Determinants

Foundational to any implementation effort is understanding factors that promote or hinder change. Promoting factors are called ‘facilitators’ and hindering factors are ‘barriers.’ Determinants of these factors may be individual, such as the preferences, needs, attitudes, and knowledge of HCP, hospital leaders, patients, and visitors. An individual may be a strong, engaged leader (a facilitator) or an unengaged obstructor (barrier). Determinants may include a team’s composition or ways of communicating, an organization’s culture and capacity, or a system’s policies and resources. ^{Reference Aarons, Hurlburt and Horwitz16} Organizationally, implementation may be facilitated or impeded by expectations and allocation of time (eg, competing priorities, data collection burden, provision of time to dedicate to an effort, fast turnaround at the expense of sustained processes), resources (eg, ease of adapting the EMR, staff capacity, and turnover), and leadership support ^{Reference Saint, Kowalski, Banaszak-Holl, Forman, Damschroder and Krein17} or follower buy-in. ^{Reference Saint, Kowalski, Banaszak-Holl, Forman, Damschroder and Krein18}

Facilitators and barriers affect implementation to differing degrees. For example, an individual practitioner may oppose a change (ie, be a barrier), but the supervisor may be able to facilitate to overcome the opposition. Alternatively, a practitioner may champion a change, but without the support of the leadership, they may be unable to initiate it. Additional influential factors include context, level of engagement, and reliability (see Table 1 for a glossary of terms).

Table 1. Glossary of Terms

Failure by HCP to adhere to a guideline or standard is a common basis for initiating an improvement project. The Cabana Framework ^{Reference Cabana, Rand and Powe19} is a useful tool to understand how addressing real or perceived barriers can make an implementation effort successful. The framework employs 3 domains (ie, knowledge, attitude, and behavior) to understand the spectrum of barriers. The Expert Recommendations for Implementing Change (ERIC) ^{Reference Powell, Waltz and Chinman20} is another resource to help map barriers to strategies and identify appropriate implementation models or frameworks.

Measurement

Data are essential for implementation to establish baselines, identify opportunities, measure progress, and justify use of resources to organizational leaders. No single method or measure will work for all situations, and standardized measures often are not available. Different frameworks lend themselves to specific methodologies, but any chosen method must do the following:

• Be appropriate for the question(s) it seeks to answer.

• Adhere to the method’s rules for data collection and analysis. As with any project, it may be prudent to review the analytic plan with an expert to ensure that data collected will yield a result.

Choosing measures

There are 3 general types of measures employed in implementation ²¹ :

1. 1. Outcome measure: The ultimate goal of a project, such as reduced surgical site infections or improving antimicrobial susceptibility patterns.

2. 2. Process measure: The action taken to reach the desired outcome, such as adherence to a prevention bundle or compliance with hand hygiene standards.

3. 3. Balancing measure: An undesired outcome that could be caused by changing a system, such as increased staff absences due to dry skin from a hand hygiene product or due to side effects from a required vaccine.

Ideally, all 3 types of measures are included in a project. For example, a project seeking to reduce ventilator-associated pneumonia (VAP, ie, outcome) seeks to increase early extubation (ie, process) but needs to ensure a rise in reintubations or unplanned intubations is not occurring (ie, balancing). At times, a balancing measure may be difficult to identify due to the rarity of an event or an indirect relationship between outcome or process and balancing measures. In the VAP example, using the rate of nonventilator hospital-acquired pneumonia (NV-HAP) could help identify patients who develop NV-HAP following extubation (implying they were extubated too early), but not every patient with NV-HAP will have been intubated prior to diagnosis.

Similarly, choosing an outcome measure may be difficult if the likelihood of an outcome is multifactorial or exceedingly rare. In the case of hand hygiene, improved adherence (process) should prevent nosocomial transmission (outcome). However, an overall HAI rate may not reflect the change because hand hygiene is one of many potential factors that affect nosocomial transmission. Also, it may not be possible to count all prevented transmissions. For example, a patient would not be counted if they experienced onset of an upper respiratory infection following discharge but did not require readmission. In the case of antimicrobial resistance, improved contact precaution adherence for multidrug-resistant organism (MDRO) patients (ie, process) and improved antimicrobial stewardship (ie, process) should decrease morbidity and mortality due to antimicrobial resistance (ie, outcome) but may be difficult to demonstrate in a single center or short period. ^{Reference Livorsi, Drainoni and Reisinger10} In these examples, the focus of the project might be the process and balancing measures, with attention to but not reliance on the outcome.

Often, measures that are standardized and utilized broadly are referred to as ‘benchmarks.’ Measures also may be developed locally and used in a combination with benchmark measures. For instance, a facility may start with NHSN event definitions ²² and adapt them as definitions change over time or as needed based on the suitability to their setting (eg, pediatric, long-term care, home healthcare). ^{Reference Advani, Murray, Murdzek, Aniskiewicz and Bizzarro23} As another example, facilities typically use the WHO 5 Moments process measures to identify occasions when HCP should perform hand hygiene during patient care, but the methods of measurement of adherence can vary. A facility may measure adherence to all moments, adherence to a specific moment, or the amount of hand hygiene product used. ²⁴ When possible, it is important to use the least resource-intensive means of data collection because resources are needed to feed data back to those who were monitored. Monitoring in combination with feedback has been shown to influence change and be more effective when delivered at a high frequency. ^{Reference Jamtvedt, Young, Kristoffersen, O’Brien and Oxman25}

Choosing a method

Table 2 provides a nonexhaustive list of methodologies commonly used for implementation measurement. For a research-focused overview, readers are encouraged to review the 2016 SHEA series on research methods in healthcare epidemiology and antimicrobial stewardship ^{Reference Morgan, Safdar, Milstone and Anderson26} and Livorsi et al. ^{Reference Livorsi, Drainoni and Reisinger10}

Table 2. Methods for Measurement

The 4 “Es”

Pronovost et al articulated the 4 Es (Engage, Educate, Execute, and Evaluate), which may be the most pervasive model used in US healthcare epidemiology. ^{Reference Pronovost, Berenholtz and Needham30} This model is well suited for large-scale projects that include multiple sites. Its cyclical nature allows for formative work and feedback to drive modifications and adaptations, and it provides a guide for resolving knowledge gaps through education. However, it does not include targeted strategies to address multilevel barriers that hinder putting knowledge into practice.

The following 4 E strategies guide organizational change efforts:

1. 1. Engagement: To motivate key working partners to take ownership and support the proposed interventions.

2. 2. Education: To ensure key working partners understand why the proposed interventions are important.

3. 3. Execution: To embed the intervention into standardized care processes.

4. 4. Evaluation: To understand whether the intervention is successful.

The 4Es guide improvement teams in planning to address key partners for the implementation process: senior hospital leaders, improvement team leaders, and frontline staff. Planning for and utilization of multifaceted interventions that address the 4Es, coupled with explicit efforts to improve teamwork and safety culture, ^{Reference Sexton, Berenholtz and Goeschel31} have been associated with reductions in HAIs ^{Reference Berenholtz, Lubomski and Weeks32–Reference Pronovost, Watson, Goeschel, Hyzy and Berenholtz34} and mortality ^{Reference Lipitz-Snyderman, Steinwachs, Needham, Colantuoni, Morlock and Pronovost35} and increased cost savings. ^{Reference Waters, Korn and Colantuoni36}

Behavior Change Wheel

The Behavior Change Wheel (BCW) is the result of an effort to link interventions with targeted behaviors more directly. It was developed by Michie et al, ^{Reference Michie, van Stralen and West37} who evaluated 19 existing behavior change frameworks for comprehensiveness (ie, applicability to any intervention), coherence, and link to a behavioral model. The result was a 3-layered tool:

1. 1. A behavior system composed of capability, opportunity, and motivation (COM-B)

2. 2. Nine intervention functions that can be used to affect behavioral change

3. 3. Seven policy categories that enable or support interventions to enact the desired behavior change.

One strength of the model is its nonlinearity, meaning that >1 behavioral system component, intervention function, and policy category can apply to an effort to affect change. Additionally, the model attempts to incorporate contextual influences on behavior, which the authors refer to as ‘automatic’ functions such as emotions and impulses that arise from associative learning and/or innate dispositions, as opposed to more reflective processes involving evaluations and plans. The BCW has been used widely in health promotion efforts such as smoking cessation ^{Reference Gould, Bar-Zeev and Bovill38} and obesity and sedentary behavior reduction, ^{Reference Ojo, Bailey, Brierley, Hewson and Chater39–Reference Munir, Biddle and Davies41} and COM-B has been used to investigate hand hygiene adherence ^{Reference Schmidtke and Drinkwater42,Reference Lambe, Lydon and Madden43} and antibiotic prescribing. ^{Reference Tomsic, Ebadi and Gosse44,Reference Courtenay, Rowbotham, Lim, Peters, Yates and Chater45}

Comprehensive Unit-based Safety Program (CUSP)

CUSP focuses broadly on the idea of safety culture by empowering HCP to take responsibility for safety in their area, rather than defining specific domains of impact. As described by Pronovost et al ^{Reference Pronovost, Weast and Rosenstein46} at Johns Hopkins, who developed and validated the CUSP model in intensive care settings in 2005, the program is composed of 8 steps:

1. 1. Culture of safety assessment

2. 2. Sciences of safety education

3. 3. Staff identification of safety concerns

4. 4. Senior executive adoption of a unit

5. 5. Improvements implemented from safety concerns

6. 6. Documentation and analysis of efforts

7. 7. Sharing of results

8. 8. Culture reassessment.

The US Agency for Healthcare Research & Quality (AHRQ), the federal agency that provided funding for the development of CUSP, maintains an updated version of this framework on its website. ^{Reference Pronovost, Needham and Berenholtz33,47} The AHRQ also funded “On the CUSP: Stop CAUTI,” a national program to reduce the incidence of CAUTI through technical and hospital culture adaptations rooted in the CUSP model. ⁴⁸ This program focused on culture change. ^{Reference Meddings, Chopra and Saint49} CUSP has also been used in the ambulatory setting, specifically in the AHRQ “Safety Program for Improving Antibiotic Use,” a program derived from CUSP model concepts, designed to reduce overprescribing of antibiotics in primary care. ^{Reference Keller, Caballero and Tamma50} The AHRQ has further extended CUSP into ambulatory surgery, providing a full toolkit for the prevention of surgical-site infection on their website. ⁵¹ Investigators and implementation scientists have continued to use the CUSP approach in a variety of clinical settings with mixed results. Although CUSP has shown success in preventing HAIs, such as CLABSI in US intensive care units ^{Reference Berenholtz, Lubomski and Weeks32,Reference Miller, Briody and Casey52} and CAUTI on medical-surgical floors in acute-care hospitals ^{Reference Saint, Greene and Krein53} and in nursing homes, ^{Reference Mody, Greene and Meddings54} not all interventions using a CUSP-based approach have been successful. ⁵⁵

European and Mixed Methods

The European and Mixed-Methods framework derives from the CFIR ²⁷ and originated as the ‘InDepth’ work package, ^{Reference Sax, Clack and Touveneau56} a longitudinal qualitative comparative case study within the Prevention of Hospital Infections by Intervention and Training (PROHIBIT) study. ^{Reference van der Kooi, Sax and Pittet57,Reference van der Kooi, Sax and Grundmann58} Specifically, InDepth sought to identify the role contextual factors play in barriers and facilitators to successful implementation. ^{Reference Clack, Zingg and Saint59} The framework defines 3 qualitative measures of implementation success:

1. 1. Acceptability: Satisfaction with the intervention.

2. 2. Intervention fidelity: Local implementation matched with the stated goals of the multicenter trial.

3. 3. Adaptation: Local efforts to match the intervention with local context.

The framework has not been applied beyond the PROHIBIT outcomes of CLABSI rates and hand hygiene adherence, ^{Reference van der Kooi, Sax and Pittet57,Reference van der Kooi, Sax and Grundmann58} but the reported results may be used to inform other approaches.

Getting to Outcomes (GTO)®

Getting to Outcomes (GTO)® is a means of planning, implementing, and evaluating programs and initiatives developed for community settings. GTO® seeks to build capacity for self-efficacy, attitudes, and behaviors to yield effective prevention practices. ^{Reference Chinman, Hunter and Ebener60} The process involves 10 steps:

• Steps 1–5: Assess and evaluate needs, goals, and feasibility of a proposed program.

• Step 6: Plan and deliver the program.

• Steps 7–10: Evaluate, improve, and sustain successes.

GTO® has been utilized for numerous community-based initiatives, such as evidence-based sexual health promotion, ^{Reference Chinman, Acosta, Ebener, Malone and Slaughter61} a dual-disorder treatment program for veterans, ^{Reference Chinman, McCarthy, Hannah, Byrne and Smelson62} and development of casework models for child welfare services. ^{Reference Barbee, Antle, Wandersman and Cahn63} Additionally, the RAND Corporation has published a guide to develop community emergency preparedness programs. This guide breaks down each of the 10 GTO® steps, provides materials and examples, ^{Reference Ebener, Hunter, Adams, Eisenman, Acosta and Chinman64} and may facilitate the use of GTO® in implementing infection prevention interventions.

Model for Improvement

The Model for Improvement ^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost65} was developed by the Associates for Process Improvement based on Deming’s System of Profound Knowledge. ^{Reference Deming66} It has since been adopted widely, perhaps most notably by the Institute for Healthcare Improvement (IHI) in its 100,000 and 5 Million Lives campaigns of the early 2000s. ⁶⁷ The model has been used to accelerate change in a variety of healthcare and public health settings, ^{Reference Lannon and Peterson68–Reference Harrison, Shook, Harris, Lea, Cornett and Randolph70} and subspecialists have created primers focused on their practice areas. ^{Reference Guo, Fortin, Mayo, Robinson and Lo71–Reference Gaudreault-Tremblay, McQuillan, Parekh and Noone74} The Model for Improvement begins with 3 questions:

1. 1. What are we trying to accomplish?

2. 2. How will we know that a change is an improvement?

3. 3. What change can we make that will result in improvement?

Once those questions are answered, the identified changes or interventions are tested using plan–do–study–act (PDSA) cycles. Individual tests include the following:

• Plan: Predictions of outcome.

• Do: Executed according to plan.

• Study: Analysis and evaluation.

• Act: Decision whether to keep, abandon, or modify the intervention.

PDSA findings and decisions then guide planning the next experiment, starting a new PDSA cycle. Multiple cycles are done in series called ‘ramps.’ The Model for Improvement is designed for team-driven projects, relies heavily on data analysis and interpretation, and requires training (much of which can be self-directed online).

Models for underperforming hospitals and units

Allthough national implementation studies have succeeded in preventing several different HAIs, investigators have not seen the same success when focused on facilities most in need of help—hospitals underperforming with respect to HAI prevention. The CDC-funded national prospective, interventional, quality improvement program, CDC STRIVE (States Targeting Reduction in Infections Via Engagement), focused on reducing CLABSIs, CAUTIs, Clostridioides difficile infections, and methicillin-resistant Staphylococcus aureus (MRSA) bloodstream infections in hospitals with a disproportionately high burden of HAIs. ^{Reference Popovich, Calfee and Patel98} Although nearly 400 US hospitals participated in this multimodal, multifaceted, partner-facilitated program, they did not see significantly reduced rates of CLABSI, ^{Reference Patel, Greene and Jones99} CAUTI, ^{Reference Meddings, Manojlovich and Ameling100} C difficile infection, ^{Reference Dubberke, Rohde and Saint101} or MRSA bloodstream infection. ^{Reference Calfee, Davila and Chopra102} More recently, the Agency for Healthcare Research and Quality (AHRQ) funded a national program that invited US hospitals that had at least 1 adult ICU with elevated CLABSI or CAUTI rates to participate in an externally facilitated program implemented by a national project team and state hospital associations using the Comprehensive Unit-based Safety Program (CUSP) framework. ^{Reference Meddings, Greene and Ratz103} Results from the first 2 cohorts (366 recruited ICUs from 220 hospitals in 16 states and Puerto Rico) revealed no statistically significant reductions in CLABSI, CAUTI, or catheter utilization in the 280 ICUs that completed the program. ^{Reference Meddings, Greene and Ratz103} These researchers cite a number of possible factors contributing to the disappointing result, including underutilization of training and coaching resources, lack of infrastructure, and a different selection process for participants (eg, identifying low-performing units or those that had been unsuccessful to date versus asking for volunteers for earlier CUSP work, which may have selected for early adopters and high performers). Investigations in this and similar cohorts could help elucidate why hospitals with disproportionately high HAI rates have not yet seen significant reductions in HAIs despite broad-based efforts. Increased focus on the development, adaptation, and utilization of implementation models and frameworks to infection prevention and control may help identify implementation gaps that contribute to lack of improvement and guide their closure.

Sustaining system change

A long-term goal of any implementation effort is to sustain and advance short-term gains. Ideally, sustaining gains occurs with less intensity than initial efforts, maintaining gains or improving at a slower rate and allowing resources to be directed to another effort. Characteristics of successfully sustained interventions have included those that are incorporated into the standard workflow, have effective champions to shepherd the effort and re-engage when necessary, can be modified over time, fit with an organization’s mission and procedures, provide easily perceived benefits to staff members and/or clients, and are supported by partner organizations. ^{Reference Scheirer104} It can be difficult to meet those criteria in healthcare, where changes in workflows and staff are frequent. ^{Reference Fowler, Krein, Ratz, Zawol and Saint105} Demonstrating successfully sustained implementation should include evidence of (1) sustainment, that is, sustained use of an evidence-based intervention (process measure), and (2) sustainability, that is, sustained benefits of an evidence-based intervention (outcome measure). ^{Reference Moullin, Sklar and Green106}

Linking ongoing process data to ongoing outcome data can prove challenging. In one study, CLABSI reduction in ICUs was sustained for a decade, but process measurement was not performed. ^{Reference Pronovost, Watson, Goeschel, Hyzy and Berenholtz34} A study on CAUTI prevention at a Veterans’ Affairs (VA) hospital found that, 8 years after implementation, appropriateness of urinary catheters remained high and stable. ^{Reference Fowler, Krein, Ratz, Zawol and Saint105} Catheter use decreased, but the facility was unable to report outcome data. ^{Reference Fowler, Krein, Ratz, Zawol and Saint105} These researchers hypothesized that success was due to a 3-component, evidence-based intervention, institutionalization of the interventions (ie, standardizing nursing assessments and handoffs that included the intervention), and effective champions who re-engaged when necessary. Additionally, a study on hand hygiene on 2 hospital units in Italy found that adherence dropped after 4 years (from 84.2% to 71%) despite maintaining champions and processes. ^{Reference Lieber, Mantengoli and Saint107} Recent proposals for standard definitions ^{Reference Moullin, Sklar and Green106,Reference Moore, Mascarenhas, Bain and Straus108} and modified ERIC strategies to account for sustainment and identify interventions that yield short-term and sustained improvements ^{Reference Nathan, Powell and Shelton109} can form a basis for future research and understanding.