Content uploaded by Paula A B Ribeiro
Author content
All content in this area was uploaded by Paula A B Ribeiro on May 10, 2018
Content may be subject to copyright.
ORIGINAL ARTICLE
Effects of weekly supervised exercise or physical activity
counseling on fasting blood glucose in women diagnosed
with gestational diabetes mellitus: A systematic review
and meta-analysis of randomized trials
Roberta BGEGINSKI,
1,2
*Paula A.B. RIBEIRO,
3
Michelle F. MOTTOLA
4,5
and
José Geraldo L. RAMOS
1,2
1
Obstetrics and Gynecology Department, Hospital de Clínicas de Porto Alegre,
2
Graduate Program in Health Sciences, Gynecology and
Obstetrics,
3
Postgraduate Studies Program in Cardiology, School of Medicine, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil,
4
Faculty of Health Sciences, Department of Anatomy and Cell Biology, Schulich School of Medicine and Dentistry, R. Samuel McLaughlin
Foundation Exercise and Pregnancy Laboratory, School of Kinesiology, The University of Western Ontario, and
5
Children’s Health Research
Institute, London, Canada
Correspondence
Roberta Bgeginski, Graduate Program in
Health Sciences, Gynecology and
Obstetrics, Ramiro Barcelos 2400, 90035-
903, Porto Alegre, RS, Brazil.
Tel: +55 51 99192 2961
Fax: +55 51 3343 2546
Email: robertabg@gmail.com
Received 27 March 2016; revised 11
December 2016; accepted 25
December 2016.
doi: 10.1111/1753-0407.12519
Abstract
Background: Exercise is an important part of gestational diabetes mellitus
(GDM) lifestyle management. However, no meta-analysis has analyzed the
effects of exercise programs on fasting blood glucose (FBG) in women with
GDM. A systematic review with meta-analysis was performed to evaluate the
effects of weekly supervised exercise (EXE) or physical activity counseling
(PA) in women with GDM compared with usual prenatal care (UPN) on gly-
cemic control.
Methods: Eligible trials were identified from Medline, EMBASE, Web of Sci-
ence, Scopus and SportDiscus up to December 2016. Data were retrieved from
randomized controlled trials comparing UPN with UPN plus weekly super-
vised (at least once a week) prenatal exercise or PA counseling for which FBG
values before and after intervention were available. Random-effects meta-
analysis was performed for mean difference in FBG after exercise intervention.
Results: The search yielded 781 publications, of which 82 were assessed for
eligibility and eight were included in the meta-analysis. The overall effect on
absolute FBG concentrations was not significant (P= 0.11) compared with
UPN. However, PA versus UPN showed a significant reduction in absolute
FBG concentrations (weighted mean difference −3.88 mg/dL; 95% confidence
interval −7.33, −0.42 mg/dL; I
2
48%; P
heterogeneity
< 0.15).
Conclusions: Physical activity counseling in women with GDM showed a sig-
nificant effect compared with UPN on FBG concentrations, possibly due to a
longer follow-up time compared with the EXE groups. This result highlights
Highlights
•There was no overall effect of exercise when all interventions were tested.
•The sensitivity analysis showed that physical activity counseling, which was the intervention with a longer
follow-up, had significant effects on fasted blood glucose when compared to weekly-supervised exercise in
gestational diabetes mellitus (GDM).
•Physical activity counseling added to standard care may help motivate women with GDM to be active for
longer periods and acquire benefits.
Journal of Diabetes •• (2017), ••–••
© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 1
the importance of an early intervention that lasts to delivery for best practice
of GDM management.
Keywords: exercise, glycemic control, hyperglycemia, insulin resistance,
pregnancy.
Introduction
Gestational diabetes mellitus (GDM) has been defined
as hyperglycemia detected during routine testing in
pregnancy (generally between 24 and 28 weeks) that
does not meet the criteria of diabetes mellitus in preg-
nancy.
1
The diagnosis of GDM is made using a single-
step 75-g oral glucose tolerance test when one or more
of the following results are recorded during routine test-
ing between 24 and 28 weeks of pregnancy: (i) fasting
plasma glucose ≥92 mg/dL; (ii) 1-h post-load glucose
≥180 mg/dL; and (iii) 2-h post-load glucose ≥153 mg/
dL.
2
Worldwide, approximately 7% of all pregnancies
are affected by GDM (ranging from 1% to 14% depend-
ing on the population studied and the diagnostic tests
used), with more than 200 000 cases diagnosed annu-
ally.
1
Elevated glucose concentrations are associated
with adverse pregnancy outcomes such as fetal macro-
somia, shoulder dystocia, neonatal hypoglycemia and
future type 2 diabetes (T2D).
1
Exercise is an important part of GDM lifestyle mana-
gement because physical activity (PA) enhances maternal
insulin sensitivity, helps maintain daily fasting and post-
prandial normoglycemia and decreases the need for
insulin.
3–6
Exercise and an active lifestyle may also
decrease the risk of developing T2D for women with
GDM in the future.
7,8
However, structured exercise pro-
grams are not available for all women with GDM, and
most women will receive only counseling to increase PA
from their primary care providers or healthcare team.
A Cochrane review examined the effects of exercise
programs compared with no specific program or with
other therapies in women with GDM and concluded that
no significant difference was found between exercise and
the other regimens in several outcomes evaluated, such as
gestational age at delivery, preterm delivery (<37 weeks),
birthweight at delivery, cesarean section and macrosomia
(>4000 g).
9
However, fasting blood glucose (FBG) was
not evaluated and the authors did not analyze the effects
of different supervised exercise modalities or PA counsel-
ing in maternal and perinatal outcomes.
9
To our knowl-
edge, there is no systematic review that summarizes the
effects of exercise programs on FBG, specifically in
women with GDM.
The aim of the present study was to conduct a sys-
tematic review with a meta-analysis of randomized
controlled trials (RCTs) to evaluate the effect of weekly
supervised exercise (EXE) or PA counseling in women
diagnosed with GDM compared with usual prenatal
care (UPN) on glycemic control.
Methods
The present systematic review and meta-analysis were
performed according to the Preferred Reporting Items
for Systematic Reviews and Meta-Analysis (PRISMA)
statement.
10
Search strategy and study selection
Eligible trials were identified by a structured electronic
search in Medline, EMBASE, Web of Science, Scopus
and SportDiscus up to December 2016. Electronic data-
bases were searched using similar search strategies
focusing on four main groups of terms: “exercise,”
“pregnancy,”“gestational diabetes mellitus”and “ran-
domized controlled trial.”These terms were adjusted to
fit the requirements specified in each database. The
complete search strategy used for the PubMed database
is shown in Table S1 available as File S1. The lists of
references from each study were also searched to iden-
tify additional studies. Relevant review articles were
evaluated for information on additional trials.
Eligibility criteria
Studies including pregnant women diagnosed with
GDM in which treatment allocation was randomized,
with a control group receiving standard prenatal care
for GDM, were considered for further eligibility assess-
ment. The RCTs were required to have at least one
intervention arm of either EXE, defined as exercise ses-
sions performed with study personnel at least once a
week throughout the program, or PA counseling in
which women with GDM received counseling for PA or
performed exercise without any direct supervision. For
inclusion, studies were required to provide before- and
after-intervention absolute values of FBG or differences
between means and dispersion values. Excluded studies
were: (i) interventions consisting solely of pelvic floor
exercises, stretching or relaxation; (ii) studies regarding
pregnant women diagnosed with type 1 diabetes or
T2D before pregnancy; (iii) studies that analyzed
2© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Exercise and gestational diabetes R. BGEGINSKI et al.
exercise interventions in risk factors or prevention of
GDM; (iv) interventions that lasted less than 4 weeks;
(v) studies that did not provide data on blood glucose;
and/or (vi) studies not published in English, Spanish,
Italian or Portuguese. No restrictions regarding the year
of publication were applied.
Study selection and data extraction
Two authors (RB and PABR) independently reviewed
the titles and/or abstracts to exclude any clearly irrelevant
studies. Reviewers were not blinded to journal of publica-
tion, author names, or their institutions. All abstracts that
provided insufficient information on the inclusion and
exclusion criteria were selected to evaluate the full text.
Corresponding authors of potentially eligible studies were
contacted if the required data could not be located in the
published report. The full texts of the remaining studies
were then retrieved and read by the same authors to deter-
mine whether the studies met the inclusion criteria. Dis-
agreements were solved by a third author (JGLR), who
independently examined the studies.
Using a standardized data extraction sheet, the follow-
ing information (if available) was extracted in duplicate
and recorded from studies: authors, year of publication,
country of origin, total sample size, maternal age, criteria
for GDM diagnosis, FBG concentrations, trial duration
(in weeks), insulin requirements, prepregnancy body mass
index (BMI), weight gain during pregnancy, gestational
age at birth, birth weight and length, and APGAR scores
at 1 and 5 min. Information on adherence or compliance
to protocols and drop-out rates was also extracted.
Assessment of risk of bias and quality of studies
Methodological quality was explored using an approach
similar to that recommended by The Cochrane Collabo-
ration in assessing risk.
11
The criteria for methodological
quality of studies included the following dimensions: ade-
quate sequence generation, allocation concealment,
blinding of participants or personnel, blinding of out-
come, description of losses and exclusions, intention-to-
treat analysis, incomplete outcome data, and selective
reporting. Studies without a clear description of these
features were considered as unclear. The complete assess-
ment of risk of bias is shown in Table S2 in File S1.
Data synthesis and analysis
A meta-analysis was performed using REVIEW MANAGER
5.3 software (The Nordic Cochrane Centre, The
Cochrane Collaboration, Copenhagen, Denmark).
11
Baseline FBG values did not differ between groups for
the RCTs included. Based on this information, the
post-intervention values of FBG were considered the
primary outcome of the present study. For continuous
outcomes, mean differences between exercise and con-
trol groups were examined. Gestational age, birthweight
at delivery, prepregnancy BMI, weight gain during
pregnancy, macrosomia, preterm delivery, and cesarean
section were considered secondary outcomes. Random-
effects models were used because clinical heterogeneity
between trials was expected (considering the different
intervention regimens). Results of the meta-analysis are
expressed as weighted mean differences (WMD) for
continuous outcomes and as odds ratios (OR) for
dichotomous outcomes, both with 95% confidence inter-
vals (CI), and with I
2
values as markers of intertrial het-
erogeneity. The assumption of homogeneity of true
effect sizes was assessed by the Cochran’sQ-test and
the degree of inconsistency across studies was calculated
I
2
.
12
In the case of I
2
> 40%, heterogeneity was
explored with subgroup and sensitivity analysis and the
overall result was presented using the random-effect
model. The meta-analysis comprised the comparisons
of overall effect (EXE + PA vs UPN). Sensitivity analy-
sis considering the type of intervention (EXE vs PA)
and additional nutritional counseling (combined or not
with the intervention) was performed. The presence of
publishing bias for the hypothesis of an association
between supervised exercise training during pregnancy
and blood glucose control was assessed informally by
visual inspection of funnel plots (see Fig. S1). An α
value ≤0.05 was considered statistically significant.
Results
Aflow diagram of the search results is shown in Fig. 1.
The initial literature search resulted in 781 potentially
relevant citations. After exclusion of duplicates, title
and abstract review, and full text review, eight studies
were included in the meta-analysis (Table 1); five studies
were RCTs of EXE groups.
3,5,6,13,14
and three were
RCTs of PA counseling.
4,15,16
The study characteristics for the EXE and PA counsel-
ing groups are given in Tables 1 and 2. The eight RCTs
included in the analysis totaled 469 pregnant women with
GDM, of whom 91 engaged in EXE and 143 were given
PA counseling. Four trials had an exclusively aerobic
supervised intervention
3,6,13,14
and one study examined
the effects of a supervised circuit-type resistance exercise
alone.
5
For the PA counseling analysis, only one trial had
an exclusively aerobic exercise counseling intervention.
15
The UPN groups received standard prenatal care for
women with GDM, and all groups received nutritional
counseling added to the interventions.
© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 3
R. BGEGINSKI et al. Exercise and gestational diabetes
For the EXE trials, the frequency of exercise sessions
ranged from two to three times per week and the time
of each supervised session ranged from 20 to 45 min.
The duration of the interventions ranged from 4 to
7 weeks, ending in the late third trimester or at delivery.
The mode of supervised exercise included walking, sta-
tionary cycling, arm cycling, and circuit-type resistance
training. Exercise intensity was assessed by heart rate
monitors
3,6,13,14
and the Borg’s Scale of Perceived
Exertion.
5,6,13
For the PA counseling trials, the frequency of PA
ranged from two to seven times per week and the time
of each PA session ranged from 15 to 20 min. The
duration of the interventions ranged from 8 to
14 weeks, ending in the late third trimester or at deliv-
ery. The mode of PA included walking, yoga and
circuit-type resistance training and the exercise intensity
was assessed by perceived exertion.
Quality assessment (risk of bias)
Among the EXE studies, four of five studies (80%) pre-
sented adequate sequence generation,
3,5,6,13
two (40%)
reported allocation concealment,
5,6
one (20%) had
blinding of participants or personnel,
5
none had blinded
assessment of outcomes, all (100%) described losses to
follow-up and exclusions, one (20%) used the intention-
to-treat principle for statistical analyses,
5
all (100%)
described incomplete outcome data, and four (80%)
were free of selective reporting.
3,5,6,13
Among the PA
counseling studies, all three (100%) presented adequate
sequence generation, all reported allocation conceal-
ment, one (33.3%) had blinding of participants or
personnel,
15
one (33.3%) had blinded assessment of
outcomes,
15
all three described losses to follow-up and
exclusions, one (33.3%) used the intention-to-treat prin-
ciple for statistical analyses,
15
all described incomplete
outcome data, and two (66.7%) were free of selective
reporting.
15,16
The nature of the interventions did not
allow blinding of the subjects and personnel to the type
of intervention.
Effect of interventions on FBG
The overall effect of exercise (supervised or PA counsel-
ing) on FBG concentrations was not significantly
Records identified through
database searching
(n = 781)
Additional records identified
through other sources
(n = 0)
Records after duplicates removed
(n = 521)
Records screened
(n = 521)
Records excluded
(n = 439)
Full-text articles excluded, with reasons (n = 74)
Reason for exclusion:
• Not GDM: n = 36
• Language other than English, Spanish,
Italian or Portuguese: n = 4
• Not the eligible intervention: n = 28
• No blood glucose data: n = 5
• Not RCT: n = 1
Studies included in
qualitative synthesis
(n = 8)
Studies included in PA
counselling meta-analysis
(n = 3)
Studies included in supervised
exercise meta-analysis
(n = 5)
Full-text articles assessed
for eligibility
(n = 82)
Included Eligibility Screening Identification
Figure 1 Flow diagram of studies included in the present analysis. GDM, gestational diabetes mellitus; RCT, randomized controlled trial;
PA, physical activity.
4© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Exercise and gestational diabetes R. BGEGINSKI et al.
Table 1 Characteristics of the weekly supervised exercise and physical activity counseling studies included in the present meta-analysis
Study and
country
Sample
size (n)
Age
(years)
GA at
entry
(weeks)
Exercise intervention
Diet
advice Caloric intake
Frequency
(times/week) Intensity
Time
(min) Type
Follow-up
duration
(weeks) Adherence
No.
drop-
outs
Control
group
Weekly supervised exercise
Avery et al.
13
(US)
I: 14 I: 32.2 4.9 I: 28.7 3.0 2 (with
supervision)
70% HR
max
and RPE
30 Cycle
ergometer
or walking
NR NR I: 1 Dietary therapy
and usual
physical
activity
Yes I: 2301 550
kcal/day
C: 14 C: 30.4 5.1 C: 26.3 8.1 1–2 (without
supervision)
C: 3 C: 2190 472
kcal/day
Bung et al.
14
(US)
I:21 I:31.0 4.5 I: 30.3 1.9 3 50% of the
last VO
2max
45 Recumbent
bicycle
Minimum
of 4
>90% I: 4 Insulin therapy
and diet
protocol
NR 30 kcal/kg diet
C:20 C:32.0 5.7 C: 30.3 2.0 C: 3
de Barros et al.
5
(Brazil)
I:32 I: 32.4 5.4 I:31.0 2.3 2 (with
supervision)
RPE 5–6
(“somewhat
hard”)
30–40 Circuit type
resistance
training
7 NR I: 1 Prenatal
routine
care
Yes 35 kcal/kg ideal
weight/day
and 300 kcal/
day were added
in the 2nd and
3rd trimesters
C: 32 C:31.8 4.8 C: 31.5 2.2 1 (without
supervision)
C: 1
Halse et al.
6
(Australia)
I: 20 I: 34 5 I: 28.8 0.8 3 (with
supervision)
Interval:
55–85%
age-
predicted
HR
max
.
and
RPE 9–16
25 to 45 Cycle
ergometer
61 96% 0 Usual
physical
activity
regimen
Yes/ I: Week 1
7252 2060 kJ,
final week
6897 1932 kJ
C: 20 C: 32 3 C: 28.8 1.0 2 (without
supervision)
C: Week 1
7414 2412 kJ,
final week
7005 2067 kJ
Jovanovic-
Peterson et al.
3
(US)
I: 10 I:29.5 2.5 I: 28 3 70% reserve
HR
20 Arm
ergometer
6 NR 0 Diet Yes 24–30 kcal/kg/24
h dietC: 9 C:31.1 2.8 C: 28
Physical activity counseling
Bo et al.
15
(Italy)
I: 51 I: 35.9 4.8 Range
24–26
7 RPE 12–14 At least
20
Walking 12–14 68.80% I: 0 Diet Yes I: 2116.2 267.7
kcal/dayC: 0C: 50 C: 33.9 5.3
C: 2115.9 383.0
kcal/day
Brankston et al.
4
(Canada)
I: 16 I: 30.5 4.4 I: 29.0 2.0 2.0 0.9 “Somewhat
hard”
NR Circuit type
resistance
training
26–32
weeks
pregnancy
to
delivery
NR 6 Standard
diabetic
diet
Yes 24 to 30 kcal/kg
per day diet.C: 16 C: 31.3 5.0 C: 29.6 2.1
Youngwanichsetha
et al.
16
(Thailand)
I: 85 I: 32.5 5.0 Range
24–30
5NR15–20 Yoga 8 >80% I: 5 Standard
diabetes
care
Yes NR
C: 85 C: 31.2 4.5 C: 5
Where appropriate, data are given as the mean s.d.
I, intervention group; C, control group; GA, gestational age; BMI, body mass index; RPE, ratings of perceived exertion; NR, not reported; HR, heart rate; Max., maximum; VO
2
, oxygen consumption.
© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 5
R. BGEGINSKI et al. Exercise and gestational diabetes
different (P= 0.11) to that of UPN (WMD = −2.76 mg/
dL, 95% CI −6.13, 0.61; I
2
87%; P
heterogeneity
<0.00001;
Fig. 2). However, when the different exercise interven-
tions were tested, PA counseling showed a significant
decrease in FBG concentrations of −3.88 mg/dL com-
pared with control (WMD = −3.88, 95% CI −7.33,
−0.42; I
2
48%; P
heterogeneity
< 0.15; Fig. 2). The effects of
EXE interventions compared with UPN were not
Table 2 Newborn characteristics of the weekly supervised exercise and physical activity counseling studies included in the present meta-
analysis
Study and country Gestational age at birth (weeks) Birth weight (g) Birth length (cm)
APGAR score
1 min 5 min
Weekly supervised exercise
Avery et al.
13
(US) I: 39.4 1.2 I: 3419 528 NR I: 8 I: 9
C: 39.7 0.9 C: 3609 428 C: 9 C:9
Bung et al.
14
(US) I: 38.9 1.7 I: 3369 534 I: 49.3 2.0 I: >8 I: >8
C: 38.3 2.0 C: 3482 502 C: 51.0 3.0 C: >8 C: >8
de Barros et al.
5
(Brazil) I: 38.5 1.2 I: 3230 450 NR NR NR
C: 38.6 1.1 C: 3330 490
Halse et al.
6
(Australia) I: 38.6 1.7 I: 3176 526 I: 49.8 2.5 I: 9 1I:90
C: 38.7 1.2 C: 3319 478 C: 49.9 2.9 C: 9 1C:90
Jovanovic-Peterson et al.
3
(US) I: range 39.5–40.5 I: 3634 317 NR NR NR
C: range 39.4–40.0 C: 3465 343
Physical activity counseling
Bo et al.
15
(Italy) NR NR NR NR NR
Brankston et al.
4
(Canada) NR NR NR NR NR
Youngwanichsetha et al.
16
(Thailand) NR NR NR NR NR
Where possible, data are given as the mean s.d.
I, intervention group; C, control group; NR, not reported.
Supervised exercise training
Subtotal (95% CI)
Avery et al.13 9.8%
13.7%
13.3%
16.5%
11.3%
64.5%
1.00 [-5.91, 7.91]
5.04 [1.10, 8.98]
2.59 [-1.65, 6.83]
-1.80 [-2.92, -0.68]
-17.50 [-23.26, -11.74]
-1.92 [-7.50, 3.65]
-1.60 [-5.54, 2.34]
-12.61 [-23.29, -1.93]
-3.96 [-6.31, -1.61]
-3.88 [-7.33, -0.42]
-2.76 [-6.13, 0.61]
13.7%
6.2%
15.5%
35.5%
100.0%
0
Exercise Control
25-25
Bung et al.14
de Barros et al.5
Halse et al.6
Jovanovic-Peterson et al.3
Heterogeneity: Tau2 = 34.86; Chi2 = 44.58, d.f. = 4 (P < 0.00001); I2 = 91%
Test for overall effect: Z = 0.68 (P = 0.50)
Heterogeneity: Tau2 = 4.34; Chi2 = 3.81, d.f. = 2 (P = 0.15); I2 = 48%
Test for overall effect: Z = 2.20 (P = 0.03)
Heterogeneity: Tau2 = 17.56; Chi2 = 51.93, d.f. = 7 (P < 0.00001); I2 = 87%
Test for subgroup differences: Tau2 = 0.34, d.f. = 1 (P = 0.56), I2 = 0%
Test for overall effect: Z = 1.61 (P = 0.11)
Bo et al.15
BRANKSTON et al.4
Youngwanichsetha et al.16
Physical activity counseling
Subtotal (95% CI)
Subtotal (95% CI)
Study or subgroup
Mean difference
IV, Random, 95% CI Weight
Mean difference
IV, Random, 95% CI
Figure 2 Mean weighted difference and 95% confidence intervals (CI) for fasting blood glucose (in mg/dL) for weekly supervised exercise
versus prenatal standard care, and physical activity counseling versus prenatal standard care in women diagnosed with gestational diabetes
mellitus. Squares represent study-specific estimates; diamonds represent pooled estimates of random-effects meta-analyses.
6© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Exercise and gestational diabetes R. BGEGINSKI et al.
significantly different (WMD = −1.92 mg/dL, 95% CI
−7.50, 3.65; I
2
91%; P
heterogeneity
< 0.00001; Fig. 2).
Sensitivity analyses and exploration of heterogeneity
Sensitivity analyses for nutritional counseling, type of
intervention, and training volume were performed to
explore the heterogeneity in the studies. Studies that
had nutritional advice integrated with the exercise inter-
vention were analyzed (seven interventions,
444 patients).
3–6,14–16
The overall effect of exercise
(supervised or PA counseling) integrated with the exer-
cise intervention on FBG concentrations was not signif-
icantly different (P= 0.08) compared with control
(Fig. S2a). However, when the different exercise inter-
ventions were tested, PA counseling showed a signifi-
cant decrease in FBG concentrations of 3.88 mg/dL
compared with control (Fig. S2a). The effect of an EXE
intervention on FBG concentrations compared with
UPN was not significantly different (Fig. S2a).
Absolute changes of FBG with type of intervention
(exercise modality) were examined. Aerobic exercise
only was chosen for the majority of studies and the
overall effect on FBG concentrations was not signifi-
cantly different (five interventions, 219 patients;
3,6,13–15
P= 0.30; Fig. S2b) compared with control.
Sensitivity analyses of data according to the weekly
amount of exercise (sessions duration ×frequency =
training volume) were performed according to Ameri-
can Congress of Obstetrics and Gynecology (ACOG)
guidelines that recommend moderate-intensity exercise
for at least 20–30 min/day on most days of the week for
pregnant women.
17
This recommendation leads to a
range of 100–150 min/week exercise (20–30 min/day ×
5 days/week). However, participants reached the upper
limit of 150 min/week exercise in only one study.
6
Thus,
analyses were performed using the lower limit of
100 min/week exercise, with the results indicating that
the overall effect of this weekly amount of exercise (five
interventions, 182 patients),
3,5,6,13,14
supervised or PA
counseling, was not significantly different (P= 0.46)
compared with UPN (Fig. S2c).
Effects of interventions on secondary outcomes
Secondary outcomes were analyzed only for the weekly
supervised exercise trials because the PA counseling trials
did not report these data. Overall, EXE had no effect on
gestational age of delivery (WMD = 0.01 weeks, 95% CI
−0.33, 0.36; I
2
0.0%; P
heterogeneity
=0.75)andbirthweight
at delivery (WMD = −69.41 g, 95% CI −202.05, 63.22;
I
2
0.0%; P
heterogeneity
= 0.52). In addition, EXE had no
significant effect on the odds of having a macrosomic
newborn (birthweight >4000 g; OR 0.79, 95% CI 0.24,
2.62; I
2
20.0%; P
heterogeneity
= 0.57), a preterm delivery
(<37 weeks gestation; OR 1.16, 95% CI 0.39, 3.41;
I
2
20.0%; P
heterogeneity
= 0.93) or a cesarean delivery
(OR 0.75, 95% CI 0.37, 1.55; I
2
20.0%; P
heterogeneity
= 0.95). Only two weekly supervised exercise trials
6,14
reported data on prepregnancy BMI and weight gain dur-
ing pregnancy; analysis of these secondary outcomes was
not performed.
Discussion
The present systematic review provides the summarized
effects from eight RCTs involving 469 pregnant
participants.
3–6,13–16
The overall effect of EXE and PA
counseling on FBG concentrations was not significantly
different compared with UPN. Because UPN for
women with GDM includes some type of PA recom-
mendation, these results are not surprising. However,
when the different exercise interventions were tested,
PA counseling showed a significant decrease in absolute
FBG concentrations of 3.88 mg/dL compared with con-
trol. Although a reduction in FBG concentrations may
be an important indicator of the efficacy of the inter-
vention program for women with GDM, especially if
these values are brought below the suggested target for
FBG concentrations, the exact target remains a globally
controversial issue. We were not expecting PA counsel-
ing to be more successful in reducing FBG than control
over a program of direct PA intervention, which was
contrary to our hypothesis. We expected that direct
interventions would show better improvements than
counseling, as demonstrated previously for HbA1c in
T2D subjects.
18
The results of the present study are
probably due to the timing of the PA counseling inter-
ventions, which were longer than direct exercise inter-
ventions (PA counseling interventions 8–14 weeks vs
exercise interventions 4–7 weeks). Furthermore, PA
counseling started in early pregnancy and was not
delayed until a diagnosis of GDM, when the metabolic
imbalance was already established. Moreover, to our
knowledge, the present study represents the first system-
atic review and meta-analysis of RCTs that stratifies the
studies by the type of PA intervention, which can help
health practitioners improve adherence in this
population.
It is important to point out methodological differ-
ences between the present study and the Cochrane
review.
9
Fasting blood glucose was not considered as
an outcome in the latter, but was analyzed as a primary
outcome in the present study. We analyzed the effects
of different exercise interventions on maternal and peri-
natal outcomes in eight trials.
3–6,13–16
Most of the meta-
© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 7
R. BGEGINSKI et al. Exercise and gestational diabetes
analysis comparisons in the previous review
9
were per-
formed by including one study only, decreasing the
power of the results of that review.
Persistent hyperglycemia during pregnancy is associ-
ated with increased risk of fetal malformations, macro-
somia, and neonatal hypoglycemia at delivery.
2,19
The
literature reports that exercise affects body composition,
carbohydrate and lipid metabolism, and stimulates glu-
cose uptake, lowering blood glucose concentrations.
20,21
Exercise increases the rate of glucose uptake into the
skeletal muscle, a process that is regulated by the trans-
location of the glucose transport protein GLUT-4.
22–24
Thus, it would be logical to introduce exercise to neu-
tralize the negative effects of GDM, thereby improving
health during pregnancy.
In order to understand the possible factors underlying
the lack of an overall effect of exercise on FBG, we used
sensitivity analysis stratified by nutritional advice inte-
grated with exercise, type, and volume of the exercise
intervention. Concerning nutritional advice as a co-inter-
vention, a significant reduction in FBG of 3.55 mg/dL
was observed when nutritional counseling was integrated
with PA counseling. A partial explanation for these results
is that the number of participants and the duration of
follow-up was greater for the PA counseling trials. In
addition, the heterogeneity of the PA counseling trials was
smaller than the EXE trials (48% vs 93%, respectively).
In almost all studies, patients were under dietary
advice as part of routine prenatal care. The aim of die-
tary advice for women with GDM is to optimize glyce-
mic control, thus preventing maternal hyperglycemia
and reducing postprandial glucose concentrations. The
diet is based on ingestion of low-glycemic index foods,
which induce a gradual increase in blood glucose due to
their slow digestion and absorption. The studies
included are considerably homogeneous regarding
caloric intake, with a range of 24–35 kcal/kg per day,
as recommended for successful pregnancy outcomes.
1
We did not observe a clear trend for the weekly vol-
ume of exercise or the type of exercise that affected
FBG. The ACOG guidelines recommend moderate-
intensity exercise for at least 20–30 min/day on most of
days of the week for pregnant women.
17
Only few preg-
nant women meet the upper range of 150 min/week
exercise.
25
Borodulin et al.
26
reported that the preva-
lence of sufficient activity varied between 3% and 38%,
depending on the type of activity and measurement of
intensity included. The present review showed that only
one study
6
met the current recommendation of
150 min/week exercise. It is clear that further research is
needed to tailor exercise programs that reach the target
amount of exercise recommended, and then verify the
effect of the exercise program on GDM and perinatal
outcomes. There is evidence that to achieve ≥150 min/
week exercise is advantageous for blood glucose con-
trol, as shown in previous studies with T2D in non-
pregnant individuals.
18
The analyses of secondary outcomes were performed
with EXE trials only because data were not available
for the PA counseling studies. The EXE interventions
did not affect gestational age, or the odds of having a
preterm delivery or a cesarean delivery compared with
controls. To date, no studies have shown an increased
risk of preterm labor or an increased incidence of pre-
mature rupture of membranes among exercising preg-
nant women that were not at risk for these conditions.
The PA counseling trials evaluated did not report data
regarding gestational age and birthweight at delivery,
incidence of macrosomia, prepregnancy BMI, and ges-
tational weight gain. It is strongly encouraged that
future RCTs report these important outcomes to under-
stand the extension of the effect of non-supervised inter-
ventions in terms of public health.
The findings of the present review suggest that addi-
tional studies are required before reaching specific con-
clusions regarding the effect of supervised exercise or
PA counseling on blood glucose control in women with
GDM. Considering the small number of total partici-
pants (<100 for the exercise intervention trials and
143 for the PA counseling) and that half the studies did
not show significant effects, we do believe that larger
RCTs with longer intervention times would reduce the
heterogeneity of the studies and help establish consen-
sus regarding exercise intervention in women with
GDM. We understand that direct supervision will not
be available to all women with GDM and, in this situa-
tion, PA counseling should be encouraged or, when
possible, both interventions should be combined
throughout pregnancy.
The present study has strong methodological points,
such as the careful selection of the studies, data extrac-
tion, and analysis of the methodological quality of the
articles. Thus, trials were only reviewed if the partici-
pants were diagnosed with GDM, the analyses were stra-
tified by type of intervention (EXE vs PA counseling),
and the included articles exhibited geographical diversity
leading to high generalizability, with trials in North
America, South America, Europe, Asia, and Oceania.
The limitations of the present study result mainly
from the quality of the included studies (Table S2 in
File S1) and the substantial evidence of heterogeneity
and potential publication bias. The random-effects
model was chosen to account for heterogeneity of the
studies, which was considered high for most of the ana-
lyses. Overall, the general quality of the studies was
low, reflecting an increased risk of bias in many of the
8© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Exercise and gestational diabetes R. BGEGINSKI et al.
studies included. The oldest study included was per-
formed 26 years ago, and the criteria to diagnose GDM
have changed substantially over recent years. Despite
these limitations, the present systematic review with
meta-analysis provides a general overview of the
research literature addressing the role of exercise, with
or without supervision, in blood glucose control in
women with GDM.
In conclusion, the overall effects of exercise, super-
vised or through PA counseling, on FBG concentra-
tions in women diagnosed with GDM were not
significantly different compared with control interven-
tions. When the different exercise interventions were
tested, PA counseling showed a significant decrease in
absolute FBG concentrations. There is insufficient evi-
dence to recommend, or advice against, women with
GDM enrolling in exercise programs. Physical activity
counseling sessions in addition to standard care may
help motivate women with GDM to be more active,
whereas structured exercise may be more difficult to
achieve. For future studies, we suggest that larger RCTs
be designed to compare early intervention of supervised
exercise with PA counseling (or at least be of similar
length), with longer follow-up time to delivery for best
practice of GDM management in order to directly eval-
uate the different effects of these types of interventions
on glycemic control in women with GDM.
Acknowledgements
The authors thank CAPES (BEX 14179/13–0), The
University of Western Ontario, CNPq, and FIPE/
HCPA for financial support. The authors also thank
Rhiannon E. HALSE, (School of Sport Science, Exer-
cise and Health, The University of Western Australia,
Perth, WA, Australia) and Simona B, (Department of
Medical Sciences, University of Turin, Turin, Italy) for
providing further data to complete the meta-analysis,
and Bruna BARROSO (School of Physical Education,
Physiotherapy and Dance, Universidade Federal do
Rio Grande do Sul, Porto Alegre, Brazil) for revision of
the extracted data.
Disclosure
The authors have no conflicts of interest to declare.
References
1. Hod M, Kapur A, Sacks DA et al. The International
Federation of Gynecology and Obstetrics (FIGO) initia-
tive on gestational diabetes mellitus: A pragmatic guide
for diagnosis, management, and care. Int J Gynecol
Obstet. 2015; 131 (Suppl. 3): S173–211.
2. Metzger BE, Gabbe SG, Persson B et al. International
Association of Diabetes and Pregnancy Study Groups
recommendations on the diagnosis and classification of
hyperglycemia in pregnancy. Diabetes Care. 2010; 33:
676–82.
3. Jovanovic-Peterson L, Durak EP, Peterson CM. Rando-
mized trial of diet versus diet plus cardiovascular condi-
tioning on glucose levels in gestational diabetes.
Am J Obstet Gynecol. 1989; 161: 415–9.
4. Brankston GN, Mitchell BF, Ryan EA, Okun NB.
Resistance exercise decreases the need for insulin in over-
weight women with gestational diabetes mellitus.
Am J Obstet Gynecol. 2004; 190: 188–93.
5. de Barros MC, Lopes MAB, Francisco RPV,
Sapienza AD, Zugaib M. Resistance exercise and glyce-
mic control in women with gestational diabetes mellitus.
Am J Obstet Gynecol. 2010; 203: 551–6.
6. Halse RE, Wallman KE, Newnham JP, GuelfiKJ.
Home-based exercise training improves capillary glucose
profile in women with gestational diabetes. Med Sci
Sports Exerc. 2014; 46: 1702–9.
7. Manson JE, Rimm EB, Stampfer MJ et al. Physical
activity and incidence of non-insulin-dependent diabetes
mellitus in women. Lancet. 1991; 338: 774–8.
8. Tuomilehto J, Lindstrom J, Eriksson JG
et al. Prevention of type 2 diabetes mellitus by changes
in lifestyle among subjects with impaired glucose toler-
ance. N Engl J Med. 2001; 344: 1343–50.
9. Ceysens G, Rouiller D, Boulvain M. Exercise for diabe-
tic pregnant women. Cochrane Database Syst Rev. 2006;
3: CD004225.
10. Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA
Group. Preferred reporting items for systematic reviews
and metaanalyses: The PRISMA statement. PLoS Med.
2009; 6: e100009.
11. Higgins J, Green S. Cochrane Handbook for Systematic
Reviews of Interventions. John Wiley & Sons, Chichester,
2011.
12. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Mea-
suring inconsistency in meta-analyses. BMJ. 2003; 327:
557–60.
13. Avery MD, Leon AS, Kopher RA. Effects of a partially
home-based exercise program for women with gesta-
tional diabetes. Obstet Gynecol. 1997; 89:10–5.
14. Bung P, Artal R, Khodiguian N et al. Exercise in gesta-
tional diabetes. No optional therapeutic approach? Dia-
betes. 1991; 40: 182–5.
15. Bo S, Rosato R, Ciccone G et al. Simple lifestyle recom-
mendations and the outcomes of gestational diabetes. A
2×2 factorial randomized trial. Diabetes Obes Metab.
2014; 16: 1032–5.
16. Youngwanichsetha S, Phumdoung S, Ingkathawornwong
T. The effects of mindfulness eating and yoga exercise on
blood sugar levels of pregnant women with gestational
diabetes mellitus. Appl Nurs Res. 2014; 27: 227–30.
17. ACOG Committee opinion no. 650: Physical activity
and exercise during pregnancy and the postpartum
period. Obstet Gynecol. 2015; 126: e135–42.
© 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 9
R. BGEGINSKI et al. Exercise and gestational diabetes
18. Umpierre D, Ribeiro PA, Schaan BD, Ribeiro JP. Vol-
ume of supervised exercise training impacts glycaemic
control in patients with type 2 diabetes: A systematic
review with meta-regression analysis. Diabetologia. 2013;
56: 242–51.
19. Thompson D, Berger H, Feig D, Canadian Diabetes
Association Clinical Practice Guidelines Expert Commit-
tee. Diabetes and pregnancy. Can J Diabetes. 2013; 37
(Suppl. 1): S168–83.
20. Artal R. Exercise: The alternative therapeutic interven-
tion for gestational diabetes. Clin Obstet Gynecol. 2003;
46: 479–87.
21. Ruchat SM, Davenport MH, Giroux I et al. Effect of exer-
cise intensity and duration on capillary glucose responses
in pregnant women at low and right risk for gestational
diabetes. Diabetes Metab Res Rev. 2012; 28:669–78.
22. Goodyear LJ, Kahn BB. Exercise, glucose transport, and
insulin sensitivity. Annu Rev Med. 1998; 49: 235–61.
23. Hayashi T, Wojtaszewski JF, Goodyear LJ. Exercise
regulation of glucose transport in skeletal muscle.
Am J Physiol. 1997; 273: E1039–51.
24. Ryder JW, Chibalin AV, Zierath JR. Intracellular
mechanisms underlying increases in glucose uptake in
response to insulin or exercise in skeletal muscle. Acta
Physiol Scand. 2001; 171: 249–57.
25. Evenson KR, Wen F. Prevalence and correlates of objec-
tively measured physical activity and sedentary behavior
among US pregnant women. Prev Med. 2011; 53:39–43.
26. Borodulin KM, Evenson KR, Wen F, Herring AH,
Benson AM. Physical activity patterns during pregnancy.
Med Sci Sports Exerc. 2008; 40: 1901–8.
Supporting information
Additional Supporting Information may be found in the
online version of this article:
Table S1. Literature search strategy used for the
PubMed database.
Table S2. Risk of bias of studies included in the meta-
analysis: weekly-supervised exercise and physical activ-
ity counseling groups.
Figure S1. Funnel plot of each trial’s observations for
the effects of supervised exercise training and physical
activity counseling on changes in fasting blood glucose.
Figure S2. Weighted mean difference in fasting blood
glucose and 95% confidence intervals for (a) studies
with dietary counseling integrated with the exercise
intervention, (b) aerobic exercise only performed during
pregnancy, and (c) weekly amount of exercise.
10 © 2016 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Exercise and gestational diabetes R. BGEGINSKI et al.
