Ann Scheck McAlearney, Julie Robbins
Central line-associated blood stream infections (CLABSIs) are a leading cause of healthcare-associated infections (HAIs), preventable deaths, and excess health care costs in U.S. hospitals. To address this issue, many hospitals are adopting evidence-based interventions designed to reduce and prevent CLABSIs, but success with intervention implementation has been mixed. Our study was designed to explore the role of management practices in facilitating successful adoption of evidence-based interventions among hospitals participating in statewide CLABSI prevention collaboratives. We completed a multi-site case study investigation including eight hospitals, matched on level of reduction of CLABSI infection rates achieved during their participation in State collaboratives. Over the eight 2-day site visits, we conducted 194 in-person interviews with a mix of clinical and administrative key informants. We found that while our qualitative methodology provided rich insight into factors that affect successful adoption of evidence-based CLABSI prevention standards, we also encountered methodological challenges associated with study design, implementation, and analysis. Taking into account lessons learned—such as the importance of maintaining flexibility and developing an ability to "learn on the fly" during site visits, and being sensitive to the need to manage a large quantity of data and prioritize data analyses—can help guide future studies of HAI prevention and research focusing on implementation of evidence-based interventions. Conducting site visits guided by rigorous qualitative methods is an effective way to study HAI prevention efforts, but future investigations could take into account lessons learned in our study to increase the likelihood that research aims will be achieved.
Central line-associated blood stream infections (CLABSIs) are a leading cause of healthcare-associated infections (HAIs), preventable deaths, and health care costs in hospitals.1,2 In the past decade, the number of CLABSI infections that occur in intensive care units (ICUs) has been reduced significantly due to widespread adoption of evidence-based practices, such as provider education, standardization of processes, and the use of checklists to ensure practice consistency.1–4 Widespread implementation of these practices has been supported through coordinated quality improvement initiatives at the State and local levels.5 These efforts have contributed significantly to an estimated 58 percent decline in CLABSIs among patients hospitalized in ICUs in the United States between 2001 and 2009.5 However, the results have not been uniform; some hospitals have virtually eliminated CLABSIs in their ICUs, while others continue to struggle in their efforts to prevent these infections.6
In an effort to better understand this variation and the factors that contribute to hospitals' successful implementation of evidence-based CLABSI prevention practices, we examined whether and how management and organizational factors facilitate successful CLABSI prevention, an emphasis that has not been considered in the literature. Given the exploratory nature of our inquiry, we used a multi-site case study design and rigorous qualitative methods for this research. We conducted in-depth case studies of CLABSI prevention efforts in eight hospitals to obtain insight into both the facilitators of and barriers to these efforts.
Our study design and methodology constitute an approach that may have applicability for future efforts to study the prevention of CLABSIs and other HAIs. The purpose of this paper is to share insights about the value of this methodological approach for studying HAI prevention in hospitals and to highlight issues that may influence the implementation and impact of this approach. The information and insights we provide should be of immediate use to those involved in efforts to study factors contributing to the success and failures of HAI prevention initiatives, and may have particular value to investigators and policymakers attempting to make sense of the variable levels of success appearing across hospitals, ICUs, and initiatives.
Case Study Sample
As a first step in designing this study, our research team sought to study organizations that had equal access to similar evidence-based CLABSI prevention guidelines. Consequently, we focused our study on hospitals that had participated in the national, AHRQ-funded project "On the CUSP: Stop BSI" (henceforth referred to as CUSP). This project provided funding for State efforts to disseminate and support implementation of evidence-based CLABSI prevention guidelines.
To minimize differences in timing of CUSP implementation and maximize the opportunity to track related CLABSI outcomes, we decided to study organizations from States that had participated in the first two cohorts of CUSP and completed their participation prior to the commencement of our research. Working with the CUSP project leadership team at the Healthcare Research and Educational Trust (HRET), we identified and selected four States that would (1) provide variation with respect to implementation experience and (2) be most likely to support this research and participate in the study.
To select hospitals for the study, we used CUSP project data that included CLABSI rates and hospital characteristics to categorize hospitals from the four targeted States. We aimed to develop a final sample of "contrasting cases."7
Thus, we identified pairs of potential sites that had contrasting CLABSI outcomes but were matched on key organizational characteristics (i.e., size, rural vs. urban, teaching status). Our review of the CLABSI outcomes data showed that hospitals indeed had differential CLABSI prevention outcomes, despite following similar evidence-based intervention protocols; we were able to classify hospitals as "good" or "less good" on this basis. In general, the "good" sites had been able to achieve zero CLABSIs and then sustain this rate, except for the occasional anomaly for a few sites during and following CUSP implementation. By contrast, although many of the "less good" sites had reduced CLABSI rates over time, their data trends suggested that they had experienced difficulty reaching zero CLABSIs, had inconsistent results over time, and/or had inconsistent rates across their ICUs. As a final step, we obtained subjective input from State-level CUSP coordinators to confirm that the sites we selected were appropriate for study and would be likely to agree to participate.
Case Study Key Informants
The main source of data for the study was interviews with key informants, conducted during 2-day site visits to each of the case study organizations; all eight site visits were completed between June 2011 and October 2012. At each participating hospital, the research team identified a single contact person and then worked with that person to coordinate the site visit and the key informant interviews. Appropriate key informants were identified by the hospital contact person, guided by a list of target key informants that had been prepared by the research team. Target key informants for each site included executive leaders (e.g., chief executive officers, chief medical officers), managers involved in CLABSI prevention efforts (e.g., ICU directors and nurse managers), and frontline clinicians (e.g., ICU physicians and nurses) and staff (e.g., infection prevention, quality improvement) directly involved in or impacted by these efforts.
Case Study Data Collection
To ensure consistency in the data collection, we used a standard guide to conduct interviews, while recognizing that the specific focus of the questions would vary based on the informant's role in the organization. The interviews included questions about the organization's approach to CLABSI prevention, facilitators of and barriers to these efforts, considerations about sustainability, and lessons learned. The interviews were recorded and transcribed to ensure accuracy and reliability.
The primary goal of this paper is to provide insight about the applicability of a qualitative case study methodology to investigate CLABSI prevention; the specific findings from our overall research study will be presented elsewhere. Thus, for the results we report in this paper, members of the research team reviewed project notes and reflected on our experiences conducting the multiple site visits and considered the opportunities, issues, and lessons we learned with respect to applying case study methods to investigate HAI prevention initiatives.
In the sections that follow, we present our findings that highlight the value of the case study approach for investigating CLABSI prevention efforts and identify specific challenges and lessons learned that may be applicable to similar efforts in the future.
Value of Case Study Research
First, our sample of eight hospitals proved to be sufficient and robust, offering a breadth of perspectives and rich insight into CLABSI prevention efforts. As a result, we were able to complete in-depth analyses of the key issues that had been the initial focus of our study and to explore issues that emerged throughout the project. The sample also provided sufficient evidence to support "theoretical replication"of themes across sites, thus enhancing the validity of our analyses and the potential generalizability of our findings.7
Second, we found that we reached a point of saturation when conducting our key informant interviews; by the end of each site visit, we were no longer learning any new information from interviews at that site; instead, we were gathering additional evidence that strengthened our findings.8 As a result, we are confident not only that we spoke with a sufficient number of informants at each site, but that we spoke with the "right" informants. This finding confirms the robustness of the sample and adds credibility to our conclusions.
Finally, our research was designed to provide flexibility to explore themes that emerged during the course of the site visits, and this indeed occurred. For instance, while we did not initially expect that exploring differences between ICUs would be feasible or informative, we found that this was possible and enhanced our findings. Overall, our use of a semi-structured interview guide permitted us to maintain consistency in our inquiry across sites but also enabled us to be flexible as we sought to explore emergent issues and themes.
Challenges and Lessons Learned
We encountered challenges throughout all phases of this case study research project—particularly around study design, study implementation, and analysis. We present these challenges, and our strategies for overcoming them, as valuable lessons learned that (1) improved the reliability and validity of our findings, and (2) may inform future, similar research efforts. We summarize these challenges and lessons learned in Table 1 and further describe each individually below.
Table 1. Overview of key challenges and how they were addressed, by research phase
|Research Phase||Key Challenges||How Addressed|
Study Design Challenges
Selecting States. Despite having objective criteria, we quickly realized that subjective factors were equally important in the site selection process. For instance, given that an important factor impacting our research was the ability to successfully recruit States/hospitals to participate in our study, we sought guidance from the CUSP national project office staff (HRET) to identify States that would provide the best fit for our study. As a result, the subjective factors that we considered when selecting States for participation included State-level coordinator receptiveness and willingness to help, competing priorities and initiatives within a State, and perceptions about the level of hospital engagement in CUSP within each State. These factors proved important and influenced our selection process. One State coordinator expressed reluctance to participate in our project because he did not want to ask the participating hospitals to "do one more thing." Another State coordinator expressed enthusiasm for participating in our project but took nearly a year to follow up and recruit individual hospitals because of competing priorities within the State. We addressed these challenges either by working directly with the State coordinators to resolve them or by circling back to the CUSP national project office staff to consider alternatives.
Defining "good" vs. "less good." Although we initially sought to use CUSP project data to identify "good" and "less good" project sites, we quickly realized that we could not rely on the CLABSI rates alone to identify well-matched hospital pairs that would satisfy our research objectives. For example, while many hospitals in the sample had zero CLABSIs during the 18-month study followup period, many of these hospitals also did not have any infections during the 12-month baseline period. This was due largely to their small size and low patient acuity, which contributed to a relatively low risk for CLABSIs overall. Thus, while these hospitals met the criterion for having "good" outcomes, their experiences were clearly not aligned with our research objective; as a result, hospitals with zero CLABSIs during the baseline period were excluded from participation.
A more detailed analysis of the data on CLABSI rates suggested that hospital size or type could affect both baseline and followup infection rates. For instance, many large academic medical centers had baseline CLABSI rates that were well over five infections per 1,000 line-days, while many smaller community hospitals' baseline rates were lower than two infections per 1,000 line-days. Consequently, we decided first to group the hospitals on the basis of these organizational characteristics and then to select the pairs of "good" and "less good" hospitals from these groupings. We also sought subjective input from State-level coordinators, asking for their perspective about which hospital pairs would be a good match for our study and also would be likely to participate. Our experience highlighted that even when objective data are available, a more purposive selection approach focused on overall study goals may be an important consideration if the aims of a study are to be achieved.
Study Implementation Challenges
Securing site participation. We identified several common challenges in attempting to secure hospitals' participation as case study sites. First, our site contacts were typically individuals who had served as lead representatives for their hospitals' participation in the statewide CUSP project. Although very knowledgeable about the CLABSI prevention efforts at their organizations, they typically were staff-level professionals (e.g., infection control, quality improvement) with little research experience; more importantly, they were not empowered to make a hospital-level commitment to participate in our study. To address this challenge, we developed several communication resources (e.g., a one-page description of the project and the benefits of participation) that our site contacts could use to encourage project participation within their hospitals; these resources were well-received and made a difference with respect to our ability to secure commitments to participate.
Another frequent challenge in securing site participation was related to competing priorities within the hospitals and/or their ICUs. For instance, several of the hospitals that we approached either resisted or delayed participation due to conflicting priorities, such as accreditation activities, unit-based electronic health record (EHR) system implementation, and other pressing clinical priorities. In nearly all cases, we were able to secure participation by recognizing the importance of these conflicts, being flexible with our timeline, and being politely persistent when reaching out to sites.
Finally, although our research team had institutional review board (IRB) approvals from our own institutions, at least three of the sites required that we obtain approval from their hospital IRBs prior to beginning our research. This created an extra step in the process and resulted in time delays that had to be accommodated in the project timeline.
Identifying the "right" key informants. Although we had a list of target key informants based on job titles, as previously described, we learned that there was considerable institution-specific variation with respect to identifying the "right" informants who would be able to discuss CLABSI prevention efforts. Key considerations included the degree to which the project was a top leadership priority, physician involvement, and the structure and role of professional staff departments such as infection control and quality improvement. As we gained experience across the sites, we modified our list of target key informants, but we also realized that the best strategy was to clearly communicate that we were interested in talking to the "right" informants at all levels and then to defer to the site-based contacts with respect to their judgment and institutional insight about whom we should interview.
Interviewing frontline staff. Because implementation of evidence-based CLABSI guidelines requires significant changes to workflow and clinical relationships, we were very interested in interviewing frontline clinicians and staff. At the same time, we recognized that the clinical demands on these staff would, and should, take precedence over our research. For example, given the intensity of the ICU workload, it was often difficult for frontline nurses to know whether they could spare 30 minutes to speak with us on any given day or to predict a good time to do so. We addressed this challenge by developing a flexible strategy in which the research team would schedule a block of time to complete interviews in a unit-based location (e.g., conference or break room in the ICU). Then, instead of meeting with a predetermined list of individuals, we were available to interview any staff member(s) that unit leaders had identified as having some slack time during our time block. These interviews often took the form of small group sessions in which participants came in and out based on their clinical demands at the time. In general, this flexibility resulted in some trade-offs between depth and breadth of perspectives (e.g., some interviews were shorter than we would have liked), but the research team ultimately concluded that this approach offered an unobtrusive way to successfully obtain perspectives from frontline clinicians.
Two process management strategies contributed to our success in applying this approach. First, the primary interviewer had to be flexible during the interviews, prioritizing questions based on participants' availability. Second, another member of the research team served as a "greeter" for new participants, introducing the project and completing the informed consent process in a side conversation so as not to interrupt the flow of the ongoing interview. While this interview approach might not be possible for all qualitative investigators, the experience of our research team and the adaptability of our individual interviewers enabled us to successfully modify our standard interview protocol and accomplish the aims of our study.
Learning "on the fly." All members of the research team were scholars with doctorate-level training in health care management and had extensive practical and research experience in a variety of clinical settings. While this mix of skills and experience qualified us to conduct our inquiry into the management factors related to successful CLABSI implementation, no member of the team had specific experience related to ICUs or infection control. Therefore, one of the biggest challenges for the team at the outset of our investigation was the clinical nature of the research context and the highly technical focus of many of our informants (e.g., frontline nurses). To prepare for the site visits, we conducted extensive background research to learn about CLABSI prevention and familiarize ourselves with key clinical terms. Yet, once we started the interviews, we realized that there was often a local clinical language and shorthand (e.g., acronyms and institution-specific terms) that we could not have studied or anticipated prior to visiting the individual sites. We managed this challenge in three ways, by: (1) emphasizing at the outset of each interview that we were not clinicians, thus setting the stage for what we might know and not know; (2) asking clarifying questions when necessary for us to effectively conduct our inquiry regarding the organizational dynamics related to CLABSI prevention; and (3) simply listening and learning "on the fly," distilling what was important in the course of the interview and the site visit through repetition and context. For instance, during one of the early site visits, we were introduced to the term "scrubbing the hub," a shorthand term for a specific infection control practice for maintaining central lines. Although we did not fully understand the clinical details of this practice, we asked for clarification and were able to quickly assess that this was an important component of CLABSI prevention efforts and, in some cases, represented a major change in practice. We carried our learning forward when we heard this term being used in many of the subsequent site visits.
Unexpected variation in sample. Despite our best efforts in the site selection process, we encountered unexpected variation among the hospitals in our sample. Most notable was that one of the sites we had selected as "less good" had made a complete turnaround with respect to CLABSI prevention practices by the time of our site visit. Although this site had struggled during the formal CUSP initiative, the hospital had made major changes since its completion. As a result, by the time of our site visit, this hospital appeared to have become a "good" site, based on its own CLABSI outcomes data and insights about the changes made in the intervening months. While our study benefited from the opportunity to learn from a site that had experienced a turnaround in performance with its CLABSI prevention efforts, this finding challenged our "contrasting cases" study design. As a result, we reclassified this site as "good" in all of our analyses, leaving us with three matched pairs and an additional two "good" sites.
We also observed considerable variation in perceptions between, and even within, the study hospitals regarding their role and participation in the CUSP initiative. In some hospitals, and for some informants, project participation was widely recognized and acknowledged; however, in other hospitals, mentioning anything about CUSP resulted in blank stares from many interviewees. This surprising finding challenged our assumption that selecting sites on the basis of CUSP participation would ensure similar baseline levels of access to information and resources regarding CLABSI prevention. We are conducting more in-depth analyses to determine whether there are any systematic differences in CUSP awareness between the case study sites with "good" and those with "less good" CLABSI outcomes.
Focus and priority. Our analytic priority from the outset was to focus on contrasts between the sites with "good" outcomes and those with "less good" outcomes to identify management factors that influence successful CLABSI prevention efforts. However, as with all exploratory studies, some of the most interesting analytic points emerged as the study progressed—for example, issues related to unit-level variation, differences in perceptions based on organizational roles, and differences in implementation of specific practices. While we knew that the study would provide insight about CLABSI prevention efforts, the research team was able to confirm the potential broader applicability of our findings for other HAI prevention and clinical quality improvement efforts.
Quantity of data. With nearly 200 key informant interviews and over 1,500 pages of transcripts representing a wide variety of perspectives from eight different case study sites, the research team has a very robust qualitative database that will support our ability to complete reliable and valid analyses and widely disseminate our findings to relevant audiences. At the same time, the research team faces the (welcome) challenge of managing and making sense of our abundant data; as a result, we have developed a prioritized list of dissemination targets based on project goals, timeliness and potential impact of results, and breadth of scope. While we have not yet published findings from this recently completed study, we have several manuscripts under review and additional analyses underway. Our overarching analytic priority is to identify the factors that differentiate the hospitals with "good" from those with "less good" outcomes and to develop nuanced insight into the factors that support successful CLABSI prevention efforts. From a management perspective, we have had to develop a strong data management system that includes a systematic process for protecting confidential information, efficiently managing data files, and expanding and building the capabilities of our analytic team to support the necessary work.
Discussion and Conclusion
In this paper, we present an important approach to studying factors that influence hospitals' efforts to prevent and reduce CLABSIs. Our qualitative case study methodology produced data that increased our understanding of the organizational factors and management strategies that affect CLABSI prevention efforts. These findings have direct applicability to organizations seeking to prevent CLABSIs but also may provide insight into other similar efforts related to HAIs or other patient safety priorities.
The goal of our paper has been to share what we have learned about the value of this method, the associated challenges, and the "lessons learned." Based on our experience with this research, we believe that our study design and method could be effectively employed by researchers seeking to study the implementation of other evidence-based clinical interventions; further, this approach may be applicable in non-hospital contexts to the extent that consistent clinical guidelines exist and are widely disseminated in those settings. We have highlighted some of the ways in which site visits and qualitative methods can help achieve the aims of HAI prevention research and hope that these insights will be useful for other researchers seeking to pursue this kind of research.
This project was funded under contract no. HHSA290200600022 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services. 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. Conduct of this research was reviewed and approved by the institutional review boards of the respective authors.
We thank co-investigator Andrew Garman, Professor of Health Systems Management at Rush University, for his contribution and insights throughout this research, and Alexandra Moss and Pamela Beavers, both affiliated with the Department of Family Medicine at The Ohio State University, for their help preparing data for this study. In addition, we are very grateful to the hospitals and individuals who participated in this study.
Department of Family Medicine, College of Medicine (ASM, JR); Division of Health Services Management and Policy, College of Public Health (ASM); The Ohio State University, Columbus, OH.
Address correspondence to: Ann Scheck McAlearney, Sc.D., Professor and Vice Chair for Research, Department of Family Medicine, College of Medicine, The Ohio State University, 2231 North High Street, 273 Northwood and High Building, Columbus, OH 43201; Email: Ann.McAlearney@osumc.edu.
1. Berenholtz SM, Pronovost PJ, Lipsett PA, et al. Eliminating catheter-related bloodstream infections in the intensive care unit. Crit Care Med 2004 Oct;32(10):2014-20. PMID: 15483409.
2. Pronovost P, Needham D, Berenholtz S, et al. An intervention to decrease catheter-related bloodstream infections in the ICU. New Engl J Med 2006 Dec 28;355(26):2725-32. PMID: 17192537.
3. Agency for Healthcare Research and Quality. Healthcare Innovations Exchange. A National Campaign to Scale up and Spread CUSP-CLABSI: Interviews with James B. Battles and Stephen Hines. www.innovations.ahrq.gov/content.aspx?ID=3356&utm_source=issueanc&utm_medium=email&utm_campaign=20111207. Accessed December 7, 2011.
4. Pronovost PJ, Berenholtz SM, Goeschel C, et al. Improving patient safety in intensive care units in Michigan. J Crit Care 2008 Jun;23(2):207-21. PMID: 18538214.
5. 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):243-8. www.cdc.gov/mmwr/preview/mmwrhtml/mm6008a4.htm.
6. Agency for Healthcare Research and Quality. Eliminating CLABSI, A National Patient Safety Imperative: A progress report on the national On the CUSP: Stop BSI project. AHRQ Pub. 11-0037-EF. Apr 2011. www.ahrq.gov/professionals/quality-patient-safety/cusp/onthecusprpt/index.html. Accessed August 25, 2011.
7. Yin RK. Case Study Research: Design and Methods. 4th ed. Thousand Oaks, CA: Sage Publications; 2008.
8. Corbin, J, Strauss, A. Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. 3rd ed. Thousand Oaks, CA: Sage Publications; 2007.