Integrated motivational interviewing and cognitive behaviour therapy can increase physical activity and improve health of adult ambulatory care patients in a regional hospital: the Healthy4U randomised controlled trial

Design

The Healthy4U study was a single-blind randomised controlled trial designed and reported in line with the CONSORT recommendations for reporting (Fig. 1) [28]. The trial was registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12616001331426) prior to patient recruitment.

Participants

Participants were recruited from an ambulatory, secondary care clinic in a major tertiary hospital in regional Victoria. These outpatients receive medical care from specialities in general surgery, orthopaedic surgery and endocrinology after a referral from a GP. Throughout the recruitment phase of the study, recruitment flyers were available at the clinic and patients who were potentially interested in participating made direct contact with the research team using information provided on the flyer.

Participants were included if they were between 18 and 69 years, and reported being insufficiently physically active, defined as obtaining less than 150 min/week of moderate-to-vigorous physical activity (MVPA) [29]. A single item question “As a rule, do you do at least half an hour of moderate or vigorous exercise (such as walking or a sport) on five or more days of the week?” - was used to identify insufficiently physically active individuals [30]. The following exclusion criteria were applied: sufficiently physically active [29]; an existing medical condition that contraindicated PA (indicated by the Physical Activity Readiness Questionnaire); a diagnosis of diabetes; deaf/hearing impaired; disabling neurological disorder; severe mental illness such as psychosis; learning disability; dementia; registered blind; housebound or resident in nursing home; non-ambulant; pregnancy; advanced cancer.

Randomisation and allocation

Participants who fulfilled the inclusion criteria and consented to take part in the trial were randomly allocated to either the intervention or the control group based on a random number sequence produced by a computer generated program (randomizer.org). Assignments were prepared and sealed in sequentially numbered opaque envelopes. Assignment was made by opening the next envelope in the sequence, after the recruiter had determined eligibility for the study, participants had consented to take part, attendance at an education session was confirmed and baseline measurements were completed.

Procedure

Participants’ characteristics and outcome measures were recorded at baseline, after 3 months of intervention (post-intervention) and at 6 months (follow-up) by assessors blinded to the study group assignment. The extension of outcome measures from baseline to 6 months, which included a 3-month period where no contact with participants was made, was designed to investigate behaviour change maintenance using a previously accepted follow-up duration [31]. A recent systematic review highlighted the need for behaviour change interventions to distinguish between initial behaviour change and behaviour change maintenance [32].

Intervention

All enrolled participants attended an education session prior to group allocation. The education session was a facilitated learning session based around self-management and lifestyle modification and was carried out using a self-determination theory (SDT) framework [33]. SDT is a general theory of human motivation that defines motivation as “psychological energy directed at a particular goal” [34]. Individuals are more likely to engage in certain behaviours, physical activity for example, if they both value that behaviour, and have motivation for change [34]. SDT was used in this group setting to support, educate and motivate participants around positive lifestyle choices [33].

The intervention group completed a telephone-based, integrated MI-CBT intervention, delivered in eight 30-min sessions over 12 weeks. The intervention was delivered using an MI framework, where MI microskills (open-ended questions, affirmations, reflections and summaries) were used in all sessions to progress participants through the MI processes of change (engagement, focusing, evocation, and planning) [16]. Throughout intervention sessions 1 to 4, MI was predominantly delivered in isolation (i.e., without CBT), exploring participants’ feelings about change and evoking intentions to change [16]. Where MI was used alone, the person delivering the intervention refrained from discussing any specific change-oriented strategies, and instead focused on exploring participant feelings and specific ambivalence regarding barriers to physical activity [16]. In the subsequent sessions, the integrated MI-CBT phase, more specific focus was directed on the identified drivers of ambivalence and resistance, leading to the formulation of goal-directed action plans [16, 35]. The CBT treatment built upon a number of evidence-based protocols with adaptations to focus on the goal of change in PA [35]. The CBT component focused more explicitly on individual determinants of PA such as PA experiences, PA outcome expectations, and PA self-efficacy [35]. The CBT strategies, which included goal setting, action planning, self-monitoring, personal feedback and relapse prevention, were incorporated within this MI framework for supporting PA change and maintenance [36]. The intervention used the integration of MI with CBT in two ways: (i) the underlying spirit of MI was used as a foundational platform from which to conduct CBT, and (ii) during more action orientated sessions therapists could switch back to MI in response to identified markers of ambivalence or resistance [35, 36].

The intervention was delivered by an experienced allied health clinician trained in MI-CBT, including workshop attendances, and one-on-one coaching from an experienced practicing psychologist. The intervener’s proficiency in using motivational interviewing was confirmed via role-play sessions, one at the commencement and one at midpoint of the intervention. Proficiency was confirmed by an independent assessor using the validated Motivational Interviewing Integrity scale 3.1.1 [37]. The intervener’s proficiency in using motivational interviewing was rated as competent on the global clinician rating at both assessments. All participants enrolled into the control arm attended the education session. Apart from contact regarding follow-up outcome measures, participants of the control group received no further contact initiated by the research team.

Outcome measures

The primary outcome, MVPA (minutes/day) was assessed by accelerometry (wGT3X-BT; Actigraph, USA) during all waking hours over 7 consecutive days. PA was calculated using the manufacturers software (Actilife; Actigraph, USA) with cut points by Freedson Adult (1998) used to provide daily measures of MVPA (> 1951 cpm) [38]. Accelerometer wear time was based on activity counts per minute. Non-wear time was defined as 60 min or more of consecutive activity counts of zero, with a spike tolerance of 2 min and 100 cpm. Accelerometer data were considered valid if the accelerometer was worn > 10 h per day for at least 5 of 7 days including at least 1 weekend day [39]. The accelerometer was worn on a waist band, over the right hip. In adults, the hip-mounted ActiGraph has demonstrated high inter-device reliability (r = 0.98) and validity against indirect calorimetry (r = 0.56, p < 0.001) [40]. Participants used logbooks to report significant PA events (e.g., attending exercise class, going for a walk, heavy gardening) and periods of accelerometer non-wear. Participants returned the PA logbook along with the accelerometer within 48 h of the last accelerometer day. PA data were verified manually against the PA logbooks.

Anthropometric measures were taken objectively in accordance with International Standards for Anthropometric Assessment [41]. Waist circumference (WC) was measured to the nearest 0.1 cm using a rigid anthropometric measuring tape (Lufkin, US). Body mass was recorded to the nearest 0.1 kg using a calibrated scale (model 813; Seca, Germany). Free standing stature was recorded to the nearest 0.1 cm using a calibrated equipment with the participant barefoot (Portable stadiometer; Seca, Germany). Body mass index (BMI) was calculated by dividing body mass by the square of height. Self-efficacy to be physically active was measured using the physical activity self-efficacy survey [42]. The survey measures confidence related to undertaking physical activity over a continual timeframe with a higher score indicating a higher degree of self-efficacy. The survey has support for both its reliability and validity [42]. Health-related quality of life (HrQoL) was measured using the Medical Outcomes Study Short Form 12 Health Survey (SF-12) [43]. The SF-12 is a valid and reliable tool with published psychometric support [44]. A single index score on a scale of 0 to 1 was calculated for all participants, with a higher score indicating a more favourable health state [45].

The Australian type 2 diabetes risk assessment tool (AUSDRISK) was used to measure risk of type 2 diabetes [46]. This 12-item tool has been validated in a number of Australian studies [46]. The AUSDRISK includes questions on age, gender, waist circumference, and family history of diabetes. Potential scores range from 0 to 38 and relate to the probability of developing diabetes within the next 5 years [46]. For scores of 12–15, approximately one person in every 14 will develop diabetes [46]. For scores of 20 and above, approximately one person in every 3 will develop diabetes [46]. Demographic data were collected on participant postcode, employment status, smoking status and medical history.

Study size

In order to detect between-group differences of 30 ± 50 (mean ± SD) minutes, the standardized mean difference, or effect size required is 0.60 [47]. A sample size of 30 participants per arm was calculated to be sufficient to detect an effect size of 0.60 or greater, with the alpha set at 0.05, and the power set at 0.80. Protecting against a drop-out rate of 20% over the 6-month study duration, 36 participants were recruited and randomised into each arm.

Data analyses

Analyses were carried out using IBM SPSS Statistics for Windows (Version 23.0; IBM Corp., USA) and statistical significance was set at an alpha of 0.05. Data were assessed for normal distribution by Shapiro-Wilk’s [48]. Homogeneity of variances and covariances were assessed by Levene’s test and Box’s M test, respectively. Grouped data are presented as mean ± standard deviation. For the main analyses, a series of mixed-model ANOVAs (within: time; between: intervention) were used to assess the effects of the integrated MI-CBT intervention on each of the outcome variables separately. Mauchly’s test was consulted and Greenhouse–Geisser correction was applied if the assumption of sphericity was violated [48]. A significant interaction effect was interpreted to demonstrate that the change in dependent variables was influenced by intervention. Where data were in breach of Shapiro-Wilks test of normality, sensitivity analyses were performed. Data were explored for significant outliers and repeat sensitivity analyses were undertaken on data with outliers removed. Repeated sensitivity analyses provided no indication that the outliers had a significant effect on the outcome; therefore, all data were included in analyses.

A full intention-to-treat approach was used. For participants with missing data at 6-month follow-up (n = 2 in both groups), the last-observation-carried forward approach was adopted [49]. Repeat sensitivity analyses were undertaken on data with and without imputing the last-observation-carried forward value. The repeated sensitivity analyses provided no indication that the imputed values had a significant effect on the outcome.

Changes and maintenance in physical activity

The Healthy4U trial recruited participants deemed to be insufficiently active via self-report, and seeking to become more physically active. At baseline, daily MVPA was 31 ± 10 min, indicating moderate levels of PA. All participants completed their baseline measures after attending the education session. Interventions based on SDT have been shown to result in short-term increases in PA [50]. Therefore, it is likely that the education session contributed to moderate amounts of MVPA across all participants at baseline [50]. Despite the moderate level of PA observed at baseline, the intervention group significantly increased MVPA at 3 months (post-intervention) by a further 15.3 min/day (95%CI: 9.7 to 21.0 min/day), with similar MVPA recorded at follow-up (12.9 min/day; 95%CI: 6.5 to 19.5 min/day). These results indicate a positive change in behaviour that was maintained at 6 months. To put this change into perspective, an additional 15 min of PA is associated with a reduction in all-cause mortality of 4%, regardless of age [51]. In contrast, compared to baseline, the control group decreased MVPA at follow-up by 9.9 min/day (95%CI: -3.7 to -16.0 min/day). This reduction in MVPA in the control group is likely to reflect change from an elevated baseline value resulting from the initial education session. Decreases in PA have been shown to occur following the end of an exercise intervention, demonstrating the challenges of implementing prescriptive exercise interventions into the real world [52]. Studies have attempted to address this through the incorporation of self-monitoring strategies to prevent PA recidivism following the completion of exercise interventions [53]. A recent meta-analysis recommended that lifestyle interventions utilise self-help strategies for health related behaviour change [54]. The integration of MI and CBT provides strategies for participants to identify and overcome barriers to initiate and maintain behaviour change [20]. The observed changes in MVPA in the intervention group is suggestive of the effectiveness of integrated MI-CBT for PA change and maintenance.

Changes and maintenance in anthropometrics

Compared to control, integrated MI-CBT resulted in significant changes in anthropometric outcomes. From baseline to follow-up, body mass reduced by 2.7 kg (95%CI: -2.1 to -3.3 kg) in the intervention group. Although the mean reduction in body mass did not exceed a clinically significant value of 5% [55], weight loss over the duration of the study is noteworthy as most middle aged individuals continue to gain weight each year [56]. The average gain in body mass of 0.5 kg (95% CI: -0.3 to 1.0) in the control exemplified the normal pattern of middle-aged weight gain [56]. The 2.5 cm decrease in WC observed in the MI-CBT group (95%CI: -1.8 to -3.1 cm) is also promising, as the baseline value of mean WC suggest an increased risk of obesity related diseases [57]. There is limited evidence for what constitutes a minimally important change in WC; however, a change of between 1.8 and 4.1 cm has been proposed as a marker of maintained change [58]. Furthermore, high waist circumference is positively associated with higher mortality rates at all levels of BMI from 20 to 50 kg/m² [59]. A meta-regression analysis found that a 1 cm increase in WC can increase the relative risk of cardiovascular events by 2% [60]. The observed decrease in WC in this study is in contrast with longitudinal data on Australian adults, which reported an annualised increase in waist circumference, that did not slow over time [61]. The observed pattern of increased WC occurred with a simultaneous decrease in body mass over the period [61]. NHANES data also indicated that over time WC increased to a larger extent that expected, relative to changes in body mass over the same period [62]. These findings suggest that excess body weight over time is resulting in an increase in central adiposity, which is associated with a greater risk of cardiometabolic diseases [62]. Although recruitment into this study was based upon changing PA, not body composition, the positive changes in anthropometric measures are of clinical importance as these risk factors are strong indicators of metabolic dysfunction, and associated with development and worsening of cardiovascular disease and diabetes [63]. The changes in both WC and body mass found in this study may strengthen confidence in the MI-CBT intervention for change and maintenance in anthropometric measures.

Changes and maintenance in health-related outcomes

The average baseline scores for PA self-efficacy indicated that the sample had moderate belief in their ability to be physically active. Lack of confidence and self-belief are strongly associated with low rates of PA [64]. Despite having had higher levels of PA self-efficacy at baseline, the mean PA self-efficacy in the control group decreased at post-intervention and even further at follow-up, which diametrically opposed the trajectory of the intervention group. This increase in PA self-efficacy is a potential mediator for the improvement and maintenance in PA levels among the intervention group. The contrast in patterns of PA self-efficacy between the groups might be explained by the exposure of the intervention group to the MI-CBT treatment. MI-CBT strategies focus on increasing self-efficacy for behaviour change, as well as developing strategies for planning and relapse prevention [16, 20]. The changes in self-efficacy found in the study participants exhibited the same pattern of change as PA. The integrated MI-CBT group increased their self-efficacy and PA, while the control group demonstrated decreases in these outcomes.

Integrated MI-CBT resulted in small but significant changes in type 2 diabetes risk. The mean baseline AUSDRISK score of 14 indicates that participants are at high risk of developing type 2 diabetes within 5 years [46]. Early screening for 2 diabetes risk, and subsequent lifestyle modification reduces the risk of type 2 diabetes and other chronic diseases [65]. Research indicates that almost 40% of individuals with prediabetes, if left untreated, will progress to diabetes in 4 years [66]. Lifestyle interventions can decrease the percentage of those with prediabetes who go on to develop diabetes to 20% [66]. The observed increase in PA and reduction in WC in the intervention group contributed to a reduction in risk score, whereas no change in risk score was observed in the control group. At follow-up the intervention group had a mean risk score of 13, which still indicates a high risk developing type 2 diabetes [46]. The long-term maintenance of behaviour change is important for the effectiveness of lifestyle interventions to prevent type 2 diabetes.

The integrated MI-CBT intervention also resulted in small but significant changes between the groups for HrQoL. A consistent positive association has been demonstrated between PA and HrQoL [67]. Changes in HrQoL in the intervention group were significant at follow-up, but not at post-intervention. Changes in multidimensional quality of life measures have been shown to be less responsive than measures of specific patient outcomes, in this example PA change, particularly where interventions are aimed at achieving a particular outcome [68]. This might account for the slower degree of change in HrQoL exhibited by the intervention group [68].

Strengths

This study was unique in that it enrolled participants from secondary care in a public hospital, integrating preventative health into secondary care. Patients presenting to secondary care have higher rates of chronic disease than the general population, and the targeting of high risk groups is essential to address the rising prevalence of chronic diseases in Australia [69]. While hospital patients have expressed a desire for health behaviour interventions [5], research from practicing hospitals doctors indicate that they do not have sufficient time to spend giving preventive advice to patients [27]. The Healthy4U intervention was implemented to address this identified gap, to supplement clinical practice in secondary care, facilitating a process where clinicians under time duress were able to direct patients into a health behaviour intervention. With rates of lifestyle related disease projected to continue to rise in the future it is important to develop and evaluate innovative ways to address this concern [70]. For a regional hospital the delivery of the health coaching via telephone was important as it can extend reach to both geographically and socially disadvantaged areas, which commonly have higher risk of chronic diseases [71]. In RCT’s on behaviour change, the maintenance of physical activity behaviour change is not often reported, influencing the generalisability of findings [31]. The Healthy4U trial was purposely designed to assess the effect of the intervention for behaviour change and maintenance by extending outcome measures to 6 months from baseline. The use of objectively measured PA at all time points was a considerable strength of the study. Objective measures offer more precise estimates of activity intensity while removing many of the issues associated with participant recall and response bias [72]. A recent meta-analysis on MI for PA change demonstrated that the effect size of the intervention was smaller in trials using objective measures, relative to self-reported data [73]. The objectively measured changes in PA strengthen the confidence in the findings [72].

The participant retention rate in this study was high, with only 4 participants (2 from both groups) lost at follow-up. All participants were required to attend the education session, which was designed to motivate behavioural change. Mandatory attendance at the education session might have positively influenced participation in the study for individuals allocated to both groups. This study used a Manual of Operating Procedures (MOP) to facilitate consistency in protocol implementation and data collection across participants. The MOP transformed the study protocol into a guideline describing the procedure for initial and subsequent contacts with participants. The MOP included standardised procedures for reminding participants of their upcoming commitments in relation to attendance for assessments and accelerometer use, which might partially explain the strong compliance rates.

Intervention adherence rate was also high, with 100% of participants receiving all eight sessions of integrated MI-CBT. Intervention-led health behaviour change relies on mutual understanding and trust between the intervener and the participant, which can only emerge when sufficient time is given [74]. The individual delivering the intervention was confirmed as being fidelity proficient.

Limitations

Although the target number of participants was modest, the sample size was large enough to detect significant differences between groups in all outcome measures. Recruitment of volunteers into this study meant that all participants were already interested in becoming more active. Although this might limit the transferability of findings to all community-dwelling adults, it does not influence interpretation about the effectiveness of the intervention when compared against the control, due to the robust nature of the RCT study design. As the study was confined to a 6-month timeframe, it is not clear if improvements were sustained beyond that measurement point. Nevertheless the intervention resulted in increases in MVPA that were sustained at follow-up, which is indicative of behaviour change maintenance [31], and the exhibited patterns of behaviour change found in the outcomes were not indicative of recidivism following the intervention completion [75]. Lastly, the broad generalizability of these findings might be difficult because the study was conducted in one regional location and the majority of participants were female and obese.