The Long-Term Effects of Non-Pharmacological Interventions on Diabetes and Chronic Complication Outcomes in Patients With Hyperglycemia: A Systematic Review and Meta-Analysis

Objective This study aimed at examining the long-term effects of non-pharmacological interventions on reducing the diabetes incidence among patients with prediabetes and chronic complications events among patients with hyperglycemia (pre-diabetes and diabetes) by performing a systematic review and meta-analysis of randomized controlled trials (RCTs). Methods PubMed, MEDLINE, EMBASE, the Cochrane Library, and the Web of Science Core Collection were searched for studies published between January 1990 and November 2021, looking for RCTs to evaluate the effects of non-pharmacological interventions on preventing the incidence of diabetes and chronic complications in comparison with medical therapy, placebo, or usual diabetes care. Two independent reviews extracted relevant data and quality assessment. Any discrepancies were resolved by a third reviewer. Results In total, 20 articles involved 16 RCTs (follow-up ranged from 2 to 30 years) were included. Pooled analysis of intervention studies demonstrated clearly that non-pharmacological interventions have a significant effect on reducing the diabetes events in patients with prediabetes (RR 0.62; 95% CI 0.54, 0.71). Pooled analysis of extended follow-up studies showed that non-pharmacological interventions could effectively reduce the diabetes incidence in patients with prediabetes (RR 0.78; 95% CI 0.63, 0.96). Meta-regression and subgroup analysis indicates that the diabetes incidence of the long-term group (duration > 3 years) was clearly reduced by 0.05% compared with the relatively short-term group (duration ≤ 3 years). The incidence of microvascular complications in patients with hyperglycemia was effectively lowered by non-pharmacological interventions (RR 0.60; 95% CI 0.43, 0.83). Conclusion Non-pharmacological interventions have a long-term effect on reducing the diabetes incidence among prediabetic patients and effectively preventing microvascular complications on hyperglycemia. Systematic Review Registration https://www.crd.york.ac.uk/prospero/.


INTRODUCTION
The condition of hyperglycemia is one of the most common chronic metabolic disorders, including prediabetes and diabetes. Prediabetic state is a high-risk group of diabetes and a potential risk factor for diabetes and cardiovascular disease, and impaired glucose tolerance (IGT) is more common than impaired fasting glucose (IFG) (1). According to the American Diabetes Association (ADA) expert group, up to 70% of patients with prediabetes will eventually develop diabetes (2). Without effective intervention, it will be more likely to worsen and develop into diabetes mellitus with macro-and microvascular conditions (3,4). Diabetes complications, particularly macroand microvascular diseases, are the leading cause of reduction in the quality of life of patients and increase in diabetic mortality. Previously large-scale meta-analyses have indicated that prediabetes is associated with an increased risk of cardiovascular diseases (5)(6)(7)(8). Therefore, prevention of diabetes and its severe complications is urgently needed.
Currently, a growing number of clinics and patients pay more attention to non-pharmacological strategies due to the hypoglycemic agents having a limited role in the progression of diabetes and its complications (9). Different non-pharmacological strategies have been reported including lifestyle change, dietary modification, physical activity, and exercise with different intensity, with favorable and unfavorable records on diabetes and its complication prevention and development (10)(11)(12). Several large RCTs with a long-term follow-up such as Diabetes Prevention Program (DPP), the Finnish Diabetes Prevention Study, and Da Qing Diabetes Prevention have been reported, stating that adopting a healthy lifestyle would obtain long-term effects on preventing diabetes and its complications (13)(14)(15). Moreover, many recent systematic reviews have emphasized the important role in glycemic control and diabetes incidence (16)(17)(18). However, these systematic reviews only evaluated the effects on diabetes events among intervention studies, without assessing the effects of extended follow-up studies of intervention completed. In addition, few systematic reviews have reported the important role of non-pharmacological strategies in microvascular complications.
Therefore, it might be beneficial to conduct a systematic review and meta-analysis to whether comprehensive nonpharmacological interventions would be long-term effectiveness for preventing the diabetes and diabetes-related complications compared to other proposed treatments. The objective of this study, obtained by a comprehensive systematic review and meta-analysis of randomized controlled trials (RCTs), was to evaluate the long-term effects of non-pharmacological interventions on reducing the diabetes incidence in prediabetic patients and microvascular complications in patients with pre-diabetes and diabetes.

MATERIALS AND METHODS
This systematic review and meta-analysis was performed according to the Cochrane Handbook for Systematic Reviews of Interventions (https://gdt.gradepro.org/app/handbook/ handbook.html). Data were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) Guidelines (19). The PRISMA checklist of this study is provided in Supplementary Table S1. The study was registered in the International prospective register of systematic reviews (CRD42021240826).

Data Sources and Search Strategy
The five databases including PubMed, MEDLINE, EMBASE, the Cochrane Library, and the Web of Science Core Collection were searched for eligible trials using the keywords "diabetes" or "hyperglycemia" or "Impaired glucose tolerance"; "Lifestyle intervention" or "physical activity" or "exercise"; "Macrovascular complications" or "Microvascular complications" or "diabetic nephropathy" or "Diabetic retinopathy" or "Diabetic peripheral neuropathy" or "Diabetic foot"; "diabetes incidence". The initial search was performed in March 2021, and an updated search of five databases was performed in November 2021 using the same search terms. The search strategy for this study is provided in Supplementary Table S2.

Study Selection
To be included, studies had to be RCTs, which made direct comparisons of non-pharmacological interventions with medical therapy, placebo, usual care, or standard care; included RCTs had at least a 2-year duration of intervention; patients should be of any age with hyperglycemia (i.e., type 1 diabetes mellitus, type 2 diabetes mellitus, prediabetes, impaired glucose tolerance, impaired fasting blood glucose) and without chronic complications of diabetes; included RCTs reported at least one of the main outcomes of interest (i.e., diabetes incidence, cardiovascular complications, microvascular complications); and for the articles in the same study, the article with the longest follow-up duration was included. We excluded studies where both the intervention group and the control group were non-pharmacological interventions, non-RCTs, and publications without original data or with incomplete data. A second author (YC) confirmed that all articles that met the inclusion and exclusion criteria were included in the meta-analysis.

Data Extraction
Two independent investigators (RZ and (YC) used a standardized form to extract data from RCTs that met the inclusion/exclusion criteria, and any discrepancies were resolved by a third author (FL). The following relevant data from each article were extracted, including authors and year of publication; country; study design; study intervention and follow-up duration; patients' data including type of hyperglycemia; intervention measures of treatment group and control group; diagnostic criteria of outcomes; and outcomes of diabetes incidence and chronic complications. When the outcomes were reported as the percentage of incidence, a conversion in number of diabetes and chronic complications has been made in order to conduct an analysis.

Data Synthesis and Analysis
The primary outcomes of this meta-analysis were the diabetes incidence among prediabetic patients and microvascular events in patients with hyperglycemia. The meta-analysis was carried out using Review Manager 5.4 (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark).
Meta-analysis was performed using a random-effect model because of possible clinical heterogeneity. Data were pooled into relative risks (RRs) for dichotomous outcomes and RCTs with 95% CI. Heterogeneity was assessed using the I 2 statistics, where I 2 <30% was considered as low heterogeneity, I 2 values of 30-70 were considered as moderate heterogeneity, and I 2 >70% was considered as high heterogeneity.
Sensitivity analyses for the study quality and certain study characteristics (i.e., intervention duration, follow-up duration, and type of treatment measures) were performed through removing each individual trial. After recalculating pooled-effect estimates and heterogeneity, changing the significance of the effect or altering the effect size by 10% or more was considered influential. The meta-regression and subgroup analysis was used to examine the effect of follow-up duration on heterogeneity between the studies. We assessed publication bias by visual inspection of funnel plots for any outcomes>10 articles.

Quality Assessment
The Cochrane risk-of-bias tool was used to assess the RCT risk of bias. We assessed risk of bias in random sequence generation and allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting. A sensitivity analysis was conducted to exclude the articles with relevant weaknesses in trial design or execution. The overall quality of the evidence was also assessed using Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group guidelines (20). The quality of each RCT was assessed blinding by two reviews (RZ and (YC), and disagreements were resolved by a third author (FL).

Literature Search
The initial search assessed 4,495 publications. After the removal of duplicates, a total of 4,000 studies were selected and 3,936 publications were excluded by screening the titles and abstracts. Of the remaining 64 publications, 44 were excluded (Supplementary Table S3) because of study duration less than 2 years (7 publications), comparison of 2 treatments of nonpharmacological interventions (4 publications), unsuitable endpoints (11 publications), non-randomized study (9 publications), without original data (2 publication), unsuitable participants (5 publications), and articles being from the same trial and without the outcome of interest (6 publications). Finally, 20 suitable publications were included in the quantitative synthesis meta-analysis ( Figure 1).

Long-Term Effects on the Prevention of Diabetes in Intervention Studies
This analysis selected the results of the study that reported the longest follow-up time of diabetes events in terms of the same RCT. In overall analysis of 15 studies, as Figure 2 shows, the non-pharmacological interventions led to a diabetes incidence decrease significantly greater than those of comparator groups (0.62; 95% CI 0.54 to 0.71, p < 0.00001), with low heterogeneity between studies (I 2 = 29%, p = 0.14). Figure 3 shows the long-term effect on lowering microvascular events among patients with hyperglycemia. Compared with usual care and medical treatment, those with nonpharmacological interventions among hyperglycemic individuals had a lower incidence of microvascular complications (0.60; 95% CI 0.43 to 0.83, p = 0.002); heterogeneity across articles was moderate with an I 2 of 60% (p = 0.06). In sensitivity analysis, the heterogeneity of combined estimates did mark a change with the exclusion of Gong et al. or the DPP group ( Table 2), indicating that the source of

Long-Term Effects on the Prevention of Diabetes in Extended Follow-Up Studies
The three articles (Gong et al.; Nanditha et al.; Lindström et al.) were reported with diabetes outcomes of extended follow-up of intervention completed. As Figure 4 shows, the overall diabetes incidence was clearly reduction (0.78; 95% CI 0.63 to 0.96, p = 0.07), with evidence of substantial heterogeneity (I 2 = 62%, p = 0.0). It was indicated that the past non-pharmacological interventions still had a long-term legacy effect to lower diabetes incidence. However, the heterogeneity across articles was moderate with an I 2 of 60%. In sensitivity analysis, the heterogeneity of combined estimates did mark a change with the exclusion of Gong

Meta-Regression and Subgroup Analysis
Of the 21 articles, the 17 articles (the length of duration ranged from 2 to 6 years) were used to analyze the effect on diabetes prevention in intervention studies. To further evaluate the impact of duration on diabetes incidence, a random-effect meta-regression was conducted. The results of meta-regression analysis showed that duration (coefficient: -0.312, 95% CI: -0.457 to 0.166. p: 0.000) had a significant effect on prevention of diabetes. Based on the results of the meta-regression analysis, the subgroup analysis was performed for diabetes outcomes by follow-up duration (≤3 years, >3 years). As Figure 5 shows, the overall diabetes incidence was clearly reduced (0.62; 95% CI 0.54 to 0.71, p<0.01), with evidence of low heterogeneity (I 2 = 29%, p = 0.14). Moreover, the diabetes incidence of the long-term duration group was clearly reduced by 0.05% compared with the relatively short-term-duration group.

Publication Bias
Publication bias was assessed for diabetes outcome of 15 intervention studies. Details of quality of bias assessment of the included studies are listed in Table 4, and the funnel plot of 15 articles is provided in Supplementary Table S4.

DISCUSSION
In this systematic review and meta-analysis of 16 RCTs comparing non-pharmacological interventions with usual care, standard care, and medical therapy in prediabetes and diabetes, use of non-pharmacological interventions led to a long-term effect on reduction in the overall diabetes incidence and microvascular complications. Our forest plot of meta-analysis and sensitivity analysis demonstrated clearly that nonpharmacological interventions have such a reliable and longterm effect on the reduction in the microvascular complications among patients with hyperglycemia.
Our findings are consistent with several previous metaanalysis results which also indicated that the comprehensive non-pharmacological interventions showed significant effects on prevention of diabetes incidence and its risk of complications. Among them, lifestyle modification accounted for a large proportion, and previous meta-analysis studies have demonstrated a significant impact of lifestyle intervention on reduction of diabetes incidence and diabetes-related complications, as well as benefit in risk factors of cardiovascular disease (40)(41)(42)(43). Moreover, several articles of meta-analysis show that physical activity and diet modification are also associated with a decrease in blood glucose and diabetes incidence (17,44,45). In addition, international guidelines recommended the year rate of diabetes incidence in diabetes prevention studies. In our meta-analysis, we have included that the follow-up duration of these studies was more than 2 years and even 20 years, and we not only analyzed the long-term effects of non-pharmacological therapies on diabetes incidence but also analyzed the efficacy of preventing microvascular events. Importantly, different intensities of intervention may affect the outcomes among participants. Low intensity of intervention may change the outcomes weakening the effects of interventions in these patients. However, with the increase of age, the intensity of intervention strategies and patients' compliance may weaken as time goes by. We aimed to evaluate the relationship of unpharmacotherapy with pharmacotherapy or placebo, and we did not evaluate the effects between different intensities of lifestyle intervention or physical activity in our analysis. The results of this pooled analysis show that non-pharmacological therapies (lifestyle intervention, physical activity, exercise, etc.) have a significant effect on the reduction in diabetes incidence in intervention studies and extended follow-up studies among prediabetic patients, and protection of diabetes from microvascular diseases. It is indicated that even a low intensity of intervention could lead to a significant effect and past interventions have had a very long-term effect and metabolic memory. Some previous meta-analyses have reported relating outcomes of diabetic complications, such as the risk of cardiovascular events, and fewer systematic reviews have reported the chronic complications of diabetes in nonpharmacological studies. Since there are fewer studies that reported cardiovascular events in non-pharmacological studies, we have only analyzed the results of microvascular events. Originally, we have included RCTs of non-pharmacologicalinterventions in traditional Chinese medicine and bariatric surgeries. Compared with the medical therapies, bariatric and metabolic surgeries have been shown to be effective at preventing microvascular complications among patients with obesity and T2DM (46,47). Bariatric and metabolic surgeries are not longterm interventions. Therefore, we updated the inclusion criteria and did not include surgery-related studies. In addition, bariatric and metabolic surgeries were invasive therapies through gastrectomy. Its necessity, safety, and postoperative complications should be carefully considered. Moreover, durations of non-pharmacological interventions of traditional Chinese medicine in diabetes prevention studies, such as acupuncture and Qigong, have been reported for less than 1 year, so we exclude them.
The strengths of this study include that it is a comprehensive non-pharmacological intervention (included RCTs more than 2 years and extended follow-up studies) that evaluates the effects on diabetes incidence and microvascular events. It was indicated that the past non-pharmacological interventions have had a long-term metabolic memory to prevent diabetes and its complications. This study has several limitations. First,      regarding intervention strategies, only four intervention measures (lifestyle intervention, exercise, diet modification, and physical activity) met the inclusion criteria of this meta-analysis. Second is the variation in follow-up among the studies, ranging from 2 to 20 years. Last, due to the different treatment courses and different populations of patients, the heterogeneity remains relatively high among studies that evaluate the effect of reduction of diabetes incidence and microvascular complications.

CONCLUSION
Overall, non-pharmacological strategies implemented are promising approaches for preventing diabetes among prediabetic patients and microvascular complications among patients with hyperglycemia, with a long-term significant effect. More prospective randomized clinical trials and extended followup are needed to evaluate the non-pharmacological strategies on diabetes and microvascular events.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
FL conceived and designed the analysis. RZ and YC extracted the data and performed the analysis. RZ, YC, and YhZ wrote the paper and made the figures and tables. JD, XA, YD, YqZ, and XK provided the critical revision of the manuscript. All authors contributed to the article and approved the submitted version.