Background
Diabetes is a common chronic disease worldwide; in 2011, it affected 366 million people [
1,
2]. It is estimated that 592 million people will have the condition by 2035 [
3]. This growing prevalence is related to increasing economic growth, urbanization, and lifestyle alteration characterized by risk factors such as obesity and sedentary activity [
4,
5]. Prediabetes, i.e. impaired fasting blood glucose or impaired glucose tolerance, often occurs about 5 years before the development of type 2 diabetes mellitus (T2DM) [
6]. People with prediabetes are at 5 to 15% greater risk of progression to T2DM [
7].
Large randomized controlled trials (RCTs) from Finland, India, China, and the US have shown that lifestyle interventions can decrease the incidence of T2DM from 58 to 29% in high-risk populations [
8‐
11], with maintenance up to 20 years [
12,
13]. Currently, 7 of the top 10 countries with the greatest number of people living with diabetes are low- and middle-income countries (LMICs), including China, India, Brazil, Pakistan, Indonesia, and Bangladesh [
3,
14‐
17]. In the last decade, the prevalence of diabetes has escalated more in LMICs compared to high-income countries (HICs). The numbers of people with diabetes differ substantially by country income group [
3]. By 2030, the number of people with diabetes is projected to increase by 92% in low-income countries, 57% in lower-middle income countries, 46% in upper-middle income countries, and 25% in HICs. The rate of increase is inversely associated with the current income status of countries [
3].
One of the challenges in many countries is to establish effective and low-cost interventions to prevent the development of T2DM that can be successfully implemented and sustained [
18]. Community-based programs are practical, relatively low-resource, and often involve educational programs aiming at lifestyle change. They address various aspects of health, including diet, physical activity, and health behaviors and have demonstrated significant benefits for the management of chronic diseases such as cardiovascular disease (CVD) and T2DM [
19]. Community-based programs are appealing, since they can reach people outside of conventional healthcare settings, and usually target all groups in the community. If a program succeeds with a positive effect on behavioral change, it probably can achieve considerable and widespread risk reduction in the community [
20]. In recent years, several systematic reviews have reported on the positive effects of diabetes prevention programs, mainly in HICs [
20‐
22], but are now drawing more attention in LMICs. The data encourage policymakers at local and national levels to collaborate in order to mitigate the rising prevalence of T2DM in their populations. Community-based programs mainly provide health behavioral interventions through group-based (and sometimes also individual) educational sessions to achieve health-related behavioral modification in the communities. Since busy healthcare settings do not have adequate capacity to offer intensive behavioral interventions [
23], programs with greater expertise and resources to offer effective interventions and enhance health-related behavioral modification should be offered to the whole community [
24].
Three landmark clinical trials of people at risk of diabetes have indicated a remarkable relative risk reduction in the progression to T2DM with health-related behavioral modification [
11‐
13,
25‐
27]. However, it is still unclear how these behavioral changes influence the risk of complications related to T2DM. Despite the continued development and use of community-based programs, there is currently limited evidence to support or refute their effectiveness in LMICs. Most studies reported in recent systematic reviews have been based in HICs, while few studies based in LMICs have been performed [
21,
22,
28]. One review did cover diabetes prevention programs in developing countries [
29], but the researchers performed literature searches only until September 2009. To our knowledge, few reviewers to date have tried to assess the effectiveness of community-based programs in risk reduction of T2DM in LMICs. Generalizing evidence from HICs to LMICs needs to be considered with caution given cultural, ethnic, and economic differences, as well as the differences in targeted populations [
30]. This systematic review and meta-analysis aims to review the last decade of evidence on the effectiveness of community-based programs to prevent or reduce the risk of developing T2DM in at-risk populations in LMICs.
Discussion
This review summarized the evidence of six randomized controlled trials conducted in the last 10 years with 2574 participants analyzed to assess the effect of community-based interventions on the primary prevention of T2DM in low- and middle-income countries (LMICs). Overall, these results are optimistic that community-based interventions can modify several risk factors for T2DM, including anthropometric indices (weight, BMI, and waist circumference) and glycemic control (fasting blood glucose and HbA1C), and may be effective in risk reduction of T2DM in communities as well. Our findings indicated that the risk of developing T2DM (the incidence rate of diabetes) was lower in intervention groups, with a relative risk reduction of 0.57 (95% CI; 0.03 to 1.06). Therefore, 77 per 1000 fewer people developed diabetes after participating in these interventions (See Fig.
5). However, this moderate-quality evidence was not statistically significant, probably due to the heterogeneity that resulted in the wide confidence intervals.
In anthropometric indices, our analysis demonstrates that community-based programs probably have a beneficial effect on weight, BMI, and waist circumference. The pooled results of moderate-quality evidence showed a significant reduction of 2.3 kg weight (1.19 to 3.4 kg lower). In addition, our results revealed a reduction of 1.27 kg/m2 BMI (0.44 to 2.1 kg/m2 lower) and 1.66 cm in waist circumference. However, the quality of the evidence was low in these to outcomes due to inconsistency among the results of our studies and risk of bias in some of them. Two glycemic control outcomes also showed a significant improvement. HbA1C and FBS reduced significantly by 1.17% (1.51 to 0.82 lower) and 5.3 mg/dl (1.98 to 8.69 lower), respectively. The effect of community-based programs on blood pressure was limited in our review with wide thresholds that did not exclude appreciable benefit or harm.
The observed heterogeneity in the pooled effects of some outcomes is probably related to differences in methods and duration of the interventions. The interventions in our included studies used a variety of approaches ranging from physical activity and dietary interventions to lifestyle education through cell-phone text messages. The difference in the intensity of the interventions is expected to affect their effectiveness in participants’ lifestyle change. The duration of these interventions varied from 6 months to 2 years. Due to the low number of RCTs, we could not perform a subgroup analysis based on the duration of interventions to compare the effect of long-term interventions with short-term ones.
Although our review addresses the effect of community-based programs in populations at risk of diabetes only in LMICs, it highlighted several remarkable findings that confirm earlier reviews of lifestyle-related interventions on the prevention or risk reduction of T2DM in the world (HICs and LMICs). In 2012 Rawal L.B., et al., [
29] assessed the effect of lifestyle-related interventions in developing countries on the incidence rate of diabetes, though their review did not carry out a meta-analysis and only reported the results of each trial. Similar to our findings of a descriptively lower relative risk reduction in intervention groups, they reported that lifestyle interventions can result in significant reductions of risk of the development of T2DM in people with impaired glucose tolerance or impaired fasting glucose. In China [
10,
13], a 42% relative risk reduction of T2DM was reported with a diet and exercise intervention, comparable to our pooled effect T2DM risk reduction of 43%. Kerrison et al. [
72] assessed nine trials and reported that lifestyle interventions reduced the incidence of diabetes more than standard treatment (in 8 of 9 studies reviewed) and increased weight reduction, but no meta-analysis was performed. In relation to anthropometric outcomes, Dunkley et al. (2014) [
28] reviewed 22 experimental and observational studies of lifestyle interventions aimed at risk reduction or prevention of T2DM with the primary outcome of weight change. In contrast to our review, all studies were carried out in HICs (Europe and North America). Their pooled results showed a mean weight loss of 2.32 kg (95% CI [− 2.92, − 1.72]), very similar to our findings indicating a significant loss of 2.3 kg (95% CI [− 3.40, − 1.19]). Zhang [
73] showed a 3.99% reduction in weight (95% CI [− 4.69, − 3.29]) in their review that is also similar to our results. In relation to glycemic control indicators, Qing-Hai Gong in 2015 reviewed nine trials (two from LMICs) and reported that lifestyle modification programs (physical or dietary interventions or both) were associated with significant improvements in 2-h blood glucose (SMD [95% CI]; − 0.56 [− 1.01 to − 0.10],
p = 0.000) and FBS levels (SMD [95% CI]; − 0.27; [− 0.38 to − 0.15],
p = 0.042) in patients with impaired glucose tolerance (IGT) [
74]. Their result is in contrast with ours in relation to the significant decline of 2-h blood glucose. However, the FBS pooled result is comparable to our study findings. A review by Zhang [
73] had similar findings showing lifestyle interventions reduced FBS and HbA1C significantly with FBS (MD [95% CI]; − 0.14 mmol/L [− 0.19, − 0.10]), and HbA1c (MD [95% CI]; − 0.06% [− 0.09, − 0.03]). One systematic review [
75] including eight trials found that interventions combining physical activity and diet or behavioral modification in LMICs significantly reduced both the systolic blood pressure (SBP) (MD; 95%CI, − 6.1 mmHg; − 8.9 to − 3.3) and diastolic blood pressure (DBP) (MD; 95% CI -2.4 mmHg; − 3.7 to − 1.1). This result is different from our finding that interventions had no or limited effect on SBP and DBP. However, they found that the interventions were effective in lowering SBP and DBP only in the studies where participants received antihypertensive drugs. In our review, none of the three trials which reported the effect on blood pressure reported using antihypertensive drugs.
Overall, similar to our results, the previous systematic reviews’ findings suggest that behavioral, educational, or lifestyle modification interventions – such as community-based interventions – maybe effective to prevent or reduce the risk of T2DM in many countries, including LMICs.
Limitations and potential biases in the review
We conducted our review based on the recommendations provided in the Cochrane Handbook for Systematic Review of Interventions [
33] with a comprehensive literature search across major databases and two trial registries to identify published and unpublished trials. However, we did not conduct a broad search for gray literature and we limited our search results to English language literature. We planned to use funnel plots to evaluate potential publication bias if we included at least 10 trials, but due to the limited number of included RCTs (6 studies), we were not able to produce a funnel plot to assess publication bias that is the study limitation. Some of the included trials were relatively small and mostly described one-year follow-up, and some did not report all of our outcomes of interest. The limited number of included trials that reported each outcome made it impossible to carry out subgroup analysis.
The generalizability of our review is limited by the small number of included trial settings, age and condition of the participants, and method of intervention delivery. Half of the included trials were carried out in India and the other three in China, Brazil, and Vietnam. Thus, our review did not provide evidence across most LMICs. The recruited participants’ ages ranged from 30 to 76 years old. Therefore, our review result may not be applicable to younger adults, though T2DM is relatively rare among that age group. The content of interventions was reasonably similar, but the method of interventions varied from participation in a walking group and educational sessions, to cellphone text messages, to delivery of health behavioral modification. Since educational methods were different in the interventions, we cannot apply their results to all community-based programs.