Physical activity implementation in the curricula of healthcare professions: a systematic review

Search methods and eligibility criteria

We conducted a unregistered systematic review in line with the Centre for Reviews and Dissemination recommendations for undertaking systematic reviews¹³ and the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement.¹⁴ Inclusion criteria were determined utilising the PICOS (Population, Interventions, Comparison, Outcomes, Study design) format (Table 1).

The systematic review was limited to the publications written in the English language. We excluded studies based solely on PA interventions without an educational component. Interventions that combined PA with other health behaviours (e.g., healthy eating) were also excluded if outcomes were not available for PA knowledge, behaviour, and confidence.

PICOS inclusion informed the selected search terms. Searches were performed by a specialist librarian of the following databases: Cochrane Central Register of Controlled Trials (Issue 9 of 12, September 2020), CINAHL (1982-present), Embase (1974-2020), ERIC (1966-2020), Medline (1946-2020) and SPORTDiscus. Examples of search terms (abstract, keyword’s, MeSH term, subject heading, title) were ‘students’, ‘education’, ‘curriculum’, ‘exercise’, ‘physical activity’, ‘professional’, ‘medical’ ‘nursing’ or ‘allied health’, ‘lifestyle’, ‘systematic review’ and ‘randomised controlled trial’. Reference lists of included articles returned from the database search were scanned for relevant publications. No time limit was specified, original search was completed in 2020 and an updated search was conducted in Oct 2022. The search algorithm can be found in Appendix 1.⁵⁴

Study collection and data extraction

All records generated by the searches were imported onto Rayyan¹⁵ and duplicate records were removed. Two reviewers (RP and IML) independently screened the titles and abstracts from the literature search according to the eligibility criteria (Table 1). Full texts of articles which were consistent with study eligibility, were obtained and assessed independently by three reviewers (RP, IML and SB). Full-text articles that met the eligibility criteria were included in the review. Disagreement on selection of articles was solved by discussion until consensus was reached, another author (GM) was available to arbitrate if needed. Reasons for exclusion of non-eligibility studies were recorded, see Appendix 2.⁵⁴

After full-text screening, RP and SB extracted data using the 2016 Guideline for Reporting Evidence-based practice Educational interventions and Teaching (GREET).¹⁶ The GREET checklist includes items recommended for reporting educational interventions for facilitating foundation knowledge and skills in evidence-based practice. Specifically, questions are asked to address the why, what, who, how, where and the how well of the educational intervention. For each eligible study, information on the characteristics of studies, study population, PA educational intervention and outcomes were extracted. Characteristics of studies included study design, country of origin and year of publication. Characteristics of study population included sample size, pre-registration/post-registration, and professional background. Characteristics of PA intervention included content, mode of delivery and duration. Data was also extracted on outcomes assessed, statistical analysis and effect of intervention on outcomes. The data extracted from the eligible studies are in line with the PRISMA guidelines, summarised in the characteristics of eligible studies Table 2.

Risk of bias assessment

The purpose of the risk of bias assessment was to assess the methodological quality of the eligible studies and to determine the extent to which they have addressed the possibility of bias in its design, conduct and analysis. Two authors (SB and RP) independently assessed risk of bias using Joanna Briggs Institute critical appraisal tools.¹⁷ Any disagreements were resolved through discussion, with a third reviewer (IML) acting as an arbitrator when necessary.

Data synthesis

Meta-analysis was not possible due to the variation in quantitative study designs and the widely varying methods of measurement and analysis employed by the analytical studies, and/or insufficient data reported. Consequently, a narrative synthesis was undertaken because it can be used in systematic reviews focusing on a range of questions, not only those relating to the effectiveness of a particular intervention.¹⁸ To assess the overall effectiveness of the reviewed educational interventions, we first evaluated each study’s design and then we assessed the strength of evidence of measured outcomes: knowledge, perceptions, confidence, attitudes, beliefs, and changes in behaviour/practice (where available).

Search outcomes

In total 1,542 articles were identified and screened. Following title and abstract screening, 1,411 articles were excluded. After full-text screening of 57 articles, 25 studies were included in the systematic review. An updated search was undertaken in Oct 2022, a total of 863 articles were retrieved across various databases; 67 duplicates were removed, 796 were abstract screened, 743 were excluded which left 53 articles for full-text screening. Of the 53, seven additional articles were eligible for inclusion. Therefore, the total eligible number of studies was 32. The screening process and reasons for exclusion of full text articles are shown in Figure 1 using the PRISMA 2020 flow diagram for new systematic reviews. Reference to excluded studies and reasons can be found in Appendix 2.

Figure 1. PRISMA 2020 flow diagram.

Characteristics of eligible studies

The characteristics of the included studies are reported in Table 2. The eligible studies included the following populations: medical students (n=12), medical doctors (n=8), pharmacy students (n=2), physiotherapy students (n=1), physician assistant students (n=1), nursing students (n=1), nurses (n=1), and a mixed cohort (n=6). The sample size by profession included, medical students (n=1801), medical doctors (n=452), pharmacy students (n=160), physiotherapy students (n=32), PA students (n=40), nursing students (n=50), nurses (n=117), and mixed cohorts (n=717).

Of the included studies, 11 were held in the United States of America (USA), six in Australia four in the UK and Canada, two in Ireland and one study in each of the following countries: Brazil, Mexico, Israel, India and Spain. All studies were published in a range of academic journals between 1999-2022. Seven of the studies had a randomised controlled trial (RCT) design, eight were non-randomised controlled trials, and 17 were single group pre-post-test designs. All trials included a single publication except for Jadczak et al. (2018)¹⁹ which included two publications. Total size of sample was 3,256 participants, pre- and post-registration learners accounted for 2370 and 844, respectively.

Characteristics of learning approaches included in eligible studies

Duration of the PA education interventions varied and ranged from a two-hour lecture to a 16-hour intervention over an eight-month period. Overview of commonly used learning pedagogical approaches are presented in Table 3.

Where a control group existed, no intervention, traditional intervention (i.e. delivery of existing curriculum), or usual practice were the most common types. In most cases they were from the same population as the study sample except for Kotecki and Calyton (2003),²⁰ who compared pharmacy students with practising pharmacists.

The eligible studies used a variety of outcome measures. The most reported educational outcomes were knowledge (n=16), perceptions (n=3), confidence (n=10), attitudes (n=3), beliefs (n=1), and changes in behaviour/practice (n=10). Most of the outcome measures were self-developed, some were based on theoretical models such as Theory of Planned behaviour, and the only validated measure used was Kirkpatrick’s model of training evaluation. Two of the studies used structured clinical examination using a standardised patient to assess changes in practice.^21,22

Risk of bias assessment outcomes

Reporting of the studies was generally good with all studies clearly stating their research aims and appropriately describing methodology. Recruitment and data collection were described by all studies, convenience sampling was the most used sampling technique, and 11 studies adequately described loss to follow up.

None of the studies described a relationship between the participants and the researchers, making it impossible to establish what relationship, if any, existed. Eight of the studies lacked detail on who delivered the educational intervention making it difficult to establish if there may be a risk of bias introduced through professional association. Most of the outcome measures were self-developed and therefore, a key limitation of the studies, appropriate statistical tests were applied for all the studies.

For most of the eligible studies, PA specific content was adequately described, and frequency and duration of sessions were often well described, what was less clear was the scheduling of sessions and time in between them.

For the seven eligible RCTs, only two^23,24 reported “true randomization”, while only one²³ reported “concealment” of the allocation to groups. Two RCTs^21,25 reported no “similarity of treatment groups at baseline”, while all RCTs reported: a) blinding of participants to the treatment assignment, b) blinding of those delivering the treatment assignment, and c) blinding of outcomes assessors to the treatment assignment. Also, all RCTs reported that treatment groups treated identically other than the intervention of interest and that participants analysed in the groups to which they were randomized. Finally, three RCTs^23,26,27 reported complete follow-up or adequately described the differences between groups adequately, while the latter information was missing from the remaining four RCTs. The outcomes of the risk of bias assessment of the eligible RCTs can be found in Table 4, while a summary of them can be found in Figure 2.

Figure 2. Summary of risk of bias assessment outcomes of randomized controlled trials.

Only two^22,28 out of the 25 eligible quasi-experimental and single-arm design studies did not report the “cause” and the “effect” of interest in their research question. Six studies reported similar participant characteristics in comparison groups,^29–34 one study²⁸ reported differences in participants across groups, and in three studies^22,35,36 this information was unclear. In the remaining 15 single group studies this risk of bias component was not applicable.

Six studies^{22,29,32,34–36} reported that apart from the allocated intervention all participants received similar treatment, one study²⁸ reported differences in received, in two studies^30,31 this information was unclear, while in the remaining 16 this risk of bias component was not applicable.

Also, there was a control group in nine^{20,28–31,33–36} out of the 25 studies, while all studies reported multiple measurements of outcome (pre-post). In follow up completeness and adequately description component, nine studies^{20,22,29–32,34,37,38} reported yes, in one study²⁸ this information was unclear, while the remaining 10 studies reported no. Regarding the component of outcomes of participants to be included in same way, in two studies^31,35 this information was unclear, nine studies^{20,22,28–30,32–34,36} reported yes, while in the remaining 14 studies this risk of bias component was not applicable. In the outcomes measurements reliability component, two studies^39,40 reported no, in seven studies^{20,22,28–31,35} this information was unclear, while the remaining 16 studies reported yes. Finally, all studies reported appropriate statistical analysis. The outcomes of the risk of bias assessment of the e eligible quasi-experimental and single arm design can be found in Table 5, while a summary of them can be found in Figure 3.

Figure 3. Summary of risk of bias assessment outcomes of quasi-experimental and single arm design studies.

Data synthesis outcomes

Didactic input

Of the three studies^21,35,41 which reported a solely didactic approach to education delivery, two^35,41 reported statistical improvements in confidence and knowledge in students’ ability to assess and counsel about PA as well as students’ beliefs regarding the importance of PA in managing disease. Maloney et al 2011,²¹ was the only RCT and reported no statistical difference in the knowledge and change in practice between the intervention and the control group.

Training workshops

Interactive training workshops were the most common method of educational delivery. Of the 13 eligible studies, four (of which one was a RCT and two were non-randomised control trials) did not reach statistically significant findings in relation to knowledge, behaviour, and perceptions. The remainder were single group pre-post-test design and statistical improvements in confidence, attitudes, beliefs, knowledge, and behaviour were observed.

Self-directed e-learning

Self -directed on-line learning was exclusively used in three of the studies, all of which were single group pre-post-test design.^8,38,40 Raj, Khan and Nair (2020) study⁴⁰ did not reach statistical significance in knowledge scores, Pugh et al (2020) did achieve statistical improvements in relation to students’ knowledge and confidence and Webb et al.³⁸ revealed a statistical increase in the frequency of discussion on PA.

Didactic with Simulation-based learning

Didactic with simulation-based learning was used in three of the studies, of which two^20,30 were non-randomised control trials and the remaining study⁴² was a single group pre-post-test design. Of the three studies, Jadczak, Tam and Visvanathan (2018)³⁰ reported a statistical improvement in perceived competence, Kotecki and Clayton (2003)²⁰ reported statistical improvements in confidence and Bass iii et al (2004)⁴² noted a statistical improvement in knowledge and self-efficacy for exercise prescription.

Didactic with experiential learning

Experiential learning for Hasson et al (2018)³⁹ was in the form of pedometer use and whilst the intervention did not impact knowledge amongst public health nurses it did result in statistically significant positive changes in attitudes and behaviours when compared to baseline data.

Experiential learning

Of the four studies that included interventions comprising experiential learning, three were RCTs (Cooke; Marques-Sule; Shields 2010), and one was a single group pre-post-test design (Jadczak et al., 2017). Marques-Sule and colleagues (2022) found a statistical within-group improvement in Physiotherapists PA knowledge in the intervention group but no change in the control group, while Cooke et al. (2015) found improvements in UK medical students perceived behavioural control in both intervention and control groups. Another RCT found that contact with adolescents with Down syndrome during a physiotherapy clinical placement statistically improved students’ attitudes towards exercise for people with Down syndrome.²⁷ In a single-group study, Jadczak, and colleagues (2017) observed that an intervention consisting of medical student-led interviews of older people participating in a community-based exercise class led to statistical improvements in students perceived competence in counselling older people about exercise.

Flipped classroom

Of the three eligible studies that used a flipped classroom approach, only Ockene et al (2021)²⁶ was a RCT, Pasarica and Kay (2020)²² was a non-randomised control trial and Mattison and Nemec (2014)⁴³ was a single group pre-post-test design. In relation to perceived skill, Ockene et al.’s study revealed no statistical difference between the intervention and control group. Pasarica and Kay (2020) revealed no statistical difference in perceived confidence and Mattison and Nemec (2014) noted a statistically significant improvement in knowledge following the intervention.

Other

Freene et al. (2022) found that the use of PA champions to encourage access to a PA resource across 13 health disciplines in one Australian university, resulted in no statistical changes in PA promotion preparedness, PA knowledge, or the amount of PA content delivered, despite a statistical increase in the use of teaching resources.

Recommendations/Implications for practice and research

Although various e-learning approaches were used in the appraised studies, such as online self-learning modules, interactive exercises, and web-based discussion forums, they could have been developed further to include virtual consultations. The interventions that did compare in-person versus web-based approaches produced similar outcomes, suggesting that from a time and cost efficiency perspective, e-learning approaches may form e a growing evidence base.

Instructors involved in the delivery of the educational interventions included research study teams, strength and conditioning specialist, medical doctors, academics, pharmacist, exercise physiologists and nurses, calling for a multi-disciplinary team approach to teaching and learning.

Strengths and Limitation

The current systematic review sheds light on the teaching and learning approaches used in delivering PA education for health care professionals. Whilst this review has captured teaching and learning approaches for pre- and post-registration provision across all health care professions, this review does have several limitations. First, this review built exclusively on published studies, whereas unpublished studies, grey literature and non-peer-reviewed literature were excluded. Although including unpublished, grey, and non-peer-reviewed literature has potential benefits in terms of comprehensiveness, it can introduce greater bias in the results of the systematic review. Unpublished studies are usually of lower methodological quality than published studies. The educational intervention studies within this review generally have design limitations which limit the ability to support replication and draw any firm conclusions.

Given the heterogeneity and reliance on self-developed, self-report survey measures, caution must be applied to the validity of the reported outcomes due to the risk of response bias.⁵³ Most of the studies lacked an adequate sample size, and, in many cases, there was a high loss to follow up which will result in selection bias representing a threat to the internal validity of the studies. Similar design limitations have been reported elsewhere.¹²