DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
Contents available at ScienceDirect
Diabetes Research and Clinical Practice journ al h ome pa ge : www .elsevier.co m/lo cate/diabres
A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes Jane E. Yardley a,b, Jacqueline Hay a, Ahmed M. Abou-Setta c,d, Seth D. Marks e, Jonathan McGavock a,* a
Manitoba Institute of Child Health, University of Manitoba, Winnipeg, MB, Canada University of Alberta, Augustana Campus, Camrose, AB, Canada c George & Fay Yee Centre for Healthcare Innovation, University of Manitoba, Winnipeg, MB, Canada d Winnipeg Regional Health Authority, Winnipeg, MB, Canada e Department of Pediatrics, University of Manitoba, Winnipeg, MB, Canada b
article info
abstract
Article history:
Aims: Conflicting evidence exists regarding the benefits of physical activity for long-term
Received 18 June 2014
blood glucose control in adults with type 1 diabetes (T1D). The object of this systematic
Received in revised form
review was to determine the effects of physical activity on long-term blood glucose control
27 August 2014
in T1D adults.
Accepted 15 September 2014
Methods: PubMed/Medline, Embase, CENTRAL, SPORTdiscus, Global Health and ICTRP were
Available online xxx
searched up to October 2013 for randomized trials of aerobic or resistance exercise training
Keywords:
months. The primary outcome was glycated hemoglobin (HbA1c). Secondary outcomes
in T1D adults. Exercises had to be performed at least twice weekly for a minimum of two Physical activity
included cardiorespiratory fitness and insulin dose.
Glycemic control
Results: Six randomized trials were identified (323 adults); sample sizes ranged from n = 6 to
Type 1 diabetes
n = 148 participants receiving the intervention. Five trials had an unknown risk of bias; one trial was deemed to be at high risk of bias. Exercise frequency varied from twice weekly to daily, with intensities (50–90% VO2peak), and session durations (20–120 min) varying widely. Four trials reported HbA1c, which decreased with exercise training (mean difference [MD] 0.78% (9 mmol/mol), 95% CI 1.14 (13 mmol/mol) to 0.41 (5 mmol/mol); p < 0.0001; I2 0%) compared with controls. Exercise training improved cardiorespiratory fitness by 3.45 ml/ kg/min (95% CI 0.59 to 6.31, p = 0.02, I2 0%) compared with controls. One trial reported an effect on insulin dose (MD 0.4 U/kg, 95% CI 0.53 to 0.27, p < 0.00001) compared to controls. Conclusion: There are currently insufficient well-designed studies to ascertain the true effect of exercise training on HbA1c in individuals with T1D, but current results are promising. # 2014 Published by Elsevier Ireland Ltd.
1.
Background
Regular physical activity is associated with multiple health benefits for individuals with type 1 diabetes (T1D) [1].
Increased physical activity is associated with a lower risk of complications and an increased life expectancy [2], however, more than 60% of adults with T1D do not achieve recommended levels of physical activity [3]. Children with T1D also face barriers to achieving recommended physical activity
* Corresponding author. Tel.: +1 204 480 1359. E-mail address: [email protected] (J. McGavock). http://dx.doi.org/10.1016/j.diabres.2014.09.038 0168-8227/# 2014 Published by Elsevier Ireland Ltd.
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8
2
diabetes research and clinical practice xxx (2014) xxx–xxx
targets [4]. In addition to conventional barriers to exercise (i.e. lack of time, work pressure, environmental conditions, low energy etc.), fear of hypoglycaemia [5] and a lack of evidence for any beneficial effect of exercise on long-term glycemic control likely contribute to low physical activity rates in individuals with T1D. Clinical practice guidelines recommend 150 min of moderate (50–70% of person’s maximum heart rate) to vigorous (>70% of person’s maximum heart rate) physical activity weekly for adults with T1D [6,7]. Due to the dearth of evidence available on this topic, the appropriate dose (type, duration, frequency, intensity) of physical activity needed for improved glycemic control in this population remains unclear. In contrast to the well-documented benefits of physical activity for glycemic control among individuals with type 2 diabetes (T2D) [8,9], experimental studies of exercise training in T1D individuals have yielded conflicting results. Recent systematic reviews of trials of physical activity for glycemic control in persons with T1D failed to address these inconsistencies as they have either (1) included quasi experimental trials [1,10]; (2) pooled data from short term training studies (2–8 weeks) and acute studies [11] and/or (3) included studies of interventions that combined diet and exercise training [1,10,12]. One recent meta-analysis of physical activity and/or sedentary behavior intervention studies in children with T1D, included only randomized controlled trials [13] and found a significant improvement in glycemic control (HbA1c) with exercise training [mean difference 0.85% (95% CI, 1.45 to 0.24%)]. A comparable systematic review and meta-analysis in adults is currently unavailable in the literature. Therefore, a focused systematic review of the literature was conducted to examine the state of the scientific literature regarding randomized trials of exercise training interventions lasting more than eight weeks with exercise performed at least twice weekly in individuals with type 1 diabetes.
2.
Methods
Using a previously published protocol [14], a systematic review applying methodological approaches outlined in the Methodological Expectations of Cochrane Intervention Reviews [15] was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses criteria [16]. An expert panel from multiple fields (pediatrics, endocrinology, clinical epidemiology, exercise physiology) formulated the review question, reviewed the search strategies and review methods, and provided input throughout the review process.
2.1. Populations, interventions, comparators, outcome measures, settings and study designs (PICOS)
Inclusion criteria for trials included in the review were: (1) the majority of participants (80%) had to be between 15 and 50 years old and diagnosed with type 1 diabetes mellitus; (2) trials had to be prospective, randomized and controlled with a non-exercising T1D control group; and (3) the exercise intervention consisted of supervised or unsupervised aerobic, resistance or combined physical activity offered at least twice weekly for longer than eight weeks. Trials were excluded if they (1) involved animals; (2) included participants with cystic fibrosis, type 2 diabetes mellitus, monogenic forms of diabetes, secondary diabetes, or with HbA1c < 7.0% (53 mmol/mol); (3) included co-dietary intervention, behavioral modification not directly related to physical activity, or acute exercise only (i.e. single exercise session); (4) used a quasi-experimental design (e.g., alternate randomization, randomization according to hospital number, non-randomized trials (e.g. cohort and casecontrol studies, cross-over or cluster randomized trials The primary outcome measure was HbA1c assessed at the end of the active intervention. Secondary outcomes included maximal oxygen uptake, weight (kg), daily insulin dose, cardiovascular risk factors, body mass index and occurrence of adverse events in both groups.
2.2.
Trials without language restriction published up to October 2013 that met inclusion criteria were identified using individualized search strategies prepared for the following databases: PubMed (National Library of Medicine), EMBASE (Ovid), CENTRAL (the Cochrane Library—Wiley), CINAHL (EbscoHost), Global Health (Ovid), and SPORTDiscus (EbscoHost). The PubMed strategy is presented in Appendix 1. Further, a forward search was performed in Scopus and Web of Science to identify additional citations. To identify ongoing or planned trials, we searched the World Health Organization’s International Clinical Trials Registry Platform for relevant registrations. Finally, abstracts, conference proceedings and the references lists of relevant narrative and systematic reviews were searched and included trials were hand-searched for possible relevant citations. Two reviewers independently screened the titles and abstracts (when available) of search results to determine if studies met the inclusion criteria. Reports were classified as: include, exclude, unclear, or duplicate of another citation. The full text of citations classified as ‘‘include’’ or ‘‘unclear’’ by either reviewer were retrieved for formal review. Two reviewers independently assessed the full text of each potentially included trial by using a standardized form outlining the predetermined inclusion/exclusion criteria. Disagreements were resolved by discussion between the two reviewers or by another reviewer’s adjudication, as needed.
2.3. The primary research question for this analysis was: ‘‘In adolescents and adults (aged 15–50 years) living with T1D, does a structured physical activity intervention lead to clinically meaningful reductions in glycated hemoglobin (HbA1c) compared to a non-exercising control condition?’’ To address this question, only randomized, controlled trials of adults diagnosed with T1D were included.
Search strategy for identification of studies
Data extraction and management
Two independent reviewers (JY, JH) extracted information using standardized, piloted forms. Disagreements were resolved through consensus and with the assistance of another reviewer (JM), if consensus was not achieved. From each trial the following information was extracted: author, year and language of publication, source of funding, study
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
design, study population, patient characteristics, intervention and its comparator, and outcomes of interest.
2.4.
Assessment of risk of bias
Internal validity of included trials was assessed using the Cochrane Collaboration Risk of Bias tool [17,18], which consists of six domains (sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and ‘‘other’’ sources of bias) and a categorization of the overall risk of bias. Each domain was rated as ‘‘low risk’’, ‘‘unclear risk’’, or ‘‘high risk.’’ Overall assessments were based on the responses to individual domains, judged by two independent reviewers (AA, JY) following the recommendations detailed in the Cochrane Handbook for Systematic Reviewers and related publications [17,18]. The criteria used to assess each domain are summarized in Appendix 2. If one or more domains were judged as having a high risk of bias, the trial was classified as being at high risk of bias. If all components of a trial were classified as having a low risk of bias, the trial was considered to be at low risk of bias. In cases of mixed assessments of low and unclear risk of bias, or where all assessments were unclear risk of bias, overall classification was unclear risk of bias.
2.5.
Measures of treatment effect
Data from included trials were analyzed using Review Manager (RevMan version 5.2, The Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark), and Microsoft Excel (Excel v.14, Microsoft Corp., Redmond, WA, USA). Pooled
Records identified through database searching (n = 14,660 citations)
continuous data were expressed as mean differences, with 95% confidence intervals (random-effects model). Statistical heterogeneity of the data was quantified using the I-squared test with 95% uncertainty intervals [19]. Publication bias was assessed by viewing the overlap of confidence intervals, and using funnel plot techniques [20]. All tests of statistical inference reflect a 2-sided a of 0.05.
2.6.
Subgroup analysis
For the primary outcome of reduction in post-treatment HbA1c, the following a priori subgroup analyses were performed, where data were available: participants with baseline HbA1c > 8.5% (69 mmol/mol) versus HbA1c < 8.5% (population); and short (12 months) trial durations (timing). Aggregate results were also examined to determine whether large effect sizes were related to high risks of bias. Finally, whether or not sources of funding influenced results was examined.
3.
Results
Of the 14,663 citations obtained from electronic and handsearches, 51 trial reports that potentially examined the effect of exercise on glycemic control in T1D individuals were identified. Forty-three reports were excluded: inappropriate population (n = 5); inappropriate intervention/control (n = 2); inappropriate study design (n = 21); full-text manuscripts unavailable for review through library services (n = 13); unable to translate non-English language (n = 2) (Fig. 1). Six unique
Additional records identified through other sources (n = 2 citations)
Records after duplicates removed (n = 11,772)
Records screened for eligibility (n = 11,772)
Full-text articles assessed for eligibility (n = 50)
3
Records excluded (n = 11,722)
Full-text articles excluded: (n = 43) Inappropriate population (n = 5) Inappropriate intervention/control (n = 2) Inappropriate study design (n = 21) Unavailable for review through library services (n = 13) Non-English language – unable to translate (n = 2)
Studies included in review (n = 6 primary publication) (n = 1 companion publications)
Fig. 1 – Modified PRISMA flow-chart. Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
Study
N (training/ control)
Intervention
Program duration
Weekly exercise frequency and duration
Adherence to training
Mean age (years)
Mean HbA1c (%)
Baseline
9/7
Laaksonen [22]
28/28
Landt [23]
9/6
Maggio [24]
15/12
Salem [26]
148/48
WallbergHenriksson [12]
6/7
Supervised high intensity aerobic (80–90% HRR) and ‘‘strength’’ training (exercises not described) Supervised aerobic (50–80% VO2peak) exercise. Intensity increased through intervention. Supervised aerobic and high intensity training (80–85% maximum HR) Supervised ‘‘weight bearing physical activity’’ (HR 140 beats/ minute) Supervised aerobic (65–85% HRR), high intensity and resistance (50–100% 10RM) exercise Unsupervised aerobic exercise (60–90% VO2peak–intensity increased progressively)
Training
Control
Training
Control
Control
6 months
2 times/week 120 min/session (supervised) Once/week 60 min/session (unsupervised)
62–100% (supervised) 52–89% (unsupervised)
15.9 1.5
16.3 1.2
8.0 1.4
8.0 1.4
Not reported
Not reported
3–4 months
3–5 times/week; 30–60 min/session
Not reported
32.5 5.7
29.5 6.3
8.2 1.1
8.3 1.3
8.0 1.0
8.5 1.6
3 months
3 times/week; 65 min/session
Not reported
16.1 2.4
15.9 0.7
12.1 3.0
12.1 2.5
12.0 3
12.0 2.5
9 months
2 times/week; 90 min/session
80%
10.5 2.0
10.5 2.9
7.9 0.6
7.9 0.8
Not reported
Not reported
6 months
1–3 times/week; 65 min/session
100%
14.6 2.3
15.0 2.4
8.9 1.5
8.3 2.1
8.0 1.1
8.9 1.4
5 months
Daily; 20 min/ session
74%
36.0 4.9
35.0 5.3
10.4 1.5
10.6 1.6
10.8 1.4
10.6 1.5
HR = heart rate; HRR = heart rate reserve; VO2peak = peak oxygen uptake; RM = repetition maximum.
diabetes research and clinical practice xxx (2014) xxx–xxx
Heyman [21]
Post-intervention
DIAB-6193; No. of Pages 8
4
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
Table 1 – Study characteristics.
DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
Table 2 – Summary risks of bias. Bias domain
Risk of bias
Adequate sequence generation
High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk
Allocation concealment Blinding (participants and personnel) Blinding (outcome assessors) Incomplete outcome data addressed Free of selective reporting Free of other bias
N studies (% of total) 0 5 1 0 5 1 0 4 2 0 5 1 1 2 3 0 1 5 0 0 6
(0.0%) (83.3%) (16.7%) (0.0%) (83.3%) (16.7%) (0.0%) (66.7%) (33.3%) (0.0%) (83.3%) (16.7%) (16.7%) (33.3%) (50.0%) (0.0%) (16.7%) (83.3%) (0.0%) (0.0%) (100.0%)
trial reports [12,21–25] (plus data from two companion publications [26,27]) (Table 1) were identified and retained in the final analyses. The mean age of patients ranged from 10.5 to 35.5 years and 28% were males. The six included trials were single-center trials conducted in Europe [12,21,22,24], USA [23] and Egypt [25]. All trials were published in English-language journals between 1985 and 2012. Five trials were classified as having an unclear risk of bias; one trial [12] was classified as having high risk of bias (Table 2). Only one trial was industrysponsored [21], while the others were sponsored by their institutions, governments or did not report any sponsorship. The primary outcome measure for this analysis was reported in four of the six trials. There was a great deal of variability in the number of participants involved in the studies as well as the dose of exercise provided in the interventions (Table 1). The studies included in the analysis enrolled between 6 [12] and 196 [25] participants (median, 16; interquartile range (IQR), 14 to 35). The frequency of exercise was as little as twice weekly [24] in one study to daily in another [12], with sessions lasting from 20 min [12] to 120 min [24]. In addition, exercise intensity varied both within and among studies (from 50 to 90% VO2peak). At baseline, three studies (using self-report by questionnaire) reported that participants on average exercised 4.8 h/ week (range, 1.8 to 6.8), while one additional study reported
5
less than one exercise session per week. The average preintervention HbA1c was 9.2% [(range, 7.9 to 12.1%); 77.1 mmol/ mol (range 62.8 to 108.8 mmol/mol)] and patients were receiving 1 U/kg insulin (range, 0.7 to 1.2).
3.1.
Primary outcome measure
Information was available for 323 participants with T1D that were randomly assigned to either exercise training (n = 215) or control (n = 108). Of the four trials (280 participants—191 assigned to exercise training; 89 assigned to control) that reported a measure of HbA1c in the trial, a significant reduction in post-treatment HbA1c was observed in the exercise group compared with standard care [Mean difference (MD) 0.78% (9 mmol/mol), 95% CI 1.14 (13 mmol/mol) to 0.41 (5 mmol/mol); p < 0.0001; Fig. 2]. It should be noted that 148 of the 191 participants assigned to exercise training were from a single study. The I2 statistic (0%) revealed no heterogeneity among the six studies. Subgroup analyses revealed greater reductions in HbA1c [MD 0.7% (8 mmol/mol); 95% CI 1.4 to 0.04 (15 to 1 mmol/mol), p = 0.04)] in patients having a baseline HbA1c > 8.5% [69 mmol/ mol (n = 3 trials)] compared to patients with a baseline HbA1c < 8.5% (69 mmol/mol) [MD 0.5 (6 mmol/mol); 95% CI 1.3 to 0.3 (14 mmol/mol to 4 mmol/mol), p = 0.22) (n = 1 trial)], but the difference between groups was not significant. As there were no studies lasting longer than 12 months, it was not possible to perform a subgroup analysis involving trial duration.
3.2.
Secondary outcome measures
Participants randomized to exercise displayed significantly greater improvements in maximal oxygen uptake compared with standard care [3.45 ml/kg/min (95% CI 0.59 to 6.31 ml/kg/ min; p = 0.02; I2 0%; 3 trials; 70 participants)] at trial end. Improvements in fitness were accompanied by differences in post-treatment body weight (MD 1.10, 95% CI 0.11 to 2.10 kg; p = 0.03; I2 0; 2 trials; 43 participants) but not body mass index (MD 0.02, 95% CI 0.40 to 0.37 kg/m2; p = 0.93; I2 57%; 3 trials; 265 participants) between groups (Table 3). Data for cardiovascular risk factors (LDL, HDL, VLDL, TG, TC, Apo-A, Apo-B) were not consistently available for all studies. Where they were available, no effect of exercise was observed at the end of trial follow-up (Table 3). Insulin requirements were reported by two trials with one showing significant reduction in daily insulin dosage (U/kg) in the exercise group of patients with HbA1c > 8.5% (69 mmol/ mol) [MD 0.40%, 95% CI 0.53 to 0.27%, p < 0.00001], while the other trial showed no difference in patients with
Fig. 2 – Post-treatment glycated hemoglobin (HbA1C)*. * Boxes and horizontal lines represent point estimates, varying in size according to the weight in the analysis, and 95% confidence intervals. Chi2 = Chi-squared df = degrees of freedom; CI = confidence interval; I2 = I-squared; IV = inverse variance; p = p-value; Tau2 = Tau-squared; Z = Z score. Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8
6
diabetes research and clinical practice xxx (2014) xxx–xxx
Table 3 – Secondary study outcomes. Outcome Measure Maximal oxygen uptake [12,22,23] Weight (kg) [21,24] Body mass index [22,24,26] Insulin dose (U/kg) [22,26] High density lipoprotein [12,21,22,26] Low density lipoprotein [12,21,22,26] Very low density lipoprotein [12] Total cholesterol [12,21,26] Total triglycerides [12,21,26] Apolipoprotein (a) [12] Apolipoprotein (b) [22]
Trials
Intervention
Control
3 2 3 2 4 4 1 3 3 1 1
35 24 183 168 183 183 6 163 163 9 20
35 19 82 70 84 84 7 62 62 7 22
Effect estimate (95% CI)
I2 (uCI)
MD 3.45 (0.59, 6.31) MD 1.10 (0.11, 2.10) MD 0.02 (0.40, 0.37) MD 0.21 (0.58, 0.16) SMD 0.34 (0.56, 1.23) SMD 0.02 (0.29, 0.25) MD 0.00 (0.14, 0.14) SMD 0.72 (1.70, 0.27) SMD 0.57 (1.19, 0.06) MD 0.13 (0.05, 0.31) MD 0.07 (0.19, 0.05)
0% (0%, 88%) 0% 57% (0%, 88%) 94% 85% (63%, 94%) 0% (0%, 70%) NE 77% (24%, 93%) 48% (0%, 85%) NE NE
CI = confidence intervals; I2 = I-squared; MD = mean difference; SMD = standardized mean difference; uCI = uncertainty intervals around the Isquared statistic.
HbA1c < 8.5% (69 mmol/mol) [MD 0.02%, 95% CI 0.14 to 0.10%, p = 0.75] compared with standard care. It should be noted that participants in the first study had a mean dosage of 1.2 U/kg at baseline, while the mean at baseline in the second study was 0.68 U/kg. The pooled estimate showed no significant difference between groups (Table 3).
3.3.
Adverse outcomes
Adverse event data were sparsely reported in the trials and only one trial reporting no post-exercise hypoglycemic events (n = 15) [24].
4.
Discussion
The current systematic review and meta-analysis supports observations of previous meta-analyses [10,11,13]. First, very few (only four) published randomized controlled trials could be identified comparing the effects of longer-term (i.e. >8 weeks) exercise training on HbA1c to performing no exercise in adults with T1D. Second, regular exercise training performed at least twice weekly for a minimum of eight weeks would appear to contribute to a significant absolute reduction in HbA1c in T1D individuals compared to controls receiving usual care. Finally, exercise training elicits significant improvements in fitness and may reduce the required insulin dose among persons with T1D. These finding are similar to those in previous reviews [10,11,13]. The effect of exercise training observed in this metaanalysis should be interpreted with caution as the majority of the participants (75%) included in the analysis derived from a single study with overwhelmingly positive findings [25]. Improvements in HbA1c were modest in two of the other three studies included [22,23], and a worsening of blood glucose control was actually found in one study [12]. In addition, two of the four studies included in the analysis were performed in the 1980s [12,23], since which considerable improvements in diabetes management have taken place (e.g. use of insulin analogues, insulin pumps, etc.). Taken together, it would seem that there is still insufficient evidence to confirm that structured exercise training elicits a significant improvement in glycemic control in persons with T1D. These preliminary results are, nonetheless, encouraging.
Recent systematic reviews and meta-analyses of exercise training in this population have found positive effects on glycemic control and cardiorespiratory fitness while using slightly different inclusion criteria [10,11,13]. Two of these analyses found statistically significant [0.3% [13] and 0.9% [11] (3 to 9 mmol/mol)] reductions in HbA1c with exercise training, while one did not [10]. The effect size in the present meta-analysis was almost identical to a recent systematic review of 11 exercise interventions [MD 0.9% (10 mmol/mol); CI: 1.5 (16 mmol/mol) to 0.3% (3 mmol/mol)] that included trials of unsupervised exercise and alternative forms of exercise (i.e. pilates) in children and adolescents with type 1 diabetes [13]. The similar effect size between meta-analyses suggests that improved glycemic control is possible with various forms of exercise training regardless of age. Systematic reviews including quasi-experimental trials [11], and sub-group analyses of trials restricted to adults [11] revealed significantly smaller effect sizes [0.3 to 0.4% (3 to 5 mmol/mol) absolute reductions]. The results presented here, while limited by a small sample size, seem to suggest that regular supervised exercise training could yield significant reductions in HbA1c in adult T1D patients. It also suggests that those with poor glycemic control at study entry may experience a greater improvement in glycemic control with increased supervised physical activity. The latter observation should be interpreted with caution as the results are driven largely by a single trial whose methods would not meet current standards. In addition, the observation that most studies included in the analysis suffered from an unclear risk of bias reinforces previous calls for adequately powered and rigorously designed trials of exercise training on glycemic control in T1D patients [28]. The aim of the current analysis was to examine the effect of regular, frequent exercise training on measures of glycemic control and cardiometabolic risk factors. Therefore, studies lasting less than eight weeks and studies with fewer than two training sessions weekly were excluded as those would be either too short or too infrequent to observe meaningful changes in metabolic health outcomes. Despite the rigorous inclusion criteria and the small number of studies included, there was still considerable variation in the intensity (50–90% VO2peak), frequency (2–7 times/week), duration (20–120 min) of the exercise sessions delivered and baseline physical activity levels in the six trials included in the final analysis (Table 1). As mean baseline physical activity ranged from 1.8 to 6.8 h
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
weekly, it is possible that the efficacy of the intervention was attenuated in those who were most active at baseline, thereby adding to the variability in the treatment effect. The variability in the dose of exercise delivered between trials may have also contributed to the variability in the effect sizes of the trials presented here and the differences observed between systematic reviews. It is generally well-accepted that increasing the time spent exercising per week and increasing the intensity of regular exercise is associated with greater longterm health benefits in individuals without type 1 diabetes [29,30]. Unfortunately, due to the relative lack of power in the current systematic review, it was not possible to perform multivariate regression analyses to determine the effects of exercise dose on glycemic control or cardiovascular risk factors for persons living with T1D, as others have done for trials of exercise in patients with T2D [31]. As such, little empirical evidence exists for an appropriate dose of physical activity for improving glycemic control in persons living with T1D.
4.1.
Limitations
In an effort to understand the effects of long-term exercise training on glycemic control, trials lasting less than eight weeks and interventions offered less than twice weekly were excluded from analyses. Additionally, few studies described the randomization process, none described concealment or allocation and fewer than half of the studies reported the number of patients withdrawing from the study. As a result, the risk of bias for these trials is unclear. As patient adherence rates to exercise training were not included in half of the trials included in the analysis where HbA1c was measured it is unclear if failure to adhere to the protocol contributed to null findings in these trials. In addition, this analysis unable to properly assess the frequency of hypoglycemia, a known risk and deterrent of exercise in T1DM. Finally, as most of the studies included in the analysis involved supervised exercise training, it is difficult to distinguish between the impact of exercise per se and the impact of having more frequent contact between the patient and the supervising trainer. In conclusion, while exercise is associated with a multitude of health benefits, the evidence to date regarding the benefits of physical activity for glycemic control in individuals with type 1 diabetes lacks strength. While the present analysis would suggest that regular exercise training is associated with a significant improvement in glycemic control and cardiorespiratory fitness in T1D patients, this mostly reflects the outcomes of a single study with a relatively large number of participants. There is little evidence to indicate the ideal duration, intensity or type of exercise that would be most appropriate for this population. The results presented here highlight the need for properly designed, adequately powered, randomized controlled trials of exercise on glycemic control in persons with T1D to guide clinical practice guidelines.
Contribution statement The authors contributed in the following ways: conception and design (AMAS, JM, SM), literature screening and conflict
7
resolution (JH, JM, JY), analysis and interpretation of data (AMAS, JM), drafting the article (AMAS, JH, JY) or providing important revisions (JM, SM). All authors approved of the final version of the article.
Funding sources Financial support for this systematic review was provided through the Robert Wallace Cameron Chair in Evidence-based Child Health (Grant # PEDS-311740), Manitoba Health Research Council and the Lawson Foundation (Grant # GRT 2012-059). JY was supported by a Manitoba Health Research Council/ Manitoba Institute of Child Health Postdoctoral Fellowship and a Canadian Institutes of Health Research Postdoctoral Fellowship. (MFE: 131495) JMM is supported by a New Investigator Award from the Canadian Institutes of Health Research (Grant #MSH-104361) and holds the Robert Wallace Cameron Chair in Evidence-based Child Health.
Conflicts of Interest statements The authors have no conflicts of interest to declare.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.diabres.2014.09.038.
references
[1] Chimen M, Kennedy A, Nirantharakumar K, Pang TT, Andrews R, Narendran P. What are the health benefits of physical activity in type 1 diabetes mellitus? A literature review. Diabetologia 2012;55(3):542–51. [2] Tielemans SM, Soedamah-Muthu SS, De Neve M, Toeller M, Chaturvedi N, Fuller JH, et al. Association of physical activity with all-cause mortality and incident and prevalent cardiovascular disease among patients with type 1 diabetes: the EURODIAB Prospective Complications Study. Diabetologia 2013;56(1):82–91. [3] Plotnikoff RC, Taylor LM, Wilson PM, Courneya KS, Sigal RJ, Birkett N, et al. Factors associated with physical activity in Canadian adults with diabetes. Med Sci Sports Exerc 2006;38(8):1526–34. [4] Newton KH, Wiltshire EJ, Elley CR. Pedometers and text messaging to increase physical activity: randomized controlled trial of adolescents with type 1 diabetes. Diabetes Care 2009;32(5):813–5. [5] Brazeau AS, Rabasa-Lhoret R, Strychar I, Mircescu H. Barriers to physical activity among patients with type 1 diabetes. Diabetes Care 2008;31(11):2108–9. [6] Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. The Canadian Diabetes Association clinical practice guidelines: physical activity and diabetes. Can J Diabetes 2013;37:S40–4. [7] American Diabetes Association. Physical activity/exercise and diabetes. Diabetes Care 2004;27(Suppl 1):S58–62. [8] Boule NG, Haddad E, Kenny GP, Wells GA, Sigal RJ. Effects of exercise on glycemic control and body mass in type 2
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8
8
diabetes research and clinical practice xxx (2014) xxx–xxx
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19] [20]
diabetes mellitus: a meta-analysis of controlled clinical trials. JAMA 2001;286(10):1218–27. Snowling NJ, Hopkins WG. Effects of different modes of exercise training on glucose control and risk factors for complications in type 2 diabetic patients: a meta-analysis. Diabetes Care 2006;29(11):2518–27. Kennedy A, Nirantharakumar K, Chimen M, Pang TT, Hemming K, Andrews RC, et al. Does exercise improve glycaemic control in type 1 diabetes? A systematic review and meta-analysis. PLoS One 2013;8(3):e58861. Tonoli C, Heyman E, Roelands B, Buyse L, Cheung SS, Berthoin S, et al. Effects of different types of acute and chronic (training) exercise on glycaemic control in type 1 diabetes mellitus: a meta-analysis. Sports Med 2012;42(12):1059–80. Wallberg-Henriksson H, Gunnarsson R, Rossner S, Wahren J. Long-term physical training in female type 1 (insulindependent) diabetic patients: absence of significant effect on glycaemic control and lipoprotein levels. Diabetologia 1986;29(1):53–7. MacMillan F, Kirk A, Mutrie N, Matthews L, Robertson K, Saunders DH. A systematic review of physical activity and sedentary behavior intervention studies in youth with type 1 diabetes: study characteristics, intervention design and efficacy. Pediatr Diabetes 2014;15(3):175–89. The Effectiveness of Structured Physical Activity for Reducing HbA1c in Patients with Type 1 Diabetes: A Meta-analysis. hhttp://www.crd.york.ac.uk/PROSPERO/ display_record.asp?ID=CRD42012003509i. Chandler J, Churchill R, Higgins J, Lasserson T, Tovey D. Methodological standards for the conduct of new Cochrane Intervention Reviews (Version 2.2). In: Methodological expectations of Cochrane intervention reviews (MECIR). The Cochrane Collaboration; 2012. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med 2009;6(7):e1000100. Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions version 5.0.1: the Cochrane collaboration; 2008. ˜ , David M, Higgins JPT, Douglas GA, Peter CGT, Peter JA Andrew DO, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d:5928. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21(11):1539–58. Ioannidis JP, Trikalinos TA. The appropriateness of asymmetry tests for publication bias in meta-analyses: a large survey. CMAJ 2007;176(8):1091–6.
[21] Heyman E, Toutain C, Delamarche P, Berthon P, Briard D, Youssef H, et al. Exercise training and cardiovascular risk factors in type 1 diabetic adolescent girls. Pediatr Exerc Sci 2007;19(4):408–19. [22] Laaksonen DE, Atalay M, Niskanen LK, Mustonen J, Sen CK, Lakka TA, et al. Aerobic exercise and the lipid profile in type 1 diabetic men: a randomized controlled trial. Med Sci Sports Exerc 2000;32(9):1541–8. [23] Landt KW, Campaigne BN, James FW, Sperling MA. Effects of exercise training on insulin sensitivity in adolescents with type I diabetes. Diabetes Care 1985;8(5):461–5. [24] Maggio AB, Rizzoli RR, Marchand LM, Ferrari S, Beghetti M, Farpour-Lambert NJ. Physical activity increases bone mineral density in children with type 1 diabetes. Med Sci Sports Exerc 2012;44(7):1206–11. [25] Salem MA, Aboelasrar MA, Elbarbary NS, Elhilaly RA, Refaat YM. Is exercise a therapeutic tool for improvement of cardiovascular risk factors in adolescents with type 1 diabetes mellitus? A randomised controlled trial. Diabetes Metab Syndr 2010;2(1):47. [26] Salem M, Abo-ElAsrar M, Elbarbary N, El-Hilaly R, Refaat Y. Is exercise a therapeutic tool for improvement of cardiovascular risk factors in adolescents with type 1 diabetes mellitus? A randomised controlled trial. Pediatr Diabetes 2010;11:35–113. [27] Maggio AB, Ferrari S, Kraenzlin M, Marchand LM, Schwitzgebel V, Beghetti M, et al. Decreased bone turnover in children and adolescents with well controlled type 1 diabetes. J Pediatr Endocrinol Metab 2010; 23(7):697–707. [28] Yardley J, Mollard R, Macintosh A, Macmillan F, Wicklow B, Berard L, et al. Vigorous intensity exercise for glycemic control in patients with type 1 diabetes. Can J Diabetes 2013;37(6):427–32. [29] Wen CP, Wai JP, Tsai MK, Yang YC, Cheng TY, Lee MC, et al. Minimum amount of physical activity for reduced mortality and extended life expectancy: a prospective cohort study. Lancet 2011;378(9798):1244–53. [30] Slentz CA, Houmard JA, Johnson JL, Bateman LA, Tanner CJ, McCartney JS, et al. Inactivity, exercise training and detraining, and plasma lipoproteins. STRRIDE: a randomized, controlled study of exercise intensity and amount. J Appl Physiol 2007;103(2):432–42. [31] Umpierre D, Ribeiro PA, Kramer CK, Leitao CB, Zucatti AT, Azevedo MJ, et al. Physical activity advice only or structured exercise training and association with HbA1c levels in type 2 diabetes: a systematic review and metaanalysis. JAMA 2011;305(17):1790–9.
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
Contents available at ScienceDirect
Diabetes Research and Clinical Practice journ al h ome pa ge : www .elsevier.co m/lo cate/diabres
A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes Jane E. Yardley a,b, Jacqueline Hay a, Ahmed M. Abou-Setta c,d, Seth D. Marks e, Jonathan McGavock a,* a
Manitoba Institute of Child Health, University of Manitoba, Winnipeg, MB, Canada University of Alberta, Augustana Campus, Camrose, AB, Canada c George & Fay Yee Centre for Healthcare Innovation, University of Manitoba, Winnipeg, MB, Canada d Winnipeg Regional Health Authority, Winnipeg, MB, Canada e Department of Pediatrics, University of Manitoba, Winnipeg, MB, Canada b
article info
abstract
Article history:
Aims: Conflicting evidence exists regarding the benefits of physical activity for long-term
Received 18 June 2014
blood glucose control in adults with type 1 diabetes (T1D). The object of this systematic
Received in revised form
review was to determine the effects of physical activity on long-term blood glucose control
27 August 2014
in T1D adults.
Accepted 15 September 2014
Methods: PubMed/Medline, Embase, CENTRAL, SPORTdiscus, Global Health and ICTRP were
Available online xxx
searched up to October 2013 for randomized trials of aerobic or resistance exercise training
Keywords:
months. The primary outcome was glycated hemoglobin (HbA1c). Secondary outcomes
in T1D adults. Exercises had to be performed at least twice weekly for a minimum of two Physical activity
included cardiorespiratory fitness and insulin dose.
Glycemic control
Results: Six randomized trials were identified (323 adults); sample sizes ranged from n = 6 to
Type 1 diabetes
n = 148 participants receiving the intervention. Five trials had an unknown risk of bias; one trial was deemed to be at high risk of bias. Exercise frequency varied from twice weekly to daily, with intensities (50–90% VO2peak), and session durations (20–120 min) varying widely. Four trials reported HbA1c, which decreased with exercise training (mean difference [MD] 0.78% (9 mmol/mol), 95% CI 1.14 (13 mmol/mol) to 0.41 (5 mmol/mol); p < 0.0001; I2 0%) compared with controls. Exercise training improved cardiorespiratory fitness by 3.45 ml/ kg/min (95% CI 0.59 to 6.31, p = 0.02, I2 0%) compared with controls. One trial reported an effect on insulin dose (MD 0.4 U/kg, 95% CI 0.53 to 0.27, p < 0.00001) compared to controls. Conclusion: There are currently insufficient well-designed studies to ascertain the true effect of exercise training on HbA1c in individuals with T1D, but current results are promising. # 2014 Published by Elsevier Ireland Ltd.
1.
Background
Regular physical activity is associated with multiple health benefits for individuals with type 1 diabetes (T1D) [1].
Increased physical activity is associated with a lower risk of complications and an increased life expectancy [2], however, more than 60% of adults with T1D do not achieve recommended levels of physical activity [3]. Children with T1D also face barriers to achieving recommended physical activity
* Corresponding author. Tel.: +1 204 480 1359. E-mail address: [email protected] (J. McGavock). http://dx.doi.org/10.1016/j.diabres.2014.09.038 0168-8227/# 2014 Published by Elsevier Ireland Ltd.
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8
2
diabetes research and clinical practice xxx (2014) xxx–xxx
targets [4]. In addition to conventional barriers to exercise (i.e. lack of time, work pressure, environmental conditions, low energy etc.), fear of hypoglycaemia [5] and a lack of evidence for any beneficial effect of exercise on long-term glycemic control likely contribute to low physical activity rates in individuals with T1D. Clinical practice guidelines recommend 150 min of moderate (50–70% of person’s maximum heart rate) to vigorous (>70% of person’s maximum heart rate) physical activity weekly for adults with T1D [6,7]. Due to the dearth of evidence available on this topic, the appropriate dose (type, duration, frequency, intensity) of physical activity needed for improved glycemic control in this population remains unclear. In contrast to the well-documented benefits of physical activity for glycemic control among individuals with type 2 diabetes (T2D) [8,9], experimental studies of exercise training in T1D individuals have yielded conflicting results. Recent systematic reviews of trials of physical activity for glycemic control in persons with T1D failed to address these inconsistencies as they have either (1) included quasi experimental trials [1,10]; (2) pooled data from short term training studies (2–8 weeks) and acute studies [11] and/or (3) included studies of interventions that combined diet and exercise training [1,10,12]. One recent meta-analysis of physical activity and/or sedentary behavior intervention studies in children with T1D, included only randomized controlled trials [13] and found a significant improvement in glycemic control (HbA1c) with exercise training [mean difference 0.85% (95% CI, 1.45 to 0.24%)]. A comparable systematic review and meta-analysis in adults is currently unavailable in the literature. Therefore, a focused systematic review of the literature was conducted to examine the state of the scientific literature regarding randomized trials of exercise training interventions lasting more than eight weeks with exercise performed at least twice weekly in individuals with type 1 diabetes.
2.
Methods
Using a previously published protocol [14], a systematic review applying methodological approaches outlined in the Methodological Expectations of Cochrane Intervention Reviews [15] was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses criteria [16]. An expert panel from multiple fields (pediatrics, endocrinology, clinical epidemiology, exercise physiology) formulated the review question, reviewed the search strategies and review methods, and provided input throughout the review process.
2.1. Populations, interventions, comparators, outcome measures, settings and study designs (PICOS)
Inclusion criteria for trials included in the review were: (1) the majority of participants (80%) had to be between 15 and 50 years old and diagnosed with type 1 diabetes mellitus; (2) trials had to be prospective, randomized and controlled with a non-exercising T1D control group; and (3) the exercise intervention consisted of supervised or unsupervised aerobic, resistance or combined physical activity offered at least twice weekly for longer than eight weeks. Trials were excluded if they (1) involved animals; (2) included participants with cystic fibrosis, type 2 diabetes mellitus, monogenic forms of diabetes, secondary diabetes, or with HbA1c < 7.0% (53 mmol/mol); (3) included co-dietary intervention, behavioral modification not directly related to physical activity, or acute exercise only (i.e. single exercise session); (4) used a quasi-experimental design (e.g., alternate randomization, randomization according to hospital number, non-randomized trials (e.g. cohort and casecontrol studies, cross-over or cluster randomized trials The primary outcome measure was HbA1c assessed at the end of the active intervention. Secondary outcomes included maximal oxygen uptake, weight (kg), daily insulin dose, cardiovascular risk factors, body mass index and occurrence of adverse events in both groups.
2.2.
Trials without language restriction published up to October 2013 that met inclusion criteria were identified using individualized search strategies prepared for the following databases: PubMed (National Library of Medicine), EMBASE (Ovid), CENTRAL (the Cochrane Library—Wiley), CINAHL (EbscoHost), Global Health (Ovid), and SPORTDiscus (EbscoHost). The PubMed strategy is presented in Appendix 1. Further, a forward search was performed in Scopus and Web of Science to identify additional citations. To identify ongoing or planned trials, we searched the World Health Organization’s International Clinical Trials Registry Platform for relevant registrations. Finally, abstracts, conference proceedings and the references lists of relevant narrative and systematic reviews were searched and included trials were hand-searched for possible relevant citations. Two reviewers independently screened the titles and abstracts (when available) of search results to determine if studies met the inclusion criteria. Reports were classified as: include, exclude, unclear, or duplicate of another citation. The full text of citations classified as ‘‘include’’ or ‘‘unclear’’ by either reviewer were retrieved for formal review. Two reviewers independently assessed the full text of each potentially included trial by using a standardized form outlining the predetermined inclusion/exclusion criteria. Disagreements were resolved by discussion between the two reviewers or by another reviewer’s adjudication, as needed.
2.3. The primary research question for this analysis was: ‘‘In adolescents and adults (aged 15–50 years) living with T1D, does a structured physical activity intervention lead to clinically meaningful reductions in glycated hemoglobin (HbA1c) compared to a non-exercising control condition?’’ To address this question, only randomized, controlled trials of adults diagnosed with T1D were included.
Search strategy for identification of studies
Data extraction and management
Two independent reviewers (JY, JH) extracted information using standardized, piloted forms. Disagreements were resolved through consensus and with the assistance of another reviewer (JM), if consensus was not achieved. From each trial the following information was extracted: author, year and language of publication, source of funding, study
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
design, study population, patient characteristics, intervention and its comparator, and outcomes of interest.
2.4.
Assessment of risk of bias
Internal validity of included trials was assessed using the Cochrane Collaboration Risk of Bias tool [17,18], which consists of six domains (sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and ‘‘other’’ sources of bias) and a categorization of the overall risk of bias. Each domain was rated as ‘‘low risk’’, ‘‘unclear risk’’, or ‘‘high risk.’’ Overall assessments were based on the responses to individual domains, judged by two independent reviewers (AA, JY) following the recommendations detailed in the Cochrane Handbook for Systematic Reviewers and related publications [17,18]. The criteria used to assess each domain are summarized in Appendix 2. If one or more domains were judged as having a high risk of bias, the trial was classified as being at high risk of bias. If all components of a trial were classified as having a low risk of bias, the trial was considered to be at low risk of bias. In cases of mixed assessments of low and unclear risk of bias, or where all assessments were unclear risk of bias, overall classification was unclear risk of bias.
2.5.
Measures of treatment effect
Data from included trials were analyzed using Review Manager (RevMan version 5.2, The Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark), and Microsoft Excel (Excel v.14, Microsoft Corp., Redmond, WA, USA). Pooled
Records identified through database searching (n = 14,660 citations)
continuous data were expressed as mean differences, with 95% confidence intervals (random-effects model). Statistical heterogeneity of the data was quantified using the I-squared test with 95% uncertainty intervals [19]. Publication bias was assessed by viewing the overlap of confidence intervals, and using funnel plot techniques [20]. All tests of statistical inference reflect a 2-sided a of 0.05.
2.6.
Subgroup analysis
For the primary outcome of reduction in post-treatment HbA1c, the following a priori subgroup analyses were performed, where data were available: participants with baseline HbA1c > 8.5% (69 mmol/mol) versus HbA1c < 8.5% (population); and short (12 months) trial durations (timing). Aggregate results were also examined to determine whether large effect sizes were related to high risks of bias. Finally, whether or not sources of funding influenced results was examined.
3.
Results
Of the 14,663 citations obtained from electronic and handsearches, 51 trial reports that potentially examined the effect of exercise on glycemic control in T1D individuals were identified. Forty-three reports were excluded: inappropriate population (n = 5); inappropriate intervention/control (n = 2); inappropriate study design (n = 21); full-text manuscripts unavailable for review through library services (n = 13); unable to translate non-English language (n = 2) (Fig. 1). Six unique
Additional records identified through other sources (n = 2 citations)
Records after duplicates removed (n = 11,772)
Records screened for eligibility (n = 11,772)
Full-text articles assessed for eligibility (n = 50)
3
Records excluded (n = 11,722)
Full-text articles excluded: (n = 43) Inappropriate population (n = 5) Inappropriate intervention/control (n = 2) Inappropriate study design (n = 21) Unavailable for review through library services (n = 13) Non-English language – unable to translate (n = 2)
Studies included in review (n = 6 primary publication) (n = 1 companion publications)
Fig. 1 – Modified PRISMA flow-chart. Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
Study
N (training/ control)
Intervention
Program duration
Weekly exercise frequency and duration
Adherence to training
Mean age (years)
Mean HbA1c (%)
Baseline
9/7
Laaksonen [22]
28/28
Landt [23]
9/6
Maggio [24]
15/12
Salem [26]
148/48
WallbergHenriksson [12]
6/7
Supervised high intensity aerobic (80–90% HRR) and ‘‘strength’’ training (exercises not described) Supervised aerobic (50–80% VO2peak) exercise. Intensity increased through intervention. Supervised aerobic and high intensity training (80–85% maximum HR) Supervised ‘‘weight bearing physical activity’’ (HR 140 beats/ minute) Supervised aerobic (65–85% HRR), high intensity and resistance (50–100% 10RM) exercise Unsupervised aerobic exercise (60–90% VO2peak–intensity increased progressively)
Training
Control
Training
Control
Control
6 months
2 times/week 120 min/session (supervised) Once/week 60 min/session (unsupervised)
62–100% (supervised) 52–89% (unsupervised)
15.9 1.5
16.3 1.2
8.0 1.4
8.0 1.4
Not reported
Not reported
3–4 months
3–5 times/week; 30–60 min/session
Not reported
32.5 5.7
29.5 6.3
8.2 1.1
8.3 1.3
8.0 1.0
8.5 1.6
3 months
3 times/week; 65 min/session
Not reported
16.1 2.4
15.9 0.7
12.1 3.0
12.1 2.5
12.0 3
12.0 2.5
9 months
2 times/week; 90 min/session
80%
10.5 2.0
10.5 2.9
7.9 0.6
7.9 0.8
Not reported
Not reported
6 months
1–3 times/week; 65 min/session
100%
14.6 2.3
15.0 2.4
8.9 1.5
8.3 2.1
8.0 1.1
8.9 1.4
5 months
Daily; 20 min/ session
74%
36.0 4.9
35.0 5.3
10.4 1.5
10.6 1.6
10.8 1.4
10.6 1.5
HR = heart rate; HRR = heart rate reserve; VO2peak = peak oxygen uptake; RM = repetition maximum.
diabetes research and clinical practice xxx (2014) xxx–xxx
Heyman [21]
Post-intervention
DIAB-6193; No. of Pages 8
4
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
Table 1 – Study characteristics.
DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
Table 2 – Summary risks of bias. Bias domain
Risk of bias
Adequate sequence generation
High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk High risk Unclear risk Low risk
Allocation concealment Blinding (participants and personnel) Blinding (outcome assessors) Incomplete outcome data addressed Free of selective reporting Free of other bias
N studies (% of total) 0 5 1 0 5 1 0 4 2 0 5 1 1 2 3 0 1 5 0 0 6
(0.0%) (83.3%) (16.7%) (0.0%) (83.3%) (16.7%) (0.0%) (66.7%) (33.3%) (0.0%) (83.3%) (16.7%) (16.7%) (33.3%) (50.0%) (0.0%) (16.7%) (83.3%) (0.0%) (0.0%) (100.0%)
trial reports [12,21–25] (plus data from two companion publications [26,27]) (Table 1) were identified and retained in the final analyses. The mean age of patients ranged from 10.5 to 35.5 years and 28% were males. The six included trials were single-center trials conducted in Europe [12,21,22,24], USA [23] and Egypt [25]. All trials were published in English-language journals between 1985 and 2012. Five trials were classified as having an unclear risk of bias; one trial [12] was classified as having high risk of bias (Table 2). Only one trial was industrysponsored [21], while the others were sponsored by their institutions, governments or did not report any sponsorship. The primary outcome measure for this analysis was reported in four of the six trials. There was a great deal of variability in the number of participants involved in the studies as well as the dose of exercise provided in the interventions (Table 1). The studies included in the analysis enrolled between 6 [12] and 196 [25] participants (median, 16; interquartile range (IQR), 14 to 35). The frequency of exercise was as little as twice weekly [24] in one study to daily in another [12], with sessions lasting from 20 min [12] to 120 min [24]. In addition, exercise intensity varied both within and among studies (from 50 to 90% VO2peak). At baseline, three studies (using self-report by questionnaire) reported that participants on average exercised 4.8 h/ week (range, 1.8 to 6.8), while one additional study reported
5
less than one exercise session per week. The average preintervention HbA1c was 9.2% [(range, 7.9 to 12.1%); 77.1 mmol/ mol (range 62.8 to 108.8 mmol/mol)] and patients were receiving 1 U/kg insulin (range, 0.7 to 1.2).
3.1.
Primary outcome measure
Information was available for 323 participants with T1D that were randomly assigned to either exercise training (n = 215) or control (n = 108). Of the four trials (280 participants—191 assigned to exercise training; 89 assigned to control) that reported a measure of HbA1c in the trial, a significant reduction in post-treatment HbA1c was observed in the exercise group compared with standard care [Mean difference (MD) 0.78% (9 mmol/mol), 95% CI 1.14 (13 mmol/mol) to 0.41 (5 mmol/mol); p < 0.0001; Fig. 2]. It should be noted that 148 of the 191 participants assigned to exercise training were from a single study. The I2 statistic (0%) revealed no heterogeneity among the six studies. Subgroup analyses revealed greater reductions in HbA1c [MD 0.7% (8 mmol/mol); 95% CI 1.4 to 0.04 (15 to 1 mmol/mol), p = 0.04)] in patients having a baseline HbA1c > 8.5% [69 mmol/ mol (n = 3 trials)] compared to patients with a baseline HbA1c < 8.5% (69 mmol/mol) [MD 0.5 (6 mmol/mol); 95% CI 1.3 to 0.3 (14 mmol/mol to 4 mmol/mol), p = 0.22) (n = 1 trial)], but the difference between groups was not significant. As there were no studies lasting longer than 12 months, it was not possible to perform a subgroup analysis involving trial duration.
3.2.
Secondary outcome measures
Participants randomized to exercise displayed significantly greater improvements in maximal oxygen uptake compared with standard care [3.45 ml/kg/min (95% CI 0.59 to 6.31 ml/kg/ min; p = 0.02; I2 0%; 3 trials; 70 participants)] at trial end. Improvements in fitness were accompanied by differences in post-treatment body weight (MD 1.10, 95% CI 0.11 to 2.10 kg; p = 0.03; I2 0; 2 trials; 43 participants) but not body mass index (MD 0.02, 95% CI 0.40 to 0.37 kg/m2; p = 0.93; I2 57%; 3 trials; 265 participants) between groups (Table 3). Data for cardiovascular risk factors (LDL, HDL, VLDL, TG, TC, Apo-A, Apo-B) were not consistently available for all studies. Where they were available, no effect of exercise was observed at the end of trial follow-up (Table 3). Insulin requirements were reported by two trials with one showing significant reduction in daily insulin dosage (U/kg) in the exercise group of patients with HbA1c > 8.5% (69 mmol/ mol) [MD 0.40%, 95% CI 0.53 to 0.27%, p < 0.00001], while the other trial showed no difference in patients with
Fig. 2 – Post-treatment glycated hemoglobin (HbA1C)*. * Boxes and horizontal lines represent point estimates, varying in size according to the weight in the analysis, and 95% confidence intervals. Chi2 = Chi-squared df = degrees of freedom; CI = confidence interval; I2 = I-squared; IV = inverse variance; p = p-value; Tau2 = Tau-squared; Z = Z score. Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8
6
diabetes research and clinical practice xxx (2014) xxx–xxx
Table 3 – Secondary study outcomes. Outcome Measure Maximal oxygen uptake [12,22,23] Weight (kg) [21,24] Body mass index [22,24,26] Insulin dose (U/kg) [22,26] High density lipoprotein [12,21,22,26] Low density lipoprotein [12,21,22,26] Very low density lipoprotein [12] Total cholesterol [12,21,26] Total triglycerides [12,21,26] Apolipoprotein (a) [12] Apolipoprotein (b) [22]
Trials
Intervention
Control
3 2 3 2 4 4 1 3 3 1 1
35 24 183 168 183 183 6 163 163 9 20
35 19 82 70 84 84 7 62 62 7 22
Effect estimate (95% CI)
I2 (uCI)
MD 3.45 (0.59, 6.31) MD 1.10 (0.11, 2.10) MD 0.02 (0.40, 0.37) MD 0.21 (0.58, 0.16) SMD 0.34 (0.56, 1.23) SMD 0.02 (0.29, 0.25) MD 0.00 (0.14, 0.14) SMD 0.72 (1.70, 0.27) SMD 0.57 (1.19, 0.06) MD 0.13 (0.05, 0.31) MD 0.07 (0.19, 0.05)
0% (0%, 88%) 0% 57% (0%, 88%) 94% 85% (63%, 94%) 0% (0%, 70%) NE 77% (24%, 93%) 48% (0%, 85%) NE NE
CI = confidence intervals; I2 = I-squared; MD = mean difference; SMD = standardized mean difference; uCI = uncertainty intervals around the Isquared statistic.
HbA1c < 8.5% (69 mmol/mol) [MD 0.02%, 95% CI 0.14 to 0.10%, p = 0.75] compared with standard care. It should be noted that participants in the first study had a mean dosage of 1.2 U/kg at baseline, while the mean at baseline in the second study was 0.68 U/kg. The pooled estimate showed no significant difference between groups (Table 3).
3.3.
Adverse outcomes
Adverse event data were sparsely reported in the trials and only one trial reporting no post-exercise hypoglycemic events (n = 15) [24].
4.
Discussion
The current systematic review and meta-analysis supports observations of previous meta-analyses [10,11,13]. First, very few (only four) published randomized controlled trials could be identified comparing the effects of longer-term (i.e. >8 weeks) exercise training on HbA1c to performing no exercise in adults with T1D. Second, regular exercise training performed at least twice weekly for a minimum of eight weeks would appear to contribute to a significant absolute reduction in HbA1c in T1D individuals compared to controls receiving usual care. Finally, exercise training elicits significant improvements in fitness and may reduce the required insulin dose among persons with T1D. These finding are similar to those in previous reviews [10,11,13]. The effect of exercise training observed in this metaanalysis should be interpreted with caution as the majority of the participants (75%) included in the analysis derived from a single study with overwhelmingly positive findings [25]. Improvements in HbA1c were modest in two of the other three studies included [22,23], and a worsening of blood glucose control was actually found in one study [12]. In addition, two of the four studies included in the analysis were performed in the 1980s [12,23], since which considerable improvements in diabetes management have taken place (e.g. use of insulin analogues, insulin pumps, etc.). Taken together, it would seem that there is still insufficient evidence to confirm that structured exercise training elicits a significant improvement in glycemic control in persons with T1D. These preliminary results are, nonetheless, encouraging.
Recent systematic reviews and meta-analyses of exercise training in this population have found positive effects on glycemic control and cardiorespiratory fitness while using slightly different inclusion criteria [10,11,13]. Two of these analyses found statistically significant [0.3% [13] and 0.9% [11] (3 to 9 mmol/mol)] reductions in HbA1c with exercise training, while one did not [10]. The effect size in the present meta-analysis was almost identical to a recent systematic review of 11 exercise interventions [MD 0.9% (10 mmol/mol); CI: 1.5 (16 mmol/mol) to 0.3% (3 mmol/mol)] that included trials of unsupervised exercise and alternative forms of exercise (i.e. pilates) in children and adolescents with type 1 diabetes [13]. The similar effect size between meta-analyses suggests that improved glycemic control is possible with various forms of exercise training regardless of age. Systematic reviews including quasi-experimental trials [11], and sub-group analyses of trials restricted to adults [11] revealed significantly smaller effect sizes [0.3 to 0.4% (3 to 5 mmol/mol) absolute reductions]. The results presented here, while limited by a small sample size, seem to suggest that regular supervised exercise training could yield significant reductions in HbA1c in adult T1D patients. It also suggests that those with poor glycemic control at study entry may experience a greater improvement in glycemic control with increased supervised physical activity. The latter observation should be interpreted with caution as the results are driven largely by a single trial whose methods would not meet current standards. In addition, the observation that most studies included in the analysis suffered from an unclear risk of bias reinforces previous calls for adequately powered and rigorously designed trials of exercise training on glycemic control in T1D patients [28]. The aim of the current analysis was to examine the effect of regular, frequent exercise training on measures of glycemic control and cardiometabolic risk factors. Therefore, studies lasting less than eight weeks and studies with fewer than two training sessions weekly were excluded as those would be either too short or too infrequent to observe meaningful changes in metabolic health outcomes. Despite the rigorous inclusion criteria and the small number of studies included, there was still considerable variation in the intensity (50–90% VO2peak), frequency (2–7 times/week), duration (20–120 min) of the exercise sessions delivered and baseline physical activity levels in the six trials included in the final analysis (Table 1). As mean baseline physical activity ranged from 1.8 to 6.8 h
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8 diabetes research and clinical practice xxx (2014) xxx–xxx
weekly, it is possible that the efficacy of the intervention was attenuated in those who were most active at baseline, thereby adding to the variability in the treatment effect. The variability in the dose of exercise delivered between trials may have also contributed to the variability in the effect sizes of the trials presented here and the differences observed between systematic reviews. It is generally well-accepted that increasing the time spent exercising per week and increasing the intensity of regular exercise is associated with greater longterm health benefits in individuals without type 1 diabetes [29,30]. Unfortunately, due to the relative lack of power in the current systematic review, it was not possible to perform multivariate regression analyses to determine the effects of exercise dose on glycemic control or cardiovascular risk factors for persons living with T1D, as others have done for trials of exercise in patients with T2D [31]. As such, little empirical evidence exists for an appropriate dose of physical activity for improving glycemic control in persons living with T1D.
4.1.
Limitations
In an effort to understand the effects of long-term exercise training on glycemic control, trials lasting less than eight weeks and interventions offered less than twice weekly were excluded from analyses. Additionally, few studies described the randomization process, none described concealment or allocation and fewer than half of the studies reported the number of patients withdrawing from the study. As a result, the risk of bias for these trials is unclear. As patient adherence rates to exercise training were not included in half of the trials included in the analysis where HbA1c was measured it is unclear if failure to adhere to the protocol contributed to null findings in these trials. In addition, this analysis unable to properly assess the frequency of hypoglycemia, a known risk and deterrent of exercise in T1DM. Finally, as most of the studies included in the analysis involved supervised exercise training, it is difficult to distinguish between the impact of exercise per se and the impact of having more frequent contact between the patient and the supervising trainer. In conclusion, while exercise is associated with a multitude of health benefits, the evidence to date regarding the benefits of physical activity for glycemic control in individuals with type 1 diabetes lacks strength. While the present analysis would suggest that regular exercise training is associated with a significant improvement in glycemic control and cardiorespiratory fitness in T1D patients, this mostly reflects the outcomes of a single study with a relatively large number of participants. There is little evidence to indicate the ideal duration, intensity or type of exercise that would be most appropriate for this population. The results presented here highlight the need for properly designed, adequately powered, randomized controlled trials of exercise on glycemic control in persons with T1D to guide clinical practice guidelines.
Contribution statement The authors contributed in the following ways: conception and design (AMAS, JM, SM), literature screening and conflict
7
resolution (JH, JM, JY), analysis and interpretation of data (AMAS, JM), drafting the article (AMAS, JH, JY) or providing important revisions (JM, SM). All authors approved of the final version of the article.
Funding sources Financial support for this systematic review was provided through the Robert Wallace Cameron Chair in Evidence-based Child Health (Grant # PEDS-311740), Manitoba Health Research Council and the Lawson Foundation (Grant # GRT 2012-059). JY was supported by a Manitoba Health Research Council/ Manitoba Institute of Child Health Postdoctoral Fellowship and a Canadian Institutes of Health Research Postdoctoral Fellowship. (MFE: 131495) JMM is supported by a New Investigator Award from the Canadian Institutes of Health Research (Grant #MSH-104361) and holds the Robert Wallace Cameron Chair in Evidence-based Child Health.
Conflicts of Interest statements The authors have no conflicts of interest to declare.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.diabres.2014.09.038.
references
[1] Chimen M, Kennedy A, Nirantharakumar K, Pang TT, Andrews R, Narendran P. What are the health benefits of physical activity in type 1 diabetes mellitus? A literature review. Diabetologia 2012;55(3):542–51. [2] Tielemans SM, Soedamah-Muthu SS, De Neve M, Toeller M, Chaturvedi N, Fuller JH, et al. Association of physical activity with all-cause mortality and incident and prevalent cardiovascular disease among patients with type 1 diabetes: the EURODIAB Prospective Complications Study. Diabetologia 2013;56(1):82–91. [3] Plotnikoff RC, Taylor LM, Wilson PM, Courneya KS, Sigal RJ, Birkett N, et al. Factors associated with physical activity in Canadian adults with diabetes. Med Sci Sports Exerc 2006;38(8):1526–34. [4] Newton KH, Wiltshire EJ, Elley CR. Pedometers and text messaging to increase physical activity: randomized controlled trial of adolescents with type 1 diabetes. Diabetes Care 2009;32(5):813–5. [5] Brazeau AS, Rabasa-Lhoret R, Strychar I, Mircescu H. Barriers to physical activity among patients with type 1 diabetes. Diabetes Care 2008;31(11):2108–9. [6] Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. The Canadian Diabetes Association clinical practice guidelines: physical activity and diabetes. Can J Diabetes 2013;37:S40–4. [7] American Diabetes Association. Physical activity/exercise and diabetes. Diabetes Care 2004;27(Suppl 1):S58–62. [8] Boule NG, Haddad E, Kenny GP, Wells GA, Sigal RJ. Effects of exercise on glycemic control and body mass in type 2
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
DIAB-6193; No. of Pages 8
8
diabetes research and clinical practice xxx (2014) xxx–xxx
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19] [20]
diabetes mellitus: a meta-analysis of controlled clinical trials. JAMA 2001;286(10):1218–27. Snowling NJ, Hopkins WG. Effects of different modes of exercise training on glucose control and risk factors for complications in type 2 diabetic patients: a meta-analysis. Diabetes Care 2006;29(11):2518–27. Kennedy A, Nirantharakumar K, Chimen M, Pang TT, Hemming K, Andrews RC, et al. Does exercise improve glycaemic control in type 1 diabetes? A systematic review and meta-analysis. PLoS One 2013;8(3):e58861. Tonoli C, Heyman E, Roelands B, Buyse L, Cheung SS, Berthoin S, et al. Effects of different types of acute and chronic (training) exercise on glycaemic control in type 1 diabetes mellitus: a meta-analysis. Sports Med 2012;42(12):1059–80. Wallberg-Henriksson H, Gunnarsson R, Rossner S, Wahren J. Long-term physical training in female type 1 (insulindependent) diabetic patients: absence of significant effect on glycaemic control and lipoprotein levels. Diabetologia 1986;29(1):53–7. MacMillan F, Kirk A, Mutrie N, Matthews L, Robertson K, Saunders DH. A systematic review of physical activity and sedentary behavior intervention studies in youth with type 1 diabetes: study characteristics, intervention design and efficacy. Pediatr Diabetes 2014;15(3):175–89. The Effectiveness of Structured Physical Activity for Reducing HbA1c in Patients with Type 1 Diabetes: A Meta-analysis. hhttp://www.crd.york.ac.uk/PROSPERO/ display_record.asp?ID=CRD42012003509i. Chandler J, Churchill R, Higgins J, Lasserson T, Tovey D. Methodological standards for the conduct of new Cochrane Intervention Reviews (Version 2.2). In: Methodological expectations of Cochrane intervention reviews (MECIR). The Cochrane Collaboration; 2012. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med 2009;6(7):e1000100. Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions version 5.0.1: the Cochrane collaboration; 2008. ˜ , David M, Higgins JPT, Douglas GA, Peter CGT, Peter JA Andrew DO, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d:5928. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21(11):1539–58. Ioannidis JP, Trikalinos TA. The appropriateness of asymmetry tests for publication bias in meta-analyses: a large survey. CMAJ 2007;176(8):1091–6.
[21] Heyman E, Toutain C, Delamarche P, Berthon P, Briard D, Youssef H, et al. Exercise training and cardiovascular risk factors in type 1 diabetic adolescent girls. Pediatr Exerc Sci 2007;19(4):408–19. [22] Laaksonen DE, Atalay M, Niskanen LK, Mustonen J, Sen CK, Lakka TA, et al. Aerobic exercise and the lipid profile in type 1 diabetic men: a randomized controlled trial. Med Sci Sports Exerc 2000;32(9):1541–8. [23] Landt KW, Campaigne BN, James FW, Sperling MA. Effects of exercise training on insulin sensitivity in adolescents with type I diabetes. Diabetes Care 1985;8(5):461–5. [24] Maggio AB, Rizzoli RR, Marchand LM, Ferrari S, Beghetti M, Farpour-Lambert NJ. Physical activity increases bone mineral density in children with type 1 diabetes. Med Sci Sports Exerc 2012;44(7):1206–11. [25] Salem MA, Aboelasrar MA, Elbarbary NS, Elhilaly RA, Refaat YM. Is exercise a therapeutic tool for improvement of cardiovascular risk factors in adolescents with type 1 diabetes mellitus? A randomised controlled trial. Diabetes Metab Syndr 2010;2(1):47. [26] Salem M, Abo-ElAsrar M, Elbarbary N, El-Hilaly R, Refaat Y. Is exercise a therapeutic tool for improvement of cardiovascular risk factors in adolescents with type 1 diabetes mellitus? A randomised controlled trial. Pediatr Diabetes 2010;11:35–113. [27] Maggio AB, Ferrari S, Kraenzlin M, Marchand LM, Schwitzgebel V, Beghetti M, et al. Decreased bone turnover in children and adolescents with well controlled type 1 diabetes. J Pediatr Endocrinol Metab 2010; 23(7):697–707. [28] Yardley J, Mollard R, Macintosh A, Macmillan F, Wicklow B, Berard L, et al. Vigorous intensity exercise for glycemic control in patients with type 1 diabetes. Can J Diabetes 2013;37(6):427–32. [29] Wen CP, Wai JP, Tsai MK, Yang YC, Cheng TY, Lee MC, et al. Minimum amount of physical activity for reduced mortality and extended life expectancy: a prospective cohort study. Lancet 2011;378(9798):1244–53. [30] Slentz CA, Houmard JA, Johnson JL, Bateman LA, Tanner CJ, McCartney JS, et al. Inactivity, exercise training and detraining, and plasma lipoproteins. STRRIDE: a randomized, controlled study of exercise intensity and amount. J Appl Physiol 2007;103(2):432–42. [31] Umpierre D, Ribeiro PA, Kramer CK, Leitao CB, Zucatti AT, Azevedo MJ, et al. Physical activity advice only or structured exercise training and association with HbA1c levels in type 2 diabetes: a systematic review and metaanalysis. JAMA 2011;305(17):1790–9.
Please cite this article in press as: Yardley JE, et al. A systematic review and meta-analysis of exercise interventions in adults with type 1 diabetes. Diabetes Res Clin Pract (2014), http://dx.doi.org/10.1016/j.diabres.2014.09.038
