Comparative effectiveness of different modes of exercise interventions in diabetics with frailty in China: a systematic review and a network meta-analysis

PubMed, Cochrane Library, Embase, Web of Science, CNKI, WANFANG, VIP and SinoMed were searched by computer. Randomized controlled trials (RCTs) in patients with both diabetes and frailty were collected. The search time limit was from the establishment of the database to May 21, 2023, and the language type was Chinese or English. In addition, the references of the included literature were traced back to supplement the acquisition of relevant literature. The search was conducted by combining subject words and free words. (Additional file 1: Annexes 1).

Study selection and data extraction records identified from databases and manual searches were imported into Endnote X9.0. According to the inclusion and exclusion criteria, two investigators respectively evaluated and screened the titles and abstracts recorded in Endnote X9.0 to finally determine the included literature. Analyses were resolved by discussion between the two authors, and a third author was invited to adjudicate when necessary. Data extraction included: (1) RCT characteristics (e.g., authors, year of publication, study location, sample size, intervention and control measures); (2) characteristics of the participants (e.g., gender, age); And (3) outcomes (e.g., FBG, HbA1c, PBG). For multiple articles with data from the same RCT, the article with the longest follow-up period was selected, and the remaining articles were supplemented. For multiple sets of experiments, each set of data will be extracted.

The Cochrane risk of bias tool for randomised trials (RoB 2) was used to assess the inexpensive risk of RCTs [32]. Two researchers independently evaluate the quality of the article, and when there is a disagreement between the two researchers, a third researcher participates in the discussion and helps to resolve the disagreement.

The network package of Stata 14.0 and the netmeta package of version 2.8–2 of R4.3.1 were used in this study. If the outcome was a continuous variable, the mean difference (MD) was used as the effect size indicator and the 95% confidence interval (CI) was provided. In R4.3.1 software, the netmeta package of 2.8–2 is used for data analysis. The level of heterogeneity between studies was assessed using the I² statistic, and when I² > 50%, the heterogeneity between studies was considered statistically significant, in which case a random-effects model was used; otherwise, a fixed-effect model was used [33]. In the stata 14.0 software, the network package is used for data analysis. Heterogeneity between studies was assessed by calculating Cochran’s Q statistic, and a fixed-effect model was chosen when the P value of the Q statistic was > 0.05; otherwise a random-effects model was used [34].

For data expressed in quartiles and medians, conversion to the format of mean and standard deviation [35, 36].When P < 0.05, the difference was considered statistically significant. Node analysis was used for inconsistency test, and if P > 0.05, the consistency model was used for analysis. Otherwise, the inconsistency model was used. The effect of exercise intervention was ranked by the best probability ranking chart.

Pooled data using fixed-effect and random-effects models were used to assess the robustness of network meta-analysis. We use the global inconsistency method for inconsistency testing, and the node-splitting method for local inconsistency. If the P > 0.05, there is no significant difference in the results, and the consistency model is used, otherwise, the inconsistency model is used.

Funnel plots were used to test the included literature for publication bias for each outcome [37].

The search revealed a total of 1067 studies in various databases. These included PubMed (257), Cochrane Library (198), Embase (192), Web of Science (80), CNKI (132), VIP (69), WANFANG (84), SinoMed (49), and 6 articles retrieved by other sources. After screening, 415 duplicate studies were excluded, 251 studies were excluded because their content did not match this study, and 334 studies were excluded after reading the title and abstract sections. After reading the remaining 67 studies in full, 52 studies were excluded (45 studies lacked a basis for the diagnosis of frailty or diabetes, 5 studies were unable to extract data, 1 study did not meet study population requirements and 1 study did not meet regional requirements). As a result, a total of 15 RCTs were included in this study. (Fig. 1).

The basic characteristics of the included studies are shown in Table 1. A total of 1550 patients were included in our network meta-analysis. All studies included in this study included two arms. The included studies were published between 2019 and 2023 and had an intervention duration of 2 to 6 months. There are five main types of training in the literature: resistance training, aerobic training, comprehensive training, Otago training, and Pilates training. (Table 1).

A total of 3 articles were judged to be high risk and 12 articles were judged to be some concerns in this assessment. Of the randomization process evaluations, 2 were judged to be high risk. 3 of the Deviations from intended interventions were rated as high risk. In the evaluation of Mising outcome data, 1 was rated as high risk, and the rest were low risk. All studies were assessed as low risk in the Measurement of the outcome. In the Selection of the reported result, all studies were assessed as having some concerns. (Fig. 2, Additional file 1: Fig. 1: Results on risk of bias (using RoB2) of including RCTs).

The literature included in this study has not yet formed a ring structure, so an indirect comparison method was used in this study. The consistency analysis was performed for each outcome index, and it was found that the scale reduction factor parameter value was close to 1.00, indicating good convergence. In data analysis, the I² of all studies was less than 50%, the consistency model was used for network Meta-analysis.

The improvement of fasting blood glucose and glycosylated hemoglobin in diabetic patients with frailty was represented by a funnel plot. The funnel plot shows that the points are not completely symmetrical, indicating that there may be some publication bias. (Additional file 1: Annexes 1).

Multi-group training is recommended by the American College of Sports Medicine as an effective alternative to traditional training training to improve the physical function of the elderly [53]. Studies [54, 55] have shown that multi-group training can not only regulate glucose metabolism and lipid metabolism in diabetic patients, reduce the risk of cardiovascular disease, but also improve frailty, cognitive function, emotional state of the elderly. However, there are few network meta-analyses including multi-group training interventions, and it is impossible to obtain the horizontal comparison results of multi-group training.

Pilates training, invented by Joseph Hubertus Pilates, is a kind of training that integrates the training of body muscles and body functions with the concept of breathing and mental concentration. It is an training method that improves the body's strength, flexibility and core body control [56], while emphasizing the role of breathing on the body [57]. Previous studies have shown that Pilates training has a good effect on pain relief [58], while recent studies have shown that Pilates training has a good effect on improving the balance of the body, reducing the risk of falls [59], controlling blood pressure [60] and relieving bad emotions [61]. A study that included 21 subjects [62] showed that Pilates training was the best way to improve Proactive balance. However, another systematic review on elderly patients with type 2 diabetes [63] showed that Pilates training was superior to other trainings in improving glycosylated hemoglobin, cholesterol, triglyceride and other indicators. This study shows that integrated training is the best training for blood glucose control and lipid reduction. Other studies have shown that it is the best measure for lowering blood lipids and glycated protein [64, 65], but not Pilates training. This may be because: (1) There were not enough Pilates training in the included studies, but mainly divided into three categories: resistance training, aerobic training, and the combination of resistance training and aerobic training (integrated training); (2) Different patient population: previous studies mostly focused on patients with diabetes or type 2 diabetes, while this study focused on patients with diabetes and frailty.

Although the American College of Sports Medicine recommends resistance training as a safe and effective lifestyle intervention for the elderly [66], and both resistance training and aerobic training are related to the improvement of systemic insulin sensitivity and blood glucose control in the elderly [67], their training action time is relatively short, and there is a certain gap compared with other training methods that take into consideration endurance training.

Otago training involves muscle strength, balance, walking and other aspects of training, which takes into account aerobic, resistance and balance, but it is a home-based, progressive training for the prevention of frailty and falls. It has been proved to effectively improve the cognitive function, balance ability, lower limb muscle strength and functional physical fitness of the elderly [68], and prevent the elderly from falling [69]. Accelerated recovery of physical function and reduced economic costs [70, 71] due to the greater emphasis on strength training and balance training. However, compared with other types of training, Otago training pays more attention to fall prevention [72, 73], so it may be inferior to other sports in terms of training volume and cannot provide enough stimulation to the body, so the effect is relatively weak.

Integrated training can take into consideration the requirements of strength training and endurance during training, which has been proved to be beneficial to the improvement of vascular function [74]. At the same time, it can improve cardiac dysfunction by reducing inflammation and oxidative stress [75]. Another study showed that integrated training is beneficial to the improvement of metabolic syndrome in patients. However, the lack of reasonable allocation of time arrangement in the implementation process of the included studies may affect the intervention effect. As for the research on frailty, some studies have concluded that resistance training has the greatest potential to improve frailty of patients [76]. However, all the studies included in this paper used resistance training for the intervention of frailty, so Meta-analysis is not possible.