Factors related to use of the centers for disease control guidelines for prevention of central line infections in bone marrow transplant patients

Central vascular catheters (CVCs) are essential for patients receiving a bone marrow transplant (BMT). Central line infections (CLI) pose a serious threat to BMT patients, causing increases in morbidity, financial expenses, and even treatment-related mortality. Determining best practices in prevention of CLI in BMT patients is vital but understudied. The Centers for Disease Control and Prevention (CDC) has published guidelines, including four specific performance indicators (Pis), for prevention of intravascular catheter-related infections. To date, no published study has examined use of the Pis in the CDC guidelines in the care of BMT patients. The purpose of this study was to examine practice patterns among US BMT centers and to determine whether select recommendations in the CDC guidelines are used to prevent CLI in BMT patients. Specific aims of this study were to (a) describe selfreported practice patterns for prevention of CLI in participating US BMT centers, (b) to determine the extent of use of the Pis, (c) to determine which other practice factors are associated with self-reported use of the Pis, and (d) to describe self-reported rates of CLI in participating BMT centers. A survey was mailed to all US BMT centers listed on the BMT InfoNet website. Sixty of 189 BMT centers (32%) participated by completing and returning the survey. The survey assessed practices related to prevention of CLI in BMT patients. Results of the study indicated (a) that practice patterns for CLI prevention vary among participating BMT centers, (b) that overall self-reported use of three of the four CDC Pis was generally high, (c) that six factors were associated with self-reported use of the Pis, and (d) that most participants could not or would not report CLI incidence rates for their BMT centers. CLI in BMT patients can be a matter of life and death. Improving patient outcomes by decreasing rates of CLI depends upon tracking and trending rates of CLI, strengthening roles and coordination within the BMT care team, and strengthening BMT center healthcare worker education programs.

FACTORS RELATED TO USE OF THE CENTERS FOR DISEASE CONTROL GUIDELINES FOR PREVENTION OF CENTRAL LINE INFECTIONS IN BONE MARROW TRANSPLANT PATIENTS by Carolyn S. Bearnson A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy College of Nursing The University of Utah December 2009 Copyright © Carolyn S. Bearnson 2009 All Rights Reserved THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by Carolyn S. Bearnson This dissertation has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. '! i j o <{ O' tf*-f11 Bob Wong 6 / C # Clyde D. Ford m l o < j Matthew H. Samore THE UNIVERSITY OF UTAH GRADUATE SCHOOL FINAL READING APPROVAL To the Graduate Council of the University of Utah: I have read the dissertation of Carolyn S. Bearnson in i t s f m a l form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. v W i ? Date Anita Kinney Chair: Supervisory Comr Approved for the Major Department Approved for the Graduate Council g, David S. Chapman Dean of The Graduate School ABSTRACT Central vascular catheters (CVCs) are essential for patients receiving a bone marrow transplant (BMT). Central line infections (CLI) pose a serious threat to BMT patients, causing increases in morbidity, financial expenses, and even treatment-related mortality. Determining best practices in prevention of CLI in BMT patients is vital but understudied. The Centers for Disease Control and Prevention (CDC) has published guidelines, including four specific performance indicators (Pis), for prevention of intravascular catheter-related infections. To date, no published study has examined use of the Pis in the CDC guidelines in the care of BMT patients. The purpose of this study was to examine practice patterns among US BMT centers and to determine whether select recommendations in the CDC guidelines are used to prevent CLI in BMT patients. Specific aims of this study were to (a) describe self-reported practice patterns for prevention of CLI in participating US BMT centers, (b) to determine the extent of use of the Pis, (c) to determine which other practice factors are associated with self-reported use of the Pis, and (d) to describe self-reported rates of CLI in participating BMT centers. A survey was mailed to all US BMT centers listed on the BMT InfoNet website. Sixty of 189 BMT centers (32%) participated by completing and returning the survey. The survey assessed practices related to prevention of CLI in BMT patients. Results of the study indicated (a) that practice patterns for CLI prevention vary among participating BMT centers, (b) that overall self-reported use of three of the four CDC Pis was generally high, (c) that six factors were associated with self-reported use of the Pis, and (d) that most participants could not or would not report CLI incidence rates for their BMT centers. CLI in BMT patients can be a matter of life and death. Improving patient outcomes by decreasing rates of CLI depends upon tracking and trending rates of CLI, strengthening roles and coordination within the BMT care team, and strengthening BMT center healthcare worker education programs. This study is dedicated to the bone marrow transplant patients who have enriched my life and contributed to my deep caring about improving the quality of their care. TABLE OF CONTENTS Page ABSTRACT iv LIST OF FIGURES ix LIST OF TABLES x ACKNOWLEDGMENTS xii Chapter I. INTRODUCTION AND STATEMENT OF THE PROBLEM 1 Central Lines in Bone Marrow Transplant Patients 1 Specific Aims 3 Conceptual Framework 5 II. LITERATURE REVIEW 10 Best Practice and Benchmarking 10 Best Practice vs. Evidence-based Practice 10 Best Practice in Prevention of CLI in BMT Patients 11 CLI in the General Patient Population 14 CLI in ICU Patients 17 CLI in BMT Patients 19 Evidence of Practice Pattern Variation in Prevention of CLI in BMT Patients 21 Guidelines for Prevention of CLI 30 Other Factors 31 Use of Guidelines in Prevention of CLI 34 III. DESIGN AND METHODOLOGY 40 Study Design 40 Research Questions 42 Population and Setting 43 Instrument 57 Chapter Page Pilot Study 65 Main Study 69 Procedure 83 Data Management 90 Data Analysis 91 IV. RESULTS 103 Population and Sample Descriptions 103 Analysis of Survey Data 106 Summary of Results 132 V. DISCUSSION 134 Study Findings 134 Implications of the Study Findings 139 Limitations 151 Conclusion 156 Appendices A. BMT LIFE SURVEY 158 B. COVER LETTER 163 C. CDC GUIDELINES 166 REFERENCES 210 viii LIST OF FIGURES Figure Page 1. Conceptual model of infection prevention 6 2. Conceptual model of infection prevention showing bone marrow transplant center structure, process and outcomes related to select recommendations in the CDC guidelines for prevention of intravascular catheter-related infections 8 3. Conceptual model for the study design 44 4. Percent of participants who self-reported use of each of the five CDC-recommended sterile barrier items when line insertions were performed in the operating room, in radiology, and at the patient's bedside 119 5. Bar chart showing frequencies of central line infections (CLIs) per 1000 catheter days reported for the year 2005 by 14 participants 133 6. Points for research and practice improvement throughout the conceptual model of infection prevention used for the present study 149 LIST OF TABLES Table Page 1. Performance Indicators in the CDC Guidelines for the Prevention of Intravascular Catheter-Related Infections (2002a) 4 2. Alignment of the Study Conceptual Model of Infection Prevention, the CDC Recommendations Related to the Performance Indicators, and the Survey Content Areas 9 3. Results of a Literature Search for Best Practice Studies of Central Line Infections in Bone Marrow Transplant Patients 12 4. Incidence, Attributable Mortality, and Cost of Central Venous Catheter-Related Infection 16 5. Rates of Central Line Infections Reported in Forty-Two Studies of Bone Marrow and Stem Cell Transplantation 22 6. Summary of Infection Descriptions and Rates of Central Line Infection in 42 Studies 29 7. Results of a Literature Search on CLI and the Use of Guidelines 35 8. Results of a Literature Search on CLI and the Use of Guidelines in BMT Patients 39 9. Response Rates in Recent Nursing and Allied Healthcare Studies Using Mailed Surveys 47 10. Response Rates in Recent Nursing and Allied Healthcare Studies Using Telephone Surveys 48 11. Response Rates in Studies Using Dillman's Tailored Design Method for Mailed Surveys of Healthcare Providers 51 12. Detectable Effect Size (r) for Various Levels of Response Rate with Alpha = .05 (two-tailed) and Power = .80 56 Table Page 13. Question(s) Used to Calculate Use of Each CDC Recommendation 72 14. Alignment of Survey Questions with Specific CDC Recommendations and the Performance Indicators 78 15. Response Rate of Participants Following Mailings and Follow-up Contacts 89 16. Use Scores to be Calculated from the Data Collected in the BMT LIFE Survey 93 17. Analyses Comparing Participants and Nonparticipants According to Demographic and Organizational Characteristics of the Bone Marrow Transplant Centers 105 18. Relationships Between the Aims of the Study, the Practice Factors (IVs) and the Performance Indicators (Pis) in the 2002 CDC Guidelines for Prevention of Intravascular Catheter-Related Infection, and the Analytic Approach for this Study 107 19. Participant Responses to BMT LIFE Survey Questions About Performance Indicator 1 - Education of Healthcare Workers 116 20. Participant Responses to BMT LIFE Survey Questions About Performance Indicator 2 - Use of Maximal Sterile Barriers 118 21. Spearman Correlations Among All the Factors (Independent Variables) 124 22. Spearman Correlations Between the Independent Variables (Demographic and Organizational Factors and Practice Factors) and the Dependent Variables (Use of the Four Performance Indicators and Overall Mean Use-Total) 125 23. Multiple Linear Regression of the Significantly Related Independent Variables on the Four CDC Performance Indicators (Pis) and Overall Mean Use-Total (OMU-T) 127 24. Comparison of Raw p-Values and p-Values Adjusted by the Bonferroni-Holm Procedure for n = 6 Associations Between Performance Indicators and the Independent Variables 131 xi ACKNOWLEDGMENTS I gratefully acknowledge the contributions of several individuals who guided and supported the work of this dissertation. Anita Kinney provided wise direction and counsel through every phase of the work. She continually both encouraged and prodded, as needed, helping me to grow into the work and to set and reach attainable goals along the way. Without her persistent help and encouragement this study would never have reached completion. Kathi Mooney helped in the initial phase of planning the study and gave formative critiques of the work as it progressed. Clyde Ford first caught the investigator's interest in searching and reading the literature about central line infections. He remained interested and encouraging throughout the process. Matthew Samore helped me to get out of the details and look at the bigger picture. His comments and suggestions were thought provoking and insightful. Bob Wong spent considerable time coaching me in the process of analyzing and interpreting the study data. His patient, ongoing assistance was invaluable. Linda Edelman enthusiastically joined my supervisory committee in the last stages and made thoughtful and appropriate suggestions for improvement of the dissertation. My husband, Lee, cheerfully served as research assistant and consultant, and both he and my children were unfailingly enthusiastic and encouraging in their support to me in this whole long process. I express grateful appreciation for financial support of the study which I received from the Brigham Young University (BYU) College of Nursing Research Committee and from the Dr. Elaine Dyer Research Endowment at BYU. xiii CHAPTER I INTRODUCTION AND STATEMENT OF THE PROBLEM Central Lines in Bone Marrow Transplant Patients Autologous bone marrow transplant (BMT) patients, persons who receive their own bone marrow or stem cells, are given life-threatening high doses of chemotherapy to destroy cancer cells. Allogeneic BMT patients, persons who receive donated bone marrow or stem cells, also receive life-threatening doses of chemotherapy and, additionally, receive chemotherapy and/or radiotherapy to destroy their immune systems so they will not reject the donor cells. High doses of drugs or anticancer agents are given intravenously (IV) and must be given in large veins where they are quickly diluted with blood to avoid irritation to the veins and infiltration of chemotherapy into the surrounding tissues, causing severe tissue damage. In many treatment regimens, multiple infusions of chemotherapy agents are given at the same time. Using peripheral veins for chemotherapy may require frequent changes of IV sites, as small veins do not tolerate toxic agents well. Central lines used for BMT patients generally have three ports (access sites) and lumens (separate channels through which fluids may be infused simultaneously), so multiple infusions can be given at the same time without mixing with each other. The use of central lines avoids the risks of small vein and tissue damage and the need for multiple IV sites. Throughout treatment, BMT patients have frequent blood draws for lab tests. These blood draws can be taken from a central line without inflicting the pain of a needle stick on the patient. Multiple supportive infusions are an important part of the BMT treatment regimen. Supportive infusions include (a) antibiotics, (b) fluids and electrolytes, (c) total parenteral nutrition (TPN) during periods of extreme nausea and mucosal damage, (d) blood products, and (e) various other infusions. Accommodating all of these therapies without a central line is not feasible. Thus central lines are not an option; they are essential. Central line infections (CLI) in BMT patients pose serious threats to patients. Due to the high-dose chemotherapy and/or radiotherapy, BMT patients are immunosuppressed and at considerable increased risk for CLIs during a significant portion of their time in treatment. A CLI can result in (a) increased morbidity, (b) delay of treatment, (c) increased financial expenses related to increased length of stay (LOS) and use of hospital resources, and even (d) increased treatment-related mortality. Determining best practices in prevention of CLI in BMT patients is essential to improving both the care and the survival of patients in this significantly at-risk patient population. Use of guidelines for the prevention of CLI is presumed to effectively reduce rates of CLI. In 2002, the Centers for Disease Control and Prevention (CDC) published Guidelines for the Prevention of Intravascular Catheter-Related Infections (hereafter known as the CDC guidelines [2002a]). These guidelines contain four performance indicators (Pis) to be used to evaluate the impact of the CDC guidelines on individual institutions. The four Pis are shown in Table 1. There is no currently published literature to support the presumption that use of the CDC guidelines reduces rates of CLI in the BMT patient population. Specific Aims Identifying and implementing the most effective interventions for preventing CLI in BMT patients is critical. Preventing CLI will decrease the mortality, morbidity, and financial cost attributable to CLI in BMT patients. The purpose of this study was twofold: (a) to examine variations in practice patterns among US BMT centers and (b) to determine whether select recommendations in the 2002 CDC guidelines are used in a specific, high-risk population of patients, namely BMT patients. The specific aims of this study are (a) to describe self-reported practice patterns for prevention of CLI in participating US BMT centers, (b) to determine the level of self-reported use of the Pis in the 2002 CDC guidelines among participating US BMT centers, (c) to determine which of the practice factors are associated with self-reported levels of use of the Pis in the 2002 CDC guidelines, and (d) to describe self-reported incidence rates of CLI in participating BMT centers. Findings of this study will provide information not currently available in any published study. No published study of US BMT centers has examined the extent or the outcomes of use of either the broad range of practices or to the Pis in the CDC guidelines for prevention of CLI (2002a). The findings of this study may be used to identify specific performance targets for improvement of nursing and medical Table 1. Performance Indicators in the CDC Guidelines for the Prevention of Intravascular Catheter-Related Infections (2002a) Performance Indicators Abbreviations Essential Elements Described in the CDC Guidelines (2002a, p. 14) Educating Healthcare Workers EHW Implement both didactic and interactive educational programs for healthcare providers who insert and maintain catheters. Maximal Sterile MSB Use maximal sterile barrier precautions when Barriers placing catheters. 2% Chlorhexidine 2%C Use chlorhexidine to prep skin at insertion site. Discontinuing Catheters DC Track rates of removing catheters when they are no longer medically necessary. management of central lines in BMT patients. The ultimate outcome of this study is identification of and support for those practices and practice changes that can effectively reduce morbidity, LOS in the hospital, financial expenses, and mortality due to CLIs in BMT patients. Conceptual Framework The study was based on a conceptual model derived from two of the three levels of infection prevention described by Valanis (1999), as shown in Figure 1. This model acknowledges that primary and secondary prevention are important concepts in CLI. While primary prevention (preventing occurrence of CLI) is the main focus of this study, secondary prevention (early detection, appropriate treatment, and on-going education regarding CLI) are also vital parts of reducing the significant deleterious sequelae of CLI in BMT patients. Tertiary prevention (limiting disability and addressing rehabilitation) is not part of this study and, therefore, is not depicted in the conceptual model. This model has not previously been used in studies of CLI in BMT patients. Recommendations in the CDC guidelines deal mostly with three aspects of the model of infection prevention: (a) breaking the chain of transmission (one aspect of primary prevention), (b) education (one aspect of secondary prevention), and (c) early detection (another aspect of secondary prevention). Three of the Pis address breaking the chain of transmission, one PI addresses education, and one additional CDC recommendation addresses early detection. The study addressed use by participating US BMT centers of the CDC guidelines by examining self-reported BMT center structure, processes, and outcomes related to Figure I. Conceptual model of infection prevention. Note: Highlighted boxes are the preventive strategies that were studied in this research. 7 select recommendations in the guidelines, as shown in Figure 2. This figure is based on Donabedian's model of structure, process, and outcome (1980). Structure, in this context, refers to the physical and human resources, such as BMT center administration, policies, and staff, which influence delivery of healthcare. Process refers to functions, such as quality improvement, staff education, and data collection, which utilize the structures. Outcomes are the results of the processes, which include practice patterns and rates of CLI in BMT patients (Kunkel, Rosenqvist, & Westerling, 2007). A survey instrument (BMT LIFE) was developed specifically for this study. Each CDC recommendation that relates to one of the Pis is addressed in the BMT LIFE. Alignment of the conceptual model, the CDC recommendations related to the Pis, and the survey content areas is shown in Table 2. INFECTION PREVENTION PRIMARY PREVENTION- Breaking the Chain of Transmission STRUCTURE SECONDARY PREVENTION Education STRUCTURE External -P- Internal •CDC Guidelines and •Hospital performance policies and indicators (Pis) for procedures prevention of central (P&P) for CLI line infections (CLI) prevention External - • Internal •Literature on evidence-based practice •Nursing programs •Medical library •Education department •Bone marrow transplant center staff PROCESS PROCESS T OUTCOME Decreased CLI rates Early Detection STRUCTURE External -•Internal •National/region al data banks and surveillance systems, such as NNIS and NHSN •Infection control department •Infection surveillance PROCESS •Assess and assure use of P&P and Pis INTERMEDIATE OUTCOMES •Educate staff about prevention of CLI •Assess CVC practice patterns •Practice patterns for prevention of C L I , •Use of P&P and of Pis •CLI rates available •Calculate and track CLI incidence rates Figure 2. Conceptual model of infection prevention showing bone marrow transplant center structure, process and outcomes related to select recommendations in the CDC guidelines for prevention of intravascular catheter-related infections. Abbreviations: CDC = Centers for Disease Control and Prevention; NNIS = National Nosocomial Infection Surveillance System: CVC = Central venous catheter 9 Table 2. Alignment of the Study Conceptual Model of Infection Prevention, the CDC Recommendations Related to the Performance Indicators, and the Survey Content Areas Conceptual Model of Infection Prevention 2002 CDC Recommendations Related to the Performance Indicators BMT LIFE Survey Content Areas 1 - Primary Prevention Approaches A - Breaking the Chain of Transmission B - Inactivating the Infectious Agent C - Increasing Host Resistance 2 - Secondary Prevention Approaches A - Case Finding B - Education General - VIII. B. Prompt removal of nonessential IV catheters CVC - IV. Maximal sterile barrier precautions General - VI. A. Cutaneous antisepsis N/A N/A N/A General - I. A. Educating healthcare workers General - I. B. Assessing knowledge and usage of healthcare workers Section D. Catheter insertion, removal, and days in place Section D. Catheter insertion, removal, and days in place Section E. Catheter site and catheter care Section A. Healthcare workers Section A. Healthcare workers Note: General = general recommendations for use with all intravascular catheters; CVC = recommendations to be use specifically with central venous catheters; N/A = not addressed by the performance indicators; Roman numerals refer to numbered sections in the CDC guidelines. CHAPTER II LITERATURE REVIEW Best Practice and Benchmarking The concepts of best practice and benchmarking originated in the business world, where best practice refers to documented strategies used by companies considered "best-in- class" in a specific area. "Benchmarking" is a process of seeking and studying best practice strategies for use in quality improvement programs (Bogan & English, 1994). Best Practice vs. Evidence-based Practice A common trigger for nursing and medical research is the discovery of varying rates of successful outcomes related to a given procedure. Studies comparing outcomes of different methods for performing the procedure may produce evidence indicating which of several methods is associated with the best outcomes, after controlling for extraneous variables. When the evidence is compelling enough to change practice, or to establish that one method is clearly superior to all others, the superior method may be considered best practice. Application of best practices that are supported by a significant body of careful research results in evidence-based practice. Development of evidence-based guidelines is a high priority in any industry concerned with the safety of employees and/or consumers. Healthcare applications of evidence-based guidelines are being utilized in many settings to improve the outcomes of patient care. Quality improvement efforts are particularly appropriate where patient morbidity and mortality and high costs for healthcare services are at issue. Best Practice in Prevention of CLI in BMT Patients A literature search using Pub Med (National Library of Medicine, n.d.) was conducted to determine the extent of published research or clinical information related to best practice in prevention of CLI in BMT patients. Limits imposed on the search were publication date from January 1, 1990 to April 9, 2007, only items with abstracts, in English, and with human subjects. The results, shown in Table 3, demonstrate the current lack of published information about benchmarking, best practice, or evidence-based practice related to prevention of CLI in BMT patients. Search number seven in Table 3 included search terms related to (a) evidence-based practice, practice pattern variation, best practice, critical appraisals, and bench marking, plus (b) multiple terms describing central line infections. Forty-seven articles met both sets of terms; 10 were studies, 19 were review articles, 4 were clinical reports, and 14 turned out to be totally unrelated or only minimally related to best practice in CLI prevention. The study topics included implementation of evidence-based practice (Hatler et al., 2006; Warren, Cosgrove, et al., 2006; Young, Commiskey, & Wilson, 2006); decreases in CLI and CLI-related mortality and costs with use of evidence-based guidelines (Berenholtz et al., 2004; Render et al., 2006); lack of coordination between research findings, policies for infection prevention, and current best practice (Jones, 2006; Warren, Yokoe, et al., 2006); the effects of a behavioral educational intervention to increase use of evidence-based CVC management guidelines (Coopersmith et al., 2004); the use of prophylactic antibiotics for gastrointestinal procedures in children with CVCs 12 Table 3. Results of a Literature Search for Best Practice Studies of Central Line Infections in Bone Marrow Transplant Patients Search Number Search Strategy 1 Evidence based practice OR practice pattern varia* OR best practice OR critical appraisal OR bench mark* 21,403 2 Bone marrow transplant OR stem cell transplant OR blood and marrow transplant OR blood cell transplant 32,545 3 Central venous access device OR central line OR central vascular access device OR CVAD OR Hickman OR right atrial catheter OR central venous catheter 16,316 4 Central line infection OR intravascular catheter related infection OR intravascular device related infection OR catheter related bloodstream infection OR exit site infection OR tunnel infection OR pocket infection OR catheter colonization OR catheter associated bloodstream infection OR infusate related bloodstream infection 3,253 5 #1 AND #2 78 6 #1 AND #3 117 7 #1 AND #4 47 8 #2 AND #3 306 9 #2 AND #4 71 10 #2 AND #3 AND #4 56 11 #1 AND #3 AND #4 33 12 #1 AND #2 AND #3 2 13 #1 AND #2 AND #4 0 Note: varia* was used as a search term to pick up derivatives such as variation, variability, etc.; mark* was used as a search term to pick up mark, marking, etc. Number of Articles 13 or other infection risk factors (Snyder & Bratton, 2002); and the optimum time for removing central lines in bacteremic neonates (Benjamin et al, 2001). The review articles focused on (a) risk factors for CLI in pediatric patients (Kline, 2005); (b) cause, diagnosis, and management of CLI (Lee & Johnston, 2006; Mermel et al., 2001; Slaughter, 2004); (c) specific methods of infection prevention (Adams & Elliot, 2007; Bagnall-Reeb, 2004; Bearman, Munro, Sessler, & Wenzel, 2006); (d) evidence and guidelines for CLI prevention strategies (O'Grady, 2002; O'Grady et al., 2002a; O'Grady et al., 2002b); (e) an analysis of patient safety practices (Shojania et al., 2001); (f) the practice of scheduled replacement of CVCs (Timsit, 2000); (g) an evidence-based practice model for CVC selection and site care (Woods, Nass, & Deisch, 2000); (h) the need for literature which addresses the nursing aspects of CVC management (Cook, 1999); (i) the accountability of healthcare providers for control of CLI (Vost & Longstaff, 1997); (j) the best method for culturing catheters (Rello, Jubert, Esandi, & Valles, 1997); (k) placement choices for CVCs in children (Hollyoak, Ong, & Leditschke, 1997); (1) the use of personal digital assistant infectious disease applications (Miller, Beattie, & Butt, 2003); and (m) an institutional protocol for obtaining blood cultures in cancer patients suspected of having an infection (Penwarden & Montgomery, 2002). The clinical reports addressed (a) positive outcomes of using evidence-based central line care practice in pediatric intensive care units (Morgan & Thomas, 2007); (b) the need for standardization in infection control policy and practice (Morritt et al., 2006); (c) the relationship between compliance with best practice and (1) nurse/physician management and (2) quality improvement programs (Earsing, Hobson, & White, 2005); and (d) the possibility that sequential antibiotic administration for other infections may also decrease the incidence of CLI (Wilcox, 1998). In general, these studies, review articles, and clinical reports shed light on the problems associated with CLI and on the use of evidence-based practices which show promise in reducing rates of CLI. The more recent articles, those published since 2002, show a beginning trend to link evidence-based practices to reductions in CLI. Searches number 12 and 13 yielded meager results. Using the search terms related to evidence-based practice, BMT, and CVCs, yielded two articles: a review of guidelines for managing high-dose-therapy-induced neutropenia (West & Mitchell, 2004) and a case study and review of the literature on pinch-off syndrome - which is a precursor to and cause of mechanical failure in CVCs (Fazeny-Dorner et al., 2003). No article met the search criteria of best practice for CLI in BMT patients. Following completion of data collection and analysis, a follow-up literature review showed one article which did meet the criteria of best practice for CLI in BMT patients (Warren, Cosgrove, et al., 2006). However, the study was a multicenter study of 12 intensive care units and 1 BMT unit, and the 13 units were identified only as Unit A, Unit B, Unit C, and so on. Thus the deidentified unit data precluded relating the findings specifically to BMT patients. The particular vulnerability of BMT patients to CLI highlights the need for examining the use of evidence-based practice to reduce CLI in BMT patients. CLI in the General Patient Population CLI causes significant morbidity, mortality, and financial expense in hospitalized patients. This is a particular concern for nursing because use and care of central lines is generally the responsibility of nurses (Roach, Larson, Cohran, & Bartlett, 1995). Morbid effects reported from CLI range from simple phlebitis to septicemia. Estimates are that mortality from nosocomial line infections exceeds 20% in all populations (Charalambous, Swoboda, Dick, Perl, & Lipsett, 1998; Crow, 1996). Table 4 summarizes incidence (the rate of occurrence), attributable mortality (the excess rate of mortality which is due, in a specific patient population, to CLI), and cost data related to CLI reported in one meta-analysis of 13 studies conducted in the United States (n = 8), Germany (n = 2), England (n - 2), and Western Australia (n = 1); and similar data from three US research studies. Nosocomial bloodstream infections (BSI) increased two- to three-fold from 1985 to 1995, with the biggest increases occurring in large teaching hospitals. The rising incidence rates are generally attributed to the increased use of central lines, as about 90% of IV catheter-related infections occur in patients with central lines (Cohran, Larson, Roach, Blane, & Pierce, 1996; Crow, 1996). The CDC (2002a) estimated there were approximately 80,000 central venous catheter- (CVC-) associated BSIs each year in the United States, costing $296 million to $2.3 billion annually for the care of these patients. The Pis in the CDC guidelines (2002a) provide a convenient means for assessing adoption of the guidelines. The literature addressing central line infections and the CDC Pis is sparse. A search using those terms (the central line infection search terms used in search #4 shown in Table 2, and the search term "performance indicators") yielded six publications. Two articles presented the 2002 CDC guideline document (O'Grady et al., 2002a; O'Grady et al., 2002b). One article (Mermel et al. 2001) gives guidelines and Pis for management, but not prevention, of CLI. The Pis in this article were related to the diagnosis and treatment of CLI. An article, by Rudy, Lucke, Whitman, and Davidson Table 4. Incidence, Attributable Mortality, and Cost of Central Venous Catheter-Related Infection Type of Article: Authors Incidence Meta-Analysis: Saint, Veenstra, 42.7% & Lipsky (2000) Research Report: Pittet, Tarara, N/A & Wenzel (1994); Pittet & Wenzel [letter] (1994) Research Report: Smith, N/A Meixler, & Simberkoff (1991) Research Report: Digiovine, N/A Chenoweth, Watts, & Higgins (1999) Attributable Cost of Cost of Mortality Local Bloodstream Infection Infection 4-20% $399 $6,005-9,738 25% N/A $28,690 28% N/A N/A 4.4% N/A $16,000 (2001), described three different methods of benchmarking for CLI in eight hospitals within a healthcare system in 1998. Braun et al. (2003) used bloodstream infection rates as performance indicators for an evaluation of processes and indicators in infection control (EPIC) study. The variations they found in specifications of infection rates limited the usefulness of rates for comparison and improvement. The sixth article addressed performance status rather than performance indicators. No study published since the release of the CDC 2002 guidelines describes the use of all four of the 2002 CDC Pis to evaluate the impact of adoption of the CDC guidelines. CLI in ICU Patients The National Nosocomial Infections Surveillance (NNIS) System, a division of the CDC, tracked and reported rates of hospital-acquired infections. Organized in 1970, the NNIS System began with 62 United States hospitals and grew to around 300 hospitals. The NNIS System unit of measurement for reporting CLI was central line-associated BSI per 1000 catheter days. The NNIS reports for 2001 through 2004 (CDC, 2001b; CDC, 2002b; CDC, 2003; CDC, 2004) show pooled mean central line-associated BSI rates ranging from 2.9 to 7.9, 2.9 to 7.9, 2.9 to 7.8, and 2.7 to 7.4, respectively, in nine different categories of ICUs for the period from January 1995 to June 2001, June 2002, June 2003, and June 2004 respectively. The overall pooled mean can be calculated for the same time periods at 5.1, 5.0, 4.9, and 4.3, respectively. This steady decrease over the 4 years, 2001, 2002, 2003, and 2004, represents an important trend. Having a uniform way of reporting CLI makes it possible to track trends and investigate possible related factors. Tracking rates of infection over time is recommended in the CDC guidelines. The calculated central line-associated BSI rates of 5.1, 5.0, 4.9, and 4.3 exclude the burn and respiratory ICU data for all four time periods. NNIS calculations exclude those categories of ICUs for which there were fewer than 20 units included in the NNIS report. There were fewer than 20 respiratory ICUs for all four time periods. Burn units, for the four time periods, numbered 18 for 2001, 19 for 2002, 21 for 2003, and 14 for 2004. Because the burn unit data were not included in calculations for 2001, 2002, and 18 2004 they were also excluded from calculations for 2003 in order to accurately reflect the BSI rate trend in the other units over 4 years' time. The burn units had the highest rates of BSI for the three years, 2001, 2002 and 2003 with rates of 9.7, 8.8, and 8.5, respectively. In 2004 their CLI rates dropped to 7.0. NNIS has not tracked the rate of NNIS hospitals' use of CDC guidelines or of the Pis. The slight decrease in rates of CLI in NNIS hospitals from 2001 to 2004 data releases, may be related to a higher level of use of guidelines, or it may simply be due to the effect of having an infection control professional (ICP) in the hospital monitoring rates of CLI. Minimum requirements for NNIS hospitals included having 100 or more beds, 1 fulltime ICP for the first 100 occupied beds, and an additional fulltime ICP for each additional 250 beds (CDC, 2001a). During 2004-05 NNIS was combined with two other national health surveillance systems to create the National Healthcare Safety Network (NHSN). Data collection is now carried out and reported by NHSN (Tokars et al., 2004). Replacing the former NNIS data summary, NHSN issued its first data summary report, the summary for 2006, in June, 2007 (Edwards et al., 2007). NHSN offered the caveat that comparison of the NHSN data with the last NNIS data could be misleading because units reporting to NHSN are a subset of former members of NNIS. Thus, changes in CLI rates could be due to (a) a change in the characteristics of the units reporting to NHSN, or to (b) an actual change in the rate of CLI. Therefore, NHSN data were not included in the literature review. 19 CLI in BMT Patients The problem of CLI in BMT patients is critical and understudied. Published studies of CLI in BMT patients are relatively few. Forty-two studies published between 1990 and 2007 were reviewed. They focused on (a) comparisons of kinds of central line dressings and timing of dressing changes (Benhamou et al., 2002; Brandt, DePalma, Irwin, Shogan, & Lucke, 1996; Shivnan et al., 1991); (b) IV tubing change practices (deMoissac, & Jensen, 1998); (c) types of central venous access devices (Biffi et al., 1999; Biffi et al., 2004; Brodwater et al., 2000; Haire, Stephens, Kotulak, Schmit- Pokorny, & Kessinger, 1995; Lazarus et al., 2000; Leibundgut, Miiller, Miiller, Ridolfi- Luthy, & Hirt, 1995; Madero et al., 1996; Platzbecker et al., 2001; Restrepo et al., 2002; Ulz et al., 1990); (d) use of femoral CVCs (Lazarus, Creger, Bloom & Shenk, 1990; Sovinz et al., 2001); (e) complications associated with CVCs in stem cell and BMT (Meisenberg et al., 1997; Moosa, Julian, Rosenfeld, & Shadduck, 1991; Uderzo et al., 1992); (f) incidence of nosocomial infection, including CLI, and associated pathogens (Adler et al., 2006; Aksu et al., 2001; gelebi, Akan, Ak§aglayan, Ustiin, & Arat, 2000; Dettenkofer et al., 2003; Engelhard et al., 1996; Elishoov, Or, Strauss, & Engelhard, 1998; Keung et al., 1995; Marena et al., 2001; Ninin et al., 2001); (g) risk factors for CLI in home infusion therapy (Tokars et al., 1999); (h) methods of diagnosing catheter-relation bloodstream infections (Abdelkefi et al., 2005; Abdelkefi et al., 2006; Krause et al., 2004); (i) the use of prophylactic antibiotics to prevent CVC-related infections (Lim, Smith, Machin, & Goldstone, 1993; Ljungman, Hagglund, Bjorkstrand, Lonnqvist, & Ringden, 1997; Vassilomanolakis et al., 1995); (j) the use of prophylactic urokinase to prevent complications with Hickman catheters (Solomon, Moore, Arthur, & Prince, 2001); (k) decreased infectious morbidity associated with stem cell transplantation in the outpatient setting (Chandrasekar et al., 2001); (1) timing and techniques associated with central line insertion and replacement (Apsner et al., 1998; Martinez et al., 1999; Miceli et al., 2005; Muhm et al., 1997; Richard-Smith & Buh, 1995); (m) time of occurrence of CLIs (Romano et al., 1999); and (n) patient education (Richard-Smith & Buh, 1995). Taken together, these studies provide useful information on selection and placement of various types of CVCs, tubing and dressings; risk factors, incidence, and diagnosis of CLI; and use of prophylactic medications and patient education to decrease CLI incidence. What is missing is an assessment of the adoption and the impact of use of the CDC guidelines for preventing CLI among BMT patients. There are no published studies of CLI in BMT patients addressing the CDC Pis. No data on attributable mortality, morbidity, and cost of CLI in BMT patients were found in a search using Pub Med. With a mortality rate in the general patient population reported at 10- 20% of hospitalized patients who acquire IV catheter-related infections (Charalambous et al., 1998; Crow, 1996), it is possible that the mortality rate in BMT patients who acquire a CLI may be even higher. Due to their immunosuppressed state, BMT patients may be more likely to acquire a catheter-related infection and die from the infection than are patients in the general patient population. The need for effective nursing and medical practice strategies in CLI prevention for this high risk patient population is significant. Evidence of Practice Pattern Variation in Prevention of CLI in BMT Patients The 42 studies of CLI in BMT patients from 1990 to 2007 are summarized in Table 5. It is immediately apparent that the studies used multiple different descriptions and definitions for CLI. The terms used to quantify rates of infection were also diverse. Ten of the studies did report results using the terms entry or exit infection (3), catheter-associated bloodstream infection (2) or catheter-related bloodstream infection (5), which are now among the CDC-approved terms for CLI, as found in the CDC guidelines. Only one of those studies calculated the CLIs as the number of infections per 1000 catheter days, the CDC-approved rate of measurement. Therefore, no two studies can be compared in a useful way. The general lack of standardization in literature describing and defining CLI, and in expressing rates of CLI, makes it difficult to draw conclusions from a review of the studies (Zitella, 2003). Table 6 summarizes how CLIs were described and their rates in the 42 studies. Only 6 studies used numbers of infections per 1000 catheter days; 11 studies used the CDC-recommended terms for infection (clinical sepsis, exit site infection, tunnel infection, pocket infection, CA-BSI, CR-BSI); and only 2 studies used both. While useful in showing the extent of the problem of CLI in BMT patients, these studies clearly demonstrate the need for a common language to describe and to quantify CLI in BMT patients. Even with acknowledgement of differences in methods of quantifying and reporting CLI, it seems apparent that some BMT centers have lower rates of CLI than other BMT centers - markedly lower in a few cases. What makes the difference? Some Table 5. Rates of Central Line Infections Reported in Forty-Two Studies of Bone Marrow and Stem Cell Transplantation 22 Study Variable of Interest N Description and Rate of Infection Ulz et al. (1990) Complications in Hickman right-atrial catheters Lazarus et al. (1990) Use of femoral site for CVC 111 Infectious complications - 44% of catheters 5 Infectious complications - 1 catheter-related Strep infection 1 E. coli bacteremia Moosa et al. (1991) Complications of indwelling CVCs in BMT recipients 123 Catheter infection - 15.8% of catheters Catheter removal due to infection - 10.8% of catheters Shivnan et al. (1991) Uderzo et al. (1992) Lim et al. (1993) Haire et al. (1995) Dry sterile gauze dressing changed daily versus transparent adherent dressing changed every 4 days CVC-related complications after BMT in children with hematological malignancies Hickman catheter-related sepsis following prophylactic teicoplanin Use of double-lumen inferior vena cava catheters for both apheresis and peripheral stem cell transplantations 98 Catheter-related infection - 1% of patients 53 CVC-related infections - 31% of CVCs, or 3.1 per 1000 catheter days 88 Lower incidences of sepsis with teicoplanin 20 Catheter-related infections - 15% of catheters/patients 23 Table 5. (Continued) Study Variable of Interest N Description and Rate of Infection Keung et al. (1995) Richard-Smith, & Buh (1995) Vassilomanolakis et al. (1995) Central venous catheter- 11 related infections Effects of timing of catheter 10 placement, and patient education on central line catheter infections Effects of vancomycin 40 prophylaxis on CVC- and related infections after 46 BMT in patients with CVCs malignancies Catheter removal due to infection - 57% of catheters (11.5 infections per 1000 catheter days) Pre-intervention infections - 50% of patients; Post-intervention infections - 16%, 4%, 0% at next 3 consecutive quarterly intervals respectively Infected CVCs - 9/46 (20%) CVCs Exit site infections - 5/46 (11%) CVCs CVC-related bacteraemia - 7/46(15%) CVCs Brandt, DePalma, Irwin, Shogan, & Lucke(1996) Dry sterile gauze dressing changed daily versus moisture vapor permeable dressings changed weekly 101 Reasons for line removal (percent of patients): Central vascular catheter (CVC) sepsis - 3% CVC sepsis and tunnel infection - 3% Tunnel infection - 5% Suspected CVC sepsis - 14% Engelhard et al. (1996) Documented catheter-related infections 242 Septicemia - incidence 7% Colonization - incidence 7% Exit site infection - incidence 3.7%; 24 Table 5. (Continued) Study Variable of Interest N Description and Rate of Infection Elishoov, Or, Strauss, & Engelhard (1998)(same study as Engelhard et al., 1996) Documented catheter-related infections (same study as above) Septicemia episodes per 1000 catheter days: Catheter-related 5.28 Unknown origin 4.86 Exit site infection 2.59 Leibundgut et al. (1996) Madero et al. (1996) Ljungman et al. (1997) Meisenberg et al. (1997) Complications associated 32 with Broviac catheters in children Complications in non- 62 tunneled catheters in children undergoing BMT Gram-positive infections 65 after prophylactic teicoplanin Complications associated 156 with CVCs used for stem cell collection followed by high-dose chemotherapy and autologous stem cell rescue Septic episodes - 0.26 per 100 catheter days Entry site infection - 9.6% of patients Catheter-related infection - 12.9% of patients Catheter-related sepsis - 14.5% of patients No differences between teicoplanin group and control group Early and late presumptive or documented infections of the CVC-6.5% of patients Muhm et al. (1997) Percutaneous nonangiographic insertion of Hickman catheters by anesthesiologists and intensivists 53 Suspected or documented infection - 26% of catheters (resulting in removal of 14 catheters per 3333 catheter days) 25 Table 5. (Continued) Study Variable of Interest N Description and Rate of Infection Apsner et al. (1998) Fluoroscopic guidance for Hickman catheter placement 81 Positive blood cultures - 2.1 / 1000 catheter days Suspected infection - 23 (28.4%) Tunnel infection - 2 (2.5%) deMoissac & Jensen (1998) IV tubing change every 24 versus 48 hours 50 Colonized infusate - 5% of IV sets Biffiet al. (1999) Use of totally implantable access ports 68 Port removal due to infection - 2.8% of devices Martinez et al. (1999) Romano et al. (1999) Tokars et al. (1999) Brodwater et al. (2000) CVC exchange by guidewire 17 for treatment of catheter-related bacteraemia in BMT or intensive chemotherapy patients Bloodstream infections in 442 children receiving allogeneic BMT Risk factors for bloodstream 827 infection in patients with home infusion therapy Converting implanted ports 67 to tunneled CVCs Catheter-related bacteraemia - episodes 14 CVC-related bloodstream infections - 6% of all CVCs Bloodstream infections - 0.99/ 1000 catheter days Persistent neutropenic fever - 2 (3%) Proven catheter infection - 2 (3%) Overall infection rate - 1.2 per 1000 catheter days 26 Table 5. (Continued) Study Variable of Interest N Description and Rate of Infection gelebi et al. (2000) Lazarus et al. (2000) Aksu et al. (2001) Febrile neutropenia in 145 patients receiving conventional chemotherapy (CCT) and patients receiving peripheral blood stem cell transplantation (PBSCT) Use of a single catheter for 112 both collection and transplantation Aerobic bacterial and fungal 74 infections in peripheral blood stem cell transplants Catheter infections: CCT group (50 pts) - not disclosed PBSCT groups: Alio-6/50 pts (12%) Auto - 3/45 pts (6.7%) Infectious complications - 2 exit site infections 17 bacteremias Catheter-related infections - 26 Catheter removal due to catheter-related infections - 10 Chandrasekar et al. (2001) M arena et al. (2001) Ninin et al. (2001) Platzbecker et al. (2001) Infectious morbidity after 105 outpatient autologous stem cell transplantation for women with breast cancer Incidence of nosocomial 143 infections in stem cell transplant patients Bacterial, viral, and fungal 446 infections in adult BMT recipients Use of double-lumen port 26 Catheter site infections - 5 patients Lines removed for suspicion of infection - 6 patients Device-associated infection rate for central line-associated bloodstream infections - 0.016 per 1000 catheter days Catheter-related infections - 9 (2%) of patients 2 port systems removed due to early pocket infections 27 Table 5. (Continued) Study Variable of Interest N Description and Rate of Infection Solomon et al. Prophylactic urokinase (2001) Sovinz et al. (2001) Using tunneled femoral CVCs in children 100 Septicaemic events - 8/52 in urokinase group 9/48 in heparin group Exit site infections - 27/52 in urokinase group 28/48 in heparin group Septic thrombosis - 2/52 in urokinase group 4/48 in heparin group 9 Catheter removal due to infection - 1 catheter Benhamou et al. (2002) Restrepo et al. (2002) Dettenkofer et al. (2003) Biffi et al. (2004) Effects of less frequent 112 catheter dressing changes in pediatric candidates for BMT Performance of a hybrid 82 CVC used for both stem cell harvest and transplant support of autologous stem cell transplant patients Nosocomial infections in 351 adult bone marrow and stem cell transplant recipients Use of an implanted port for 376 both high-dose therapy and transplantation Bloodstream infections - 13 (11.6%) patients Catheter site infections - 18 (16.1%) patients Catheter-related bloodstream infections (using CDC criteria for CR-BSI) - 18(22%) patients Catheter- associated infections - 20 (5.7%) patients Device-related complications - 2 pocket infections (0.01 per 1000 catheter days 3 port-related bacteremias (0.016 per 1000 catheter days) 28 Table 5. (Continued) Study Variable of Interest N Description and Rate of Infection Krause et al. Use of differential time-to- 51 Rate of catheter-related (2004) positivity method and Gram stain-acridine orange leukocyte cytospin test to detect catheter-related bloodstream infection bloodstream infections - 31% of patients Abdelkefi et al. Use of differential time-to- Catheter-related (2005) positivity method to diagnose catheter-related bloodstream infection bloodstream infections - 22 patients had catheter-related bacteremia 16 patients had noncatheter-related bacteremia Miceli et al. (2005) Leaving previously inserted ports in place for stem cell transplantation 86 Patient morbidity - No increase in morbidity Increased used of antibacterial and antifungal agents Abdelkefi et al. Use of Gram stain-acridine 245 Test ability to detect (2006) orange leukocyte cytospin test to detect catheter-related bloodstream infection catheter-related bloodstream infection - Gram stain detected only 2 Differential time-to-positivity detected 26 Adler et al. (2006) Comparing implantable ports and Hickman catheters 281 Rates of bloodstream infections - Implantable ports - 1.451 per 1000 catheter days Hickman catheters - 4.656 per 1000 catheter days Table 6. Summary of Infection Descriptions and Rates of Central Line Infection in 42 Studies Type of Infection Percent of Patients Percent of Catheters Infections/1000 Catheter Days Other Colonized infusate (1) Bacteremia (1) Persistent neutropenic fever or febrile neutropenia (2) 3-12 Suspicion of or suspected CVC sepsis or infection (2) 14 - 28.4 Infectious complications (2) 20 Catheter-related sepsis (1) 14.5 Catheter infection, or catheter-related infection, or 1-35 port-related infection, or CVC-related infection (8) Catheter or port removal due to infection or sepsis (7) 8-11 Suspected or documented infection (2) 8 Septic episodes (1) Entry or exit site infections (3) 5 - 16.1 Tunnel infections (1) 2.5 Bloodstream infections (2) 11.6 CA-BSI (2) 5.7 CR-BSI (5) 22-31 5% of IV sets 44 16-31 2.8-57 26 0.01 - 11.5 14/3333 0.26 / 100 0.99 0.016 14 episodes 2 ports Number of patients - 22 Abbreviations: CVC = central vascular catheter; CA-BSI = catheter associated bloodstream infection; CR-BSI = catheter related bloodstream infection differences may be related to variability, among centers, in aggregate patient demographics, in number and types of BMTs performed, and in practice patterns related to prevention of CLI. No published study has described the variation in practice patterns of CLI prevention among US BMT centers. Guidelines for Prevention of CLI In 1995, the CDC issued guidelines for prevention of CLI. The guidelines were later published (Pearson, 1996) and made available to healthcare workers and the general public. The following year, three organizations (the CDC, the Infectious Diseases Society of America, and the American Society for Blood and Marrow Transplantation) co-sponsored and began working on a set of guidelines aimed at preventing opportunistic infections in hematopoietic stem cell transplant recipients. Main CVC infection prevention recommendations of this guideline were to (a) implement the CDC guidelines (1996), (b) avoid tap water contamination of the CVC site, and (c) protect the end cap of needleless IV access devices from tap water contamination during bathing (Centers for Disease Control and Prevention, Infectious Disease Society of America, American Society of Blood and Marrow Transplantation, 2000; Dykewicz, 2001). The CDC issued new CLI prevention guidelines in its Morbidity and Mortality Weekly Report of August 9, 2002 (CDC, 2002a), and these have now replaced the guidelines published in 1996. Each of the two CDC documents represents a synthesis of hundreds of studies to determine optimum infection prevention practices for insertion and care of intravascular catheters. Although these guidelines are not specifically directed at preventing infections in the BMT patient population, they may represent at least 31 minimum standards of evidence-based practice for insertion and care of central lines in BMT patients. The CDC categorized the individual recommendations (2002a, p. 13) according to the level and kind of evidence which support each recommendation, as follows: "Category IA. Strongly recommended for implementation and strongly supported by well-designed experimental, clinical, or epidemiologic studies. "Category IB. Strongly recommended for implementation and supported by some experimental, clinical, or epidemiologic studies, and a strong theoretical rationale. "Category IC. Required by state or federal regulations, rules or standards. "Category II. Suggested for implementation and supported by suggestive clinical or epidemiologic studies or a theoretical rationale. "Unresolved issue. Represents an unresolved issue for which evidence is insufficient or no consensus regarding efficacy exists." Only category 1A recommendations were surveyed in the present study. An important addition to the 2002 CDC guidelines was the identification of the Pis, which were intended to be used to monitor local implementation of the new guidelines. Other changes included the addition or deletion of some guidelines, the reassignment of some guidelines to different levels of emphasis (categories) based on the updated literature review, and other minor modifications. Other Factors Other factors, either not addressed or not fully specified in the CDC guidelines, may be related to CLI. Literature supports several such factors, including nurse staffing 32 patterns, educational preparation of the RNs, use of mid-level providers, prophylactic anticoagulant use, and availability of BMT center CLI rates. Factors related to CLI rates are assumed to be related to levels of use of guidelines for prevention of CLI. Nurse Staffing Patterns With the current nursing shortage, nurse managers in hospitals have had to create new staffing matrices, delegating some nursing duties to unlicensed assistive personnel, or increasing the patient-to-nurse ratio. Understaffing has been linked to nurse reports of low quality patient care (Aiken, Clarke, & Sloane, 2002), higher risk of post-surgical complications (Kovner & Gergen, 1998; Pronovost, et al., 2001), higher rates of mortality and failure to rescue (Aiken, Clarke, Sloane, Sochalski, & Silber, 2002), and increases in both hospital-acquired infections, in general (Vicca, 1999), and CVC-related bloodstream infections, in particular (Fridkin, Pear, Williamson, Galgiani, & Jarvis, 1996). Understaffing also increases the likelihood of failure to adhere to protocols, policies, and guidelines that may require additional patient-care and documentation time (Pittet, Mourouga, & Perneger, 1999). Educational Preparation of the RNs A nursing shortage stimulates development of new programs to prepare nurses. Two-year associate degree programs see greater increases in enrollment because they can produce new RNs more quickly than 4-year baccalaureate programs. A landmark study of outcomes for over 200,000 general, orthopedic, and vascular surgery patients in 168 hospitals reported lower mortality and failure-to-rescue rates in hospitals with higher proportions of nurses with baccalaureate or higher degrees (Aiken, Clarke, Cheung, Sloane, & Silber, 2003). Percentage of RNs with baccalaureate (BSN) preparation may be associated with levels of use of guidelines. Use of Mid-level Providers Mid-level providers are used in many hospital settings and include nurse practitioners (NPs), clinical nurse specialists (CNSs), and physicians' assistants (PAs). Both NPs and CNSs are advanced practice nurses (APNs). PAs may be but generally are not nurses. Use of mid-level providers is increasing in many clinical areas, and current literature is particularly supportive of the APN roles and their positive effects on patient outcomes and cost of care (Brooten et al., 2002; Fulton & Baldwin, 2004). Use of mid-level providers may be associated with levels of use of guidelines. Use of Prophylactic Anticoagulants Low-dose anticoagulants have been shown to reduce the occurrence of venous catheter-associated thrombosis, a risk for patients with central lines (Boraks, et a., 1998). Use of prophylactic anticoagulants to prevent catheter-related thrombosis may help to prevent clot formation on the distal end of the catheter and thereby prevent the development of an environment in which bacteria readily grow (Polderman & Girbes, 2002; Randolph, Cook, Gonzales, & Andrew, 1998; Timsit et al., 1998). Use of prophylactic anticoagulants may also be associated with use of guidelines. Availability of BMT Center Rates of CLI Having access to CLI rates is essential to improvement in central line care. ICUs in the NNIS System had access to their rates of CLI, and the trends showed incremental improvement. The slight downward trend in their CLI rates in the last 3 years of NNIS reports (CDC, 2001b; CDC, 2002b; CDC, 2003) may have been associated with knowledge of their rates and a corresponding increase in use of guidelines. Use of Guidelines in Prevention of CLI A literature search on CLI and the use of guidelines yielded disappointing but not unexpected results. The search strategy included various terms for CLI plus various terms for use of, or adherence to, guidelines. The result of the search was a list of 66 journal articles, 11 of which were at least somewhat related to guidelines for prevention of CLI. Table 7 shows the topics of the articles and the guidelines mentioned in each article. Two reviews and one clinical trial addressed the use of prophylactic antibiotics (Bagnall-Reeb, 2004; Sandoe et al., 2003; van de Wetering & van Woensel, 2003). Two studies examined educational interventions to increase use of guidelines (Coopersmith et al., 2004; Eggimann et al., 2000). One study and one presentation of guidelines focused on diagnosis of CLI (Fatkenheuer et al., 2003; Raad et al., 2004). Finally, two review articles encouraged the use of guidelines (Parker, 2002; Rosenthal, 2003); one study compared local to national rates of CLI (Askarian, Hosseini, Kheirandish & Memish, 2003); one study used electronic prompts to increase use of guidelines (Rijnders, Vandecassteele, Wijngaerden, Munter, & Peetermans, 2003); and one study compared the outcomes of insertion of tunneled central lines by a trained clinical nurse specialist (CNS) to the outcomes of central line insertion by multiple junior medical staff persons (Fitzsimons et al., 1997). No article mentioned use of the 2002 CDC guideline Pis; however, one article presented two other guidelines for preventing 35 Table 7. Results of a Literature Search on CLI and the Use of Guidelines Topics Guidelines Mentioned References by Type (Country, if not US) Antibiotic lock technique - considering broader use than the CDC recommends CDC Guidelines (2002) Review - Bagnall- Reeb(2004) Using behavioral educational interventions to improve compliance with evidence-based guidelines of CVC management Diagnosing catheter-related bloodstream infection (CR-BSI) by differential time to positivity, using a specific guideline to define CR-BSI Use of evidence-based guidelines advocated, but no specific guideline identified 2002 CDC Guidelines cited, but not named in the text, in reference to incidence and prevention of catheter-related bloodstream infections 1996 CDC Guidelines cited, but not named, in reference to use of stopcocks Infectious Diseases Society of America guidelines (2001) Study - Coopersmith et al. (2004) Study - Raad et al. (2004) Comparing two methods of perioperative antibiotic prophylaxis for prevention of catheter-related colonization and infection, in spite of guidelines which state no prophylaxis is needed Presenting guidelines for diagnosis and management of catheter-related infections in neutropenic patients UK Department of Health guidelines (2001) Guidelines of the Infectious Diseases Working Party (AGIHO) of the German Society of Hematology and Oncology (DGHO) (2003) Clinical trial - Sandoe et al. (2003) (UK) Practice guidelines - Fatkenheuer et al. (2003) 36 Table 7. (Continued) Topics Guidelines Mentioned References by Type (Country, if not US) Using electronic prompts to improve physician compliance with guidelines for treatment of catheter-related bloodstream infections Comparing local device-associated nosocomial infection rates to rates published in the National Nosocomial Infections Surveillance System reports Infectious Diseases Society of America (IDSA) - Guidelines for the Management of Intravascular Catheter-related Infections (2001) Study - Rijnders, Vandecasteele, Wijngaerden, Munter, & Peetermans (2003) (Belgium) National Nosocomial Infections Study - Askarian, Surveillance System (NISS) methods for monitoring and observations of catheter-related bloodstream infections (NISS report for January 1992 to June 2001) (2001) Determining efficacy of Hospital Infection Control administering prophylactic antibiotics to prevent Gram-positive catheter-related infections Practices Advisory Committee (HICPAC) Recommendations for the Prevention of Nosocomial Intravascular Device Related Infections (1990) (use of guidelines not specifically addressed, but statement made that infections are increasing despite availability of the international HICPAC guidelines; implication that guidelines are either not being followed or are not sufficient) Hosseini, Kheirandish, & Memish (2003) (Iran) Review (Cochrane Database of Systematic Reviews) - van de Wetering & van Woensel (2003) Highlighting, and urging CDC guidelines (2002a) compliance with, three different guidelines or standards for preventing catheter-related infections American Association of Blood Banks's Technical Manual (1999) Intravenous Nurses Society's Infusion Nursing Standards of Practice (2000) Review - Rosenthal (2003) 37 Table 7. (Continued) Topics Guidelines Mentioned References by Type (Country, if not US) Highlighting and recommending use of three guidelines for preventing catheter-related bloodstream infections CDC guidelines (Draft) (2001) Department of Health (2001) Infection Control Nurses Association (ICNA, 2001) Review - Parker (2002) Examining effects of an educational campaign to increase compliance with guidelines for hand hygiene, site care, and line insertion, replacement, and removal Institutional guidelines prepared by staff within the study hospital Study - Eggimann et al. (2000) Institutional guidelines - Study - Fitzsimons et Guidelines for practice by the al. (1997) CNS CVCs, using predetermined guidelines Having a specially trained clinical nurse specialist (CNS) insert percutaneous tunneled infectious complications following chemo- and or radiotherapy and stem cell transplantation (Bertz et al., 2003). A second literature search, adding various search terms for BMT to the terms for CLI and use of guidelines, produced only one article. The previously mentioned article by Bertz et al. (2003) and the article from this second search are shown in Table 8. The first article is merely a presentation of a set of guidelines. The second article describes a study implementing diagnostic guidelines or criteria for nontuberculous mycobacteria infection (Weinstock, Feinstein, Sepkowitz, & Jakubowski, 2003). No article was found to describe the practice, or the results, of implementing the 2002 CDC guideline Pis. Translating research evidence into evidence-based practice for prevention of CLI is challenging. It requires persuasive and on-going education of care providers; changes in protocols for central line selection, insertion, use, and care; and on-going surveillance of practice patterns for use of new protocols and guidelines. Evaluation of CLI prevention guidelines must be measured by two endpoints - (a) institutional adoption of the guidelines and (b) changes in institutional rates of CLI. The literature reviewed for the present study makes it clear that the CDC guidelines, though they are the most complete and well researched guidelines published currently, have not been widely adopted in BMT centers. The present study focused on identifying patterns in care of central lines, patterns in use of the CDC Pis, factors associated with those patterns, and patterns in rates of CLI. 39 Table 8. Results of a Literature Search on CLI and the Use of Guidelines in BMT Patients Topics Guidelines Specified References by Type (Country, if not US) Presenting guidelines for Guidelines of the Infectious antibiotic therapy for infectious complications following high-dose chemo- and or radiotherapy and stem cell transplantation Implementing guidelines/criteria to distinguish nontuberculous mycobacteria (NTM) infection from NTM colonization in hematopoietic stem cell transplant patients Diseases Working Party (AGIHO) of the German Society of Hematology and Oncology (DGHO) (2003) American Thoracic Society and CDC diagnostic criteria for NTM infection (no dates specified) Practice Guidelines - Bertz et al. (2003) (Germany) Study - Weinstock, Feinstein, Sepkowitz, & Jakubowski (2003) CHAPTER III DESIGN AND METHODOLOGY Study Design This study used an exploratory, descriptive approach to examine variations in self-reported practice patterns among US BMT centers with regard to prevention of CLI, in general, and to use of specific recommendations in the 2002 CDC guidelines for preventing central line infections (2002a) in particular. The study also examined factors related to self-reported use of the CDC guidelines, as well as self-reported rates of central line infections in US BMT patients. In keeping with the epidemiologic focus and the conceptual model based on epidemiologic principles of infection prevention shown in Figure 1, this study was population-based and observational. Retrospective cross-sectional data from a cohort of US BMT centers were collected and analyzed using a descriptive correlational design. Wood and Brink (1998) identify the following six basic assumptions for correlational designs, (a) Covariance of the study variables in similar populations has not been demonstrated in previous studies, (b) An applicable conceptual framework supports the relationships between the study variables, (c) No existing tested theory predicts the relationships to be studied between the variables, (d) The study variables are known to exist and can be studied in the target population, (e) The study sample represents the target population, (f) The variables are studied as they exist and are not manipulated. This study met the six assumptions listed above for a correlational design, as follows: (a) The question of use of CDC guidelines, specifically use of all four Pis in the 2002 CDC guidelines, among BMT centers has not been studied. That is to say, use of various guidelines, including the CDC guidelines, has been encouraged and recommended, but no single published study of CLI has addressed usage or factors related to usage of the group of recommendations identified as Pis in the CDC guidelines (2002a). (b) The variables identified as Pis in the 2002 CDC guidelines were organized using the conceptual model for infection prevention, as shown in Table 1. (c) There is no published study that tests predictive theories about possible relationships between the variables, (d) The variables exist in the BMT patient population and can be studied, (e) The targeted population was US BMT centers, and all 202 US BMT centers listed on the BMT InfoNet website were invited to participate, (f) The variables were studied as they existed, with no manipulation. The study population was US BMT centers, and the outcome of interest, or dependent variable (DV), was use of the Pis in the 2002 CDC guidelines. The predictors, or independent variables (IVs), were classified as (a) four demographic and organizational factors and (b) five practice factors. The demographic and organizational factors variables were (a) type of transplant unit, (b) type(s) of transplants performed in 2005, (c) years in practice, and (d) number of transplants performed to 2005. The practice factors, which may be associated with levels of use of the Pis, were (a) staffing patterns, (b) educational preparation of RNs, (c) use of mid-level providers, (d) use of prophylactic anticoagulants, and (e) availability of BMT center CLI rates. 42 The investigator surveyed US BMT centers regarding use of the 2002 (the most current) CDC Pis and central line care practice recommendations, other CLI prevention strategies, and self-reported rates of CLI. The data were used to detect variations in practice patterns, levels of use of the 2002 CDC Pis, and rates of CLI among the studied BMT centers. Surveys were addressed, by name, to a BMT coordinator in each BMT center. Specifically, the information collected was used to identify (a) self-reported practices for prevention of CLI in participating US BMT centers, (b) self-reported behaviors representing levels of use of the Pis and other CDC line care recommendations in participating BMT centers, (c) factors which may be associated with self-reported levels of use of the 2002 CDC Pis in participating BMT centers, and (d) self-reported rates of CLI in participating BMT centers. Research Questions This study proposed to answer four research questions. 1. What are the self-reported practices for prevention of CLI in participating US BMT centers? 2. What are the self-reported levels of use of the Pis in the 2002 CDC guidelines among participating US BMT centers? 3. Which factors are associated with self-reported levels of use of the Pis in the 2002 CDC guidelines in participating US BMT centers? 4. What are the self-reported rates of CLI in participating US BMT centers? Figure 3 shows the conceptual model for the study design. Research question 3 was answered by analyzing the variables and the associations among the variables as depicted in the figure. Population and Setting The population for this study consisted of all US BMT centers listed on the BMT InfoNet website (BMT InfoNet, n.d.) as of June, 2006. Participants were nurses or other personnel in the BMT centers. The setting was the BMT center. Participants completed a mailed survey and returned it by mail to the investigator. A list of all 202 US BMT centers registered with the BMT InfoNet was obtained from the BMT InfoNet website on June 26-27, 2006. A survey was sent to each transplant center. Aggregate data were collected from the BMT InfoNet website and from each participating BMT center, and no individual patient data were solicited or used. Therefore, the study participants are the BMT centers and not individual BMT patients. All US BMT centers listed on the BMT InfoNet website at June 26-27, 2006 were invited to participate. The only exclusion criteria were non-US BMT centers, US BMT centers not listed on the BMT InfoNet website, BMT centers listed on the website but with inadequate demographic data, and BMT centers listed on the website but found to be no longer in business. The informant in each participating BMT center had to be able to read, write, and speak English. In all, 189 BMT centers met inclusion criteria. Minimum Sample Size Sample size, when converted to a response rate, determines, to a large extent, the usefulness of study results. Response rate, an umbrella term, encompasses a family of 44 VARIABLES: Figure 3. Conceptual model for the study design. DEPENDENT V ARlABLES: Levels of Usage to the Performance Indicators (API) in the CDC Guidelines, namely: I. Usage to Recommendations related to Educating Healthcare Workers (API-EHW) DATA 2. Usage to Recommendations Maximal Sterile Barriers (rAelPatIe-dM tSoB ) COLLECTI ON 3. Usage to Recommendatio ns related to 2% AND Chlorhexidine (API-2%C) 4. Usage to Recommendations related to ANALYSIS Discontinuing Catheters (API-DC) 5. Overall Mean Usage to the 4 PIs (OMU) Measurement of Relatio nships and Interpretation of IN DEPENDENT VARIABLES: Meaning Demographic and Organi zational Factors: I. Type of transplant unit (pedi atric, adult , or combined pediatric and adult) 2. Type(s) of transplants performed (autologous onl y or autologous and allogeneic) 3. Years in practice 4. Total transpla nts performed to 2005 Practice factors: l. Staffing patterns 2. Educatio nal preparation of RNs 3. Use of mid-level providers 4. Use of pro phylactic anticoagulants 5. Availabilit y of BMT center rates of CLI 45 rates. The American Association for Public Opinion Research (AAPOR) defined 12 components of outcome rates and combined them in formulae as response rates, cooperation rates, refusal rates, and contact rates (2006). This study calculated the response rate as the number of completed surveys plus partially completed surveys divided by the number of completed and partial surveys plus the number of refusals, non-contacts and contacts with failure to return surveys. BMT sites found to be out of business were not included in the denominator. AAOPR identifies this rate as Response Rate 6, or "the maximum response rate" (p. 33). Because not all 202 centers could be expected to respond, it was important to explore the minimum sample size needed for this study. A decision-based strategy (Pedhazur & Schmelkin, 1991) is often used to determine how many participants to attempt to recruit. This strategy is based on first identifying effect size (ES), Type I error (alpha), and Type II error (beta); plotting them on a matrix or master table (Kraemer & Thiemann, 1987); and then locating the corresponding sample size on the table. It would be ideal to calculate the number of participants needed to have sufficient power to detect an association between levels of use of the Pis in the 2002 CDC guidelines and factors thought to be related to levels of usage, and then to recruit exactly that number of participants. However, there was a limited number of US BMT centers, and it would have been difficult to recruit the optimum number of participants. Given the relatively small number of potential participants and the fixed size of the target population, it was more reasonable to begin with a discussion of sample size and work back, through Type I and Type II error, to identify a detectable ES (Cohen, 1988, p. 15). 46 Survey research using mailed surveys tends to have a lower participant response rate than research using face-to-face or telephone interviews, sometimes achieving only a 25-30% response rate (Burns & Grove, 1997; Waltz, Strickland & Lenz, 1991). Table 9 shows the results of a review of recent nursing and allied healthcare studies using mailed surveys. Of the nine studies, only three had response rates greater than 50%. Response rates ranged from 24 to 77%, with an average of 47.3%. Burns and Grove suggest that a response rate less than 50% imposes serious question as to the representativeness of the sample, thereby limiting the generalizability of a study's findings. Table 10 presents a review of recent nursing and allied healthcare studies using telephone surveys. In the seven studies cited, most response rates were greater than 50%, and the average response rate was 58.4%. While the higher rate of response is appealing, the format and content of a survey weighs in on the decision of which mode to use. Some of the data for this study had to be calculated or obtained from someone at the BMT center other than the informant. At least two telephone calls, and potentially several more, would have been required to gather all of the data by telephone. In addition, a consent form for use of the data would need to have been mailed to each participant and then returned to the investigator. Using a mailed survey with telephone follow-up seemed a more practical approach for this study. Improving Response Dillman (2000) has identified several methods for increasing the response rate for mailed surveys to the 70 to 78% range. A 50 to 70% response rate, a relatively high response rate for mailed surveys, would yield 101 to 143 of the 202 BMT centers. Key concepts for improving the rate of survey return include providing rewards, reducing 47 Table 9. Response Rates in Recent Nursing and Allied Healthcare Studies Using Mailed Surveys Study - Journal N RR Subject of study Comments Siem, Wipke-Tevis, 68 77% Skin assessment and Rantz, & Popejoy pressure ulcer care in (2003) - Ostomy/Wound hospital-based skilled Management nursing facilities Martin & Larson (2003) 263 53% Chemotherapy- Incentive used: a - Oncology Nursing handling practices of raffle of two Forum outpatient and office- annual Oncology based oncology Nursing Society nurses memberships Vande Vusse, Hanson, 334 24% Couples' views of Fehring, Newman, & the effects of natural Fox (2003) - Journal of family planning on Nursing Scholarship marital dynamics Criste (2003) -AANA 133 30% Do nurse anesthetists Journal demonstrate gender bias in treating pain? Carr, Gareis, & Barnett 98 50% Characteristics and (2003) - Journal of outcomes for women Women's Health physicians who work reduced hours Porterfield et al. (2003) 1273 59% Caring for the - Journal of Health underserved: Current Care for the Poor and practice of alumni of Underserved the National Health Service Corps. Lyons, Lapin, & Young 787 49% Job satisfaction of (2003) - Journal of nursing and allied Allied Health health graduates from a mid-Atlantic university McKenna, Smith, Poole, 551 47% Horizontal violence & Coverdale (2003) - (or bullying): Journal of Advanced experiences of RNs in Nursing their first year of practice Haugsdal, & Scherb 414 37% Using nursing (2003) - Journal of the interventions American Academy of classification to Nurse Practitioners describe the work of the nurse practitioner 48 Table 10. Response Rates in Recent Nursing and Allied Healthcare Studies Using Telephone Surveys Study - Journal N RR Subject of study Comments Hsu etal., (2004) 695 69% - Medical Care Johnson, Saha, 6299 54% Arbelaez, Beach, & Cooper (2004) - Journal of General Internal Medicine Patients' knowledge of cost-sharing and its influence on behavior Patients' perceptions of bias and cultural competence in health care McCormack, & Uhrig (2003) - Medical Care Braun et al. (2003) - Medical Care Peerson, Aitken, Manias, Parker, & Wong, (2002) - Journal of Advanced Nursing Kozlowski et al. (2002) - American Journal of Epidemiology Brogger, Bakke, Eide, & Gulsvik, (2002) - American Journal of Epidemiology 3738 606 30 hospitals 6 agencies 3383 171 76% Beneficiary knowledge of the Medicare program 65% Patients' use of self-referral vs. primary care physician referral 43% Hospital and agency managers' 23% perceptions of agency nursing 70% Feasibility of using random digit dialing telephone interviews to locate participants for a study of smoking-related behaviors 67% Comparison of the use of telephone and mail surveys in epidemiology Participants - 1% random sample of previous mail surveys. More complete answers given by telephone than by mail. Nature of the questions asked may affect completeness of participants' answers. social costs to participants, and establishing trust. Each of the following strategies, recommended by Dillman, was used to increase the response rate in this study: 1. Providing a reward a. Demonstrating positive regard in the cover letter b. Providing a small reward for the person in each BMT center who fills out the survey instrument c. Entering participants in a raffle for 1 of 10 annual memberships or membership renewals in the Oncology Nursing Society (worth $92.00) d. Sharing results of the study with participating BMT centers 2. Decreasing social costs a. Using language that does not subordinate the participant b. Enclosing a return envelope 3. Increasing trust a. Emphasizing the importance of the study b. Identifying the investigator as an oncology and BMT nurse Dillman maintains that providing a small token of appreciation up front is more effective than offering to send a gift upon completion of a mailed survey. Evans, Peterson, and Demark-Wahnefried (2004), using a six-page mailed survey, found no difference in their study of 1402 prostate cancer survivors. They achieved an overall response rate of 60% with no difference in response rate between the group that received a 30-minute prepaid phone card with the survey, and the group that received the phone card after completing the survey. Table 11 shows response rates and use of Dillman's (2000) Tailored Design method in nine studies of healthcare providers. The range of response rates was 31 to 90%, with an average response rate of 60%. Only two of the nine studies reported inclusion of a monetary reward with the cover letter, and those two studies had response rates of 49% and 47%, well below the average response rate for the studies. The most consistently used element of Dillman's method was inclusion of a stamped self-addressed return envelope with the mailed survey. Five studies used this element, but response rates among those five ranged from 31 to 82%. No other element was consistently used among the studies. Number of follow-up contacts ranged from one to "up to five contacts," not further described and over an unspecified period of time (Russell, Injeyan, Verhoef, & Eliasziw, 2004, p. 372) with no apparent increase in response rate associated with increase in number of follow-up contacts. Two studies used prestudy contacts with potential participants, as recommended by Dillman. Gallagher et al. (2007) sent an email notification prior to mailing hard copies of their surveys to participants, and they achieved a 90% response rate. Gutmanis et al. (2007) mailed prestudy letters 1 week before mailing surveys to participants, and they achieved a 47% response rate. Both studies also used follow-up contacts, as shown in Table 10, to increase response rates. Factors not associated with the Dillman tailored design method (2000) may account for the wide range in response rates. Gallagher et al. (2007) achieved the highest rate with a 90% response. Participants were dental students who were surveyed by their college following their fifth year final exam. Students may have felt vulnerable to perceived pressure to participate. Table 11. Response Rates in Studies Using Dillman's Tailored Design Method for Mailed Surveys of Healthcare Providers Elements of Dillman's Tailored Design Method Used N* RR^ (Dillman, 2000) Study - Journal Subject of Study Participants Sampling Frame SMR1 SASEZ FUJ Mailings FUJ Calls Other Vigod & Stewart (2002) Management of Family 622 Ontario No Yes FU reminder with N/R N/R 194 31% - BMC Women's Health abnormal uterine Physicians Physicians with new copy of survey bleeding highest rates of at 3 weeks and at 8 hysterectomy weeks Russell, Injeyan, Beliefs and Chiropractic All chiropractors N/R N/R "up to five contacts" N/R N/R 503 78% Verhoef, & Eliasziw behaviours of registered to with "the principles (2004) - Vaccine chiropractors about practice in Alberta described by immunization Dillman" Ruggiero (2005) - Health, work Critical Care Systematic $2.00 Yes FU reminder N/R N/R 247 49% Journal of Nursing variables, and job RNs selection (every postcards after 1 Administration satisfaction among 7th) from a random week nurses selection of 3500 RNs Sempowski, Rungi, & Urban Canadian Family All General and No Yes One FU mailing N/R N/R 89 82% Seguin (2006)-BMC family physicians' Physicians Family Family Practice minor office Practitioners in procedures Kingston, Ontario Silcox, Ashbury, Van Attitudes of Anesthesia All residents and No Yes One FU email after N/R N/R 283 60% DenKerkhof & Milne residents and Residents Program Directors 2 months; (2006) - Anesthesia and program directors and Program at a teaching replacement surveys Analgesia about research Directors hospital 1 month later during training Baird (2007) - Illinois Clinical Random selection No Yes Not reported N/R N/R 306 37% Professional psychologists' Psychology of Random Psychology: Research attitudes about selection of 25% and Practice prescriptive of licensed privileges psychologists in Illinois Table 11. (Continued) Elements of Dillman's Tailored Design Method Used N* RRS (Dillman. 2000) Study - Journal Subject of Study Participants Sampling Method SMR1 SASE" FUJ FU' Other Mailings Calls Gallagher, Patel, Views of final year Dental All final year N/R N/R FU reminder after N/R Pre- 126 90% Donaldson, & Wilson dental students Students dental students 3 weeks; FU letter study (2007) - BMC Oral about their and new survey in email Health professional career 6 weeks; FU letter in 8 weeks Gutmanis, Beynon, Factors influencing Physicians Random selection $2.00 N/R FU reminder letter N/R Pre- 931 47% Tutty, Wathen, & identification of and Nurses of 1000 MDs and gift with new survey study MacMillan (2007) - and response to in Women's 1000 nurses from card after 3 weeks letter BMC Public Health intimate partner Care Areas professional for violence of Practice directories coffee shop Skedros, Hunt, & Pitts Variations in Physicians Orthopaedic No N/R FU mailing after 6 N/R N/R 169 64% (2007) - BMC steroid/anesthetic surgeons and weeks Musculoskeletal injections for "selected non- Disorders painful shoulder surgeon sub-conditions among 4 specialists and different specialty specialty groups of physicians" in physicians Utah, Idaho and Wyoming Note. 'SMR = small monetary reward enclosed; 2SASE = self-addressed stamped envelope enclosed; 3FU = follow-up; 4N = Number; ?RR = Response Rate; N/R = not reported K> Length of the survey is probably a determinant of response rates. The Vigod and Stewart study (2002) used a five-page questionnaire and had a 31% response rate, the lowest response rate of the nine studies. The nature of the information requested in the survey may also affect participant response rates. Interestingly, Gutmanis et al. (2007) mailed surveys to 1000 nurses and 1000 physicians in practice areas preidentified as specifically including women's care. Their study elicited information about participants' identification of and response to intimate partner violence against women. Of the targeted population, 60% of the nurses and 33% of the physicians responded, for an overall response rate of 47%. This suggests that nurses may be more interested in the study topic than are physicians. However, it could also suggest that physicians and nurses may differ on other characteristics, such as likeliness to respond to surveys or gender. While evidence suggests the Dillman tailored design method (2000) may improve response rates to surveys, it is also evident that other factors influence potential participants' decisions to participate or not to participate. The publications containing descriptions of the survey instruments and the recruiting methods used in the nine studies in Table 10 did not supply sufficient information to make valid comparisons of all the possible factors contributing to the wide difference in response rates among the nine studies. Type I Error Type 1 error (alpha) was set at .05, just as a matter of convention (Hinlke, Wiersma, & Jurs, 1998). Although this was an exploratory study, it was important to set alpha at a generally respected level in order to prevent jumping to an incorrect conclusion that the null hypothesis should be rejected (that is, that associations existed between practice factors and self-reported use of the CDC Pis, when, in consideration of the studied evidence, they did not). Type II Error The minimal acceptable power for a study is .80 (Hinkle et al., 1998). A lower power increases the possibility of a Type II error (beta), failing to detect an actual difference (i.e., failing to detect a studied relationship which, in fact, did exist among the studied variables). Effect Size The significance of this study is two-fold and is based on the assumed inverse relationship between rates of CLI in BMT patients and levels BMT centers' use of the CDC's 2002 research-based recommendations for prevention of CLI, specifically use of the CDC's PI recommendations. Hence the study sought to (a) highlight variations in practice patterns relative to the CDC Pis and (b) identify factors related to levels of use of the Pis. Therefore, effect size was defined as r, or the correlation between use of the Pis and the practice factors, after controlling for effects of the demographic and organizational factors. Low levels of use of the Pis would provide targets for practice improvement, and significant correlations between low levels of usage and any of the predictor variables would identify potential areas for changes in BMT centers' structure and/or process for prevention of CLI. Such changes could lead to improved outcomes, namely decreased rates of CLI in BMT patients. Determining the ES that could be detected in this study was a critical process because a significant effect could represent a potential for reducing the rate of CLI in BMT patients through implementation of appropriate practice changes. Detecting a statistically significant effect/correlation could, therefore, be a first step toward a clinically significant outcome. A Conventional Approach to Estimating Attainable Effect Size The effect to be considered in this study was an association between self-reported use of the Pis and the predictor variables previously identified as the practice factors (staffing patterns, educational preparation of RNs, use of mid-level providers, use of prophylactic anticoagulants, and availability of BMT center rates of CLI) and the demographic and organizational factors. This effect was measured as the strength of the association (r) between self-reported levels of use of the Pis and the predictors. The null hypothesis is r = 0. Using Cohen's conventional definition, a medium effect size would be r = .30 (Cohen, 1983, p. 60). Working backward through Cohen's (1983, p. 529) power tables, and imputing intermediate values not included in the tables, yields the ES that could be detected at alpha = .05 and power = .80 for a range of possible survey response rates, shown in Table 12. Using Table 12, then, we could project, for example, that with a sample size of 40% of the study population, or 76 participants, it would be possible to detect an association on the order of r = .31 between the predictor variables and self-reported levels of use of the Pis. Carrying that a step further, the predictor variables would explain approximately 9% (f - .306 x .306) of the variance in level of use of the Pis for a participating BMT center. With a higher response rate, for instance a 70% response, or 56 Table 12. Detectable Effect Size (r)for Various Levels of Response Rate with Alpha = .05 (two-tailed) and Power = .80 Response Number of Effect Rate Participants Size 20% 38 .43 30% 57 .36 40% 76 .31 50% 95 .28 60% 113 .26 70% 132 .25 80% 151 .23 132 participants, it would be possible to detect an even smaller effect size - a relationship on the order of r = .25, between the predictor variables and self-reported use of the Pis if the effect does, in fact, exist. A Different Approach to Response Estimation A different way of determining the minimum response rate needed, which confirms the results shown above, is given by Browner, Newman, Cummings, and Hulley (2001). When the study data will be analyzed with a correlation coefficient, a table prepared by Browner et al. shows that an effect size of r- .45 with a two-tailed a = .05 and p = .20 can be detected with a sample of 36. Reaching a comparative conclusion on minimum response rates by using two different tables from two different sources confirmed that a response rate of 20 to 30% would be adequate to detect a significant effect in this population. Instrument Instrument Development Construction and Initial Review of the Instrument There is no published instrument to measure or estimate use of the CDC guidelines or use of the Pis in the guidelines. Therefore, an instrument, the BMT LIFE (Bone Marrow Transplant Line Infection Factor Evaluation) was developed for this study. The BMT LIFE was based on a review of the 2002 CDC guidelines and constructed to measure self-reported use of the Pis. Each usage question was based on one of the specific CDC (2002a) recommendations related to the four Pis (educating healthcare workers, using maximal sterile barriers, using 2% chlorhexidine skin prep, and discontinuing catheters as soon as medically indicated), for content validity. Content validity of the BMT LIFE was defined as the ability of the survey to actually measure the level of self-reported usage by a participating BMT center of each of the individual CDC recommendations addressed in the survey. The BMT LIFE had not been used before, so it was imperative to know that the questions would elicit responses that truly represent the essential behaviors which constitute use of each of the CDC recommendations addressed by the study. If, and only if, the content were valid could appropriate, meaningful, and useful inferences be made from analysis of the collected data (Pedhazur & Schmelkin, 1991). Survey content. Content areas of the questions in the survey are similar to those in the CDC guidelines, but the questions were organized in sections with slightly different names than the names of the sections in the CDC guidelines. The investigator did not want participants to recognize any exact phrases from the CDC guidelines because this might cause them to alter their responses based on the perceived desirability of answers reflecting the CDC guidelines. (Please see a copy of BMT LIFE, which is provided in Appendix A). Some additional questions were included in the survey to furnish other information that may be useful. Responses to these questions were analyzed separately using descriptive statistics and correlations. Those responses that were significantly correlated with any of the dependent variables were included in the multiple linear regression analysis of self-reported use of the Pis. The additional questions are questions A.5-13, B.l-2, C.l-3, D.6, E. 1,4-11, and F.l. Questions A.5-8 ask about use of trained CVC personnel for CVC insertion, CVC maintenance, and supervision of CVC trainees, and about staff levels. Respondents were also asked about adequacy of nursing staff levels to minimize incidence of CLIs. Questions A.9-11 ask about staffing patterns. Information derived from these questions was used to describe nurse-to-patient ratios in participating BMT centers. Comparisons of staffing patterns were then made between participating BMT centers and with use of the Pis. The CDC does not define the term "appropriate nursing staff levels" (2002a, p. 15) used in the recommendation which addresses staffing. Questions A. 12 and 13 ask about the educational preparation of RNs and the use of mid-level providers; both staffing levels and RN educational preparation have been shown to affect patient outcomes (Aiken et al., 2003; Aiken, Clarke, & Sloane, 2002; Aiken et al., 2002; Pronovost et al., 2001; Safdar, Kluger, & Maki, 2002; Tourangeau et al., 2007). Use of mid-level providers may also affect patient outcomes (Brooten et al., 2002; Fulton & Baldwin, 2004). Answers to these questions were used to describe patterns in educational preparation of RNs and other staff in participating US BMT centers. Staffing levels, RN educational preparation, and use of mid-level providers may also be associated with levels of self-reported use of the Pis (Wallace, MacKenzie, & Weeks, 2006). Questions B. 1 and 2 ask for the CLI rate in each BMT center for the calendar year 2005. Question C.l requests aggregate data about the number and kinds of transplants performed. Question D-6 requests data about the length of time central lines are generally in use. Answers to these questions could be used to calculate the rates of CLI per 1000 catheter days for those BMT centers that do not express their CLI rates in that Questions C.2 and 3 request aggregate data about patient age. That information, along with the aggregate data collected in Question C. 1 about number and kinds of transplants performed, and demographic data supplied on the BMT InfoNet website, was used to control for the effects of the demographic and organizational factors. Question E.l asks for information about the use of prophylactic anticoagulants. The 2002 CDC guidelines discuss (2002a, p. 9) but do not recommend the use of prophylactic anticoagulants. Questions E.4-11 ask respondents to identify the frequency with which they use preparations other than 2% chlorhexidine for skin antisepsis. The CDC guidelines list four preparations, including chlorhexidine, which may be used for skin antisepsis, with the expressed preference for use of chlorhexidine (2002a, p. 16). Question F.l is open-ended and requests that participants list the most important things they do at their BMT center to prevent CLI. This question was intended to show whether or not BMT center personnel identify the Pis or other CDC recommendations as being most important practices to prevent CLI. Survey format. BMT LIFE contains rating scale, multiple choice, multiple answer, and open-ended questions regarding CLI prevention practices addressed in the CDC guidelines and other selected practices, not mentioned in the CDC guidelines, which may be associated with levels of use of the 2002 CDC Pis. The survey also includes questions about aggregate patient age groups, number and types of BMT performed, and an open-ended question asking respondents to list the most important things they do at their BMT center to prevent CLI. The BMT LIFE survey was reviewed by content experts for face validity (Nunnally and Bernstein, 1994, p. 110). The content experts were one bachelor's prepared oncology RN, one master's prepared oncology RN, one BMT physician who is board certified in internal medicine/medical oncology, and one hospital epidemiologist who is board certified in internal medicine. The two oncology nurses were asked to complete the survey and comment on (a) the amount of time it took, (b) the ease with which they could answer the questions, and (c) any questions that needed to be revised. They returned the completed surveys with written comments. Although they had no difficulty answering the questions, both nurses suggested that the length of the survey would result in a low rate of participant response, 61 and one nurse warned of the probability of incomplete responses by participants who just got tired of answering the questions. The two physicians provided verbal feedback on questions specifically related to labeling and quantifying CLI. Focused Review of Selected Aspects of the Survey Process by Telephone Contact To further improve the survey process, a telephone contact was made with eight potential participants to get their reactions to questions regarding survey length, ability and willingness of participants to get and disclose data related to CLI rates, and incentives to be given to participants. A schedule of questions was created as a table with questions listed across the top of the table, names of 11 BMT centers with their contact information listed in the left-hand column and open boxes for answers below the questions. The investigator selected BMT Info Net as the source for a list of US BMT centers, based on the amount and type of demographic and organizational information available to the public on the BMT InfoNet website (BMT, n.d.). BMT InfoNet is a patient support organization which provides lists of US and Canadian BMT centers and other patient support resources for prospective BMT patients. The website contains a page with contact information and demographic and organizational data for each affiliated BMT center. The investigator decided to contact 5% of the potential US study participants for feedback on questions about the study. To obtain an unbiased sample, the investigator used the alphabetical list of US BMT centers on the BMT InfoNet Website and stratified the list by patient type - pediatric BMT centers, adult BMT centers, and combined pediatric and adult BMT centers. Approximately 5% (or every 20th) of the 202 potential US study participants were systematically selected for the telephone survey, using the following method. Each number from 1 to 20 was written on a piece of paper. The papers were folded and placed in a bowl, and a paper withdrawn blindly from the bowl contained the number 5. Then, using 5 as the starting number in each of the three stratified lists of BMT centers, the 5th center in each of the three patient-type groups and the succeeding 20th (25th, 45th, 65th, etc.) centers in each group were selected for inclusion in the telephone survey. Eleven centers in all were selected, with the resulting telephone survey group containing 2 of the 35 pediatric centers, 6 of the 107 adult centers, and 3 of the 60 combined pediatric and adult centers. Thus the distribution of participants among the three types of BMT centers approximated the distribution of all 202 BMT centers used for the main study. The investigator wanted a telephone survey group which would reasonably represent the larger group of which it was a part. Cold calls to the centers were made during the first week in May, 2005. The investigator asked to speak to a BMT coordinator in each center and introduced herself to the coordinator as an oncology nurse doing a study of central line infections in bone marrow transplant patients. It took two or three calls to each BMT center to reach someone who could provide answers to the telephone survey questions. By the end of the week, individuals (usually BMT coordinators) in eight of the eleven centers - one pediatric, five adult, and two combined pediatric and adult - were reached and queried about survey length, ability and willingness of the informant at each center to provide answers to the questions about rates of CLI, and opinions of informants about the desirability of three different incentives - a $5.00 bill attached to the survey, a drawing for one of five certificates for a 1-year memberships in ONS, and a copy of the study findings. The information obtained from the eight informants was considered adequate to meet the objectives of the telephone survey, so no further attempts were made to contact anyone at the remaining three sites. All eight of the potential study participants said they would not have time, or would have difficulty finding time, to complete a 12-page survey questionnaire. However, all of them said they were very interested in research on prevention of CLI in BMT patients. Seven of eight people surveyed thought they could get, or would be willing to try to get, CLI rate data for the survey. The same seven people said they would just need to get approval before giving the rate information. The eighth person said that only one or two transplants had been done at his facility in 2003 and none in 2004 or 2005. He stated that the facility was not currently transplanting but would be starting again that summer (the summer of 2005). Discussion of incentives was interesting. The first three people the investigator spoke to all said they would not be able to accept cash, such as a $5 bill attached to the survey. They could accept gifts, but cash would have to be given to the hospital. The investigator asked if they would be able to accept a $5 Wal-Mart gift card, and they said they could accept that. For the rest of the calls, the cash incentive was not suggested. A $5 Wal-Mart gift card was suggested instead. Most thought that a $5 Wal-Mart gift card would be acceptable. When asked about being placed in a drawing for a paid ONS membership, everyone agreed that would be a highly desirable incentive. Everyone also expressed interest in receiving a copy of the study results. 64 The 11 sites used for the telephone interviews were retained in the population for the main study. No BMT LIFE survey data were collected from the eight informants during the telephone survey, but the comments of the informants were used to improve the format and content of the BMT LIFE instrument and the cover letter. Of the eight informants in the telephone survey group, two became participants in the study, five did not complete and return surveys, and one did not participate because that site was no longer performing bone marrow transplants. Revisions to the Instrument Originally, BMT LIFE was a 12-page survey containing 76 questions assessing self-reported use of 51 CDC recommendations, 10 questions exploring other factors which may be related to CLI in BMT patients, and additional questions requesting aggregate patient and BMT center data. Following the telephone survey, the BMT LIFE survey was shortened by restricting the usage questions mainly to questions aligned with the four Pis in the CDC guidelines. That cut the survey down from 12 pages to 4 pages. Admittedly, the survey instrument lost the ability to discriminate between use and nonuse of the CDC recommendations that were omitted from the survey, but it may be that self-reported use of the CDC recommendations related to the Pis is representative of overall use and not just use of the four specific areas addressed in the Pis. The trade-off was the potential for an increase in the response rate, versus obtaining more detailed data from only a few respondents. The anticipated increased response rate was seen as a way to increase both the sample size and the power of the study to demonstrate significant relationships between some of the study variables. The investigator originally planned to attach a new $5.00 bill to each survey, for the main study, and to enter all participants in a drawing for one of five ONS memberships. However, based on the comments about not being able to accept cash, the modest interest in the $5.00 Wal-Mart gift card, and the overwhelming interest in the ONS memberships, the investigator decided to increase the number of ONS memberships in the drawing from five to ten, omit attaching the gift cards to the survey, and send a gift card to only those participants whose names were not drawn for ONS memberships. Pilot Study Next a pilot study was conducted to identify potential problems with data collection (Waltz, Strickland, & Lenz, 1991). Additionally, response rates and the quality of the data collected were reviewed. Selection of Participants The investigator decided to use the Canadian BMT centers listed on the BMT InfoNet website for the pilot study to preserve all possible US centers for the main study. A cohort of 15 Canadian BMT centers was identified. Only English-speaking Canadian centers were selected. The person completing each survey had to be able to read, write, and speak English. The BMT InfoNet website provides the same kinds of information and data for both US and Canadian BMT centers, and the Canadian centers were deemed to be similar to US BMT centers in terms of being able to answer the survey questions and give helpful feedback. Dillman's Tailored Design Method for Improving Response Rates Dillman (2000) describes survey research as constituting a social exchange. His Tailored Design method recommends specific practices to establish trust and increase response rates in survey research. Key concepts for improving the rate of survey return include providing rewards, reducing social costs to participants, and establishing trust. A modified Dillman (2000) approach was used to encourage potential participants to respond. The Dillman strategies, previously discussed, were used with minor changes. A Canadian $5.00 bill was attached to the cover letter and BMT LIFE survey sent to each BMT center in the Canadian cohort, as the cohort was small enough to make that incentive affordable, and it was not possible to procure Canadian Wal-Mart gift cards in this country. Surveys were mailed to potential participants on April 20, 2006. Implementing Dillman's Method In the implementation phase of survey research, Dillman (2000) describes five needed elements which can increase response rates: (a) a respondent-friendly survey instrument, (b) four contacts by first class mail and an additional contact made by telephone, if feasible, or if needed, (c) using real first-class stamps on return envelopes, (d) personalizing all correspondence, and (e) providing small financial incentives in advance. Modifications to these elements related to recruiting participants internationally. Prior to mailing the 15 pilot surveys, the investigator streamlined the survey instrument and made it as clear and user-friendly as possible. The oncology nurses who had reviewed the survey found no difficulty in reading, interpreting and answering the questions. Follow-up contacts for the pilot survey were by email. The investigator provided her email address in the cover letter to the pilot survey recipients. The Canadian participants were asked to email the investigator with any questions. A self-addressed return envelope was included for ease in returning the completed survey. The envelopes were not stamped because the investigator did not have access to Canadian postage. However, the attached $5 bills provided more than enough cash to reimburse participants for purchasing postage. Interestingly, all of the respondents mailed the surveys using their BMT center mail system, as evidenced by the metered postage on each of the return envelopes, so the respondents bore no personal burden of time or expense for purchasing postage. The envelopes containing the surveys were personally addressed to the BMT coordinator in each BMT center whose name appeared on the BMT InfoNet website. Likewise, the cover letters were also personally addressed. The $5 bill attached to each cover letter was for the person completing the survey. It was described as a token of appreciation from the investigator. Response to the Mailing As previously stated, the purposes of the pilot study were to identify problems with data collection and the quality of the data collected and to project the probable response rate for the main study. By May 16, 2006, 16 days following mailing of the surveys, three of the surveys had been completed and returned. Two additional surveys came in within the next few days. No follow-up contacts were used to increase the response rate, such as were planned for the main study, because the pilot surveys which had been received by May 20th had adequately met the purposes of the pilot study. There were no problems with the data collection or the quality of the data collected, and the response rate for the pilot was 33% without extensive follow-up contacts, suggesting the possibility of a higher response rate with additional follow-up contacts. Following data collection, names of the 5 participants in the pilot study were placed in a bowl, and one name was drawn for the 1-year Oncology Nursing Society membership. The investigator telephoned to inform the participant of the award and to verify the correct mailing address. The membership certificate was mailed to the winner, and no further contact was made with any of the participants. Following completion of the main study, all five Canadian participants were placed on a list to receive copies of the study findings. Findings of the Pilot Study Five of the 15 Canadian recipients completed and returned the BMT LIFE survey, for a response rate of 33.33%. Participants answered each question without apparent difficulty, with the exception of the questions asking for rates of CLI. No Canadian participant was able to supply that information. That finding was not unexpected and is significant in terms of use of the CDC guideline (CDC, 2002a) about surveillance of rates of CLI. No material change was made to the BMT LIFE survey on the basis of the Canadian pilot study. Procedural differences between the pilot study and the main study were as follows: Pilot Study Main Study Return envelope: Self-addressed, no stamp affixed Self-addressed, US postage affixed 69 Incentive: Canadian $5.00 bill attached $5.00 Wal-Mart gift card promised to participants Contacts: No follow-up mailings; email Follow-up contacts, contacts provided; no telephone including post cards, contacts telephone calls, email, and additional surveys Main Study Concept Clarification CDC Guidelines Theoretical definition. In this study, the term "CDC guidelines" referred to the Guidelines for Prevention of Intravascular Catheter-Related Infections (CDC, 2002a). Operational definition. Only recommendations in the CDC-defined categories IA and IB are addressed by questions in the BMT LIFE survey instrument. These two categories, as described in Chapter II, are strongly recommended by the CDC because they are supported by research. The other categories of recommendations were excluded from the survey because, by definition, they are not supported by research, and they cannot be considered evidence-based. The CDC guidelines define Category IC recommendations as requirements imposed by regulatory agencies, Category II recommendations as having suggestive clinical, epidemiologic or theoretical support, and Unresolved Issues as having insufficient evidence regarding efficacy (CDC, 2002a, p. In the 2002 CDC guidelines, the first section contains general recommendations which are applicable to insertion, maintenance and care of all intravascular catheters. 15). Following the general section are sections specific to (a) peripheral catheters in adult and pediatric patients, (b) central venous catheters (CVCs) in adult and pediatric patients, (c) peripheral arterial catheters and devices for monitoring pressure in adult and pediatric patients, and (d) umbilical catheters. Only the first section of general recommendations and the section with recommendations specifically for CVCs are addressed by questions in the BMT LIFE instrument. Theoretical definition. The terms "use" and "utilization" and "implementation" have been employed in studies to assess or identify a level of acceptance or adoption of recommended behaviors, such as new practice models (Lindqvist, Torsson, Almqvist, & Bjorgell, 2008) and clinical practice guidelines (Kontos & Poland, 2009), and to assess the outcomes of utilization of new procedures (Hermansen et al., 2008). Use in this study is the extent to which participating US BMT centers perform procedures for prevention of CLI described in select recommendations from the 2002 CDC guidelines for central line insertion, care, and maintenance. Operational definition. Use in this study is operationally defined as self-reported performance of behaviors recommended in the Pis in the CDC guidelines (2002a), or the level of compliance, self-reported by each participating BMT center, to selected Category IA and Category IB CDC recommendations for practice. Self-reported use was scored as a mean use score for each PI. The use scores for each participant were then summed to create an Overall Mean Use-Total (OMU-T) score for each participant, and then all the participant scores were summed and averaged to create the OMU-T score for the participants as a whole. The specific question(s) in the BMT LIFE instrument, and the corresponding answers, that were employed to calculate the level of self-reported use of each CDC recommendation are identified in Table 13. Calculating use. Each survey question has points attached to each possible answer. Self-reported use scores were calculated using the points earned for each of the answers related to a recommendation. This method was employed for all four types of questions - rating scale (RS), multiple answer (MA), multiple choice (MC), and open-ended (OE). The RS questions were on a 5-point likert-type scale, but only Always (4 points), Usually (3 points), Sometimes (2 points), and Rarely (1 point) were assigned points. Never was worth 0 points. The following example illustrates the calculation of self-reported use of a CDC recommendation. The first CDC (2002a) PI is "implementation of educational programs that include didactic and interactive components for those who insert and maintain catheters" (p. 13). CDC general recommendation I - "Health-care worker education and training" (CDC, 2002a, p. 13) has three parts - A, B, and C - which address education and training of healthcare workers. BMT LIFE contains four questions with a total of 10 items which assess use of this performance indicator. The coding for the possible answers to these questions is Always (4 points), Usually (3 points), Sometimes (2 points), Rarely (1 point), and Never (0 points). Participant responses were calculated for each individual item, with a mean calculated for each item and an overall mean calculated for each PI. The second PI is use of maximal sterile barriers during CVC insertion. This multiple answer question has seven possible answers, with five correct answers required for a maximum of five points. Some BMT centers reported having CVCs inserted in just 72 Table 13. Question(s) Used to Calculate Use of Each CDC Recommendation CDC BMT LIFE Question Response That Recommendation Question(s) Type Indicates the Highest Level of Self-reported Use of the 2002 CDC Guidelines I.A. Educate health-care A - 1: a. RS Always workers about IV catheter b. RS Always use, insertion, maintenance, c. RS Always and infection prevention. d. RS Always A - 2 : a. RS Always b. RS Always I. B. Assess knowledge of and A - 3 : a. RS Always use of guidelines. b. RS Always A - 4 : a. RS Always b. RS Always VI. A.l. Use 2% chlorhexidine E - 2 RS Always for cutaneous antisepsis E - 3 RS Always before catheter insertion and during dressing changes. VIII. B. Promptly remove IV D - 5 OE Criteria for line removal catheters that are no longer must include language essential. equivalent to "Promptly remove" when "no longer essential for medical management" (CDC, 2002a, p. 14, 16). IV. A. Use maximal sterile D- 1 MA a, b, d, e, and g barrier precautions during D - 2 MA a, b, d, e, and g catheter insertion. D - 3 MA a, b, d, e, and g D - 4 MA a, b, d, e, and g Note: General = Refers to general recommendations for use with all intravascular catheters; CVC = Refers to recommendations to be used specifically with central venous catheters; Category IA or IB = Refers to CDC guidelines categories of recommendations 73 Table 13. (Continued) (CDC, 2002a), where Category IA recommendations are strongly supported by research, and Category IB recommendations are supported by some research and "a strong theoretical rationale"; RS = rating scale; MA = multiple answer; OE = open-ended; Always = The response Always indicates that the participant always does the specific behavior mentioned in the survey question. Always = 4 points; Usually = 3 points; Sometimes = 2 points; Rarely = 1 point; Never = 0 points. one setting in the BMT facility, while other BMT centers used more than one setting. To make the scores equivalent for all centers, scores for each setting were added together and divided by the number of settings. This resulted in an overall mean usage score across all settings for each BMT center. The third PI is use of 2% chlorhexidine for skin antisepsis before CVC insertion and during dressing changes. Most participants answered the question about use during dressing changes, but some could not answer the question about use prior to CVC insertion. Therefore, the answers were summed and then divided by the number of questions answered, similar to the scoring described for the second PI. The fourth PI is prompt discontinuation of CVCs no longer medically necessary. This open-ended question was scored from zero to three points, based on specific required language. Central Line Infection Theoretical definition. There are several terms and definitions which fit under the umbrella term "central line infection." The terms used in the BMT LIFE survey include exit site infection, tunnel infection, catheter-associated BSI, catheter-related BSI, pocket infection, and clinical sepsis. Each of these terms is defined precisely in the CDC guidelines (2002a), and the CDC definitions are described in the survey. Operational definition. For the purposes of this study, a central line infection is defined by the CDC (2002a) recommended definitions for each of the CLI terms listed above. Description and Operationalization of Practice Factors Staffing Patterns Staffing patterns refers to the nurse-to-patient ratio, to ensuring appropriate nursing staff levels, and to designating trained CVC personnel for specific CVC-related procedures. Nurse-to-patient ratio was operationalized as the number of patients per RN, LPN, or NA on a typical day shift for each participating BMT center and was collected in that format from the survey. The day shift was used, rather than other shifts, to be consistent with previous studies of the outcomes of nurse staffing levels (Aiken, Clarke, & Sloane, 2002; Pronovost et al, 2001). Ensuring appropriate nursing staff levels and designating trained CVC personnel were measured on a 5-point likert-type scale and operationalized as a score of 0, 1, 2, 3, or 4, reflecting the frequency of having appropriate nursing staff levels (ranging from never [0] to always [4]). All of the staffing pattern items are ordinal level data. Educational Preparation of RNs Educational preparation of RNs was operationalized as the percentage of RNs holding a bachelor's degree. It was obtained in that format from the BMT LIFE survey. The level of these data is ratio. Use of Mid-level Providers Use of mid-level providers was operationalized as either No (0 points) or Yes (1 point). In addition, three dummy variables for use of clinical nurse specialists (CNSs), nurse practitioners (NPs), and physicians' assistants (PAs) were created to assess use of specific mid-level providers. This data level is categorical. Use of Prophylactic Anticoagulants Use of prophylactic anticoagulants was measured using a 5-point likert-type scale with Always = 4 and Never = 0. This is interval level data. Participants were also asked to list the anticoagulant if one were used. Access to BMT Center Rates of CLI Access to BMT center rates or numbers of CLI is dichotomous and was operationalized as the ability (or lack of ability) of a participant to obtain and report CLI rate or number data on the BMT LIFE survey. Participants who supplied CLI data in any format were deemed to have access to BMT center rates or numbers of CLI. Those who checked the box indicating that their institution does not provide CLI data specifically for their BMT centers were deemed not to have access to BMT center rates or numbers of CLI. This variable was dummy-coded as 0 (No) or 1 (Yes) and is categorical. Most Important Practices to Help Prevent CLI The variable most important practices to help prevent CLI was operationalized as self-reported most important practices to help prevent CLI listed by participants in answer to question one in Section F - Practices Which Help to Prevent Central Line Infections. Responses to this question were not analyzed. Such analysis was beyond the scope of this study and may be the subject of future research. Description and Operationalization of the Demographic and Organizational Factors All of the information for the demographic and organizational factors was obtained from the BMT InfoNet website. 77 Type of Transplant Unit Type of transplant unit was operationalized as either adult, pediatric, or combined adult and pediatric. This is a categorical variable. Type(s) of Transplants Performed Type(s) of transplants performed is a categorical variable and was operationalized as either autologous only or autologous and allogeneic. BMT centers performing only autologous transplants were coded 0, and centers performing both autologous and allogeneic transplants were coded 1. Years in Practice Years in practice was derived from the year listed as the year the transplant unit opened. It is ratio level data. Number of Transplants Performed Number of transplants performed was operationalized as the number of transplants listed for each BMT center to 2005. The data is ratio level. Measures of Validity Instrument Validity Content validity. Questions in the BMT LIFE were designed to assess self-reported use of concrete behaviors recommended by the 2002 CDC guidelines, or to assess variations in practice patterns of participating BMT centers. Table 14 shows the alignment of survey questions with the individual CDC recommendations and with the Table 14. Alignment of Survey Questions with Specific CDC Recommendations and the Performance Indicators BMT LIFE Content Areas and Question Numbers CDC Recommendations Performance Indicators Section A - Healthcare Workers 1 - 2 Gener