Motivations to exercise in young men following a residential weight loss programme conducted in National Service - a mixed methods study

The qualitative component of our study was conducted from March to August 2017 in military installations across Singapore. Our research team comprised one female Masters of Public Health (MPH) student and one male military physician pursuing a PhD. Both researchers had experience in qualitative methods, neither had a history of excess body weight. Analysis was guided and supported by two senior co-authors with expertise in qualitative methods and physical activity promotion.

To elicit the instructors’ perspectives we scheduled three separate focus groups involving a convenience sample of five to nine instructors each. Eligible discussants would have instructed in at least one full run of the residential programme. The focus groups were conducted onsite in a meeting room each lasting approximately on hour. We maintained the integrity of groups, meaning military leaders, fitness trainers and ancillary staff would not mix. The facilitator gave a short three-minute overview highlighting the problem of weight gain following the residential programme. Using a topic guide developed by our team (see Supplementary Table 1) the facilitator subsequently prompted participants to share their thoughts on what could be done differently to prevent relapse. A second member of the research team observed and took hand-written notes. The facilitator recorded key issues on a large sheet of poster paper for all participants to review. The session was concluded with a participatory learning and action exercise where participants were allowed to rank order and confirm key priorities in weight loss management using a ten-seed technique, which has been shown to stimulate participation and foster group consensus [39]. Audio-recordings were not made in order to promote more open sharing by participants who were all employees of the military.

To explore participant experiences further, we recruited a second convenience sample of 20 individuals who were randomly selected from the list of trainees who had completed the programme in December 2016. Our aim was to interview participants in-depth 6 months after completion of the programme within a 6-week period. Starting in May 2017 we wrote to each of their superior officers to schedule interviews that would be conducted behind closed doors near the place of work. Participation was voluntary and informed consent was required. No incentives were provided for participation in the interviews.

Key priorities identified by the focus groups informed the topic guide (see Supplementary Table 2) used in the in-depth interviews. The introduction to the study started with the process of obtaining informed consent. Subsequently the participant and the two researchers conducted a short ice-breaker comprising a brief self-introduction and an exercise of taking each other’s weight on a portable scale. Thereafter one researcher commenced the interview while the second team member observed and compiled field notes. All interviews were recorded on an audio device. The topic guide was structured to cover the interviewees’ lifestyles with an emphasis on physical activity and diet before, during, and after the residential programme. Participants were encouraged to reflect on their school, work and home environments. Before concluding the interview, participants were asked if they had any questions or concerns. After each interviewee was dismissed the researchers took time to reflect on the field notes and identify how the data added to the overall narrative. This process was refined over the first three interviews and ultimately helped determine when saturation had been reached.

Data from the focus groups and interviews were organised using the NVIVO 12 software package. The qualitative dataset was coded by one researcher using a thematic framework (see Table 1) comprising the six constructs of exercise regulations proposed by Deci & Ryan and embedded in the BREQ-3 instrument [40‐42]. Interviewees were categorised, based on their weight trajectories as maintainers or relapsers, where a maintainer was defined as not having regained more than 50% of weight lost during the programme. The cross-tabulation function in the NVIVO 12 was used to compile relevant data for comparative analyses.

We compiled a survey instrument that would match the key themes identified in the preliminary analyses of qualitative data. The initial questionnaire contained segments on demographic, lifestyle (for example smoking and eating habits), health data (height and weight), the 24-item BREQ-3 and the participants’ most recent fitness test results. The items on the BREQ-3 segment were rated on a Likert scale of five options ranging from ‘not true for me’ to ‘very true for me’. The questionnaire was designed to be anonymous and self-administered by a participant comfortable in the use of the English language. Height, weights and fitness test results would be self-reported. The questionnaire was pilot-tested and refined with the help of age-matched conscripts supporting our team.

A full cohort of approximately 600 members of the residential programme were targeted for recruitment during the penultimate week of the residential programme in February 2019. This period coincided with briefings by military staff in preparation for the dispersal of programme participants to various roles following basic training. Conscripts were invited by an independent member of staff, not related to the residential programme or research team, to participate in the self-administered pen-and-paper survey. Potential participants were informed that the survey was anonymous, that their responses would not be tracked and that there would be no direct incentives for their contribution. Persons who declined to participate were advised to return an empty questionnaire at the end of the session.

BREQ-3 items were subjected to confirmatory factorial analysis (see Supplementary file 1) to test whether the six-factor structure adequately fit the data collected from our population as described in two recent publications [43, 44]. Then we computed composite scores deriving numerical averages for each item under the six BREQ-3 constructs. Summaries of survey items and the composite scores ranging from a minimum 1.0 to maximum 5.0 were compiled as parametric outcome data. Initial BMI was classified into overweight and obese categories using the WHO’s international cut-off value of 30 kg/m². We categorised weight loss according to three groups signifying less than 10, 10 to 15% and more than 15% weight loss. The fitness test results were treated as a binary variable according to whether the participant had achieved 50 points or higher during the final test administered at the end of the programme. Our epidemiological analyses examined the relationship between BREQ-3 scores and binary exposure variables – BMI and IPPT results – using independent t-tests. The relationship between BREQ-3 scores and the three weight loss categories was examined using ANOVA. Where parametric statistical tests showed differences greater in significance than p < 0.05, analyses were repeated under the assumption of non-parametric outcome data using the Wilcoxon rank-sum and Kruskal-Wallis tests respectively. Our reporting of comparative analyses reflects the outcomes of non-parametric statistical tests exclusively. We used the STATA 14 programme for all statistical analyses.

We have used the Consolidated Criteria for Reporting Qualitative Research (COREQ) [45] and the Good Reporting of a Mixed Methods Study (GRAMMS) [46] checklists to ensure the completeness of this manuscript.

The priorities expressed by the three focus groups comprising fitness trainers (n = 8), military leaders (n = 6) and ancillary staff (n = 7) are summarised in Table 3 showing a predilection for extrinsic controls to drive exercise and dietary behaviours. Among ancillary staff, however, the need to cultivate better habits, hence autonomous regulation, was given highest priority. Specific quotes from in-depth interviews have been cross-tabulated (see Supplementary Table 5). Analysis of qualitative data revealed the following narratives on motivations to exercise.

Confirmatory factorial analysis results (see Supplementary file 1) supported adequate convergent and discriminatory validity of the BREQ-3 scales in our sample. Overall, survey respondents rated items describing amotivation lowest of all six constructs (Table 4), followed by external then integrated regulation. Items describing identified regulation were scored highest with intrinsic motivation and introjected regulation ranked second and third respectively. This rank order did not differ when data were stratified by initial BMI or weight loss categories. However, the rating of all six constructs differed significantly when stratified by fitness test result; fitter respondents expressed stronger agreement with constructs of autonomous regulation (introjected, identified, integrated and intrinsic) and rejected statements of amotivation and external regulation.

While our study design was exploratory and sequential in nature, our analyses were explanatory and convergent [47]. Using triangulation within qualitative and quantitative data sources [48] we elicited three key themes on the motivations to exercise: (1) persistent barriers to exercise; (2) diminishing extrinsic motivation; and (3) unidentified benefits of exercise.

While the residential programme had suppressed exercise amotivation, it was unable to eliminate physical and social barriers to exercise that resurfaced after transition. The literature on exercise prescription advocates a structured approach to raising levels of physical activity through step-wise increases in exercise duration, then frequency and lastly intensity [49]. The pain and discomfort experienced by relapsers in our study might be interpreted as an inability to tolerate a large sustained increase in the level of activity resulting in over-training and increased risk of illness and injury despite best efforts to ensure gradual progression [50]. Concerning social barriers to exercise, environments that stigmatise persons suffering obesity can foster a strong sense of exercise avoidance [51]. In our sample, introjected regulation through camaraderie generally helped participants overcome the fear of engaging in exercise. However, as is the case with conventional weight loss programmes, this sense of community either receded or was completely lost after transition [52].

Consistent with the literature our survey findings suggest that the relationship between fitness and motivations to exercise was likely a direct one not mediated by initial BMI or amount of weight lost [53, 54]. While we are unable to comment on the direction of the relationship, this finding suggests that interventions to promote autonomous motivations to exercise might be better directed at the least fit and not the heaviest segment of the population. Interventions which have treated activity or fitness as the primary outcome and relegated weight loss to a secondary outcome have gained traction in recent years [55]. For men such approaches have the advantage of leaning on common perceptions of masculinity while weight loss maintenance and slimming typically evokes femininity [56].

The success of residential programmes has been attributed to the structured environments they can build around the needs of persons suffering overweight and obesity [57, 58]. Our data even suggests that there was a demand for a framework of incentives and penalties that extended past the end of the programme. Previous systematic reviews have shown that the impact of financial incentives for weight control in obesity was at best unclear [59, 60]. A rigorous trial in military veterans showed that financial incentives could produce tangible results, nevertheless the risk of relapse was high after disbursement had completed [61]. One novel approach to promoting exercise under free-living conditions involves mobile health interventions [62]. A recent randomised controlled trial showed that activity trackers administered in a workplace context on their own helped sustain a higher level of physical activity, even in absence of overt incentives [63].

Our survey data indicated that programme participants agreed most strongly with identified regulation as a motivation for exercise, meaning that they would have assigned a conscious value or benefit to exercise. From the qualitative data we were able to elicit two sub-themes detailing this benefit: (1) to control body weight and (2) to improve physical fitness. The former carried the outward benefit of a leaner, healthier appearance. The latter, however, was recognised almost in hindsight, as an enabler of exercise participation. A minority of voices discussed chronic disease risks but tended to recognise these as late consequences of high body weight, not inactive lifestyles.

Current scientific consensus is that long-term weight loss requires a combination of diet and exercise [64]. Exercise alone has been shown to produce smaller, more short-lived changes in weight [65]. The strong focus on body weight in our sample suggests an incomplete appreciation for the direct benefits of exercise that include improved chronic disease risk and better mental health [66, 67].

Our qualitative findings seem to suggest that while the programme helped participants overcome barriers to exercise, it did not prepare them for independent weight management. To do so would require a deliberate pivot from predominantly extrinsic to a more autonomous motivations to exercise, where the benefits of exercise would help sustain an active lifestyle. More importantly, our quantitative findings did not fully converge onto this picture as we uncovered that the young men in our population might have been at risk of overestimating their propensity for autonomous regulation. Therefore, to help explain this discrepancy, we integrated quantitative with qualitative results to discuss key reinforcing and deleterious patterns across different BREQ-3 constructs [68].

Low fitness and undesirable physical appearance were recurring themes in our study. High body weight before the programme and relapse after transition would produce either discouragement, physical discomfort or social distress, which in turn would reinforce inactive behaviour. Conversely, improved fitness and lean appearance were constructed as pre-requisites to enhanced enjoyment of exercise and sports, thus reinforcing the motivation to exercise. It has been suggested that in order to promote greater exercise autonomy, exercise programmes should avoid body weight and physical appearance as motive [69]. A framing that focuses on active lifestyles and fitness might foster stronger identified and introjected motivations to exercise. This could be achieved for example by monitoring of strength and cardiorespiratory fitness or by using gadgets and smartphone technologies to track activity levels [70, 71]. Finally, medical literature tends to favour health risks as vehicles for identified regulation [72]. Our study participants were more likely to register the real problems of disease in family members rather than in their distant-future selves. A significant minority of participants even shared their experiences with premature death due to noncommunicable disease in the family. These first-hand experiences could influence a personal sense of vulnerability and offer themselves as relevant topics for health promotion messaging [73].

From the qualitative data it emerged that young men living with obesity had to negotiate a fine balance between asserting self-esteem and responding to external cues that encouraged weight loss. A strong sense of camaraderie in the programme helped alleviate the social stigma of obesity. However, it might have inadvertently prevented a critical discussion on the utility and risks of negative self-image in overweight and obese individuals [74‐76]. Prevalent framing of changes in appearance as a source of inspiration and identified regulation in this population offers an opportunity to deepen the discussion on how to promote healthy lifestyles without diminishing self-esteem. One alternative avenue could be to emphasise the tangible benefits of exercise over the intangible improvements associated with lower body weight [77].

In our qualitative sample weight gain trajectories were related to the nature of work participants engaged in after completing the residential programme. Transport drivers routinely prioritised work obligations over their own needs leading to a permanent disruption of meal and exercise routines. These findings were in keeping with patterns observed in UK adolescents transitioning into adulthood [78]. Considering that a significant number of former participants were assigned to roles in security and logistics, it follows that success in transition might be bolstered by interventions at specific workplaces. Finally, it should not be overlooked that interviewees generally and begrudgingly accepted exercise as a legitimate means of instilling discipline. By associating exercise with punishment, however, the programme might have inadvertently suppressed the development of autonomous motivations to exercise, as reflected by participants who lamented the loss of strong external regulation after transition.

A key strength of this study was its mixed methods design, which allowed us to examine diverse lived experiences and assess the magnitude and frequency of motivations to exercise [79]. Our findings in a military population might lack generalisability to community-based weight loss programmes, the study nevertheless offers an opportunity to better understand the after-effects of an intensive lifestyle-intervention programme. Furthermore, our programme was designed to meet the needs of a wide segment of the population, including participants who might not have been captured at all by a voluntary programme.

Our study was designed to explore a specific period of transition with a focus on exercise. This meant that our methods prevented us from probing deeper into sensitive psychosocial themes such as bullying which constitutes a significant part of the literature on youth suffering from obesity [80]. Furthermore, we used a deductive approach framed by self-determination theory, thus limiting our ability to explore alternative behavioural constructs. The selection of participants for the qualitative study was through convenience sampling, which might not have provided the same depth and clarity as key-informant and snow-ball selection techniques. There were non-responders in both the qualitative and quantitative portions of the study who might concealed important data that would have expanded the proposed narrative. Finally, it was not feasible to review our results with the study participants meaning the participants themselves were not able to lend support to, or refute our findings and recommendations.

The findings of data triangulation and integration have allowed us to formulate the following recommendations which would help the residential programme reframe its approach to exercise and physical activity promotion: