Search results

A total of 12,693 articles were provided by electronic searches, and 28 articles were obtained through other sources. Based on titles, abstracts, or duplicated articles, the 12090 of them were excluded. Eventually, 10 articles [20,21,22,23,24,25,26,27,28,29] were included for meta-analysis. The specific screening process for articles is shown in Fig. 1.

Fig. 1Characteristics of the included studies

All included studies were published between 2004 and 2022. There were 913 patients in the included articles, containing 463 patients in the intervention groups and 450 patients in the control groups. Patients in 8 articles exercised at home by themselves; patients in 2 articles exercised with family members. The duration of the intervention ranged from 3 to 12 months. Nine articles reported on fasting glucose or HbA1c; five articles reported on BMI or blood pressure; four articles reported on Low-density lipoproteins and triglycerides; and three articles reported on 2-h postprandial glucose. The specific basic characteristics are summarized in Table 1.

Table 1 Basic characteristics of the included studiesQuality Appraisal

The details of the assessment are shown in Fig. 2. The method of generating random sequences was reported in 9 studies and 1 article [24] was given unclear risk because of not reporting randomization. Only 4 articles [20, 21, 25, 27] mentioned allocation concealment and none of the rest mentioned specific allocation concealment resulting in unclear risk. Only 5 articles [21, 22, 26, 27, 34] were blinded to patients and investigators, and 1 article [28] was given the high risk for not blinding, while none of the rest reported whether they were blinded to investigators and participants. Six articles [20,21,22, 24, 27, 28] were assessed as low risk, as objective measurement instruments were used in the outcome assessment, while the rest of the articles had subjective assessments for outcome measures, so it was unclear whether the risk of bias was present. Two articles [21, 28] were assessed as high risk because of 20% or more dropouts for attrition and missing data analysis, and 2 articles [23, 26] were assessed as the unclear risk because of lack of data on attrition rate, while the rest were assessed as low risk. All of the included articles reported prespecified outcomes and all were assessed as low risk of bias. Only 1 article[26] was assessed as unclear risk of bias for other bias, and the rest were assessed as low risk.

Fig. 2

Risk of bias (ROB) assessment (A) ROB summary. (B) ROB graph

According to the GRADE system, low levels of certainty were shown in the results of HbA1c and FBG and very low levels of certainty were shown in the results of 2 h PG and LDL (Table 2). The downgrade was mainly due to the unclear allocation concealment in most studies, the high risk of bias in some studies, small sample sizes and wide confidence intervals, and too few studies to assess publication bias.

Table 2 GRADE Summary of EvidencePrimary outcomes (HbA1c)HbA1c (%)

Results from 9 studies [20, 22,23,24,25,26,27,28,29] with 786 patients were integrated to determine the effect of interventions on HbA1c levels (Fig. 3). The results indicated that, as a whole, exercise interventions based on family management or self-management significantly lowered HbA1c levels compared to the control groups (Z = 3.90; 95% CI MD = -0.81; − 1.21 to − 0.40; P < 0.0001). The results showed large heterogeneity (I2 = 87%). Sensitivity analysis revealed that 3 studies [24, 25, 27]were heterogeneous. After removing these 3 studies [24, 25, 27], the heterogeneity decreased to 41% with a significant effect of 1.04 (Z = 8.77; 95% CI − 1.27 to − 0.81; P < 0.00001) (Fig. 3).

Fig. 3

Forest plot of HbA1c on exercise interventions based on family management or self-management. HbA1c, glycosylated haemoglobin

Secondary outcomes (fasting blood glucose, 2-h plasma glucose, BMI, blood pressure, lipids)Fasting blood glucose (mmol/L)

The results of 9 studies [20,21,22,23,24,25,26, 28, 29] with 764 patients were combined to identify the effect of the Interventions on fasting blood glucose levels (Fig. 4). The results showed that, as a whole, exercise interventions based on family management or self-management significantly reduced fasting blood glucose levels compared to the control groups (Z = 4.63; 95% CI MD = − 1.17; − 1.67 to − 0.68; P < 0.00001). There was large heterogeneity in results (I2 = 81%). Sensitivity analysis identified 3 studies [21, 23, 25] that were heterogeneous. After removing these 3 studies [21, 23, 25], the heterogeneity decreased to 43% with a significant effect of 1.71(Z = 6.24; 95% CI − 2.25 to − 1.18; P < 0.00001) (Fig. 4).

Fig. 4

Forest plot of FBG on exercise interventions based on family management or self-management. FBG, Fasting blood glucose

2-h plasma glucose (mmol/L)

Three studies [20, 23, 24] reporting 2-h plasma glucose levels had a total of 304 patients included in the studies. The results revealed a significant difference between the intervention and control groups (Z = 5.53; 95% CI; MD = − 1.84; − 2.50 to − 1.19; P < 0.00001). There was heterogeneity between studies (I2 = 51%) (Fig. 5).

Fig. 5

Forest plot of 2 h PG on exercise interventions based on family management or self-management. 2 h PG, 2-h plasma glucose

Body mass index (BMI) (kg/m.2)

Five studies [22, 25,26,27,28] reporting on BMI with a total of 388 patients were included in the studies. The results of the Meta-analysis showed no significant differences between the intervention and control groups (Z = 1.59; 95% CI MD = − 0.70; − 1.57 to 0.16; P = 0.11). And there was almost no heterogeneity between studies (I2 = 1%) (Fig. 6).

Fig. 6

Forest plot of BMI on exercise interventions based on family management or self-management

Blood pressure (BP) (mmHg)

The meta-analysis for systolic blood pressure (SBP) and diastolic blood pressure (DBP) included 5 studies [21, 26,27,28,29] with 464 patients in total. Meta-analysis of SBP showed no significant differences between the intervention and control groups (Z = 1.79; 95% CI MD = − 7.30; − 15.29 to 0.69; P = 0.07). The results showed large heterogeneity (I2 = 85%). Sensitivity analysis revealed that the 2 studies [21, 27] were heterogeneous. After removing these 2 studies [21, 27], the heterogeneity decreased to 39% and there was a significant difference between the intervention and control groups (Z = 3.88; 95% CI MD = − 14.45; − 21.74 to − 7.16; P = 0.0001) (Fig. 7).

Fig. 7

Forest plot of SBP on exercise interventions based on family management or self-management. SBP, Systolic blood pressure

Meta-analysis of DBP showed no significant differences between the intervention and control groups (Z = 0.93; 95% CI MD = − 2.04; − 6.36 to 2.27; P = 0.35). The results showed large heterogeneity (I2 = 77%). Sensitivity analysis revealed that the 2 studies [26, 29] were heterogeneous. After removing these 2 studies [26, 29], heterogeneity was completely reduced (I2 = 0%), but there was still no significant difference between the intervention and control groups (Z = 1.04; 95% CI; MD = -1.02; -2.95 to 0.91; P = 0.30) (Fig. 8).

Fig. 8

Forest plot of DBP on exercise interventions based on family management or self-management. DBP, Diastolic blood pressure

Lipids (mmol/L)

The meta-analysis for Low-density lipoproteins consisted of 4 studies [25, 26, 28, 29] with a total of 236 patients. The results indicated a significant difference between the intervention and control groups (Z = 3.73; 95% CI MD = − 0.38; − 0.58 to − 0.18; P = 0.0002). The results showed no significant heterogeneity between studies (I2 = 9%) (Fig. 9).

Fig. 9

Forest plot of LDL on exercise interventions based on family management or self-management. LDL, Low-density lipoproteins

The meta-analysis for High-density lipoproteins consisted of 2 studies [25, 26] with a total of 73 patients. The Meta-analysis of High-density lipoproteins showed no significant differences between the intervention and control groups (Z = 0.02; 95% CI MD = 0.00; − 0.25 to 0.25; P = 0.99). There was heterogeneity between studies (I2 = 59%) (Fig. 10).

Fig. 10

Forest plot of HDL on exercise interventions based on family management or self-management. HDL, High-density lipoproteins

The meta-analysis for triglycerides included 4 studies [25, 26, 28, 29] with a total of 236 patients. The Meta-analysis of triglycerides showed no significant differences between the intervention and control groups (Z = 1.24; 95% CI MD = − 0.16; − 0.40 to 0.09; P = 0.22). The results showed no significant heterogeneity between studies (I2 = 22%) (Fig. 11).

Fig. 11