Introduction

People with spinal cord injury (SCI) experience numerous barriers (e.g., lack of transportation, negative attitudes, and increased cost to participate) to being physically active [1]. Considering these salient barriers, it is not surprising people with SCI participate in less physical activity (PA) compared with their able-bodied counterparts, as well as other populations with chronic physical conditions [2]. Notably, large declines in PA occur following discharge from rehabilitation [3].

Considering these large declines in PA following discharge, physiotherapists may be influential interventionists to support people with SCI to be physically active. In fact, evidence supports that physiotherapists are perceived by people with SCI as having the training, time, and confidence needed to provide information to help their patients become more physically active [4, 5]. However, the lack of knowledge and resources are two barriers that significantly impact physiotherapist-led PA promotion [6]. Thus, strategies and interventions are needed to address these barriers and meet the needs of physiotherapists to support their patients with SCI to be physically active.

Behavioural interventions have demonstrated improvements in PA among people with SCI [7]; however, the use of theory in those interventions was inconsistent and the development process lacked engagement of end-users. The Medical Research Council has recommended that the development and evaluation of complex interventions include the use of theory, take the local context into consideration (an integral component of integrated knowledge translation [IKT]), and be systematic [8]. These three recommendations have been well-supported empirically: larger effects on PA behaviour have been observed following interventions that were developed using theory vs. those that were not [9, 10]; consideration of the local context increases the relevance and maximizes the “potential fit” within the stakeholder’s context [11]; and systematic intervention development using the best available evidence ensures the implementation of the highest quality interventions [12]. In order to address these recommendations for intervention development [8], we employed (a) a health behaviour change theory to provide a blueprint of constructs to target individual behaviour change, (b) an IKT framework to involve two end-user groups (people with SCI and physiotherapists) in developing the specific strategies used to target theoretical constructs, and (c) a tool to systematically guide quality intervention development. The effects of this intervention on leisure-time PA (activity one chooses to do in their free time [13]) among people with SCI have been reported previously where we demonstrated very large effects on PA. This was the first behavioural intervention to significantly improve fitness amongst people with SCI [14].

This process is a pioneering example of using both behaviour change theory and IKT to develop a leisure-time PA behaviour change intervention for people with SCI that can be delivered by physiotherapists. The purpose of this paper is to describe five phases of intervention co-development, guided by theories, frameworks, and tools, and evaluate the feasibility of the PA intervention to be delivered in the rehabilitation setting.

Methods

Theory

In order to address key recommendations for intervention development [7], three guiding frameworks were used to help inform the development of this PA intervention: the health action process approach (HAPA [15]) model, the knowledge-to-action cycle (KTAC [16]), and The Appraisal of Guidelines, Research, and Evaluation II (AGREE-II [12]). Briefly, the HAPA model acknowledges that different psychosocial variables (e.g., outcome expectancies, self-efficacy, action planning, and action control) are salient depending on the individual’s stage of motivation (i.e., pre-intender, intender, and actor [15]). Interventions guided by the HAPA model have been implemented among people with SCI and yielded medium to large effects for changes in PA [17, 18]. The HAPA model was used to guide which strategies physiotherapists should implement within the PA intervention based on their patient’s stage of motivation. The KTAC [16] is a framework used to translate research-based or experiential knowledge into implementation (e.g., in practice, policies, and programmes [16]). The KTAC was used to guide the engagement of stakeholders (people with SCI and physiotherapists) throughout all phases of the PA intervention development process. The AGREE-II is the internationally accepted, gold-standard protocol for clinical guideline assessment, development, and reporting [12] and it has also been used to guide stakeholder involvement in the formulation of a PA promotion resource for people with SCI [19]. The 23-item AGREE-II tool was used as a framework to guide the production of stakeholder recommendations for the PA intervention.

PA intervention development process

People with SCI and physiotherapists (our two key stakeholders) were consulted to inform the PA intervention content and delivery. Specifically, people with SCI who took part in our previous studies or who were members of adapted exercise facilities, and physiotherapists specializing in the care of patients with SCI informed the methodology and interpretation of the findings. Guided by the HAPA model, the KTAC, and AGREE-II, the PA intervention was developed and evaluated in five phases (see Fig. 1 and Supplementary File 1 for how the frameworks were applied). Phases 1–4 were used to inform the content, format, and delivery of the PA intervention. Phase 5 describes an evaluation of changes in implementation determinants amongst physiotherapists following training on the developed intervention.

Fig. 1: Summary of the use of theories and frameworks to develop the intervention.
figure 1

AGREE-II appraisal of the guidelines, research, and evaluation, HAPA health action process approach model. Adapted from reference [16].

Phase 1: Systematic reviews and meta-analysis

The first author was a collaborator on two recent systematic reviews [1, 7] and a meta-analysis [9], which provided the evidence base for the PA intervention. A mix of SCI-specific and general physical disability evidence was used because of the limited availability of high-quality SCI-specific information. This evidence base included:

  1. (1)

    The factors that affect PA participation among people with physical disability [1].

  2. (2)

    The interventions, behaviour change techniques, and modes of delivery used to increase PA behaviour and its antecedents among people with SCI [7] and people with physical disability [9].

Phase 2: Key informant interviews with people with SCI

Individual interviews were conducted with 26 adults with SCI who had recently participated in a larger study [20]. Participants had injury levels ranging from C5 to L2, were 31–64 years old, and 1.2–43.0 years post injury. Open-ended questions were administered to understand participants’ experiences or recommendations for strategies that were or were not helpful for engaging in PA from their physiotherapists (Supplementary File 2). The strategies identified were listed for discussion in Phase 4.

Phase 3: National survey of physiotherapists

A national survey was employed from July to August 2015 to assess: (a) whether physiotherapists wanted an intervention to promote PA; (b) physiotherapists’ intervention needs and barriers to promoting PA; and (c) their intervention delivery preferences. Survey questions were adapted from previous physiotherapist PA promotion surveys from other countries [21, 22] and were pilot-tested by three physiotherapists. Invitations to participate in the survey were sent out via email by the Canadian Physiotherapy Association to practicing (>1 year) physiotherapists across Canada. Survey results were summarized using frequency statistics (Supplementary File 3).

Phase 4: Expert panel meeting

Stakeholders were engaged in developing the PA intervention prototype. A panel of experts was formed consisting of people with SCI (paraplegia and tetraplegia, n = 5), inpatient, outpatient, and private practice physiotherapists (n = 5), a physiatrist, and behaviour change researchers (n = 2). A meeting was scheduled to discuss and identify the most relevant results from Phases 1 to 3, highlight missing information, and develop strategies for disseminating the PA intervention. Prior to the meeting, members were provided with a summary of the results of Phases 1–3. An adapted version of the AGREE-II (Supplementary File 1) was followed to guide the meeting agenda. A checklist of panel recommendations was generated from the meeting and a coder external to the project checked the PA intervention prototype to confirm the recommendations were applied and a separate reviewer confirmed that AGREE-II checklist items were met. Revisions of the PA intervention prototype were sent to expert panel members until all members were satisfied with the content.

Phase 5: PA intervention content evaluation—randomized controlled trial of intervention training and implementation determinants among physiotherapists

Participants

Based on a previous examination of a single, 1.5 h education session among university students, a very large-sized effect on knowledge was expected (Cohen’s d = 1.8 [23]). A total sample size of 12 was needed to have 80% power to detect a very large effect at p < 0.05. To be conservative for the difference in population sampled (i.e., university students vs. physiotherapists), a total of 20 physiotherapists were recruited. Participants were recruited from February to April 2017 by contacting participants from past studies who provided consent to be contacted for future studies. Inclusion criteria were physiotherapists who can read and communicate in English, are currently practicing in Canada, and have been practicing for greater than 1 year.

Design

Feasibility of implementing the developed PA intervention was tested using a two-group, pre-test and post-test design. Physiotherapists were matched by the level of experience (years and number of patients serviced/year) in working with people with SCI. Using a random numbers generator, physiotherapists were randomly assigned to an experimental group (n = 10) or a wait-list control (n = 10) group by the first author (JM; Fig. 2). Questionnaires were administered online by a research assistant blinded to the participants’ group assignment. Analysis was also performed blinded to group assignment by JM.

Fig. 2
figure 2

CONSORT flow diagram of participants.

PA intervention training

To assess the feasibility of implementing the developed PA intervention, physiotherapists were trained in the PA intervention content in a 1 h, individual education session. JM delivered the training individually over video conference where the physiotherapist could participate at their preferred location. Physiotherapists were provided with an electronic copy of the developed PA intervention which included a 50-page toolkit outlining intervention strategies (described in the “Results”) at the end of the training.

Measures

A modified theoretical domains framework (TDF) measure [24] was used to evaluate implementation determinants (i.e., barriers identified in Phase 3 such as knowledge, confidence, and resources). Support for the internal consistency reliability and discriminant validity of the TDF measure has been previously demonstrated amongst physiotherapists [24]. A modified affordability, practicability, effectiveness, acceptability, safety, and equity (APEASE)-criteria measure was implemented to assess participants’ perceptions on the feasibility of implementing the PA intervention in the physiotherapist setting [25]. Participants responded to each item on a seven-point Likert scale (1 = strongly disagree and 7 = strongly agree). Lastly, a test was administered comprised of 20 true or false questions to assess knowledge of SCI-specific PA information (e.g., exercise safety considerations, exercise guidelines and effective-behaviour change techniques). The true or false questions were not evaluated at both baseline and post-training to avoid a recall bias on performance.

Procedure

All measures were completed online using Fluid Surveys [Survey Monkey, Ottawa, ON]. At baseline, physiotherapists randomized to the experimental group completed the TDF measure. Immediately following, JM sent them an electronic copy of the PA intervention and explained it through a one-on-one, single-hour lecture and case study via video-conferencing technology. Immediately following training, experimental-group participants completed the APEASE measure and were re-contacted 1-week later for post-testing, which consisted of the knowledge test, re-administration of the TDF measure, and a semi-structured interview to provide feedback on the PA intervention content. Participants in the control group completed the knowledge test and TDF measure at baseline. They returned one-week later to complete the TDF measure, were then delivered the PA intervention resource in the same manner as the intervention group, and completed the APEASE measure and a semi-structured interview.

Statistical analyses

Cronbach’s α for the TDF domain items were each >0.90 and warranted the use of aggregate scores. ANCOVAs with baseline scores as the covariate were conducted to compare TDF scores between groups. Independent samples t-tests were conducted to compare knowledge scores between groups and descriptive statistics were calculated to summarize APEASE scores.

Results

The following is a summary of the key findings from each phase of the PA intervention development and evaluation process.

Phase 1: Systematic reviews and meta-analysis

The systematic review of reviews demonstrated that there are over 200 barriers to participate in PA and that transportation and finances are particularly salient barriers in the SCI population [1]. A key recommendation from this review was a call for multi-sectoral (e.g., researchers, healthcare providers [e.g., physiotherapists], and recreation providers) and multi-level (i.e., individual, inter-individual, community, institutional, and policy) PA interventions [1].

The systematic review and meta-analysis of PA interventions highlighted the most effective-behaviour change techniques for changing PA behaviour and its antecedents in people with SCI and other physical disabilities. These techniques included self-monitoring, feedback on behaviour, goal setting, action planning, problem-solving, reviewing behavioural goals, and information about health consequences [7, 9].

Phase 2: Key informant interviews with people with SCI

Key recommendations included providing referrals to adapted programmes or facilities, connecting with peers, tailoring or asking patients for their activity preferences, educating on financial support, and providing exercise prescriptions. All participant response themes are included in Supplementary File 2.

Phase 3: National survey of physiotherapists

Respondents included 204 physiotherapists who served a general patient population and 35 who had direct experience working with patients with SCI (total n = 239). Respondents with experience treating patients with SCI were primarily female (80%), had been practicing for 1–5 years (40%), and mostly practiced in neurology (49%). Importantly, 90% of respondents indicated that they would use a resource outlining how to promote PA to their patients. Key barriers to PA promotion included lack of knowledge, confidence, and resources (e.g., time, tools, and programmes). These barriers informed the theoretical domains (knowledge, skills, confidence, and environmental resources) that were assessed in the intervention prototype pilot-test (Phase 5).

Regarding intervention development, physiotherapists preferred to receive the PA intervention resource from a physiotherapist; however, researchers and representatives with SCI were also deemed to be acceptable. The delivery format of the PA intervention resource was preferred to be online or an in-person workshop outside of their clinic. The most desired content included behaviour change techniques, example exercises, and a list of benefits and barriers to PA. The most feasible options for delivery of the PA intervention to patients were either brief, one-on-one counselling integrated into regular consultations or resource distribution. A summary of all survey questions and response frequencies is included in Supplementary File 3.

Phase 4: Expert panel

A list of recommendations for the PA intervention format, content, and delivery was made following the panel meeting and confirmed by the panel members (Supplementary File 4). An important outcome of the expert panel meeting was the selection of the HAPA model as the guiding framework for tailoring strategy selection when implementing the PA intervention among people with SCI. Panel members described the need to select strategies based on participants’ level of motivation and emphasized the importance of using behavioural strategies (e.g., self-regulation) to translate intention to action. The behavioural researchers discussed these points and decided that the HAPA model was a good fit to guide the PA intervention based on the panel’s recommendations and previous literature [17, 26].

Phase 5: Randomized controlled trial of PA intervention implementation feasibility among physiotherapists

Twenty physiotherapists (16 female) who had been practicing for an average of 16.6 years participated in this study, 15 of whom had previously worked with patients with SCI. There were no significant differences in baseline characteristics between groups.

Results of the t-tests on the TDF constructs demonstrated that scores for perceived knowledge, skills, confidence, and environmental resources were significantly greater in the experimental group compared with the control group (all ps < 0.05; see Table 1). Intentions were not significantly different between groups. Independent samples t-tests showed significantly higher scores on the test of SCI-specific PA knowledge for the experimental group (M = 16.1 ± 2.1) compared with the control group (M = 11.3 ± 1.6), t = 5.8, 95% CI of the difference = 3.1–6.5, Cohen’s d = 2.57, p < 0.001.

Table 1 Baseline-adjusted post-test scores for the theoretical domains framework (TDF) subscales.

One participant was missing APEASE data due to malfunction of the survey software. Therefore, 19 respondents were included in the descriptive statistics summarized in Fig. 3. Almost all physiotherapists either strongly agreed or agreed that the PA intervention was affordable, practicable, effective, acceptable, had no side effects/safety concerns, and was equally beneficial for physiotherapists across different settings. All physiotherapists either strongly agreed or agreed that they would recommend the PA intervention to their colleagues. Changes to the PA intervention resource were made following recommendations from the semi-structured interviews. This included the addition of a section on wheelchair seating, recommendations for adapted programmes, facilities, and resources, and a summary “cheat sheet”.

Fig. 3: Physiotherapist (n = 19) response frequencies (%) to APEASE questionnaire items.
figure 3

Affordability (one item), practicality (four items), effectiveness (one item), acceptability (one item), safety (one item) and equity (one item) of the intervention were assessed and item scores were averaged for each domain [23].

Summary of the developed PA intervention

A final version of the PA intervention was developed following the five phases of development and evaluation. Briefly, the intervention guides physiotherapists to tailor PA-enhancing strategies based on the patient’s stage of motivation (pre-intender, intender and actor), their context (e.g., resources, barriers and preferences) and in consideration of what is feasible within the physiotherapist’s setting (e.g., inpatient, outpatient and community). Three overarching PA intervention strategies are used:

  1. (1)

    Education includes safety information, basics of PA, the SCI PA guidelines [27, 28], and behaviour change techniques (e.g., action planning, monitoring, and using prompts/cues).

  2. (2)

    Link and refer provide resources or referrals to address key barriers such as finances and transportation, local resources and facilities, information on developing peer connections, and suggest consideration of other supports such as family, occupational therapists, and SCI-specific physiotherapists.

  3. (3)

    Tailored prescription specifies instructions for tailoring strategies to the patient, examples of different activities organized by level of commitment and competition, basic exercise prescription, sample exercise programmes, and ways to adapt common exercise equipment.

Discussion

This project used theory and frameworks to guide the aggregation of stakeholder recommendations to develop an intervention to assist physiotherapists in promoting PA to their clients with SCI. Almost 300 people with SCI and physiotherapists were engaged throughout the five-phase process to inform the development and implementation of the PA intervention. Preliminary testing of the PA intervention content demonstrates its feasibility to be implemented by physiotherapists in the rehabilitation setting.

Novelty of the methods

The five-phase process used in this project contributes uniquely to the literature on the development of physiotherapist-led PA intervention methods in two ways. First, it serves as a model for how to involve both physiotherapists and people with SCI using the KTAC. This dual engagement required consideration of both the barriers and facilitators for physiotherapists to deliver the PA intervention and for people with SCI to perform the behaviour. Other studies have used the KTAC to develop interventions; however, consideration of the interventionist’s (i.e., physiotherapist) context is often overlooked [29]. Involvement of stakeholders has been shown to be the best predictor of the translation of research into practice [11]. For effective translation, both the interventionist and recipient of the intervention should be equally considered. Our intervention development process provides a model for doing so.

Second, the five-phase process was rooted in theory and validated frameworks. The Medical Research Council calls for interventions to be developed using theory, to consider the local context, and for the development process to be systematic [8]. This development process selected a behaviour change theory (HAPA) to guide physiotherapists to select strategies when delivering the intervention, a knowledge translation framework (KTAC) to engage stakeholders in the development process, and a tool to help ensure the evidence-based and systematic development of a quality intervention (AGREE-II). Previous interventions have used the HAPA model [17, 26], the KTAC [30], and the AGREE-II [19] to guide resource or intervention development independently. Using all three meant that the advantages of using theory, co-creation of knowledge, and rigorous systematic development were combined. The intervention’s preliminary evidence for usability in the physiotherapist context supports the utility of this development approach.

Results from the randomized controlled trial of the PA intervention implementation feasibility in the physiotherapist setting demonstrated improvements in perceived and tested knowledge, confidence, skills and resources to deliver an SCI-specific PA intervention. Although intentions to promote PA were not increased, this finding is not surprising as scores were already high, likely demonstrating a ceiling effect. Confidence has been shown to be the strongest predictor of whether a physiotherapist prescribes PA [31]. Experimental-group physiotherapists reported very large-sized increases in confidence to promote PA compared with the control group, suggesting an increased likelihood that physiotherapists would promote PA in practice. The intervention also addressed skills and resources, which were two other important barriers to PA promotion identified by the physiotherapists in the national survey. Addressing barriers, combined with demonstrated improved knowledge and confidence to promote PA supports the potential for uptake and implementation in the real-world practice setting [11, 31]. The PA intervention has been summarized in a toolkit, which has been made available in PDF format on the SCIActionCanada.com website.

Limitations

Some limitations to the development and evaluation process should be acknowledged. First, the semi-structured interview phase only included individuals who used a manual wheelchair as the interviews were conducted as part of a larger study [20]. Power wheelchair users and ambulatory individuals with SCI may have unique perspectives that were not captured in these interviews. Second, the generalizability of the PA intervention to other healthcare providers is currently unknown. Our stakeholders suggested occupational therapists, recreational therapists, and personal trainers may be other interventionists that could deliver the intervention. Third, there was no comparison intervention that used alternative methods to understand whether this process was more effective than other intervention development methods. To date, there has been no established consensus on the optimal method for intervention development; however, exploration of new methods has been encouraged [11]. This research provides one such example of intervention development.

This is the first PA intervention among people with SCI to be developed using behaviour change theory, an IKT approach, and a quality development tool. Importantly, the five-phase process facilitated the involvement of almost 300 end-users which is anticipated to increase the adoption and effectiveness of the intervention when it is implemented in the community. Given the numerous barriers people with physical disability report, intervention developers are encouraged to involve end-users and tailor intervention content and delivery accordingly.