Effect of Frequency of Changing Point-of-Use Reminder Signs on Health Care Worker Hand Hygiene Adherence

This cluster randomized trial investigates the effect of frequency of changing hand hygiene reminder signs on hand hygiene adherence among health care workers.

interventions by developing an evidence-based bundle of effective strategies. While many have 23 advocated for a multifaceted bundle as the most effective approach to improving hand-hygiene 24 compliance, no studies have systematically identified an optimal combination of interventions. 25 As such, infection prevention programs continue to use a variety of hand-hygiene interventions 26 without clear evidence of which components in isolation or combination are most effective. 27 The two specific aims and associated hypotheses of CREATE Project 2 include: 28 1) Identify combinations of hand-hygiene intervention strategies that optimize hand-hygiene 29 compliance and that could form an evidence-based hand-hygiene bundle for VHA 30 implementation. 31 Hypothesis 1: Combinations of interventions will increase compliance rates more than single 32 interventions. 33 Aim 1 will entail a 30-month cluster-randomized controlled trial that will sequentially test three 34 individual hand-hygiene interventionshand-hygiene point-of-use reminder signs to serve as 35 an environmental cue to action, individual hand sanitizers, and health care worker hand cultures 36 to identify an optimal combination of interventions to increase hand-hygiene compliance. The 37 trial will be conducted in 59 hospital units in 10 VA hospitals in order to test the efficacy of 38 individual and then sequentially added interventions to determine their incremental impact on 39 hand-hygiene compliance. 40 41 2) Identify institutional, organizational, ward/ICU, and individual level facilitators and barriers to 42 implementing hand-hygiene interventions. 43 Hypothesis 2: Facilitators and barriers will pattern around contextual factors such as level of 44 leadership support and organization of infection control programs. 45 Aim 2 will entail a qualitative process evaluation that includes site visits to purposefully selected 46 sites, semi-structured interviews, and observations to examine barriers and facilitators to the 47 interventions and develop contextual insight for implementing and scaling-up the intervention at 48 additional sites as a national initiative. 49 50 51

Introduction 131
A. Background 132 Hand hygiene is key to preventing and controlling methicillin-resistant Staphylococcus aureus 133 (MRSA) and other hospital-acquired infections (HAIs). This project will advance the science of 134 hand hygiene compliance interventions by developing an evidence-based bundle of effective 135 strategies. While many have advocated for a multifaceted bundle as the most effective 136 approach to improving hand hygiene compliance, no studies have systematically identified an 137 optimal combination of interventions. 2, 7-9 As such, infection prevention programs continue to use 138 a variety of hand hygiene interventions without clear evidence of which components in isolation 139 or combination are most effective. 140 141 Hand Hygiene, Infection Control, and Compliance Rates: Hand hygiene is widely considered the 142 most effective method of preventing HAIs. [1][2] Multiple studies have demonstrated that increases 143 in hand hygiene compliance lower the rates of infection in hospital settings. 10-13 Yet, hand 144 hygiene compliance remains persistently low. [1][2]14 In the latest systematic review of hand 145 hygiene compliance, the average compliance rate was 40%. 3 In our own work, we have 146 obtained similar findings. In one study of VA and non-VA hospitals, hand hygiene compliance 147 averaged 32.9% for room entry and 50.8% for exit. 15 However, the average compliance rate 148 reported by infection control programs at the same sites was 87.2%. Thus, rates obtained via 149 rigorous observation methods following established guidelines were much lower than those 150 found in routine practice, where observation methods: 1) are much more variable, 2) frequently 151 do not follow recommended methods, [16][17][18]  project-performance was collectively at 48%...It's interesting that a number of the hospitals 156 were misled by faulty data to believe that they were doing as well as, say, 85%, at baseline 157 rather than 48%." 21 The 2012 TJC National Patient Safety Goal for hand hygiene 158 (NPSG.07.01.01) now states: "Comply with either the current Centers for Disease Control and 159 Prevention (CDC) hand hygiene guidelines or the current World Health Organization (WHO) 160 hand hygiene guidelines. Set goals for improving compliance with hand hygiene guidelines. 161 Improve compliance with hand hygiene guidelines based on established goals." However, this 162 same goal does not require facilities to follow evidence-based guidelines on measuring hand 163 hygiene, despite TJC's publication of such guidelines. 18 In addition, little guidance exists in the 164 TJC's goal on which intervention strategies improve hand hygiene compliance rates-and 165 prevent infectious disease. 166 167 Hand Hygiene Clinical Practice Guidelines and VHA Directives: Currently, VHA Directive (2011-168 007) "Required Hand Hygiene Practices" provides regulations on hand hygiene practice in VHA. 169 This Directive is primarily a reiteration of CDC and WHO guidelines for a VHA context, including 170 when to use antimicrobial soap or alcohol-based hand rub, hand hygiene practices related to 171 glove usage, availability and placement of hand hygiene products, and where to obtain VA 172 Infection: Don't Pass It On campaign materials from the Office of Public Health 173 (http://www.publichealth.va.gov/infectiondontpassiton/). It also requires all VA facilities to have a 174 written hand hygiene policy. What is missing in the Directive is guidance on the best 175 interventions for improving hand hygiene. WHO guidelines provide some recommendations for 176 interventions, but focus on multimodal, promotional campaigns at the hospital-system or country 177 level, while also acknowledging science is lacking on which components of the campaign are 178 essential for hand hygiene improvement. 2 VHA has made it a priority to develop a web-based 179 Hand Hygiene Toolkit for Infection Control Professionals (ICPs) in the field, 22  hygiene compliance in an additive or synergistic manner, or do they detract from one another? 212 The study only observed the combined effect of all components together. In addition, Pittet and 213 colleagues' 12 work is frequently cited for its innovation, sustainability, and measured effect on 214 HAIs, but it was a single-center study with potential lack of generalizability outside the particular 215 Swiss hospital in which it was conducted. What is needed to advance the science of hand 216 hygiene are well-designed, multi-site studies that disentangle the components of bundled hand  217  hygiene interventions.  218  219  Identifying Three Hand Hygiene Intervention Strategies: Based on our own systematic review,  220 the VHA-wide hand hygiene survey, and our pilot data,(see Preliminary Studies Section ) we 221 have selected three interventions that are most likely to improve hand hygiene compliance. 222 These interventions are 1) hand hygiene point-of-use reminder signs to serve as an 223 environmental cue to action; 2) individual hand sanitizers, and 3) health care worker ( care? Answers to these questions will help address the broader question of: What is the "hand 276 hygiene culture" of the facilities? In turn, this information will assist in tailoring the 277 implementation plans for the interventions to address some of the known barriers and take 278 advantage of existing facilitators. 279 280 Besides focusing on organizational and structural aspects that influence hand hygiene 281 practices, we also draw on the HBM in the design of our intervention targeting individual HCW 282 behavior and to guide our research questions in Aim 1. The HBM incorporates six constructs: 283 perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to 284 action, and self-efficacy. 58 The traditional focus of the HBM has been to change patients' health 285 behavior; however, the intended behavior change of this intervention involves HCWs. Work in 286 this area suggests a need to shift perceived susceptibility from the person performing the 287 behavior (HCW) to the individual most susceptible (patient). Therefore, the intervention 288 concentrates on changing HCW behavior by focusing on their perceptions regarding the threat 289 that pathogens carried on their own hands pose to the susceptible patient. To change the 290 HCWs' behavior, the intervention relies on environmental cues to action (hand hygiene signs 291 and hand cultures), reinforcing perceived benefits (using signs containing gain-framed 292 messages), removing barriers (systematic distribution of personal hand sanitizer), and HCWs' 293 perceived susceptibility and severity of the patient and their role in increasing their patients' risk 294 (patient-focused signs and display of hand cultures). 295 296 Signs and the display of HCW hand cultures are environmental cues to action to remind workers 297 to disinfect their hands. The messages each portrays will also be based on psychology and 298 health communication theory to enhance the effect of the cue to action. Specifically, the signs 299 will contain gain-framed messages that emphasize the benefits of proper hand hygiene (as 300 opposed to the harms associated with poor compliance). Although this distinction may seem 301 subtle, accumulating evidence suggests that gain-framed messages are more effective than 302 loss-framed messages at promoting certain health behaviors such as hand hygiene. 33, 59-60 The 303 signs will also be patient-focused based on recent research and our own pilot work. 34  use signs had a non-significant 11.68% increase in exit (p=0.44) compliance. Importantly, the 369 sign with patient-focused and gain-framed language had the highest entry compliance 53% vs 370 29% for the other signs (p=0.042) and the highest exit compliance 79% vs 52% for the other 371 signs (p=0.36). Although the signs led to minimal improvement in HH compliance overall, the 372 gain-framed, patient-focused sign was associated with substantial increases in entry and exit 373 compliance. This finding highlights the potential importance of the specific type of messaging 374 strategy that is used. Therefore, after completing additional observations to verify the effect, we 375 will test variations of gain-framed, patient-focused signs. Given positive findings from gain-376 framed messages 33 and patient-focused messages 34 on hand hygiene signs, we expect using 377 all gain-framed, patient-focused signs will increase the effect of cue-to-action signs overall. independently assessed each article and a third investigator resolved any discrepancies. Two 444 data extractors recorded data from each selected study on data collection forms designed for 445 the systematic review. Recorded data included study design (e.g., cluster randomized trial, pre-446 post design, time series), interventions, outcome, adjustment for potential confounders, and 447 potential biases. We rated the study quality using standard methods. 26, 81 448 449 Overall, 12 of the 14 articles assessed met inclusion criteria. All but one of the studies was a 450 before and after quasi-experimental design. Two-thirds (n=8) of the articles had less than 5,000 451 hand hygiene observations and only 2 had more than 10,000. Four of the 12 studies were 452 conducted in the United States, while a majority (n=6) were conducted in European countries. 453 Most studies (n=11) were conducted in acute-care facilities in general medical wards and 454 intensive care units (ICUs), although 3 were conducted throughout entire hospitals. Following 455 the recommendations of many in the field, only one study tested a single intervention (signs) 456 and the remainder included multiple intervention components. The majority (n=9) of the studies 457 included 4 or more types of intervention strategies, with 2 studies including over 7. None of the 458 studies attempted to disentangle the individual effect of the interventions. The most commonly 459 bundled intervention components were signs (n=9), followed by education (n=9), audit and 460 feedback (n=8), and increased access to hand hygiene products (n=8). The final meta-analysis 461 has not been completed, but hand hygiene compliance rates in pre-intervention periods ranged 462 from 6.5% to 54.3% and from 49.9% to 95.6% in the post-intervention period. Under Secretary of Health stated in his response to the OIG report that the results of these 527 studies would be used to inform a web-based hand hygiene toolkit maintained by the Office of 528 Public Health, IDPIO. Findings from this proposed study will be directly disseminated through 529 the hand hygiene toolkit. Drs. Reisinger and Vander Weg will develop the materials for 530 dissemination which will include the most effective hand hygiene signs, detailed information 531 about the "bundled" intervention and its overall effect, and recommendations for tailoring and 532 implementing the intervention based on the Aim 2. Thus, this project will lead to immediate 533 impact by providing products for dissemination to improve quality of care. 534 535 The effort to improve hand hygiene practices continues to expand in VHA. The proposed study will use a parallel, mixed-methods design that will integrate qualitative 568 research (Aim 2) with a cluster-randomized controlled trial (Aim 1). 569 570 Specific Aims: The two specific aims and associated hypotheses of the project include: 571 1) Identify combinations of hand hygiene intervention strategies that optimize hand hygiene 572 compliance and that could form an evidence-based hand hygiene bundle for VHA 573 implementation. 574 Hypothesis 1: Combinations of interventions will increase compliance rates more than single 575 interventions. 576 Aim 1 will entail a 30-month cluster-randomized controlled trial that will sequentially test three 577 individual hand hygiene interventions to identify an optimal combination of interventions to 578 increase hand hygiene compliance. The trial will be conducted in 59 hospital units in 10 VA 579 hospitals in order to test the efficacy of individual and then sequentially added interventions to 580 determine their incremental impact on hand hygiene compliance. 581 582 2) Identify institutional, organizational, ward/ICU, and individual level facilitators and barriers to 583 implementing hand hygiene interventions. 584 Hypothesis 2: Facilitators and barriers will pattern around contextual factors such as level of 585 leadership support and organization of infection control programs. 586 Aim 2 will entail a qualitative process evaluation that includes site visits to purposefully selected 587 sites, semi-structured interviews, and observations to examine barriers and facilitators to the 588 interventions and develop contextual insight for implementing and scaling-up the intervention at 589 additional sites as a national initiative. 590

Resources and Personnel 592
The central site will be the Iowa City VA Health Care System. Research staff at Iowa City will be 593 responsible for patient data pulled from VA administrative datasets and compiling and securely 594 storing all hand hygiene observation data for Aim 1. Iowa City research staff will be the only staff 595 performing Aim 2 (qualitative process evaluation). For Aim 2, the Iowa City research staff will 596 recruit VA personnel to participate in individual and group semi-structured interviews. They will 597 also be responsible for the informed consent protocol and conduct all of the interviews. 598 The nine other participating sites will contribute to efforts of Aim1 of the study. Each site will be 599 in charge of 1) conducting observations of potential HCW hand hygiene opportunities (Site 600 Research Coordinator), 2) hanging up the hand hygiene signs (Site PI/Hospital Epidemiologist), 601 3  The table also  615 indicates whether the site is in VISN 23 and whether Aim 2 will be conducted at the site. The 616 sites were carefully selected to represent geographical variation-covering the Northeast to 617 Northwest and Midwestern, Southern, and Western states. In addition, the hospitals represent 618 small, medium, and large facilities within VA. Finally, this study is embedded in a larger set of 619 integrated studies as part of a CREATE application focused on combating MRSA and other 620 HAIs. One of the broader aims of these studies is to develop a VISN-level computer model to 621 assist leadership in policy decision-making around the spread of MRSA and other HAIs. 622 Therefore, we selected all three tertiary care hospitals in VISN 23 to ensure we have a rich set 623 of data for a single VISN. The 6 sites for Aim 2 were selected to balance the need for depth of 624 data in a single VISN (Iowa City, Minneapolis, and Omaha), geographic variation (Baltimore, 625 Portland, Miami, and VISN 23), and successful collaboration on previous projects (Baltimore, 626 Iowa City, Portland). All general medicine and surgical wards and ICUs will participate in the 627 study. (Psychiatric wards were excluded due to restrictions regarding alcohol-based hand 628 sanitizer.) 629 MRSA infection rates will be collected monthly throughout the project for each of the 59 units.

642
Operationalization of Primary Outcome: Nurses, physicians and other staff will be observed for 643 10 consecutive minutes of clinical activity on the study wards and ICUs. Observers will be 644 research staff specifically trained and validated in hand hygiene observations. (See Appendix  645 for standardized observation worksheet.) The observation periods will be 10 minutes long so 646 that observers can be transiently on the wards/units without their presence being noticed, which 647 should minimize the Hawthorne effect. 14 As Dr. Perencevich has done in his previous study, the 648 observers will be given a "cover story" so that the true purpose of their observations will be 649 hidden. 14 Therefore, while observers will be on the wards, the HCWs will not know they are 650 being watched for infection control purposes, so they will not artificially increase their 651 compliance. In ICU settings, we will state that the observers are collecting data for a severity of 652 illness aggregate measure, such as the APACHE score. In non-ICU, medical/surgical ward 653 settings the observer will state that they are doing a patient movement and activity of daily living 654 survey to determine the proportion of time patients who are isolated or non-isolated receive their 655 trays of food or are transported outside of their rooms to physical therapy or radiology, etc. The 656 standardized observation instructions and training protocol has been successfully implemented 657 in other studies by the team. 14-15 Hand hygiene opportunities at entry and exit of a patient's room 658 will be recorded. We are using entry and exit because it is the one of the most common "hand 659 hygiene opportunities" reported in the literature and by VA facilities. However, we will also 660 record observations of the 5 Moments of Hand Hygiene as current hand hygiene observation 661 guidelines recommend. 2, 54, 80 Observations will occur during the day shift (75%) and night shift 662 (25%). In addition, observations will include weekends (20%) and weekdays (80%). Interventions: Based on our systematic review, VA-wide survey, and our own pilot data we have 677 identified the three interventions with the greatest likelihood of improving hand hygiene 678 compliance and also in need of further study individually and in combination. The three 679 interventions are: 1) theoretically-based and empirically-tested hand hygiene signs to serve as 680 environmental cues to action and to reinforce of HCWs' perceptions of patient susceptibility, 2) 681 personal hand sanitizers to decrease barriers and improve accessibility to hand sanitizer, and 3) 682 HCW hand cultures to test the combination of individual feedback, environmental cues to action, 683 and HCW perceptions of patient susceptibility and severity. This proposal will identify an optimal 684 combination of interventions to increase and sustain hand hygiene compliance. 685 686 helping patients appear to be most effective in changing HCW behavior. However, evidence 699 also strongly suggests that repeated exposure to the same cues leads to a process of 700 habituation such that their effectiveness at prompting behavior is diminished over time. 61-64 701 Therefore, it is important to identify strategies for maintaining the salience of hand hygiene cues. 702 One approach is to periodically modify the cues to increase their novelty and enhance the 703 potential that HCWs will attend to and act on them. 54 For the current proposal, we will study how 704 to sustain their power as a cue to action by varying the frequency of changing the signs. 705 706 The specific signs that we will use in the proposed study will be based on those found to be 707 most effective from our ongoing pilot work. We will use signs from the next phase of our pilot 708 work in which we are testing multiple gain-framed, patient-focused signs with slight variations on 709 message verbiage, color, and HCW and patient demographics. The six most effective signs will 710 be changed at varying frequency depending on their randomization assignment. Although 711 evidence suggests that periodically changing reminders is likely to reduce habituation and 712 enhance their impact, little is known about the frequency at which this should occur. Therefore, 713 signs will be changed at varying frequencies: every week, once a month, or displayed for the 714 entire 6 months of this phase of the study, with effects on hand hygiene compliance compared 715 across conditions. We will continue to use point-of-use to determine the placement of the signs. 716 During the first site visits the qualitative team will document current practices surrounding the 717 display of hand hygiene posters. They will also discuss how to best implement point-of-use 718 signs with the infection control (IC) team at each site. We have worked with Baltimore VA, Iowa 719 City VA, and Portland VA to create solutions to displaying signs next to dispensers on hospital 720 units. For those sites at which we do not conduct site visits as part of Aim 2, we will conduct 721 interviews over the phone to gather the necessary information. We will conduct the phone 722 interviews after the site visits in ensure we will be able to integrate all necessary questions into 723 the interview guide. hygiene observation protocol, we will also document 752 the source of the hand sanitizer that is used (e.g., 753 individual hand sanitizer, wall dispenser at point of 754 care, etc.), as well as whether the individual hand sanitizer is visibly present on the HCW, to 755 determine the extent to which this approach contributes to overall hand hygiene. effectiveness of individual feedback with environmental cues to action. Furthermore, direct 771 visual evidence regarding the presence of pathogens on one's hands (and that of their 772 colleagues) is likely to provide an especially salient affective or emotional source of motivation 773 for improving hand hygiene (i.e., increases HCWs' sense of placing patients at risk). 774 775 For this intervention, the Site PI/Hospital Epidemiologist will be responsible for collecting blood 776 agar culture plates of HCWs' hands on each unit once a month at the intervention hospitals. The 777 Hospital Epidemiologist or a designated IC staff will go to each ward/unit and ask HCWs on the 778 ward/unit at the time to place their hand on a blood agar plate. The HCW will be told the results 779 (pictures of culture plates after organisms have been grown) will be confidentially displayed in 780 staff work areas; however, if HCWs would like to their specific results, they can provide a four 781 digit number that will be associated with their agar plate. We will not record their names or track 782 the number. The number will be written on the culture plate and then on the digital photo for 783 display. The research team will not record the number in any other way and will not record HCW 784 names or any other identifying informations.  areas, but out of patient view. During the initial site visits, the qualitative team will observe 816 current feedback mechanisms on the wards and units and work with the IC team to plan a way 817 of displaying the feedback. This information will be used in interviews and planning sessions 818 conducted with the four sites the qualitative team does not visit. Since the collected of culture 819 plates is an intervention and not being done for data collection purposes, all culture plates will 820 be destroyed via proper disposal techniques and digital photographs of the culture plates will be 821 used for educational purposes only. 822 823 Study Procedure: The trial will entail 6 phases. (See Figure 3.) As an overview, Phases 2, 4 and 824 6 will be intervention periods and Phases 3 and 5 will establish a new baseline and serve as 825 washout periods. Importantly, Phase 2 randomization will occur at the ward/ICU level since our 826 pilot data suggests room entry signs can be randomized at the ward/ICU level. However, to 827 avoid crossover effects, the other interventions will be randomized at the hospital level in 828 Phases 4 and 6. 829 830 Phase 1 will obtain baseline rates of hand hygiene compliance and MRSA on each unit at each 831 hospital. Phase 2 will compare different strategies for changing hand hygiene signage on 832 hospital units to determine if periodically varying signage more effectively modifies hand 833 hygiene behavior than using static environmental cues. The 59 individual hospital units will be 834 randomized to 3 groups (~19 units per group): 1) no change in signs; 2) changing signs 835 monthly; and 3) changing signs weekly. To improve power, we will use stratified randomization 836 with matching based on baseline hand hygiene compliance rates during Months 1-6. 69 Block 837 randomization will be used to ensure that baseline compliance rates will be evenly distributed 838 across conditions. To do this, we will rank the units by compliance rate and then randomize by 839 blocks of three using a computer-generated list of random numbers prepared by the Data 840 Manger in consultation with the biostatistician. During Phase 3, the sign change strategy that 841 was associated with the highest hand hygiene compliance will be universally implemented in the 842 59 study units in all 10 hospitals, this will set a new baseline and serve as a wash-out period. In 843 our current study of hand hygiene interventions, we were able to detect a difference within six 844 weeks. Thus, 3 months is likely enough time for wash-out. 845 846 Phase 4 will compare the additional benefits of the other 2 interventions-distribution of 847 individual hand sanitizers to healthcare workers and obtaining cultures of healthcare workers' 848 hands with visual feedback on hand contamination. In Phase 4, hospitals will be randomized to 849 3 groups: 1) signage only (4 hospitals); 2) signage and hand sanitizers (3 hospitals); and 3) 850 signage and hand cultures (3 hospitals). Group assignment will again be stratified using block 851 randomization (4, 3, 3), based on hand hygiene compliance rates measured during Phase 3. 852 Individual strategies will be randomized by hospital to decrease contamination between 853 wards/units. During Phase 5, the Phase 4 intervention associated with the highest compliance 854 rates will be universally implemented; if both interventions were equally efficacious, the signage 855 only hospitals will be randomized to receive either the hand sanitizer or hand culture 856 intervention. Phase 5, like phase 3 will also serve as a washout period. Phase 6 will determine 857 the potential incremental benefits of all 3 interventions in combination by re-randomizing 858 hospitals to 2 groups: 1) signage and one of the other interventions (5 hospitals); and 2) all 3 859 interventions (5 hospitals). Again, the interventions will be randomized by hospital to decrease 860 likelihood of contamination. 861 862 Given that signs are already a frequently utilized, well-accepted, inexpensive, and easy to 863 implement strategy, we are starting the intervention with theoretically-based, empirically-864 supported signs and testing the rate at which they should be changed in order to most 865 effectively enhance HCW hand hygiene. Individual hand sanitizers are also relatively low cost 866 and are currently available to HCWs in most VA hospitals. However, little is known about their 867 impact on improving hand hygiene compliance rates as a stand-alone intervention. Finally, 868 displaying the results of HCWs' blood culture plates is an innovative new strategy that has 869 developed primarily in quality improvement projects, but it has never been directly tested 870 against (or in combination with) other interventions. It is also relatively low cost at $1 per plate 871 and involve little lab processing. Collectively, this approach will allow us to evaluate the 872 individual and combined effects of the different intervention strategies in order to identify the 873 optimal hand hygiene bundle. 874 875 Aim 2 is a qualitative process evaluation at 6 sites in the study. It involves pre-intervention and 876 post-intervention site visits by the PI and Qualitative Analyst, which will include observations, 877 semi-structured interviews with the IC team, and focus groups with ward/ICU staff. Semi-878 structured phone interviews will also be conducted with the IC team after each intervention is 879 implemented. 880 881 During the first 6 months (prior to the baseline period [Phase 1]), Drs. Reisinger and Moeckli will 882 travel to each of the 6 hospitals selected for Aim 2 and oversee a qualitative process evaluation 883 of hand hygiene practices. The evaluation will include observations on individual units, semi-884 structured interviews with IC staff and leadership, and focus groups with staff on one high and 885 one low performing unit at each hospital. Semi-structured interviews will be conducted 886 individually interventions to try to improve compliance. The qualitative team will return to Iowa City to 906 analyze the data and develop case studies of each hospital's hand hygiene practices, including 907 compliance data collected for the study. This will serve as a baseline description of the "hand 908 hygiene culture" of each facility. 909 910 After each interventional phase is implemented, semi-structured interviews will be conducted 911 over the phone with the IC team and nurse managers at each hospital (during the washout 912 periods and during the post-intervention qualitative evaluation). The interviews will focus on 913 perceived barriers and facilitators to implementation, staff reaction to the strategies, perceived 914 impact on workflow and patient/provider relationships, and perceptions of each intervention's fit 915 with the culture of the hospital. 916 917 Toward the end of Phase 6, Drs. Reisinger and Moeckli will return to the sites and replicate the 918 same qualitative evaluation protocol. The qualitative evaluation will identify changes in "hand 919 hygiene cultures" and document feedback about the interventions from frontline staff during unit 920 focus groups. 921 922 All interviews will be audiorecorded on digital encrypted recorders or over the phone directly to a 923 secure server. They will then be transcribed by in-house transcribers and reviewed for accuracy. 924 All textual data (field notes and transcripts) will then be imported into MAXQDA, a qualitative 925 data management and analysis software program. 91 926 927 928  929  AIM I:  930 With regards to observations of hand hygiene compliance, no individual identifying information 931 will be collected from healthcare workers or patients. Data regarding MRSA and hospital-932 acquired infection rates will be collected from the VA database, IPEC. All data will be collected 933

Risk protection
anonymously. There is no patient-level data for the observation portion of the study; healthcare 934 worker data will be anonymously observed and then aggregated. Removal of access to 935 research study data will be accomplished for study personnel when they are no longer part of 936 the research team and any incidents involving theft of data, etc., will be reported immediately. 937 MRSA and hospital-acquired infection outcome data will be analyzed using VA datasets. Given 938 the number of veterans included in this retrospective analysis and the use of existing data, 939 informed consent will not be obtained from patients. This practice is consistent with practices for 940 studies that use existing administrative data and medical records data. However, our data 941 management protocol will ensure the protection of the protected health information thereby 942 minimizing risks to individual veterans. A waiver of consent is included. 943 Risk will be minimized by limiting access to data, maintaining study paper files in locked offices, 944 and storing data on password-protected computers and on servers, which are secured in locked 945 invitation from the PI of the proposed research project. The letter will contain elements of 964 consent. The PI or a member of her study team will then review the letter before conducting an 965 interview with any member of the IC team. Site visits will take place during the baseline period 966 and at the end of the interventions. 967 968 Four designated sites will not participate in site visits, but will participate in phone interviews 969 with at baseline and post-intervention. Each member of the IC Team will be emailed a letter of 970 invitation from the PI of the proposed research project. The letter will contain elements of 971 consent. The PI or a member of her study team will then review the letter on the phone before 972 conducting an interview with any member of the IC team. 973 974 In addition, phone interviews will be conducted at all 10 sites during the washout periods to 975 gather any feedback from the IC team on implementation barriers for the previous intervention 976 conducted at their site. 977 978 VA Healthcare Workers (HCWs): With assistance from the Site PI, the Iowa City Research 979 Coordinator will contact the appropriate supervisors and inform them of the Qualitative Team's 980 site visit. She will arrange a time for the focus group on two different units. At the time of the 981 focus group, participants will be given an informational sheet and the PI will review the elements 982 of consent (see focus group script with elements of consent and information sheet) with the 983 participants in the focus group. They will be given the opportunity to ask questions and have all 984 their questions answered. Participation in the focus group will be an indication of consent. This 985 consent process has worked well in other focus group projects the team as conducted. 986 987 Data Collection: The study personnel will be responsible for reviewing elements of consent 988 before each point of data collection. 989 Interviews and focus groups will be recorded on encrypted recorders and files will be uploaded 990 directly to a secure server. Recordings will be transcribed and reviewed for accuracy. All data 991 (recordings, transcripts) will be stored on a secured server. Each participant will be assigned a 992 unique study identifier which will be used on all data. Subject identities (e.g., names, addresses, 993 professional title) will be kept separate from the data and linkage information will only be 994 available to the PI and Research Coordinator. 995 996 For proposed research, confidentiality risk will be minimized by limiting access to data to study 997 personnel, removing unique participant identifiers (e.g., names) from analytical files, maintaining 998 study data forms and files in locked offices, and storing study databases on password-protected 999 computer files housed on a secure server. This server is configured and maintained by VA IRM 1000 staff and meets all VA security standards. The server is in a secure location at the Iowa City 1001 VAMC, where it is in a video-monitored area behind locked doors. All analyses will be Power to Detect a 5% Increase in Hand Hygiene Rate Phase 2 60% 99% Phase 4 65% 99% Phase 6 70% 99% 1108 Analysis: Following methods we have outlined previously, 90 we will use quasi-Poisson mixed 1109 effects models to account for matching and the cluster-randomization design, as observations 1110 over time within individual rooms, units and hospitals are expected to be correlated. The 1111 dependent variable in each model will be the monthly hand hygiene compliance rate. The 1112 numerator will be the number of times a HCW disinfects their hands while the denominator will 1113 be the number of opportunities for hand hygiene. The ideal denominator is the WHO 5 Moments 1114 of Hand Hygiene 2 and where possible, we will collect these data. However, our current IIR 09-1115 099 suggests that the most reliably measurable moments are room entry and room exit. 1116 Separate models will be run for the WHO 5 moments (combined as one dichotomous measure 1117 and for each of the individual "moments"), room entry compliance, and room exit compliance. 1118 Independent variables will include the tested intervention, the bed size and type of unit, monthly 1119 census, estimated monthly nurse to patient ratios, and other known confounders associated 1120 with hand hygiene compliance. The results will advance understanding of the effectiveness of 1121 individual interventions and their combined effectiveness as a bundle. 1122 1123 AIM 2: 1124 Analysis: After the initial site visits, the study team will review transcripts and conduct an initial 1125 thematic content analysis. 92-93 Drs. Reisinger, Moeckli and Ms. Páez will read a subset of 1126 transcripts and generate a preliminary codebook through a series of meetings. The thematic 1127 codes will include a priori codes such as hand hygiene compliance data collection, hand 1128 hygiene interventions, facilitators and barriers to hand hygiene compliance, as well as inductive 1129 codes which are likely to emerge during data collection. The codebook will consist of top-level 1130 thematic codes, which will be used to "tag" textual data with the main themes. 94 The codebook 1131 will then be tested against subsequent interviews. 1132 1133 To test the codebook, Dr. Reisinger and the coding team will code one IC team member 1134 interview and one staff focus group and enter coding in MAXQDA10 individually. We will then 1135 meet together to review results. MAXQDA10 allows for side-by-side comparison of coding and a 1136 measure of agreement. We will systematically compare coding and discuss coding until 1137 consensus is reached. Final consensus will be entered in MAXQDA10. If measure of agreement 1138 is 80% or higher between coders, the coding team will divide the transcripts and will continue to 1139 code the remainder of the focus groups and interviews, periodically coding the same transcript 1140 to test whether agreement continues at the 80% level. If the agreement is not 80% for all codes, 1141 Dr. Reisinger will continue to work with the team and revisions to the codebook will be 1142 discussed. The process will be repeated until 80% agreement is reached. At least 10% of the 1143 transcripts will be consensus coded during the coding period. 1144 1145 During Phases 2-6, the qualitative team will continue to apply the codebook to the newly 1146 conducted interviews, but it is likely the codebook will expand to include specific codes about 1147 the study interventions. The coding team will continue to test agreement as the codebook is 1148 revised. 1149 1150 In the final Phase, the coding team will code the transcripts from the post-intervention site visits. 1151 The analysis will assist in writing a second case study of each hospital. The revised case 1152 studies will highlight changes in the "hand hygiene culture". It will also be used to develop a 1153 report on the facilitators and barriers to the interventions within the context of the intervention 1154 sites. This report will be disseminated with the intervention results to improve implementation of 1155 the final hand hygiene bundle. 1156 1157 After top-level coding is complete, the research team will discuss which codes warrant more 1158 detailed subcoding. Our team has followed Miles and Huberman's 95 approach to subcoding, 1159 called matrix analysis, in several projects. In matrix coding, we develop a list of subcodes under 1160 a top-level theme and then code both for the presence or absence of the subcode, as well as 1161 content. MAXQDA10 has several features which aid in this level of analysis, including 1162 hierarchical coding, weighting of coded segments, and multi-factor queries in matrices. This 1163 coding will provide more detailed analysis on specific issues the research team finds most 1164 relevant and will primarily be used in the development of manuscripts for peer-reviewed 1165 publications Subjects are entitled to decide to end their participation during an interview without any loss of 1175 benefits to which they are entitled. Their interview will be ended. If the participant would like to 1176 withdraw after the interview has already occurred, for data collected prior to withdrawal, the 1177 study team may continue to review the data already collected for the study but cannot collect 1178 further information, except from public records. 1179 1180 1181