Dissemination and implementation strategies for physical activity guidelines among adults with disability, chronic conditions, and pregnancy: a systematic scoping review

Guidelines and/or recommendations (i.e., statements informed by a systematic review of evidence or based on expert appraisal of synthesized evidence, respectively) [34] for improving fitness levels or health status through PA and/or exercise among specific populations were identified. Eligible guidelines were those released in the past 10 years in Canada or those intended for use in Canada. Guidelines international in scope must have been led by a researcher working in Canada and endorsed by a Canadian organization that promotes PA for eligibility in this study. Guidelines were restricted to the English language due to a lack of funding for translation. Guidelines were identified through targeted website searching followed by a survey of published literature (Supplement 1) and were reviewed by all authors. Note that authors four and five have established research programs that include PA guideline development and promotion efforts for several of the included populations. All 10 guidelines were deemed relevant (Table 1).

Records were sought via published literature, organizational releases, guideline messaging, web-pages, and any other records meeting pre-defined eligibility criteria (Table 2). Records must have reported a strategy for disseminating and/or implementing one of the PA guidelines in Table 1 and targeted adults (18 + years) of the identified specific population or their stakeholders (i.e., healthcare professionals). Included records must have used a human participant pool, been delivered in English, and been published from 2011 onwards. This date reflects recent advancements in the field of dissemination, whereby a greater range of strategies (e.g., social media) are being used for health promotion [35]. All study designs were considered in answering RQs 1, 2, 4, and 5. Experimental, quasi-experimental, pre/post, and prospective designs were considered in answering RQs 3 and 6.

In line with Tomasone et al. [23] and D’Urzo et al., [30] five search approaches were used: (1) peer-reviewed literature databases [Web of Science (Web of Science Core Collection, BIOSIS Previews, KCI-Korean Journal Database, MEDLINE®, Russian Science Citation Index, SciELO Citation Index) and Google Scholar (Publish or Perish (https://​harzing.​com/​resources/​publish-or-perish)] [36], (2) grey literature database (Thesis & Dissertation – ProQuest Dissertations Online), (3) custom Google search engines [Carleton University’s MADGIC search engine (http://​subject-guides.​uwaterloo.​ca/​c.​php?​g=​695548&​p=​4931873) and Ontario Public Health Libraries Association (http://​www.​ophla.​ca/​resources.​htm)], (4) targeted web-based searches, and (5) content expert consultation. Peer-reviewed literature searches, grey literature searches, and custom Google searches were developed and carried out by a professional librarian (third author). The peer-reviewed literature searches were conducted in Web of Science databases as the cited reference feature was used to find articles that referenced the pre-specified list of guidelines (Table 1). Unlike traditional reviews, which find limited value in citation searching, this method was preferred as it is unlikely that articles about guideline D&I fail to reference the guideline in question. Targeted web-based searches were carried out by the second author and audited by the first author. Search terms (Supplement 1) were modified per data source but remained specific to the pre-specified guidelines (Table 1) and to dissemination and/or implementation. The first four search approaches took place from November 2019 to March 2020.

For the targeted web-based searches, the website for the publishing organization of each guideline in Table 1, and for organizations that promote PA to specific populations in Canada (e.g., ParticipACTION), were searched. Additional websites were identified by entering each guideline name into the search bar and scanning the first 100 hits. Two methods were used to identify potential reports of guideline D&I strategies: (i) the 2-click method (i.e., records within 2 clicks of the guideline page were included), and (ii) search strings were developed and run through website search bars (Supplement 1).

The content expert consultation was conducted following the completion of other searches to identify any additional records for each guideline that were potentially missed or not publicly available. Primary authors involved in the development of each guideline were identified as potential content experts. Where no primary author could be identified, the publishing organization was contacted directly (i.e., Diabetes Canada, Parkinson Canada). Where a single researcher led multiple guidelines, one email was sent regarding all relevant guidelines. Initial emails were sent to 8 content experts in February 2020 and included a copy of Table 1 and any reports of dissemination and/or implementation of the guideline identified through the four other search methods. Experts were asked to provide (a) PDFs, citations, and/or links to published manuscripts, or (b) (un)published organizational reports of guideline dissemination and/or implementation, and to nominate an alternate contact if they felt that individual could better respond to the request. Two weeks from initial contact, a reminder was sent offering an additional two weeks to respond. In total, 11 content experts (representing all 10 guidelines) were contacted, from which 10 responses (representing all but the pregnancy guidelines) were received.

Published literature, grey literature, and custom Google search results were screened using the Covidence software [37]. The first and second author independently screened each title and abstract for potential eligibility. Specific to targeted web-based searches, titles of up to the first 100 records from each search were screened [30]. Relevant titles and their corresponding URLs were compiled in a Microsoft Excel spreadsheet for full-text screening.

Full-text screening for peer reviewed literature, grey literature, and custom Google search results was performed by the first and second authors. Discrepancies were resolved at both stages through consultation between the two reviewers and the last author where necessary. Full-texts identified by targeted web-based searches and content experts were screened by the second author and audited by the first author.

The data extraction table in the present review was modelled after Tomasone et al. [23] to extract data pertaining to record characteristics (i.e., title, year, author, participant characteristics, and study design), intervention characteristics (i.e., setting, relevant guideline, strategy and evaluation format if evaluated), dissemination outcomes (i.e., awareness, attitude, and knowledge of the guideline and any other non-specified dissemination outcome), and implementation outcomes (i.e., self-efficacy, intentions, and/or behaviours relating to the guideline, and any other non-specified implementation outcome). The data extraction table was pilot tested by the second author and audited by the first author.

Extracted strategies were placed into seven classifications according to Tomasone et al.’s [23] framework for movement guideline D&I strategies (Table 3), which was adapted from Leeman et al.’s [38] framework for classifying implementation strategies for evidence-based healthcare interventions.

Following data extraction, records were synthesized into two tables, representing D&I strategies. Within each table, records were organized by strategy class (e.g., dissemination process strategy, implementation scale-up strategy), then by strategy type (e.g., distribution of guideline materials, mass media/communication campaigns), then by number of RQs addressed, then by the guideline they aimed to disseminate or implement.

As anticipated, heterogeneity of results pertaining to the effectiveness of strategies precluded the utilization of meta-analysis.

A risk of bias (ROB) assessment was conducted for all records that evaluated the effectiveness of a dissemination or implementation strategy (i.e., RQs 3 and/or 6). ROB assessments were conducted by the second author and audited by the first author. Randomized controlled trial (RCT) study designs were assessed according to the Cochrane Collaboration ROB tool [128]. Non-randomized study designs were assessed according to the Cochrane ROB Assessment Tool for Non-Randomized Studies [129].

Peer reviewed literature searches, grey literature searches, custom Google search engines and targeted-web search approaches identified 528 records. As Web of Science’s cited reference feature allowed us to search multiple databases simultaneously, duplicates from these databases were not recorded. With grey literature searches, duplicates were also not recorded. Thus, zero duplicates were recorded during title and abstract screening. Of the 528 records, 163 records were deemed potentially relevant and were retrieved for full-text screening. Of the 11 contacted content experts, 10 participated and contributed an additional 58 records, all of which were retrieved for full-text screening. The full texts of 221 records were screened and 81 were included in data synthesis (Fig. 1).

The content expert consultation search approach had the highest yield at 44 unique records, followed by targeted web-based searches (n = 27). All other records were identified through peer-reviewed and grey literature searches (n = 10). The custom Google searches identified zero records.

Of the 81 records, 79 pertained to single guidelines while two related to two guidelines (i.e., for cancer survivors and pregnancy) [77, 78]. Forty-two records reported only dissemination strategies, 28 reported only implementation strategies, and 11 reported both D&I strategies, for a total of 53 records discussing dissemination strategies and 39 discussed implementation strategies. There were 23 instances where multiple strategies were used in a single record (e.g., “counselling” and “planning tools”) [106], resulting in 109 instances of the use of a strategy. Members of specific populations were targeted in 51 records, stakeholders (e.g., healthcare professionals) were targeted in 23 records, and 7 records targeted both.

Of the six records evaluating six dissemination strategies, all but one used a non-randomized study design and were rated as “serious” ROB (see last column in Table 4) because of missing data (n = 3) [40, 41, 90], subjective outcome measurement (n = 1) [90], and confounding bias (n = 1) [90]. The one RCT was rated as “high” ROB (see last column in Table 4) due to blinding participants and assessments, attrition bias, and large initial between-group differences [89].

Of the 16 records that evaluated 23 evaluated implementation strategies, 13 were non-randomized study designs and were all deemed as “serious” ROB (see last column in Table 5) due to confounding bias (n = 11) [42, 97, 101‐103, 106‐108, 117‐119], sampling (n = 4) [97, 102, 103, 107], intervention measurement (n = 4) [42, 102, 106, 108], outcome measurement (n = 6) [42, 97, 101, 102, 106, 117], and missing data (n = 1) [130].

The remaining three records evaluating implementation strategies were all RCTs deemed as “high” ROB (see last column in Table 5) due to lack of blinding participants and assessments (n = 3) [98, 109, 110] and attrition bias (n = 2) [98, 109]. Full ROB ratings can be seen in Supplement 2 (RCTs) and Supplement 3 (non-randomized, quasi-experimental, and observational study designs).

Six dissemination process strategies involved “formative research” to support the dissemination of two guidelines (i.e. for MS and Alzheimer’s) [3, 10]. Further, we identified 49 dissemination strategies – a greater number than identified in Tomasone et al.’s [23] review of guideline D&I for the general population. “Distribution of guideline materials” was the most commonly used dissemination strategy type (n = 30), while “mass media/communications campaigns” was the least common (n = 9). Only four records evaluated two types of dissemination strategies (i.e., “distribution of guideline materials” and “education”) and no included record evaluated “mass media/communications campaigns”. Many included records reported distributing fact sheets as a dissemination strategy, potentially due to their simplicity and low production cost relative to other dissemination strategies (e.g., scientific reports, mass media campaigns) [131]. While no included study evaluated a fact sheet, prior research has found fact sheets to improve Australian health professionals’ knowledge and intentions to advise against alcohol consumption during pregnancy [132] and to enhance North American and Western European women’s knowledge and attitudes about congenital infections as effectively as educational videos [133].

As the majority of dissemination strategies were not evaluated, it is difficult to reliably conclude how effective these strategies are at enhancing awareness, attitudes, and knowledge toward PA guidelines. Similarly, we cannot determine whether any of the evaluated strategies were more or less effective than others due to the heterogeneous evaluation measures used. Nevertheless, our findings align with Tomasone et al., [23] suggesting that low levels of guideline awareness and knowledge persisted despite the use of abovementioned strategies to disseminate PA guidelines. Formative research (e.g., engaging stakeholders) is warranted for future studies to identify how dissemination strategies may be best enacted in real-world settings and evaluation is warranted to determine which have the most utility for improving guideline awareness and knowledge among specific populations and their stakeholders.

Nearly as many implementation strategies were identified as dissemination strategies (n = 53 and n = 55, respectively). The most-used implementation strategy was “planning tools” (n = 16) with “feedback”, “alerts”, and “financial incentives” being the least-often used (ns = 1). Unique from Tomasone et al. [23] was our identification of multiple records discussing multiple implementation strategies for a single intervention (i.e., the Changing Minds, Changing Lives intervention promoted the uptake and use of the SCI guidelines among stakeholders) [90, 117‐119]. The researchers of the Changing Minds, Changing Lives intervention evaluated “education” and “stakeholder training” strategies, which informed multiple intervention iterations. Possibly, sequentially applying strategies in a single intervention may provide greater opportunities to assess the effectiveness of, and thus improve upon, D&I efforts.

Nevertheless, few implementation strategies were evaluated and evaluations applied a variety of measures, particularly for self-efficacy. This makes cross-comparison a challenge, rendering conclusions on relative effectiveness impossible. Of the evaluated implementation strategies, “counselling” and “skills training” seem to have been most successful at enhancing self-efficacy, intention, and behaviour to meet PA guideline benchmarks among members of specific populations and “stakeholder training” seemed to have had an influence on stakeholders’ self-efficacy to engage in implementation process strategies. “Counselling”, “skills training”, and “stakeholder training” were often delivered in-person, which is likely to be resource intensive. In light of the COVID-19 pandemic, remote PA counselling and skills training may be a more viable option [134]. Recent work has found that online PA interventions are not only more feasible to implement broadly in absence of research funding, but may also be as effective as in-person formats [135]. Indeed, remote formats may be beneficial for specific populations in overcoming their unique barriers to engaging in PA such as concerns overs accessibility and transportation [109].

Finally, 16 planning tools were implemented among specific populations, such as fillable handouts and checklists to assist individuals in setting PA goals, but only three were evaluated. Planning has been identified as an important, effective step for health behaviour change among the general population [136]. While singular implementation strategies, such as planning tools, may be lower cost [136] and easier to deliver than multifaceted interventions (e.g., combined PA counselling and PA training) [137], evaluating such tools may not be feasible for guideline developers if funding is limited [138]. Still, research utilizing planning tools should aim to investigate their impact on specific population health behaviour when possible.

Eleven records used D&I strategies concurrently, more than was identified in Tomasone et al. [23] This may be because stakeholders of PA guideline D&I for specific populations (e.g., guideline developers, organization members) are more integrated with their target communities than stakeholders of general population PA guidelines. For instance, “formative research”, “distribution of guideline materials”, “counselling”, and “planning tools” were used for both the MS Get Fit Toolkit and the SCI Get Fit Toolkit [43, 54] to simultaneously disseminate and implement the guidelines. However, no record concurrently evaluated D&I outcomes [43, 52]. It would be helpful to understand the impacts of concurrent use of D&I strategies, such as how targeting PA guideline awareness first may augment increases PA behaviour in line with the guidelines. From a theoretical stance, behavioural determinants, such as awareness or knowledge, are antecedents to longer-term outcomes, such as PA [23, 139]. Accordingly, future work should evaluate concurrent uses of D&I strategies to clarify their interplay in promoting the uptake and use of guidelines.

Interestingly, many identified records pertained to the PA guidelines for persons living with and beyond cancer (n = 29), and for persons with SCI (n = 21). Comparatively, the osteoporosis guidelines [11] were reflected in three records. Organizations could benefit from enhanced communication amongst each other and with researchers to endorse strategies deemed to be most effective, which can help advance guideline uptake and use [140]. Indeed, our content expert consultation identified many unpublished and non-public records (n = 43), which may be hampering important advances in the field. Thus, we recommend that researchers in this area make all guideline D&I strategies available, such as with open science, to improve communication.

Despite the smaller scope of our review (i.e., smaller geographic scope, smaller-sized populations, English only, PA guidelines only), more records of specific population PA guidelines were identified (n = 81) than the Tomasone et al. [23] review of PA, sedentary behaviour, and sleep guidelines (n = 47). Perhaps, strategies in specific populations are developed and enacted more frequently because specific populations are more defined, connected with health professionals (e.g., oncologists), and likely to ask for guideline resources relative to the general population [141]. Moreover, PA guidelines for specific populations are often disseminated and implemented with dedicated research grant funding with expectations for end-of-grant knowledge translation efforts by guideline development groups through existing networks with end-users. Nevertheless, no firm conclusions can be drawn regarding whether the higher number of identified records in this review resulted from specific population PA guidelines having been developed and supported by special interest groups. The high variability in methods used to evaluate D&I strategies created a challenge for determining strategies’ relative effectiveness. Unfortunately, some strategy classes (i.e., implementation scale-up strategies) and some types (i.e., “mass media/communications campaign”, “human resources”) were not evaluated, similar to Tomasone et al., [23] and dissemination scale-up strategies were not represented at all in our review. Finally, despite the usefulness of D&I process strategies in establishing the needs of the target audience [23], only six identified implementation strategies in our review were either paired with “stakeholder engagement” (i.e., integration [98]; capacity-building) [97] or “human resources” [42, 100‐102].

These findings bear practical and knowledge implications for future study and reporting in D&I. First, the present results may inform future guideline D&I work in Canada or analogous countries (i.e., high-income status, English-speaking). Specifically, it appears that “counselling”, “skills training”, and “planning tools” are deemed to be effective at enhancing self-efficacy, intentions, and PA behaviour in line with guidelines [42, 98, 101, 102, 106, 108‐110] and “stakeholder training” is deemed to be effective at improving self-efficacy to perform implementation process strategies [97, 107, 117‐119]. Further, limited evidence suggests that “formative research” may be useful in determining the quality of dissemination efforts [40, 41] and “education” may be effective at increasing awareness and knowledge of PA guidelines [90]. Thus, to promote PA among specific populations, it is recommended that “formative research” and “education” be used to enhance awareness and knowledge and “counselling”, “skills training”, and “planning tools” be used to enhance self-efficacy, intentions, and PA behaviour in line with the guidelines. Future research should continually investigate their effectiveness, to support refinement and positive evaluations of these strategies among the general population [23], and should begin to evaluate other types of strategies (e.g., “mass media/communications campaigns”, “implementation toolkits”) to determine which contribute to positive outcomes. In all populations, future studies should incorporate more formative research on strategy development and improvement and continue using integrated knowledge translation approaches that engage relevant stakeholder groups from project outset [142], which may improve the potency of future guideline D&I efforts [143]. Overall, there is a need for further use and investigation of the effectiveness of a wider range of D&I strategies for both the general population and specific populations.

This systematic scoping review has several notable strengths. Primarily, the methodology used was rigorous and adhered to published standards [30, 31, 33]. The comprehensive search strategy identified 81 relevant records reporting D&I strategies. The majority of records were located through the content expert (n = 44) and targeted web-based (n = 27) search approaches, demonstrating the appropriateness of the systematic scoping review methodology used. Secondly, this review applied the framework by Tomasone et al., [23] enabling the classification of six dissemination process strategies, which showcased the utility of collaborative, formative approaches for enhancing dissemination strategies. The present study also highlighted a need for greater frequency and consistency in the evaluation of D&I strategies and their outcomes. Moving forward, comparative evaluations could help determine whether certain strategies are more effective than others within a given population, or are more effective in one population than another. Finally, while this review identified strategies for disseminating and implementing PA guidelines to only eight specific populations, the present findings could apply to other researchers disseminating or implementing PA guidelines more broadly. Along with findings from Tomasone et al. [23], it appears that “feedback”, “financial incentives”, “counselling”, and “planning tools” may be effective for PA guideline implementation regardless of the population of interest.

This review is not without its limitations. Specifically, our inclusion criteria may have limited the number of relevant records identified. For example, because some of the included guidelines are international in scope, such as the 2017 SCI guidelines [9], it is possible that they could have been disseminated and/or implemented by other countries; thus, our review precluded inclusion of such records (c.f., [144]) despite them utilizing the same guideline. Second, our review provides a snapshot of D&I strategies for specific population PA guidelines at a specific point in time, which is subject to change as additional D&I efforts are made. However of interest, even the content experts who led guideline D&I were not aware of any additional, recent records, suggesting that the majority of D&I efforts happen soon after a guideline is released. Further, the few evaluated strategies and large variation in evaluation methods did not support a comprehensive evaluation of the relative effectiveness of D&I strategies. Finally, the ROB tools used may have been a limitation as they identified a high degree of bias in the included studies, when the problem may instead be systemic. For instance, large sample sizes and control groups may not be possible or ethical when conducting PA research among specific populations as PA opportunities for these individuals tend to be highly sought-after yet rare [145].