Exploring maintenance of physical activity behaviour change among people living with and beyond gastrointestinal cancer: a cross-sectional qualitative study and typology

Objectives In the last decade, there has been a rapid expansion of physical activity (PA) promotion programmes and interventions targeting people living with and beyond cancer (LWBC). The impact that these initiatives have on long-term maintenance of PA remains under-researched. This study sought to explore the experiences of participants in order to characterise those who have and have not successfully sustained increases in PA following participation in a PA intervention after a diagnosis of gastrointestinal (GI) cancer, and identify barriers and facilitators of this behaviour. Design Cross-sectional qualitative study. Semi-structured interviews with participants who had previously taken part in a PA programme in the UK, explored current and past PA behaviour and factors that promoted or inhibited regular PA participation. Interviews were audio-recorded, transcribed verbatim and analysed using thematic analysis. Themes and subthemes were identified. Differences between individuals were recognised and a typology of PA engagement was developed. Participants Twenty-seven individuals (n=15 male, mean age=66.3 years) with a diagnosis of GI cancer who had participated in one of four interventions designed to encourage PA participation. Setting UK. Results Seven themes were identified: disease processes, the role of ageing, emotion and psychological well-being, incorporating PA into everyday life, social interaction, support and self-monitoring and competing demands. A typology with three types describing long-term PA engagement was generated: (1) maintained PA, (2) intermittent PA, (3) low activity. Findings indicate that identifying an enjoyable activity that is appropriate to an individual’s level of physical functioning and is highly valued is key to supporting long-term PA engagement. Conclusion The typology described here can be used to guide stratified and personalised intervention development and support sustained PA engagement by people LWBC.


INTRODUCTION
There is robust evidence that physical activity (PA) can provide improvements in physiological and psychosocial outcomes in people living with and beyond cancer (LWBC), including enhanced quality of life, physical function and reductions in cancer-related fatigue [1]. Furthermore, observational data suggest PA may reduce cancer recurrence, death from cancer and other causes in some cancer types. Therefore, cancer survivors potentially stand to benefit considerably from engagement in PA and consequently it is recommended that regular PA be promoted as part of the cancer care pathway [2]. The most recent guidelines from the American College of Sports Medicine suggest those LWBC should avoid inactivity and aim to achieve public health guidelines of 150 minutes per week of aerobic exercise and twice a week resistance exercise. This level of PA can also reduce the risk of developing chronic diseases, delay decline in physical and mental functioning associated with aging, and extend life [3]. The majority of cancer survivors do not meet these recommendations, and are less active than those without a history of cancer [4]. There has been a rapid expansion of PA promotion programmes and interventions targeting people LWBC to address this imbalance. Evidence suggests on-site supervised, home-based, web-based, individual and group and peer-support interventions can successfully increase PA in the short and medium term [5][6][7]. Furthermore, a recent systematic review and meta-analysis of long-term PA behaviour reported a moderate impact at least 3 months post-intervention completion [8]. However, there is wide variation of impact at an individual level with large standard deviations in self-reported and/or objectively recorded PA. This is elegantly illustrated by Morey and colleague's trajectory analysis of the RENEW study, a broad-reach intervention delivered by printed materials and telephone to older adults with a diagnosis of cancer [9]. The authors report a small proportion (7%) of participants did not engage in any PA at any point during the study. This is in contrast to 33% who increased PA at the end of the intervention but declined during the observational follow-up and around 60% who maintained increases in PA during the intervention and 12-month follow-up period. If interventions are to be developed that successfully support as many individuals as possible to maintain a physically active lifestyle we need to understand factors that promote and inhibit this maintenance of PA.
Individuals diagnosed with gastrointestinal (GI) cancers, such as colon, rectal, stomach and oesophageal are a particularly important group to consider. Affecting both men and women, cancers of the colon and rectum are particularly prevalent in older age which means individuals frequently present with other comorbidities. Individuals LWB GI cancer may also experience troublesome symptoms such as stool frequency and urgency, incontinence, fatigue and pain[10] making adoption and maintance of PA particularly challenging.
A recent mixed methods review investigated barriers, facilitators and preferences for PA among people LWBC and found that common barriers were treatment-related side effects, lack of time and fatigue [11]. Facilitators included exercising to gain control over health and feelings of well-being. This paper reports findings from a qualitative study that sought to explore the experiences of participants in order to characterise those who have and have not successfully sustained increases in PA following participation in a PA intervention after a diagnosis of GI cancer, and identify factors that promote and inhibit this behaviour.

Recruitment and procedure
Eligible participants had a previous diagnosis of colorectal, oesophageal or stomach cancer and had completed a PA programme more than 6 months before participating in the current study.
Participants were recruited from four PA promotion programmes in England and Northern Ireland (see Table 1). The coordinators of the programmes identified eligible participants and sent a letter of invitation to take part in the current study. Those who were willing were asked to return a reply slip to the coordinating centre at the University of Southampton. A lifestyle programme for those diagnosed with breast, prostate or colorectal cancer. This was a distance-based lifestyle intervention, delivered through printed materials and personally tailored telephone discussions underpinned by habit theory [28]. See published protocol [29]. Identified participants had a ENCOURAGE participants took part in a structured, supervised hospital-based exercise programme before during and after cancer treatments and prior to surgery.

Active Everyday
A PA referral programme based in Sheffield, UK. Participants received personalised behaviour change support and signposting to local PA opportunities [30]. Colorectal cancer only. Macmillan Move More Programme -Northern Ireland Participants receive personalised support with one-to-one consultations and advice to increase their activity levels and appropriate sign-posting to local opportunities including groupbased exercise programmes for people LWBC cancer. Colorectal cancer only.

Ethical statement
Ethical approval was obtained from the Health Research Authority and the University of Southampton Research Ethics committee prior to recruitment and data collection. Participants were provided with information sheets and gave written consent prior to interview.

Procedure
The study was reported according to the Consolidation Criteria for Reporting Qualitative Research (COREQ) [12]. See supplementary file 1.
All interviews were conducted by CG (an experienced qualitative researcher) and audio-recorded (with consent) using an encrypted electronic device. CG is a member of the trial steering committee for the ASCOT study but did not have direct involvement with the delivery of the intervention.
Participants were informed that the current study was independent of the study in which they had previously taken part.  Participants were also asked to complete a short demographics questionnaire to capture, age, sex, cancer type, marital status, ethnicity, level of education, home and car ownership, occupation and caring responsibilities.

Analysis
A thematic analysis was conducted as described by Braun & Clarke [13]. This was a recursive process moving back and forth between stages with continuous interpretation of the data from the outset. 1) Transcription: interviews were audio-recorded using an encrypted audio recorder and transcribed verbatim by an independent, experienced transcriber. Transcripts were checked against audio recorders, were anonymised and pseudonyms attributed. 2) Familiarisation with data: transcripts and field notes were read and re-read with analytical notes, thoughts and impressions recorded freehand. 3) Generating initial codes: transcripts were read line by line and initial codes applied. TC, CG and CRM coded a sample of transcripts (n=3). A data analysis workshop was then held to discuss codes and their labels. TC and CG coded a further 3 transcripts followed by a second analysis workshop before CG coded the remaining interviews. 4) Searching for themes: codes were then sorted into themes and sub-themes by CG and an initial thematic map generated and 5) Reviewing, defining and naming themes: themes were data driven, identified at a semantic level within the realist paradigm. A further analysis workshop was held between TC, CG and CRM to discuss the themes, sub-themes and the relationships between them, while also interrogating the data between cases. Refinements were made and CG revisited the transcripts to determine if themes and subthemes were an accurate representation of the data. NVIVO was used to manage the data.
Analysis indicated that groups of cases (individuals) existed and it was possible to develop a typology. A typology provides a useful way of describing groups of individuals with different clusters of behaviours, values or attitudes. Kluge's [14] four stage process of construction of empirically grounded types was followed. Stage 1; 'development of relevant analysing dimensions' -was achieved through the thematic analysis described above; here 'dimensions' are themes and sub- 3; 'analysis of meaningful relations and type construction' -relationships between the themes of the groups were analysed and heterogeneity between groups checked, i.e. sufficient variation in data existed between the groups. Stage 4; 'characterisation of the constructed types' -through an iterative process, including discussion with the wider research team, the final typology of PA maintenance was characterised.
Critical reflection was practiced throughout, and peer de-briefing was used to explore researcher bias.

Patient and public involvement
Patients and carers were involved in the conception of this study which forms part of a National Institute for Health Research post-doctoral fellowship. They reviewed the letters of invitation, patient information sheets and consent forms and suggested amendments. Patients and carers will also be involved in the dissemination of results, supporting the writing of a participant report and using their networks to disseminate results to wider patient communities.

Characteristics of respondents
Letters of invitation were sent to 124 individuals who had previously completed one of the four aforementioned programmes. A total of 48 reply slips were returned; yielding a response rate of 39%. Three individuals were found to be ineligible (no diagnosis of GI cancer) and seven could not be contacted. To ensure variation in the sample purposive sampling of the 38 viable responses was employed considering age, sex, socio-economic status and level of PA. Twenty-seven individuals were selected for interview.  Table 2. Just over half of the sample were male with an average age of 66.3 years, all were white British. The majority (89%) were married and owned their own home (93%), 56% were retired.

Thematic analysis results
Seven themes were identified, each is set out below with exemplar quotes presented in Table 3.
Disease processes:

Disease limits activities.
For some, late effects of cancer treatment inhibited engagement in certain activities. Limitations as a result of a stoma were described, for example, avoiding activities that involved bending, or swimming due to issues of body esteem and concerns about bag leakage. Some also experienced bowel urgency and described avoiding certain activities where toilet facilities were unavailable, as well as adapting activities due to concerns about leaks or accidents.
Other comorbidities/ill health impacted PA participation, including arthritic conditions and back problems. Chronic obstructive pulmonary disease, asthma and breathlessness were reported to restrict walking outside, particularly in cold weather. More acute periods of illness or injury were also described including cough/colds and muscle/joint strains which interrupted PA participation.
PA to improve comorbidities/late effects of cancer.
In contrast, PA was also described as essential in maintaining good health and well-being, avoiding illness and physical deterioration and this was a key motivating factor for ongoing participation.
Participants also recalled the role of PA in alleviating late effects of cancer treatment, for example,   Table 3).
The role of ageing Factors related to ageing were frequently referenced in relation to PA participation.

Ageing perceived as inhibiting PA.
In some cases, advancing age was cited as a barrier to engaging in PA. Modification to activities were described such as using the bus rather than walking, as well as a process of 'slowing down', which in some cases was associated with retirement. Some perceived that more structured exercise was not appropriate for older adults.
PA to combat consequences of ageing.
In contrast participation in PA was also reported to be important to preserve fitness and mobility and slow the physical decline associated with ageing. It was felt that regular exercise helped to ensure a future where the individual can engage in activities they enjoy and honour commitments to family and loved ones for as long as possible. Maintaining independence and avoiding burdening others was also important to some participants. Comparisons were also made to older relatives who they had witnessed decline in health and a desire to avoid such reductions in mobility and independence were expressed.
Emotion and psychological well-being The association between PA and psychological well-being and emotion was commonly described, both in a facilitative and deleterious way.
Psychological well-being.
Participants who engaged regularly in PA frequently described enjoying the activities as well as feeling a positive impact on psychological well-being. In some instances, PA also acted as a distraction from concerns about their health and alleviated anxieties. Others spoke of feeling empowered by taking exercise, using it as a vehicle to take control over their health following their cancer diagnosis and treatment.

Low mood, low motivation and lack of enjoyment.
For others feelings of low mood accompanied apathy and impacted on motivation leading to a reduction in activity levels. Boredom was also expressed with some participants struggling to find ways to increase their activity level in a manner that interested them and they enjoyed.
Incorporating PA into everyday life Participants described a variety of ways in which they incorporated PA into their lives.

Incidental activities.
Typically, individuals who participated in little or no structured exercise described a desire to avoid long periods of sitting. Participants recalled engaging in incidental activities such as domestic chores and gardening as a means of staying active.

Planned activities.
For some, engagement was structured, organised and part of their routine, such as attending regular exercise classes or active commuting. Participants expressed commitment to these activities and for  Others describe periods of increased PA engagement, for example   during the PA programme or prompted by other factors such as a desire to lose weight, which subsequently reduced or stopped. Some participants describe how they re-engaged in activities weeks, months or even years later.

Social interaction
Engagement in PA provided an important opportunity to socialise.

Avoiding isolation.
For some individuals living alone and/or who were retired, attending exercise classes provided an opportunity to meet with others and socialise, reducing feelings of isolation.
Friendship: Others talked about how participation in PA presented opportunities to develop meaningful and long-lasting friendships with those they exercised with. For those participants engaging in regular PA was an opportunity to spend time with friends and acted as a motivation to continued engagement.

Support and self-monitoring
For some participants continued engagement in PA was facilitated by personalised support. This was described by individuals who had previously engaged in regular exercise prior to cancer diagnosis, as well as those who were novices. The one-to-one specialised support provided during involvement in a PA promotion programme helped individuals find activities they enjoyed and were appropriate for them, enabling continued participation. For example, one participant described how a practice nurse 'lectured' her about taking more exercise with little effect. However, following a consultation with the Active Everyday practitioner, she was able to find an activity that suited her needs, and which she had engaged in weekly for more than 2 years. See Table 3 for quotes.
Self-monitoring techniques were important for some. The use of technology to monitor activity, particularly steps per day, was described as useful. Participants in the ASCOT trial were asked to wear a pedometer at the beginning of the study. Some reported that they were surprised at how few steps there were doing at baseline, and this provided motivation for change. Some continued to use a device to measure their steps, setting a goal, such as walking 10,000 steps a day and regularly self-monitoring their behaviour.
Further, despite ensuring no advice/recommendations to increase activity were made during the interviews some described their participation in the current study as acting as a prompt to consider re-engaging with PA.

Competing demands
Responsibilities such as caring for family members, family commitments and work schedules made consistent and continued engagement in PA challenging for some. Others were able to use these commitments as an opportunity to increase their activity levels. For example, playing with grandchildren and active commuting.

DISCUSSION
This study provides a novel insight into the experiences of sustaining long-term PA behaviour change following participation in a PA promotion programme among people LWBC. Seven key themes were identified, presenting key factors that promote or inhibit regular participation in PA; disease From these findings describing experiences, barriers and facilitators to long-term PA engagement it was possible to generate a typology consisting of three multidimensional types "maintained PA", "intermittent PA" and "low activity". This typology, informed by the themes identified, characterises the varied experiences of our sample helping to expand our understanding of factors that influence an individual's success, or not, of incorporating regular PA into their everyday lives (see Table 4 for characteristics by type and accompanying vignettes).
The first type is "Maintained PA" (see Figure 1.1). Participants described themselves as routinely active, most often engaging in structured moderate and/or vigorous PA at regular times/days.
Examples include attending the gym, Nordic walking and group exercise classes. Some had participated in regular PA at other periods of life and the PA intervention supported them to reengage with these activities. Others established new routines following participation. The defining features of this group include finding enjoyment and pleasure in regular PA which is planned and prioritised. Physical and psychological benefits of regular PA in preserving health, mobility and independence were described and highly valued. Interruptions to activity tended to be brief and resuming activity was not problematic or effortful. PA as a means of socialisation was important for some but not all in this type.
Characteristics of the "maintained PA" type are comparable to findings from a systematic review and meta-synthesis of the acceptability of PA interventions in older adults [15]. This review included evidence from studies of older adults (> 65 years old) from non-clinical populations. The authors described the importance of fun and enjoyment as a motivator for engagement, as well as descriptions of functional and psychosocial improvements, related to PA participation which increase perceived value of participation, and consequently enacted behaviour. Devereux-Fitzgerald et al. include the two factors of 'enjoyment' and 'value' as central to their analytic model describing  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59   Sue talks about the importance of physical activity to maintain health and describes its importance due to being overweight and experiencing a myocardial infarction. Sue describes periods where she exercised regularly but is currently inactive and says 'I know I should do more'. Following completion of cardiac rehabilitation Sue purchased a stationary exercise bike, the intention was to use it every day which she reports doing for a few weeks before stopping. Sue describes a lack of enjoyment and little change in weight after using her stationary bike. She describes having low mood and a lack of motivation to continue. Sue also talks about regaining weight she had lost which  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   20 PA behaviour as seen in the 'maintained PA' type [12]. The enjoyment of social interaction (as well as of PA participation in and of itself) was also a common theme in the published literature captured in this review. This is supported in part by the current study but social support was not a universal facilitator of continued PA engagement. Other reviews have challenged this assumption that social support is a key facilitator to engagement in PA, reporting it to be important for some but not all [16][17].
In one of the few qualitative studies investigating long-term PA behaviour in people LWBC, Midtgaard et al.[18] interviewed individuals who had participated in the Copenhagen PACT study, a 12 month programme to promote PA after cancer. Those interviewed described the importance of prioritising PA and planning ahead to ensure regular engagement. They also described exercise as a prerequisite for 'feeling and staying well'. However, as others have argued [15], a focus solely on the role of PA for long-term health is unlikely to encourage long-term participation of many older adults.
This is in keeping with the current study whereby the majority of participants describes knowledge of the value of PA for health but do not necessarily participate regularly. This is evident when we consider the 'intermittent PA' type (See Figure 1.2).
Participants in this type described irregular engagement in structured PA, often resulting in cycles of action and inaction. For example, starting an exercise class or purchasing new exercise equipment, participating regularly for a period of time and then subsequently experiencing an interruption to that behaviour. Typically, these participants discussed the known benefits of exercise and described taking part because 'they should', often expressing extrinsically driven motivations such as a desire to lose weight. Reasons for interruptions to PA participation varied and included intrinsic factors, for example, boredom or low mood and extrinsic factors, such as a stressful event, ill health or caring responsibilities. Once these barriers were no longer relevant participants often found it difficult to re-engage due to a lack of motivation/apathy. This group also described deriving less pleasure and intrinsic reward from engaging in PA than those in the maintained PA type. work and home responsibilities and lack of motivation as key barriers to engagement [11]. Others describe the deleterious effect of older age, physical decline and lack of time [16,19]. Although most studies in these reviews focused on adoption of PA rather than maintenance, the few studies that do consider long-term behaviour revealed similar obstacles. Brunet et al. [20] interviewed women with a previous diagnosis of breast cancer regarding maintenance of self-directed PA following participation in an intervention. Physical barriers were common, including arthritis and long-term effects of cancer treatment, such as fatigue as well as a lack of motivation and work and caring responsibilities. Similarly, in a study exploring experiences of PA over 5 years since diagnosis, Hefferon et al. [21] reported lack of motivation and apathy as well as the deleterious effects of the ageing process and co-morbidities and practical barriers such as proximity to exercise facilities and competing priorities as prohibitive to regular long-term PA engagement.
When considering the cycle of action and inaction as seen in the 'intermittent PA' type it is helpful to consider theories of motivation and behaviour change that may help to explain this phenomenon.
Kwasnicka et al. [22] present a systematic review of behaviour theories with a focus on maintenance of behaviour change and identify five key themes of which self-regulation is one. They propose that individuals differ in their ability to self-regulate their behaviour and those with lower self-regulatory capacity may have weaker intention-behaviour relationships (pg 285). It is suggested that when a behaviour falls short of a relevant goal, for example an individual does not attend an exercise class as planned, the individual must then exert more cognitive effort to 'correct' this behaviour and bring about satisfaction of goal achievement, or they disengage with the goal. At this stage the new behaviour is effortful, requiring significant self-regulation. Kwasnicka et al. [22] describe selfregulation as a limited cognitive resource which can be depleted through things such as stress and illness, and our self-regulatory reserves fluctuate throughout life. Habit formation has been proposed as a strategy to support behaviour maintenance (Gardner et al., in press). Habit is a process whereby a cue triggers an impulse to act because a mental association has been learned between a cue and an action [23,24]. Habits can support maintenance of behaviour because they are controlled through an associative reflexive system, as opposed to an information-processing system [25], meaning they continue to guide behaviours even when self-regulatory resources are low. Habit formation however can be a long process [26] and maintaining motivation for long enough to develop strong habits can be challenging, therefore long-term behaviour change will only be achieved once the behaviour is habitual, requiring less conscious control. These suggested mechanisms bear many similarities to the narratives presented by those in the intermittent activity type. It is also proposed that motives for sustained behaviour must bring about regular gratification and will be more successful if they are intrinsic, such as the enjoyment described by those in the 'maintained MVPA' type in this study.
The third and final type described here is "Low activity" (see Figure 1.3). Participants in this group take part in little or no structured PA but do describe avoiding long periods of sedentary time by engaging in activities of low intensity such as jobs around the home. Reference is also made to incidental activities such as climbing stairs in the home which are felt to maintain mobility and fitness levels. As with the maintained PA type, the importance of PA to maintain mobility was discussed. Those in the lower activity type felt this was achieved with lower intensity activities such as walking and household chores. Levels of PA participation were felt to be appropriate for their age and perceived capability. Reference was also made to reductions in activity levels compared to earlier in the life. Some are content with this and attribute it as a natural consequence of ageing.
Others had to stop or adapt activities less willingly as a result of comorbidities and/or consequences of cancer treatment. Comparisons were also made to others of a similar age with participants evaluating their activity levels to be higher than their peers. Those in the low activity group describe being able to carry out necessary activities of daily living and infrequently express a desire to engage in additional PA. Demographic characteristics of individuals by type (see Table 4) show those in the low activity group to be slightly older that those in the other types.  [27]. They synthesised the evidence for community-dwelling older adults who had not been involved in PA interventions and were therefore not necessarily active or willing to be active. They describe how older adults in these studies interpreted PA as a 'by-product' of other activities and a perception that more structured and purposeful PA wasn't necessary or appropriate for them. Similar concepts were described in a review of factors influencing PA participation in people with a diagnosis of lung cancer with 'usual activities' preferred over structured exercise [16].

Implications for practice and future research
Findings from this study have important implications for practice. Participants from all four PA programmes were represented in the three types described here demonstrating successful (or not) behaviour change was not limited to one approach. The typology reveals the need for targeted interventions to support long-term behaviour change rather than a 'one size fits all' approach. Those who have a history of exercising and/or who find activities that bring them pleasure and who hold strong values of the importance of PA will likely need minimal support to sustain engagement. In contrast individuals who experience cycles of success and failure of engagement need further support with self-regulatory process. More frequent contacts may help them identify appropriate activities, develop action plans and problem solve when they face challenges. However further research is needed to determine how best to support individuals to self-regulate independently when they experience recurrent cycles of action and inaction in order to prevent reliance on external support.
Raising awareness of the importance and relevance of PA that goes beyond participating in incidental and daily activities is necessary in the low activity type. They also require support to alter their self-identity, to see participation in PA as an appropriate behaviour for them, which may also A priority for future research is therefore to develop processes which can identify these key characteristics and determine the most appropriate support. Lastly, the striking similarities between themes presented here and those in the wider literature, particularly among clinical and non-clinical populations of older adults, suggests conclusions and therefore recommendations may be appropriate to other cancer types and non-cancer populations.

Strengths and limitations
Inclusion of a large sample consisting of men and women with variation in current activity levels and an age range that is representative of this disease cohort are strengths of this study. However transferability is reduced by the absence of any non-white participants, inclusion of only those with GI cancers with the vast majority married or co-habiting and most owning their own home suggesting financial security and a lack of socio-economic diversity despite attempts to maximise this during sampling.
It is also important to note that we present thematic descriptions of the data and higher-level conceptual models in the form of a typology. There are multiple causal mechanisms at play, often beyond the control of the individual which are not described in the typology presented here. Future work should look at mechanisms associated with these types and how these might inform the design of future PA promotion programmes. The authors also note that this typology sits within environmental and structural contexts which were not explored.
• How are things now in relation to your physical activity habits?
• For those who describe increased/regular physical activity engagement: o What has helped you to make these changes? o In what way is physical activity important to your everyday life?
• For those who describe no change in physical activity levels: o Would you like to be doing anything differently?
o If yeswhat are the main things that prevent you from being more active?
• For those who describe an initial increase but then a decline in physical activity levels: o Why do you think your activity levels reduced after the trial?
o Would you like to be doing anything differently?
o If yeswhat are the main things that prevent you from doing so?
• How do your physical activity habits compare to times before your cancer diagnosis?

Abstract
Objectives: In the last decade there has been a rapid expansion of physical activity (PA) promotion programmes and interventions targeting people living with and beyond cancer. The impact that these initiatives have on long-term maintenance of PA remains under-researched. This study sought to explore the experiences of participants in order to characterise those who have and have not successfully sustained increases in PA following participation in a PA intervention after a diagnosis of gastrointestinal cancer, and identify barriers and facilitators of this behaviour.
Design: Cross-sectional qualitative study. Semi-structured interviews with participants who had previously taken part in a physical activity program in the UK explored current and past PA behaviour and factors that promoted or inhibited regular PA participation. Interviews were audio recorded, transcribed verbatim and analysed using thematic analysis. Themes and sub-themes were identified.
Differences between individuals were recognised and a typology of PA engagement was developed.
Participants: Twenty-seven individuals (N = 15 male, mean age = 66.3 years) with a diagnosis of GI cancer who had participated in one of four interventions designed to encourage PA participation.
Setting: United Kingdom.
Results: Seven themes were identified: disease processes, the role of ageing, emotion and psychological well-being, incorporating PA into everyday life, social interaction, support and selfmonitoring and competing demands. A typology with three types describing long-term PA engagement was generated 1) maintained PA, 2) intermittent PA, 3) low activity. Findings indicate that identifying an enjoyable activity that is appropriate to an individual's level of physical functioning and is highly valued is key to supporting long-term PA engagement.
Conclusion: The typology described here can be used to guide stratified and personalised intervention development and support sustained PA engagement by people living with and beyond cancer.

Strengths and limitations of this study:
 This qualitative study using in-depth interviews with individuals engaging in varying levels of physical activity provides novel insight into the barriers and facilitators to long-term physical activity engagement after diagnosis and treatment of cancer.
 The typology presented can help inform the design and delivery of personalised physical activity interventions to support long-term behaviour change.
 Findings may be limited because participants were recruited from completed physical activity interventions and therefore do not reflect the views and experiences of individual who were unwilling or unable to take part in such a trial.

INTRODUCTION
There is robust evidence that physical activity (PA) can provide improvements in physiological and psychosocial outcomes in people living with and beyond cancer (LWBC), including enhanced quality of life, physical function and reductions in cancer-related fatigue [1]. Furthermore, observational data suggest PA may reduce cancer recurrence, death from cancer and other causes in some cancer types. Therefore, cancer survivors potentially stand to benefit considerably from engagement in PA and consequently it is recommended that regular PA be promoted as part of the cancer care pathway [2]. The most recent guidelines from the American College of Sports Medicine suggest those LWBC should avoid inactivity and aim to achieve public health guidelines of 150 minutes per week of aerobic exercise and twice a week resistance exercise. This level of PA can also reduce the risk of developing chronic diseases, delay decline in physical and mental functioning associated with aging, and extend life [3]. The majority of cancer survivors do not meet these recommendations, and are less active than those without a history of cancer [4]. There has been a rapid expansion of PA promotion programmes and interventions targeting people LWBC to address this imbalance. Evidence suggests on-site supervised, home-based, web-based, individual and group and peer-support interventions can successfully increase PA in the short and medium term [5][6][7]. Furthermore, a recent systematic review and meta-analysis of long-term PA behaviour reported a moderate impact at least 3 months post-intervention completion [8]. However, there is wide variation of impact at an individual level with large standard deviations in self-reported and/or objectively recorded PA. This is elegantly illustrated by Morey and colleague's trajectory analysis of the RENEW study, a broad-reach intervention delivered by printed materials and telephone to older adults with a diagnosis of cancer [9]. The authors report a small proportion (7%) of participants did not engage in any PA at any point during the study. This is in contrast to 33% who increased PA at the end of the intervention but declined during the observational follow-up and around 60% who maintained increases in PA during the intervention and 12-month follow-up period. If interventions are to be developed that successfully support as many individuals as possible to maintain a physically active lifestyle we need to understand factors that promote and inhibit this maintenance of PA.  [11]. Facilitators included exercising to gain control over health and feelings of well-being. However, the majority of studies in this review focused on uptake and/or participation in interventions, with little attention paid to experiences of independent and sustained (or not) behaviour change post-intervention. Furthermore, only 4 of the 19 studies included individuals with GI cancer.
This paper reports findings from a qualitative study that sought to explore the experiences of participants in order to characterise those who have and have not successfully sustained increases in PA following participation in a PA intervention after a diagnosis of GI cancer, and identify factors that promote and inhibit this behaviour.

Recruitment and procedure
Participants were recruited from four PA promotion programmes in England and Northern Ireland (see Table 1). The coordinators of the programmes identified eligible participants: previous diagnosis of colorectal, oesophageal or stomach cancer and had completed the PA programme more than 6 months before participating in the current study and sent a letter of invitation to take part in the current study. Those who were willing were asked to return a reply slip to the coordinating centre at the University of Southampton. A lifestyle programme for those diagnosed with breast, prostate or colorectal cancer. ASCOT was a distance-based lifestyle intervention, delivered through printed materials and personally tailored telephone discussions underpinned by habit theory [12].  [13]. Identified participants had a previous diagnosis of colorectal cancer and had set at least one PA goal during the study. Efficacy of aN exercise training programme during Concurrent neoadjuvant canceR treAtments) trial (ENCOURAGE) ENCOURAGE participants took part in a structured, supervised hospital-based exercise programme before during and after cancer treatments and prior to surgery.

Active Everyday
A PA referral programme based in Sheffield, UK. Participants received personalised behaviour change support and signposting to local PA opportunities [14]. Colorectal cancer only. Macmillan Move More Programme -Northern Ireland Participants receive personalised support with one-to-one consultations and advice to increase their activity levels and appropriate sign-posting to local opportunities including groupbased exercise programmes for people LWBC cancer. Colorectal cancer only.

Ethical statement
Ethical approval was obtained from the Health Research Authority and the University of Southampton Research Ethics committee prior to recruitment and data collection. Participants were provided with information sheets and gave written consent prior to interview.

Procedure
The study was reported according to the Consolidation Criteria for Reporting Qualitative Research (COREQ) [15]. See supplementary file 1.
All interviews were conducted by CG (an experienced qualitative researcher) and audio-recorded (with consent) using an encrypted electronic device. CG is a member of the trial steering committee for the ASCOT study but did not have direct involvement with the delivery of the intervention.
Participants were informed that the current study was independent of the study in which they had previously taken part.
One-to-one telephone interviews took place between January and June 2019 and lasted between 40 and 75 minutes. Semi-structured Interviews were carried out using a topic guide comprising open Participants were also asked to complete a short demographics questionnaire to capture, age, sex, cancer type, marital status, ethnicity, level of education, home and car ownership, occupation and caring responsibilities.

Analysis
A thematic analysis was conducted as described by Braun & Clarke [16]. A recursive process was employed, moving back and forth between stages with continuous interpretation of the data from the outset. 1) Transcription: interviews were audio-recorded using an encrypted audio recorder and transcribed verbatim by an independent, experienced transcriber. Transcripts were checked against audio recorders, were anonymised and pseudonyms attributed. 2) Familiarisation with data: transcripts and field notes were read and re-read with analytical notes, thoughts and impressions recorded freehand. 3) Generating initial codes: transcripts were read line by line and initial codes applied. TC, CG and CRM coded a sample of transcripts (n=3). A data analysis workshop was then held to discuss codes and their labels. TC and CG coded a further 3 transcripts followed by a second analysis workshop before CG coded the remaining interviews. 4) Searching for themes: codes were then sorted into themes and sub-themes by CG and an initial thematic map generated and 5) Reviewing, defining and naming themes: themes were data driven, identified at a semantic level within the realist paradigm. A further analysis workshop was held between TC, CG and CRM to discuss the themes, sub-themes and the relationships between them, while also interrogating the data between cases. Refinements were made and CG revisited the transcripts to determine if themes and subthemes were an accurate representation of the data. NVIVO was used to manage the data.  [17] four stage process of construction of empirically grounded types was followed. Stage 1; 'development of relevant analysing dimensions' -was achieved through the thematic analysis described above; here 'dimensions' are themes and subthemes. Stage 2; 'grouping of cases and analysis of empirical regularities' -individual cases were compared and contrasted and their position within the themes and sub-themes determined. Stage 3; 'analysis of meaningful relations and type construction' -relationships between the themes of the groups were analysed and heterogeneity between groups checked, i.e. sufficient variation in data existed between the groups. Stage 4; 'characterisation of the constructed types' -through an iterative process, including discussion with the wider research team, the final typology of PA maintenance was characterised.
Critical reflection was practiced throughout, and peer de-briefing was used to explore researcher bias.

Patient and public involvement
Patients and carers were involved in the conception of this study which forms part of a National Institute for Health Research post-doctoral fellowship. They reviewed the letters of invitation, patient information sheets and consent forms and suggested amendments. Patients and carers will also be involved in the dissemination of results, supporting the writing of a participant report and using their networks to disseminate results to wider patient communities.

Characteristics of respondents
Letters of invitation were sent to 124 individuals who had previously completed one of the four aforementioned programmes. A total of 48 reply slips were returned; yielding a response rate of 39%. Three individuals were found to be ineligible (no diagnosis of GI cancer) and seven could not be contacted. To ensure variation in the sample purposive sampling of the 38 viable responses was employed considering age, sex, socio-economic status and level of PA. Twenty-seven individuals were selected for interview.
Participant demographics are described in Table 2. Just over half of the sample were male with an average age of 66.3 years, all were white British. The majority (89%) were married and owned their own home (93%), 56% were retired.

Thematic analysis results
Seven themes were identified, each is set out below with exemplar quotes presented in Table 3.
Disease processes: Participants described the processes of disease, both cancer and comorbidities, and the role this played in relation to their PA participation.

Disease limits activities.
For some, late effects of cancer treatment inhibited engagement in certain activities. Limitations as a result of a stoma were described, for example, avoiding activities that involved bending, or swimming due to issues of body esteem and concerns about bag leakage. Some also experienced bowel urgency and described avoiding certain activities where toilet facilities were unavailable, as well as adapting activities due to concerns about leaks or accidents.
Other comorbidities/ill health impacted PA participation, including arthritic conditions and back problems. Chronic obstructive pulmonary disease, asthma and breathlessness were reported to restrict walking outside, particularly in cold weather. More acute periods of illness or injury were also described including cough/colds and muscle/joint strains which interrupted PA participation.
PA to improve comorbidities/late effects of cancer.
In contrast, PA was also described as essential in maintaining good health and well-being, avoiding illness and physical deterioration, and was a key motivating factor for ongoing participation.
The role of ageing Factors related to ageing were frequently referenced in relation to PA participation.

Ageing perceived as inhibiting PA.
In some cases, advancing age was cited as a barrier to engaging in PA. Modification to activities were described such as using the bus rather than walking, as well as a process of 'slowing down', which in some cases was associated with retirement. Some perceived that more structured exercise was not appropriate for older adults.

PA to combat consequences of ageing.
In contrast participation in PA was also reported to be important to preserve fitness and mobility and slow the physical decline associated with ageing. It was felt that regular exercise helped to ensure a future where the individual can engage in activities they enjoy and honour commitments to family and loved ones for as long as possible. Maintaining independence and avoiding burdening others was also important to some participants. Comparisons were also made to older relatives who they had witnessed decline in health and a desire to avoid such reductions in mobility and independence were expressed.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   14 The association between PA and psychological well-being and emotion was commonly described, both in a facilitative and deleterious way.

Emotion and psychological well-being
Psychological well-being.
Participants who engaged regularly in PA frequently described enjoying the activities as well as feeling a positive impact on psychological well-being. In some instances, PA also acted as a distraction from concerns about their health and alleviated anxieties. Others spoke of feeling empowered by taking exercise, using it as a vehicle to take control over their health following their cancer diagnosis and treatment.

Low mood, low motivation and lack of enjoyment.
For others feelings of low mood accompanied apathy and impacted on motivation leading to a reduction in activity levels. Boredom was also expressed with some participants struggling to find ways to increase their activity level in a manner that interested them and they enjoyed.
Incorporating PA into everyday life Participants described a variety of ways in which they incorporated PA into their lives.

Incidental activities.
Typically, individuals who participated in little or no structured exercise described a desire to avoid long periods of sitting. Participants recalled engaging in incidental activities such as domestic chores and gardening as a means of staying active.

Planned activities.
For some, engagement was structured, organised and part of their routine, such as attending regular exercise classes or active commuting. Participants expressed commitment to these activities and for  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   15 some engagement was a priority. Others describe periods of increased PA engagement, for example during the PA programme or prompted by other factors such as a desire to lose weight, which subsequently reduced or stopped. Some participants describe how they re-engaged in activities weeks, months or even years later.

Social interaction
Engagement in PA provided an important opportunity to socialise.

Avoiding isolation.
For some individuals living alone and/or who were retired, attending exercise classes provided an opportunity to meet with others and socialise, reducing feelings of isolation.
Friendship: Others talked about how participation in PA presented opportunities to develop meaningful and long-lasting friendships with those they exercised with. For those participants engaging in regular PA was an opportunity to spend time with friends and acted as a motivation to continued engagement.

Support and self-monitoring
For some participants continued engagement in PA was facilitated by personalised support. This was described by individuals who had previously engaged in regular exercise prior to cancer diagnosis, as well as those who were novices. The one-to-one specialised support provided during involvement in a PA promotion programme helped individuals find activities they enjoyed and were appropriate for them, enabling continued participation. For example, one participant described how a practice nurse 'lectured' her about taking more exercise with little effect. However, following a consultation with  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   16 the Active Everyday practitioner, she was able to find an activity that suited her needs, and which she had engaged in weekly for more than 2 years. See Table 3 for quotes.
Self-monitoring techniques were important for some. The use of technology to monitor activity, particularly steps per day, was described as useful. Participants in the ASCOT trial were asked to wear a pedometer at the beginning of the study. Some reported that they were surprised at how few steps there were doing at baseline, and this provided motivation for change. Some continued to use a device to measure their steps, setting a goal, such as walking 10,000 steps a day and regularly self-monitoring their behaviour.
Further, despite ensuring no advice/recommendations to increase activity were made during the interviews some described their participation in the current study as acting as a prompt to consider re-engaging with PA.

Competing demands
Responsibilities such as caring for family members, family commitments and work schedules made consistent and continued engagement in PA challenging for some. Others were able to use these commitments as an opportunity to increase their activity levels. For example, playing with grandchildren and active commuting.
The first type is "Maintained PA" (see Figure 1.1). Participants described themselves as routinely active, most often engaging in structured moderate and/or vigorous PA at regular times/days.
Examples include attending the gym, Nordic walking and group exercise classes. Some had participated in regular PA at other periods of life and the PA intervention supported them to reengage with these activities. Others established new routines following participation. The defining features of this group include finding enjoyment and pleasure in regular PA which is planned and prioritised. Physical and psychological benefits of regular PA in preserving health, mobility and independence were described and highly valued. Interruptions to activity tended to be brief and resuming activity was not problematic or effortful. PA as a means of socialisation was important for some but not all in this type.
The second type is "Intermittent PA" (see Figure 1.2). Participants in this type described irregular engagement in structured PA, often resulting in cycles of action and inaction. For example, starting an exercise class or purchasing new exercise equipment, participating regularly for a period of time and then subsequently experiencing an interruption to that behaviour. Typically, these participants discussed the known benefits of exercise and described taking part because 'they should', often expressing extrinsically driven motivations such as a desire to lose weight. Reasons for interruptions to PA participation varied and included intrinsic factors, for example, boredom or low mood and extrinsic factors, such as a stressful event, ill health or caring responsibilities. Once these barriers were no longer relevant participants often found it difficult to re-engage due to a lack of motivation/apathy. This group also described deriving less pleasure and intrinsic reward from engaging in PA than those in the maintained PA type.
The third and final type described is "Low activity" (see Figure 1.3). Participants in this group take part in little or no structured PA but do describe avoiding long periods of sedentary time by engaging 18 in activities of low intensity such as jobs around the home. Reference is also made to incidental activities such as climbing stairs in the home which are felt to maintain mobility and fitness levels. As with the maintained PA type, the importance of PA to maintain mobility was discussed with those in the lower activity type feeling this was achieved with lower intensity activities such as walking and household chores. Levels of PA participation were felt to be appropriate for their age and perceived capability. Reference was also made to reductions in activity levels compared to earlier in the life.
Some are content with this and attribute it as a natural consequence of ageing. Others had to stop or adapt activities less willingly as a result of comorbidities and/or consequences of cancer treatment. Comparisons were also made to others of a similar age with participants evaluating their activity levels to be higher than their peers. Those in the low activity group describe being able to carry out necessary activities of daily living and infrequently express a desire to engage in additional PA. Demographic characteristics of individuals by type (see Table 4) show those in the low activity group to be slightly older that those in the other types. There was variation in original intervention participation across all 3 types.

DISCUSSION
This study provides the first published insight into the experiences of sustaining long-term PA behaviour change following participation in a PA promotion programme among people LWB gastrointestinal cancer. Seven key themes were identified, presenting key factors that promote or inhibit regular participation in PA; disease processes, impact of ageing, emotions and psychological well-being, incorporating PA into everyday life, social interaction, competing demands support and self-monitoring. Additionally, a novel typology was developed to include three types; maintained PA, intermittent PA and low PA.
Characteristics of the "maintained PA" type where participation in PA is regular and sustained, are comparable to findings from a systematic review and meta-synthesis of the acceptability of PA  type [15]. The enjoyment of social interaction (as well as of PA participation in and of itself) was also a common theme in the published literature captured in this review and supported in part by the current study. However, social support was not a universal facilitator of continued PA engagement.
Other reviews have challenged this assumption that social support is a key facilitator to engagement in PA, reporting it to be important for some but not all [19][20].
In one of the few qualitative studies investigating long-term PA behaviour in people LWBC, Midtgaard et al. [21] interviewed individuals who had participated in the Copenhagen PACT study, a 12 month programme to promote PA after cancer. Those interviewed described the importance of prioritising PA and planning ahead to ensure regular engagement. They also described exercise as a prerequisite for 'feeling and staying well'. However, as others have argued[18], a focus solely on the role of PA for long-term health is unlikely to encourage long-term participation of many older adults.
This argument is in keeping with the current study whereby the majority of participants describes knowledge of the value of PA for health but do not necessarily participate regularly.   Sue talks about the importance of physical activity to maintain health and describes its importance due to being overweight and experiencing a myocardial infarction. Sue describes periods where she exercised regularly but is currently inactive and says 'I know I should do more'. Following completion of cardiac rehabilitation Sue purchased a stationary exercise bike, the intention was to use it every day which she reports doing for a few weeks before stopping. Sue describes a lack of enjoyment and little change in weight after using her stationary bike. She describes having low mood and a lack of motivation to continue. work and home responsibilities and lack of motivation as key barriers to engagement [11]. Others describe the deleterious effect of older age, physical decline and lack of time [19,22]. Although most studies in these reviews focused on adoption of PA rather than maintenance, the few studies that do consider long-term behaviour revealed similar obstacles. Brunet et al. [23] interviewed women with a previous diagnosis of breast cancer regarding maintenance of self-directed PA following participation in an intervention. Physical barriers were common, including arthritis and long-term effects of cancer treatment, such as fatigue as well as a lack of motivation and work and caring responsibilities. Similarly, in a study exploring experiences of PA over 5 years since diagnosis, Hefferon et al. [24] reported lack of motivation and apathy as well as the deleterious effects of the ageing process and co-morbidities and practical barriers such as proximity to exercise facilities and competing priorities as prohibitive to regular long-term PA engagement.
When considering the "intermittent PA" type with cycles of action and inaction, it is helpful to consider theories of motivation and behaviour change that may help to explain this phenomenon.
Kwasnicka et al. [25] present a systematic review of behaviour theories with a focus on maintenance of behaviour change and identify five key themes of which self-regulation is one. They propose that individuals differ in their ability to self-regulate their behaviour and those with lower self-regulatory capacity may have weaker intention-behaviour relationships (pg 285). It is suggested that when a behaviour falls short of a relevant goal, for example an individual does not attend an exercise class as planned, the individual must then exert more cognitive effort to 'correct' this behaviour and bring about satisfaction of goal achievement, or they disengage with the goal. At this stage the new behaviour is effortful, requiring significant self-regulation. Kwasnicka et al. [25] describe selfregulation as a limited cognitive resource which can be depleted through things such as stress and illness, and our self-regulatory reserves fluctuate throughout life. Habit formation has been proposed as a strategy to support behaviour maintenance [26]. Habit is a process whereby a cue triggers an impulse to act because a mental association has been learned between a cue and an action [27,28]. Habits can support maintenance of behaviour because they are controlled through an associative reflexive system, as opposed to an information-processing system [29], meaning they continue to guide behaviours even when self-regulatory resources are low. Habit formation however can be a long process [30] and maintaining motivation for long enough to develop strong habits can be challenging, therefore long-term behaviour change will only be achieved once the behaviour is habitual, requiring less conscious control. These suggested mechanisms bear many similarities to the narratives presented by those in the intermittent activity type. It is also proposed that motives for sustained behaviour must bring about regular gratification and will be more successful if they are intrinsic, such as the enjoyment described by those in the 'maintained MVPA' type in this study.
A number of the components of the 'low activity' type, characterised by engagement in little or no structured PA, share similarities with evidence from a recent review by McGowan and colleagues who conducted a systematic review and meta-synthesis of the acceptability of PA to older adults [31]. They synthesised the evidence for community-dwelling older adults who had not been involved in PA interventions and were therefore not necessarily active or willing to be active. They describe how older adults in these studies interpreted PA as a 'by-product' of other activities and a perception that more structured and purposeful PA wasn't necessary or appropriate for them.
Similar concepts were described in a review of factors influencing PA participation in people with a diagnosis of lung cancer with 'usual activities' preferred over structured exercise [22].

Implications for practice and future research
Findings from this study have important implications for practice. In addition to providing novel data of the experiences of maintance of PA behaviour change in PLWB gastrointestinal cancer, generation of the unique typology argues for targeted interventions to support long-term behaviour change rather than a 'one size fits all' approach. Participants from all four PA programmes were represented in the three types described here demonstrating successful (or not) behaviour change was not limited to one approach. Those who have a history of exercising and/or who find activities that bring them pleasure and who hold strong values of the importance of PA will likely need minimal support to sustain engagement. In contrast individuals who experience cycles of success and failure of engagement need further support with self-regulatory process. More frequent contacts may help them identify appropriate activities, develop action plans and problem solve when they face challenges. However further research is needed to determine how best to support individuals to self-regulate independently when they experience recurrent cycles of action and inaction in order to prevent reliance on external support.
Raising awareness of the importance and relevance of PA that goes beyond participating in incidental and daily activities is necessary in the low activity type. They may also require support to alter their self-identity, to see participation in PA as an appropriate behaviour for them, which may also be facilitated by alteration of social norms. They are likely to need further support to find activities appropriate to their physical capabilities, considering comorbidities and long-term consequences of cancer. True for all is the importance of finding activities that are meaningful and enjoyable.
A priority for future research is therefore to develop processes which can identify these key characteristics and determine the most appropriate support. Lastly, the striking similarities between themes presented here and those in the wider literature, particularly among clinical and non-clinical populations of older adults, suggests conclusions and therefore recommendations may be appropriate to other cancer types and non-cancer populations.

Strengths and limitations
Inclusion of a mixed sample consisting of men and women with variation in current activity levels and age is a strength of this study. However, transferability is reduced by the absence of any non- white participants, and the vast majority were married or co-habiting and most owning their own home suggesting financial security and a lack of socio-economic diversity despite attempts to maximise this during sampling. Furthermore, only 7 participants were classified as the 'low activity' type and therefore, despite our sampling strategy, are underrepresented in this study.
It is also important to note that we present thematic descriptions of the data and higher-level conceptual models in the form of a typology. There are multiple causal mechanisms at play, often beyond the control of the individual which are not described in the typology presented here. Future work should look at mechanisms associated with these types and how these might inform the design of future PA promotion programmes. The authors also note that this typology sits within environmental and structural contexts which were not explored.

Conclusions
People LWBC experience numerous challenges to sustaining increases in PA. These include cancer specific consequences such as presence of a stoma and issues with bowel habit, however many are similar to those described among non-cancer populations. The novel typology presented here supports the need for stratified approaches to long-term support for PA behaviour change. This approach is in keeping with emerging models of cancer care which call for personalised approaches.
Future research should explore whether tailoring of interventions using these typologies produces better maintenance of PA.         • How did you incorporate this into your day to day life?

CONFLICT OF INTEREST: None to declare
• How are things now in relation to your physical activity habits?
Differences between individuals were recognised and a typology of PA engagement was developed.
Participants: Twenty-seven individuals (N = 15 male, mean age = 66.3 years) with a diagnosis of GI cancer who had participated in one of four interventions designed to encourage PA participation.
Setting: United Kingdom.
Results: Seven themes were identified: disease processes, the role of ageing, emotion and psychological well-being, incorporating PA into everyday life, social interaction, support and selfmonitoring and competing demands. A typology with three types describing long-term PA engagement was generated 1) maintained PA, 2) intermittent PA, 3) low activity. Findings indicate that identifying an enjoyable activity that is appropriate to an individual's level of physical functioning and is highly valued is key to supporting long-term PA engagement.

Strengths and limitations of this study:
 This qualitative study using in-depth interviews with individuals engaging in varying levels of physical activity provides novel insight into the barriers and facilitators to long-term physical activity engagement after diagnosis and treatment of cancer.
 The typology presented can help inform the design and delivery of personalised physical activity interventions to support long-term behaviour change.
 Findings may be limited because participants were recruited from completed physical activity interventions and therefore do not reflect the views and experiences of individual who were unwilling or unable to take part in such a trial.

INTRODUCTION
There is robust evidence that physical activity (PA) can provide improvements in physiological and psychosocial outcomes in people living with and beyond cancer (LWBC), including enhanced quality of life, physical function and reductions in cancer-related fatigue [1]. Furthermore, observational data suggest PA may reduce cancer recurrence, death from cancer and other causes in some cancer types. Therefore, cancer survivors potentially stand to benefit considerably from engagement in PA and consequently it is recommended that regular PA be promoted as part of the cancer care pathway [2]. The most recent guidelines from the American College of Sports Medicine suggest those LWBC should avoid inactivity and aim to achieve public health guidelines of 150 minutes per week of aerobic exercise and twice a week resistance exercise. This level of PA can also reduce the risk of developing chronic diseases, delay decline in physical and mental functioning associated with aging, and extend life [3].  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   4 The majority of cancer survivors do not meet these recommendations, and are less active than those without a history of cancer [4]. There has been a rapid expansion of PA promotion programmes and interventions targeting people LWBC to address this imbalance. Evidence suggests on-site supervised, home-based, web-based, individual and group and peer-support interventions can successfully increase PA in the short and medium term [5][6][7]. Furthermore, a recent systematic review and meta-analysis of long-term PA behaviour reported a moderate impact at least 3 months post-intervention completion [8]. However, there is wide variation of impact at an individual level with large standard deviations in self-reported and/or objectively recorded PA. This is elegantly illustrated by Morey and colleague's trajectory analysis of the RENEW study, a broad-reach intervention delivered by printed materials and telephone to older adults with a diagnosis of cancer [9]. The authors report a small proportion (7%) of participants did not engage in any PA at any point during the study. This is in contrast to 33% who increased PA at the end of the intervention but declined during the observational follow-up and around 60% who maintained increases in PA during the intervention and 12-month follow-up period. If interventions are to be developed that successfully support as many individuals as possible to maintain a physically active lifestyle we need to understand factors that promote and inhibit this maintenance of PA.
Individuals diagnosed with gastrointestinal (GI) cancers, such as colon, rectal, stomach and oesophageal are a particularly important group to consider. Affecting both men and women, cancers of the colon and rectum are particularly prevalent in older age which means individuals frequently present with other comorbidities. Individuals LWB GI cancer may also experience troublesome symptoms such as stool frequency and urgency, incontinence, fatigue and pain[10] making adoption and maintenance of PA particularly challenging.
There is a paucity of published literature exploring challenges and enablers to long-term PA behaviour change among cancer populations. A recent mixed methods review investigated barriers, facilitators and preferences for PA among people LWBC and found that common barriers were  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   5 treatment-related side effects, lack of time and fatigue [11]. Facilitators included exercising to gain control over health and feelings of well-being. However, the majority of studies in this review focused on uptake and/or participation in interventions, with little attention paid to experiences of independent and sustained (or not) behaviour change post-intervention. Furthermore, only 4 of the 19 studies included individuals with GI cancer.
This paper reports findings from a qualitative study that sought to explore the experiences of participants in order to characterise those who have and have not successfully sustained increases in PA following participation in a PA intervention after a diagnosis of GI cancer, and identify factors that promote and inhibit this behaviour.

Recruitment and procedure
Participants were recruited from four PA promotion programmes in England and Northern Ireland (see Table 1). The coordinators of the programmes identified eligible participants: previous diagnosis of colorectal, oesophageal or stomach cancer and had completed the PA programme more than 6 months before participating in the current study and sent a letter of invitation to take part in the current study. Those who were willing were asked to return a reply slip to the coordinating centre at the University of Southampton. A lifestyle programme for those diagnosed with breast, prostate or colorectal cancer. ASCOT was a distance-based lifestyle intervention, delivered through printed materials and personally tailored telephone discussions underpinned by habit theory [12]. See published protocol [13]. Identified participants had a previous diagnosis of colorectal cancer and had set at least one PA goal during the study. Efficacy of aN exercise training programme during Concurrent neoadjuvant canceR treAtments) trial (ENCOURAGE) ENCOURAGE participants took part in a structured, supervised hospital-based exercise programme before during and after cancer treatments and prior to surgery.

Active Everyday
A PA referral programme based in Sheffield, UK. Participants received personalised behaviour change support and signposting to local PA opportunities [14]. Colorectal cancer only. Macmillan Move More Programme -Northern Ireland Participants receive personalised support with one-to-one consultations and advice to increase their activity levels and appropriate sign-posting to local opportunities including groupbased exercise programmes for people LWBC cancer. Colorectal cancer only.

Ethical statement
Ethical approval was obtained from the Health Research Authority and the University of Southampton Research Ethics committee prior to recruitment and data collection. Participants were provided with information sheets and gave written consent prior to interview.

Procedure
The study was reported according to the Consolidation Criteria for Reporting Qualitative Research (COREQ) [15]. See supplementary file 1.
All interviews were conducted by CG (an experienced qualitative researcher) and audio-recorded (with consent) using an encrypted electronic device. CG is a member of the trial steering committee for the ASCOT study but did not have direct involvement with the delivery of the intervention.
Participants were informed that the current study was independent of the study in which they had previously taken part.
Participants were also asked to complete a short demographics questionnaire to capture, age, sex, cancer type, marital status, ethnicity, level of education, home and car ownership, occupation and caring responsibilities.

Analysis
A thematic analysis was conducted as described by Braun & Clarke [16]. A recursive process was employed, moving back and forth between stages with continuous interpretation of the data from the outset. 1) Transcription: interviews were audio-recorded using an encrypted audio recorder and transcribed verbatim by an independent, experienced transcriber. Transcripts were checked against audio recorders, were anonymised and pseudonyms attributed. 2) Familiarisation with data: transcripts and field notes were read and re-read with analytical notes, thoughts and impressions recorded freehand. 3) Generating initial codes: transcripts were read line by line and initial codes applied. TC, CG and CRM coded a sample of transcripts (n=3). A data analysis workshop was then held to discuss codes and their labels. TC and CG coded a further 3 transcripts followed by a second analysis workshop before CG coded the remaining interviews. 4) Searching for themes: codes were then sorted into themes and sub-themes by CG and an initial thematic map generated and 5) Reviewing, defining and naming themes: themes were data driven, identified at a semantic level within the realist paradigm. A further analysis workshop was held between TC, CG and CRM to discuss the themes, sub-themes and the relationships between them, while also interrogating the data between cases. Refinements were made and CG revisited the transcripts to determine if themes and subthemes were an accurate representation of the data. NVIVO was used to manage the data.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  Critical reflection was practiced throughout, and peer de-briefing was used to explore researcher bias.

Patient and public involvement
Patients and carers were involved in the conception of this study which forms part of a National Institute for Health Research post-doctoral fellowship. They reviewed the letters of invitation, patient information sheets and consent forms and suggested amendments. Patients and carers will also be involved in the dissemination of results, supporting the writing of a participant report and using their networks to disseminate results to wider patient communities.

Characteristics of respondents
Letters of invitation were sent to 124 individuals who had previously completed one of the four aforementioned programmes. A total of 48 reply slips were returned; yielding a response rate of 39%. Three individuals were found to be ineligible (no diagnosis of GI cancer) and seven could not be contacted. To ensure variation in the sample purposive sampling of the 38 viable responses was employed considering age, sex, socio-economic status and level of PA. Twenty-seven individuals were selected for interview.
Participant demographics are described in Table 2. Just over half of the sample were male with an average age of 66.3 years, all were white British. The majority (89%) were married and owned their own home (93%), 56% were retired.

Thematic analysis results
Seven themes were identified, each is set out below with exemplar quotes presented in Table 3.
Disease processes: Participants described the processes of disease, both cancer and comorbidities, and the role this played in relation to their PA participation.

Disease limits activities.
For some, late effects of cancer treatment inhibited engagement in certain activities. Limitations as a result of a stoma were described, for example, avoiding activities that involved bending, or swimming due to issues of body esteem and concerns about bag leakage. Some also experienced bowel urgency and described avoiding certain activities where toilet facilities were unavailable, as well as adapting activities due to concerns about leaks or accidents.
Other comorbidities/ill health impacted PA participation, including arthritic conditions and back problems. Chronic obstructive pulmonary disease, asthma and breathlessness were reported to restrict walking outside, particularly in cold weather. More acute periods of illness or injury were also described including cough/colds and muscle/joint strains which interrupted PA participation.
PA to improve comorbidities/late effects of cancer.
In contrast, PA was also described as essential in maintaining good health and well-being, avoiding illness and physical deterioration, and was a key motivating factor for ongoing participation.
Participants also recalled the role of PA in alleviating late effects of cancer treatment, for example,  Practice nurses with my GP that---those sort of people lectured me for years about doing a bit more physical activity but they're not offering me anything.
We (Active Everyday trainer) talked about whether or not---you know, even when I was younger, I would have played any sport or done anything like that. We---we had a general conversation and that's---that as far as I know is a big part of their role. Is to fit people in with things that they feel you ' 13 exercising daily to improve gut mobility and bowel habit. PA was also reported as important to manage other conditions such as Type II diabetes, avoiding weight gain/promoting weight loss and alleviating joint stiffness caused by arthritic conditions. For one participant, engaging in regular PA revolutionised the way he managed chronic back pain (see Table 3).
The role of ageing Factors related to ageing were frequently referenced in relation to PA participation.

Ageing perceived as inhibiting PA.
In some cases, advancing age was cited as a barrier to engaging in PA. Modification to activities were described such as using the bus rather than walking, as well as a process of 'slowing down', which in some cases was associated with retirement. Some perceived that more structured exercise was not appropriate for older adults.

PA to combat consequences of ageing.
In contrast participation in PA was also reported to be important to preserve fitness and mobility and slow the physical decline associated with ageing. It was felt that regular exercise helped to ensure a future where the individual can engage in activities they enjoy and honour commitments to family and loved ones for as long as possible. Maintaining independence and avoiding burdening others was also important to some participants. Comparisons were also made to older relatives who they had witnessed decline in health and a desire to avoid such reductions in mobility and independence were expressed. The association between PA and psychological well-being and emotion was commonly described, both in a facilitative and deleterious way.

Emotion and psychological well-being
Psychological well-being.
Participants who engaged regularly in PA frequently described enjoying the activities as well as feeling a positive impact on psychological well-being. In some instances, PA also acted as a distraction from concerns about their health and alleviated anxieties. Others spoke of feeling empowered by taking exercise, using it as a vehicle to take control over their health following their cancer diagnosis and treatment.

Low mood, low motivation and lack of enjoyment.
For others feelings of low mood accompanied apathy and impacted on motivation leading to a reduction in activity levels. Boredom was also expressed with some participants struggling to find ways to increase their activity level in a manner that interested them and they enjoyed.
Incorporating PA into everyday life Participants described a variety of ways in which they incorporated PA into their lives.

Incidental activities.
Typically, individuals who participated in little or no structured exercise described a desire to avoid long periods of sitting. Participants recalled engaging in incidental activities such as domestic chores and gardening as a means of staying active.

Planned activities.
For some, engagement was structured, organised and part of their routine, such as attending regular exercise classes or active commuting. Participants expressed commitment to these activities and for 15 some engagement was a priority. Others describe periods of increased PA engagement, for example during the PA programme or prompted by other factors such as a desire to lose weight, which subsequently reduced or stopped. Some participants describe how they re-engaged in activities weeks, months or even years later.

Social interaction
Engagement in PA provided an important opportunity to socialise.

Avoiding isolation.
For some individuals living alone and/or who were retired, attending exercise classes provided an opportunity to meet with others and socialise, reducing feelings of isolation.
Friendship: Others talked about how participation in PA presented opportunities to develop meaningful and long-lasting friendships with those they exercised with. For those participants engaging in regular PA was an opportunity to spend time with friends and acted as a motivation to continued engagement.

Support and self-monitoring
For some participants continued engagement in PA was facilitated by personalised support. This was described by individuals who had previously engaged in regular exercise prior to cancer diagnosis, as well as those who were novices. The one-to-one specialised support provided during involvement in a PA promotion programme helped individuals find activities they enjoyed and were appropriate for them, enabling continued participation. For example, one participant described how a practice nurse 'lectured' her about taking more exercise with little effect. However, following a consultation with  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   16 the Active Everyday practitioner, she was able to find an activity that suited her needs, and which she had engaged in weekly for more than 2 years. See Table 3 for quotes.
Self-monitoring techniques were important for some. The use of technology to monitor activity, particularly steps per day, was described as useful. Participants in the ASCOT trial were asked to wear a pedometer at the beginning of the study. Some reported that they were surprised at how few steps there were doing at baseline, and this provided motivation for change. Some continued to use a device to measure their steps, setting a goal, such as walking 10,000 steps a day and regularly self-monitoring their behaviour.
Further, despite ensuring no advice/recommendations to increase activity were made during the interviews some described their participation in the current study as acting as a prompt to consider re-engaging with PA.

Competing demands
Responsibilities such as caring for family members, family commitments and work schedules made consistent and continued engagement in PA challenging for some. Others were able to use these commitments as an opportunity to increase their activity levels. For example, playing with grandchildren and active commuting.
The first type is "Maintained PA" (see Figure 1.1). Participants described themselves as routinely active, most often engaging in structured moderate and/or vigorous PA at regular times/days. Examples include attending the gym, Nordic walking and group exercise classes. Some had participated in regular PA at other periods of life and the PA intervention supported them to reengage with these activities. Others established new routines following participation. The defining features of this group include finding enjoyment and pleasure in regular PA which is planned and prioritised. Physical and psychological benefits of regular PA in preserving health, mobility and independence were described and highly valued. Interruptions to activity tended to be brief and resuming activity was not problematic or effortful. PA as a means of socialisation was important for some but not all in this type.
The second type is "Intermittent PA" (see Figure 1.2). Participants in this type described irregular engagement in structured PA, often resulting in cycles of action and inaction. For example, starting an exercise class or purchasing new exercise equipment, participating regularly for a period of time and then subsequently experiencing an interruption to that behaviour. Typically, these participants discussed the known benefits of exercise and described taking part because 'they should', often expressing extrinsically driven motivations such as a desire to lose weight. Reasons for interruptions to PA participation varied and included intrinsic factors, for example, boredom or low mood and extrinsic factors, such as a stressful event, ill health or caring responsibilities. Once these barriers were no longer relevant participants often found it difficult to re-engage due to a lack of motivation/apathy. This group also described deriving less pleasure and intrinsic reward from engaging in PA than those in the maintained PA type.
The third and final type described is "Low activity" (see Figure 1.3). Participants in this group take part in little or no structured PA but do describe avoiding long periods of sedentary time by engaging  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   18 in activities of low intensity such as jobs around the home. Reference is also made to incidental activities such as climbing stairs in the home which are felt to maintain mobility and fitness levels. As with the maintained PA type, the importance of PA to maintain mobility was discussed with those in the lower activity type feeling this was achieved with lower intensity activities such as walking and household chores. Levels of PA participation were felt to be appropriate for their age and perceived capability. Reference was also made to reductions in activity levels compared to earlier in the life.
Some are content with this and attribute it as a natural consequence of ageing. Others had to stop or adapt activities less willingly as a result of comorbidities and/or consequences of cancer treatment. Comparisons were also made to others of a similar age with participants evaluating their activity levels to be higher than their peers. Those in the low activity group describe being able to carry out necessary activities of daily living and infrequently express a desire to engage in additional PA. Demographic characteristics of individuals by type (see Table 4) show those in the low activity group to be slightly older that those in the other types. There was variation in original intervention participation across all 3 types.

DISCUSSION
This study provides the first published insight into the experiences of sustaining long-term PA behaviour change following participation in a PA promotion programme among people LWB gastrointestinal cancer. Seven key themes were identified, presenting key factors that promote or inhibit regular participation in PA; disease processes, impact of ageing, emotions and psychological well-being, incorporating PA into everyday life, social interaction, competing demands support and self-monitoring. Additionally, a novel typology was developed to include three types; maintained PA, intermittent PA and low PA.
Characteristics of the "maintained PA" type where participation in PA is regular and sustained, are comparable to findings from a systematic review and meta-synthesis of the acceptability of PA  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  and 'value' as central to their analytic model describing PA behaviour as seen in the 'maintained PA' type [15]. The enjoyment of social interaction (as well as of PA participation in and of itself) was also a common theme in the published literature captured in this review and supported in part by the current study. However, social support was not a universal facilitator of continued PA engagement.
Other reviews have challenged this assumption that social support is a key facilitator to engagement in PA, reporting it to be important for some but not all [19][20].
In one of the few qualitative studies investigating long-term PA behaviour in people LWBC, Midtgaard et al. [21] interviewed individuals who had participated in the Copenhagen PACT study, a 12 month programme to promote PA after cancer. Those interviewed described the importance of prioritising PA and planning ahead to ensure regular engagement. They also described exercise as a prerequisite for 'feeling and staying well'. However, as others have argued[18], a focus solely on the role of PA for long-term health is unlikely to encourage long-term participation of many older adults.
This argument is in keeping with the current study whereby the majority of participants describes knowledge of the value of PA for health but do not necessarily participate regularly.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59   Sue talks about the importance of physical activity to maintain health and describes its importance due to being overweight and experiencing a myocardial infarction. Sue describes periods where she exercised regularly but is currently inactive and says 'I know I should do more'. Following completion of cardiac rehabilitation Sue purchased a stationary exercise bike, the intention was to use it every day which she reports doing for a few weeks before stopping. Sue describes a lack of enjoyment and little change in weight after using her stationary bike. She describes having low mood and a lack of motivation to continue. work and home responsibilities and lack of motivation as key barriers to engagement [11]. Others describe the deleterious effect of older age, physical decline and lack of time [19,22]. Although most studies in these reviews focused on adoption of PA rather than maintenance, the few studies that do consider long-term behaviour revealed similar obstacles. Brunet et al. [23] interviewed women with a previous diagnosis of breast cancer regarding maintenance of self-directed PA following participation in an intervention. Physical barriers were common, including arthritis and long-term effects of cancer treatment, such as fatigue as well as a lack of motivation and work and caring responsibilities. Similarly, in a study exploring experiences of PA over 5 years since diagnosis, Hefferon et al. [24] reported lack of motivation and apathy as well as the deleterious effects of the ageing process and co-morbidities and practical barriers such as proximity to exercise facilities and competing priorities as prohibitive to regular long-term PA engagement.
When considering the "intermittent PA" type with cycles of action and inaction, it is helpful to consider theories of motivation and behaviour change that may help to explain this phenomenon.
Kwasnicka et al. [25] present a systematic review of behaviour theories with a focus on maintenance of behaviour change and identify five key themes of which self-regulation is one. They propose that individuals differ in their ability to self-regulate their behaviour and those with lower self-regulatory capacity may have weaker intention-behaviour relationships (pg 285). It is suggested that when a behaviour falls short of a relevant goal, for example an individual does not attend an exercise class as planned, the individual must then exert more cognitive effort to 'correct' this behaviour and bring about satisfaction of goal achievement, or they disengage with the goal. At this stage the new behaviour is effortful, requiring significant self-regulation. Kwasnicka et al. [25] describe selfregulation as a limited cognitive resource which can be depleted through things such as stress and illness, and our self-regulatory reserves fluctuate throughout life. Habit formation has been proposed as a strategy to support behaviour maintenance [26]. Habit is a process whereby a cue  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   23 triggers an impulse to act because a mental association has been learned between a cue and an action [27,28]. Habits can support maintenance of behaviour because they are controlled through an associative reflexive system, as opposed to an information-processing system [29], meaning they continue to guide behaviours even when self-regulatory resources are low. Habit formation however can be a long process [30] and maintaining motivation for long enough to develop strong habits can be challenging, therefore long-term behaviour change will only be achieved once the behaviour is habitual, requiring less conscious control. These suggested mechanisms bear many similarities to the narratives presented by those in the intermittent activity type. It is also proposed that motives for sustained behaviour must bring about regular gratification and will be more successful if they are intrinsic, such as the enjoyment described by those in the 'maintained MVPA' type in this study.
A number of the components of the 'low activity' type, characterised by engagement in little or no structured PA, share similarities with evidence from a recent review by McGowan and colleagues who conducted a systematic review and meta-synthesis of the acceptability of PA to older adults [31]. They synthesised the evidence for community-dwelling older adults who had not been involved in PA interventions and were therefore not necessarily active or willing to be active. They describe how older adults in these studies interpreted PA as a 'by-product' of other activities and a perception that more structured and purposeful PA wasn't necessary or appropriate for them.
Similar concepts were described in a review of factors influencing PA participation in people with a diagnosis of lung cancer with 'usual activities' preferred over structured exercise [22].

Implications for practice and future research
Findings from this study have important implications for practice. In addition to providing novel data of the experiences of maintance of PA behaviour change in PLWB gastrointestinal cancer, generation of the unique typology argues for targeted interventions to support long-term behaviour change  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   24 rather than a 'one size fits all' approach. Participants from all four PA programmes were represented in the three types described here demonstrating successful (or not) behaviour change was not limited to one approach. Those who have a history of exercising and/or who find activities that bring them pleasure and who hold strong values of the importance of PA will likely need minimal support to sustain engagement. In contrast individuals who experience cycles of success and failure of engagement need further support with self-regulatory process. More frequent contacts may help them identify appropriate activities, develop action plans and problem solve when they face challenges. However further research is needed to determine how best to support individuals to self-regulate independently when they experience recurrent cycles of action and inaction in order to prevent reliance on external support.
Raising awareness of the importance and relevance of PA that goes beyond participating in incidental and daily activities is necessary in the low activity type. They may also require support to alter their self-identity, to see participation in PA as an appropriate behaviour for them, which may also be facilitated by alteration of social norms. They are likely to need further support to find activities appropriate to their physical capabilities, considering comorbidities and long-term consequences of cancer. True for all is the importance of finding activities that are meaningful and enjoyable.
A priority for future research is therefore to develop processes which can identify these key characteristics and determine the most appropriate support. Lastly, the striking similarities between themes presented here and those in the wider literature, particularly among clinical and non-clinical populations of older adults, suggests conclusions and therefore recommendations may be appropriate to other cancer types and non-cancer populations.

Strengths and limitations
Inclusion of a mixed sample consisting of men and women with variation in current activity levels and age is a strength of this study. However, transferability is reduced by the absence of any non-  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   25 white participants, and the vast majority were married or co-habiting and most owning their own home suggesting financial security and a lack of socio-economic diversity despite attempts to maximise this during sampling. Furthermore, only 7 participants were classified as the 'low activity' type and therefore, despite our sampling strategy, are underrepresented in this study.
It is also important to note that we present thematic descriptions of the data and higher-level conceptual models in the form of a typology. There are multiple causal mechanisms at play, often beyond the control of the individual which are not described in the typology presented here. Future work should look at mechanisms associated with these types and how these might inform the design of future PA promotion programmes. The authors also note that this typology sits within environmental and structural contexts which were not explored.

Conclusions
People LWBC experience numerous challenges to sustaining increases in PA. These include cancer specific consequences such as presence of a stoma and issues with bowel habit, however many are similar to those described among non-cancer populations. The novel typology presented here supports the need for stratified approaches to long-term support for PA behaviour change. This approach is in keeping with emerging models of cancer care which call for personalised approaches.
Future research should explore whether tailoring of interventions using these typologies produces better maintenance of PA.

COREQ (COnsolidated criteria for REporting Qualitative research) Checklist
A checklist of items that should be included in reports of qualitative research. You must report the page number in your manuscript where you consider each of the items listed in this checklist. If you have not included this information, either revise your manuscript accordingly before submitting or note N/A. • What did you do as part of the [name physical activity programme]?
• How did you incorporate this into your day to day life?
• How are things now in relation to your physical activity habits?
• For those who describe increased/regular physical activity engagement: o What has helped you to make these changes? o In what way is physical activity important to your everyday life?
• For those who describe no change in physical activity levels: o Would you like to be doing anything differently?
o If yeswhat are the main things that prevent you from being more active?
• For those who describe an initial increase but then a decline in physical activity levels: o Why do you think your activity levels reduced after the trial?
o Would you like to be doing anything differently?
o If yeswhat are the main things that prevent you from doing so?