on hemodialysis for treatment of end-stage renal disease (ESRD), suffer higher rates. Since 1993 the rate of hospitalizations for ESRD patients due to infection has increased by 31 percent, while all-cause hospitalization rates have decreased.2
Identifying the factors that contribute to HAIs is a challenge due to the nature of the health care system. Health care is a complex work system with various interrelated components that interact with one another in a dynamic fashion. Challenges affecting HAI reduction in inpatient settings are exacerbated in outpatient settings because patients are ambulatory and can come into contact with more infection-causing pathogens, both within and outside the care environment.
To analyze and address HAIs within a complex system, it is important to examine the many factors that contribute to HAIs and to implement an inclusive and comprehensive intervention plan. One framework that is particularly well-suited for addressing health care challenges (such as infections) within complex work systems is that of the sociotechnical system (STS) model.3 STS is an approach to complex organizational work design that recognizes the interaction between people and technology in workplaces. This approach aids in uncovering relationships and interactions within a system that can affect issues such as HAIs and how interventions work within the system.
Figure 1 illustrates the STS framework as applied to an ambulatory dialysis unit (ADU). In an STS analysis, one must first define the boundaries between the internal environment and the external environment. In the instance of the ADU, the external environment is composed of factors that affect what occurs within the dialysis unit but are outside its control. This includes influences such as policies, legislation and government regulations, equipment suppliers, and the regional demographics of patients. The internal environment is defined as the primary focus of the study and includes four domains:
- Organizational factors give insights into the structure of the organization, as well as the mutual relations within the focal unit necessary to accomplish the tasks and achieve the goals of the unit, such as unit policies and procedures, shift schedules, staff-to-patient ratios, work culture, and work values and beliefs.
- Physical environment factors include all aspects of the design of the physical space, such as layout and design, chair/bed spacing, air and water quality, work surfaces, resource locations.
- Technical factors are the means and methods by which work is performed, such as work processes and procedures, tools, equipment, and software.
- People factors are the characteristics and attributes of the individuals who interact with hospital staff, such as clinical staff, environmental services staff, transportation staff, patients, patient family members.
Studying a health care work system from an STS perspective helps to identify the contribution of each of the four domains, both individually and in combination, to unintended outcomes such as HAIs. Solutions can then be developed that address multiple factors within the work system, achieving greater effectiveness and sustainability than solutions that only target one factor. Once the STS model is defined, a method to apply the model is needed that will uncover system relationships in order to develop solutions that address the HAI risk factors present.
The remainder of this paper provides a description of the methods used to apply the STS model—the macroergonomic analysis and design (MEAD) method, as described by Kleiner3—in an ambulatory dialysis facility. Several STS models are discussed in the literature (see Carayon4 for a list of models and the STS components addressed by each). This method was chosen because the STS components described by the MEAD method are applicable to the health care environment and the problem addressed. For example, some models evaluated components such as supply chains or spatial interactions that were not the central focus of this study. There was the added benefit of having one of the MEAD method's authors to advise us in the application of the process. The MEAD method was utilized to identify misaligned work-system factors and to develop an intervention to realign those factors and effectively reduce variance. Kleiner defines a variance as something that significantly affects performance criteria.3 In this case, a variance may be thought of as any situation that may lead to HAIs in an ADU.
Figure 1. Health Care as a Sociotechnical System (adapted from Kleiner3)
The MEAD method was applied to an ambulatory dialysis facility that is the largest not-for-profit dialysis center in the greater Baltimore/Washington, DC region. It is equipped with 54 dialysis stations across three rooms that serve approximately 280 patients and provide more than 3,600 dialysis sessions per month. A discussion of the challenges encountered over the course of this study and lessons learned will also be presented.
Macroergonomic Analysis and Design (MEAD) Method
The MEAD method facilitates the analysis and organization of data by identifying variances (nonoptimal situations) within STS domains and misalignments between STS domains. MEAD consists of 10 major analyses (Table 1). Phases I through IV are the STS domain initial analyses. Phases V through VIII use the data from the initial domain analyses to determine variances and methodically identify misalignments in work-system factors. Phases IX and X support the development of the intervention and implementation and illustrate the iterative process to optimize the implementation. A more elaborate description of the goals of each phase may be found in Kleiner.3,5
|I. Environmental and Organizational Design Subsystems: Initial Scanning||1. Perform Mission, Vision, Principles Analysis
2. Perform System Scan
3. Perform Environmental Scan
4. Specify Initial Organizational Design Dimensions
|VI. Personnel Subsystem Analysis: Construct Key Variance Control Table and Role Network||1. Construct Key Variance Control Table
2. Construct Role Network
3. Evaluate Effectiveness
4. Specify Organizational Design Dimensions
|II. Technical Subsystem Analysis: Define Production System Type and Performance Expectations||1. Define Production System Type
2. Define Performance Expectations (Performance Criteria)
3. Specify Organizational Design Dimensions
4. Define Function Allocation Requirements
|VII. Function Allocation and Joint Design||1. Perform Function Allocation
2. Design Changes to the Technological Subsystem
3. Design Changes to the Personnel Subsystem
4. Prescribe Final Organizational Design
|III. Flowchart the Technical Work Process and Identify Unit Operations||1. Identify Unit Operations
2. Flowchart the Process
|VIII. Role and Responsibilities Perceptions||1. Evaluate Role and Responsibility Perceptions
2. Provide Training Support
|IV. Collect Variance Data||1. Collect Variance Data
2. Differentiate Between Input and Throughput Variances
|IX. Design/ Redesign Support Subsystems and Interfaces||1. Design/Redesign Support Subsystems
2. Design/Redesign Interfaces and Functions
3. Design/Redesign the Internal Physical Environment
|V. Construct Variance Matrix||1. Identify Relationships Among Variances
2. Identify Key Variances
|X. Implement and Improve||1. Implement
2. Perform Evaluations
Application of the MEAD Method
In preparation for conducting the MEAD analyses, HAI risk factors were defined to aid in the design of data collection and analysis. A comprehensive literature review was conducted to identify the HAI risk factors that exist within each of the four internal STS domains. In addition, a menu was developed of current interventions related to organizational, technological, people, and environmental aspects to reduce HAIs.6,7 This process revealed that HAI risk factors and potential interventions are rarely examined across STS domains. For example, articles addressing poor hand hygiene often identified the people domain as the contributing factor to this problem and suggested interventions that only targeted this factor (e.g., provide reminders to staff to wash their hands). Organizational, technical, and physical environment factors (e.g., availability and location of sinks or other hand washing stations and their relationship to workflow) were typically not addressed in the same article, demonstrating that current interventions may not be sufficient for addressing HAIs in an ADU.
Of the 213 HAI intervention studies reviewed, only one paper addressed all four of the STS factors. In this study, Pronovost and colleagues demonstrated an intervention that reduced catheter-related bloodstream infections to zero at 3 months after implementation and sustained significantly low rates for 16 to 18 months after implementation.8 Pronovost's team did not use the same STS framework used in this study; however, critical components of that intervention were to "summarize and simplify what to do; measure and provide feedback on outcomes; and improve culture by building expectations of performance standards into work processes," as described by Bosk et al.9 Closer inspection of the Pronovost et al. study revealed that the authors examined the culture of the organization; they noted needed improvements, involved leadership, took efforts to understand the personnel, redesigned work processes taking the environment into consideration, and acknowledged external environmental pressures, which is similar to using an STS framework. This demonstrates that effectiveness and sustainability of HAI interventions can be achieved when all work-system factors are analyzed.
The literature review determined that the risk factors and potential interventions encompassing all STS domains have scarcely been examined. Therefore, the literature alone could not be used to recommend complete interventions because each study focused on only one STS domain, in one particular type of health care environment. Overall qualities, such as cleanliness of surfaces and tools, proper hand hygiene, and clean air and water are essential across health care environments. However, interventions that work in an ICU may not be applicable within a hemodialysis unit because the acuity of the patients differs, the environment in which the care is provided differs, and the equipment differs. The literature review supported the strategy that designing a successful intervention would require an analysis of each of the STS domains in a health care work system. From this perspective, the team developed a data collection strategy.
Development of Data Collection Methods
Whereas the MEAD method defines the steps for information that should be gathered and for the analysis, it does not specify the methods for data collection. After advice and input were obtained from a Technical Expert Panel representing experts in nephrology, infection prevention and control, nursing, and human factors engineering, the data collection methods were determined. Table 2 shows how each method contributed to a specific domain analysis.
|Domain Definition and Analysis||Data Collection Methods|
|External Environment||Document and policy reviews (such as corporate documents, hospital policies, Federal regulatory documents and policies) and open-ended interviews (State regulatory personnel)|
|Physical Environment and Organization||Open-ended interviews with upper-level management and hospital administration (such as the vice president for medical affairs, chair of nephrology, renal nurse managers, environmental services management)|
|Organizational document review of documents pertaining to nursing work structure, patient satisfaction, and overall facility status (such as the facility nursing organizational chart, 2011 dialysis facility report, and recent satisfaction surveys)|
|Surface contamination and air quality assessment|
|Technical and Personnel||Focus groups with frontline dialysis staff and ESRD patients|
|Patient chart review to analyze patient population in terms of comorbidities, living situations, vascular access, and recent hospitalizations|
|HAI surveillance data collection to understand the most prevalent infections in the unit, identify seasonal trends vs. unexpected peaks|
|Work processes observation, including initiating patient treatment, ending patient treatment, monitoring patients, shift-change work process, and cleaning processes.|
The external environment (factors outside the ADU's control that affect the ADU work environment) was determined to include the following: external institutions and organizations (e.g., the Maryland Kidney Commission, Centers for Medicare & Medicaid Services), national guidelines and recommendations, patient lifestyle, and patient vascular access determination. Within the organization domain, the main components were identified as the ADU's mission, vision, and values; financial health; unit capacity; policies and procedures; renal team management and structure; environmental services management and structure; quality assurance management and structure; and patient education policies. The physical environment was analyzed in terms of physical layout/design, air quality, water quality, and equipment and resource location. The technical domain was analyzed in terms of work processes and procedures, tools, equipment, and software. The people domain was determined to include clinical staff, environmental services staff, transportation staff, patients, and patient family members.
Following data collection across the STS domains, the data were analyzed for emerging themes related to HAI risk factors to determine variances within each domain. Fifty-seven variances were identified across the external and internal environment through the MEAD analysis. Table 3 provides the total number of variances identified for each domain and examples of variances found within each domain. The variances within the people domain represent a large proportion of the total number of variances identified. Several of the variances overlap with variances from different internal domains, which illustrates the interconnected nature of the STS domains and the importance of considering variance within a holistic context. For example, the physical environment variance that "the physical layout of the oldest treatment area contains barriers that are difficult to work around and creates inefficient workflow patterns" may also be viewed as an organizational domain resource constraint that sufficient resources cannot be procured to remodel and/or add additional facilities. Nonetheless, the critical task is to utilize the MEAD method to thoroughly identify the variances so that variance solutions may be determined by considering options across all of the STS domains. In this case, the layout of the unit was a physical domain variance, and the organizational resources were viewed as a restraint to be considered when developing interventions.
Identification of Work-System Misalignments
The MEAD analyses conducted for Phases V through VIII allowed the team to construct the variance matrix for the ADU (Phase V) and determine where realignments needed to occur. The variances are not weighted, in an effort to identify those that may contribute more heavily to the identified problem, in this case the occurrence of HAIs in an ADU. Instead, each variance is analyzed in terms of its relationship to other variances so that "key" variances can be identified. Key variances are those that have numerous relationships with other variances3 and are important because they significantly affect performance criteria. In addition, each of the 57 variances was categorized across six HAI risk factors identified in the literature: surface contamination, workflow/work stress, hemodialysis patient-specific risks, feedback, patient education, and standards of care. This process allowed us to decide where to focus our efforts and to determine the scope of the intervention. Table 4 provides an example of the results of this process for surface contamination.
Determining the Intervention
The analysis described in the previous phases informed Phases IX and X. This led to the development of an intervention change package, the AHRQ Systematic Approach for Eliminating Risks (SAFER) Initiative. The components of the change package were selected by the ADU management and research team to meet practical criteria—such as personnel, resources, timelines and criteria, and access—that emerged during the MEAD analyses. The ADU management and research team determined that the intervention components must meet the following criteria:
- Linked directly to the risk factors and variances identified.
- Approved by dialysis facility management prior to implementation.
- Achievable and able to be implemented in a reasonable timeframe.
- Sustainable with available facility resources and not reliant upon research funds to maintain.
- Selected to create efficiencies where possible.
- Addressable to the HAIs that are most prevalent in the ADU: vascular access-related bloodstream infections, vascular access site infections, and wound infections.
|STS Domain (Total Number of Variances)||Examples of Variances Identified|
|External Environment (5)||Complex national guidelines and regulations do not provide instructions and best practices for implementing the suggested recommendations. To meet national guidelines, facilities must independently interpret the recommendations and devise plans for implementation.|
|Patients rely on a variety of transportation methods to travel to and from the dialysis unit (family member, nonfamily volunteer, public transportation, private facility transportation such as from a nursing home). Transportation schedules may result in a patient needing to start or end dialysis treatments at different times to meet transportation needs.|
|Fistulas and grafts are the preferred vascular access devices for dialysis treatment because they are safer than catheters and have a lower risk for infection.* Multiple external factors contribute to the vascular access device used by a patient, including insurance policies and various pre-existing and incompatible medical conditions.|
|Organizational (10)||Three patients are scheduled to be put onto dialysis within a 30-minute window for each dialysis shift, yet care providers have a 5-hour treatment shift that allows for more flexibility in putting patients onto dialysis.|
|Environmental services (ES) staff members are not an integrated part of the renal team. These staff members serve the entire hospital, with one ES staff member dedicated to each of the three dialysis units. This can contribute to shift delays and disjointed communication between ES and dialysis staff.|
|Consolidating quality assurance data is time consuming and difficult. To create a comprehensive report of the data, quality assurance managers must merge information from three separate sources: hospital reports, antibiotic lists, and laboratory reports.|
|Physical Environment (6)||The physical layout of the oldest treatment area contains barriers that are difficult to work around and create inefficient workflow patterns.|
|Patients gain access to treatment areas before their scheduled time or before the staff is ready to receive them. This contributes to a stressful and rushed environment.|
|Supplies and equipment are not located in easily accessible areas, and procedures may require staff to make multiple trips to different locations to obtain the needed supplies to put a patient on dialysis.|
|Technical (8)||Unexpected events while a patient is put onto and taken off of dialysis disrupt the process workflow.|
|Work processes do not support early patient wound detection.|
|Because of responsibility to other units, it is difficult for ES staff to adapt to unanticipated schedule changes, which may happen for a host of reasons.|
|People (28)||There is a lack of communication between ES and dialysis unit staff during shift change.|
|Staff feels rushed during dialysis put-on, which leads to inconsistent practices.|
|Patients are not always aware they have a wound.|
* Go to Centers for Disease Control and Prevention; 2001; http://www.cdc.gov/mmwr/preview/mmwrhtml/mm6008a4.htm.
|Organizational||Technical||People||Physical Environment||External Environment|
|Environmental Services (ES) is not part of the renal team||Availability of ES staff at shift change||Lack of communication between ES and dialysis staff during shift change||Bacteria counts on high-touch surfaces are very high||Patients may introduce contaminants from outside|
|Inadequate ES staffing for the size of the unit and tasks involved||Availability of ES staff to handle odd start/stop time patients||Knowledge of ES staff regarding types of infections in the unit|
One may note that the criteria do not address the inclusion of only interventions supported by an evidence base; however, evidence-based recommendations and guidelines, such as those issued by the Centers for Disease Control and Prevention (CDC)10 and the Association for Professionals in Infection Control and Epidemiology (APIC),11 were considered when selecting components for the change package. For example, the use of an antibiotic ointment for catheter patients was included based on CDC's recommendation; however, the staff's suggestion to have rolling carts for supplies was not included, in order to adhere to APIC guidelines. Nonetheless, while there is a body of evidence that supports interventions to reduce HAI rates in inpatient settings, the evidence base supporting interventions to reduce HAI rates in ambulatory settings is scant.
The reason for the lack of evidence in ambulatory settings is partially driven by the fact that numerous variables influence HAI rates, and these variables are extremely difficult to control in outpatient settings. For this reason, the lack of an evidence base to support interventions in ambulatory settings was not applied as a strict criterion for inclusion in the change package; however, the evidence base was considered to the extent that information was available. Furthermore, the relationship between HAIs and suspected risk factors does not always have an undisputed evidence base. For example, while there is sufficient knowledge regarding the means of transmission of bacteria and correlations between surface contamination and HAI rates, there is not yet clear evidence that reducing surface contamination reduces HAIs in an ADU. Finally, there is not an evidence-based intervention for every variance noted in the ADU. For example, while workflow and work stress were noted as correlated contributors to HAIs in some health care settings, there were no interventions in the literature that specifically addressed this ADU's particular stressors. Therefore, we felt it was appropriate to expand possible interventions to those that addressed the variances and met the other criteria.
Two risk factors were not addressed because the intervention components would not meet these criteria: patient education and standard of care. There was a lack of time and financial resources to provide additional education and training resources to the ADU's highly variable patient population. The recommended intervention components targeting the remaining four risk factors developed by the research team in collaboration with the renal team are discussed below. Table 5 lists the risk factors and recommended interventions.
Table 5. Identified risk factors and recommended intervention components of the SAFER initiative change package (pilot intervention)
|Risk Factor||Recommended Intervention|
|Surface Contamination||Provide dedicated environmental services (ES) resources to the ADU|
|Improve communication between dialysis staff and ES staff|
|Add antimicrobial materials to high-touch areas|
|Workflow/Work Stress||Install transparent privacy film on window between main treatment area and waiting room|
|Keep patients out of units until scheduled|
|Hemodialysis Patient-Specific Risk Factors||Perform foot exams for at-risk patients|
|Use antiseptic wipes to clean the patient’s arm prior to dialysis put-on|
|Use antibiotic ointment at the vascular access exit site for catheter patients|
|Feedback||Optimize HAI surveillance system for quality assessment|
|Post monthly HAI rates in waiting rooms and staff areas|
Discussion and Lessons Learned
The MEAD method provides a structured and comprehensive approach to uncover and untangle the HAI risk factors in a busy ADU. The MEAD method offers a number of advantages over previous research efforts. One advantage of the step-by-step, 10-stage process is that it is adaptable to a variety of settings. In addition, since each specific health care setting will have different internal and external environments, the STS perspective allows the mapping of information garnered using the MEAD method to a solid foundation for developing effective and sustainable intervention packages.
There were several lessons learned in the application of the MEAD method that may be noted for future use, including issues related to literature reviews, time and resources needed, the role of the environment (in this case, ambulatory units), and the importance of interpersonal relationships. Each of these areas will be discussed in further detail below.
The importance of the literature review was a key lesson learned that provided the basis of the research team's understanding of the risk factors for HAIs. To prepare for and design data collection and analysis, the time and effort spent preparing a comprehensive literature review was invaluable. Systematically reviewing previous research on this topic allowed the research team to define risk factors for HAIs, which provided the basis for the MEAD analysis.
The immense time and resource commitment that is involved in collecting and analyzing data in the very complex environment of a large ADU was another lesson learned. Once the range of risk factors present in the unit was uncovered and the types of interventions evaluated, it was determined that the list of data collection methods needed to systematically study each STS domain would be more extensive than originally anticipated. However, while conducting analyses of the STS domains and developing interventions to address those domains may be more resource-intensive than many methods that have been used to mitigate HAIs in the past, the advantage of the STS framework is that the application of its methods identifies all of the STS domain factors for which variances, or nonoptimal situations, exist. The necessary alignments for a successful intervention begin to surface, indicating that a set of robust intervention components is necessary to align work-system factors across domains for maximum effectiveness. Considering the great success demonstrated in the work by Pronovost et al., which incorporated an HAI intervention comprising all of the STS subcomponents,8 and the high risk and cost of HAIs to society, the rationale for using a comprehensive STS research approach is clear.
Engaging in research in an ambulatory care environment provided another lesson learned, since reducing HAIs in this environment appears to be exponentially more complicated than it would be in an inpatient environment. However, with health care delivery shifting away from inpatient hospital settings and toward a variety of ambulatory and community-based settings, understanding and identifying the challenges to reducing HAIs in ambulatory care is of paramount importance. Vulnerable patient populations rely on frequent and intensive use of ambulatory care to maintain or improve their health. For example, each year more than one million cancer patients receive outpatient chemotherapy, radiation therapy, or both.12 It is critical that all of this care be provided under conditions that minimize or eliminate the risks of HAIs.13
Many aspects of an ambulatory health care environment are difficult to control, and whether an infection is truly health care-associated, simply community-acquired, or patient-driven can be very difficult to define. While an intensive care patient is passive to his or her care, an ambulatory patient potentially can play a vital role in that care. And while the environment of an intensive care unit provides opportunity for control, an ambulatory environment introduces challenges such as patient transportation issues, patient support needs, and seasonal variation in HAI rates that direct how data must be collected and analyzed. Although this may make an HAI rate of "zero" seem daunting, systems models such as STS and methods such as MEAD begin to untangle the complex system components that support and drive tangible solutions.
A final note is that we have maintained the importance of ADU management and staff ownership of the project from its initiation. For example, we regularly share information about the project with ADU management and staff, except for confidential data and materials. While we suggested an initial set of interventions in the context of the variances that were uncovered, these were provided as suggestions. If a member of the ADU leadership did not agree with a recommended intervention, we listened to the ADU leadership reasoning and asked for a suggested alternative that would address the same variance or set of variances. We believe this method yielded a superior intervention change package that, more importantly, had the buy-in of management and staff and was thought to have a higher likelihood of sustainability after completion of data collection for the main intervention.
This project was funded under contract no. HHSA290201000021I from the Agency for Healthcare Quality and Research (AHRQ), U.S. Department of Health and Human Services, under the ACTION II contract awarded to the CNA Health ACTION Partnership [CHAP]. The findings and conclusions in this document are those of the authors, who are responsible for its content, and do not necessarily represent the views of AHRQ. No statement in this report should be construed as an official position of AHRQ or of the U.S. Department of Health and Human Services. We thank Kerm Henriksen, Ph.D., of AHRQ, for his thoughtful oversight of this project, and Amanda Borsky, MPP, of CNA, for her management and contribution to this project and manuscript. We also acknowledge the contribution of Brian Kleiner, Ph.D., who provided invaluable advice and support on the application of the MEAD approach for this project.
National Center for Human Factors in Healthcare, Washington, DC (VRL, LC, RR).
Address correspondence to: Vicki R. Lewis, Ph.D., National Center for Human Factors in Healthcare, 3007 Tilden St. NW, Suite 7M, Washington, DC 20008; Email: Vicki.R.Lewis@medicalhfe.org.
1. Centers for Disease Control and Prevention. Vital signs: central line-associated blood stream infections—United States, 2001, 2008, and 2009. Morb Mortal Wkly Rept 2011 Mar 4;60(8):243-248. www.cdc.gov/mmwr/preview/mmwrhtml/mm6008a4.htm.
2. United States Renal Data System. 2011 USRDS Annual Data Report. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Division of Kidney, Urologic, and Hematologic Diseases; 2011.
3. Kleiner BM. Sociotechnical systems in health care. In: Carayon P, ed. Handbook of Human Factors and Ergonomics in Health Care and Patient Safety. Mahwah, NJ: Lawrence Erlbaum Associates; 2007:79-95.
6. Stephens RJ, Lewis VR, Clark L, et al. Identifying and Aligning Work-System Factors to Reduce Healthcare-Associated Infections (HAIs). Washington D.C.: National Center for Human Factors Engineering in Healthcare; 2012.
7. Lewis VR, Parker SH, Stephens RJ, et al. Why Aren't we achieving better results? A literature review of healthcare associated infection interventions. Proceedings of the Human Factors and Ergonomics Society Annual Meeting. 2012 Sep;56(1):829-33. doi/10.1177/1071181312561173.
10. Centers for Disease Control and Prevention. Dialysis Safety Guidelines and Recommendations. 2013. Available at: http://www.cdc.gov/dialysis/guidelines/index.html
12. Halpern MT, Yabroff KR. Prevalence of outpatient cancer treatment in the United States: estimates from the Medical Panel Expenditures Survey (MEPS). Cancer Invest 2008 Jul;26(6):647-51. PMID: 18584358.
13. National Center for Emerging and Zoonotic Infectious Diseases, Division of Healthcare Quality Promotion. Guide to Infection Prevention for Outpatient Settings: Minimum Expectations for Safe Care. Atlanta, GA: Centers for Disease Control and Prevention; 2011.