gestational diabetes mellitus and diet: a …...gestational diabetes mellitus (gdm) is one of the...

Gestational Diabetes Mellitus andDiet: A Systematic Review andMeta-analysis of RandomizedControlled Trials Examining theImpact of Modified DietaryInterventions onMaternalGlucoseControl andNeonatalBirthWeightDiabetes Care 2018;41:1346–1361 | https://doi.org/10.2337/dc18-0102

OBJECTIVE

Medical nutrition therapy is a mainstay of gestational diabetes mellitus (GDM)treatment. However, data are limited regarding the optimal diet for achievingeuglycemia and improved perinatal outcomes. This study aims to investigatewhether modified dietary interventions are associated with improved glycemiaand/or improved birthweight outcomes inwomenwith GDMwhen comparedwithcontrol dietary interventions.

RESEARCH DESIGN AND METHODS

Data from published randomized controlled trials that reported on dietary com-ponents, maternal glycemia, and birth weight were gathered from 12 databases.Data were extracted in duplicate using prespecified forms.

RESULTS

From 2,269 records screened, 18 randomized controlled trials involving 1,151women were included. Pooled analysis demonstrated that for modified dietaryinterventions when compared with control subjects, there was a larger decreasein fasting and postprandial glucose (24.07 mg/dL [95% CI27.58,20.57]; P = 0.02and 27.78 mg/dL [95% CI 212.27, 23.29]; P = 0.0007, respectively) and a lowerneed formedication treatment (relative risk 0.65 [95%CI 0.47, 0.88];P=0.006). Forneonatal outcomes, analysis of 16 randomized controlled trials including 841 par-ticipants showed that modified dietary interventions were associated with lowerinfant birth weight (2170.62 g [95% CI 2333.64, 27.60]; P = 0.04) and lessmacrosomia (relative risk 0.49 [95%CI 0.27, 0.88];P=0.02). Thequality of evidencefor these outcomes was low to very low. Baseline differences between groups inpostprandial glucose may have influenced glucose-related outcomes. As well,relatively small numbers of study participants limit between-diet comparison.

CONCLUSIONS

Modified dietary interventions favorably influenced outcomes related to maternalglycemiaandbirthweight.This indicates that there is roomfor improvement inusualdietary advice for women with GDM.

1Division of Endocrinology and Metabolism, De-partment of Medicine, University of Calgary,Calgary, Canada2Norfolk and Norwich University Hospitals, Nor-folk, U.K.3Department of Endocrinology and Nutrition,Hospital Mutua de Terrassa, Terrassa, Spain4Institute of Biomedical Research, Hospitalde la Santa Creu i Sant Pau, Barcelona,Spain5Rabin Medical Center, Tel Aviv University, Tel Aviv,Israel6Iberoamerican Cochrane Centre, Hospital de laSanta Creu i Sant Pau, Barcelona, Spain7CIBER Epidemiologıa y Salud Publica, Institutode Salud Carlos III, Madrid, Spain8Department of Epidemiology, Hospital de laSanta Creu i Sant Pau, Barcelona, Spain9Department of Pharmacology, Therapeutics andToxicology, Universitat Autonoma de Barcelona,Bellaterra, Spain10Nutricia Research, Utrecht, the Netherlands11Department of Pediatrics, University MedicalCentre Groningen, University of Groningen, Gro-ningen, the Netherlands12Nestle Research Center, Lausanne, Switzerland13Department of Obstetrics and Gynecology,University of Helsinki, Helsinki, Finland14Helsinki University Hospital, Helsinki, Finland15Institute of Biomedicine, University of Turku,Turku, Finland16Turku University Hospital, Turku, Finland17Statens Serum Institut, Copenhagen, Denmark18King’s College London, London, U.K.19Research and Development Department, Ab-bott Nutrition, Granada, Spain20Department of Nutritional Sciences, Universityof Vienna, Vienna, Austria

Jennifer M. Yamamoto,1 Joanne E. Kellett,2

Montserrat Balsells,3

Apolonia Garcıa-Patterson,4 Eran Hadar,5

Ivan Sola,4,6,7 Ignasi Gich,7,8,9

Eline M. van der Beek,10,11

Eurıdice Casta~neda-Gutierrez,12

Seppo Heinonen,13,14 Moshe Hod,5

Kirsi Laitinen,15,16 Sjurdur F. Olsen,17

Lucilla Poston,18 Ricardo Rueda,19

Petra Rust,20 Lilou van Lieshout,21

Bettina Schelkle,21 Helen R. Murphy,2,22,23

and Rosa Corcoy24,25,26

1346 Diabetes Care Volume 41, July 2018

RECONSIDER

INGPREG

NANCY

WITHDIABETES

https://doi.org/10.2337/dc18-0102

http://crossmark.crossref.org/dialog/?doi=10.2337/dc18-0102&domain=pdf&date_stamp=2018-06-07

Gestational diabetes mellitus (GDM) isone of the most common medical com-plications in pregnancy and affects anestimated 14% of pregnancies, or one inevery seven births globally (1). Womenwith GDM and their offspring are at in-creased risk of both short- and longer-term complications, including, for mothers,later development of type 2 diabetes, andfor offspring, increased lifelong risks ofdeveloping obesity, type 2 diabetes, andmetabolic syndrome (2–6). The adverseintrauterine environment causes epige-netic changes in the fetus that maycontribute to metabolic disorders, theso-called vicious cycle of diabetes (7).The mainstay of GDM treatment is

dietary and lifestyle advice, which in-cludes medical nutrition therapy, weightmanagement, and physical activity (8).Women monitor their fasting and post-meal glucose levels and adjust their in-dividual diet and lifestyle to meet theirglycemic targets. This pragmatic approachachieves the glycemic targets in approx-imately two-thirds of women with GDM(8). However, despite the importance ofmedical nutrition therapy and its wide-spread recommendation in clinical prac-tice, there are limited data regarding theoptimal diet for achieving maternal eu-glycemia (8–11). It is also unknownwhether the dietary interventions forachieving maternal glycemia are alsoeffective for reducing excessive fetalgrowth and adiposity (12).Different dietary strategies have been

reported including low glycemic index(GI), energy restriction, increase or de-crease in carbohydrates, and modifica-tions of fat or protein quality or quantity(12–14). Three recent systematic reviewshave been performed examining specificdiets and pregnancy outcomes (15–17).Viana et al. (16) and Wei et al. (15)concluded that low-GI diets were asso-ciated with a decreased risk of infantmacrosomia. However, the most recent

systematic review from Cochrane, in-cluding 19 trials randomizing 1,398women, found no clear difference inlarge for gestational age or other primaryneonatal outcomes with the low-GI diet(17). The primary maternal outcomeswere hypertension (gestational and/orpreeclampsia), delivery by cesarean sec-tion, and type 2 diabetes, outcomes forwhich most trials lacked statisticalpower, even when dietary subgroupswere combined. Remarkably, no sys-tematic reviewsexamined the impactofmodified dietary interventions on thedetailedmaternal glycemic parameters,includingchange inglucose-relatedvar-iables, the outcomes that are most di-rectly influenced by diet.

To address this knowledge gap, weperformed a systematic review and meta-analysis of randomized controlled trialsto investigate whether modified dietaryinterventions (defined as a dietary in-tervention different from the usual oneused in the control group) inwomenwithGDM offer improved glycemic controland/or improved neonatal outcomeswhen compared with standard diets.

RESEARCH DESIGN AND METHODS

In accordance with a published protocol(PROSPERO CRD42016042391), we per-formed a systematic review and meta-analysis. Reporting is in accordance withthe Preferred Reporting Items for System-atic Reviews and Meta-Analyses (PRISMA)guidelines (18). An international panel ofexperts was formed by the InternationalLife Sciences Institute Europe. This paneldeterminedthereviewprotocolandcarriedout all aspects of the review.

Data Sources and Search StrategyThe following databases were searchedfor all available dates using the searchtermsdetailed in Supplementary Table 1:PubMed, MEDLINE, Cochrane CentralRegister of Controlled Trials (CENTRAL),

Embase, Cumulative Index to Nursingand Allied Health Literature (CINAHL),Webof Science Core Collection, Applied SocialSciences Index and Abstracts, ProQuest,ProQuest Dissertations & ThesesdA&Iand UK & Ireland, National Institutefor Health and Care Excellence evidencesearch, Scopus, UK Clinical Trials Gate-way, ISRCTN, and ClinicalTrials.gov. Theinitial search was performed in July2016. An updated search of MEDLINE,Embase, CENTRAL, and CINAHL was per-formed on 3 October 2017 using thesame search terms.

A hand-search of relevant reviewsand all included articles was conductedto identify studies for potential inclu-sion. As well, experts on the panel wereconsulted for the inclusion of additionalarticles. Reference management wascarried out using EndNote.

Study SelectionAll titles and abstracts were assessedindependently and in duplicate to iden-tify articles requiring full-text review.Published studies fulfilling the followingcriteria were included: randomized con-trolled trials, evaluated modified dietaryinterventions on women with GDM, glu-cose intoleranceorhyperglycemiaduringpregnancy, reported-on primary mater-nal and neonatal outcomes, includedwomen aged 18–45 years, had a durationof 2 weeks or more, and were publishedin English, French, Spanish, Portuguese,Italian, Dutch, German, or Chinese. Weexcluded studies that included partic-ipants with type 1 or type 2 diabetes ifdata for participants with GDM werenot presented independently, if dietarycharacteristics were not available, if thestudy was in animals, or if the study didnot report outcomes of interest. We didnot include studies of nutritional supple-ments such as vitamin D or probiotics asrecent reviews have addressed thesetopics (19,20).

21International Life Sciences Institute Europe,Brussels, Belgium22Cambridge University Hospitals NHS Founda-tion Trust, Cambridge, U.K.23Norwich Medical School, University of EastAnglia, Norwich, U.K.24DepartmentofMedicine,UniversitatAutonomade Barcelona, Bellaterra, Spain25CIBER Bioengineering, Biomaterials and Nano-technology, Instituto de Salud Carlos III, Madrid,Spain

26Department of Endocrinology and Nutrition, Hos-pital de la Santa Creu i Sant Pau, Barcelona, Spain

Corresponding author: Helen R. Murphy, [email protected].

Received13January2018andaccepted10March2018.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc18-0102/-/DC1.H.R.M. and R.C. contributed equally to thiswork.

© 2018 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered. More infor-mation is available at http://www.diabetesjournals.org/content/license.

See accompanying commentary, p. 1343.See accompanying articles, pp. 1337,1339, 1362, 1370, 1378, 1385, 1391,and e111.

care.diabetesjournals.org Yamamoto and Associates 1347

http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc18-0102/-/DC1

mailto:[email protected]

mailto:[email protected]



http://www.diabetesjournals.org/content/license

http://www.diabetesjournals.org/content/license

http://care.diabetesjournals.org

All citations identified after title andabstract assessment were full-text re-viewed in duplicate. Reasons for exclu-sion at the full-text review stage wererecorded. Any disagreements betweenreviewers were resolved by consensusand with consultation with the expertgroup when required.

Data ExtractionData from included studies were ex-tracted in duplicate using prespecifieddata extraction forms. Extracted dataelements included study and participantdemographics, study design, diagnosticcriteria for GDM, glucose intolerance orhyperglycemia, funding source, descrip-tion of modified dietary intervention andcomparator, and maternal and neonataloutcomes. For studies with missing data,inconsistencies, orotherqueries, authorswere contacted. Record managementwas carried out usingMicrosoft Excel andRevMan.For articles providing information on

maternal weight, fasting glucose, post-prandial glucose, HbA1c, or HOMA insulinresistance index (HOMA-IR) at baselineand postintervention but not theirchange, change was calculated as thedifferencebetweenpostinterventionandbaseline. Standard deviations were im-puted using the correlation coefficientobserved in articles reporting full infor-mation on the variable at baseline andpostintervention and its change or acorrelation coefficient of 0.5 when thisinformation was not available (21). Asstudies differed in postprandial glucoseat baseline, glycemic control at studyentrywasnotconsideredtobeequivalentin both arms, and thus continuous glu-cose-related variables at follow-up arereported as change from baseline.

Data SynthesisThe primary outcomes were maternalglycemic outcomes (mean glucose, fast-ing glucose, postprandial glucose [afterbreakfast, lunch, and dinner and com-bined], hemoglobin A1c [HbA1c], assess-ment of insulin sensitivity by HOMA-IR,and change in these parameters frombaseline to assessment; medicationtreatment [defined as oral diabetesmed-ications or insulin]) and neonatal birthweight outcomes (birth weight, macro-somia, and large for gestational age).Data were pooled into relative risks

(RRs) or mean differences with 95% CI

for dichotomous outcomes and contin-uousoutcomes,respectively.Meta-analysiswas performed using random-effectsmodels. A prespecified analysis stratifiedby type of diet and quality assessment wasperformed to explore potential reasonsfor interstudy variation. Heterogeneitywas assessed using I2 statistics. Smallstudy effects were examined for usingfunnel plots. Analyses were conductedusing RevMan version 5.3. Pooled esti-mation of birth weight in the study andcontrol arms, both overall and accordingto the specific diet intervention, wasperformed using Stata 14.0.

Quality AssessmentMethodological quality and bias assess-ment was completed by two reviewers.Risk of bias was assessed using theCochrane Collaboration tool, which ratesseven items as being high, low, or unclearfor risk of bias (21). These items includedrandom sequence generation, allocationconcealment, blinding of participantsand personnel, blinding of outcome as-sessment, incomplete outcome data, se-lective outcome reporting, and otherpotential sources of bias (21). A sensi-tivity analysis was performed excludingarticles with relevant weaknesses in trialdesign or execution.

The overall quality of the evidencewasalso assessed using Grading of Recom-mendations Assessment, Developmentand Evaluation (GRADE) working groupguidelines (21). GRADE was assessedfor all primary and secondary outcomes,bothmaternal and neonatal, but withoutsubgroup analysis per different dietaryintervention for each outcome measure.

RESULTS

We screened 2,269 records for poten-tial inclusion, and 126 articles were re-viewed in full (Supplementary Fig. 1).Eighteen studies (12–14,22–36) were in-cluded in themeta-analysiswitha total of1,151 pregnant women with GDM.

Study CharacteristicsThe types of modified dietary interven-tion included low-GI (n = 4), Dietary Ap-proaches to Stop Hypertension (DASH)(n = 3), low-carbohydrate (n = 3), fat-modification (n = 2), soy protein–enrichment (n = 2), energy-restriction(n = 1), high-fiber (n = 1), and ethnic diets(i.e., foods commonly consumed ac-cording to participant’s ethnicity) (n = 1)

and behavioral intervention (n = 1).Details of the study characteristics areincluded in Table 1. Most trials were sin-gle centered and had small sample sizes(range 12–150). Only two trials (one eachfrom Spain and Australia) included over100 participants, nine had 50–100 par-ticipants, and seven studies had fewerthan 50 participants. They were per-formed in North America, Europe, orAustralasia and all had a duration of atleast 2 weeks. The ethnicity of partic-ipants was reported in seven studies(12,13,26,29,31,32,34).

Most studies assessed individual di-etary adherence using food diaries (13,23–36). Althoughmost studies did reportan overall difference in dietary compo-sition between the intervention diet andcontrol diet, few studies reported a de-tailed assessment of dietary adherence.Only five studies used a formal measure ofadherence (24,25,29,33,34), and four ofthem reported data (25,29,33,34). Ad-herence ranged from 20% to 76% in thecontrol groups and 60% to 80% in theintervention groups.

Participant CharacteristicsWhen baseline characteristic data werepooled,women in the intervention groupwere older than women in the controlgroup (pooled mean difference 0.60years [95% CI 0.06, 1.14]) and had higherpostprandialglucose(5.47[0.86,10.08]),most influenced by the DASH and ethnicdiet studies. There was no overall sig-nificant difference between the inter-vention and control groups for BMI,gestational age at enrollment, fastingglucose, HbA1c, or HOMA-IR.

Maternal Glycemic Outcomes for AllModified Dietary InterventionsPooled risk ratios in 15 studies involving1,023womendemonstratedalowerneedfor medication (RR 0.65 [95% CI 0.47,0.88]; I2 = 55) (Table 2). Thirteen studies(n = 662 women) reported fasting glu-cose levels, nine (n = 475) reported com-binedpostprandial glucosemeasures, andthree (n = 175) reported post-breakfastglucose measures. Pooled analysis dem-onstrated a larger decrease in fasting, com-bined postprandial, and post-breakfastglucose levels in modified dietary inter-ventions (mean 24.07 mg/dL [95% CI27.58,20.57],I2=86,P=0.02;27.78mg/dL[212.27, 23.29], I2 = 63, P = 0.0007;and24.76mg/dL [29.13,20.38], I2 =34,P = 0.03, respectively) compared with

1348 GDM and Diet Diabetes Care Volume 41, July 2018


Table

1—Characteristicsofstudiesincluded

Author,year

(ref.)

Country

nEstimated

sample

size

Defi

nition

ofGDM

Durationof

dietary

interven

tion

Gestationalagein

weeks

aten

rollm

ent

(mean6

SD)

BaselineBMI,

kg/m

2(m

ean6

SD)

Meanmaternal

age,

years

(mean6

SD)

Dietary

interven

tion

Dietcomposition

(mean6

SD)*

Low-GIdiet

Grant,

2011

(26)

Canada

4750

todetect

a0.6mmol/L

difference

incapillary

glucose;nnot

achieved

CanadianDiabetes

Association(40)

28weeks

until

delivery

Control:

296

2.35

Interven

tion†:

296

3.21

Control:26

64.69

Interven

tion:

276

4.58

(prepregnancy)

Control:

346

0.46

Interven

tion:

346

5.16

Low

GI:Women

were

provided

withalist

ofstarch

choices

specificto

either

interven

tion(low

GI)orcontrol

Control:GI

58.0

60.5

Interven

tion:

GI49

.06

0.8

Louie,

2011

(29)

Australia

99120to

detect

a26

0-g

difference

inbirth

weight

(stoppe

dearly

because

of

smallerthan

expectedSD

)

Australasian

Diabetes

inPregnancySociety

criteria

(41)

Randomization

untildelivery

Control:

29.7

63.5

Interven

tion:

296

4.0

Control:

24.1

65.7

Interven

tion:

23.9

64.4

(prepregnancy)

Control:

32.4

64.5

Interven

tion:

346

4.1

LowGI:Target

GI#

50butotherwise

similarcomposition

tothecontroldiet

Control:en

ergy

1,93

46

465;

carbohydrate

40.3

68.3;

protein

22.2

67.5;

fat

35.1

616

.9;GI

53.0

66.5

Interven

tion:

energy

1,83

66

403;

carbohydrate

38.7

68.3;

protein

23.4

65.8;

fat

34.9

611

.0;GI

47.0

66.5

Ma, 20

15(30)

China

95Notreported

ChineseMed

ical

Associationand

American

Diabetes

Association(42)

24–26

weeks

until

delivery

Control:

27.9

61.1

Interven

tion:

27.5

61.1

Control:

21.156

2.75

Interven

tion:

21.906

3.14

(prepregnancy)

Control:

30.0

63.5

Interven

tion:

30.1

63.8

Low

GI:Women

provided

withan

exchange

listfor

starch

choices

specificto

either

interven

tion(low

GI)orcontrol

Control:en

ergy

2,03

06

215;

carbohydrate

49.8

66.8;

protein

18.8

62.5;

fat

31.8

63.8;

GI

53.8

62.5

Interven

tion:

energy

2,00

66

215;

carbohydrate

48.566

7.0;

protein

18.9

62.9;

fat32

.16

4.1;

GI

50.1

62.2

Con

tinu

edon

p.13

50



Table

1—Continued

Author,year

(ref.)

Country

nEstimated

sample

size

Defi

nition

ofGDM

Durationof

dietary

interven

tion

Gestationalagein

weeks

aten

rollm

ent

(mean6

SD)

BaselineBMI,

kg/m

2(m

ean6

SD)

Meanmaternal

age,

years

(mean6

SD)

Dietary

interven

tion

Dietcomposition

(mean6

SD)*

Moses,

2009

(13)

Australia

63Notreported

Australasian

Diabetes

inPregnancySociety

(41)

28–32

weeks

until

delivery

Control:

29.9

61.11

Interven

tion:

30.3

61.11

Control:32

.86

7.92

Interven

tion:

32.0

66.68

(aten

rollm

ent)

Control:

31.3

64.52

Interven

tion:

30.8

63.90

LowGI:Women

asked

toavoid

specific

high-GIfoodsand

wereprovided

with

abooklet

outlining

carbohydrate

choices

Control:en

ergy

1,65

66

433;

carbohydrate

36.2

68.2;

protein

24.0

64.4;

fat

34.3

69.9;

GI

52.2

66.0

Interven

tion:

energy

1,71

36

368;

carbohydrate

36.7

66.1;

protein

23.9

63.9;

fat

33.4

66.12

;GI

48.0

65.0

DASH

diet

Asemi,

2013

(22)

Iran

3432

for“keyvariable

serum

HDL”

50-gglucose

challenge

.14

0mg/dL→

100gOGTT;G

DM

iftw

oormore

of

fasting.95

mg/dL,

1-h18

0mg/dL,2-h

155mg/dL,or3-h

140mg/dL

4weeks

Notreported

Control:

31.4

65.7

Interven

tion:

29.0

63.2(at

enrollm

ent)

Control:

29.4

66.2

Interven

tion:

30.7

66.7

DASH

diet:dietrich

infruit,vegetables,

wholegrains,and

low-fat

dairy;lowin

saturatedfats,

cholesterol,refined

grains,andsw

eets

Control:en

ergy

2,39

26

161;

carbohydrate

54.0

66.9;

protein

17.6

62.8;

fat

29.3

65.6

Interven

tion:

energy

2,40

06

25;

carbohydrate

66.8

62.2;

protein

16.8

61.2;

fat

17.6

60.9

Asemi,

2014

(23)

Iran

5242

todetecta75

-gdifference

inbirth

weight

Asabove

4weeks

Control:

25.9

61.4

Interven

tion:

25.8

61.4

Control:31

64.9

Interven

tion:

29.2

63.5

(aten

rollm

ent)

Control:

30.7

66.3

Interven

tion:

31.9

66.1

DASH

diet:as

above

Control:en

ergy

2,35

26

163;

carbohydrate

54.2

67.1;

protein

18.2

63.4;

fat

28.5

65.6

Interven

tion:

energy

2,40

76

30;

carbohydrate

66.4

62.04

;protein

17.0

61.3;

fat17

.46

1.0

Con

tinu

edon

p.13

51


Table

1—Continued

Author,year

(ref.)

Country

nEstimated

sample

size

Defi

nition

ofGDM

Durationof

dietary

interven

tion

Gestationalagein

weeks

aten

rollm

ent

(mean6

SD)

BaselineBMI,

kg/m

2(m

ean6

SD)

Meanmaternal

age,

years

(mean6

SD)

Dietary

interven

tion

Dietcomposition

(mean6

SD)*

Yao, 2015

(36)

China

3342

todetecta75

-gdifference

inbirthweight;not

achieved

50-gglucose

challenge

→10

0gOGTT

resultswithtw

oor

more

of

fasting.95

mg/dL,

1-h$18

0mg/dL,

2-h$15

5mg/dL,or

3-h$14

0mg/dL

4weeks

Control:

25.7

61.3

Interven

tion:

26.9

61.4

Control:30

.96

3.6

Interven

tion:

30.2

64.1

(aten

rollm

ent)

Control:

28.3

65.1

Interven

tion:

30.7

65.6

DASH

diet:same

asabove

Control:en

ergy

2,38

66

174;

carbohydrate

52.3

67.2;

protein

18.0

63.3;

fat

28.3

65.1

Interven

tion:

energy

2,40

86

54;

carbohydrate

66.7

62.3;

protein

16.9

61.2;

fat

17.176

1.16

Low-carbohydrate

diets

Cypryk,

2007

(25)

Poland

30Notreported

WorldHealth

Organization

criteria

2weeks

29.2

65.4

Notreported

28.7

63.7

Low

(interven

tion)vs.

high(control)

carbohydrate

(45%

vs.60

%oftotal

energy,

respectively)

Control‡:

carbohydrate

60;

protein

25;fat

15Interven

tion‡:

carbohydrate

45;

protein

25;fat30

Hernandez,

2016

(12)

U.S.

12Pilotstudyto

estimateSD

Carpen

terandCoustan

criteria

(43)

30–31

weeks

until

delivery

Control§:

31.7

62.45

Interven

tion:

31.2

60.98

Control:34

.36

3.92

Interven

tion:

33.4

63.43

(aten

rollm

ent)

Control:

306

2.45

Interven

tion:

286

4.90

Low

carbohydrate

(interven

tion)vs.

higher-complex

carbohydrate/

lower

fat(control)

Control‡:

carbohydrate

60;

protein

15;fat

25Interven

tion‡:

carbohydrate

40;

protein

15;fat45

Moreno-

Castilla,

2013

(31)

Spain

152

152todetecta

22%

difference

inneedforinsulin

2006

National

Diabetes

and

Pregnancy

Clinical

Guidelines

(44,45)

#35

weeks

until

delivery

Control:

30.1

63.5

Interven

tion:

30.4

63.0

Control:26

.66

5.5

Interven

tion:

25.4

65.7

(prepregnancy)

Control:

32.1

64.4

Interven

tion:

30.4

63.0

Low

carbohydrate

(interven

tion)vs.

control(40%vs.55%

oftotaldieten

ergy

ascarbohydrate)

Control‡:en

ergy

1,80

0minim

um;

carbohydrate

55;

protein

20;fat

25Interven

tion‡:

energy

1,80

0minim

um;

carbohydrate

40;

protein

20;fat40

Con

tinu

edon

p.13

52



Table

1—Continued

Author,year

(ref.)

Country

nEstimated

sample

size

Defi

nition

ofGDM

Durationof

dietary

interven

tion

Gestationalagein

weeks

aten

rollm

ent

(mean6

SD)

BaselineBMI,

kg/m

2(m

ean6

SD)

Meanmaternal

age,

years

(mean6

SD)

Dietary

interven

tion

Dietcomposition

(mean6

SD)*

Soyprotein–en

richmen

tdiets

Jamilian,

2015

(27)

Iran

6856

(minim

um

clinical

difference

not

reported

)

One-step

75gOGTT,

American

Diabetes

Association(46)

6weeks

Notreported

Control:28

.46

3.4

Interven

tion:

28.9

65.0

Control:

29.3

64.2

Interven

tion:

28.2

64.6

Soyprotein

diethad

thesameam

ountof

protein

ascontrol

dietbuttheprotein

portionwas

made

upof35

%anim

alprotein,35

%soy

protein,30

%other

plantproteins

Control:en

ergy

2,42

66

191;

carbohydrate

54.6

67.1;

protein

14.4

61.7;

fat

32.1

65.4

Interven

tion:

energy

2,30

86

194;

carbohydrate

54.6

67.3;

protein

15.0

62.6;

fat

30.3

64.7

Sarathi,

2016

(14)

India

62Notreported

International

Associationof

Diabetes

and

Pregnancy

Study

Groupscriteria

(47)

From

diagnosis

untildelivery

Control:

25.566

1.69

Interven

tion:

25.196

1.92

Notreported

Control:29

.176

3.38

Interven

tion:

29.436

2.98

Soyprotein

diet:25

%ofcereal

partof

high-fiber

complex

carbohydrates

replacedwithsoy

Control‡:en

ergy

1,60

0–2,00

0;minim

um

carbohydrate

175g

Interven

tion‡:

energy

1,60

0–2,00

0;minim

um

carbohydrate

175g

Fat-modificationdiets

Lauszus,

2001

(28)

Den

mark

2720

todetect

adifference

incholesterolof

0.65

mmol/L

3-h75

gOGTT

with

bloodsamples

takenevery

30minutes,GDM

if2ormore

glucoses.3SD

above

themean

34weeks

until

delivery

Notreported

Control:32

.26

5.61

Interven

tion:

35.3

68.65

(aten

rollm

ent)

Control:

296

3.74

Interven

tion:

316

3.61

High monounsaturated

fattyacids:source

was

hybrid

sunflower

oilwith

high-contentoleic

acid

andsnacks

of

almondsand

hazelnuts

Control:en

ergy

1,727;

carbohydrate

50.0

63.6;

protein

19.0

63.6;

fat

30.0

67.2

Interven

tion:

energy

1,98

2;carbohydrate

466

3.5;

protein

166

3.5;

fat37

63.5

Con

tinu

edon

p.13

53


Table

1—Continued

Author,year

(ref.)

Country

nEstimated

sample

size

Defi

nition

ofGDM

Durationof

dietary

interven

tion

Gestationalagein

weeks

aten

rollm

ent

(mean6

SD)

BaselineBMI,

kg/m

2(m

ean6

SD)

Meanmaternal

age,

years

(mean6

SD)

Dietary

interven

tion

Dietcomposition

(mean6

SD)*

Wang,

2015

(35)

China

84Notreported

International

Associationof

Diabetes

and

Pregnancy

Study

Groupscriteria

(47)

;27

weeks

until

delivery

Control:

27.3

61.96

Interven

tion:

27.4

61.52

Control:22

.26

3.6

Interven

tion:

21.4

63.0

(prepregnancy)

Control:

29.7

64.64

Interven

tion:

30.3

64.17

Polyunsaturatedfatty

acid

meals(50–54

%carbohydrate,31–

35%fatw

ith45–40

gsunflower

oil)

Control:en

ergy

1,97

86

107;

carbohydrate

55.4

62.0;

protein

17.9

61.0;

fat

26.7

61.3

Interven

tion:

energy

1,96

06

90;

carbohydrate

47.7

60.7;

protein

18.0

60.7;

fat

34.3

60.2

Other

diets

Bo, 2014

(24)

Italy

99in

diet

study

(total

n=20

0)

200todetecta

10%

difference

infastingglucose

(based

on

exercise

portion

oftrial)

75gOGTT

24–26

weeks

until

delivery

Notreported

Control:26

.86

4.1

Interven

tion:

26.9

64.6

Control:

33.9

65.3

Interven

tion:

35.1

64.4

Beh

avioraldietary

recommen

dations:

individual

recommen

dations

forhelpingdietary

choices

Control:en

ergy

2,11

66

383;

carbohydrate

46.9

65.9;

protein

15.6

62.6;

fat

37.4

64.2

Interven

tion:

energy

2,15

66

286;

carbohydrate

47.8

64.9;

protein

15.5

62.4;

fat

36.7

63.9

Rae, 2000

(32)

Australia

124

120to

detect

adecreasein

insulin

use

from

40%to

15%and

adecreasein

macrosomia

from

25%to

5%

OGTT

fasting

glucose.5.4mmol/

Land/or2-h

glucose.7.9mmol/

L(48)

,36

weeks

until

delivery

Control:

28.3

64.6

Interven

tion:

28.1

65.8

Control:38

.06

0.7

Interven

tion:

37.9

60.7(at

diagnosis)

Control:30

.6Interven

tion:

30.2

(SDnot

reported

)

Moderateen

ergy

restriction(1,590–

1,776kcal/day)vs.

control(2,010–

2,220kcal/day)

Control:en

ergy

1,63

06

339;

carbohydrate

41.0

64.6;

protein

24.0

62.3;

fat

34.0

65.3

Interven

tion:

energy

1,56

66

289;

carbohydrate

42.0

65.7;

protein

25.0

62.4;

fat

31.0

65.7

Con

tinu

edon

p.13

54



control group. There were no significantdifferences in change in HbA1c (sevenstudies), HOMA-IR (four studies), or inpost-lunch or -dinner glucose levels (twostudies).

Neonatal Birth Weight Outcomesfor All DietsPooled mean birth weight was 3,266.65 g(95% CI 3,172.15, 3,361.16) in the modi-fied dietary intervention versus 3,449.88 g(3,304.34, 3,595.42) in the control group.Pooled analysis of all 16 modified die-tary interventions including 841 partic-ipants demonstrated lower birth weight(mean2170.62g[95%CI2333.64,27.60],I2=88;P=0.04)and lessmacrosomia (RR0.49 [95% CI 0.27, 0.88], I2 = 11; P = 0.02)compared with conventional dietary ad-vice (Table 2 and Fig. 1). There was nosignificant difference in the risk of large-for-gestational-age newborns in modi-fied dietary interventions as comparedwith control diets (RR 0.96 [95% CI 0.63,1.46], I2 = 0; P = 0.85).

Subgroup Meta-analysis by Typesof Dietary InterventionsPooled analysis of low-GI diets showeda larger decrease in fasting (26,29,30),postprandial, and post-breakfast glu-cose compared with control diets (26,30)(Table 2). However, the pooled analysis ofthe DASH diet showed significant favor-able modifications in several outcomes,including change in fasting (22,36) andpostprandial glucose (22), HOMA-IR (35),HbA1c (22), medication need (22,23,36),infant birth weight (23,36), and macro-somia (23,36) (Tables 2 and3). Last, pooledanalysis of the soy protein–enriched dietdemonstrated a significant decrease inmedication use and birth weight (14,27)(Tables 2 and 3). One soy–protein inter-vention (n = 68 participants) describedsignificantly lowerHOMA-IR(27) (Table2).

Behavioral (one study) and ethnic-specific modified dietary interventions(one study) were included. The behav-ioral changedietary interventionreportedsignificant differences in change in post-prandial glucose and in HbA1c (Table 2)(24). The ethnic diet study demonstrateda significantly larger decrease in fastingand postprandial glucose (Table 2) (34).Fat-modification, low-carbohydrate, andenergy-restriction diets were not asso-ciated with a significant difference inour primary outcomes in the stratifiedanalysis.

Table

1—Continued

Author,year

(ref.)

Country

nEstimated

sample

size

Defi

nition

ofGDM

Durationof

dietary

interven

tion

Gestationalagein

weeks

aten

rollm

ent

(mean6

SD)

BaselineBMI,

kg/m

2(m

ean6

SD)

Meanmaternal

age,

years

(mean6

SD)

Dietary

interven

tion

Dietcomposition

(mean6

SD)*

Reece,

1995

(33)

U.S.

50Po

sthoc

calculation

Notreported

24–29

weeks

until

delivery

Notreported

Notreported

Notreported

Fiber-enriched

diet:

fiber

takenas

fiber-

rich

foods

(40g/day)

andahigh-fiber

drink(40g/day)

Control‡:

carbohydrate50

;fat

30;fiber

20g/day

Interven

tion‡:

carbohydrate60

;fat

20with80

gfiber/

day

Valen

tini,

2012

(34)

Italy

20Notrep

orted

(pilot

study)

FourthInternational

Workshop

Conferen

ceon

Gestational

Diabetes

Mellitus

(49)

From

diagnosis

(screeningat

24–28

weeks)

untildelivery

Control

27.1

65.9

Interven

tion:

21.3

66.8

Control:24

.16

4.7

Interven

tion:

25.7

63.6

(prepregnancy)

Control:

30.2

64.7

Interven

tion:

28.9

63.3

Ethnicmealplan:foods

commonly

consumed

per

participant’s

ethnicity

withthe

samekcal

and

nutrient

compositionas

the

controldiet

Control‡:

carbohydrate

53;

protein

18;fat28

;fiber

26g/day

Interven

tion‡:

carbohydrate

55;

protein

17;fat28

;fiber

21g/day

Unless

otherwisestated

,theunitsarekcal/day

foren

ergy,%

forcarbohydrate,protein,andfat.OGTT,oralglucose

tolerance

test.*R

eported

actual

dietary

intake.When

notreported

,prescribed

dietary

intake

isreported

.†Interven

tionisdefi

ned

asdietary

interven

tiondifferentfrom

theusual

dietary

interven

tionusedin

thecontrolgroup.‡Indicates

prescribed

diet.§Thecontroland

interven

tiongroupswerereversed

forthepurpose

ofmeta-analysisso

itcould

beincluded

inthelow-carbohydrate

group.


Table 2—Pooled analyses of primary maternal glycemic and infant birth weight outcomes

Outcome Diet subgroup N of studies N of women Effect estimate I2 (%)

Maternal glycemic outcomes

Mean [95% CI]

Change in fasting glucose (mg/dL) All diets 13 662 24.07 [27.58, 20.57] 86Low GI (26,29,30) 3 195 25.28 [26.83, 23.73] 0DASH (22,36) 2 67 211.55 [214.00, 29.09] 0Low carbohydrate (12,25) 2 42 3.81 [24.29, 11.92] 69Fat modification (28,35) 2 109 4.87 [20.44, 10.18] 0Soy protein (14,27) 2 130 27.47 [220.28, 5.34] 91Behavior (24) 1 99 21.50 [25.66, 2.66] dEthnic (34) 1 20 225.34 [237.57, 213.11] d

Change in postprandial glucose (mg/dL) All diets 9 475 27.78 [212.27, 23.29] 63Low GI (26,30) 2 121 27.08 [212.07, 22.08] 4DASH (22) 1 34 245.22 [268.97, 221.47] dLow carbohydrate (25) 1 30 23.00 [210.06, 4.06] d

Fat modification (28,35) 2 109 26.43 [213.08, 0.22] 0Soy protein (14) 1 62 21.05 [211.03, 8.93] d

Behavior (24) 1 99 26.90 [211.68, 22.12] dEthnic (34) 1 20 216.28 [222.83, 29.73] d

Change in post-breakfast glucose (mg/dL) All 3 175 24.76 [29.13, 20.38] 34Low GI (30) 1 83 28.6 [214.11, 23.09] d

Low carbohydrate (25) 1 30 23.00 [28.15, 2.15] dSoy protein (14) 1 62 21.05 [29.73, 7.63] d

Change in post-lunch glucose (mg/dL) All 2 92 4.50 [21.90, 10.90] 0Low carbohydrate (25) 1 30 4.00 [24.56, 12.56] d

Soy protein (14) 1 62 5.14 [24.51, 14.79] d

Change in post-dinner glucose (mg/dL) All 2 92 1.81 [25.28, 8.90] 13Low carbohydrate (25) 1 30 1.00 [28.14, 10.14] d

Soy protein (14) 1 62 3.03 [28.20, 14.26] d

Change in HOMA-IR (mIU/mL 3 mmol/L) All 4 212 21.10 [22.26, 0.07] 90DASH (36) 1 33 21.90 [22.36, 21.44] dLow carbohydrate (12) 1 12 0.60 [21.90, 3.10] d

Soy protein (27) 1 68 22.00 [23.17, 20.83] d

Behavior (24) 1 99 20.30 [20.71, 0.11] d

Change in HbA1c (%) All 7 407 20.05 [20.13, 0.02] 84Low GI (29,30) 2 167 0.01 [20.02, 0.03] 0DASH (22) 1 34 20.25 [20.42, 20.08] d

Fat modification (28) 1 25 0.10 [20.14, 0.34] d

Soy protein (14) 1 62 20.01 [20.07, 0.05] dBehavior (24) 1 99 20.19 [20.26, 20.12] d

Ethnic diet (34) 1 20 20.05 [20.27, 0.17] d

RR [95% CI]

Medication treatment All 15 1023 0.65 [0.47, 0.88] 55Low GI (13,26,29,30) 4 293 0.80 [0.55, 1.14] 34DASH (22,23,36) 3 119 0.29 [0.17, 0.50] 0Low carbohydrate (31) 1 150 1.00 [0.75, 1.34] d

Energy restriction (32) 1 117 1.05 [0.47, 2.34] d

Fat modification (35) 1 84 Not estimable dSoy protein (14,27) 2 130 0.44 [0.21, 0.91] 0Behavior (24) 1 99 0.61 [0.15, 2.42] d

Ethnic (34) 1 20 2.00 [0.21, 18.69] dFiber (33) 1 11 Not estimable d

Infant birth weight outcomes

Mean [95% CI]

Birth weight (g) All 16 841 2170.62 [2333.64, 27.60] 88

Low GI (13,26,29,30) 4 276 254.25 [2178.98, 70.47] 0

DASH (22,23,36) 3 119 2598.19 [2663.09, 2533.30] 0

Low carbohydrate (12,25) 2 42 57.73 [2164.93, 280.39] 0


Fat modification (28,35) 2 109 2139.61 [2294.80, 15.58] 0

Soy protein (14,27) 2 131 2184.67 [2319.35, 249.98] 0

Continued on p. 1356



Secondary OutcomesWeight gain from inclusion was lowerfor low-carbohydrate diets and cesareanbirth for DASH diets (SupplementaryTable 2). Specific diet interventions didnot show significant between-group dif-ferences in maternal gestational weightgain throughout pregnancy, preeclamp-sia/eclampsia, neonatal hypoglycemia asdefined by the authors, preterm birth,neonatal intensive care unit admission,or small-for-gestational-age newborns(Supplementary Tables 2 and 3).

Sensitivity Analysis of PrimaryOutcomesSensitivity analysis was performed toexplore reasons for heterogeneity andto assess outcomes when studies withmethodological concernswere removed.We were unable to include four studies(22,23,34,36), including all the DASHdiet studies, where clarification of certainaspects of the results could not be ob-tained,evenafteradirectapproachtotheauthors. The authors of the ethnic dietstudy responded to queries but did notprovide the required information re-garding gestational age at randomiza-tion (34). After these studies are removed,the changes in postprandial glucose(mean 25.90 mg/dL [95% CI 27.93,23.88], I2 = 0; P = 0.0001), post-breakfastglucose levels (24.76 mg/dL [29.13,20.38], I2 = 34; P = 0.03), and birth weight(274.88 g [2144.86, 24.90], I2 = 1; P =0.04) remained significant when alldietswere combined (Table 3). Further-more, the heterogeneity in most primary

outcomes decreased after removal ofthese four studies.

When dietary subgroups were as-sessed, low-GI diets had significant differ-ences in changes in fasting (mean 25.33mg/dL [95%CI26.91,23.76]) (26,29,30),postprandial (27.08 mg/dL [212.07,22.08]) (26,30), and post-breakfast(28.6 mg/dL [214.11,23.09]) glucose(26,30). The soy protein–enriched diethad differences in change of HOMA-IR(mean 22.00 [95% CI 23.17, 20.83])(27), required less medication use (RR0.44 [95% CI 0.21, 0.91]), and had alower birth weight (mean 2184.67 g[95% CI 2319.35, 249.98]) (14,27). Thebehavior modification diet had signifi-cant differences in change in postpran-dial glucose (mean26.90 mg/dL [95% CI29.85, 23.95]) and in HbA1c (20.19%[20.26, 20.12]) (24) (Table 3).

Assessment of Bias and Quality of theEvidenceNone of the included studies were as-sessed as having a low risk of bias in allseven items of the Cochrane Collabora-tion tool (Supplementary Fig. 2). Moststudies were high risk for blinding ofparticipants and personnel and for othersources of bias (Supplementary Fig. 3).Studies scored high risk for other sourcesof bias for concerns such as baseline differ-ences and industry funding. Most studieshad an unclear risk of bias for selectiveoutcome reporting and very few had reg-istered protocols (Supplementary Fig. 3).

GRADE assessment for the outcomesof interest reveals overall low to very

lowquality of evidence (SupplementaryTable 4). Considerations to downgradequality of evidence involved the entirespectrum, including limitations in thestudy design, inconsistency in study results,and indirectness and imprecision in effectestimates.

Evaluation for Small Study EffectFunnel plots of means and RRs of theprimary outcomes for the main analysisare shown in Supplementary Figs. 4 and5 and for the sensitivity analysis in Sup-plementary Figs. 6 and 7. Overall, funnelplot asymmetry improves with the sen-sitivityanalysiscomparedwiththemainanalysis for neonatal birth weight out-comes.

CONCLUSIONS

In this meta-analysis, we pooled resultsfrom 18 studies including 1,151 womenwith a variety of modified dietary inter-ventions. Remarkably, this is the firstmeta-analysis with a comprehensiveanalysisonmaternalglucoseparameters.Despite the heterogeneity between stud-ies, we found a moderate effect of dietaryinterventions on maternal glycemic out-comes, including changes in fasting, post-breakfast, and postprandial glucose levelsand need for medication treatment, andon neonatal birth weight. After removal offour studies with methodological con-cerns, we saw an attenuation of the treat-ment effect. Nonetheless, the change inpost-breakfast and postprandial glucoselevels and lowering of infant birth weight

Table 2—Continued


Ethnic diet (34) 1 20 2370.00 [2928.87, 188.87] d

Fiber (33) 1 22 294.00 [2446.68, 258.68] d

RR [95% CI]

Large for gestational age All 8 647 0.96 [0.63, 1.46] 0Low GI (13,26,29) 3 193 1.33 [0.54, 3.31] 0Low carbohydrate (31) 1 149 0.51 [0.13, 1.95] d

Energy restriction (32) 1 123 1.17 [0.65, 2.12] dSoy protein (14) 1 63 0.45 [0.04, 4.76] d

Behavior (24) 1 99 0.73 [0.25, 2.14] d

Ethnic diet (34) 1 20 0.14 [0.01, 2.45] d

Macrosomia All 12 834 0.49 [0.27, 0.88] 11Low GI (13,26,29,30) 4 276 0.46 [0.15, 1.46] 0DASH (23,36) 2 85 0.12 [0.03, 0.51] 0Low carbohydrate (25,31) 2 179 0.20 [0.02, 1.69] d

Energy restriction (32) 1 122 1.56 [0.61, 3.94] dFat modification (35) 1 84 0.35 [0.04, 3.23] d

Soy protein (27) 1 68 0.60 [0.16, 2.31] d

Ethnic diet (34) 1 20 0.20 [0.01, 3.70] d














remained significant. Given the inconsis-tencies between the main and sensitivityanalyses, we consider that conclusionsshould be drawn from the latter. Thesedata suggest that dietary interventionsmodified above and beyond usual dietaryadviceforGDMhavethepotentialtoofferbetter maternal glycemic control andinfant birth weight outcomes. However,the quality of evidencewas judged as lowto very low due to the limitations in thedesign of included studies, the inconsis-tency between their results, and theimprecision in their effect estimates.

Previous systematic reviews havefocused on the easier-to-quantify out-comes, such as the decision to startadditional pharmacotherapy and glucose-related variables at follow-up, but did notaddress change from baseline (15–17). Themost recently published Cochrane sys-tematic review by Han et al. (17) did notfind any clear evidence of benefit otherthan a possible reduction in cesareansection associated with DASH diet.The very high-carbohydrate intake (;400g/day) and 12 servings of fruit andvegetables in the DASH diet (22,23,36)

limit its clinical applicability and general-izability to women from lower socioeco-nomic, inner city backgrounds inWesterncountries. The Cochrane review sharedone of our primary outcomes, largefor gestational age (17). Neither meta-analysis detected a significant differ-ence in risk of large for gestational agebecause the trialswith a larger effect onbirth weight (the three DASH studies)did not report on large for gestationalage.

Our findings regarding pooled analy-sis of low-GI dietary interventions are

Figure 1—Forest plot of birth weight for modified dietary interventions compared with control diets in women with GDM. Reference citationsfor studies can be found in Table 1. CHO, carbohydrate; IV, inverse variance.



Table 3—Sensitivity analysis of primary maternal glycemic and infant birth weight outcomes


Maternal glycemic outcomes

Mean [95% CI]

Change in fasting glucose (mg/dL) All diets 10 575 21.98 [25.41, 1.45] 74Low GI (26,29,30) 3 195 25.33 [26.91, 23.76] 0DASH 0 0 Not estimable dLow carbohydrate (12,25) 2 42 3.66 [24.42, 11.73] 57Fat modification (28,35) 2 109 4.88 [21.45, 11.21] 0Soy protein (14,27) 2 130 27.51 [220.31, 5.30] 90Behavior (24) 1 99 21.50 [26.47, 3.47] dEthnic 0 0 Not estimable d

Change in postprandial glucose (mg/dL) All diets 7 421 25.90 [27.93, 23.88] 0Low GI (26,30) 2 121 27.08 [212.07, 22.08] 4DASH 0 0 Not estimable dLow carbohydrate (25) 1 30 23.00 [28.15, 2.15] d

Fat modification (28,35) 2 109 24.85 [213.32, 3.62] 40Soy protein (14) 1 62 21.05 [29.73, 7.63] d

Behavior (24) 1 99 26.90 [29.85, 23.95] dEthnic 0 0 Not estimable d

Change in post-breakfast glucose (mg/dL) All diets 3 175 24.76 [29.13, 20.38] 34Low GI (30) 1 83 28.6 [214.11, 23.09] d

Low carbohydrate (25) 1 30 23.00 [28.15, 2.15] dSoy protein (14) 1 62 21.05 [29.73, 7.63] d

Change in post-lunch glucose (mg/dL) All diets 2 92 4.50 [21.90, 10.90] 0Low carbohydrate (25) 1 30 4.00 [24.56, 12.56] d

Soy protein (14) 1 62 5.14 [24.51, 14.79] d

Change in post-dinner glucose (mg/dL) All diets 2 92 1.81 [25.28, 8.90] 0Low carbohydrate (25) 1 30 1.00 [28.14, 10.14] d

Soy protein (14) 1 62 3.03 [28.20, 14.26] d

Change in HOMA-IR (mIU/mL 3 mmol/L) All diets 3 179 20.74 [22.09, 0.61] 75DASH 0 0 Not estimable dLow carbohydrate (12) 1 12 0.60 [21.90, 3.10] d

Soy protein (27) 1 68 22.00 [23.17, 20.83] d

Behavior (24) 1 99 20.30 [20.71, 0.11] d

Change in HbA1c (%) All diets 5 353 20.03 [20.11, 0.05] 87Low GI (29,30) 2 167 0.01 [20.02, 0.03] 0DASH 0 0 Not estimable d

Fat modification (28) 1 25 0.10 [20.14, 0.34] d

Soy protein (14) 1 62 20.01 [20.07, 0.05] dBehavior (24) 1 99 20.19 [20.26, 20.12] d

Ethnic diet 0 0 Not estimable d

RR [95% CI]

Medication treatment All diets 11 884 0.82 [0.65, 1.04] 24Low GI (13,26,29,30) 4 293 0.80 [0.55, 1.14] 34DASH 0 0 Not estimable d

Low carbohydrate (31) 1 150 1.00 [0.75, 1.34] d


Fat modification (35) 1 84 Not estimable dSoy protein (14,27) 2 130 0.44 [0.21, 0.91] 0Behavior (24) 1 99 0.61 [0.15, 2.42] d

Ethnic 0 0 Not estimable dFiber (33) 1 11 Not estimable d

Infant birth weight outcomes

Mean [95% CI]

Birth weight (g) All diets 12 702 274.88 [2144.86, 24.90] 1

Low GI (13,26,29,30) 4 276 254.25 [2178.98, 70.47] 0

DASH 0 0 Not estimable d

Low carbohydrate (12,25) 2 42 57.73 [2164.93, 280.39] 0


Fat modification (28,35) 2 109 2139.61 [2294.80, 15.58] 0

Soy protein (14,27) 2 131 2184.67 [2319.35, 249.98] 0

Continued on p. 1359


broadly consistent with those of Vianaet al. (16) and Wei et al. (15). Viana et al.(16) noted decreased birth weight andinsulin use based on four studies of low-GIdiet among 257 women (mean difference2161.9 g [95%CI2246.4,277.4] and RR0.767[95%CI0.597,0.986], respectively).Wei et al. (15) also reported decreasedrisk of macrosomia with a low-GI diet infive studies of 302 women (RR 0.27 [95%CI 0.10, 0.71]). In our analyses of fourstudies in a comparable number of par-ticipants (n = 276), we found the samedirection of these effect estimates, with-out significant between-group differen-ces. This ismost likely due to thedifferentstudies included. For example, we wereunable to obtain effect estimates strat-ified by type of diabetes in the study byPerichart-Perera et al. (which includedwomen with type 2 diabetes) and there-fore did not include this study (37). Animportant difference between our anal-yses and that ofWei et al. (15) is that theyincluded DASH diet as a low-GI dietarysubtype.We also included a recent studyby Ma et al. (30) not included by theprevious reviews.Our sensitivity analyses highlighted con-

cerns regarding some studies included inprevious reviews. Notably, after removalof the studies with the most substantialmethodological concerns in the sensitiv-ity analysis, differences in the change infasting plasma glucose were no longersignificant. Although differences in thechange in postprandial glucose and birthweight persisted, they were attenuated.

This review highlights limitations ofthe current literature examining dietaryinterventions in GDM. Most studies aretoo small to demonstrate significant dif-ferences in our primary outcomes. Sevenstudies had fewer than 50 participantsand only two had more than 100 partic-ipants (n = 125 and 150). The shortduration of many dietary interventionsand the late gestational age atwhich theywere started (38) may also have limitedtheir impactonglycemicandbirthweightoutcomes. Furthermore, we cannot con-clude if the improvements in maternalglycemia and infant birth weight are dueto reduced energy intake, improved nu-trient quality, or specific changes in typesof carbohydrate and/or protein.

We have not addressed the indirectmodifications of nutrients. For example,reducing intake of dietary carbohydratestodecreasepostprandial glucosemaybecompensated by a higher consumptionof fat potentially leading to adverse ef-fects on maternal insulin resistance andfetal body composition. Beneficial oradverse effects of other nutrients such asn-3 long-chain polyunsaturated fatty acid,vitamin D, iron, and selenium cannot beruled out.

Our study has important strengths andweakness. To our knowledge, ours is thefirst systematic review of dietary inter-ventions inGDMcomprehensively exam-ining the impact of diet on maternalglycemic outcomes assessing the changein fasting and postprandial glucose,HbA1c, and HOMA-IR from baseline.

This is especially important given thatgroups were not well balanced at base-line. Our review also benefits from therigorousmethodology used aswell as thescientific, nutritional, and clinical exper-tise from an international interdisciplin-arypanel.However, italsohaslimitations.Baseline differences between groups inpostprandial glucose may have influ-enced glucose-related outcomes. Fur-thermore, three of the included trialswere pilot studies and therefore notdesigned to find between-group differ-ences (12,26,34). The low number ofstudies reporting on adherence clearlyillustrates that the quality of the evi-dence is far fromideal. Theheterogeneityof the dietary interventions even withina specific type (varied macronutrientratios, unknown micronutrient intake,and short length of some dietary inter-ventions) and baseline characteristics ofwomen included (such as prepregnancyBMI or ethnicity) may have also affectedourpooledresults. It shouldalsobenotedthat the relatively small numbers of studyparticipants limit between-diet compar-isons. Last, we were unable to resolvequeries regarding potential concerns forsources of bias because of lack of authorresponse to our queries. We have ad-dressed this by excluding these studiesin the sensitivity analysis.

Modified dietary interventions favor-ably influenced outcomes related to ma-ternal glycemia and birth weight. Thisindicates that there is room for improve-ment in usual dietary advice for women

Table 3—Continued



Fiber (33) 1 22 294.00 [2446.68, 258.68] d

RR [95% CI]

Large for gestational age All diets 7 627 1.00 [0.66, 1.53] 0Low GI (13,26,29) 3 193 1.33 [0.54, 3.31] 0Low carbohydrate (31) 1 149 0.51 [0.13, 1.95] d

Energy restriction (32) 1 123 1.17 [0.65, 2.12] dSoy protein (14) 1 63 0.45 [0.04, 4.76] d

Behavior (24) 1 99 0.73 [0.25, 2.14] d


Macrosomia All 9 729 0.73 [0.40, 1.31] 0Low GI (13,26,29,30) 4 276 0.46 [0.15, 1.46] 0DASH 0 0 Not estimable 0Low carbohydrate (25,31) 2 179 0.20 [0.02, 1.69] d

Energy restriction (32) 1 122 1.56 [0.61, 3.94] dFat modification (35) 1 84 0.35 [0.04, 3.23] d

Soy protein (27) 1 68 0.60 [0.16, 2.31] d




with GDM. Although the quality ofthe evidence in the scientific literatureis low, our review highlights the key roleof nutrition in the management of GDMand the potential for improvement ifbetter recommendations based on ad-equately powered high-quality studieswere developed. Given the prevalenceof GDM, new studies designed to eval-uate potential dietary interventions forthese women should be based in largerstudy groups with appropriate statisti-cal power. As most women with GDMare entering pregnancy with a high BMI,evidence-based recommendations re-garding both dietary components andtotal energy intake are particularly impor-tant for overweight and obese women.The evaluation of nutrient quality, in ad-dition to their quantity, as well as dietarypatterns such as Mediterranean diet (39)wouldalsoberelevant. Inparticular, thereisanurgentneedforwell-designeddietaryintervention studies in the low- and mid-dle-income countries where the globalhealth consequences of GDM are greatest.

Funding.H.R.M.wasfundedbytheU.K.NationalInstitute for Health Research (CDF 2013-06-035).This work was conducted by an expert group ofthe European branch of the International LifeSciences Institute (ISLI Europe). This publicationwas coordinated by the ISLI Europe Early Nutri-tion and Long-Term Health and the Obesity andDiabetes task forces. Industry members of thesetask forcesare listedonthe ILSI Europewebsiteatwww.ilsi.eu. Experts are not paid for the timespent on this work; however, the nonindustrymembers within the expert group were offeredsupport for travel and accommodation costsfrom the Early Nutrition and Long-Term Healthand the Obesity and Diabetes task forces to at-tend meetings to discuss the manuscript and asmall compensatory sum (honoraria) with theoption to decline. The expert group carried outthe work, i.e. collecting and analyzing data andinformation and writing the scientific paper,separate to other activities of the task forces.The research reported is the result of a scientificevaluation in line with ILSI Europe’s frameworkto provide a precompetitive setting for public-private partnership. ILSI Europe facilitated sci-entific meetings and coordinated the overallproject management and administrative tasksrelating to the completion of this work.The opinions expressed herein and the con-

clusions of this publication are those of theauthors and do not necessarily represent theviews of ILSI Europe nor those of its membercompanies. For further information about ILSIEurope, please email [email protected] or call+32 2 771 00 14.Duality of Interest. E.M.v.d.B. works part-timefor Nutricia Research. E.C.-G. works full-time forNestec. R.R. works full-time for Abbott Nutrition.

No potential conflicts of interest relevant to thisarticle were reported.Author Contributions. J.M.Y. contributed todata extraction, statistical analyses, and writingthe first draft manuscript. J.E.K. contributed todata extraction and writing the first draft sum-mary tables. M.B. and A.G.-P. contributed toliterature extraction, statistics, and manuscriptrevision. E.H. contributed to data extraction andGRADE assessments. I.S. and I.G. contributed tostatistics and manuscript revision. E.M.v.d.B.,E.C.-G., S.H., and S.F.O. contributed to conceptand design, data extraction, and manuscriptreview. M.H. contributed to concept and designanddraftmanuscriptevaluation.K.L. contributedto concept and design, data extraction, andcritical review for intellectual content. L.P. con-tributed to concept and design and manuscriptreview. R.R., P.R., and H.R.M. contributed toconcept and design, data extraction, and revisingthe draft manuscript. L.v.L. contributed to dataextraction and draft summary tables. B.S. con-tributed to data extraction and critical reviewfor intellectual content. R.C. contributed to lit-erature extraction, statistical analyses, and re-vising the draftmanuscript. R.C. is the guarantorof this work and, as such, had full access to allthe data in the study and takes responsibilityfor the integrity of the data and the accuracy ofthe data analysis.Prior Presentation. Parts of this work werepresentedattheDiabetesUKNationalDiabetesinPregnancy Conference, Leeds, U.K., 14 Novem-ber 2017, and the XXIX National Congress ofthe Spanish Society of Diabetes, Oviedo, Spain,18–20 April 2018.

References1. Guariguata L, Linnenkamp U, Beagley J,Whiting DR, Cho NH. Global estimates of theprevalence of hyperglycaemia in pregnancy. Di-abetes Res Clin Pract 2014;103:176–1852. Landon MB, Spong CY, Thom E, et al.; EuniceKennedyShriverNational InstituteofChildHealthand Human Development Maternal-Fetal Med-icine Units Network. A multicenter, randomizedtrial of treatment formild gestational diabetes. NEngl J Med 2009;361:1339–13483. Metzger BE, Lowe LP, Dyer AR, et al.; HAPOStudy Cooperative Research Group. Hyperglyce-mia and adverse pregnancy outcomes. N Engl JMed 2008;358:1991–20024. Silverman BL, Rizzo TA, Cho NH, Metzger BE;The Northwestern University Diabetes in Preg-nancy Center. Long-term effects of the intrauter-ine environment. Diabetes Care 1998;21(Suppl.2):B142–B1495. Bellamy L, Casas JP, Hingorani AD,Williams D.Type 2 diabetes mellitus after gestational di-abetes: a systematic review and meta-analysis.Lancet 2009;373:1773–17796. Boney CM, Verma A, Tucker R, Vohr BR.Metabolic syndrome in childhood: associationwith birth weight, maternal obesity, and gesta-tional diabetes mellitus. Pediatrics 2005;115:e290–e2967. Koletzko B, Brands B, ChourdakisM, et al. ThePower of Programming and the EarlyNutritionproject: opportunities for health promotionby nutrition during the first thousand days oflife and beyond. Ann Nutr Metab 2014;64:187–196

8. AmericanDiabetesAssociation.Managementof diabetes in pregnancy. Sec. 13. In Standardsof Medical Care in Diabetesd2017. DiabetesCare 2017;40(Suppl. 1):S114–S1199. Metzger BE, Buchanan TA, Coustan DR, et al.Summary and recommendations of the FifthInternational Workshop-Conference on Gesta-tional Diabetes Mellitus. Diabetes Care 2007;30(Suppl. 2):S251–S26010. National Institute for Health and Care Ex-cellence. Diabetes in pregnancy: managementfrom preconception to the postnatal period[article online], 25 February 2015. Availablefrom https://www.nice.org.uk/guidance/ng3.Accessed 2 November 201711. Thompson D, Berger H, Feig D, et al.; Cana-dianDiabetesAssociationClinical PracticeGuide-linesExpertCommittee.Diabetesandpregnancy.Can J Diabetes 2013;37(Suppl. 1):S168–S18312. Hernandez TL, Pelt RE, Anderson MA, et al.Women with gestational diabetes mellitus ran-domized to a higher-complex carbohydrate/low-fat diet manifest lower adipose tissue insulinresistance, inflammation, glucose, and free fattyacids: a pilot study.Diabetes Care2016;39:39–4213. Moses RG, Barker M, Winter M, Petocz P,Brand-Miller JC. Can a low-glycemic index dietreduce the need for insulin in gestational dia-betesmellitus?A randomized trial.DiabetesCare2009;32:996–100014. Sarathi V, Kolly A, Chaithanya Hulivana B,DwarakanathChinthamaniS. Effectof soyabasedprotein rich diet on glycaemic parameters andthyroid function tests in womenwith gestationaldiabetes mellitus. Romanian Journal of DiabetesNutrition and Metabolic Diseases 2016;23:201–20815. Wei J, HengW, Gao J. Effects of low glycemicindex diets on gestational diabetes mellitus:ameta-analysis of randomized controlled clinicaltrials. Medicine (Baltimore) 2016;95:e379216. Viana LV, Gross JL, Azevedo MJ. Dietaryintervention inpatientswithgestationaldiabetesmellitus: a systematic review and meta-analysisof randomized clinical trials on maternal and new-born outcomes. Diabetes Care 2014;37:3345–335517. Han S,Middleton P, Shepherd E, Van RyswykE, Crowther CA. Different types of dietary advicefor women with gestational diabetes mellitus.Cochrane Database Syst Rev 2017;2:CD00927518. Moher D, Liberati A, Tetzlaff J, Altman DG;PRISMA Group. Preferred reporting items forsystematic reviews and meta-analyses: thePRISMA statement. Ann Intern Med 2009;151:264–26919. AkbariM,MosazadehM, Lankarani KB, et al.The effects of vitamin D supplementation onglucose metabolism and lipid profiles in patientswith gestational diabetes: a systematic reviewand meta-analysis of randomized controlledtrials. Horm Metab Res 2017;49:647–65320. Taylor BL, Woodfall GE, Sheedy KE, et al.Effect of probiotics on metabolic outcomes inpregnantwomenwithgestationaldiabetes:asys-tematic review andmeta-analysis of randomizedcontrolled trials. Nutrients 2017;9:1–1321. Higgins JPT, Green S (Eds). Cochrane Hand-book for Systematic Reviews of Interventions,version 5.1.0 [updated March 2011]. The Co-chraneCollaboration,2011.Available fromhttp://handbook.cochrane.org/. Accessed 2 November2017


http://handbook.cochrane.org/

http://handbook.cochrane.org/

22. Asemi Z, Tabassi Z, SamimiM, Fahiminejad T,Esmaillzadeh A. Favourable effects of theDietaryApproaches toStopHypertensiondietonglucosetolerance and lipid profiles in gestational di-abetes: a randomised clinical trial. Br J Nutr2013;109:2024–203023. AsemiZ,SamimiM,TabassiZ,EsmaillzadehA.The effect of DASH diet on pregnancy outcomesin gestational diabetes: a randomized controlledclinical trial. Eur J Clin Nutr 2014;68:490–49524. Bo S, Rosato R, Ciccone G, et al. Simplelifestyle recommendations and the outcomes ofgestational diabetes. A 23 2 factorial randomizedtrial. Diabetes Obes Metab 2014;16:1032–103525. Cypryk K, Kaminska P, KosinskiM, Pertynska-Marczewska M, Lewinski A. A comparison of theeffectiveness, tolerability and safety of high andlow carbohydrate diets in women with gesta-tional diabetes. Endokrynol Pol 2007;58:314–31926. Grant SM, Wolever TMS, O’Connor DL,NisenbaumR, Josse RG. Effect of a low glycaemicindex diet on blood glucose in women withgestational hyperglycaemia. Diabetes Res ClinPract 2011;91:15–2227. Jamilian M, Asemi Z. The effect of soy intakeon metabolic profiles of women with gestationaldiabetes mellitus. J Clin Endocrinol Metab 2015;100:4654–466128. Lauszus FF, Rasmussen OW, Henriksen JE,et al. Effect of a highmonounsaturated fatty aciddietonbloodpressureandglucosemetabolisminwomen with gestational diabetes mellitus. Eur JClin Nutr 2001;55:436–44329. Louie JC, Markovic TP, Perera N, et al. Arandomized controlled trial investigating theeffects of a low-glycemic index diet onpregnancyoutcomes in gestational diabetes mellitus. Di-abetes Care 2011;34:2341–234630. Ma W-J, Huang Z-H, Huang B-X, et al. In-tensive low-glycaemic-load dietary interventionfor the management of glycaemia and serumlipids among women with gestational diabetes:a randomized control trial. Public Health Nutr2015;18:1506–151331. Moreno-Castilla C, HernandezM, BerguaM,etal. Low-carbohydratediet for the treatmentof

gestational diabetes mellitus: a randomizedcontrolled trial. Diabetes Care 2013;36:2233–223832. RaeA,BondD, EvansS,NorthF, RobermanB,WaltersB.Arandomisedcontrolledtrialofdietaryenergy restriction in the management of obesewomen with gestational diabetes. Aust N Z JObstet Gynaecol 2000;40:416–42233. Reece EA, Hagay Z, Gay LJ, O’Connor T,DeGennaro N, Homko CJ. A randomized clinicaltrialofafiber-enricheddiabeticdietvs. thestandardAmerican Diabetes Association-recommended dietin the management of diabetes mellitus in preg-nancy. J Maternal Fetal Invest 1995,1:8–1234. Valentini R, DalfraMG,MasinM, et al. A pilotstudy on dietary approaches in multiethnicity:two methods compared. Int J Endocrinol 2012;2012:98513635. WangH, JiangH, Yang L, ZhangM. Impacts ofdietary fat changes on pregnant women withgestational diabetes mellitus: a randomizedcontrolled study. Asia Pac J Clin Nutr 2015;24:58–6436. YaoJ,CongL,ZhuBL,WangT.Effectofdietaryapproaches to stop hypertension diet plan onpregnancy outcome patients with gestationaldiabetes mellitus. Bangladesh J Pharmacol 2015;10:732–73837. Perichart-PereraO,Balas-NakashM,Rodriguez-CanoA, Legorreta-Legorreta J, Parra-CovarrubiasA, Vadillo-Ortega F. Low glycemic index carbo-hydrates versus all types of carbohydrates fortreating diabetes in pregnancy: a randomizedclinical trial to evaluate the effect of glycemiccontrol. Int J Endocrinol 2012;2012:29601738. Sovio U, Murphy HR, Smith GC. Acceleratedfetal growth prior to diagnosis of gestationaldiabetes mellitus: a prospective cohort study ofnulliparouswomen. Diabetes Care 2016;39:982–98739. Esposito K, Maiorino MI, Ciotola M, et al.Effects of aMediterranean-style diet on the needfor antihyperglycemic drug therapy in patientswith newly diagnosed type2 diabetes: a random-ized trial. Ann Intern Med 2009;151:306–31440. Canadian Diabetes Association Clinical Prac-tice Guidelines Expert Committee. Canadian

Diabetes Association 2008 clinical practiceguidelines for the prevention and managementof diabetes in Canada. Can J Diabetes 2008;32:S1–S20141. Hoffman L, Nolan C, Wilson JD, Oats JJ,Simmons D; The Australasian Diabetes in Preg-nancy Society. Gestational diabetes mellitus–management guidelines. Med J Aust 1998;169:93–9742. Obsetrics Unit of Obstetrics and GynecologyBranch, ChineseMedical Association. Guidelinesfor recommendations of clinical diagnosis andtreatment for GDM. Zhongguo Shiyong Fuke YuChanke Zazhi 2007;23:475–47743. American College of Obstetricians and Gy-necologists. Gestational diabetes mellitus. Prac-tice bulletin no. 137. Obstet Gynecol 2013;122:406–41644. Ricart W, Lopez J, Mozas J, et al.; SpanishGroup for the Study of the Impact of Carpen-ter and Coustan GDM thresholds. Potentialimpact of American Diabetes Association (2000)criteria for diagnosis of gestational diabetesmellitus in Spain. Diabetologia 2005;48:1135–114145. Metzger BE. Summary and recommendationsof the Third International Workshop-Conferenceon Gestational Diabetes Mellitus. Diabetes 1991;40(Suppl. 2):197–20146. American Diabetes Association. Diagnosisand classification of diabetes mellitus. DiabetesCare 2014;37(Suppl. 1):S81–S9047. Metzger BE, Gabbe SG, Persson B, et al.;International Association of Diabetes and Preg-nancy Study Groups Consensus Panel. Interna-tional Association of Diabetes and PregnancyStudyGroups recommendationson thediagnosisand classification of hyperglycemia in pregnancy.Diabetes Care 2010;33:676–68248. Martin FI; Ad Hoc Working Party. The di-agnosis of gestational diabetes.Med J Aust 1991;155:11249. Metzger BE, Coustan DR; The OrganizingCommittee. Summary and recommendations ofthe Fourth International Workshop-Conferenceon Gestational Diabetes Mellitus. Diabetes Care1998;21(Suppl. 2):B161–B167



gestational diabetes mellitus and diet: a …...gestational diabetes mellitus (gdm) is one of the...

Documents