jurnal referat 2
Click here to load reader
-
Upload
nunki-aprillita -
Category
Documents
-
view
214 -
download
0
description
Transcript of jurnal referat 2
-
Physical Activity Before and DuringPregnancy and Risk of GestationalDiabetes MellitusA meta-analysis
DEIRDRE K. TOBIAS, SM1
CUILIN ZHANG, MD, PHD2
ROB M. VAN DAM, MD, PHD1,3
KATHERINE BOWERS, PHD2
FRANK B. HU, MD, PHD1,4
OBJECTIVE Gestational diabetes mellitus (GDM) is one of the most common complica-tions of pregnancy and is associated with a substantially elevated risk of adverse health outcomesfor both mothers and offspring. Physical activity may contribute to the prevention of GDM andthus is crucial for dissecting the vicious circle involving GDM, childhood obesity, and adulthoodobesity, and diabetes. Therefore, we aimed to systematically review and synthesize the currentevidence on the relation between physical activity and the development of GDM.
RESEARCH DESIGN AND METHODS Medline, EMBASE, and Cochrane Reviewswere searched from inception to 31 March 2010. Studies assessing the relationship betweenphysical activity and subsequent development of GDM were included. Characteristics includingstudy design, country, GDM diagnostic criteria, ascertainment of physical activity, timing ofexposure (prepregnancy or early pregnancy), adjusted relative risks, CIs, and statistical methodswere extracted independently by two reviewers.
RESULTS Our search identified seven prepregnancy and five early pregnancy studies,including five prospective cohorts, two retrospective case-control studies, and two cross-sectional study designs. Prepregnancy physical activity was assessed in 34,929 total participants,which included 2,813 cases of GDM, giving a pooled odds ratio (OR) of 0.45 (95% CI 0.280.75) when the highest versus lowest categories were compared. Exercise in early pregnancy wasassessed in 4,401 total participants, which included 361 cases of GDM, and was also significantlyprotective (0.76 [95% CI 0.700.83]).
CONCLUSIONS Higher levels of physical activity before pregnancy or in early pregnancyare associated with a significantly lower risk of developing GDM.
Diabetes Care 34:223229, 2011
G estational diabetes mellitus (GDM)is one of the most common compli-cations of pregnancy, affecting7% of all pregnancies in the U.S. (i.e.,200,000 cases annually) (1), and thisnumber is increasing as the prevalence ofobesity among women at reproductive
age escalates (24). GDM is associatedwith a significantly elevated risk for short-term and long-term complications forboth mothers and offspring. Women withGDM have an increased risk for perinatalmorbidity and impaired glucose toleranceand type 2 diabetes in the years after preg-
nancy (5,6). Children of women withGDM are more likely to be obese and haveimpaired glucose tolerance and diabetesin childhood and early adulthood (1). In arecent meta-analysis of randomized trialson the effect of treatment for GDM, vari-ous interventions for blood glucose con-trol, including diet, glucose monitoring,insulin use, and pharmaceutical interven-tions, did not significantly reduce the riskfor adverse perinatal and neonatal endpoints, including cesarean section andperinatal or neonatal death (7). Collec-tively, these data indicate that preventionof GDM altogether could be crucial foravoiding its associated adverse healthoutcomes.
Physical activity has long beenknown for its role in improving glucosehomeostasis through its direct or indi-rect impact on insulin sensitivity viaseveral mechanisms. For instance,physical activity has independent ef-fects on glucose disposal by increasingboth insul in-mediated and noninsulin-mediated glucose disposal(8,9). Physical activity can also exertlong-term effects on improvement in in-sulin sensitivity through increased fat-free mass (10). Furthermore, thebenefits of preventing or delaying theonset of type 2 diabetes among non-pregnant individuals have been re-ported repeatedly (11,12). Therefore,physical activity may have the potentialfor preventing GDM and related adversehealth outcomes. However, evidencefor its impact on GDM has not been sys-tematically synthesized. The aim of thissystematic review and meta-analysiswas to assemble the current evidencefor the relationship between physicalactivity and the development of GDM.
RESEARCH DESIGN ANDMETHODS
Data collectionRelevant published English-language ar-ticles were identified by searching theMedline database (National Library of
From the 1Departments of Nutrition and Epidemiology, Harvard School of Public Health, Boston, Massa-chusetts; the 2Epidemiology Branch, Division of Epidemiology, Statistics, and Prevention Research, EuniceKennedy Shriver National Institute of Child Health and Development, Bethesda, Maryland; the 3Depart-ment of Epidemiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore; the4Department of Public Health and Medicine, Yong Loo Lin School of Medicine, National University ofSingapore, Singapore; and the 5Channing Laboratory, Department of Medicine, Brigham and WomensHospital and Harvard Medical School, Boston, Massachusetts.
Corresponding author: Deirdre K. Tobias, [email protected] 20 July 2010 and accepted 16 September 2010. Published ahead of print at http://care.
diabetesjournals.org on 27 September 2010. DOI: 10.2337/dc10-1368. 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly
cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be herebymarked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
R e v i e w s / C o m m e n t a r i e s / A D A S t a t e m e n t sM E T A - A N A L Y S I S
care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 223
-
Medicine, Bethesda, MD), EMBASE, andCochrane Reviews (The Cochrane Collab-oration) through March 2010 and by amanual bibliography check. The Medlinesearch was as follows, with similar termsfor other databases: (diabetes, gestational[MeSH]) AND (lifestyle OR risk factorOR physical activity [MeSH] OR exer-cise), where MeSH stands for MedicalSubject Headings. Bibliographies of ac-cepted studies as well as recent reviewswere screened to ensure a complete studylisting.
For relevant abstracts, full publica-tions were retrieved for evaluation on thebasis of criteria that were established apriori. All original research articles wereconsidered except case reports. Wesought to include studies that assessed therelationship between physical activityand the risk of developing GDM. Studiesreporting only impaired glucose toleranceor an impaired glucose tolerance andGDM combined end point were not in-cluded. Our criteria did not restrict on themeasurement of physical activity (fre-quency, intensity, type, and others) or theexposure period (prepregnancy or earlypregnancy). One exception to this was theexclusion of studies evaluating physicalactivity during the total pregnancy pe-riod, such as retrospective questionnairesthat did not specifically probe a pre-GDMgestational age time period. The rationalebehind this exclusion is that women withthe diagnosis of GDM may undergo ther-apy for glucose control that includesphysical activity recommendations; thus,there is a potential for reverse causation(13,14). Our criteria did not restrict toparticular populations or countries. Arti-cles were independently screened formeeting the eligibility criteria by two re-viewers (D.K.T. and K.B.).
For each accepted article, study char-acteristics, including authors, publicationyear, study design, country, and GDMscreening and diagnostic criteria, were ex-tracted independently by two researchers(D.K.T. and K.B.). Details of the exposureincluded the specific time period underinvestigation and the method of ascertain-ment. The unadjusted and adjusted rela-tive risks and 95% CIs were extracted asreported by authors. Statistical methodswere noted, including which covariableswere considered and adjusted for. Au-thors were contacted for clarification ofany of the above extracted data points ifneeded.
Statistical analysisA random-effects meta-analysis was con-ducted to combine the relative risks re-ported for the original studies (15).Separate analyses were done for theprepregnancy and early pregnancy timeperiods. We chose to use a random-effectsmeta-analysis, which takes into accountbetween-study heterogeneity, becausethe study design and exposure dose andintensity were not uniform across studies;therefore, similar effect sizes were not as-sumed. To facilitate a comparable expo-sure across studies, we analyzed therelative risks for the highest physical ac-tivity category versus the lowest (refer-ence) category. When studies used thehighest amount of activity as their refer-ence group, we exponentiated the nega-tive of the log odds ratio (OR) and 95% CIto convert the direction of the effectestimate.
The Cochrane Q test was used to eval-uate the presence of heterogeneity, with anull hypothesis that the treatment effect isequal across all studies (16). We consid-ered heterogeneity to be significant at P0.1, a conservative standard for meta-analyses (17). In addition, we calculatedthe I2 statistic and 95% CIs to evaluate thepercentage of heterogeneity that was dueto between-study variation (18). In thepresence of heterogeneity, sensitivityanalyses were performed to evaluate effectmodification by study-level characteris-tics including study design, GDM diag-nostic criteria, physical activity measures(METs vs. frequency only; number ofquantiles), and country of study (19).These were done by performing a ran-dom-effects meta-regression for eachstudy-level variable. Stratified pooled ef-fect estimates were calculated and re-ported if there was evidence of effectmeasure modification by a given charac-teristic. The influence of outliers was alsoassessed to evaluate the impact of theirremoval and the robustness of themeta-analysis.
Publication bias was assessed throughEgger and Begg tests, using a significancelevel of P 0.05 to indicate significantasymmetry (20,21). We also performed avisual inspection of the funnel plot forpublication bias, looking for a skewed(nonsymmetric) distribution of standarderrors around the study-level effectestimates.
Analyses were conducted in Stata(version 10.0; StataCorp, College Station,TX). We used the METAN command tocalculate the pooled effect estimates and
the tes t s for heterogene i ty . TheMETAREG and HETEROGI commandswere used to conduct analyses forheterogeneity.
RESULTS Our literature searchproduced 442 citations, of which we se-lected 18 for further review of the full text(Fig. 1). Ten studies were excluded forreasons listed in Fig. 1. Therefore, eightpublications met our criteria for inclusionin this meta-analysis and review (2229)(Table 1). Findings for the OMEGA Studyprospective cohort and Alpha Study case-control population were presented inthree different publications (2628). Weassessed which outcomes were reportedmore than once to avoid inclusion of du-plicate effect estimates in our meta-analyses. Two publications by Dempseyet al. (27,28) reported results for theOMEGA Study and Alpha Study popula-tions, including both prepregnancy andearly pregnancy exercise exposures. In amore recent, single publication, Rudra etal. (26) updated the results for both studypopulations, but for the prepregnancy ex-posure only. Therefore, the publicationsby Dempsey et al. (27,28) were includedin our meta-analysis for their early preg-nancy results only, and the relative risksfrom Rudra et al. (26) were included forits prepregnancy results.
Ultimately, the eight studies in ouranalysis (prepregnancy k 7; early preg-nancy k 5) represented a total of 34,929subjects (prepregnancyN 34,929; earlypregnancy N 4,401), with 2,855 totalcases of GDM (prepregnancy n 2,813;early pregnancy n 361) (2229). Theseincluded five prospective cohort studies(22,2426,28), two retrospective case-control studies (26,27), and two cross-sectional surveys (23,29). (The totalnumber of study designs is greater thanthe total number of publications becauseRudra et al. [26] presented results for twodistinct studies in the same study.) Allstudies were conducted among U.S.women except one, which was conductedby Harizopoulou et al. (23) among Greekparticipants. In the prospective cohortstudies (22,2426,28), physical activityinterviews or questionnaires for bothprepregnancy and early pregnancy habitswere administered before participants re-ceived their diagnosis of GDM. For theretrospective case-control and cross-sectional studies (23,26,27,29), partici-pants were asked about their physicalactivity during their postpartum hospitalstay, with the exception of the study by
Physical activity and GDM: a meta-analysis
224 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org
-
Redden et al. (29), which collected expo-sure data 27 months postpartum. Theprepregnancy time period was defined insix studies as 1 year before the index preg-nancy (2328), in one study as 3 monthsbefore the index pregnancy (29), and inone study as the average exposure overseveral years of follow-up before the in-dex pregnancy (22). All but one study(29) reported use of a validated physicalactivity questionnaire to assess exposure,although only one of these questionnaireswas specifically validated in pregnantwomen, with satisfactory results (24).GDM was physician-diagnosed in all butone study, which used validated self-report of having received a physicians di-agnosis (22). Other relevant studycharacteristics are tabulated in Table 1.
Units of physical activity varied andincluded frequency (hours per week), en-ergy expenditure (MET-hours per week),and level of exertion or intensity. Physicalactivity types included total physical ac-tivity as well as specific activities (walk-ing, climbing stairs, and others). In themeta-analyses of total physical activity,five of the eight studies analyzed physicalactivity in units of energy expenditure(22,23,2628), which incorporates bothfrequency and intensity, whereas three ofthe eight studies analyzed physical activ-ity in units of frequency only (24,25,29).All but the two cross-sectional studies re-ported relative risks across quantiles ofexposure (23,29).
Total physical activityPrepregnancy. Seven studies reportedthe association between total prepreg-nancy physical activity and GDM (2226,29). A meta-analysis of relative risksindicated a 55% lower risk of GDM forwomen in the highest physical activityquantiles compared with those in the low-est (pooled OR 0.45 [95% CI 0.280.75];P 0.002) (Fig. 2). The Cochrane Q sta-tistic indicated significant heterogeneityin study results (Q 32.6, P 0.001),with an I2 value estimating that 82% (6391%) of the variance is due to between-study differences.
We conducted additional sensitivityanalyses to evaluate potential sources of het-erogeneity in the results. Meta-regressiondid not show a significant difference in ef-fect estimates among studies with a pro-spective versus retrospective study design(meta-regressionP0.54) (supplementaryFig. 1, available in an online appendix athttp://care.diabetesjournals.org/cgi/content/full/dc10-1368/DC1). Likewise,meta-regression results indicated a lack ofeffect measure modification by GDM di-agnosis criteria (P 0.58), physical activ-ity analysis by energy expenditure versusfrequency (P 0.40), study size being100 cases (P 0.15), and country ofstudy (P 0.078). When we ran themeta-regression on the total number ofexposure categories, there was a border-line significant association (P 0.053);however, the number of studies in each
strata was few. When we stratified bywhether studies adjusted for specific con-founders, we did not find a statisticallysignificant difference among effect esti-mates that controlled for family history ofdiabetes (P 0.97), smoking status (P0.23), race or ethnicity (P 0.33), parity(P 0.67), or socioeconomic status co-variables (P 0.47). Finally, to evaluatethe robustness of the main pooled effectestimate, we removed the largest study byZhang et al. (22), which accounted for62% of the total study participants. Thisdid not substantially alter the pooled ORor significance level (pooled OR 0.39[0.200.73]; P 0.004).Early pregnancy. Five studies reportedeffect estimates for the association betweenearly pregnancy physical activity and devel-opment of GDM (2325,27,28). Results foractivity during this time period indicateda significant 24% lower risk of GDM as-sociated with the highest activity groupcompared with the lowest activity group,as shown in Fig. 2 (OR 0.76 [0.700.83];P 0.0001). The Q test was not signifi-cant for heterogeneity but was possiblyunderpowered because of few studies(Q 1.83; P 0.77). Despite a pointestimate of 0%, the I2 statistic suggestedthat heterogeneity was possible, given thewide CI (95% CI 079%). In a sensitivityanalysis we removed the study by Harizo-poulou et al. (23) because it contributedto 96% of the weight in the pooled OR.The pooled OR remained statistically sig-nificant with a similar magnitude of effect(0.65 [0.430.98]; P 0.04).
Finally, the Egger and Begg tests forthe primary analyses did not indicate thepresence of publication bias in the analy-sis of total physical activity (prepregnancyP 0.30; early pregnancy P 0.81). Vi-sual inspection of the funnel plot was inagreement with the statistical test, with noapparent asymmetry.
WalkingThe association between walking andGDM risk was evaluated in three studies(22,25,27). Two studies analyzed theassociation between walking durationand GDM risk (Oken et al.:2 h/day vs.2 h/day; Dempsey et al.:3 miles/dayvs. 1 mile/day) (25,27). Overall theredid not seem to be an association be-tween walking duration and GDM risk(prepregnancy: pooled OR 0.95 [95%CI 0.50 1.83]; early pregnancy: 0.77[0.511.16]). However, when the jointeffect of walking duration and usualwalking pace was analyzed, there was
Figure 1Study attrition diagram. PA, physical activity.
Tobias and Associates
care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 225
-
Table1
Studycharacteristics
Cha
san-
Tab
eret
al.,
2008
(24)
Dem
psey
etal
.,20
04(2
8)D
emps
eyet
al.,
2004
(27)
Har
izop
oulo
uet
al.,
2009
(23)
/O
ken
etal
.,20
06(2
5)R
edde
net
al.,
2010
(29)
Rud
raet
al.,
2006
(26)
Zha
nget
al.,
2006
(22)
Stud
yde
sign
Pros
pect
ive
coho
rtPr
ospe
ctiv
eco
hort
Ret
rosp
ecti
veca
se-
cont
rol
Cro
ss-s
ecti
onal
surv
eyPr
ospe
ctiv
eco
hort
Cro
ss-s
ecti
onal
surv
eyPr
ospe
ctiv
eco
hort
Ret
rosp
ecti
veca
se-
cont
rol
Pros
pect
ive
coho
rt
Tot
alsu
bjec
ts(N
)1,
006
909
Cas
esof
GD
M(n
)33
4215
540
9180
842
216
1,42
8C
ount
ryof
stud
yU
.S.
U.S
.U
.S.
Gre
ece
U.S
.U
.S.
U.S
.U
.S.
U.S
.St
udy
popu
lati
onLa
tina
Preg
nanc
ySt
udy
OM
EG
ASt
udy
Alp
haSt
udy
Hos
pita
l-ba
sed
part
icip
ants
Proj
ect
Viv
aPR
AM
SO
ME
GA
Stud
yA
lpha
Stud
yN
urse
sH
ealt
hSt
udy
IIPr
epre
gnan
cym
eta-
anal
ysis
Ear
lypr
egna
ncy
met
a-an
alys
is
Tot
alac
tivi
tym
easu
reE
nerg
yex
pend
itur
e(K
PAS
scor
e4
20)
Ene
rgy
expe
ndit
ure
(ME
T-h
/wee
k)E
nerg
yex
pend
itur
e(M
ET
-h/w
eek)
Ene
rgy
expe
ndit
ure
(ME
T-h
/wee
k)Fr
eque
ncy
(h/
wee
k)Fr
eque
ncy
(day
s/w
eek)
Ene
rgy
expe
ndit
ure
(ME
T-h
/wee
k)E
nerg
yex
pend
itur
e(M
ET
-h/w
eek)
Ene
rgy
expe
ndit
ure
(ME
T-h
/wee
k)E
xpos
ure
grou
psPr
epre
gnan
cy:7
.9,
9.5,
10.8
,12.
3;ea
rly
preg
nanc
y:6.
8,8.
1,9.
2,10
.7(q
uart
ilem
edia
ns)
Non
e(R
ef),
28,
28
Non
e(R
ef),
0.1
9.9,
101
9.9,
29.9
,30
010
(Ref
),
100
2(R
ef),
36,
713
,14
01
(Ref
),1
4,
5N
one
(Ref
),0.
114
.9,
15.0
Non
e(R
ef),
0.1
14.9
,15.
029
.9,
30
0.2
4.8
(Ref
),4.
911
.3,1
1.4
21.3
,21
.44
0.3,
40
.4
GD
Mdi
agno
stic
crit
eria
Not
repo
rted
ND
DG
ND
DG
ND
DG
AD
ASe
lf-re
port
ND
DG
Car
pent
eran
dC
oust
anV
alid
ated
self-
repo
rt(v
ario
us)
Exp
osur
eas
sess
men
tm
etho
d(s)
KPA
SSt
anfo
rdSe
ven-
Day
PAR
ecal
l;M
inne
sota
Leis
ure-
Tim
ePA
Q;B
org
rati
ngsc
ale
Stan
ford
Seve
n-D
ayPA
Rec
all;
Min
neso
taLe
isur
e-T
ime
PAQ
;Bor
gra
ting
scal
e
IPA
QPA
SESt
anda
rdiz
edqu
esti
onna
ire
Stan
ford
Seve
n-D
ayPA
Rec
all;
Min
neso
taLe
isur
e-T
ime
PAQ
;Bor
gra
ting
scal
e
Stan
ford
Seve
n-D
ayPA
Rec
all;
Min
neso
taLe
isur
e-T
ime
PAQ
;Bor
gra
ting
scal
e
Stan
dard
ized
ques
tion
nair
e
Mod
elco
vari
able
sA
ge,p
repr
egna
ncy
BMI
Age
,rac
e/et
hnic
ity,
prep
regn
ancy
BMI,
pari
ty
Age
,rac
e/et
hnic
ity,
prep
regn
ancy
BMI,
nulli
pari
ty,
smok
ing
duri
ngpr
egna
ncy
Age
,pre
preg
nanc
yBM
I,pr
egna
ncy
wei
ght
gain
,inc
ome,
educ
atio
n,oc
cupa
tion
,re
side
nce,
fam
ilyhi
stor
yof
diab
etes
,pr
ior
gluc
ose
into
lera
nce,
prev
ious
infa
ntw
ith
mac
roso
mia
,cur
rent
glyc
osur
ia
Age
, race
/eth
nici
ty,
prep
regn
ancy
BMI,
hist
ory
ofG
DM
,m
ater
nal
hist
ory
ofdi
abet
es
Age
, race
/eth
nici
ty,
prep
regn
ancy
BMI,
pari
ty,
no.,
pren
atal
care
visi
ts,
inco
me,
alco
hol
cons
umpt
ion
duri
ngpr
egna
ncy
Age
,rac
e/et
hnic
ity,
prep
regn
ancy
BMI,
nulli
pari
ty,
prep
regn
ancy
hype
rten
sion
Age
,rac
e/et
hnic
ity,
prep
regn
ancy
BMI,
nulli
pari
ty,
prep
regn
ancy
hype
rten
sion
Age
,rac
e/et
hnic
ity,
prep
regn
ancy
BMI,
pari
ty,
smok
ing
stat
us,
fam
ilyhi
stor
yof
diab
etes
,alc
ohol
inta
ke,t
otal
calo
ric
ener
gyin
take
,cer
eal
fiber
,gly
cem
iclo
ad,t
otal
gram
sof
fat
AD
A,A
mer
ican
Dia
bete
sA
ssoc
iati
on;I
PAQ
,Int
erna
tion
alPh
ysic
alA
ctiv
ity
Que
stio
nnai
re;K
PAS,
Kai
ser
Phys
ical
Act
ivit
ySu
rvey
;ND
DG
,Nat
iona
lDia
bete
sD
ata
Gro
up;P
A,P
hysi
calA
ctiv
ity;
PAQ
,Phy
sica
lAct
ivit
yQ
uest
ionn
aire
;PA
SE,P
regn
ancy
Ris
kA
sses
smen
tan
dM
onit
orin
gSy
stem
;PR
AM
S,Pr
egna
ncy
Ris
kA
sses
smen
tan
dM
onit
orin
gSy
stem
;Ref
,ref
eren
t.
Physical activity and GDM: a meta-analysis
226 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org
-
an inverse association in the prepreg-nancy time period. In the studies byDempsey et al. (27) and Zhang et al.(22), women who reported a brisk usualwalking pace and walked for a longerduration (Dempsey et al.:2 miles/day;Zhang et al.: 30 min/day) were asso-ciated with a lower risk of GDM, com-pared with women reporting a casualusual walking pace and shorter dura-tion (pooled OR 0.59 [95% CI 0.30 0.87]). This association was slightlyattenuated in early pregnancy, as re-ported by Dempsey et al. but did notreach statistical significance (OR 0.83[95% CI 0.48 1.45]). Although onlythree studies reported associations be-tween walking and GDM risk, findingswere consistent for an inverse associa-tion with intensity of walking pace, al-though it is unclear whether walkingduration (distance or time) has similarbenefits.
Stair climbingTwo studies assessed the association be-tween stair climbing and GDM risk as thenumber of flights of stairs climbed per dayduring the prepregnancy period (22,27).They each found a significant inverse as-sociation between GDM and the highestcategory of stair climbing (Dempsey et al.:10 flights/day; Zhang et al.:15 flights/day) compared with women who did notclimb stairs, after adjustment for severalpotential confounders, includingprepregnancy BMI (Dempsey et al.: OR0.47 [95% CI 0.260.93]; Zhang et al.:0.50 [0.270.90]; pooled OR 0.49 [95%CI 0.260.72]). Dempsey et al. (27) alsoassessed stair climbing in early pregnancyand found a similar inverse association(OR 0.26 [95% CI 0.130.52]).
Vigorous activityFour studies evaluated physical activity ofvigorous intensity (22,2527). Overall,
there was an inverse association betweenparticipation in vigorous activity com-pared with no vigorous activity inprepregnancy (pooled OR 0.47 [95% CI0.190.75]). Two studies also reportedan association of GDM and vigorous ac-tivity intensity in early pregnancy(25,27). The pooled effect estimate sug-gests an inverse association with vigorousphysical activity (0.55 [0.211.43]), al-though this did not reach statisticalsignificance.
Physical inactivityFew studies addressed the association ofsedentary or inactive lifestyle in prepreg-nancy or early pregnancy with the risk ofGDM. In the prospective cohort by Okenet al. (25), those who reported being sed-entary (2 h/week total physical activity)has a nonsignificantly higher risk of GDMfor both time periods (prepregnancy: OR1.4 [95% CI 0.73.0]; early pregnancy:
Figure 2Results of meta-analyses. A: Prepregnancy physical activity. B: Early pregnancy physical activity.
Tobias and Associates
care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 227
-
1.4 [0.8 2.6]). Hours spent watchingtelevision was not associated with GDMrisk in two prospective cohort studies(Oken et al.: relative risk 1.03 [95% CI0.6 1.8]; Zhang et al.: not reported)(22,25).
CONCLUSIONS The results fromour systematic review and meta-analysesindicate that greater total physical activitybefore pregnancy or during early preg-nancy was significantly associated with alower risk of GDM. The magnitude of thisassociation was greatest for prepregnancyphysical activity with women in the high-est quantiles of activity experiencing a55% reduction in risk, compared withthat for women with the lowest activity.Heterogeneity in study results were sub-stantial, suggesting that differencesamong study populations or methodol-ogy may have affected the results. Ouranalyses to detect sources of heterogene-ity were probably underpowered becausefew studies were in each stratum. How-ever, removal of individual influentialstudies did not dramatically alter ourfindings, supporting the robustness of thepooled estimate. Early pregnancy physi-cal activity was also associated with a sta-tistically significant 25% lower risk forwomen participating in high levels ofphysical activity.
The course of a normal pregnancy in-cludes increased metabolic stress and dis-turbances in l ip id and g lucosehomeostasis in the third trimester(30,31). There is marked insulin resis-tance in maternal muscle with the intentto increase glucose supply for the devel-oping fetus. The development of GDMmight reflect an impaired capacity to han-dle such metabolic challenges, such asunderlying -cell dysfunction (32).Therefore, women more equipped to han-dle metabolic stress might be more likelyto maintain normal glucose levels (33).The inverse association we observed be-tween physical activity and developmentof GDM is biologically plausible. Researchamong nonpregnant individuals hasshown that exercise-induced improve-ments in glycemic control may be due toincreases in GLUT4, a glucose transportprotein (34,35). Physical activity also hasdirect effects on oxidative stress and en-dothelial function (11,12). Researchershave demonstrated that physical activitymay also have an indirect and potentiallymore long-term role in glucose tolerancethrough changes in body composition(36,37). Decreases in fat mass and in-
creases in muscle mass have been shownto have positive effects on glycemic con-trol (37). In our literature search we didnot identify results of any randomizedclinical trials evaluating the effect of phys-ical activity on prevention of GDM risk.However, it is reasonable to infer thatphysical activity might prevent GDMthrough similar pathways.
Although the findings in this meta-analysis give support for physical activityin the prevention of GDM, there are somelimitations. Assessment of physical activ-ity was done via self-report in question-naires; thus, misclassification is plausible.For the prospective studies in our analy-sis, misclassification of prepregnancyphysical activity in the prospective stud-ies is likely to be random with respect toexposure, because women were unawareof their GDM diagnosis at the time of as-sessment; however, attenuation of the ef-f ec t es t imates may lead to anunderestimation of the true association.In addition, the inverse association be-tween physical activity and GDM risk didnot differ by study design (i.e., prospec-tive vs. retrospective) in our meta-regression, alleviating the concern forrecall bias among the retrospective stud-ies. Adjustment for major confounderswas consistent across studies, althoughunknown or residual confounding is pos-sible. The small number of publishedstudies makes it difficult to assess hetero-geneity in the pooled ORs. There is alsothe chance for publication bias, when re-searchers are less likely to publish null oruninteresting findings. The methods usedin this review did not suggest publicationbias. Finally, although we were able toanalyze the prepregnancy and early preg-nancy physical activity periods sepa-rately, our analysis is unable to determinethe independent biological relevance ofthe two exposure periods. Prepregnancyphysical activity is one of the strongestpredictors of physical activity in earlypregnancy; thus, it is difficult to knowwhich one, or whether both, could becontributing to the inverse associationsseen in our analyses, because of their highcorrelation (38). Much of the benefit thatwe observed for pregravid physical activ-ity could also reflect continued activityduring pregnancy and vice versa.
In summary, results from this system-atic review and meta-analyses demon-strate that greater total physical activitybefore or during early pregnancy is signif-icantly associated with lower risk ofGDM, with the magnitude of the associa-
tion being stronger for prepregnancyphysical activity. Given the consistent ev-idence across several studies, promotingphysical activity among women of repro-ductive age may represent a promisingapproach for the prevention of GDM andsubsequent complications of childrenborn from pregnancies affected by GDM.It is still unknown whether beginning anexercise routine in early pregnancyamong previously sedentary or minimallyactive women incurs GDM prevention,and further research is warranted to de-termine the joint and independent effectsof physical activity before and duringearly pregnancy.
Acknowledgments C.Z. and K.B. weresupported by the Intramural Research Pro-gram of the Eunice Kennedy Shriver NationalInstitute of Child Health and Human Develop-ment, National Institutes of Health.
No potential conflicts of interest relevant tothis article were reported.
D.K.T. performed a literature search, inde-pendent review of articles, and extraction ofdata, analyzed data, and wrote the manuscript.C.Z., R.M.v.D., and F.B.H. reviewed/editedthe manuscript. K.B. performed independentreview of articles and extraction of data.
References1. American Diabetes Association. Gesta-
tional diabetes mellitus. Diabetes Care2004;27(Suppl. 1):S88S90
2. Albrecht SS, Kuklina EV, Bansil P, Jamie-son DJ, Whiteman MK, Kourtis AP, Pos-ner SF, Callaghan WM. Diabetes trendsamong delivery hospitalizations in theU.S., 19942004. Diabetes Care 2010;33:768773
3. Dabelea D, Snell-Bergeon JK, HartsfieldCL, Bischoff KJ, Hamman RF, McDuffieRS. Increasing prevalence of gestationaldiabetes mellitus (GDM) over time and bybirth cohort: Kaiser Permanente of Colo-rado GDM Screening Program. DiabetesCare 2005;28:579584
4. Ferrara A, Kahn HS, Quesenberry CP, Ri-ley C, Hedderson MM. An increase in theincidence of gestational diabetes mellitus:Northern California, 19912000. ObstetGynecol 2004;103:526533
5. Langer O, Mazze R. The relationship be-tween large-for-gestational-age infantsand glycemic control in women with ges-tational diabetes. Am J Obstet Gynecol1988;159:14781483
6. McMahon MJ, Ananth CV, Liston RM.Gestational diabetes mellitus. Risk fac-tors, obstetric complications and infantoutcomes. J Reprod Med 1998;43:372378
7. Horvath K, Koch K, Jeitler K, Matyas E,
Physical activity and GDM: a meta-analysis
228 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org
-
Bender R, Bastian H, Lange S, SiebenhoferA. Effects of treatment in women with ges-tational diabetes mellitus: systematic re-view and meta-analysis. BMJ 2010;340:c1395
8. Annuzzi G, Riccardi G, Capaldo B, KaijserL. Increased insulin-stimulated glucoseuptake by exercised human muscles oneday after prolonged physical exercise. EurJ Clin Invest 1991;21:612
9. Devlin JT. Effects of exercise on insulinsensitivity in humans. Diabetes Care1992;15:16901693
10. Shulman GI, Rothman DL, Jue T, Stein P,DeFronzo RA, Shulman RG. Quantitationof muscle glycogen synthesis in normalsubjects and subjects with non-insulin-dependent diabetes by 13C nuclear mag-netic resonance spectroscopy. N EnglJ Med 1990;322:223228
11. Boule NG, Haddad E, Kenny GP, WellsGA, Sigal RJ. Effects of exercise on glyce-mic control and body mass in type 2 dia-betes mellitus: a meta-analysis ofcontrolled clinical trials. JAMA 2001;286:12181227
12. Jeon CY, Lokken RP, Hu FB, van DamRM. Physical activity of moderate inten-sity and risk of type 2 diabetes: a system-atic review. Diabetes Care 2007;30:744752
13. American Diabetes Association. Gesta-tional diabetes mellitus. Diabetes Care2003;26(Suppl. 1):S103S105
14. Guideline Development Group. Manage-ment of diabetes from preconception tothe postnatal period: summary of NICEguidance. BMJ 2008;336:714717
15. Lau J, Ioannidis JP, Schmid CH. Quanti-tative synthesis in systematic reviews.Ann Intern Med 1997;127:820826
16. Cochran W. The combination of esti-mates from different experiments. Bio-metrics 1954;10:101129
17. Petitti DB. Approaches to heterogeneity inmeta-analysis. Stat Med 2001;20:36253633
18. Higgins JP, Thompson SG. Quantifyingheterogeneity in a meta-analysis. Stat Med2002;21:15391558
19. Berlin JA, Antman EM. Advantages andlimitations of metaanalytic regressions ofclinical trials data.Online J Curr Clin Trials1994;Doc No 134
20. Begg CB, Mazumdar M. Operating char-acteristics of a rank correlation test forpublication bias. Biometrics 1994;50:
1088110121. Egger M, Davey Smith G, Schneider M,
Minder C. Bias in meta-analysis detectedby a simple, graphical test. BMJ1997;315:629634
22. Zhang C, Solomon CG, Manson JE, HuFB. A prospective study of pregravidphysical activity and sedentary behaviorsin relation to the risk for gestational dia-betes mellitus. Arch Intern Med2006;166:543548
23. Harizopoulou VC, Kritikos A, Papaniko-laou Z, Saranti E, Vavilis D, Klonos E, Pa-padimas I, Goulis DG. Maternal physicalactivity before and during early preg-nancy as a risk factor for gestational dia-betes mellitus. Acta Diabetol. 18 July2009 [Epub ahead of print]
24. Chasan-Taber L, Schmidt MD, Pekow P,Sternfeld B, Manson JE, Solomon CG,Braun B, Markenson G. Physical activityand gestational diabetes mellitus amongHispanic women. J Womens Health(Larchmt) 2008;17:9991008
25. Oken E, Ning Y, Rifas-Shiman SL,Radesky JS, Rich-Edwards JW, GillmanMW. Associations of physical activity andinactivity before and during pregnancywith glucose tolerance. Obstet Gynecol2006;108:12001207
26. Rudra CB, Williams MA, Lee IM, MillerRS, Sorensen TK. Perceived exertion inphysical activity and risk of gestational di-abetes mellitus. Epidemiology 2006;17:3137
27. Dempsey JC, Butler CL, Sorensen TK, LeeIM, Thompson ML, Miller RS, FrederickIO, Williams MA. A case-control study ofmaternal recreational physical activityand risk of gestational diabetes mellitus.Diabetes Res Clin Pract 2004;66:203215
28. Dempsey JC, Sorensen TK, Williams MA,Lee IM, Miller RS, Dashow EE, Luthy DA.Prospective study of gestational diabetesmellitus risk in relation to maternal recre-ational physical activity before and duringpregnancy. Am J Epidemiol 2004;159:663670
29. Redden SL, Lamonte MJ, FreudenheimJL, Rudra CB. The association betweengestational diabetes mellitus and recre-ational physical activity. Matern ChildHealth J. 6 March 2010 [Epub ahead ofprint]
30. Knopp RH, Bergelin RO, Wahl PW, Wal-den CE, Chapman M, Irvine S. Popula-tion-based lipoprotein lipid reference
values for pregnant women compared tononpregnant women classified by sexhormone usage. Am J Obstet Gynecol1982;143:626637
31. Knopp RH, Warth MR, Charles D, ChildsM, Li JR, Mabuchi H, Van Allen MI. Li-poprotein metabolism in pregnancy, fattransport to the fetus, and the effects ofdiabetes. Biol Neonate 1986;50:297317
32. Metzger BE, Buchanan TA, Coustan DR,de Leiva A, Dunger DB, Hadden DR, HodM, Kitzmiller JL, Kjos SL, Oats JN, PettittDJ, Sacks DA, Zoupas C. Summary andrecommendations of the Fifth Interna-tional Workshop-Conference on Gesta-tional Diabetes Mellitus. Diabetes Care2007;30(Suppl. 2):S251S260
33. Mottola MF. The role of exercise in theprevention and treatment of gestationaldiabetes mellitus. Curr Diab Rep 2008;8:299304
34. Yung LM, Laher I, Yao X, Chen ZY, HuangY, Leung FP. Exercise, vascular wall andcardiovascular diseases: an update (part2). Sports Med 2009;39:4563
35. Ryder JW, Chibalin AV, Zierath JR. Intra-cellular mechanisms underlying increasesin glucose uptake in response to insulin orexercise in skeletal muscle. Acta PhysiolScand 2001;171:249257
36. Ivy JL. Role of exercise training in theprevention and treatment of insulin re-sistance and non-insulin-dependent di-abetes mellitus. Sports Med 1997;24:321336
37. Marwick TH, Hordern MD, Miller T,Chyun DA, Bertoni AG, Blumenthal RS,Philippides G, Rocchini A, Council onClinical Cardiology, American Heart As-sociation Exercise, Cardiac Rehabilita-tion, and Prevention Committee, Councilon Cardiovascular Disease in the Young,Council on Cardiovascular Nursing,Council on Nutrition, Physical Activity,and Metabolism, Interdisciplinary Coun-cil on Quality of Care and Outcomes Re-search. Exercise training for type 2diabetes mellitus: impact on cardiovascu-lar risk: a scientific statement from theAmerican Heart Association. Circulation2009;119:32443262
38. Ning Y, Williams MA, Dempsey JC, So-rensen TK, Frederick IO, Luthy DA. Cor-relates of recreational physical activity inearly pregnancy. J Matern Fetal NeonatalMed 2003;13:385393
Tobias and Associates
care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 229