Use and effects of implementation strategies for practice guidelines in nursing: a systematic review

Five electronic databases (Medline, EMBASE, CINAHL, PsycINFO, AMED) and the Cochrane EPOC registry were systematically searched using a search strategy developed with nursing library scientists at Queen’s University. The team used search terms and medical subject headings (MeSH) relevant to “clinical guidelines” AND “implementation” AND “nursing” AND “randomized controlled trial” (Supplemental File 2). The search was run up to September 30th, 2020 with no restrictions. We also used the search strategies developed by Grimshaw and colleagues [22] for their investigation of the effectiveness and efficiency of guideline dissemination and implementation strategies in the context of medicine. These search strategies were adjusted to focus on nursing and rerun against MEDLINE, CINAHL, and EMBASE. We also scanned reference lists of papers identified for inclusion for any additional references not captured.

Eligibility criteria.

Inclusion criteria for this review were that studies had to (a) be written in English and published in a peer-reviewed journal, (b) use a randomized controlled trial (RCT) design, and (c) evaluate the implementation of a guideline targeted to nurses or a multidisciplinary team with a focus on nursing outcomes. Implementation strategies are defined as “methods or techniques used to enhance the adoption, implementation, and sustainability of a clinical program or practice” [23]. Guidelines are defined as “systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances” [17]. A more recent definition includes benefits and harms (but no longer includes the goal of assisting practitioner and patient decisions); “Clinical practice guidelines are statements that include recommendations intended to optimize patient care that are informed by a systematic review of evidence and an assessment of the benefits and harms of alternative care options” [5]. Substitute terms for practice guidelines included “protocol,” “standard,” “algorithm,” and “clinical pathway.” The primary comparator was “usual practice” or “usual care” which indicates no distinct implementation strategy was used to change nursing practice or a traditional approach to dissemination was used (e.g., guidelines were available for use in practice setting, nurses received a copy of the guidelines). Categories of nurses included advanced practice nurses (APN); nurse practitioners (NPs); clinical educators (CE); clinical nurse specialists (CNS); registered nurses (RN); and licensed practical nurses (LPN)/registered practical nurses (RPN).

We excluded studies that were (a) not a RCT design; (b) abstracts, conference proceedings; (c) targeting implementation strategies to patients, administrators, and other health care providers (when outcomes could not be attributed to nurses); and (d) not focused on implementing practice guidelines.

Outcomes of interest were primary outcomes focused on process or outcome of care provided by nursing professionals. Outcomes were grouped into five categories (professional knowledge, professional practice, patient health status outcomes, resource use, and expenditures). Professional knowledge outcomes related to level of nursing knowledge. Professional practice outcomes related to nursing process of care (i.e., adherence to practice guidelines). Patient health status outcomes included physical health and treatment outcomes (i.e., pain, quality of life, incontinence). Economic outcomes related to resource use and measured costs and cost savings (expenditures) associated with guideline implementation [24].

Two reviewers independently screened the titles and abstracts for inclusion in Covidence [25]. Next, two reviewers independently screened the full-text articles against the inclusion criteria. When there was disagreement, a third and independent reviewer assessed the study. Where separate papers reported on different aspects of the same study (for example, one paper describing the effect of the intervention on professional practice and another, the effect on patient health status outcomes), we treated them as one study with two companion papers.

Two reviewers independently abstracted data from included studies using a standardized form adapted from the EPOC data collection checklist. The following data were extracted from each eligible full-text study: (a) study design, (b) participants, (c) setting, (d) data collection methods, (e) practice guideline, (f) use of theory, (g) types of implementation strategies, (h) outcome measures, and (i) study results. Two reviewers independently piloted the data extraction form with two studies and revisions were made. Where there was disagreement in data abstraction, a third and independent reviewer assessed the study and resolved the conflict. For those studies where separate papers reported on different outcomes of the same study, only one data abstraction template was completed.

We initially used the Cochrane Effective Practice and Organization Care Review Group (EPOC) taxonomy to describe the implementation strategies included in this review. The EPOC taxonomy has been used in previous reviews of implementation strategies with a similar scope but with different practice settings [16, 26]. To classify implementation strategies, we started by deductively categorizing strategies into the EPOC taxonomy’s section on implementation strategies [27]. Next, for those strategies not included in the EPOC taxonomy, we used an inductive thematic analysis approach to group these strategies and generate additional implementation strategy categories [28].

Two reviewers independently assessed the risk of bias using the EPOC Risk of Bias 2.0 checklist in Covidence [29]. For each of the included studies, risk of bias was assessed as a judgment of high, low, or unclear risk across nine domains. Discrepancies were resolved by consensus with a third reviewer when necessary.

We conducted a narrative synthesis after identifying methodological and clinical heterogeneity in the studies of this review; this indicated that meta-analysis was not appropriate. A narrative synthesis allows for a description of implementation strategies and their effects in achieving outcomes for guideline implementation in nursing [30]. The frequency of each EPOC taxonomy category, outcome measure category, and outcome effect is reported. For outcome effect, we describe whether the outcomes were reported as a statistically significant positive effect or had no effect. Separate comparisons were made for categories of implementation strategies and compared to grouped study results to determine whether they were related to positive and significant improvement in: professional knowledge outcomes, professional practice outcomes, patient health status outcomes, resource use outcomes, and/or expenditure outcomes.

All database searches and hand-searching of reference lists of included studies yielded a total of 38,172 citations. No studies were found using the Grimshaw et al. (22) search strategy. After removal of 4890 duplicates, 33,282 citations were screened, and 924 potential articles were identified. From this set, 878 articles did not meet our inclusion criteria and were excluded from analyses. The majority were excluded for not using an experimental study design (RCT), not implementing a guideline, and not including a nursing population. Forty-one studies (reported in 46 papers due to five companion reports) met final inclusion criteria (Fig. 1). All studies included were conducted between January 1996 and September 2020.

Full details of study-level characteristics are described in Table 1. The final set included 41 randomized controlled trials (RCTs), of which 24 were reported to be cluster randomized controlled trials (cRCT). The number of intervention arms ranged from two to three (two arms = 36 studies, three arms = 5 studies). In total, the 41 studies included a minimum of 5429 nurses (2 studies did not specify the number of participating nurses). In addition, a minimum of 231,218 patients were implicated in 27 studies, with 14 studies not reporting on the number of participating patients.

Eighteen studies were conducted in North America (16 in the USA, 2 in Canada), 5 in the UK, 13 in continental Europe (4 in the Netherlands, 3 in Germany, 2 in Finland, 2 in Italy, 1 in Austria, 1 in Belgium), 2 in South Africa, and 3 in Asia (1 in Thailand, 1 in China, and 1 in South Korea) (Table 1).

Key guideline topics included respiratory (asthma, pneumonia) (n = 7), heart conditions (n = 5), pain (n = 3), cancer (n = 2), diabetes (n = 2), pressure ulcer prevention (n = 2), maternity care (n = 2), and urinary incontinence (n = 2) (Table 1). Various nursing practices and health care settings were identified, including intensive care units, nursing homes, and in the community. The use of theory to inform the intervention and/or implementation strategy was used in 17 studies, including identifying behavioral and environmental determinants, identifying intervention targets, selecting implementation methods and delivery strategies, and informing measurement and evaluation.

Methodological quality of the studies varied (Table 2). Most of the studies had a low risk of bias in their allocation concealment (n = 33, 80%), baseline outcome measurements (n = 36, 88%) and baseline characteristic similar (n = 32, 78%), as well as a low risk in other sources of bias (n = 37, 90%). Two-thirds of the studies completed high-quality (low risk) random sequence generations (n = 27, 66%), while a number of studies had an unclear risk of randomization techniques (n = 14, 34%). Knowledge of the allocated interventions was adequately prevented in half the studies (n = 21, 50%), while in the rest of the cases it was not clear (n = 19, 46%), or not prevented (n = 2, 5%). Compared to other categories, a higher level of risk was found for incomplete outcome data (n = 11, 27%) and in protection against contamination (n = 7, 17%); however, the majority were low risk of bias in these categories (n = 31, 76% and n = 29, 71%, respectively). Lastly, the majority of studies had unclear (n = 18, 44%) and low risk of bias (n = 24, 59%) for selective outcome reporting.

Table 3 reports on implementation strategies used across all studies. Of the 41 studies, a total of 152 strategies were used. Multi-component implementation strategies were most commonly used (n = 36 studies) with only five studies reporting single component strategies.

Of the three categories of EPOC implementation strategies, all studies reported interventions targeted at healthcare workers (Table 3). The most frequently used strategies to target nursing guideline implementation were educational strategies (i.e., educational meetings (n = 33), educational materials (n = 27), and educational outreach visits (n = 12)). Audit and feedback strategies were used in 11 studies.

A wide range of educational strategies were used which highlights the challenge of classifying these interventions under the same heading (Table 4). Many studies used training sessions that included lectures, discussions, and video presentations to provide guideline information [31‐36]. Other studies employed an interactive approach to educational meetings, including case study discussions, hands-on exercises in small teams, and human patient simulation scenarios [37‐39]. The educational meetings ranged in duration and frequency—from single education sessions (i.e., 30-min training sessions) to more multi-phased interventions that included multiple educational meetings over time (i.e., one, 2-h meeting per month for 6 months) [40].

A majority of studies (n = 23/41) used implementation strategies not included in the EPOC taxonomy (Table 5), which consisted of the following: adaptation of practice guidelines to local context (n = 9), external facilitation (n = 14), and changes to organizational policy (n = 3). In addition, a sample of the included studies also reported details of the development and delivery of specific implementation strategies, indicating participatory co-creation approach during implementation (n = 3 studies). Table 5 outlines these four additional implementation strategies and compares them to the EPOC taxonomy.

One hundred and two outcomes were measured across the 41 studies. The most common outcomes were professional practice (n = 49), followed by patient health status (n = 26), professional knowledge (n = 14), expenditure (n = 8), resource use (n = 5). The majority of outcomes (60%) were reported as positive and significant, including 64% (n = 9/14) of professional knowledge outcomes, 59% (n = 29/49) of professional practice outcomes, 54% (n = 14/26) patient health status outcomes, 80% (n = 4/5) of resource use outcomes, and 63% (n = 5/8) of expenditure outcomes. A summary of study outcomes is reported in Table 3.

We grouped the implementation strategies into five mutually exclusive categories to provide a narrative synthesis of study results. We created a sixth non-mutually exclusive category to describe multi-component strategies that were composed implementation strategies not included in EPOC taxonomy.

We synthesized the findings from 41 studies (reported in 46 papers) on guideline implementation strategies for nursing practice. Multi-component educational interventions were most commonly used and included a combination of educational materials, educational meetings, and educational outreach (n = 36). Studies evaluating single implementation strategies focused on educational meetings alone (n = 5) or audit and feedback (with no educational component) (n = 1). Outcomes pertained to professional knowledge, professional practice, patient health status outcomes and less frequently, health system outcomes (resource use and expenditures). Given the combination and permutations of implementation strategies and outcomes, we were limited in the comparisons we could analyze. Meta-analysis was not possible owing in part to the heterogeneity among studies, including differences in implementation strategy content, mode of delivery, duration, and frequency, as well as outcomes collected (Table 3; Table 4). An important finding was that 56% of studies employed implementation strategies that are not included in the EPOC taxonomy, including the use of external facilitation (n = 14), multifaceted participatory approaches (n = 3), adaptation of practice guidelines to local context (n = 9), and changes to organizational policy (n = 3) (Table 5).

Our analysis suggests that educational meetings, in combination with other educational strategies (i.e., materials, outreach visits), are highly used in nursing and likely an effective implementation strategy for guideline implementation in nursing. Distribution of educational materials alone is effective but may not be sufficient to impact outcomes. The majority of studies (n = 40) evaluated educational interventions on professional knowledge outcomes (n = 14), professional practice outcomes (n = 48), patient health status outcomes (n = 26), and resource use/expenditure outcomes (n = 13). Overall, positive effects were found on the majority of professional practice outcomes (n = 29, 59%), professional knowledge outcomes (n = 9, 64%), patient health status outcomes (n = 14, 54%), resource use outcomes (n = 4, 80%), and expenditure outcomes (n = 5, 63%). Multi-component implementation strategies composed of interventions not included in the EPOC taxonomy (i.e., participatory approaches, facilitation, changes to organizational policies) also demonstrated positive trends on professional knowledge, professional practice, patient health status, resource use, and expenditure outcomes. Of the 43 outcomes measured with these participatory-based implementation strategies, 26 were reported as positive (n = 26, 60%).

Our findings on effective educational implementation strategies are in line with previous reviews of guideline implementation strategies in medicine [22], pharmacy [16], rehabilitation [13], and physiotherapy [83]. A previous review of knowledge translation interventions for promoting evidence-informed decision-making among nurses found that almost all studies identified in their review included an educational component [11]. The primary focus on educational implementation strategies assumes that nurses and other health care providers do not implement guidelines because they do not have the appropriate knowledge (i.e., barrier to guideline use is lack of knowledge of the guideline). However, many behavioral determinants, including but not limited to, an individual or group of individual’s motivation, practice context, and social influences affect the implementation process and outcomes [84]. A substantive body of implementation science research has identified modifiable behavioral determinants and/or contextual mechanisms related to implementation in health care [85, 86]. To move beyond educational implementation strategies in nursing, the field needs to be assessing barriers to guideline use, including the professional and organizational barriers to change, and use this assessment to tailor interventions to the identified barriers [87, 88].

The field of nursing is producing good quality trials. In this review, over two-thirds of studies (n = 29, 70%) were of high quality (as indicated by protection against contamination) with a low risk of bias. Comparatively, a previous review of guideline implementation studies in medicine [22] found 54.5% to be of high quality and 4.5% low quality (RCT n = 110). Despite the differences in the number of studies in medicine compared with nursing, the quality of the studies appears to be similar in both fields. Other reviews on similar topics among allied health professions and nursing report low methodological quality of the analyzed studies; however, these reviews included non-randomized control trials, quasi-experimental, and/or observational studies [11, 13, 16].

Reporting continues to be an issue identified in this review, as was also identified in previous systematic reviews in medicine [22], allied health [13, 15, 16, 83], and nursing [19]. Study quality in the nursing implementation field revealed that 25% of the risk of bias indicators were rated as unclear. Many factors are considered when assessing the methodological quality of included RCT studies such as risk of contamination, concealment of allocation, blinded assessments of outcomes, baseline measures, and follow up of professionals. Unfortunately, the documentation on how these issues are managed in a particular study is often less than adequate, thereby making it difficult to ascertain if it is inadequate reporting or inadequate trial procedures. Reporting guidelines exist for trials and intervention description (EQUATOR Network [89]), including CONSORT statements [90] and the TIDiER guidelines [91]. While some studies in this review preceded release of reporting guidelines, only few provided adequate intervention description that aligns with reporting guidelines for interventions. It was difficult to discern intervention dose to support replicability. Future implementation intervention studies should use reporting guidelines to clearly articulate intervention components and strengthen the evidence base on guideline implementation in nursing.

This review highlights several key discoveries for others involved in implementation work in nursing. First, nursing appears to be advancing the evidence base on implementation strategies. There is an increasing number of RCTs in recent years and over half (n = 23) of the studies included guideline implementation strategies not described in the EPOC taxonomy, including facilitation (n = 14), guideline adaptation to the local context (n = 9), changes to organizational policies (n = 3), and participatory approaches to research (n = 3) (Table 5). A growing body of evidence, stemming from the nursing literature, reports facilitation as an effective strategy to optimize the implementation of evidence into practice [46‐48]. A systematic review of guideline implementation in primary care found that practices supported by facilitators were 2.76 times more likely to adopt evidence-based clinical practice guidelines [46]. Further, guideline adaptation to local context is an implementation strategy that relates to the planned action phases [88, 92, 93] and highlight efforts to align an implementation strategy to the local context and build on existing knowledge locally about effective strategies to increase uptake of evidence-based practice [92]. Similarly, previous research has shown participatory research approaches, that focus on producing knowledge and implementing evidence that is relevant to the needs of knowledge users, is an important strategy to consider when implementing evidence into practice [44].

Many of the identified implementation strategies align with the existing implementation science frameworks. Leeman et al. [94] offer a five-component classification system for implementation strategies, including dissemination strategies, implementation process strategies, integration strategies, capacity-building strategies, and scale-up strategies. The more prevalent educational implementation strategies identified in this review align with the implementation process strategies category. Our additional strategies that are not included in the EPOC taxonomy (facilitation, participatory approach, adaptation of guidelines, and changes to organizational policies) align with the integration, capacity-building, and scale-up strategies. Further, the additional implementation strategies map onto the Consolidated Framework for Implementation Research (CFIR) framework [95] which illustrates how these strategies target known behavioral and contextual determinants, including the intervention characteristics, process, and inner setting domains. This differs from educational implementation strategies that target the characteristics of the individual (i.e., knowledge and beliefs about the intervention). By mapping the identified implementation strategies onto existing frameworks, it is clear that multi-component educational and participatory strategies are useful to address multiple stages of the implementation process, and target multi-level behavioral and contextual determinants of guideline implementation.

Identification of these additional implementation strategies is an important finding for nursing implementation research and practice and may help to move beyond traditional educational approaches to implementation. Facilitation, guideline adaptation, changes to organizational policy, and participatory approaches are strategies that target guideline implementation within teams, units, or organizations. In a nursing context of 24/7 care, the decision-making process often occurs in a team or group context, as well as with individual practitioner decision-making [18, 96]. This differs from previous reviews of implementation strategies in medicine, pharmacy, and dentistry where individual-based implementation strategies are singularly used (i.e., reminders, audit and feedback). These may not function in the same way within a team context. The concern is that by only categorizing implementation strategies similar to reviews with other health care providers, we may be missing an opportunity to understand how implementation works in nursing contexts. To this point, the majority of included nursing trials (n = 23/41) used implementation strategies not included in the EPOC review and showed positive impacts on patient, provider, and health system outcomes.

Overall, implementation researchers and practitioners may find results of our systematic review helpful moving forward. First, for those undertaking similar reviews in the future, there are other taxonomies that capture more of the strategies we identified through our inductive thematic analysis. Numerous taxonomies and classification schemes have been published that describe a range of implementation strategies [97]. For example, the Expert Recommendations for Implementing Change (ERIC) project provides a compilation of implementation strategies, including strategies such as facilitation, promote adaptability, assess for readiness, identify barriers and facilitators, and develop academic partnerships [42]. Further, Slaughter et al. 2017 [98] provide practical resources for implementation researchers that includes a variety of classification schemes for knowledge translation interventions. Nursing implementation researchers should aim to explore the effectiveness and feasibility of these additional types of interventions in future work. Second, nurses in practice settings or other disciplines that work primarily in teams may benefit from using a taxonomy that includes team-based implementation strategies to plan and execute their implementation projects.

A second important discovery is that pre-existing search strategies for implementation strategy literature in other health professions does not work for locating guideline implementation studies in the nursing literature. We duplicated the search strategies described in a previous review of implementation strategies used in medicine [28]; however, these strategies did not locate any relevant nursing literature. We then crafted our own extensive search strategy and located the papers included for this review. This may be related to the medicine-focused search strategy not including CINAHL or other nursing databases (e.g., Proquest Nursing and Allied Health or British Nursing Index) and not including all types of common nursing roles (e.g., registered nurses). As such, efforts are needed to go beyond pre-existing search strategies and taxonomies to capture nursing trials and the strategies employed. The nursing implementation science field could be advanced by a separate review of non-RCT studies to understand a broader base of implementation strategies. Findings from this present review highlight that the state of science is not mature enough for solely analyzing RCTs; useful information from other types of studies would supplement findings from RCTs. The current review provides the start of a search framework to be used and expanded on by others to explore additional guideline implementation strategies in nursing in quasi or non-experimental research designs and qualitative research studies.

A key finding from this review is the number of studies (n = 23) that used participatory-based implementation strategies (Table 3; Table 5). Unfortunately, due to heterogeneity of implementation strategies, direct comparisons between the 13 studies that only included EPOC taxonomy implementation strategies versus these 23 studies is not possible. Future research is needed to explore the effectiveness of these types of implementation strategies not included in the EPOC taxonomy. This will help to move this field beyond educational interventions and understand how different components of these strategies work in a nursing context and with what effect. Further, it will help practitioners select the most appropriate, feasible, and effective implementation strategies for their specific nursing context. As previously discussed, we recommend examining these implementation strategies in the context of descriptive and qualitative studies to understand what works, for whom, and in what context.

Importantly, the implementation science literature generally recommends the use of theory to guide intervention design and implementation. Use of theory supports development of interventions that target behavioral determinants and lead to potentially stronger effects [84, 99]. Further, theory use leads to evaluations that are more robust in developing a theoretical understanding of intervention effects [84]. Despite its benefits, to date, reviews in medicine and other allied health professions have not reported extensive use of theory in intervention design. In a review of guideline dissemination and implementation strategies in the cancer care context, only one of 33 included studies used theory to directly inform the design of the intervention [100]. Similarly, only 14 out of 158 studies included in a review of uptake of evidence-based interventions in maternity care reported the use of theory [101]. Notably, in the current review, 40% of studies used theory to inform the design of interventions including studies. It appears that nurses involved in implementation research are early adopters, with the use of theory in intervention design dating back to included studies published in 1997 [56]. This may be explained by the strong theoretical underpinning in nursing and the use of theory to inform nursing practice [102]. Regardless, this warrants further investigation to understand how theory is being used and what is its effect on implementation strategy development and outcomes in nursing.

The following limitations of this systematic review should be considered. First, we used a broad definition of guidelines so our interpretation may differ from others. Second, only studies published in English were included and potentially relevant studies published in other languages may have been missed. Third, many papers lacked detail on the implementation strategies used, which made it challenging to synthesize similar strategies and understand the duration and frequency needed to have the desired effect. Lastly, we had hoped to be able to do a meta-analysis; however, a narrative review was conducted because the methodological and clinical heterogeneity of the studies in this review revealed that meta-analysis was not appropriate. This level of heterogeneity among implementation studies has also been found in similar reviews [16, 100] of implementation strategies and speaks to the need for further work in the field to understand implementation effectiveness and improve reporting.