Association between physical exercise and all-cause and CVD mortality in patients with diabetes: an updated systematic review and meta-analysis

Objectives Physical activity is recommended in guidelines for treatment for diabetes, but the association between physical activity and mortality among diabetic patients has not been extensively studied. Methods Databases were searched from inception to July 10, 2020. Prospective studies were selected to evaluate the association between physical activity and risk for total and cardiovascular diseases (CVD) mortality among diabetic patients. Data were pooled using random-effect model to calculate the relative risks (RRs) with 95% confidence intervals (CIs). Results We included 16 eligible studies involving with 155,203 diabetic participants and 13,821 cases of death. Our study suggested that physical activity in diabetic patients may decrease risk for all-cause (RR 0.57, 95% CI 0.49–0.67) and CVD mortality (RR 0.55, 95% CI 0.34–0.68). The summary RR for CVD events was 0.65 (95% CI 0.41–1.03). Furthermore, the reductions in all-cause mortality were more significant in diabetic patients with old age (> 60 years) (RR 0.46, 95% CI 0.29 -0.75), higher body mass index (BMI ≥ 28) (RR 0.53, 95% CI 0.42–0.69) and shorter duration of diabetes (RR 0.45, 95% CI 0.24–0.84). Conclusion Physical activity reduced the risk of total and CVD mortality among patients with diabetes, in particular in diabetic patients with old age (> 60 years), obesity and shorter duration of diabetes.


Background
Diabetes was recognized as a substantial threat to public health with the fastest increasing morbidity worldwide 1 . It was reported that 415 million people were estimated to have diabetes in 2015, and will increase to 642 million by 2040 2,3 . Patients with diabetes are at high risk for adverse outcomes from its macro vascular and micro vascular complications, which account for more than 2 million deaths every year 4 and constitute the seventh most common cause of disability worldwide 5 . The absolute number of deaths from diabetes increased between 2006 and 2016 by 31.1% 6 . Among adults in China, diabetes was associated with 2-fold increased mortality compared with adults without diabetes 7

. Cardiovascular diseases (CVD)
is the most common cause of morbidity and mortality among subjects with type 2 diabetes mellitus (T2DM) 8 . Body mass index (BMI) is an independent risk factor of diabetes and CVD 9 , which were closely related with physical exercise. Physical activity is important in the prevention of the development of T2DM in people with impaired glucose tolerance (IGT) and for the control of glycaemia and related CVD complications 10,11 . Several studies have indicated that high leisure-time physical activity is associated with reduced total and CVD mortality among patients with diabetes [12][13][14][15][16][17] . However, results were not consistent 14,17 . In diabetic patients with heart diseases or other serious complication, physical activity may worsen their health conditions. Furthermore, how diabetes severity, patients' age or BMI influence this association has not been extensively studied. Several newly studies on the association between physical activity and mortality have been published. The aim of this study was to examine associations of physical activity with risk total and CVD mortality among diabetic patients. Furthermore, we analyzed the association in diabetic patients with different characteristics such as age, BMI, glycated haemoglobin A1C (HbA1c), types and duration of diabetes.

Materials and methods
This meta-analysis is reported according to Meta-analysis Of Observational Studies in Epidemiology (MOOSE) Reporting Guidelines 18 .

Search strategy
A systematic literature search for relevant studies was conducted in the databases of Pub Med and EMBASE from inception to January 15th, 2020. The search strategy was as follows: (((((((Diabetes Mellitus [MeSH Terms])) OR (diabetes)) OR (type 2 diabetes)) OR (type 1 diabetes) [All Fields])) AND ((((physical activity[All Fields])) OR (exercise)) OR (sports))) AND ((((death[All Fields])) OR (mortality)) OR (fetal)). In addition, we reviewed the references from relevant articles to identify additional relevant studies. Authors were contacted and requested to provide further data if required.

Study selection
The following inclusion criteria were required to be eligible for the meta-analysis: (1) cohort, prospective or longitudinal study with more than 5-year follow up; (2) diabetic patients with active and inactive physical activity; (3) reported relative risk (RR) estimates of mortality, such as relative risks (RR), odds ratios (OR), hazard ratios (HR) or incidence with 95% confidence intervals (CIs) for 2 or more categories of physical activity. In multiple same-population studies, we selected and included the one study with longest follow-up time.

Data extraction
For each eligible study, we recorded the following data: name of the first author, year of publication, study name, study location, participants, age at baseline, and measure-ment for physical activity, number of diabetes and death, years of follow-up and outcomes.

Statistical Methods
The pooled analyses were performed using the random-effects model to calculate RRs with 95% CIs). If several estimates were reported in the same article, we chose the most fully adjusted RR of the top category vs. the lowest category of physical activity. If the reference category used in the analyses was not the lowest category, we used the method described was by Hamling et al 19 to convert risk estimates. Study quality was assessed using Newcastle-Ottawa scale (NOS) 20 . Heterogeneity was quantified using the I2 test, where I2 > 50% indicated significant heterogeneity 21 . Publication bias was evaluated by the Egger's and Begg's test 22,23 . Sensitivity analyses were performed to assess the robustness of the findings by omitting each study from the analyses and then summarized the remains. Subgroup analyses were conducted to investigate the impact of age, study location, number of participants and case, follow-up of cohort studies, HbA1c, type and duration of diabetes on the association between physical activity and risk of total and CVD mortality among diabetes. All the analyses were conducted using Stata statistical software (version 16.0). A 2-sided P value of less than 0.05 was considered statistically significant.

Study Selection
Of 5,766 studies identified by the initial search, 178 were selected for full-text review; 162 of these were excluded, leaving 16 ( Figure S1). Two studies were from the same study 16,24, and we included the one with larger sample size 16. Figure S1: Study selection.

Study characteristics
After ineligible studies were excluded, 16 cohort studies were included in our meta-analysis. (Descriptive characteristics of studies and outcomes are shown in Table  S1) 12,13,[29][30][31][32][33][34][14][15][16][17][25][26][27][28] . It involved with 155,203 diabetic participants and 13,821 cases of death. Participants were aged 25 to 80 years, with more than half being middle-aged or older. The duration of cohort studies ranged from 5.7 to 23.8 years, with a median year of 8.7. Results of study quality assessment (score 0-9) yielded a score of 7.0 or above for 15 studies (Table S2). Time spent at each activity in hours per week was multiplied by its metabolic cost (based on the oxygen consumption required for that activity). These weighted hours were added up to get a total daily physical activity score. Participants were classified based on tertiles of the daily physical activity scores: low ＜30), moderate (30)(31)(32)(33), and high＞33) physical activity level. A metabolic equivalent index was computed by assigning a multiple of resting metabolic rate to each activity metabolic equivalent task (MET value) and all MET-hour products were summed to estimate daily physical activity. Participants were classified based on quintile of MET-hour score.

1013
Tielemans, et al 6 2013 European The EURODIAB Prospective Complications Study is a cohort including 3,250 male and female patients with type 1 diabetes.
Physically inactive participants were considered to be those who reported walking <1.5 kilometers on an average weekday, no regular bicycling and no participation in sports. Participants who reported walking ≥1.5 kilometers on an average weekday, regular bicycling or played any sport were considered to be the physically active group. For primary analyses, participants were divided into who were sedentary or undertook only mild physical activity in the prior week and those who participated in at least one session of moderate or vigorous physical activity >15 min. In subsequent analyses authors sought to clarify whether participation in mild physical activity was beneficial. Then they compared the outcomes of patients who undertook only mild exertion with those who were entirely sedentary and those who undertook moderate or vigorous exercise at the week prior to randomization. Participants were classified into one of four groups depending on whether they met each of the recommended guidelines: insufficient activity (insufficient aerobic and muscle strengthening activities), aerobic physical activity only (recommended aerobic activity and insufficient muscle strengthening activity), muscle strengthening only (insufficient aerobic activity and recommended muscle strengthening activity), and both (recommended aerobic and muscle strengthening activities).

Overall analyses
Fifteen observational studies were included in the analysis of diabetes and all-cause mortality. Significant reductions in all-cause mortality were observed for diabetic patients with physical activity (RR 0.57, 95% CI 0.49-0.67; P < 0.001; Figure 1). Obvious heterogeneity was detected among these studies (I2 = 83.0%; P < 0.001 for heterogeneity; Figure 1). Furthermore, pooled estimates of seven studies showed that diabetic patients with physical activity had lower risk for CVD mortality (RR 0.55, 95% CI 0.44-0.68; P < 0.001; Figure 2) with moderate heterogeneity (I2 = 68.1%; P = 0.003 for heterogeneity; Figure 2). There were four studies reported CVD evens in diabetic patients with active or inactive physical activity and the pooled RR for CVD was 0.58 (95% CI 0.49-0.69; Figure  S2) by a random effects model (I2 = 84.9%; Figure S2).

Sensitivity and subgroup analyses
Sensitivity and subgroup analyses were conducted to examine the stability of the primary results. In the sensitivity analyses, through omission of any individual study, the results for all-cause and CVD mortality were not significantly altered ( Figure S3-S4). However, the results for CVD were changed in the sensitivity analysis ( Figure S5), which may due to limited studies were included. We fur-ther conducted subgroup analyses for all-cause mortality stratified by age, study location, number of participants and case, follow-up of cohort studies, HbA1c, type and duration of diabetes. Results of any subgroup were consistent with our overall findings (Table S3). Importantly, the reductions in all-cause mortality were more significant in diabetic patients with old age (> 60 years) (RR 0.46, 95% CI 0.29-0.75; P = 0.002; Figure S3: Sensitivity analysis of studies for all-cause mortality.    Table S3) and shorter duration of diabetes (RR 0.45, 95% CI 0.24-0.84; P < 0.001; Table S3). There were eight studies conducted in T2DM and one study conducted in type 1 diabetes mellitus (T1DM). The remains didn't provide the information of types of diabetes. The summary RR for T2DM was 0.56 (95% CI 0.46-0.68; P < 0.001; Table S3) and the RR for T1DM was 0.66 (95% CI 0.42-1.03; P = 0.007; Table S3). Furthermore, there were four studies provided data of HbA1c and the average of HbA1c in these studies were more than 7%. The pooled RR of these 4 studies was 0.54 ( 95% CI 0.35-0.84; P < 0.001; data was not shown in Table). We didn't perform subgroup analysis for CVD mortality and events because the limited studies were included in each subgroup, which may provide some misleading information.

Publication bias
As shown in Table S4, no publication bias was observed according to Egger's and Begg's test in studies of analysis for mortality from all causes and CVD, and CVD events (all P > 0.05).

Discussion
In this meta-analysis involving 155,203 diabetic patients, we concluded that physical activity reduced risk for total and CVD mortality among patients with diabetes. Furthermore, the reductions were more significant in diabetic patients with old age (> 60 years), obesity and shorter duration of diabetes.
Several high-quality meta-analyses summarized the studies conducted in general population and conducted that physical activity may reduce the risk of all-cause, cardiovascular and cancer mortality, and incident type 2 diabetes [35][36][37][38][39][40][41] . We firstly summarized the studies conducted in diabetic patients and conducted that physical activity could reduce the risk of total and CVD mortality, as well as CVD events among diabetes. It may contributed by the beneficial effect of physical activity on several indices of cardio metabolic diseases including body weight 42 , body fat distribution 42-44 , blood pressure 45 , blood lipids 43,46 , insulin resistance47,48, endothelial function 49 and cardio-respiratory fitness 42,44 . It was also reported that moderate or high levels of physical activity were associated with a significantly reduced risk of total and CVD mortality among adults with diabetes, independent of age, education, BMI, blood pressure, total cholesterol, and smoking 12 . Aerobic and resistance training improve insulin action and plasma glucose (PG), lipids, blood pressure and cardiovascular risk 50 . Regular exercise is necessary for continuing benefit. Therefore, diet control and lifestyle interventions are recommended as the first-line treat-ments for diabetes 11 . Schellenberg et al also found that lifestyle interventions effectively decrease the incidence of type 2 diabetes in high-risk patients 51 . However, in patients who already have type 2 diabetes, there is no evidence of reduced all-cause mortality and insufficient evidence to suggest benefit on cardiovascular and micro vascular outcomes 51 . We, therefore, tried to further analyze the association in diabetic patients stratified by age, BMI, HbA1c and duration of diabetes. The results showed that the reductions of total mortality were more significant in diabetic patients with old age (> 60 years), obesity and shorter duration of diabetes. It was reported that older adults> 65 years old) are the least physically active age group 52 . In advanced age, physical activity is effective at mitigating sarcopenia, restoring robustness, and preventing/delaying the development of disability 53 . As we discussed previously, BMI is an independent risk factor of diabetes and CVD 9, therefore physical activity may have more protective effects on diabetic patients with obesity.
The significant reduction of total mortality in the patients with shorter duration of diabetes could, in part, be due to the protection of physical activity on β cell from further failure induced by lipotoxicity in early stage. Besides, the physical activity did not reduce the total mortality in patients with T1DM 17, which may due to the hypoglycemia during the immediate post exercise period 54 .
There are several limitations for the current study. Firstly, our meta-analysis was performed on summary data, thus leading to a relatively poor accuracy of assessment com-pared with individual-level analyses. Secondly, our study was limited to studies reported in English. Furthermore, significant heterogeneity was detected among all studies, may due to the different measurements of physical activity. Finally, we cannot identify the best way and the most appropriate amount physical activity for diabetes to achieve the greatest benefit.

Conclusion
The present study added to the literature by confirming the association between physical activity and risk for total and CVD mortality among diabetic patients, in particular in diabetic patients with old age (> 60 years), obesity and shorter duration of diabetes. It may provide some information for policymakers and future guidelines. Future study is needed to summarize the dose-response association of different kinds of physical activity and health outcomes in patients with diabetes.

Ethics approval and consent to participate
Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of interest
The authors declare no conflicts of interest.

Findings
This study received no specific funding for this work.
Author's contribution Mr. Liu a had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. XL and ZW selected relevant studies and extracted data from included studies. Any disagreement was resolved by discussion with a third review author to reach a consensus (NL). XL analyzed data and drafted the manuscript. NL and ZW reviewed and provided suggestions. All authors reviewed the manuscript and approved the final manuscript.