Cardiovascular disease (CVD) accounts for approximately one-third of all deaths in the United Kingdom and places a substantial economic burden on the UK National Health Service (NHS), with costs estimated at £14 (US $18.3) billion and rising [1]. Epidemiological and experimental studies have established strong links between low levels of occupational and leisure time physical activity (PA) and an increased risk of CVD [2-5], and this is supported by findings from a recent meta-analysis [6], which found significant associations between moderate occupational and leisure time PA and CVD risk in both men and women. In addition, a wealth of evidence demonstrates that participation in PA is associated with improvements in metabolic risk factors for CVD and CVD-related mortality [5,7,8].

PA affects the main risk factors for the development of CVD, including decreases in low-density lipoprotein [9] and maintenance of normal glucose tolerance [10]. In addition, there are protective effects resulting from weight loss or maintenance [11] and its effect on blood pressure in physically active individuals. There are negative physiological responses associated with a lack of PA and high levels of sedentary behavior [12,13]. Research has shown that prolonged sedentary behavior produces adverse effects on the cardiovascular system, involving cardiac function, stroke volume, cardiac output, heart rate, thromboembolic events [14,15], and glucose intolerance [16]. Taken together, these findings indicate that population-level increases in PA and reductions in sedentary behavior could be vital in reducing CVD incidence and mortality.

The importance of increasing PA and reducing sedentary behavior to prevent noncommunicable diseases has been highlighted by the Chief Medical Officers of England, Scotland, Wales, and Northern Ireland in the Start Active, Stay Active document [17]. The UK Department of Health currently recommends that adults should accumulate a minimum of 150 minutes of moderate-intensity PA each week to achieve tangible health benefits [17]. However, despite the well-known CVD-related and general benefits of PA, only a small percentage of adults meet these government guidelines, with >60% of men and >70% of women in England being insufficiently active to benefit their health [18].

At present, low levels of PA at the population level are partially addressed through primary care–based screening programs, including the NHS Health Checks program in the United Kingdom. Such health check programs screen for risk factors, including low PA, that contribute to the incidence of CVD and type 2 diabetes and then signpost individuals to appropriate interventions. For example, when low PA is identified during a health check, patients are signposted to interventions to increase PA. However, the PA interventions offered to individuals in these primary care settings rarely match the existing evidence on optimal methods for increasing PA behavior, and PA promotion is not a primary aim of primary care–based screening programs. Although PA prescription and advice giving are the most used methods of promoting PA in primary care settings [19], there is considerable evidence that these methods do not create lasting increases in PA [20,21], with up to 26 people needing to receive PA advice to meet the recommended levels of PA 6 months later [22]. In addition, even these minimal advice-giving approaches to PA promotion are only delivered to a small percentage of individuals for whom PA interventions are warranted [23,24], meaning that existing PA promotion interventions in primary care are suboptimal and delivered too infrequently.

The problems with PA promotion in practice may reflect deficits in the knowledge, skills, or motivation levels of health care providers (HCPs) regarding the delivery of effective behavior change interventions. Existing evidence indicates that interventions to increase PA are most likely to succeed when they include behavior change techniques (BCTs) derived from the Self-Regulation Theory [25], including self-monitoring, goal setting, action planning, feedback, and problem-solving. However, many HCPs working in primary care cite a lack of adequate training on how to deliver such self-regulatory interventions as a barrier to implementation [26]. Furthermore, HCPs may face additional barriers that limit the extent to which they offer PA promotion interventions to patients. For example, they may not be motivated to deliver PA promotion interventions in the first place; they may think that PA promotion is unimportant or not part of their role; or they may perceive barriers or conflicting priorities that prevent them from promoting PA, even if they are motivated to do so [27]. Helping HCPs overcome these barriers is crucial to ensuring that HCPs can deliver effective PA interventions in primary care.

As research from high-income countries indicates that approximately 70% to 80% of adults visit their general practice at least once a year [28] and that many individuals are willing to discuss changes in health behaviors with their primary health care professionals [29], primary care offers great potential for population-level PA promotion. To make full use of this reach, primary care HCPs need to have the necessary knowledge and skills to effectively target PA, be sufficiently motivated to do so, have sufficient self-efficacy for promoting PA, and have evidence-based tools at their disposal to assist patients as they attempt to increase their PA behavior. To this end, our group developed the Movement as Medicine for CVD Prevention (MaMCVD) program: a 2-tiered suite of behavior change interventions that aim to increase PA and thereby mitigate CVD risk.

Briefly, MaMCVD is a new PA promotion care pathway that comprises 2 theory-based behavior change interventions. The first intervention was delivered to health care professionals (eg, general practitioners [GPs], practice nurses, health care assistants, and health improvement practitioners) and is based on the Health Action Process Approach (HAPA) [30] and Self-Determination Theory (SDT) [31]. It aims to foster HCPs’ self-efficacy and motivation for promoting PA in clinical practice and equip them with the knowledge and skills necessary to deliver behavior change interventions beyond traditional advice-giving and prescriptive approaches. The second intervention was delivered to patients and is based on the HAPA [30], SDT [31], and Self-Regulation Theory [32]. The patient intervention included techniques derived from motivational interviewing (MI) to increase patients’ motivation for behavior change during primary care consultations and provided patients with a set of web-based tools with which they could self-regulate their efforts to become more physically active.

This study aims to examine the feasibility of MaMCVD as a PA promotion pathway in primary care settings, establish rates of recruitment and retention, and provide effect size estimates to potentially inform power calculations for a definitive randomized controlled trial (RCT). These aims are in line with phases 1 and 2 of the Medical Research Council framework for the development and evaluation of complex interventions to improve health [33].

This nonrandomized pilot feasibility study aimed to examine a newly developed PA care pathway for individuals with an increased risk of developing CVD and involved data collection at both HCP and patient levels. The protocol was registered (ISRCTN14582348) and received approval from the Newcastle and North Tyneside 1 Research Ethics Committee (reference 14/ES/0032). All methods were performed in accordance with the relevant guidelines and regulations. Recruitment for the study began in June 2014, and the final study data were collected in May 2015.

The original protocol for this study specified an RCT; however, this changed to a nonrandomized trial because of difficulties in recruiting primary care organizations. In addition, the length of the study was shortened, and the primary outcomes were altered to better reflect the nature of this study as a feasibility trial. Full details of these changes are available in the trial registration record (ISRCTN14582348).

The development of the MaMCVD primary care PA promotion pathway was informed by our group’s previous work in developing a primary care PA promotion pathway for people with type 2 diabetes [37]. The core of MaMCVD was an integrated website with separate areas for HCPs and their patients, described in detail in the following sections.

Paired t tests (2-tailed) in SPSS (SPSS Inc) were used to compare the baseline and follow-up levels of HCP outcomes. Multiple imputation (i=5) was undertaken to account for missing patient data at the 3-month follow-up. The imputation model used baseline demographic information and baseline and follow-up data of all study outcomes to predict the missing follow-up values of all study outcomes. Paired t tests were used to compare baseline and follow-up levels of patient outcomes in the pooled intention-to-treat data set. Cohen d effect sizes, and 95% CIs were also calculated to estimate the potential efficacy of the interventions and inform the sample size and power calculations for subsequent testing of this intervention [71]. Qualitative data from interviews with HCPs were examined using content analysis.

A favorable ethical opinion was granted by Newcastle and North Tyneside 1 NHS Research Ethics Committee (reference 14/ES/0032). Informed consent was obtained from all research participants.

The data sets generated and analyzed during this study are not publicly available, as ethical approval for the sharing of data was not sought or obtained; however, these are available from the corresponding author on reasonable request.

A total of 128 primary care organizations in the Northeast of England were approached for participation in the study. Of the 128 approached organizations, 5 (3.9%) primary care organizations (ie, n=4, 80% general practices and n=1, 20% community health organization) were willing to participate in the study. However, 50% (2/4) of the general practices dropped out of the study because of staff turnover and lack of available resources to administer the study.

Within the 3 remaining primary care organizations, recruitment packs were mailed to 827 patients. Of these 827 patients, 84 (10.2%) provided informed consent to participate in the study, with 83 (10%) subsequently completing baseline measures. Recruitment rates were 7%, 11%, and 12% across the 3 included primary care organizations. Detailed information on the flow of primary care organizations, HCPs, and individuals throughout the study can be found in Figure 1.

A total of 11 HCPs completed the web-based course and postcourse questionnaires. Of these 11 HCPs, 7 (64%) were female, with a mean age of 38.8 (SD 9.9; range 28-52) years. Enrolled HCPs were PA specialists (5/11, 46%), practice nurses (3/11, 27%), health care assistants (2/11, 18%), and GPs (1/11, 9%) and had an average of 7.1 years in their current role. Of the 11 HCPs, 7 (64%) had previously received some form of training in behavior change methods or BCTs.

After completing the web-based course, the HCPs reported increases in self-efficacy for promoting PA in practice and for having concrete plans of how and where to promote PA in practice. Smaller increases in habit were also reported, whereas reported changes in attitudes toward PA promotion, intention for PA promotion, and goal conflict were negligible. Quantitative results for the HCP outcomes are presented in Table 4.

Qualitative interviews with HCPs indicated broad acceptability of the web-based course, with an appreciation for the demonstration videos of MI techniques and several of the interactive educational elements. Some HCPs were unable to access the web-based course from the computer workstations in their primary care practice because of outdated browsers still being in use. In these cases, HCPs completed the web-based course using their home computer or tablet.

Of the 83 patients who provided data at baseline, 44 (53%) were female, and the average age was 57.5 (SD 10.2) years; 62 (75%) were married or cohabitating, and 41 (49%) were in part- or full-time employment, whereas 40 (48%) were unemployed or retired. Of the 83 patients, 15 (18%) had not completed high school or equivalent vocational education, whereas 25 (30%) had completed a university degree. Approximately 70% (58/83) of participants completed the 3-month follow-up measures, and there were no significant differences between patients who dropped out of the study and those who completed both baseline and follow-up assessments.

Analyses of intention-to-treat data revealed a significant increase of 9.6 minutes per day in objectively measured MVPA from baseline to follow-up (effect size Cohen d=0.77). Participants also reported favorable increases in self-reported total MVPA and MVPA during leisure time, as well as small reductions in sedentary behavior, all with effect sizes >Cohen d=0.44. Of the psychological variables assessed, patients reported favorable changes in self-efficacy for PA, PA action planning, action control, and intrinsic motivation for PA, with effect sizes between Cohen d=0.46 and Cohen d=0.60. The effect sizes for most of the other outcomes were <Cohen d=0.30. Table 5 provides for the means, SDs, and effect sizes with CIs.

Of the 75 patients who attended a face-to-face session and received log-in credentials for the patient website, 35 (47%) logged in at least once. Among these 35 individuals, 7 (20%) logged into the system weekly for at least 10 consecutive weeks, and the mean number of log-ins during the 3-month study period was 19.6 (mean log-ins per week 1.5). The most active user during the study period logged in 156 times or an average of 12 times per week. Further data on the use of individual components of the web-based intervention are presented in Table 6.

This nonrandomized pilot study assessed the feasibility of MaMCVD, a suite of 2 behavior change interventions for HCPs and patients to promote PA in primary care settings. MaMCVD was designed to provide HCPs with the skills required to increase motivation for PA among patients with CVD risk and help them address common barriers to promoting PA in primary care settings. In addition, it aimed to offer patients a set of theory- and evidence-based tools that they could use to self-regulate their efforts toward increasing PA and reducing CVD risk.

Among the patients approached for participation in this study, recruitment and retention rates were in line with expectations; however, recruitment of primary care organizations fell below expectations. This is attributable to several factors, including a major reorganization of primary care service delivery within the NHS just before the commencement of this study. In 2013, primary care trusts were disbanded and replaced by clinical commissioning groups, which created uncertainty as to whether and how participation in research studies would be reimbursed [72]. In addition, as this study was funded by a local (as opposed to a national or international) organization, it was not eligible for adoption by the National Institute for Health Research (NIHR) clinical research network [73]. Adoption to the NIHR portfolio would have provided additional financial compensation and logistical assistance to primary care organizations for taking part [74] beyond the CPD training and logistical assistance provided in this study. During recruitment, conversations with research leads from several primary care organizations indicated insufficient compensation and a lack of available staff as the most common reasons for nonparticipation in the study. Identifying additional compensation possibilities for practices and ensuring study adoption by primary care research networks will be key to improving the recruitment of primary care organizations when testing MaMCVD in a larger RCT.

The 30% rate of patient dropout in this study is similar to that reported in studies testing other (web-based) PA interventions among individuals with chronic diseases [75]. Although we attempted to gather information about the reasons why participants dropped out of the trial by mailing a postcard to those who did, none of these postcards were returned to the study team, and we were unable to gather such information. In any full-scale trial of MaMCVD, it might be worth including efforts to improve study retention, especially if outcomes are examined with a longer follow-up period. This could potentially include offering patients a choice of intervention components based on their preference (eg, web only, face-to-face only, or both options). For HCPs, the potential perceived burden of both receiving and delivering an intervention within the same study, as well as the share of total working time study-related obligations take up, should also be considered.

The web-based tools that constituted part of the MaMCVD intervention for patients were not used by all the participants. Approximately half of the participants who attended an MI session logged into the patient website, which is somewhat lower than the uptake reported in other internet-based PA interventions [76]. This could potentially be attributed to whether and how a patient’s HCP introduced the website. As the study did not include any checks of whether patients received website log-in credentials from their HCP, it is possible that HCPs may have forgotten to deliver these to some patients. This element of intervention fidelity should be assessed in any further rollout of MaMCVD. In addition, audio recordings of MI sessions early in the trial revealed that some HCPs misinformed patients about the purpose of the patient website, referring to it simply as a place where patients could obtain additional information about PA. As much of the content of face-to-face MI sessions was already informational, patients may not have been interested in receiving even more information and may therefore have avoided logging in. In a future rollout of the MaMCVD intervention, the role of HCPs in referring patients to the web-based motivational and self-regulatory tools for patients will be emphasized more concretely, and sufficient patient numbers should be included to allow for investigating relationships between the use of web-based tools and PA outcomes. In addition, in future studies, the fidelity of MI sessions delivered to patients should be investigated as a potential moderator of subsequent engagement with web-based self-regulation tools and overall intervention effectiveness.

Following completion of the web-based course, HCPs participating in this study self-reported considerable increases in self-efficacy for promoting PA in practice. This indicates that the web-based course helped HCPs become more confident that they could overcome the common barriers to promoting PA in practice. The intervention also led to moderate increases in planning and habits for PA promotion, meaning that PA promotion became somewhat more routine for HCPs after they completed the web-based course. As self-efficacy, planning, and habits have previously been shown to be strong predictors of other preventive clinical behaviors (eg, HCP behaviors in diabetes care) [77], it is reasonable to assume that HCPs who followed this web-based course might be more likely to promote PA to their patients outside the context of this study. HCPs reported no changes in their attitudes or intentions to promote PA, perhaps because of ceiling effects from the high baseline levels of these variables. Taken together, the effect sizes obtained for increases in self-efficacy and habit indicate that the HCP-facing MaMCVD intervention could help to improve PA promotion in practice. However, given the small sample size, potential bias of self-report measures, and lack of a control group in this study, these results should be interpreted with caution. Therefore, future studies may wish to investigate whether changes in self-efficacy, planning, and habit translate into objective changes in HCPs’ in-session PA promotion behaviors.

Between the baseline and posttreatment time points, objectively assessed MVPA increased by nearly 10 minutes per day among patients with an adequate accelerometer wear time. Subjective measures of PA corroborated these results, with the total self-reported MVPA increasing by approximately 45 minutes per week after the intervention. These 3-month effects on objective and subjective levels of MVPA are similar in magnitude to those obtained from primary care–delivered PA promotion interventions in general [78]. However, it should be noted that the sample of participants in this study was already highly active at baseline, with >80 minutes of objectively assessed MVPA per day. Future testing of MaMCVD should try to overcome this self-selection bias and recruit a less active sample to maximize the potential impact on CVD risk.

In addition to increases in PA, participants reported beneficial changes in important motivational and volitional predictors of PA behavior. Intention for PA, the seminal predictor of PA in many behavioral theories, also increased from baseline to posttreatment. This represents an important outcome of the intervention as, according to the Rubicon model [79], solid intentions to perform a behavior are prerequisites for engaging in self-regulatory behaviors, such as goal setting and action planning. Theoretically, this fits well with the increased use of action planning for PA participants reported between baseline and posttreatment time points. Although self-reported action planning did increase, this was not borne out in the objective data on patients’ use of web-based self-regulatory tools, wherein only 11 participants set a goal or made an action plan. Therefore, self-reported increases in action planning more likely reflect individuals’ mental conceptualizations of future PA-related actions, as opposed to any specific written action plans.

Patients also reported increases in action control for PA, which entails a greater focus on efforts to increase PA behavior. This, too, is an important finding, as self-regulatory efforts to change behavior require increased attention to the target behavior to be effective [32]. A recent meta-analysis indicated that coupling these self-regulatory techniques with opportunities for patients to engage in supervised PA could increase these effects on intention [80], a possible add-on for future iterations of the MaMCVD intervention.

Although most intervention effects occurred in the intended direction, several did not. Introjected motivation and external motivation, controlled forms of motivation that may undermine long-term PA participation [81], both increased between baseline and posttreatment time points. Participants also reported less concern about developing CVD after treatment than they did at baseline, as well as a decrease in the extent to which they thought PA could prevent CVD incidence. Although the magnitudes of these potential adverse effects were small, any subsequent iterations of MaMCVD should seek to mitigate these effects. This could be done by emphasizing and supporting participants’ autonomy in their PA journeys by ensuring that participants engage with the educational content about the links between PA and CVD incidence and by clearly communicating CVD risk.

This pilot study of the Movement as Medicine suite of behavior change interventions for HCPs and patients represents an important step in meeting the needs of HCPs tasked with promoting PA in primary care settings. The interventions were specifically designed to address HCPs’ barriers to PA promotion in practice and were designed in line with theory and evidence on how to increase motivation and PA among patients. In addition, the study used objective measurements of PA to overcome social desirability and other response biases in PA intervention studies [82].

Despite these strengths, some limitations should be considered. First, no clinical outcomes were assessed as part of this study, as we did not expect changes in these within the short 3-month study period. The possibilities of sampling bias (eg, more motivated participants enrolling in the study), ceiling effects, and response desirability bias on subjective outcome measures should also be considered as limitations of this study. Common biomedical measures associated with CVD risk, such as blood pressure, cholesterol, weight, and waist circumference, as well as objective measures where possible, should be included in future tests of this intervention. Second, the recruitment of primary care organizations fell below anticipated targets, which limits the ability to generalize patient retention rates across sites. This low uptake among primary care organizations also led to deviations from the published trial registration and a reduction in the scope of the study from an RCT to a nonrandomized pilot feasibility study. As a result, patient recruitment too fell below the initial sample size targets, meaning that we were unable to investigate the extent to which changes in the theoretical determinants of PA and patient engagement with the MaMCVD intervention contributed to levels of PA at the end of the intervention in this study. More financial resources and adoption to the NIHR portfolio would likely improve the uptake of the intervention and allow for testing of these predictive hypotheses. Finally, as this was a single-group pilot study, the reported effect sizes for changes in the variables under study should not be interpreted as causal or as estimates of the true effects. Rather, these effect size estimates should be used to calculate the sample size needed to demonstrate between-group efficacy in a larger RCT.

The MaMCVD program sought to provide primary HCPs with new skills to promote PA during brief primary care consultations and, more broadly, to offer a follow-on PA treatment pathway for individuals with elevated CVD risk. The intervention improved self-efficacy for PA promotion among HCPs and may have had knock-on effects on their day-to-day practice. For patients participating in the study, MaMCVD offered web-based tools with which they could motivate themselves and self-regulate their efforts at PA behavior change. The intervention led to increases in the self-reported determinants of PA, including self-efficacy, intrinsic motivation, action planning, and action control, as well as increases in objectively assessed MVPA. Although these preliminary results indicate support for the program, the findings should be tempered, given the small sample size, absence of a control group, and use of self-report measures. In addition, feasibility problems around the uptake of the program by primary care organizations and national health research bodies need to be addressed before any broader rollout or testing of the program can take place. The results obtained here will be useful in improving these aspects of the MaMCVD program and can inform subsequent testing of the intervention in an RCT.