A Scoping Review of Implementation Science in Adult Critical Care Settings

Supplemental Digital Content is available in the text.

Abstract

Objectives:

The purpose of this scoping review is to provide a synthesis of the available literature on implementation science in critical care settings. Specifically, we aimed to identify the evidence-based practices selected for implementation, the frequency and type of implementation strategies used to foster change, and the process and clinical outcomes associated with implementation.

Data Sources:

A librarian-assisted search was performed using three electronic databases.

Study Selection:

Articles that reported outcomes aimed at disseminating, implementing, or sustaining an evidence-based intervention or practice, used established implementation strategies, and were conducted in a critical care unit were included.

Data Extraction:

Two reviewers independently screened titles, abstracts, and full text of articles to determine eligibility. Data extraction was performed using customized fields established a priori within a systematic review software system.

Data Synthesis:

Of 1,707 citations, 82 met eligibility criteria. Studies included prospective research investigations, quality improvement projects, and implementation science trials. The most common practices investigated were use of a ventilator-associated pneumonia bundle, nutritional support protocols, and the Awakening and Breathing Coordination, Delirium Monitoring/Management, and Early Exercise/Mobility bundle. A variety of implementation strategies were used to facilitate evidence adoption, most commonly educational meetings, auditing and feedback, developing tools, and use of local opinion leaders. The majority of studies (76/82, 93%) reported using more than one implementation strategy. Few studies specifically used implementation science designs and frameworks to systematically evaluate both implementation and clinical outcomes.

Conclusions:

The field of critical care has experienced slow but steady gains in the number of investigations specifically guided by implementation science. However, given the exponential growth of evidence-based practices and guidelines in this same period, much work remains to critically evaluate the most effective mechanisms to integrate and sustain these practices across diverse critical care settings and teams.

Implementation science (IS) is a field of study that seeks to identify optimal methods for accelerating the systematic uptake of research findings and other evidence-based practices (EBPs) into routine clinical care (1). An evolving area of inquiry and growing priority for health-related funding agencies, IS applies theories borrowed from other disciplines (e.g., psychology, sociology, organizational change) as well as field-specific conceptual models and frameworks. These theories are generally used to: 1) describe the process of translating research into practice (process models), 2) understand what contextual factors serve as barriers and facilitators to implementation (determinant frameworks), 3) provide a structure for evaluating implementation endeavors and outcomes (evaluation frameworks), and 4) describe how change can occur in areas where intention to change is absent (classic theories) (2, 3). The number of theories, models, and frameworks are prolific, and integration of these into complex health systems often requires additional considerations regarding change mechanisms, meditators, moderators, and both proximal and distal outcomes (3, 4). Identification of an applicable model for implementation is critical in complex healthcare systems where sustained application of EBPs can improve quality and safety of care, limit harmful practice variability, and reduce soaring costs. Model-based approaches to evidence implementation and practice change can also be considerably effective in high priority, time sensitive initiatives, such as those recently experienced in the setting of the coronavirus disease 2019 pandemic.

IS overlaps with several other fields, including quality improvement (QI), dissemination science, and knowledge translation. Approaches are similar in terms of populations of interest, metrics to evaluate process and clinical outcomes, incorporation of existing clinical data, emphasis on stakeholder involvement, and goal of bridging the research to practice gap to improve health (1). The main distinction between fields lies in their focal intent. IS aims to produce “generalizable knowledge” and evidence about effective strategies and outcomes associated with sustained integration of established but underutilized EBPs into routine practice (1). In contrast, QI mainly aims to streamline processes and eliminate inefficient practices locally, commonly via rapid-cycle improvement. In QI, the change initiative may be linked to a best practice or scientific evidence, but this is not a requirement, as many initiatives focus solely on process improvement. The focus of QI is generally narrower than dissemination and IS. Dissemination science has a broader focus on studying the targeted distribution of information and intervention materials to a specific public health or clinical practice audience and spreading information using communication and education strategies (5). Finally, knowledge translation refers to a dynamic and iterative process of synthesis, dissemination, exchange, and application of knowledge. Knowledge translation promotes information not only from research to practice but also from practice back to research to identify gaps and guide future research priorities (1).

A key development in IS is a common nomenclature for implementation strategy terms, definitions, and categories that can be used to guide implementation research and practice across settings (6). This work evolves from leaders in IS (7) who advocate that knowledge of EBPs must be accompanied by knowledge of implementation in order to successfully integrate evidence-based interventions. Implementation strategies are organized into categories to provide guidance on approaches to successfully implement evidence-based interventions. For example, Flodgren et al (8) used select systematic reviews produced by the Cochrane Effective Practice and Organisation of Care group to identify, define and provide evidence for seven general strategies (i.e., printed educational materials, educational meetings, educational outreach, local opinion leaders, audit and feedback, computerized reminders, and tailored interventions). A separate taxonomy includes four domains (professional, financial, organizational, and regulatory) and 49 distinct strategies (9). Additionally, Powell et al (10) established a compilation of 73 implementation strategies, called the Expert Recommendations for Implementing Change (ERIC) strategies. Ultimately, local contextual need and assessment of facilitators and barriers should be primary drivers influencing selection of implementation strategies (8).

While there is evidence on the effectiveness of certain implementation strategies to increase EBP utilization in a variety of settings, the feasibility of these strategies in critical care units is unknown. In addition to the complexity of interventions, the very nature of the critical care environment poses unique considerations and contexts for EBP integration. A growing body of literature demonstrates a large proportion of critically ill patients do not receive evidence-based or guideline recommended care (11). This failure to apply EBPs during critical illness often leads to significant short- and long-term morbidity and mortality (11). Establishing the current state of knowledge regarding common strategies and IS efforts in critical care settings is an important step in addressing how to best sustain delivery of evidence-based interventions.

The purpose of this scoping review is to provide a synthesis of the available literature related to IS in critical care in terms of the volume, nature, and characteristics of studies conducted to identify existing knowledge gaps. We were specifically interested in identifying which EBPs were targeted for improvement, which implementation strategies were used to foster change, and which outcomes were evaluated.

MATERIALS AND METHODS

This scoping review was conducted by a research team with expertise in IS, critical care, EBP, and systematic review methodology. The review adhered to the process outlined by Grant and Booth (12) and the checklist for Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (13). The Covidence systematic review software was used for all stages of the review process, including title and abstract screening, full text review, and data extraction.

Research Question

The research question guiding the scoping review was, “What are the implementation strategies, associated EBPs, and outcomes evaluated in studies conducted in critical care settings?”

Search Strategy

A librarian-assisted search was performed on October 30, 2019, using three electronic databases: Medline, PubMed, and Cumulative Index to Nursing and Allied Health Literature. The databases were selected to be comprehensive and cover a range of disciplines. The following key words were used: IS, implementation, critical care, ICU, and EBP (Appendix A, http://links.lww.com/CCX/A455). The search was limited to articles published between January 1999 and July 2019 to coincide with the EBP movement and development of the relatively new field of IS. Reference lists of included articles were reviewed and hand searching was performed to identify other relevant literature not captured by the electronic search.

Inclusion and Exclusion Criteria

Articles were included if they: 1) reported outcomes of a research study, QI effort, or program aimed at disseminating, implementing, or sustaining an EBP, 2) described use of any of the 73 implementation strategies outlined in the ERIC project (8), 3) were conducted in a critical care unit, and 4) involved adults. Because of limited resources for translation, articles published in languages other than English were excluded. We also excluded reviews, commentaries, editorials, abstracts, and conference proceedings on the basis that these would not provide the level of detail sought in our review. Finally, literature that included work conducted outside the critical care setting or involving children was excluded, as evidence-based interventions/practices would likely differ by level of care and age.

Data Abstraction

Two reviewers (M.M., M.C.B.) independently screened titles and abstracts of identified articles to determine eligibility. The same two reviewers then performed full text review in duplicate, with conflicts resolved by an independent third reviewer (D.O.). Reviewers customized data extraction fields in Covidence to align with aims of the review. Two reviewers tested extraction fields for consistency and fidelity to project aims (M.M., H.R.). The same two reviewers independently extracted data from articles into required fields. Extracted data were compared between reviewers for consensus prior to finalizing the extraction forms. The following data were extracted from included articles: year and country of publication, project aim, design, theory or framework used to guide the project, setting, EBP implemented, components of the EBP intervention, implementation strategies, and clinical/implementation outcomes.

RESULTS

The combination of search terms with selection criteria and limits yielded 1,707 studies. Of these, 40 duplicates were removed, leaving 1,667 studies for title and abstract screening (Fig. 1). Of these, 1,426 were excluded for not meeting initial inclusion criteria (i.e., title or abstract indicated nonadult population, noncritical care setting, or literature review), resulting in 241 articles that underwent full text review, with a moderate level of agreement (ĸ = 0.450), of which an additional 159 were excluded for the following reasons: literature review (n = 69), non-ICU setting (n = 41), noneligible study design (n = 31), did not evaluate an implementation strategy (n = 16), or not in English (n = 2). A total of 82 studies were included in the final review.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Characteristics of Included Studies

Of the studies that met inclusion criteria for this review, some (n = 20) were specifically described by the authors as IS studies and a similar number (n = 19) designated as QI projects. The remaining studies (n = 43) were identified neither as IS or QI by their authors but rather described a variety of prospective implementation (PI) evaluations or designs, such as prospective cohort, time series, or pre/post evaluation study designs. Such studies will be referred to here as having PI designs. Figure 2 displays these design categories according to year of publication. The PI designs were the most common type in every time period, with the largest number occurring during the period 2005–2009. QI designs were used less frequently in the earlier time periods, with substantial increases in recent years. Similarly, there were few IS designs for the initial time periods, but reports have doubled within the last 5 years. None of the studies included in this review examined systematic de-implementation of low value practices.

Figure 2.

Year and type of publication.

Supplemental Table 1 (http://links.lww.com/CCX/A456) provides an overview of the studies included in the review according to the design categories. Studies using IS designs were conducted most often in the United States (n = 9) (14–22) and Canada (n = 7) (23–29) followed by Australia (n = 2) (30, 31), the Netherlands (n = 1) (32), and the United Kingdom (n = 1) (33). Within the PI category, studies were conducted most frequently in the United States (n = 25) (34–59) followed by Canada (n = 3) (60–62), Spain (n = 3) (63–65), Germany (n = 2) (66, 67), Norway (n = 2) (68, 69), and one study each from South Africa (70), Argentina (71), China (72), France (73), The Netherlands (74), Saudi Arabia (75), and the United Kingdom (56). Studies that reported using a QI design were conducted in the United States (n = 11) (76–86), Australia (n = 2) (87, 88), and one each in Brazil (89), Canada (90), China (91), Sweden (92), Taiwan (93), and the United Kingdom (94). Many studies (30/82, 36.5%) included more than one critical care unit in their investigation.

EBPs Implemented

Figure 3 displays the EBPs implemented in the 82 studies. The studies addressed a variety of EBPs, classified into 18 categories. The most common EBPs included: ventilator-associated pneumonia (VAP) prevention bundles (12/82, 14.6%) (14, 33, 39, 49, 54, 55, 57, 59, 60, 63, 73, 86), nutritional support/management protocols (9/82, 10.9%) (25, 26, 30, 35, 38, 40, 61, 88, 90), the Awakening and Breathing Coordination, Delirium Monitoring/Management, and Early Exercise/Mobility (ABCDE/ABCDEF) bundle (7/82, 8.5%) (15–19, 47, 76), central line-associated bloodstream infection (CLABSI) prevention bundles (6/82, 7.3%) (22, 37, 42, 52, 53, 80), and mechanical ventilation liberation/weaning protocols (6/82, 7.3%) (36, 62, 69, 72, 90, 93). Sixteen studies involved ICU mobilization protocols (4/82, 4.8%) (34, 46, 70, 78), use of checklists (4/82, 4.8%) (41, 81, 84, 89), hand hygiene/infection control bundles (n = 4/82, 4.8%) (21, 66, 71, 95), and initiatives targeting multiple EBPs (4/82, 4.8%) (24, 51, 56, 77). The remaining studies focused on antibiotic stewardship (3/82, 3.6%) (22, 65, 67), analgesia/sedation protocols (3/82, 3.6%) (68, 75, 96), palliative/end-of-life care bundles (3/82, 3.6%) (20, 32, 85), pressure ulcer prevention (3/82, 3.6%) (43, 83, 94), venous thromboembolism (VTE) prophylaxis (2/82, 2.4%) (28, 29), laboratory reduction guidelines (2/82, 2.4%) (48, 87), the surviving sepsis campaign bundles (2/82, 2.4%) (64, 82), interdisciplinary rounding/handover tools (2/82, 2.4%) (31, 58), and “other” EBPs (7/82, 8.5%) (27, 44, 45, 50, 74, 79, 92).

Figure 3.

Evidence-based practices evaluated. ABCDE/ABCDEF = Awakening and Breathing Coordination, Delirium Monitoring/Management, and Early Exercise/Mobility, CLABSI = central line-associated bloodstream infection, CR-BSI = catheter-related bloodstream infection, EBPS = evidence-based practices, MV = mechanical ventilation, VAP = ventilator-associated pneumonia, VTE = venous thromboembolism.

Temporal trends were noted by EBP category. For example, studies involving nutritional support and management, VAP, and CLABSI prevention practices were more commonly published between 2004 and 2011, while those focusing on palliative/end-of-life care, the ABCDE/ABCDEF bundle, and interdisciplinary rounding were more recent. Similarly, the complexity and number of components included in the EBP change initiatives appeared to change over time. For example, early VAP prevention bundles targeted the use of head of bed elevation, oral care, ventilator tubing condensate removal, and hand hygiene/glove use, while more recent bundles added sedation and mechanical ventilation liberation procedures and protocols. Similarly, more recent studies generally included EBPs involving members of an interdisciplinary team working together to deliver the EBP, rather than a single discipline.

Implementation Strategies

Figure 4 displays implementation strategies used in the included studies. Sixteen different ERIC strategies were described and include: educational meetings, audit and feedback, develop tools, local opinion leaders, develop effective materials, inform stakeholders, ongoing consultation, distribute materials, ongoing training, develop systems, facilitators and barriers, scale up, academic partnerships, financial incentives, and building a coalition. Of these strategies, educational meetings (56/82, 68.2%), auditing and feedback (55/82, 67.1%), developing tools (26/82, 31.7%), and use of local opinion leaders (18/82, 21.9%) were used most frequently. Most studies (76/82, 92.6%) reported using more than one strategy when implementing an EBP. Twelve studies reported use of a single strategy for EBP implementation. Single strategy approaches included auditing and feedback (23, 26, 27, 65, 66), facilitators and barriers (15, 16, 18, 32), financial incentives (36), building a coalition (19), and educational training (72). Across all types of designs, none specifically explored the role of mediators for their effect on implementation or clinical outcomes or as causal components in the relationship between implementation strategies and outcomes.

Figure 4.

Implementation strategies.

Outcomes Evaluated

Studies reported both clinical and implementation outcomes (Supplemental Table 2, http://links.lww.com/CCX/A457). Across all study designs, many reported only clinical outcomes (44/82, 53.6%) (22, 34, 36–38, 40, 41, 43–50, 52, 54–56, 60–66, 68, 70, 72, 75, 77–81, 83, 85, 87, 88, 90–94), while few reported solely implementation outcomes (15/82, 18.3%) (15–20, 24, 27, 28, 32, 51, 57, 67, 95, 96), and some reported both clinical and implementation outcomes (22/82, 26.8%) (14, 21, 23, 25, 26, 29, 30, 32, 39, 42, 53, 58, 59, 62, 71, 73, 74, 76, 82, 84, 86, 89). Primary clinical outcomes were most often linked to the EBP that was evaluated (e.g., VAP bundle implementation reported VAP rates), rather than solely surrogate metrics such as mortality or length of stay. The most frequently reported primary clinical outcomes included VAP rates (14, 33, 39, 54, 55, 59, 60, 62, 63, 75, 86), ventilator days/duration of mechanical ventilation (33, 34, 38, 46, 47, 62, 68, 72, 75, 78, 91), catheter-related bloodstream infection/CLABSI (21, 22, 37, 42, 52, 53, 81, 83), nutrition adequacy (25, 26, 40, 61), or time/duration of nutrition therapy (35, 38, 44, 88, 90). Primary clinical outcomes were reported as rates or continuous values, rather than through use of standardized outcome scales. Implementation outcomes are those identified by Proctor et al (97) and include acceptability, adoption, appropriateness, feasibility, fidelity, cost, penetration, and sustainability. The most frequently reported implementation outcome was fidelity, often reported as either percentage of adherence or compliance to the EBP that was implemented, rather than through use of standardized tools or scales (14, 18–26, 29, 30, 32, 33, 39, 42, 51, 53, 57–59, 62, 67, 71, 73, 74, 76, 82, 84, 86, 89, 95, 96). Other implementation outcomes included adoption (14, 22, 23, 30, 31), or qualitative evaluations of facilitators/barriers and perceptions of the EBP, and descriptions of team composition/building a coalition (15–17, 19, 26–28, 31). For both clinical and implementation outcomes, reporting of the various metrics across all study types was substantially heterogeneous by measurement type, duration, and frequency, rendering pooling of data difficult. However, the majority of studies reported positive improvements in both clinical and/or implementation outcomes, regardless of study design.

Specific IS Investigations

Given our specific interest in IS in critical care units, we identified a subset of articles (n = 20) as having a specific IS design, including an established IS framework, and evaluating effectiveness of implementation strategies specifically on implementation outcomes (97). The frameworks, aims, strategies, and outcomes of these studies are listed in Table 1. Various models or frameworks guided the implementation studies, with the Knowledge to Action Framework used most often (25, 26, 31), followed by The Grol Model of Implementation (21, 32). Other implementation models are listed in Table 1. Many of the studies (n = 11) included multiple ICUs. Specific EBPs evaluated in these studies included VAP prevention bundles (14, 33), ABCDEF bundle (15–19), nutrition therapy (25–27, 30), antibiotic use (23), communication practices (31), palliative care consultation (20), CLABSI prevention (21, 22), VTE prophylaxis (28, 29), and multiple EBPs (24). One additional study specifically evaluated readiness for and barriers to implementation of EBPs (32).

TABLE 1.

Summary of Implementation Science Specific Designs and Studies (n = 20)

References	Country	Framework	Study Aim	Strategies	Study Outcome
Abbott et al (14)	United States	Academic Center for Evidence-Based Practice star model; Predisposing, reinforcing, enabling constructs in educational diagnosis and evaluation/policy, regulatory, and organizational constructs in educational and environmental development	Determine effect of VAP prevention bundle on VAP rates	Stakeholders, audit and feedback, education, develop systems	VAP frequency: Variable across three ICUs; initial decrease below benchmark but not sustained Adoption: head of bed 77–69%; oral care 22–30%; empty condensate 94–93%; gloves 74–90%
Balas et al (15)	United States	Consolidated framework for implementation research	Identify facilitators and barriers to awakening and breathing coordination, delirium monitoring/management, and early exercise mobility bundle adoption; evaluate if bundle implementation was effective, sustainable, and conducive to dissemination	Assess readiness and barriers	Facilitators: Evidence, leadership Barriers: Adaptability, complexity, workload, staff, communication, documentation
Boehm et al (17)	United States	Outcome production model	Understand the relationship between organizational domains and provider attitudes towards implementation of the ABCDEF bundle	Develop tools, audit and feedback, local opinion leader, education	Adherence: Increased with use of tools (protocols) (r = 0.37–0.58), role clarity (r = 0.38–0.59), training (r = 0.33–0.46), local opinion leaders (r = 0.37–0.48), teamwork (r = 0.36–0.44)
Boltey et al (16)	United States	Shared mental model	Examine how components of shared mental model impact implementation of ABCDEF bundle	Assess readiness and barriers	Facilitators: Awareness of shared mental model can impact routine implementation Adherence via self-report (84% adherence sometimes; 51% routine adherence)
Carrothers et al (18)	United States	Not stated; survey	Identify which contextual factors facilitate/hinder implementation of ABCDE bundle in four San Francisco Bay ICUs	Assess readiness and barriers	Facilitators: Leadership, culture, champion, training, and support Barriers: Resources, turnover, knowledge, staffing
Costa et al (19)	United States	Not stated; survey	The purpose of this study was to describe team composition in ABCDE delivery and test the hypothesis that frequent involvement of a diverse team was associated with high levels of ABCDE implementation	Implementation teams	Adherence: Higher odds of adherence to spontaneous awakening trials (OR, 4.2), delirium management (OR, 3.6), and mobility (OR, 2.3) when team approach utilized
Doig et al (30)	Australia and New Zealand	Browman’s clinical practice guideline development cycle	Determine whether evidence-based feeding guidelines could be implemented using a multifaceted practice chance strategy to improve feeding and mortality among ICU patients	Local opinion leader education, reminders audit and feedback, ongoing training	Mortality: Similar between guideline vs control groups (28.9% vs 27.4%, respectively) Days to initiate feed: 0.75 d for guideline group vs 1.37 d for control group Adherence: 94% for guideline group; 72% for control group
Elligsen et al (23)	Canada	Not stated	Evaluate the impact of prospective audit and feedback on broad spectrum antimicrobial use among critical care patients	Audit and feedback	Antibiotic use: Decreased from 644 to 503 d of therapy per 1,000 patient-days
Hawe et al (33)	United Kingdom	Not stated	Describe the effects of an active multifaceted implementation of a VAP prevention bundle designed to improve staff compliance with evidence-based actions and reduce the frequency of VAP	Education, written materials, auditing and feedback; passive (phase I) vs active implementation (phase II)	Adherence: Increased from 0% to 54%VAP frequency: Decreased from 19.2 to 7.5 per 1,000 ventilator days Ventilator days: Decreased from 2,556 to 1,327
Ilan et al (24)	Canada	Not stated	Describe prescription rates of commonly recommended best practices for critically ill patients and determine factors associated with increased rates of prescription	Standardized order sets, specialty consultation	Adherence: VTE prophylaxis: 95.3% Antibiotic prophylaxis: 94.1% Stress ulcer prophylaxis: 89.7% Enteral nutrition: 72.4% Insulin infusion: 58.8% Low tidal ventilation: 53.8% Perioperative beta blockers: 40% Steroids for shock: 20% Specialty mattress: 17.6% Interruption of sedation: 8.3%
Jain et al (25)	Canada	Knowledge to action framework	Compare the effectiveness of active to passive dissemination of the Canadian clinical practice guidelines for nutrition support for the mechanically ventilated critically ill adult patient	Local opinion leader, education, audit and feedback, develop tools, tailor strategies, education, distribute materials, develop effective materials, ongoing consultation	Enteral nutrition adequacy: Increased from 42% to 50% Enteral nutrition initiation: Increased from 52% to 58%
Noome et al (32)	The Netherlands	Grol and Grimshaw model for implementation	Examine the effectiveness of supporting ICUs on implementing the guidelines	Assess readiness and barriers	Adherence: 0.71 mean scores for control; 0.72 mean scores for intervention group
Penrod et al (20)	United States	Provonost model for knowledge translation	Evaluate implementation of care and communication bundle for palliative care	Audit and feedback, develop tools, education, ongoing consultation	Bundle adherence: Increased from a range of 13–40% to 20–60%
Reynolds et al (21)	United States	Grol and Wensing model of implementation	Determine whether using tailored, multifaceted strategies would improve implementation of daily chlorhexidine bathing and decrease CLABSIs	Educational outreach, audit and feedback, local opinion leaders, printed educational materials	Compliance: Increased from 57% to 80% CLABSI rates: Decreased from 2.8 to 1.2 per 1,000 central line days
Sauro et al (28)	United States	TRIP model	Describe use of IS at the unit level and organizational level to guide an intervention to reduce CLABSI in BICU	Inform/engage stakeholder, develop systems, audit and feedback, develop tools	CLABSI rates: Decreased from 15.5 to 0 per 1,000 central line days
Sinuff et al (27)	Canada	Qualitative/IS	Identify clinician perspectives of auditing and feedback	Audit and feedback	Perceptions of audit and feedback: Poor transparency, feedback should be timely, communication should be continuous, encourage peer to peer discussion/leadership engagement
Sinuff et al (26)	Canada	Knowledge to action framework	Determine whether auditing practice and providing feedback in the form of benchmarked reports site reports is an effective strategy to improve adherence to nutrition guidelines	Audit and feedback, develop tools, education, develop systems, develop effective materials	Adherence: Increased from 71% to 81%
Sood et al (22)	United States	TRIP model	Describe use of IS at the unit level and organizational level to guide an intervention to reduce CLABSI in BICU	Inform/engage stakeholder, develop systems, audit and feedback, develop tools	CLABSI rates decreased from 15.5 to 0 per 1,000 central line days
Spooner et al (31)	Australia	Know to action framework	Implement and evaluate an evidence-based electronic minimum data set for nursing team leader shift to shift handover in the ICU using the knowledge to action framework	Assess readiness and barriers, tailor strategies, education, local opinion leaders, develop tools, develop effective materials, audit and feedback	Adherence: 78%
Stelfox et al (29)	Canada	Theoretical domains framework	Test whether a multicomponent intervention would increase use of low-molecular-weight heparin over unfractionated heparin for VTE prophylaxis in critically ill patients	Education, develop tools, reminders, audit and feedback	Adherence: Increased for intervention group: 45.9–78.3%; increased for control group: 37.9–53.3% VTE: Remained same for intervention group: 3–3%; decreased slightly for control group: 2.4–2.1% Deep vein thrombosis: Increased for intervention group: 1.9–2.1%; remained same for control group: 1.4–1.4% Pulmonary embolism: Decreased for intervention group: 1.3–1.1%; decreased for control group: 1.2–0.8%

ABCDEF = Awakening and Breathing Coordination, Delirium Monitoring/Management, and Early Exercise/Mobility, BICU = burn ICU, CLABSI = central line-associated bloodstream infection, IS = implementation science, OR = odds ratio, TRIP = translating research into practice, VAP = ventilator-associated pneumonia, VTE = venous thromboembolism.

A variety of implementation strategies were evaluated within this subset of IS specific designs (Table 1). The majority of studies (n = 13) evaluated more than one IS strategy, with auditing and feedback incorporated most often across studies (n = 13) (14, 17, 20–22, 25–27, 29–31, 33). Eight studies evaluated a single implementation strategy (15, 16, 18, 19, 23, 26, 27, 32), namely identifying facilitators and barriers (15, 16, 18, 32), building a coalition (19), and auditing and feedback (22, 26, 27).

Outcomes evaluated across the IS studies included implementation and clinical outcomes as described by Proctor et al (97). The most frequent implementation outcome evaluated was fidelity (reported as adherence) to the EBP (17, 19–21, 23, 24, 26, 29–33). Only one study included adoption as an implementation outcome (14). Additionally, a subset of studies specifically evaluated facilitators and barriers to implementation (15, 16), effective implementation strategies (28), or perceptions of auditing and feedback as a strategy using a qualitative approach (27). In addition to implementation outcomes, five of the studies also evaluated clinical outcomes; these included VAP rates and/or ventilator days (14, 33), mortality (30), and CLABSI rates (21, 22).

DISCUSSION

This is one of the first reviews to synthesize the available literature related to IS in critical care. We found significant variability in project nomenclature and designs aimed at evaluating strategies to implement an EBP and determine effect on outcomes. Despite these variations, findings across all project designs did demonstrate positive effects on clinical outcomes, highlighting the importance of EBPs in critical care. Within the dates of our review, the number of published studies in the critical care literature that reported being guided by IS has increased steadily, doubling in the last 5 years. The complexity and number of components included in the EBPs selected for improvement has similarly increased over time. Most of the EBPs studied required the input, decision-making, and performance of tasks by multiple ICU team members, rather than relying on a single disciple for implementation. While a variety of methods were used to foster adoption of these EBPs into critical care, a limited number of ERIC strategies were used (16/73), and few projects evaluated implementation outcomes, making it difficult to definitely conclude which strategies were most effective.

An inherent challenge to advancing IS work in critical care settings is the vast heterogeneity of strategies and outcomes used across projects that precludes pooling and synthesis of findings. Many projects in this review evaluated specific IS strategies, yet used a variety of terms to describe the project, and did not use consistent mechanisms for measuring and evaluating the strategies. The field of IS has had major gains in this area by using concept mapping, developing operational definitions and categories for implementation strategies, and establishing measurement tools to provide a consistent and scientific approach to these investigations. Nevertheless, this review confirms mixed evidence as to which design approach is most effective, which specific strategies are superior, and whether multiple strategies versus a single strategy is effective for successful practice change (6). In addition, the extent to which mediating factors are present and influence effectiveness of strategies warrants consideration, given the complexity of critical care environments. Ultimately, a systematic approach is needed using standardized designs, strategies, and both clinical and implementation outcomes in critical care settings to increase widespread adoption of EBPs.

We also found significant variation in outcomes evaluated in our review. Across all project designs, many only reported clinical outcomes. Among the few that reported implementation outcomes, these were limited to fidelity (adherence and compliance) and adoption. While these are certainly important components of initial EBP implementation, other outcomes such as acceptability, appropriateness, penetration, and sustainability would yield critical information on characteristics of the EBP that would advance widespread integration over time and across settings. Integration of these additional outcomes using an IS approach can generate important data on effective mechanisms for EBP change that is sustainable in complex critical care environments. The field of IS allows multiple approaches to project evaluation, including hybrid designs that establish priorities for clinical and implementation outcomes, and mechanisms for systematically measuring and reporting findings. Many IS approaches also include a QI component to integrate rapid cycle change and streamline processes to improve both clinical and implementation outcomes under investigation. This systematic approach to implementation becomes critical when working to develop a body of generalizable evidence on the most effective strategies to promote EBP utilization and evaluate impact on clinicians, organizations, patients, and populations.

Among the specific IS designs in critical care, the most comprehensively investigated EBP was the ABCDE/F bundle. Although investigations were guided by different frameworks or models, all systematically contributed information on factors influencing successful uptake of the ABCDE/F bundle into routine clinical care. Various strategies were evaluated and formal investigations into assessment of readiness and barriers were performed. This body of knowledge regarding implementation of the bundle has resulted in development of toolkits to guide clinicians through the implementation process.

As a result of this review, there are several priority areas that should be addressed to move the science of implementation forward in critical care. First, clinicians and scientists are encouraged to use established models or frameworks for implementation of best practices to provide a systematic approach to implementation and increase likelihood of generalizability and sustainability over time. Second, selection, measurement, and reporting of implementation strategies should align with current nomenclature to provide consistency of methods across studies and build evidence regarding effectiveness. Third, outcome reporting should extend beyond solely clinical outcomes and include measures of implementation outcomes using established terminology and mechanisms. There is a need for additional investigations to explore the role of mediators on clinical and implementation outcomes, as well as systematic approaches to de-implementation of low value or wasteful practices in critical care settings. Dissemination of these conceptual and methodological efforts is critical to advance widespread integration of EBPs into routine clinical care.

CONCLUSIONS

Having a structured and systematic approach to integrating EBPs into practice using an IS approach holds great potential in critical care settings and should remain a key component of critical care research agendas for all EBPs. The work should not cease upon publication of study findings on clinical effectiveness or upon publication of an evidence-based guideline. Rather, the next logical step scientifically is to identify optimal strategies to embed the findings into routine clinical care. Evaluation and dissemination of the effectiveness of these strategies on both clinical and implementation outcomes then generates evidence to promote sustainable practice change. This is exactly the aim of IS and must be addressed by developing critical care teams who are experienced in IS methodology and committed to advancing the science specifically related to acceleration of adoption and uptake of evolving effective critical care interventions that optimize patient outcomes.