Resistance Training and Older Adults with Type 2 Diabetes Mellitus: Strength of the Evidence

Hindawi Publishing Corporation
Journal of Aging Research
Volume 2012, Article ID 284635, 12 pages
doi:10.1155/2012/284635
Review Article
Resistance Training and Older Adults with Type 2 Diabetes
Mellitus: Strength of the Evidence
Nina Hovanec, 1 Anuradha Sawant,2 Tom J. Overend,2
Robert J. Petrella,3 and Anthony A. Vandervoort 2
1 Health and Rehabilitation Sciences Graduate Program, Western University, London, ON, Canada N6G 1H1
2 School of Physical Therapy and Center for Physical Activity and Aging, Faculty of Health Sciences, Western University,
London, ON, Canada N6G 1H1
3 Department of Family Medicine, Western University, London, ON, Canada N6G 1H1
Correspondence should be addressed to Anthony A. Vandervoort, [e-mail]
Received 30 April 2012; Revised 5 July 2012; Accepted 12 July 2012
Academic Editor: Bijan Najafi
Copyright © 2012 Nina Hovanec et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Objective. This paper analyzes the effects of resistance training (RT) on metabolic, neuromuscular, and cardiovascular functions
in older adults (mean age ≥ 65 years) with type 2 diabetes (T2DM). Research Design and Methods. A systematic review conducted
by two reviewers of the published literature produced 3 records based on 2 randomized controlled trials that assessed the effect
of RT on disease process measures and musculoskeletal/body composition measures. Statistical, Comprehensive Meta-Analysis
(version 2) software was used to compute Hedge’s g, and results were calculated using the random effects model to account
for methodological differences amongst studies. Results. Largest effect of RT was seen on muscle strength; especially lower body
strength, while the point estimate effect on body composition was small and not statistically significant. The cumulative point
estimate for the T2DM disease process measures was moderate and statistically significant. Conclusions. RT generally had a positive
effect on musculoskeletal, body composition, and T2DM disease processes measures, with tentative conclusions based on a low
number of completed RCTs. Thus, more research is needed on such programs for older adults (≥65 years) with T2DM.
1. Introduction
Type 2 diabetes mellitus (T2DM) in older adults is an emerg-
ing epidemic [1]. (For the purpose of this paper, the term
“older adults” refers to individuals who are at least 65 years
old.) It is an age-prevalent metabolic disorder, characterized
by insulin resistance with relative insulin deficiency [2, 3],
with the highest prevalence found in individuals who are
80 years or older—an estimated number of 40 million is
expected in the United States by the year 2050 [1].
Physical activity is considered to be a cornerstone
of T2DM prevention and management [2, 4], and it is
important to have accurate information for health care
organizations to integrate into their knowledge management
strategies [5]. Physical activity refers to “the expenditure
of energy above that of resting by contraction of skeletal
muscle to produce bodily movement,” while exercise is “a
type of physical activity that involves planned, structured and
repetitive bodily movement performed for the purpose of
improving physical fitness” [6, page 359]. Physical activity
and exercise will be used interchangeably in this paper.
In terms of physical activity as a management method
in populations living with T2DM, traditional focus has
been given to aerobic training (AT) interventions [7, 8].
Aerobic training activates large muscle groups to perform
activities such as swimming and running, increasing the
function of the heart, lungs, and muscle mitochondria to
meet the heightened oxygen demands, ultimately resulting
in cardiorespiratory fitness improvements [9]. Over the past
decade, interest has also emerged in conducting studies
that assess the potential effect of resistance training (RT)
interventions in older individuals with T2DM [10–12].
Resistance training activates the muscular system to generate
force against a resistive load [4]; it can be performed by
utilizing various exercise machines, lifting free-weights (e.g.,
dumbbells), or doing calisthenics such as situps, pushups,
crunches, and lunges. If RT is performed regularly, where
the weight lifted is increased to moderate (50% of 1RM

2 Journal of Aging Research
(1RM represents 1 Repetition Maximum, which refers to the
maximum weight that a person can lift once)) and high levels
of intensity (>75% 1RM), it often leads to increased muscle
mass and improvements in muscular fitness [4, 13–15].
Muscular fitness refers both to muscle strength, the amount
of force produced by a muscle, and muscle endurance,
the ability of a muscle to “exert submaximal force for an
extended period of time” [16, page 27].
Resistance training may be more appealing and feasible
than AT for people with T2DM who are often overweight
and sedentary [17], as well as for older adults, obese, and/or
frail individuals [4, 12, 18]. With advanced age, there is a
significant loss of muscle mass and strength, a phenomenon
known as sarcopenia [19]. It has recently been indicated
that older adults with T2DM tend to have greater muscle
mass loss, worse muscle quality (defined as the amount
of muscle strength per unit of regional muscle mass),
reduced upper and lower body strength, greater visceral
adipose content, as well as higher risk for functional decline
and disability than their healthy, age-matched counterparts
[20–24]. Resistance training might benefit older adults
living with T2DM through muscle hypertrophy, enhanced
muscle quality, strength gains for greater power development
with more effective mobility function, and glycemic profile
improvements [25].
Resistance training studies in populations with T2DM
were not readily available prior to 1997 [4]. The first
physical activity guidelines specifically designed for adults
with T2DM were developed by the American College of
Sports Medicine (ACSM) in the year 2000 [10]. As illustrated
in Figure 1, a modified timeline first introduced by Hills
and colleagues in 2010 [26], agencies such as the Canadian
Diabetes Association (CDA), the American Diabetes Associ-
ation (ADA), the Canadian Society for Exercise Physiology
(CSEP), and ACSM now include RT recommendations
within their physical activity guidelines [11, 27–37].
Due to the associated increases in blood pressure (BP)
that may be harmful, there could be unsubstantiated appre-
hension in recommending RT, especially at higher intensities.
The main concern is that these BP increases could lead to
a stroke, myocardial ischemia, or retinal hemorrhage [4].
This may partially explain the historical dominance of AT
interventions in populations living with T2DM. However,
there is a lack of scientific evidence that RT actually increases
any of the aforementioned risks, as no RT-related adverse
events have been reported in studies where individuals with
T2DM were assessed [4, 38]. Additionally, past researchers
have suggested that RT may actually reduce BP levels [39–
41]. Finally, there are precautions that can be employed
to avoid potentially harmful side-effects of exercise, such
as avoiding physical activity under certain circumstances
(detailed by Gordon in 2002 [7]) and conducting appropriate
preexercise screens and assessments [7, 35, 42].
Skeletal muscles are the largest postprandial glucose
uptake and glycogen storage sites in the human body and
as such are integral in maintaining glucose homeostasis.
Resistance training may reverse or at least limit some of the
aforementioned negative neuromuscular effects associated
with aging and/or T2DM [43]. Previous meta-analyses have
reported benefits of aerobic training, resistance training,
or a combination of the two on reducing HbA1c levels,
which signifies improved glycemic control [25, 38, 44–47].
A recent meta-analysis demonstrated that supervised aerobic
or resistance training led to greater declines in HbA1c
levels than exercise advice only [44]. However, no previous
meta-analysis has assessed the effects of RT in older adults
(≥65 years) with T2DM. At this time, the literature base
may benefit from such a review, since older adults often
experience detrimental neuromuscular and sensorimotor
changes associated with aging (e.g., sarcopenia) placing them
at an increased risk for mobility problems, injury from
falls, and disability [21, 48]. Furthermore, T2DM is most
common in older adults, who as a result of this disease often
experience various comorbidities [49], further reducing their
capacity to live independently (e.g., retinopathy, which may
lead to blindness; peripheral neuropathy, which may lead to
foot ulcers and amputations; nephropathy, which over time
could result in renal failure, etc.). Thus, the purpose of this
paper is to conduct a systematic review of the best available
evidence, in order to assess the effect of RT on metabolic,
neuromuscular, and cardiovascular functions in older adults
with T2DM.
2. Methods
This meta-analysis utilized the PRISMA as a framework
when selecting studies for inclusion in this paper [50]. This
meta-analysis is not registered with any institution, such
as the Cochrane Collaboration. The literature search was
conducted until the end of August 2011, using electronic
databases (Medline, EMBASE, AMED, PubMed, Scopus,
CINAHL) that generated MESH terms based on the follow-
ing keywords: resistance training, type 2 diabetes, and aged.
The search terms were entered into the databases using the
appropriate combinations of “OR” and “AND.” In order for
articles to be included in this paper, the following inclusion
and exclusion criteria needed to be satisfied.
Inclusion Criteria
(i) RCTs.
(ii) Published between the years 2000 and 2011.
(iii) RT interventions or a combination of RT and other
forms of intervention (e.g., flexibility, weight loss,
standard care, etc.).
(iv) Participants with established T2DM.
(v) Participants’ mean age ≥65 years.
Exclusion Criteria
(i) Participants with the presence of another chronic
illness (e.g., cancer).
(ii) Non-English publications.
(iii) Studies reporting effect of RT in previously trained
participants.

Journal of Aging Research 3
Table 1: Outcome measures.
Body composition measures Musculoskeletal measures Type 2 diabetes process measures
Whole body lean tissue mass (kg)
Whole body fat mass (kg)
Muscle strength
(i) Upper body strength
(ii) Lower body strength
Muscle quality (defined as 1RM strength
kg/unit lean body mass kg)
Muscle fiber size
(i) Type I cross sectional area (CSA) (μm 2 )
(ii) Type II CSA (μm 2 )
Fasting glucose (mmol/L)
Glycosylated hemoglobin (HbA1c) (%)
Blood pressure
Serum/fasting insulin (pmol/L)
Lipids
(i) Total cholesterol (mmol/L)
(ii) HDL cholesterol (mmol)
(iii) Triglycerides
(iv) Free fatty acids (FFAs) (μmol/L)for adults >50–60 2-3 d/wACSM-CDC: 30 min/dmoderate-intensity PA, 5 d/w;RT details not specified [31]Health Canada: light (60 min/d)to moderate (30 min/d) AT;204 d/w; all major musclemin; RT ligth, weights, highreps, in nealry all T2DMpatients, high RT for youngonly [11]CDA: AT 3 d/w, moderate-intense, 150 min/w (more ifable/willing): RT 3 d/w, ∼8ex.w/t major muscle groups, startADA : AT 150 min/w moderate;all major muscle groups [36]CDA: AT 150 min/w, moderate(>70% HRmax); RT 3 d/w,moderate weight, progress toto 3 sets, 8 reps at heavierweight [28]ACSM/ADA position stand: AT150 h/w, moderate to vigorous,3 d/w no more than 2consecutive daysRT 2-3 nonconsecutive d/w,moderate (50%1RM) oroptimal insulin action, 5-10exercises involving majormuscle groups (UE, LE, andCSEP: for adults >65 years ofage AT moderate-vigorous,150 min/w; RT using majormuscle groups, 2 d/w [30]1995 1998 2000 2002 2003 2006 2007 2008 2010 2011groups; 2–4 sets, 10–15 reps [32]w/t 1 set, 10–15 reps, progressto 3 sets, 8–12 reps [35]RT 3 d/w, 3 sets, 8–10 reps forstart 1 set 10–15 reps at2 sets of 10–15 reps, progressvigorous (75–80%1RM) forcore), 10–15 reps per set,progress to 8–10 reps withheavier weights [28]ACSM/AHA: for adults <50−60RT 1–3 d/w, 1 set, 8–12 reps,1 set, 10–15 reps [37]ADA position stand: AT 3–5 d/w, 55–79% HRmax, or 40–74% HRmax reserve, 20–60ACSM position stand: 10−303d/w; 8–10 ex. all major musclegroups, 1set, 10–15 reps [10]min/d 40-70 % Vo2max AT,ACSM position stand: RT 1 set;8–12 reps for 8−10 ex; 10–15reps for older adults [33](50–70% HRmax) to vigorous
Figure 1: Chronological Timeline of PA Recommendations for T2DM from Various Professional Organizations [modified from [26]].
PHAC [Public Health Agency of Canada]; CSEP [Canadian Society for Exercise Physiology]; CDA [Canadian Diabetes Association]; ACSM
[American College of Sports Medicine]; ADA [American Diabetes Association]; CDC [Centers for Disease Control and Prevention]; AHA
[American Heart Association]. PA [Physical Activity]; RT [resistance training]; AT [aerobic training]; UE [upper extremity]; LE [lower
extremity]; HR max [maximum heart rate]; VO2 max [maximal oxygen uptake/consumption]; d [days]; w [week]; w/t [with]; reps [repetitions];
ex [exercises]; h [hour]; min. [minute].
(iv) Studies reporting effect of RT on outcome measures
not relevant to this paper (see Table 1 for all relevant
outcome measures).
The aforementioned inclusion and exclusion criteria
were developed in order to obtain the most recent (2000–
2011), scientifically rigorous (RCTs) evidence on the specific
effect of resistance training in older adults with type 2
diabetes. Various studies, review articles, and commentaries
that did not satisfy the inclusion criteria were used to inform
the introduction and the discussion sections of this paper.
Furthermore, NH and AS independently reviewed and rated
the articles and any differences were resolved by discussion
or by comparison to the ratings provided on the PEDro
website. To limit redundancy, Cohen’s Kappa values were not
calculated since there were no major disagreements between
the authors (i.e., >95% agreement).
Outcome Measures. The primary outcome measures were
grouped into three major areas including body composition,
musculoskeletal, and type 2 diabetes disease process mea-
sures. Table 1 summarizes the major outcome headings and
their respective measures.
Methodological Quality of the Studies. Internal validity of
studies included in this paper was assessed using the PEDro

4 Journal of Aging Research
Table 2: Participant characteristics.
Source Group (n) Age (years) Gender
(M/F)
Whole body
fat mass (kg)
BMI
(kg/m 2 )
Diabetes
duration (years)
HbA1c
(%)
Fasting
glucose
(mmol/L)
Fasting
insulin
(pmol/L)
∗Brooks et al.
[17]
Castaneda et al.
[13]
Exercise 31
Control 31
66 ± 11.1
66 ± 5.6
10/21
19/12
35 ± 5.6
33.7 ± 13.4
30.9 ± 6.1
31.2 ± 5.6
8 ± 5.6
11 ± 5.6
8.7 ± 5.6
8.4 ± 1.7
8.79 ± 2.7
9.85 ± 3.8
116 ± 167.4
115 ± 176.9
Dunstan et al.
[53]
Exercise 16
Control 13
67.6 ± 5.2
66.5 ± 5.3
10/6
6/7
33.1 ± 7.4
35.6 ± 6.8
31.5 ± 3.7
32.5 ± 3.8
7.6 ± 5.4
8.8 ± 7.9
8.1 ± 1
7.5 ± 1.1
9.5 ± 2.3
9.4 ± 2.1
132.9 ± 63
101.9 ± 25.8
All measures are provided as means ± SD.
∗Brooks et al. [17] and Castaneda et al. [13] included the same cohort of participants.
scale—a valid [51] and reliable [52] tool to evaluate study
quality. Article ratings are included as PEDro scores listed in
Table 3, while rating criteria are detailed in Table 5.
Statistical Analyses. Statistical software (Comprehensive
Meta-Analysis—version 2) for meta-analysis of binary, con-
tinuous, and diagnostic data was used for computation
of Hedge’s g (a measure of effect size). Hedge’s g values
were used to assess the influence of strengthening exercises
on body composition, musculoskeletal measures, and type
2 diabetes disease outcomes (previously summarized in
Table 1). The effect sizes were interpreted as small, medium
and large if they were 0.2, 0.5, and 0.8, respectively [54].
A 95% confidence interval was constructed around the
point estimate of the effect size. Any standard errors that
were reported by study authors were converted to standard
deviations using the formula SD = √n ∗ SE, where SD is the
standard deviation, √ is the square root symbol, n refers to
the sample size, ∗ represents the multiplication function, and
SE is the standard error [55].
The statistical significance of the differences in the effects
of RT on body composition, muscle quality, and strength
along with moderator variables included for the effect on
disease processes was computed by Page’s L statistic with
the use of PASW 18 statistical software to calculate the
sum of squares (SS) between groups, as well as total SS.
Page’s L statistic was then calculated using the formula
L = [N − 1]r2 , where N is the total number of effect
sizes and r2 is the product of SS between/SS total . (Further
details regarding Page’s L statistic can be found in [56])
When performing meta-analysis, the overall effect of an
intervention can be influenced by use of particular outcome
measures or intervention strategies. Page’s L statistics was
utilized to elucidate such differences in the current study.
The presence of heterogeneity among the moderator
variables was evaluated by the Q statistic using a random
effects model. The studies were considered heterogeneous
if the P value of the Q statistic was <0.1, which has been
proposed as the appropriate alternative to the conventional
P < 0.05, when there is a low number of articles included
in a review [57]. Publication bias was not assessed, since
there were only three articles included, and any conclusions
that are drawn from the results that emerge from this meta-
analysis cannot be taken as definitive. The robustness of the
findings was established based on the assessment of the effect
size and its associated confidence intervals, rather than other
methods, such as the calculation of Fail Safe N, which can
lead to widely varied estimates [58]. The results reported
were calculated using the random effects model, in order to
account for methodological differences amongst studies. The
statistical significance for the effect sizes’ statistical tests (i.e.,
Hedge’s g) was set at P < 0.05.
3. Results
Three [13, 17, 53] of the 446 citations were included in
the final analysis (Figure 2). However, 2 of the citations
[13, 17] are technically considered one study, since their
findings are based on the same pool of participants, but they
are both included in the meta-analysis since each of them
provides relevant but different outcome measures. A total
of 32 effect sizes, evaluating the effect of strength training
on the disease process (20 effect sizes) and muscle quality
(12 effect sizes), were extracted from the included studies.
Participant and study characteristics are described in Tables
2 and 3 respectively.
3.1. Effect of RT on T2DM Disease Process Measures. Serum
insulin [17, 53], HbA1c [17, 53], HDL [13, 53], LDL and
total cholesterol [13, 53], fasting glucose [17, 53], and BP
[13, 53] were analysed to evaluate the effect of RT on the
disease process. The overall cumulative point estimate of this
effect size was statistically significant (Hedge’s g = −0.246;
P = 0.023; 95% CI: −0.458, −0.034).
For individual variables, the effect of RT on BP (Hedge’s
g = −0.540; P < 0.001; CI: −0.832, −0.248), insulin (Hedge’s
g = 0.505; P = 0.016; CI: 0.094, 0.916), total cholesterol,
and LDL cholesterol (Hedge’s g = 22120.464, P = 0.002; CI:
−0.760, −0.169) was statistically significant. However, the
effect of RT on fasting glucose (Hedge’s g = −0.121; P =
0.559; CI: −0.526, 0.284), HbA1c (Hedge’s g = −0.463; P =
0.145; CI: −1.084, 0.159), and HDL cholesterol (Hedge’s g =
0.134; P = 0.517; CI: −0.271, 0.539) was not as consistent
between studies in terms of magnitude of improvement and
fluctuations in control group. Also, the differences in effects
of RT on fasting glucose, insulin, HBA1c, cholesterol, HDL,
FFA, and BP were not statistically significant (L(19) = 14.109;
P > 0.05).