A Model for Physical Activity Behavioural Change in Middle Aged and Older People with Type 2 Diabetes

. Objective : To identify key determinants of increased PA level among adults with T2DM to improve a hierarchical model, based on social cognitive theory (SCT) and Ecological Models. It hypothesises and clarifies how these determinants relate to physical activity which is important to evidence-based PA interventions in middle aged and older people with T2DM. Methods : Full transcript studies in English were searched in the following databases: CINAHL, Medline on OvidSP, PubMed, and PsycINFO. Included articles were selected by following these combined terms: type 2 diabetes, physical activity, exercise, physical inactivity, correlates, barriers, theory, self-efficacy, ecological models. And they (n=124) met the following criteria: 1) T2BD, aged 35 and over; 2) reporting determinants or factors 3) indicated physical activity or exercise as an outcome variable. Additional records identified via bibliographies (n=4), duplicates were removed (n=95), non-full-text articles (n=8) and no-English (n=2) were excluded. Finally, of the 21 articles retrieved from databases (9 of them are reviewed studies), only 12 original articles including qualitative and quantitative study were reviewed. Results : The determinants are divided into four classifications; physical, psychological, social and environmental factors, adapted combined SCT with Ecological Model of physical activity with T2MD in middle aged older adults. Self-efficacy is the core mediators with physical, environmental, and social factors, that fact is the core of SCT. Self-efficacy had an indirective negative influence by physical factors particular regions with physical barriers such as cold weather, low-density land use. Therefore, managing self-efficacy is proposed to effectively change for physical activity. It is hypothesised that improve social supports from family may help people with T2DM increase the physical activity level. Conclusion : Therefore, it is evident from the above that many factors of PA in middle aged and older people with T2DM exist. And self-efficacy is an important determinant with PA. There is a need to clarify whether these variables are determinants and the causality between these variables. Nonetheless, theory-basement approach to studying PA in this population is required based upon this study


Introduction
World Health Organization (2016) reports that diabetes will become the seventh leading cause of death by 2030. One adult in ten will live with diabetes, which means that around 642 million adults will have this disease by 2040 worldwide (International Diabetes Federation diabetes atlas, 2015). The prevalence of diabetes is predicted to rise to four million in the UK by 2025 (Diabetes UK, 2015). Moreover, the treatment of diabetes and its complications is estimated will cost £14 billion pounds per year. Additionally, 90% of diabetics have Type 2 diabetes (T2DM) and the majority of them are adults in the UK. There are some risk factors of T2DM such as overweight, smoking. One of the risk factors of T2DM is age; there is an increasing tendency of risk for T2DM on age. Specifically, people who are over 40 years old (or over 25 for those are Black African, South Asian or African-Caribbean) have a significantly higher risk of T2DM (https://www.diabetes.org.uk, 2009). Physical activity (PA) plays a crucial part in the prevention of T2DM; and it is a cost-effective method (Diabetes UK, 2015). Thomas, Elliott, and Naughton (2006) reviewed 14 randomised controlled trials (included 377 participants) comparing no PA against PA in T2DM, it stated that the PA intervention obviously enhanced glycaemic control (-0.6% HbA1c, 95% confidence interval (CI) -0.9 to -0.3, P < .05). Unfortunately, the majority of adult patients with T2DM were failed to meet the minimum recommendations (150 minutes per week) for PA (ADA, 2017). According to Thomas, (2004), the levels of PA are low among the diabetes population in the UK, with only 34% of diabetic patients meet this recommended PA guideline due to the difficulties in social factors such as lack of local facilities (p = .03) or leisure-time (p= .012).
There are many studies that have explored some possible reasons for physical inactivity in the general population such as lake of time and social supports ( (Baranowski, Anderson, and Carmack, 1999, Orleans, 2008, and Trost et al., 2002. However, particular diseases need particular PA guidelines and interventions, middle aged and older adults also need the intervention guidelines for themselves because they are physically inactive and high risk of T2DM basing on above evidence. However, only a few studies have focused on this population so far. For example, a cross-sectional study (n=1928) has highlighted that patients with T2DM had lower PA efficacy and readiness than healthy adults in Canada (Grace et al., 2006). A review comments that the main reasons for inactivity in diabetic adults (aged 35 and over) were physiological predicted barriers (p< .01), without family support (p< .01), and lack of facilities (p = .03) (Thomas, 2004 Another medical sociology research suggested that there was a need to understand prevention and treatment of disease, not only regarding individuals' levels but also within the broader social context of their lives (Lawton et al., 2003). For example, one cross-sectional study (n=1580) by Ferrand, Perrin, and Nasarre, (2008) showed that accessibilities, social support, weather, have also been associated with PA based on ecological impacts. Thus, it is required to recognise the physical, physiological, environmental and social determinants and factors in increasing the PA level, particularly in middle-aged and older people (aged over 35) with T2DM. Therefore, this article aims to identify key determinants of increased PA level among adults with T2DM to improve a hierarchical model, based on social cognitive theory (SCT) (primary) and Ecological Models. It will hypothesise and clarify how these determinants relate to PA, which is important to evidence-based PA interventions in middle aged and older people with T2DM.

Methods
Full transcript studies in English were searched in the following databases: CINAHL (Cumulative Index to Nursing & Allied Health), Medline on OvidSP, PubMed, and PsycINFO. This review has chosen articles from 2003 up to now to update a recent review (from1985 to 2002) (Allen, 2004). Included articles were selected by following these combined terms: type 2 diabetes, physical activity, exercise, physical inactivity, correlates, barriers, theory, self-efficacy, ecological models. And they (n=124) met the following criteria: 1) T2BD, aged 35 and over; 2) reporting determinants or factors 3) indicated PA or exercise as an outcome variable. Additional records identified via bibliographies (n=4), duplicates were removed (n=95), non-full-text articles (n=8) and no-English (n=2) were excluded. Finally, of the 21 articles retrieved from databases (9 of them are reviewed studies), only 12 original articles (both qualitative and quantitative study) were reviewed in this systemic review. Including six cross-section studies adopting Social cognitive theory (SCT) or Ecological Models are summarised in Table 1.

Results and discussion
The determinants and evidence illustrated in the following section, which are divided into four classifications; physical, psychological, social and environmental factors, adapted combined SCT (primary) (Bandura, 1986) with Ecological Model (Sallis et al, 2006) of PA with T2MD in middle aged older adults. Many scientific studies regarding the factors which impact this population are based on cross-sectional studies (See Table 1). Hence, only associations between variables could be identified, but causality cannot be inferred. However, these correlates are valuable for guiding further studies.

3.1Physical factors
Physical factors refer to characteristics of the individual that might impact PA such as gender, education and employment status. And individual health status may have influence on individuals' level of PA. (1) Gender: An interview (n=23) presented that female participants more stressed the important of emotional support than male participants (Ferrand et al., 2008). In contrast, Trost et al. (2002) reviewed 38 studies and concluded that males are more willing to participate in PA than females.
(2) Education and employment status: Lawton et al. (2005) reported participants with more years in education to be physically more active than participants with lower educational level (high school), indicating the role of education as a facilitator of PA. A crosssectional study (n=990) illustrated that fulltime work participants (p< .001) were less frequently active than those in part-time work (Grace et al., 2006). (3) Health status: It is defined as the physical functioning of an individual (Grace et al., 2006), compare T2DM patients with non-diabetes participants, diabetic participants had a smaller range and frequency of PA (p< .001). Moreover, participants who are non-smokers were more regularly active than those who were smokers (p= .001). Due to the fact that this cross-sectional study has a small sample size (n=133) and lower self-report rate (43.3%), this result may suggest an association between health status and PA. Additionally, an interview study (n=31) reported that participants who have health problems (67.3%) had more difficulties on PA than the healthy (33.7%) (Lawton et al., 2005).

Psychological determinants
(1)Self-efficacy: SCT suggested that self-efficacy and goals setting are the main determinants of PA (see figure  1). Heiss and Petosa (2015) found self-efficacy to be main determinant in their review; sharing an association with physical, environmental, and social factors (See Table 1). A longitudinal study base on a large randomised community sample (n=1662) found the diabetic group stated expressively lower marks for selfefficacy (p< .005) (Plotnikoff, Brez, and Brunet, 2003). Participants who had T2DM with a lower PA efficacy and willingness (p< .009) compared with others who had cardiovascular disease. And the efficacy had a negative association with physical and ecological obstacles such as low-density land usage and cold weather (Grace et al., 2006). Moreover, a longitudinal study included 2311 T2DM (age over 50) (Plotnikoff et al., 2006). It explored the associations between self-efficacy and other factors (barriers, social supports, goals setting). It concluded that the association of self-efficacy and barriers (β = -.28, p< .01) and social support (β= .21, p< .01) were weak. However, a direct pathway of self-efficacy on goals was significant (β= .62, p< .005). Moreover, the same was true for the influence on PA (β= .19, p<.01). The recent review also (Allen, 2004) stressed that to increase selfefficacy on PA is an efficient method for exercise intervention. And One study used an accelerometer and a pedometer to measure the strength of PA; it found that self-efficacy for moderate to vigorous PA was higher cardiovascular disease participants than in T2DM participants (p= .01) (Grace et al., 2006). (2) Golds setting and Outcome expectation: Outcome expectancies reflect one's belief that performing certain behaviour will cause a certain result (Bandura, 1985). Stronger goals at baseline were associated with higher levels of PA at six months. Moreover, outcome expectations were meaningfully related to goals (β = . 20 IV's Self-efficacy, outcome expectancies, social support, goals setting DV: PA (Self-report Leisure-time PA) Response race: 43.3% Only report the impacts on T2DM (compared with T1DM) -association between self-efficacy and goals (β= .62, p< .001) -association between goals and outcome expectations (β = .20, p< .01) -association between social supports & leisure-time PA(β = .40, p< .01)

T2DM
Age >35 48% female French Canadians IV's: Lacking of access to facilities, predictive outcomes, social norm DV: PA (self-report) Response race: 33% The variance of intention to engage in PA -lacking of access to facilities (β = .24, p < .005) -predictive outcomes (β = .38, p < .0001) -social norm (β = .29, p < .0001) living nearby stores and walk to transports (Odds Ratio= 1.92, 99% CI=1.11-3.32) (Taylor et al., 2008). Due to the precious CI and sample size, living location may increase the time of slight PA in this situation. A prospective cohort study (n = 23,865) also presented that participants living in the greenest areas compared with less green places had a lower risk of diabetes (19%) (Dalton et al., 2016).

A hierarchical model of determinants of PA in middle aged and older people with T2DM
SCT is a valuable framework for evaluating correlates of PA (Bandura, 1986). However, SCT focuses on interventions that target individuals or small groups (Sallis et al., 2006). In contrast, Ecological Models refer to the public's interactions with their sociocultural and physical environments (Stokols, 1992). Based evidence above, this hierarchical model combined SCT (primary) and ecological model. It reduced social supports, norm, modelling to social factors and put access to facilities, walkability, weather, physical barriers into environmental factors (the main concept of Ecological Models). It hypothesises that the key mediators of PA in middle aged older people (age over 35) with T2DM and demonstrate their correlates with other moderators and factors.As shown in Figure 1, a positively di-direct influence the PA (1), moreover, it also indirect mediates via goals setting positively PA (2), which is a positively di-indirection pathway between self-efficacy and goals setting, because of their high correlations in PA growth (Plotnikoff et al., 2008). It is absolute that self-efficacy is the core mediators with physical, environmental, and social factors, due to it are the core of SCT. Self-efficacy had an indirective negative influence (4) by physical factors particular regions with physical barriers such as cold weather, low-density land use (Grace et al., 2006). Therefore, it proposes that manage self-efficacy is effective for change this behaviour. Predictive barriers were presented to negatively influence on self-efficacy through this mediating pathway (3) to cause lower selfefficacy of PA (De Greef et al., 2010). Predictive barriers reduced the positive influence via an indirect pathway (5) also reduced the walkability and access to facilities on PA based on evidence by Heiss &Petosa, (2015) and Plotnikoff et al., (2006). It is hypothesised that to intervene predictive barriers on PA may lead to change in behaviour.
It is hypothesised that outcome expectations can increase PA's level via an indirect pathway (7). According to De Greef et al., (2010) and it can also moderates goals setting for positive and weak influence on P. It is hypothesised that building social modelling and norm (6) for engaging middle age people (aged≥35) for increasing in PA is available. Nevertheless, this pathway might be mediated by predictive barriers (4) based on De Greef et al., (2010). Additionally, the social supports from family (8) is hypothesised to play a key role for achieved regular PA, Deshpande et al., (2005) mentioned that this supports more positive influence on female than male. Meanwhile, social supports could direct impact on PA (.04*) (Plotnikoff et al., 2006). It is hypothesised that improve social supports from family may help people with T2DM increase the PA level.

Limitations and Conclusion
Firstly, the usage of subjectively self-reported measurement is as an indicator of practising PA. It may cause over-reporting of PA by participants due to social desirability biases. Only De Greef et al., (2010) adopted both objective (accelerometer and pedometer) and selfreported assessment. Therefore, many of the subjective measurements are needed to assess precious PA in further studies. The second limitation is a lack of generalizability there are two potential reasons: 1) most studies adopted voluntary participants, selection bias may have occurred; 2) low response rate (33%-51.7%, see table 1). However, these studies presented in six different countries and ethnic groups that may improve the generalizability of this review. Therefore, it is evident from the above that many factors of PA in middle aged and older people with T2DM exist. And Self-efficacy is an important determinant with PA. There is a need to clarify whether these variables are determinants and the causality between these variables. Nonetheless, theory-basement approach to studying PA in this population is required based upon this study.