Kahn, Katherine L. MD*,†; Weinberg, Daniel A. PhD‡; Leuschner, Kristin J. PhD*; Gall, Elizabeth M. MHS‡; Siegel, Sari PhD‡; Mendel, Peter PhD*
The reduction and elimination of healthcare–associated infections (HAIs), the ultimate goal of Health and Human Services’ (HHS’s) Roadmap to Elimination,1 has depended in large part on the ability of lead federal agencies to measure and track the progress of HAI rates. Measuring and monitoring rates and making results accessible to healthcare workers, patients, and other stakeholders has been associated with improvements in the processes and outcomes of care.2–5
Unfortunately, measurement and validation of HAI data have been challenging. Despite the availability of multiple data systems, the Government Accountability Office’s 2008 report noted that HHS’s ability to accurately track HAIs was hindered by a lack of coordination among data sources.6 Furthermore, there were shortcomings in individual data sources, including changes over time in infection definitions, data collection tools, and analysis cohorts. An important challenge for the Action Plan was to provide a comprehensive cross-sectional and longitudinal picture of HAI rates, by using either existing or new data systems.
In this paper, we present HAI data and monitoring-related results of the IMPAQ-RAND evaluation of the Action Plan. We focus here on the first 3 elements of the Content-Input-Process-Product model: context (goals), inputs, and processes; the paper describes the selection of HAI Action Plan metrics and targets as well as issues related to data integration and interoperability. We do not discuss outcomes (products) and touch only briefly upon the selection and coordination of HAI data and monitoring systems here because these topics are covered thoroughly in other papers in this issue.7–14
We used the Content-Input-Process-Product framework,15 together with the HAI prevention system framework16 described in Kahn et al7 in this issue, to describe the transformative processes associated with data and monitoring efforts designed to support a federal approach toward HAI elimination. A supplemental Online Methods Appendix, (Supplemental Digital Content 1, http://links.lww.com/MLR/A597) provides additional detail about these methods. In this article, we focus on Action Plan goals for data and monitoring1,17 and their responsiveness to the context in which the Action Plan was developed, the Action Plan’s decision making regarding how to use available assets and resources to achieve stated goals (inputs), and strategies used to optimize implementation processes.
To lead the data and monitoring efforts, the Action Plan relied upon the leadership of 3 key working groups.8,17 The Interagency Patient Safety Working Group assumed responsibility for coordinating data and monitoring communications; seeking interfaces with other data and monitoring systems; and facilitating movement toward electronic reporting. The Incentives and Oversight Working Group sought to provide incentives for hospitals and healthcare organizations to participate in data and monitoring efforts; refined current methods of measuring performance; and enhanced surveyor training and tools. The Information Systems and Technology Working Group assisted in developing goals for the Action Plan and in developing data integration strategies.
Five goals pertinent to HAI data and monitoring were developed by the Prevention and Implementation Working Group and the Information Systems and Technology Working Group. As shown in Table 1, the first goal was to identify national metrics with corresponding 5-year prevention targets for 6 HAI priority conditions.
This created a focus for the selection of data and monitoring goals and activities in the 2009 and 2012 versions of the Action Plan. This goal was expanded in 2012 to address the development of national metrics for conditions related to HAIs in ambulatory surgical centers (ASCs) and end-stage renal disease (ESRD) centers, as well as influenza vaccination rates for healthcare personnel and metrics for long-term care.
The second goal focused on leveraging advancements in state-of-the-art information systems and technology. The Action Plan highlighted the need to build bridges between healthcare information systems used for infection control, quality improvement, and patient safety and to embed clinical decision support pertinent to HAI detection into electronic health record (EHR) systems to support HAI prevention reminders and clinical guidelines.
The third goal was to establish a foundation for HAI data integration and interoperability. Recommendations were made for incorporating HAI data elements into large-scale federal and other developmental and testing initiatives supported by agencies such as the American Health Information Community, Health Information Technology Standards, and the Office of the National Coordinator for Health Information Technology (ONC).
The fourth goal recommended that data and monitoring efforts be coordinated through an interagency working group. This working group would seek strategic opportunities to make HHS data systems interoperable.
The fifth major goal was to conduct a comprehensive HAI database inventory to guide plans for integration, interoperability, and alignment in the standardization of data elements needed to track HAIs nationally.
Inputs and Implementation
To transform the goals established by the Interagency Working Group, Action Plan leaders selected key inputs which informed decisions about how data and monitoring would be prioritized, resourced, and supported by stakeholders.
Phased-in Data and Monitoring Approaches
The Steering Committee selected 6 HAIs as priority infections for the first phase of the Action Plan: central line–associated bloodstream infections (CLABSI), catheter-associated urinary tract infections (CAUTI), surgical site infections (SSI), ventilator-associated pneumonias (VAP), Clostridium difficile infections, and methicillin-resistant Staphylococcus aureus (MRSA) infections.17–19 The phased emphasis on selected priority conditions was particularly relevant for data and monitoring, which required alignment and integration across a focused set of conditions, definitions, targets, data systems, and stakeholders. For phase I of the Action Plan, focusing on acute care hospitals, the Action Plan could draw from many existing efforts focused on data integration and interoperability. With phases 2 and 3, however, the Action Plan identified large gaps in data and monitoring systems for assessing HAIs in ambulatory and long-term care settings. This meant that new solutions would be needed to supplement any ongoing approaches that could be leveraged.
Identification of National Metrics and Targets
A second important early decision by Action Plan leadership was the commitment to link the selection of priority HAIs to highly visible national 5-year prevention targets. This decision (and ultimate accomplishment) went far beyond the formal recommendations of the Government Accountability Office, which focused on the process of prioritizing prevention practices.20 The Action Plan commitment to system-level metrics and targets illustrated the determination by Action Plan leadership to make a difference not just with processes but with outcomes. Successful implementation of this plan would require scientific and technological advances to engage stakeholders in HAI prevention throughout the healthcare system. It would require coordination and alignment of metrics, targets, and systems as well as stakeholder engagement. The 6 HAI-specific priority plan areas were associated with a total of 10 goals articulated in the 2009 Action Plan, as shown in Table 2.
CAUTIs were assigned 1 measure/metric. C. difficile, CLABSI, MRSA, and SSIs were each associated with 2 measures/metrics. No measures/metrics were assigned for VAP in 2009 because of multiple challenges identified with constructing the measure, including difficulty in verifying radiographic diagnoses for pneumonia given variability in reporting styles among radiologists and the lack of a VAP “gold standard.”21 The selected metrics were designed to align with existing measures, such as those from the Compendium of Strategies to Prevent Healthcare-associated Infections in Acute Care22 and the National Quality Forum,23 including efforts to assure that each measure of the Action Plan aligned with a National Quality Forum measure.
New Measures and Data Sources for the 2012 Action Plan
The 2012 Action Plan included revised definitions for ventilator-associated events.1,24 In addition, new metrics were introduced for ASCs and ESRD settings and for influenza vaccination of healthcare personnel.1 The latter efforts required coordination of inputs from HAI experts and also alignment with stakeholders, which was slowly and methodically gained through a series of national meetings in which Action Plan leaders and stakeholders from national, regional, state, and local venues iterated on decisions about the best strategies for developing HAI metrics.
ASC: Within the last decade, ASCs have shown an enormous expansion in the number and complexity of procedures performed,25–27 with increasing evidence that HAIs in ASCs have been underrecognized.28,29 In response to a series of infection epidemics noted in 2008,30 Centers for Medicare and Medicaid Services (CMS) and Centers for Disease Control and Prevention (CDC) have worked closely with Ambulatory Surgical Center Association to standardize data collection audit tools, to develop infrastructure and training for infection control personnel in ASCs, and to update ASC conditions for coverage to specifically address the need for infection control and prevention programs.31–33 In 2009, HHS provided $9 million in an attempt to reduce HAIs in ASCs.34 These activities, combined with ASCs being a major focus of the Action Plans’ second phase, have resulted in the scope of the problem of HAIs in ASCs being more systematically identified, an important first step toward systematic data and monitoring and beginning quality improvement.
ESRD Centers: As with ASCs, the selection of ESRD inputs for the Action Plan showed continuity with phase I of the Action Plan (ie, focusing on intravascular infection), but also diverged from phase I by newly focusing on viral hepatitis, a highly prevalent condition among dialysis patients. The 2012 Action Plan proposed 7 metrics to track HAI prevention progress in ESRD facilities. Whenever possible, these metrics were to leverage existing infrastructure, such as National Healthcare Safety Network, which had already been used for dialysis-related infection surveillance. In addition, CDC collaborated with CMS’s development of the ESRD information system, the Consolidated Renal Operations in a Web-Enabled Network (CROWNWeb), to enter and securely submit facility-level and patient-level administrative and clinical quality of care data to CMS in a timely manner. CROWNWeb is the first web-based data collection system established to allow all Medicare-certified ESRD facilities to securely submit dialysis center-level and patient-level data tracking viral hepatitis diagnostic test results and vaccination use data, and bloodstream infections across venues (ie, in dialysis centers and in hospitals) in addition to tracking other clinical and administrative data.
Influenza Vaccination of Healthcare Personnel: In this area, HHS has adopted as its goal the Healthy People 2020 objective of a 90% vaccination rate among healthcare personnel by 2020, as well as an interim target of 70% by 2015.35 These goals are considered aspirational by many because of substantial challenges associated with implementing stated goals. As with VAP, influenza vaccination efforts are limited by a lack of standardization of definitions, in this case for personnel. Alignment of the Action Plan with organizations such as the National Foundation for Infectious Diseases, the Joint Commission, the Society for Healthcare Epidemiology of America, and the Healthcare Infection Control Practices Advisory Committee increases its effectiveness in engaging personnel and healthcare facility administrators.36–37
Development of a Comprehensive HAI Database Inventory and Selection of Data Systems
The interagency working group developed a comprehensive HAI database inventory to guide plans for short-term and long-term integration and interoperability projects, and to establish the extent of definitional alignment and data element standardization needed to link HAI data nationally. This inventory is described in detail in Weinberg and Kahn13 in this issue. The inventory was an important first step in covering surveillance gaps at the national, state, and local levels.
Another set of key decisions for the Action Plan involved the selection of data systems to systematically monitor national, state, and local HAI rates. The mandate to demonstrate coordination across data systems and to show results within a short time window motivated Action Plan leaders to use existing data systems for hospital-based HAIs while acknowledging the need for new data systems or new uses of old systems for phase 2 HAIs. Among available data systems, the surveillance definitions and methodologies associated with the nonadministrative systems are generally perceived as being of higher quality than administrative data systems. However, for nonadministrative data systems, there have been notable concerns regarding databases’ representativeness and consistency in cohort and surveillance definitions. In the context of these challenges, Action Plan leaders selected nonadministrative data sources, using National Healthcare Safety Network (NHSN) for monitoring CLABSI, C. difficile, CAUTI, MRSA bacteremia in the hospital setting, and SSI. The CDC’s Active Bacterial Core Surveillance system is used to monitor invasive MRSA infections at the population level. Only the target for C. difficile hospitalizations was to be measured using the Healthcare Cost and Utilization Project administrative dataset. These issues are discussed further in Weinberg and Kahn.13
Leveraging Advances in Information Systems and Technology
The Action Plan combined forces with agencies and organizations to link healthcare records for HAI reporting across geographic levels spanning federal, state, and local venues. One example is CROWNWeb, the web-based system described above, which incorporates patient-level data for satisfying administrative, payment, clinical, epidemiological, and quality purposes for chronic dialysis patients. A second example is the use of Common Formats to support acute care hospitals and nursing homes in using standardized definitions to report events and “near misses.”
The Action Plan has also supported movement toward embedding clinical decision support into EHR systems to support context-specific HAI prevention reminders or clinical guidelines. Examples include ongoing efforts directed by the ONC to incorporate into meaningful use rules related to clinical notes, vital signs, and electronic prescriptions.
Establishing Foundations for HAI Data Integration and Interoperability
The Information Systems and Technology Working Group proposed modeling its data integration strategies on work that had been performed by existing agencies such as the American Health Information Community, the Healthcare Information Technology Standards Panel, and the ONC. The working group called for developing data architecture that works with the existing Nationwide Health Information Network and Federal Health Information Sharing Environment initiatives. This network of organizations advised HHS and Action Plan leaders about how to accelerate the development and adoption of health information technology.
Ongoing Challenges for Data and Monitoring
The aim of much of the progress described above is to develop an infrastructure and set of processes to support data validation using methods endorsed by a broad coalition of stakeholders. Validation, the assessment of data quality, should include review of data—often multiple types of data, examination of adherence to standardized terms, a shared understanding of transparency and disclosure, and a commitment to education, training, quality improvement, and adequate staffing procedures.38 An important model for data validation is the Consumer Union’s initial model bills for public reporting, which included a requirement for information to be checked for quality and accuracy.39 Funds from the American Recovery and Reinvestment Act have supported HAI data validation by trained state personnel. The Action Plan has sponsored forums for federal and local discussions about the importance of validation, emphasizing the need to engage stakeholders in developing validation strategies commensurate with the importance of topic.
Nevertheless, multiple challenges with validation remain. Both federal and state teams participate in HAI validation efforts, but these are not integrated. Although federal and state approaches could be complementary, given limited resources for validation, there is concern about competition for those resources at a time when their alignment of data sources is crucial.40
Data Integration and Interoperability
A key opportunity for HAI surveillance that has as yet to be fully realized is the use of EHRs and laboratory and other provider data systems. Various external stakeholders pointed to additional work required to reconcile and resolve competing metrics, including measures used by agencies and the private sector. Stakeholders also emphasized the need to reduce reporting burden and increase the usability and timeliness of current national HAI data and monitoring systems. Some pointed to the difficulty of implementing consistent surveillance procedures for particular HAIs, a problem that could be helped by developing automated electronic methods.
A number of external stakeholders discussed the need for greater investment to improve and expand the national HAI surveillance and reporting infrastructure. In the words of a representative from an accreditation organization, “It is very costly to collect information on HAIs; an enormous investment of resources is required—both at the local level and other levels in terms of aggregating and displaying the information.” External stakeholders highlighted the need to develop coordinated surveillance of HAIs across healthcare settings and types of facilities, including challenges in doing surveillance for conditions such as MRSA, C. difficile, and SSIs. An industry representative stated, “We’re a long way to building bridges that let us track infections the whole way through the system.” In this context, substantial enthusiasm continues for incorporating HAI reporting standards into “meaningful use” standards and incentives for providers using EHRs.
On the whole, the goals set forth by the Action Plan were responsive to the multiple contexts that informed it, including the need for prioritization and coordination in measuring HAIs. Given the state of HAI surveillance at the time of the 2009 Action Plan, including barriers to systems integration such as alignment issues and political barriers to data sharing, the level of specificity of the goals and tasks laid out in the Action Plan was appropriate.
However, although most activities have been initiated, few have been completed because the ambitious data and monitoring goals of the Action Plan will require more time before they are fully realized. Furthermore, the Action Plan was introduced to the American healthcare system at a time when transformation from traditional methods for manual HAI case finding toward electronic data collection and monitoring was still nascent. Early on, Action Plan leaders recognized that the success of data and monitoring in supporting HAI elimination depended upon a well-balanced arrangement between technological advances and human and organizational commitments; neither of these elements was satisfactory on their own.
As evaluators, we have noted the importance in advancing the technical aspects of data and monitoring of the 5 system properties described in Kahn et al7: prioritization, coordination and alignment, accountability and incentives, stakeholder engagement, and an awareness of the need for predictable resources. We highlight key areas of progress and challenges in these 5 areas.
Prioritization allowed for a focused approach that encouraged limited resources to be effectively utilized. Beginning with 6 high-priority conditions in the inpatient setting allowed a focus on refinement of metrics, standardization of data sources, and alignment of metrics and standards with clinical evidence about how HAI incidence could be reduced. Selecting one high-priority data source for tracking each HAI metric has encouraged real-time refinements in definitions, specifications, and cohorts to optimize longitudinal assessment of HAI rates.
Coordination and Alignment
Coordination and alignment with ONC and other federal information technology agencies facilitated the goal of building bridges between healthcare information systems used for infection control and those used for the broader quality improvement and safety agenda. Coordination and alignment are also believed to be important support structures for implementing validation activities across individuals and venues. Action Plan goals also addressed the importance of supporting EHR adoption across levels and venues; however, to date, the successful embedding of clinical decision support into EHR systems to support HAI prevention has been introduced in only a small fraction of healthcare settings. CMS’s introduction of CROWNWeb, a national system for information technology that depends upon coordination and alignment of definitions, standards, data systems, and quality metrics, provides a prime example of how this procedure might work more globally.
Accountability and Incentives
Despite substantial advances in the technical sides of data collection and transmission, a major challenge has been engaging hospitals and practice settings to utilize available systems as recommended, given the burden data and monitoring efforts place on providers. In response to this challenge, CDC and CMS coordinated their approaches so that surveillance data submitted to NHSN would also satisfy HAI hospital reporting that Medicare requires for annual payment update determination purposes. The efficiency associated with standardizing and streamlining data entry fields, as well as the incentives provided by linking data entry with payment, has been associated with a rapid increase in institutional submission of data to NHSN. As described further in Mendel et al,8 CMS’s FY 2011 through FY 2013 Medicare inpatient prospective payment system Final Rule includes provisions requiring hospitals to submit CLABSI, SSI, and CAUTI data by NHSN. Currently, nearly all acute care general hospitals, most dialysis units, and most long-term care settings are incentivized to submit data to NHSN using this mechanism.42–45
The Action Plan has balanced the need for a focused and timely process and set of observable outcomes with the need to engage key stakeholders. Although decision making by a few can lead to a more rapid pace, exclusion of key stakeholders can lead to alienation and missed opportunities for efficiency. Leveraging the already functioning Interagency Patient Safety Working Group to serve as an interagency meeting group for HAI information technology concerns served as an efficient means for aligning HAI challenges with a broader set of patient safety challenges while conserving personnel and fiscal resources. In addition, the Action Plan emphasized involvement of CDC, Agency for Healthcare Research and Quality, and CMS in design and implementation, aligned with key epidemiologic organizations, like Society for Healthcare Epidemiology of America and Infectious Disease Society of America, and actively involved the Office of the Assistant Secretary for Health in supporting national, regional, state, and local meetings to encourage bilateral knowledge exchange.
Progress in data and monitoring is essential to support federal, regional, state, and local obligations but also to assure patients receive needed care at the time they need it. Although this concept is widely supported, translation from conceptualization to implementation requires allocation of resources to develop or refine methods for information technology, evaluation and improvement of data reliability and validity, personnel training and support, and coordination of efforts across venues.
Multiple Specific Examples of Challenges That Require Resources
One example is that successful data and monitoring requires data sharing across healthcare venues paired with a commitment to data privacy.41 Aligning data sharing arrangements with the Health Insurance Portability and Accountability Act is an important challenge that will require resources to support a shared understanding of requirements across venues. Another example is the set of challenges faced by the Action Plan in aligning data sources and targets between the Partnership for Patients and the Action Plan, as described in Mendel, et al.8 Although states and the federal government have aligned data reporting with the inpatient prospective payment system, the multiple stakeholders who could benefit from HAI data (eg, patients, providers, public health agencies, patient safety groups) are likely to value different formats and degrees of aggregation. Since the expiration of American Recovery and Reinvestment Act funds, stable sources of funding are rare, and in many settings, the number of staff associated with technical assistance and validation efforts is now considered inadequate. Although with time, the use of rigorous metrics and standardized procedures for use with advanced information technology is likely to mitigate staff requirements, while new systems of care and monitoring are being developed and implemented, more resources are likely to be needed for technical assistance and translation between data and monitoring requirements and patient care and quality improvement efforts.
Optimal strategies for reporting data so that it can serve multiple purposes represent another challenge that is likely to require focused resources. The predictability of resources for training and delivering technical assistance are important concerns for federal agencies and state health departments and hospital associations as they build infrastructures to support reliable and valid HAI data systems.
HAI data and monitoring is part of a data supply chain that originates with a conceptual model for how data collection and reporting by local facilities can support local clinical, epidemiological, and administrative needs, as well as obligations for analysis by payors and at the state and national levels.46 An intricate and extensive set of links need to be secure for this chain to transmit timely, reliable, and valid data in a cost-effective manner. Overall, the Action Plan’s data and monitoring program has built a sound infrastructure that builds upon technological advances and embodies a firm commitment to prioritization, coordination and alignment, accountability and incentives, stakeholder engagement, and an awareness of the need for predictable resources. With time, and adequate resources, it is likely that the investment in infrastructure during the Action Plan’s initial years will reap great rewards.
2. Hibbard J, Greene J, Sofaer S, et al .An experiment shows that a well-designed report on costs and quality can help consumers choose high-value health care.Health Aff (Millwood). 2012; 31:560–568.
3. Smith M, Wright A, Queram C, et al .Public reporting helped drive quality improvement in outpatient diabetes care among Wisconsin physician groups.Health Aff (Millwood). 2012; 31:570–577.
4. Young G .Multistakeholder regional collaboratives have been key drivers of public reporting, but now face challenges.Health Aff (Millwood). 2012; 31:578–584.
5. Fung CH, Lim Y-W, Mattke S, et al .Systematic review: the evidence that publishing patient care performance data improves quality of care.Ann Intern Med. 2008; 148:111–123.
6. . US Government Accountability Office (GAO) .
Health-care-associated infections in hospitals: an overview of state reporting programs and individual hospital initiatives to reduce certain infections, GAO-08-808, September 2008
7. Kahn KL, Mendel P, Weinberg DA, et al .Approach for conducting the longitudinal program evaluation of the US Department of Health and Human Services National Action Plan to prevent healthcare-associated infections: roadmap to elimination.Med Care. 2014; 52:2 suppl 1 S9–S16.
8. Mendel P, Siegel S, Leuschner KJ, et al .The national response for preventing healthcare-associated infections: infrastructure development.Med Care. 2014; 52:2 suppl 1 S17–S24.
9. Kahn KL, Mendel P, Leuschner KH, et al .The national response for preventing healthcare-associated infections: research and adoption of prevention practices.Med Care. 2014; 52:2 suppl 1 S33–S45.
10. Siegel S, Kahn KL .Regional interventions to eliminate healthcare-associated infections.Med Care. 2014; 52:2 suppl 1 S46–S53.
11. Fischer L, Ellingson K, Jernigan J, et al .Driving change through the states: federal-level technical assistance to state health departments for healthcare-associated infection prevention.Med Care. 2014; 52:2 suppl 1 S54–S59.
12. Cataife G, Weinberg DA, Kahn KL .The effect of Surgical Care Improvement Project (SCIP) compliance on surgical site infections (SSI).Med Care. 2014; 52:2 suppl 1 S66–S73.
13. Weinberg DA, Kahn KL .An examination of longitudinal CAUTI, SSI, and CDI rates from key HHS data systems.Med Care. 2014; 52:2 suppl 1 S74–S82.
14. Mendel P, Weinberg DA, Gall EM, et al .The national response for preventing healthcare-associated infections: system capacity and sustainability for improvement.Med Care. 2014; 52:2 suppl 1 S83–S90.
15. Stufflebeam D. Stufflebeam DL, Madaus GF, Kellaghan T .The CIPP model for evaluation.Evaluation Models: Viewpoints on Educational and Human Services Evaluation. 2000; .Boston, MA:Kluwer Academic Publishers.
16. Mendel P, Weissbein D, Weinberg D, et al .Longitudinal Program Evaluation of the HHS Action Plan to Prevent Healthcare-associated Infections: year 1 report. 2011; .Santa Monica:RAND.
17. . US Department of Health and Human Services .Action Plan to Prevent Healthcare-associated Infections. 2009; .Washington, DC:US Department of Health and Human Services.
18. Klevens RM, Edwards JR, Richards CL, et al .Estimating health care-associated infections and deaths in US hospitals, 2002.Public Health Rep. 2007; 122:160–166.
20. . US Government Accountability Office .Leadership needed from HHS to prioritize prevention practices and improve data on these infections. 2008; .Washington, DC:US Government Accountability Office.
21. Klompas M, Kleinman K, Khan Y, et al .Rapid and reproducible surveillance for ventilator-associated pneumonia.Clin Infect Dis. 2012; 54:370–377.
22. Yokoe DS, Mermel LA, Anderson DJ, et al .A compendium of strategies to prevent healthcare-associated infections in acute care hospitals.Infect Control Hosp Epidemiol. 2008; 29:S12–S21.
24. Magill SS, Fridkin SK .Improving surveillance definitions for ventilator-associated pneumonia in an era of public reporting and performance measurement.Clin Infect Dis. 2012; 54:378–380.
25. Manian FA .Surveillance of in alternative settings: exploring the current options.Am J Infect Control. 1997; 25:102–105.
26. Michelson J .Improved detection of orthopaedic surgical site infections occurring in outpatients.Clin Orthop Rel Res. 2005; 433:218–224.
28. Schaefer MK, Jhung M, Dahl M, et al .Infection control assessment of ambulatory surgical centers.JAMA. 2010; 303:2273–2279.
29. Barie PS .Infection control practices in ambulatory surgical centers.JAMA. 2010; 303:2295–2297.
37. Fiore AE, Uyeki T, Broder K, et al .Prevention & control of influenza with vaccines - recommendations of the Advisory Committee on Immunization Practices (ACIP) 2010.MMWR Morb Mortal Wkly Rep. 2010; 59:RR08 1–62.
41. Yee BY, Bartsch SM, Wong KF, et al .Simulation shows hospitals that cooperate on infection control obtain better results than hospitals acting along.Health Aff. 2012; 31:2295–2303.