A systematic review of diabetes self‐management education interventions for people with type 2 diabetes mellitus in the Asian Western Pacific (AWP) region

Abstract Aims and objectives To assess the effectiveness of educational and/or psychological diabetes self‐management education (DSME) intervention for people with type 2 diabetes (T2DM) in the Asian Western Pacific (AWP) region. Background Translational research indicates that DSME is effective; therefore, it is important to look at the AWP region to see what has been implemented and what the potential barriers are for the low integration of DSME. The need for DSME is present, and programmes are being developed. Therefore, focusing a systematic review of DSME research in the AWP region would give a better understanding of which intervention approaches are associated with better clinical outcomes and are culturally acceptable. Design A systematic review. Methods A review of randomized controlled trials (RCTs) and comparative studies to evaluate the effectiveness of face‐to‐face delivery reporting educational and/or psychological interventions for people with T2DM was implemented. We conducted searches using MEDLINE, EMBASE, CINAHL, PubMed and ASSIA databases between January 1990–June 2018. Studies published in English and non‐English were included. Two reviewers independently extracted data on participant and intervention characteristics. The quality of evidence was rated on predetermined criteria. Main outcomes included glycaemic control (reduction in HbA1c level). Results We included 21 DSME programmes (17 RCTs), while 15 were group‐based approaches. Twelve studies (60%) were categorized as high quality. Three studies (25%) had a moderate (good) effect. Eight trials were effective in improving glycaemic control and reported statistically significant improvements in HbA1c levels. 50% of these were high‐intensity group‐based programmes.


| BACKG ROU N D
Diabetes self-management education (DSME) provides information and skills needed by people with T2DM to effectively self-manage their diabetes, in addition to medical management, to achieve optimum glycaemic control (Funnell & Anderson, 2004). Developed countries such as the United Kingdom (UK) and United States of America (USA) have successfully integrated DSME into their health systems, both have national guidelines for DSME provision (Haas et al., 2012;NICE Internal Clinical Guidelines Team, 2015).
In Western cultures, DSME is designed by the guiding principles: (a) it is effective for improving clinical outcomes and quality of life; (b) it evolves from theoretically based empowerment models; (c) although there is no "best" educational approach, culturally and age appropriate programmes incorporating behavioural and psychosocial strategies demonstrate improved outcomes; (d) ongoing support is critical to sustain progress and; (e) behavioural goal setting is an effective strategy (Funnell et al., 2008). Programmes such as diabetes education and self-management for ongoing and newly diagnosed (DESMOND) are associated with improved clinical outcomes (Davies et al., 2008), and the Diabetes Self-Management Program (DSMP) demonstrated improvement in depression, healthy eating habits, more effective patient-health provider relationship, communication and self-efficacy (Lorig, Ritter, Villa, & Armas, 2009) in the UK and USA, respectively. Meanwhile, also in the USA, the Diabetes Empowerment Education Program (DEEP) for Latinos reported significant improvement in glycaemic control (Castillo et al., 2010).
In 2014, the International Diabetes Federation's (IDF) global diabetes scorecard reported that less than 1% of countries in AWP region integrate DSME in their health services (International Diabetes Federation, 2013). Previous research suggests that DSME is effective; therefore, it is important to apply these findings to the AWP region, taking into account what has been implemented previously, as well as looking at what the potential reasons are for this low integration of DSME. The need for DSME is present, and programmes are being developed. Therefore, focusing a systematic review of DSME research in the AWP region would give a better understanding of which intervention approaches are associated with better clinical outcomes. The aim of this review was to synthesize the evidence for DSME programmes employing educational and/or psychological interventions for people with T2DM in the AWP region, looking at whether culturally specific techniques or sessions are incorporated.
The results from this study may direct the local health providers to develop tailored DSME programmes to suit the diverse local population.

| ME THODS
This systematic review did not require patient consent or Research Ethics Committee approval.  Table 1 demonstrates the search strategy and keywords used ("diabetes mellitus" and "diabetes education"). Exploded keywords were included and MESH terms for MEDLINE and modified truncation according to the different search platforms.

| Selection criteria
This systematic review included comparative studies. This is defined as RCTs, non-RCTs and observational studies that used a comparison group. The broad inclusion criteria ensured all studies measuring effectiveness of DSME in adults aged 18 and over with T2DM in different healthcare settings were included. No limit was made on the language of publication. Eligible papers that were written in a different language were translated. Studies where participants were diagnosed with type 1 diabetes mellitus (T1DM), gestational diabetes and a mixture of T2DM with T1DM or other chronic conditions were excluded. Furthermore, studies solely investigating pharmacological or medication adherence, diet, exercise, physical activity, web-based or peer support or telephone counselling were excluded to reduce confounding bias (Inzucchi et al., 2012). Only studies that covered more than one component of diabetes self-management were included. Besides, only studies implemented exercising face-to-face delivery approach were included as discussed in a recent review regarding effectiveness of intervention using information technology and it suggested they were not as effective as a face-to-face method (Pillay et al., 2015).
The American Association of Diabetes Educators (AADE) defines DSME as "the ongoing process of facilitating the knowledge, skill and ability necessary for prediabetes and diabetes selfcare" (Haas et al., 2012, p.620). The DSME programme employs a patient-centred approach as it helps people with T2DM to change their behaviour and achieve seven specific self-care behaviours: healthy eating, being active, monitoring, taking medication, problem-solving, healthy coping and reducing risks. DSME interventions were classified into (a) being predominantly educational interventions; or (b) predominantly psychological interventions (which include mental health). Educational interventions were defined as those that provide information on diabetes, its causes and management (medications or self-management) (Verkuijlen, Verhaak, Nelen, Wilkinson, & Farquhar, 2014) and may include didactic and facilitative teaching approaches. Didactic teaching is a traditional lecture-based teaching methodology that is teacher centred, while facilitative teaching refers to a learner-centred approach and is more flexible combining teaching methodology with practical sessions such as exercise classes and telephone followups (Prince & Felder, 2006). In addition, the facilitative teaching approach may be underpinned by behavioural theory whilst the intervention is being developed (Jackson, 1997).
Psychological DSME interventions focus on the therapeutic alliance between the therapist and the person with diabetes to improve their bio-psychosocial outcomes (Smith, 2012). We classified psychological interventions into the common psychotherapeutic models used in healthcare settings: (a) supportive or counselling therapy (Rogers, 1976); (b) cognitive behaviour therapy (Beck, 1976); (c) brief psychodynamic psychotherapy (Malan, 1963); and (d) interpersonal therapy (Klerman, Weissman, Rounsaville, & Chevron, 1984). Studies where the model of the intervention was unclear were included if they used one or more psychological techniques that could be coded into the above classification. Techniques such as relaxation, activity scheduling, problem-solving, goal setting, contract setting, cognitive restructuring and stress management were classified as cognitive behaviour therapy (CBT). Techniques such as motivational interviewing and non-directive counselling were classified under counselling therapy. We reported only one analysis for studies with several intervention groups with the most intensive intervention as the experimental one. Intensity was defined by approach (most intensive was group based rather than individual intervention), type of DSME intervention (most intense was psychological followed by educational), number of sessions and duration of the intervention.

| Data extraction
The first reviewer (AM) screened all titles from the searches to exclude studies that were irrelevant. Following this, three reviewers (AM, KW and ES) independently screened the title and abstracts using an eligibility checklist. The full texts of the potentially eligible studies were retrieved for full review and final selection. Studies TA B L E 1 Detailed search strategies for the systematic review of Diabetes Self-Management Education (DSME) interventions for people with type 2 diabetes mellitus (T2DM) in the Asian Western Pacific (AWP) region

| Quality assessment
The quality of the studies was assessed using the Jadad score (Jadad et al., 1996) by the first and second reviewers. The quality of included studies was assessed according to 3 appraisal elements: (a) selection bias (randomization procedure and allocation concealment); (b) blinding (masking of outcome assessor but not participants and therapist because DSME intervention cannot be concealed); and (c) attrition bias (withdrawals or dropouts). Studies were then scored on a scale between 0-5. Studies scoring greater than 3 demonstrated high quality.

| Data synthesis and analysis
Overview and characteristics of included studies are presented in summary table (Table 2). Meta-analysis was not conducted due to the heterogeneity of the intervention programmes, populations and outcome measurement. The primary outcome was improvement in glycaemic control (HbA1c, % or mmol/mol). Cohen's d effect size of HbA1c results was used to measure the magnitude of the difference in the outcome between the intervention and control groups, where a value of 0.2 represents a small effect size, 0.5 represents a moderate effect size, and 0.8 represents a large effect size (Cohen, 1988).
The secondary outcomes were other metabolic control measures, such as body mass index (BMI) (kg/m 2 ), cholesterol level (mmol/L), fasting plasma glucose level (mmol/L) and blood pressure (mmHg), as well as psychosocial variables such as self-reported quality of life, self-efficacy and level of depression.

| RE SULTS
A total of 1,744 non-duplicated publications were screened, 151 abstracts were assessed for eligibility, and 43 publications required full-text review before a decision could be made. Twenty-one studies fulfilled the inclusion criteria, and the search processes are illustrated in Figure 1. The interventions varied considerably according to the number and duration of sessions; however, the content was mostly similar focusing on diabetes self-management.
The duration of interventions ranged from 6 weeks-18 months.
Of the 21 studies, there were several clinical subgroups among the populations studied: (a) 2 studies (10%) were conducted among elderly people; (b) 4 (19%) included people with less than 5 years duration of T2DM; (c) 1 (5%) was conducted among overweight people with T2DM (BMI > 23kgm 3 ); (d) 7 (33%) were implemented among TA B L E 2 Overview of the eligible studies examining the effects of Diabetes Self-Management Education (DSME) interventions for people with type 2 diabetes mellitus (T2DM) in the Asian Western Pacific (AWP) region

Regimen in intervention group and speciality of therapist
Campbell/ Australia/ RCT/1996 (Campbell et al., 1996)  people with T2DM with sub-optimal diabetes control; and the remaining 7 (33%) did not specify a clinical sub-group.

| Quality assessment
In general, most of the studies included were of reasonable methodological quality and the abstract of the included studies was able to provide adequate information particularly on the aims, methods and findings of each study. Of twenty-one studies, 12 (60%, N = 21) of the reviews scored three and above against the Jadad score, which were then categorized as high in quality. Table 3 illustrates the elements of the Jadad score for the included studies. Six studies provided limited information on the randomization strategies used (Campbell et al., 1996;Shibayama et al., 2007;Sun et al., 2008;Wei et al., 2008;Yang et al., 2007;Zhou et al., 2011).
It was noted that none of the trials explicitly explained appropriate methods of blinding, although the majority did describe numbers of withdrawals and dropouts. In most of the studies, the data were analysed using the intention-to-treat principle, which helps to preserve the sample size, which is an important criterion for statistical power.
For the 15 studies with an educational intervention, the majority of the included studies were RCTs (60%; 10 studies) where the studies were relatively balanced; four studies were implemented in a clinical setting (Chao et al., 2015;Guo et al., 2014;Jaipakdee et al., 2015;Sone et al., 2010) and five studies in the community (Krass et al., 2007;Sun et al., 2008;Wei et al., 2008;Yuan et al., 2014;Zhou et al., 2011). Only one study was implemented in both clinical and community settings (Campbell et al., 1996). Six studies were classified as high quality by the Jadad score (Chao et al., 2015;Guo et al., 2014;Jaipakdee et al., 2015;Krass et al., 2007;Sone et al., 2010;Yuan et al., 2014). Of 6 studies with a psychological intervention, the majority were also RCTs (83%; 5 studies) and conducted in a clinical setting (Moriyama et al., 2009;Shi et al., 2010;Shibayama et al., 2007;Tan et al., 2011), with only one study taking place in the community . Four included studies were considered high quality by the Jadad score Moriyama et al., 2009;Shi et al., 2010;Tan et al., 2011).
Significant scores were observed in the following measures: the 15-item diabetes knowledge scale (DKNA) (Chao et al., 2015;Tan et al., 2011); SMBG (Chao et al., 2015;Tan et al., 2011); MMAS (Guo et al., 2014); the Diabetes Management Self-Efficacy Scale (DMSES) (Guo et al., 2014;Shi et al., 2010); and the Summary of Diabetes Self-Care Activities (SDSCA) (Guo et al., 2014), which were all found in both educational and psychological RCTs implemented in a clinical setting and rated as high quality. Additionally, significant scores were observed on the quality of life (QOL) measures among RCTs using an educational approach in a clinical setting (Jaipakdee et al., 2015) as well as an educational intervention of retrospective cohort studies in a community setting (Roberts et al., 2017;Zhou et al., 2011).

| Impact of the educational and psychological interventions on glycaemic control
Sixteen studies examined the HbA1c (%) as an outcome to measure the impact of DSME (educational and psychological) interventions.

| Impact of the educational and psychological interventions on psychological well-being, diabetes knowledge and self-management
Two studies (Campbell et al., 1996;Chao et al., 2015) measured diabetes knowledge using validated questionnaires. These include the DKNA and the Revised Diabetes Self-care Activities (RDSA) questionnaires modified from the Diabetes Self-care Activities Questionnaire used in Malaysia. Only one (Chao et al., 2015) demonstrated a significant difference between the intervention and the control group. Six studies (Chao et al., 2015;Guo et al., 2014;Ng & Sim, 2014;Shi et al., 2010;Song et al., 2012;Tan et al., 2011) measured at least one aspect of diabetes self-management. However, only four studies (Chao et al., 2015;Guo et al., 2014;Song et al., 2012;Tan et al., 2011) demonstrated statistically significant improvement in the intervention group.     Of 6 studies, four studies assessing psychosocial self-efficacy have shown statistically significant improvement (Guo et al., 2014;Liu et al., 2012;Moriyama et al., 2009;Shi et al., 2010) and only 1 (Moriyama et al., 2009) of these was carried out using an individual approach, while the other three were group based (Guo et al., 2014;Liu et al., 2012;Shi et al., 2010). Five studies measured the quality of life using validated questionnaires (EQ5D, WHO-QOL26, SF-36 and Chinese version of DQOL), with three of these reporting statistically significant results: one study used an individual psychological intervention (Moriyama et al., 2009) and 2 studies used group-based educational (didactic and facilitative) teaching (Jaipakdee et al., 2015;Zhou et al., 2011). However, the other two studies (Krass et al., 2007;Shibayama et al., 2007) reported no difference in the quality of life outcomes and were individual educational and psychological interventions, respectively.

| Impact of additional components, setting and speciality of the educators on glycaemic control
There were four studies (Campbell et al., 1996;Chao et al., 2015;Jaipakdee et al., 2015;Song et al., 2012) which integrated practical sessions such as exercise classes and healthy diet preparation into the DSME intervention. Only two of four studies (Campbell et al., 1996;Song et al., 2012) reported a moderate effect size for HbA1c.

| Integration of cultural sensitivity in the interventions
Most of the studies identified by this review have involved the delivery of educational, self-management and/or psychological interventions for people with T2DM from countries in the AWP region. However, from the examination of the intervention characteristics, none of them has specified specific cultural adaptations to address the needs of local individuals to support diabetes selfmanagement. The only cultural concessions made and reported were 17 studies mainly conducted in the East and South-East Asian countries which included the translation of self-report outcome measurements, such as validated questionnaires and delivery of interventions in native languages (Chao et al., 2015;Guo et al., 2014;Jaipakdee et al., 2015;Liu et al., 2012;Moriyama et al., 2009;Ng & Sim, 2014;Shi et al., 2010;Shibayama et al., 2007;Sone et al., 2010;Song et al., 2012;Sun et al., 2008;Tan et al., 2011;Wei et al., 2008;Wong et al., 2014;Yang et al., 2007;Yuan et al., 2014;Zhou et al., 2011).

| D ISCUSS I ON
It is important that people with T2DM undertake adequate selfmanagement to optimize blood glucose levels which may reduce and delay diabetes-related complications (UK Prospective Diabetes Study Group, 1998). Therefore, DSME is necessary to support people with T2DM to develop effective diabetes self-management skills. DSME is well integrated in Western developed countries, and this review highlights the growing number of programmes being developed in the AWP region and provides some evidence that DSME is effective in improving glycaemic control for people with T2DM living in this area. A total of 21 studies were identified which used interventions with various educational or psychological therapeutic approaches and modes of delivery.
It was found that most group-based DSME interventions provided a good effect on glycaemic control compared with one-to-one interventions, particularly programmes conducted for 10 hr or more (high-intensity programmes). Interventions that integrated practical sessions reported an exceptional clinical improvement in glycaemic control (moderate effect size Cohen's d > 0.5). There was a trend from 2007 onwards towards multidimensional interventions (involving facilitative teaching and psychological elements) rather than on didactic teaching alone. There is little evidence to recommend a specific theoretical model as the most effective for DSME from the available analysis; however, self-efficacy theory was widely used.
Hence, we have little understanding of how intervention components promote behavioural modification or lifestyle change which may help improve clinical outcomes as only one-third of our included studies measured at least one aspect of diabetes self-management.
Despite the small proportion of studies, they generally reported positive effects for glycaemic control.
No direct comparison can be made between the present systematic review and any other review conducted in the AWP region as there are none. However, the results from this review are consistent with the review by Chrvala, Sherr, and Lipman (2016) who reported that a combination of group and active participation in DSME improve diabetes management and outcomes in Western countries.
This review is also consistent with the review by Steed, Cooke, and Newman (2003) who reported that didactic teaching approaches alone had less overall effect on glycaemic control and Norris, Engelgau, and Narayan (2001) who concluded that interventions incorporating "hands-on" sessions were more effective than didactic approaches. The finding that high-intensity programmes appeared to be more beneficial supports the review by Pillay et al. (2015) who revealed that DSME with less than 10 hr of sessions (less intensive) is less effective compared with more intensive sessions. Another systematic review and meta-analysis by Steinsbekk, Rygg, Lisulo, Rise, and Fretheim (2012) synthesizing DSME RCTs concluded that interventions conducted with longer hours (more than 12 hr and between 6-10 sessions) have proven to be more beneficial in optimizing blood glucose level.
It is worth noting that multidisciplinary teams of DSME which involve more than one member of health professionals may contribute to effectiveness, though this has yet to be confirmed in RCTs and observational studies according to a review by Chrvala et al. (2016). It was unclear what the specific cultural elements of the included studies are, but all of the non-English speaking countries implemented the DSME interventions in their mother tongue and the questionnaires used to evaluate the programmes were translated and validated. Previous systematic reviews of culturally tailored DSME conducted among minority ethnics in Western nations reported a positive impact on behaviour change and glycaemic control when it is linguistically acceptable (Pillay et al., 2015).

| Limitations
There is a possibility of publication bias as only published data were included. Fourteen studies (64%) included in the analysis had less than 12 months of follow-up data; therefore, long-term outcomes could not be assessed. In addition, the assessment of psychosocial outcomes in all studies was based on self-report questionnaires which may introduce some bias. This review included non-randomized study designs and was carried out using a discursive analysis rather than meta-analysis, due to the heterogeneity in both the interventions and outcome measures, as well as the difference in populations and settings. In terms of outcomes, this review mainly focuses on glycaemic control (HbA1c), due to the issue of extracting or calculating the effect size when the secondary outcomes were made using different measures, such as for diabetes knowledge, quality of life, self-efficacy, adherence and physical activity.
Positive psychology is a relatively new approach that has been increasingly used as a promising technique to promote health; however, we did not encounter this approach in our search. In future searches, it might be useful to define this in the search strategy to see whether this changes the inclusion of positive psychology approaches.
Despite the Jadad scoring system being simple and easy to use with known reliability and external validity, there are some flaws in the scale. It should be noted that the scoring system does not address the appropriateness of data analysis or allocation concealment (which is one of the parameters to avoid bias in research), or the assessment of intention-to-treat analysis. It also focuses on blinding, which is challenging in RCTs with complex interventions.
Although blinding is gold standard in research design, blinding in most complex intervention RCTs is often not feasible. Complex intervention research such as DSME intervention trials is mainly conducted to determine effectiveness rather than efficacy in the traditional RCTs and often measure complex outcomes as well as consisting of multiple interactive elements that make it challenging to blind (Mustafa, 2017). Although randomization was implemented, there is still a potential of confounding bias, either consciously or unconsciously, which may have a negative impact on the integrity of the complex intervention trials.

| CON CLUS ION
This review identified and summarized the available evidence in the AWP region from the 21 studies regarding the effectiveness of DSME to improve diabetes self-management. The results suggest that overall group-based DSME is associated with improved clinical and psychosocial outcomes and interventions underpinned by behavioural theory with longer contact hours and the inclusion of active, hands-on participatory sessions may maximize the potential benefit of these programmes. Likewise, involving the participation of the multidisciplinary team may also be important. However, what we do not yet know is how to target DSME in this region so that it is culturally appropriate and whether beliefs and attitudes towards diabetes in ethnically diverse AWP communities are being addressed and how this is achieved, which suggests more research is needed.
We can conclude that standardized DSME programmes or specific guidelines in this region were limited; therefore, there is an urgent need to develop DSME and make it accessible to people with T2DM among AWP countries.

| RELE VAN CE TO CLINI C AL PR AC TI CE
This review may guide healthcare providers or policymakers in designing future culturally tailored programmes for people with T2DM in the AWP region. Successful programmes are likely to be group based, include active participation and longer contact hours, but more research is needed to determine how to address specific cultural beliefs and attitudes towards diabetes. For example, the small feast culture appears to be one of the biggest barriers to maintaining the desired glucose level in Asian cultures. This happens infrequently in Western cultures, where meals are only offered to close friends and family (Douglas, 1975). Our research aligned with an ethnographic study among Middle Eastern people in the United Arab Emirates (UAE) which demonstrated that providing meals to strangers, guests and friends is a way to minimize the social gap (Baglar, 2013). It is considered polite to finish food served to you in Asian cultures and this often hinders efforts at dietary modification.
Therefore, the following should be considered when conducting future research: (a) interventions to be tested using randomized trials; (b) interventions incorporating "hands-on" sessions and psychological techniques, problem-solving and goal setting; (c) additional outcome measurements of behavioural change and coping skills; (d) explicit use of culturally relevant materials; and (e) and cost-effectiveness analysis.