Positive Psychology Interventions to Improve Wellbeing and Health Behaviour Adherence in Patients with Type 2 Diabetes Mellitus: A Scoping Review and Meta-analyses

The aim of this review was to evaluate the effectiveness of positive psychological interventions (PPIs) to improve well-being and health behaviour adherence among patients with type 2 diabetes mellitus (T2DM). Medline, PsycINFO, the Cochrane register, EMBASE, and Google Scholar were systematically searched to find relevant studies until January 2020. The primary outcome was reduction in risk factors of cardiovascular disease including HbA1c, systolic blood pressure (SBP), and diastolic blood pressure along with improvement in positive affect, optimism, self-efficacy, and health behaviour adherence such as diet, exercise and medication. The secondary outcomes were reduction in depression, anxiety and stress. A random-effect model was used to compare group effect size at post-test. We identified a total of 11 studies (N=1594 participants) with substantial variability in the interventions. Overall, the results provide evidence that multi-component PPIs have a small but significant effect on positive affect, optimism, health behaviour, self-care and BMI. Further, the review demonstrates that PPIs can be effective in the reduction of anxiety and stress symptoms. However, studies included in this review are heterogenous due to methodological variation, therefore, in future more studies across a wide range of PP interventions needs to be included in order to validate the findings and for conclusive evidence.

with noncompliance to oral medications, along with diet and physical activity and monitoring of blood glucose [9]. However, depression interventions have not consistently led to changes in adherence or outcomes among this population [10]. In contrast to a negative syndrome, positive psychological states such as positive affect, optimism, gratitude and other related constructs play an important role in improving therapeutic adherence. In patients with diabetes, positive constructs have been shown to be positively associated with superior health outcomes including social, physical and health [11].

The Hypothesised Mechanism Underlying Positive Psychology Interventions
Previous studies which examined the link between PP exercises and clinical outcome have typically based through mental well-being, depressive symptoms and health behaviour adherence. For instance, Huffman et al. [12] theoretical review developed a framework describes the mechanism through which PP intervention promotes health behaviours which were mediated through positive emotions and positive mental health and reduces depression. Similarly, Celano [13] framework showed through positive psychological exercise improves cardiovascular health through improvement positive constructs such as gratitude and optimism and health behaviour adherence such as increased physical activity. Similarly, Kubzansky [14] model showed the direct and indirect relationship linking positive psychological well-being with better cardiovascular health through behavioural pathways (e.g. smoking, physical activity), and psychosocial pathways (mitigating depression, or stress). Positive affect unifies the experiences of contentment, joy and love, on the other hand, negative affect comprised of sadness, distress and fear [15]. This experience may expand a behavioural and cognitive repertoire, including intellectual, social and physical resources, possibly to include treatment adherence and health behaviour conversely (Lianov et al., 2019). Taken together, based on the theoretical underpinnings of PP states have shown to impact clinical outcomes in patients with T2D through positive states and behavioural factors [16].
To our knowledge, little is known about the potential impact of PP interventions for T2DM, and no existing meta-analyses have been undertaken of RCTs strictly through PP interventions focusing on both mental wellbeing and health behaviour adherence among this group. In the past, few studies have been conducted to evaluate interventions comprising both PP and other psychological enhancing components for depression and other health behaviour adherence in clinical populations [17]. One systematic review included 30 studies, including PP interventions (n=4 articles) along with other well-being interventions [18], while the results of this review study showed a promising effect in improving health outcomes.

Aims
The meta-analysis aimed to identify effective western-based PP interventions with a control or usual care strategy in adults (age ≥ 20 yr.) and to determine which components of these interventions optimized their impact on both depression and anxiety as well as health behaviour adherence such as diet, physical activity and medication adherence. Besides, this paper will also describe the methodology used and outline its main findings in terms of the quality and content of the reviews, the impact of interventions and will discuss the implications of these findings.

Search Strategy
The present review followed the framework of Arksey and O'Malley [19] for a scoping review. The preferred reporting items for Systematic Reviews and Meta-Analysis for Scoping Review (PRISMA-ScR) [20] criteria and guidelines were adhered while executing and reporting this review.

Stage 1: Review Aims and Research Questions
The present scoping review aimed to explore the impact of PPIs on well-being and health behaviour adherence for people with T2DM. This research question followed the suggestion by Arkey and O'Malley's [19] where emphasize was to start with a broad area before narrowing the search to determine what is already available. The authors, being experience in the field of PPIs, believe that there is a significant impact of PP on enhancing the adherence of health behaviour and improved clinical outcome via improving wellbeing and reducing stress and depression levels among people with T2DM. However, the existing meta-analysis is not specific in context to health behaviour adherence, and therefore this review helps describe the evidence for this PPI on adherence of health behaviour and distil key professional context, activities, and training protocols in the research of PP interventions. The following are the research questions guiding the scoping review (1) What are the demographic characteristics of the population in which PP intervention was applied for T2DM? (2) What theoretical basis has been used to design the PP intervention for T2DM? (3) What are the different primary and secondary outcome measures used in the interventions for T2DM? (4) What methods and approaches are used to deliver PP interventions for T2DM? (5) What is the retention rate and acceptability of the intervention for T2DM? (6) What are the educational background and prior experience of professionals delivering PP Intervention for T2DM?

Stage 2: Identifying Relevant Studies -Search Strategy
For this particular scoping review, the researcher executed systematic searches on the following electronic database: Scopus, PubMed, PsycINFO, Embase, and Ovid Cochrane Library. Search items, summarised, included positive psychological constructs (A): optimism, hope, gratitude, vitality, meaning, subjective well-being, happiness, self-acceptance, tranquillity, contentment, personal growth, positive affect, character strengths, emotional well-being, and cheerfulness. Interventions focused on PP include counting your blessings, practicing kindness, loving-kindness, setting a personal goal, expressing gratitude, using personal strengths, positive psychology intervention, and optimism training. During the search strategy, no time restrictions were placed where all published articles up to December 15, 2019, taken into consideration. The search strings were adapted according to the database. In addition, four meta-analyses [21][22][23][24] and six review articles on PPIs [25][26][27][28][29][30] were also cross-checked for additional references.

Primary Outcome
The primary outcome was an improvement in the mean positive affect, optimism, and life satisfaction constructs were included. Other primary outcome included health behaviour adherence such as diet, exercise, long-term glycemic control by the percentage of glycated haemoglobin (HbA1c) and body mass index (BMI), self-management and medication adherence.

Secondary Outcome
The secondary outcomes included in the review were depression, anxiety and stress.

Stage 3: Selection of Studies
The eligible studies were screened for their titles in the first phase, followed by its abstract in the second phase and the full paper in the final phase. This review paper comprised of peer-reviewed studies that were published exclusively in English and followed the PICOS (Participants, Interventions, Comparators, Outcomes and Study Design) framework. The framework utilized to set up particular criterions for inclusion and exclusion. To be part of this review, the following criteria had to be met. i) adult participants (≥18 years old) ii) diagnosed with T2DM, ii) any intervention trial (RCT, and non-RCT such as quasi-experimental study, proof-of-concept trials) that evaluated the effectiveness of PP intervention structured such as to encourage subjective well-being and health behaviour adherence iii) effect of PP intervention developed in line with the theoretical tradition of PP (Sin & Lyubomirsky, 2009) and iv) utilized a RCT. Conference paper, abstracts, case studies, observational studies and case reports were excluded. We also excluded studies that addressed a) recent acute complications or hospitalisation, b) patients with life-threatening illness, c) gestational diabetes, d) studies that described patients with T2DM or insulin-dependent diabetes, e) interventions that were primarily focused on reminiscence, mindfulness and /or meditation, f) and not published in English language peer-reviewed journals.

Stage 4: Data Extraction
The first author (GR) performed the data extraction, which was then verified by the second author (RS). Any disagreements were resolved by consensus and through consultation with the last author (RR). The GR later appraised abstracts from the rest of the papers to ensure that all the identified papers matched the objectives of this current research and also identifying any additional papers that could be excluded from this review. Lastly, whatever papers remained after the elimination was completely reviewed by the researcher. Following identification and elimination of duplicate papers, the rest of the identified papers were extensively reviewed by the GR, and any articles that were not pertinent to the current research were removed. In case of missing data, study authors were contacted.
The following data was gathered: First Author, year of publication, country of origin, participant ethnicity, type of diabetes (Type 1 or II), study objective, primary outcome, intervention type and psychological component involved, control group, theoretical basis, study design, mean age and percentage of female participants, sample size (per condition), setting, interventionists, mode of delivery, delivery form (self-help, group-based, individual therapy), module addressed and components, program phase, retention and acceptability of the intervention, PP intervention duration, follow-up period, metabolic control, physical activity, medication, diet, self-management, self-efficacy, positive outcome, subject wellbeing, depression, anxiety, and stress. We extracted means and standard deviation at post-test.

Stage 5: Evidence Synthesis
According to the guidelines recommended by Arksey and O'Malley's framework [19], a 'descriptive-analytical' method, based on the narrative tradition, was performed. Tables and graphs were created to reflect the overall summary of studies included. The main outcome of our scoping review was to answer the broad primary research question.

Risk of Bias
According to the scoping review recommendations and the Joanna Briggs Institute manual [20,31], the present review does not conduct an appraisal of the quality of included sources and an assessment of the risk of bias [32].

Meta-analysis Procedure
Based on Rosenthal [33] all statistical analyses were performed using "Review Manager (REVMAN) 5.3 Copenhagen" (The Nordic Cochrane Centre, 2014) to calculate the pooled MD (mean difference) as well as the corresponding 95% confidence interval (CI). The Q, I 2 , and T 2 were calculated to assess the heterogeneity between studies. A P value <0.01 or I 2 >50% indicated significant heterogeneity, and then, a random-effect model was used to pool the MDs.
Otherwise, a fixed-effect model was applied. T 2 provides the proportion of variability in the effect size [34]. A Z-test was used to determine the statistical significance of these pooled MDs. Funnel plot and Egger's test was used to assess the Publication bias. Separate meta-analyses were performed for Positive affect (PA), depression, BMI (Body Mass Index), HbA1c, and self-care. Since only two meta-analyses had all five variables filled, only those were considered into an account. In this meta-analysis, an adjusted estimate called Hedges' g which generates for each effect size was calculated based on its sample size as Cohen's d produces an overestimation for studies with small sample size [35,36]. The ES was considered as large, moderate and small if the size were 0.56 to 1.2; 0.33 to 0.55; and 0 to 0.32 small [37]. et al., subgroup analyses were performed by type of diabetes or the evaluation criteria of depression and anxiety. For all analyses, the P value of <0.05 indicated statistical significance. Evidence of publication bias was assessed in the following ways. First, Orwin's [38] Fail-Safe N, Egger's regression test and Kendall's tau [39] were used to assess publication bias. Second, we created a funnel plot by plotting the overall mean effect size against study size. Finally, we applied the trim-and-fill method [40] where the procedure imputes the effect sizes of missing studies and produces an adjusted effect size.

Search and Selection of Studies
The flow chart of literature search and study selection is represented in Figure 1. On the basis of a search of the selected electronic databases, a total of 888 potentially relevant citations were retrieved. After excluding duplicates and irrelevant articles, 82 potentially relevant articles remained. Among them, 67 citations were removed by scanning the title and abstracts: 17 non-original articles (reviews, letters or case reports), 29 articles were not directly relevant to diabetes or positive psychological intervention, 6 articles on adolescents or children and 15 non-RCTs were removed, according to the inclusion and exclusion criteria. After reading the full-text, 10 articles without available data were excluded. Finally, eleven studies [41][42][43][44][45][46][47][48][49][50][51] were included in this scoping review and meta-analysis. The characteristics of these included studies were listed in Table 1.

Scoping Review of PP Intervention Among T2D
The characteristics of these included studies were listed in Table 1. A total of eleven studies involving 1594 patients with diabetes mellitus (including 797 patients in PP intervention group and 797 patients in the control group) were published from 2007 to 2019. Most samples comprised of a heterogeneous mix of male and female participants with a mean age ranged from 72 to 49 years. There were no significant differences in age and sex between groups in these included studies.
Three studies [41,42,45] included 100% Americans outpatients, one study [43] included Turkish adults, Wu [51] conducted a study from the Taiwanese, Steinhardt [49] conducted a study amongst African Americans, while Pena [48] in Spanish Speaking Hispanic or Latino Adults. The study by Nolan et al., [47] conducted amongst Australian adults, Nishita [46] had a mix of Asian, Hawaiian or Pacific Islanders while Cohn et al., [44] had a mix of Caucasians, African Americans and Asian or Asian Americans, and non-white Hispanics. In terms of study design, four were RCT or single group RCT, or a pilot RCT [45], while three [42,49,51] used proof of concept pre and post-intervention. Two studies [44,50] had either intervention or nonrandomized trials. Three studies had self-efficacy [46,48,51], locus of control [41] as theoretical base while others applied revised stress and coping theory and the broaden and build theory of positive emotion [44], self-regulation and social cognitive theory [48], Rogers humanistic psychotherapy [50], and emotional intelligence theory [43] ( Table 1).
Of the included studies, two studies focused on optimism and gratitude intervention [42,45], one on positive emotion especially acts of kindness while other two (Steinhardt et al., 2009;[41] focused on resilience, and three studies [46,48,50], did focus on self-efficacy based intervention, one on positive reappraisal [47]. The duration of intervention varied from 1-week [47] to 8 months [50]. All studies had a control group except the study by Steinhardt [49] and Voscekova [50]. Single exposure duration ranged from 15 [42] to 120min [48,49]. Frequency of exposure ranged from biweekly and to one single exposure with 50min duration. Majority of the programs were delivered in-person while two studies delivered through telephone [45,51] and one online [44]. The format of the program was delivered individually in two studies [42,47], while eight studies did have a group program [41,43,[48][49][50][51], and others had self-help [44]. Majority of the programs were conducted in a clinical setting [41,47,49,51] while two in Urban Medical Centre [42,45] and one in Patients with Type 2 Diabetes Mellitus: A Scoping Review and Meta-analyses diabetes education centre [44] ( Table 2).
The retention rate was reported in four studies [ [44]. The programs were delivered by diverse experts including psychologists [42,47], research staff [45], trained life coach and pharmacist [46], registered nurse [50,51] and dietician [48], dietary nurse [50] PhD Candidate (Steinhardt et al., 2009), medical doctor psychotherapists and social worker (Voseckova et al., 2017) ( Table 3). The criteria for evaluating depression were CES-D (Centre for Epidemiological Studies scale for Depression), BDI (Beck Depression Inventory), MADRS (Montgomery-Asberg Depression Rating Scale), CGI (Clinical Global Impression) or/and PHQ-9 (PatientHealthQuestionnaire-9) in these studies. The basic values of these criteria were similar between PP and control groups in these included studies (Supplementary, Table 1).  DCP included four weekly class sessions devoted to resilience education and diabetes self-management, followed by eight biweekly support group meetings. Psychosocial process variables (resilience, coping strategies, diabetes empowerment), and proximal (perceived stress, depressive symptoms, diabetes self-management) and distal outcomes W1: The resilience model, the stress response, problem focused coping strategies, emotion focused coping strategies. W2: The responsibility model, above the line/below line behaviour, circle of influence /circle of concern. Five step process to move above the line. W3: focusing on empowering interpretations, how our thinking affects our health, ABCDE thinking model. The origin of beliefs. W4: Creating meaningful connections, the healing power of love and intimacy. Self-leadership and the circle of influence. Features of psychological thriving.

Focus of Intervention Interventionist(s) Module addressed/components Program Phase
Retention and Acceptability of the Intervention behaviour and not. Includes relaxation training.
*Parentheses indicate that classification was done by GR, RS as no model was mentioned.

Effects on Positive Appraisal
Pooled analyses of two studies (Nowlan et al., 2016; Yalcin et al., 2008) (n=62) reported no significant mean difference in positive appraisal between pre and post-test (Z=1.63, p=0.10), which corresponds to an effect size of -2.455 (ranged from -3.1594 to -1.7999). Heterogeneity was significant (χ 2 =86.28, I 2 =99%, p <0.001). After removal of the outliers, the heterogeneity was low and there was a significant improvement in the positive appraisal (Z=4.92, p<0.001) ( Table 4).  48) indicated no significant mean difference in self-efficacy between the groups (Table 5).

Subgroup Analyses
The subgroup analyses were carried out to examine the difference in the delivery setting (individual versus group intervention), intervention delivered by nurses versus physicians or psychologists, male versus female, age <60 years old versus age ≥60 years old. However, due to the nature of studies, subgroup analyses were carried out to only two primary outcomes, i.e., health behaviour adherence and individual group. Two studies [42,47] (n=38 participants) analyzed the mean change of individual group after PP intervention and control group immediately. The results showed there was no significant heterogeneity (χ 2 =0.06, I 2 =0%, p=0.80) among these included studies (Hedge's g=-0.245, 95% CI: -3.1594 to -1.7999). The pooled estimate (MD=1.01, (-0.45, 2.47), Z=1. 36, p=0.17) showed that there is difference between pre and post-test. Two studies [42,45] (n=112 participants) analyzed the mean change of health behaviour adherence after pre-test immediately. The results showed there was no significant heterogeneity (χ 2 =0.28, I 2 =0%, p=0.60>0.05) among these included studies. The pooled estimate (MD=-1.28, (-1.99, -0.58), Z=3.57, p=0.004) indicated that there is a significant mean difference in stress between pre and post-test (Hedge's g=-0.64 95% CI: -1.2191 to -0.0940) ( Table 6).

Discussion
To our knowledge, this is the first meta-analysis examining the effects of multicomponent PP interventions on well-being and health behaviour adherence amongst patients with T2DM. Our findings indicate that PPIs have a small but significant effect on well-being, especially positive affect and optimism at post-intervention. In addition, a small but significant effect size was found for health behaviour adherence, body mass index and self-care at post-intervention. Effect sizes for anxiety and stress were also significant. These findings show that PPIs have a potential not only to improve well-being but also help for self-care and health behaviour adherence and reduction in anxiety and stress. However, our meta-analyses did not show any effect on other well-being measures such as positive appraisal, self-efficacy, QoL, nor on HbA1c, or health behaviour adherence measures such as diet and physical activity adherence, and depression score in both at post-intervention or in comparison with controls.
The previous meta-analysis did report a significant substantially larger effect sizes of PPIs on subjective, psychological wellbeing, depression [52] and anxiety [52]. However, in contrast, the present meta-analysis did not find any association with depression. This may be perhaps due to the inclusion of the low-quality studies and possibly the use of a wide variety of tools to measure depression. Although we attempted to identify eight hundred and eighty-eight titles which met our initial search strategy but only eleven studies met our inclusion criteria, focusing specifically on enhancing well-being and health behaviour adherence with T2D. Majority of the studies that excluded during the title and abstract screening were studies that focused on a treatment approach for diagnosed depression among T2D. This suggests that to date, research has overwhelmingly emphasized alleviating depressive symptoms rather than enhancing wellbeing. Still, our analysis allowed us to hone on the key elements of preventative approaches, namely PP intervention incorporating diabetes self-management, problem-solving and resilience-focused approaches.
Positive psychological interventions have several advantages compared to other treatment programs. Firstly, exercises within positive psychological interventions tend to be shorter and could be finished independently. Secondly, as opposed to other treatments that are applied to patients with clinical depression, PP interventions are used instead of increased positive psychological wellbeing across the different population. Thirdly, this does not require the substantial provider training, specifically targets positive and optimism constructs, thereby linked to superior adherence and outcomes in T2DM, and there is no need for an in-person session. Fourthly, positive psychological interventions in a specific focus on behaviours and positive traits (for instance, optimism and positive effects) that have been linked with an increase in participation within health behaviours and superior medical outcomes [53,54]. Finally, positive psychological interventions exercises are validated, which target particular positive constructs other than self-efficacy or optimism which is associated with outcomes that are advantageous [55,56].
Yet there have not been many studies that extensively compared the effectiveness of an extensive range of PPIs on improving the subjective well-being and health behaviour adherence amongst patients with T2DM, especially from Western perspective. The western perspective would be a starting point to identify concepts, theories and methodological basis. This in turn would enable researchers from non-western countries such as India to develop culturally based constructs and intervention by mapping from Upanishads, Bhagavad Gita and examining historical, religious and sociological texts in conjunction with the scientific literature. Our strength of meta-analyses lay in the fact that the analysis comprised of evaluating the findings from diverse studies which allowed aggregation of information resulting in a superior statistical power with a strong point estimate. This is in stark contrast to any analysis that would be executed on the basis of a single stand-alone study. However, the findings should be interpreted within the scope of a few limitations. Firstly, the quality of the studies was not high, given the scope of this review. Secondly, evidence was not strong enough due to the small sample size. More studies with larger sample size should be performed to verify these results. Thirdly, most of the studies conducted the completers-only analysis, as opposed to intention-to-treat analysis, which in turn could have resulted in seriously biased results [57]. Fourthly, the significant heterogeneity among studies was still present, due to the mixed nature of the intervention (e.g. two studies that focused on resilience plus nutrition state, one study on emotional intelligence, self-efficacy (2 studies) and activation state). Despite strict inclusion criteria applied, different types of PP interventions are clubbed together and hence resulted in a high level of heterogeneity.
In this meta-analysis, there was little variation in regard to the intervention module, and program phase and most of the studies did not report on adherence. Studies had reported quite low adherence in self-help interventions [58] Therefore, it does not provide firm indications on how to improve its effectiveness. Although the timeline for outcome measurement was objectively defined (i.e., 6 months), due to the variability in the choice of the instrument across studies, which in turn could lead differences in terms of effects. The future meta-analysis should be carried out to assess the robustness of the results based on tools of measurement, different measures of effect size (risk ratio and odds ratio), and different statistical models (fixed-effect and random effect models). Further, in the present meta-analysis, a subgroup-analysis was not carried out on specific interventions type due to a small number of studies and too diverse studies. Future studies need more RCT to draw a firmer conclusion.

Conclusion
In conclusion, this scoping review and meta-analysis provide evidence that PPIs are effective in improving well-being and health behaviour adherence among patients with T2DM. Further, the review also demonstrates that PP intervention can be effective in the reduction of anxiety and stress symptoms. Practitioners can customize their treatment strategy based on the client needs and preferences in conjunction with other evidence-based interventions that enhance well-being such as mindfulness intervention [59], behavioural activation [60], reminiscence [61], acceptance and commitment therapy [62] and forgiveness interventions [63]. From the perspective of public health, PP interventions can be used as a preventive approach and non-stigmatized tool in two ways, such as promoting mental health through health internet portals and leaflets distribution. Secondly, PP can be part of the first step in a stepped care approach, which starts with low intensity at the same time low, or no costs empirically-based intervention [64,65] that boost total well-being. In case of failure in the first step, participants can be referred to specialized care for more intensive treatment. However, given that study quality was not assessed (e.g. using GRADE or a similar tool), and variation in study design, the effectiveness of these results needs to be interpreted cautiously due to the nature of publication bias and also accounting the limitations discussed above.

Recommendations for Further Research
Given that PP interventions are still evolving, there is a need for more high-quality and well-conducted studies across the different clinical population, in different age groups (e.g. older adults) with a different set of intervention formats customized based on culture. Majority of the studies have been conducted amongst the western population, and not many have been carried out in non-western or Asian population where more than one-half of the patients with diabetes in the world live. Only countable studies had focused on PP while majority delivered the program in conjunction with problem-based preventive interventions multi-component models; therefore, it's still not clear whether the impact PP on the outcome is due to multicomponent. In future, more studies need to be conducted by adapting the existing culture (e.g. See Martinez et al., [66]). Lastly, diabetes poses a huge economic burden to the society and to an individual patient; therefore, future studies need to account cost-effectiveness of conducting such interventions aiming to establish the society and public impact.