Evidence has been lacking for the long term effectiveness of motivational interviewing (MI) on behavioural and biomedical outcomes for overweight or obese people. The current study contributes to a gap in the literature on the sustained effects of an MI intervention on weight loss, physical activity, and CVD risk factors among primary care patients at 12-months post-intervention. The main aim of the study was to assess whether changes in weight, BMI, physical activity, blood pressure, and cholesterol levels within the MI intervention group were maintained one-year later. The second aim was to explore the effect of counselling session attendance (i.e., dose) on maintenance outcomes. The final aim was to examine the effects of motivational interviewing on outcomes for sub-groups presenting with specific CVD risk factors at baseline. This is the first study to show that an MI intervention, delivered in the primary care setting, can contribute to a reduction in cholesterol and a significant increase in walking at both 6- and 12-months post-intervention, compared to an information-only group, for a sample with high levels of overweight or obesity. Only two previous studies [9, 43] have reported the effects of MI interventions for weight loss beyond 6-months and both had substantial limitations curtailing their validity as long-term follow-up studies. The present study therefore offers important information on the potential effects that can be achieved by a targeted patient-centred lifestyle intervention delivered in a primary care setting.
In relation to the main aim to assess whether changes in weight, BMI, physical activity, and CVD risk factors within the MI intervention group were maintained at 12 months post-intervention, we found significant increases in walking in the MI intervention group that were maintained at 12 months. The MI intervention group also exhibited significant reductions in cholesterol during the intervention and this reduction was maintained 12-months later. In contrast, the minimal intervention group significantly increased cholesterol between the end of the intervention and 12-month follow-up. According to Ketola et al. , a clinically significant change in total cholesterol is classified as a decrease of 0.5 mmol/l. Therefore, the mean difference in our study of −0.16 mmol/l between groups at 18-months cannot be interpreted at clinically significant according to this index. However, the magnitude of the change found is comparable to changes reported in other studies that have examined the effectiveness of either individual lifestyle interventions [57, 58] or of MI interventions on cholesterol [59, 60]. There was also a significant increase in BMI from baseline to 18-months in the minimal intervention group but no significant change in BMI across follow-up for the MI intervention group. While the minimal intervention group increased BMI throughout the study, BMI was initially reduced in the MI group, but then reverted to baseline levels at 18 months.
In relation to the second aim, there were no significant effects for dose of MI on outcome measures from baseline to 18-months, with the exception of triglycerides, where the more sessions attended, the greater the improvement in triglycerides. Since any changes in the outcome variables were not associated with the number of sessions attended, these effects appear to be in response to a relatively low dose of MI. This indicates that a low-intensity MI can lead to significant improvements in cholesterol and physical activity up to one year following the intervention. Having a sample of overweight and mostly obese patients may have helped to give the study greater scope to detect such effects, compared with other studies that have mainly involved non-clinical populations, and this is one of the strengths of the present study.
In relation to the final aim involving sub-group analysis for those who were obese, hypertensive, hypercholesterolemic, and inactive at baseline, we found statistically significant differences in BMI at 6- and 12 months post-intervention for the MI intervention group, a trend that was not evident in the minimal intervention group. These findings are consistent with the effects of MI on weight reported in a recent meta-analysis of MI interventions targeting weight loss . It should also be noted that the 18-month follow-up data were collected 12-months after cessation of the intervention in the current study. Given that the average attendance to the MI sessions was 2.00, the current study provides good evidence that a minimal intervention is effective in bringing about meaningful changes in some of the outcome variables. To speculate, the inclusion of relatively modest, low-cost ‘booster’ sessions may further bolster the effectiveness of the current intervention in the long term [61, 62].
The ‘rebound’ effect, evident in the present study and commonly reported in behavioural intervention studies in which short-term changes in behaviour are lost at long-term follow-up [10, 11], suggests that future research needs to concentrate on promoting the maintenance of behaviour change. There is evidence to suggest that maintenance of behavioural changes is more likely in trials that involved face-to-face contact than those that do not . In addition, reviews on the maintenance of weight loss provide evidence to suggest that longer duration interventions (>6 months), that incorporate face-to-face contact are more effective in facilitating sustained weight loss [10, 64]. The effects of MI could well be enhanced via the inclusion of booster sessions, beyond 6 months [61, 62].
Also, reviews have revealed that both number and duration of MI sessions are related to behaviour change. For example, in studies that have used two MI sessions with 60 minutes per session, 81% of studies reported an effect [13, 28]. The combined findings of the present study and Hardcastle et al.  showed a clear dose–response pattern up to 6-months, but these were all but lost by 12 months post-intervention. Further research is needed to explore the effectiveness of follow-up prompts (i.e., brief contact following the main intervention) on the maintenance of weight loss behavioural changes. Interventions that involve prolonged contact may be more effective than those based on high frequency contact . The current study contributes to an insufficient body of literature on the maintenance of behavioural outcomes in physical activity and/or dietary intervention trials. Indeed, the recent review by Fjeldsoe  found that less than 20% of RCT’s published since 2000 have reported on behavioural maintenance and this was using a very conservative definition of maintenance (i.e., a follow-up period of 3 months or longer) . Of these studies, only one that targeted both physical activity and diet included a 12-month follow-up, similar to the current study ; only one study recruited primary care patients  and almost half recruited healthy adults [68–70]. Furthermore, only two of the studies to meet criteria for inclusion in the review were conducted within the UK; both of which had relatively brief follow-ups of 3 and 6 months respectively [71, 72].
Limitations, avenues for future research, and conclusions
Despite the promising findings, the present study has some limitations. One limitation of the study concerns the low participation rate (28%). However, this was not entirely surprising given the opt-in procedure used and that among those at risk of CHD, many are not ready to change important lifestyle behaviours and therefore are unlikely to volunteer for such a trial. Nevertheless, although the patients who entered the trial might be more motivated for lifestyle change, they still needed to change and were recruited on the basis that they were at risk of CVD. As such we can be confident about the transferability of our findings to other patients at risk of CVD, albeit those that are more ready to change their lifestyle. Another limitation concerns the relatively low uptake of the intervention. Limited resources meant that we were unable to follow-up those participants with a reminder. However, this meant that the intervention had strong ecological validity as interventions carried out in practice are unlikely to have the resources or capacity to include extensive prompts or incentives to promote attendance. This limitation notwithstanding, we were still able to find significant effects of our intervention on the primary behavioural and biomedical dependent variables, making the present findings a relatively conservative estimate of the efficacy of such an intervention. A further limitation is that other important biomedical markers such as insulin and HbA1C were not measured. Such data would have been valuable to link the findings to similar literature on behavioural interventions involving pre-diabetic and diabetic populations. Similarly, measures of skinfold and other body composition outcomes would have provided further insights into how the MI intervention impacted on CVD risk. Such measures were considered, but were not undertaken, to avoid burdening the participant. The MI intervention was designed to be pragmatic and delivered in an ecologically-valid primary-care context. A second potentially limiting aspect of the study was the availability of resources to conduct a thorough process evaluation to determine: (a) fidelity of the intervention (including maintaining a patient-focus within the ‘spirit’ of MI, and coding of each session) and (b) which components of the intervention were most effective. During training of the intervention practitioners, sessions were tape recorded, the data coded and analysed, and feedback provided on the delivery of the MI intervention, but resource implications prevented a complete analysis. Future research should measure the most likely social psychological and motivational predictors of behaviour change hypothesized to mediate the effects of MI interventions [73, 74]. This may include measures of self-efficacy [28, 75, 76], social support , autonomous forms of motivation from SDT , attitudes and perceived behavioural control from the Theory of Planned Behavior [79, 80], and, motivational readiness from the Transtheoretical model [41, 42] as important candidates to help explain the active ingredients of MI.
Nevertheless, a qualitative evaluation of this project  seeking to understand why some participants responded well to the intervention and maintained their reduced weight whilst others failed to change or failed to maintain change was undertaken. The qualitative evaluation pointed to differences in motivation and self-regulation. The lack of self-control identified by several participants was consistent with individual-difference theories of motivation which recognise that some individuals have lower capacity for sustained behaviour change and for resisting dominant responses. For example, recent work has demonstrated that people vary in their capacity to exercise self-control and this affects their ability to adhere to sustained and effortful behaviour change necessary to bring about weight loss through diet and physical activity [82–84]. Such individuals have low levels of self-control and are less likely to maintain behaviour change. However, such individuals are likely to benefit the most from MI interventions that are aimed at promoting autonomy, confidence, planning, and self-regulation. Indeed, research by Webber et al.  has shown that those with high levels of controlled motivation for weight loss at baseline lost significantly more weight in the MI group compared to those assigned to a standard behavioural weight loss intervention. Future work should take into account such individual differences and how to support those with lower levels of self-control. More research is needed to find the optimal number of consultations to promote autonomous forms of motivation and sustainable weight loss.
A further limitation of the study was the reliance on self-reported measures of physical activity and dietary behaviour . The use of accelerometers and emerging dietary recording technologies could have provided a more objective understanding of behavioural changes associated with the intervention. Certainly, the significant reduction in fat intake self- reported by the minimal intervention group across time was unexpected and contrary to our expectations. One explanation for this difference may be that patients were central to what was discussed during counselling sessions and, perhaps feeling well informed (from general media attention) about fat intake, tended not to focus conversation on how to change this aspect of their lifestyle. Another explanation is that following the motivational interviewing session, participants in the intervention group became more realistic in their self-reporting of fat intake whereas those in the minimal intervention group may have underestimated their overall fat intake. Despite this unexpected finding, it is important to reinforce the clinically-relevant biomedical outcomes related to lipids found in the current study. Cholesterol was significantly improved for those in the intervention group compared to the minimal intervention group for those identified with hypercholesterolemia at baseline. In addition, for those identified as obese, there was a significant increase in BMI across time among patients in the minimal intervention group.
Finally, we did not set out to determine a full the cost-benefit analysis of the intervention. However, we recognise the importance of identifying the cost of such an intervention. The estimated cost of delivering the intervention was based on the hours of staff time (16 hours/week for 40 weeks) relative to the mean number of counselling sessions attended (n = 2) by all those randomised into the counselling group who attended the baseline assessment (n = 203). On average, each patient received one hour of counselling. Therefore, the average cost of delivering the intervention per patient was between £47 and £63 (depending on the expertise and experience of the practitioner. The cost is relatively cheap, given the potential health gains possible from using such an approach. Furthermore, given that motivational interviewing tends to involve around substantially less face-to-face contact time with clients compared to traditional programmes and appears to be equally effective, as evidenced by three meta-analytic reviews [87–89], the lower time commitment involved makes MI a potentially cost-effective intervention especially when resources are limited .
The current study contributes to a gap in the literature on the sustained effects of motivational interviewing on weight loss, physical activity and CVD risk factors a year following intervention. The MI intervention group sustained improvements in both walking and cholesterol. Also, for those with obesity and hypercholesterolemia at baseline, the MI intervention had a significant net favourable effect on BMI and cholesterol levels respectively.