Dietary Intervention in Patients With Gestational Diabetes Mellitus: A Systematic Review and Meta-analysis of Randomized Clinical Trials on Maternal and Newborn Outcomes
Abstract
OBJECTIVE Diet is the cornerstone treatment of patients with gestational diabetes mellitus (GDM), but its role in maternal and newborn outcomes has been scarcely studied. The purpose of this study was to analyze the efficacy of dietary interventions on maternal or newborn outcomes in patients with GDM.
RESEARCH DESIGN AND METHODS A systematic review and meta-analysis of randomized clinical trials (RCTs) of dietary intervention in GDM or pregnancy with hyperglycemia was performed. MEDLINE, Embase, ClinicalTrials.gov, Cochrane, and Scopus were searched through to March 2014. The main evaluated maternal outcomes were proportion of patients using insulin and proportion of cesarean delivery; the newborn outcomes were proportion of macrosomia and hypoglycemia and newborn weight.
RESULTS From 1,170 studies, nine RCTs, including 884 women aged 31.5 years (28.7–33.2) with 27.4 weeks (24.1–30.3) of gestation, were eligible. We divided the RCTs according to the type of dietary intervention: low glycemic index (GI) (n = 4; 257 patients), total energy restriction (n = 2; 425 patients), low carbohydrates (n = 2; 182 patients), and others (n = 1; 20 patients). Diet with low GI reduced the proportion of patients who used insulin (relative risk 0.767 [95% CI 0.597, 0.986]; P = 0.039) and the newborn birth weight (weight mean differences −161.9 g [95% CI −246.4, −77.4]; P = 0.000) as compared with control diet. Total restriction and low carbohydrate diets did not change either maternal or newborn outcomes.
CONCLUSIONS A low GI diet was associated with less frequent insulin use and lower birth weight than control diets, suggesting that it is the most appropriate dietary intervention to be prescribed to patients with GDM.
Introduction
Gestational diabetes mellitus (GDM), defined as hyperglycemia diagnosed at pregnancy, has been associated with many adverse maternal and newborn outcomes (1), especially increased number of cesarean deliveries, newborns large for gestational age, and macrosomia (2). According to diagnostic criteria (3), the prevalence of GDM ranges from 1.7 to 11.6% (4,5). The prevalence of GDM could be as high as 18% in some regions if the criteria of the International Association of Diabetes and Pregnancy Study Groups Consensus Panel are used (6).
Recent systematic reviews (7,8) reinforce that the treatment of GDM is effective in reducing specific adverse maternal and newborn outcomes without evidence of short-term harm (7). Dietary intervention was simultaneously evaluated with the use of insulin as needed in both systematic reviews. Although dietary therapy is considered the cornerstone treatment for GDM, data on diet intervention as the sole GDM treatment are limited and its actual role in maternal and newborn outcomes has scarcely been studied (9). Moreover, in clinical practice, most of the dietary recommendations for patients with GDM have been based mainly on glucose control, through glucose monitoring data, instead of being based on data from hard maternal or newborn outcomes (10,11).
The recent recommendations of the guidelines of the Endocrine Society for diabetes and pregnancy to attain desired glycemic goals are based on the reduction of carbohydrate intake only to reach 35–45% of total daily energy with or without calorie restriction (11). Other dietary aspects, such as modification of the proportion and quality of dietary macronutrients, were not taken into account, although there is some evidence of beneficial effect of these interventions in both maternal and newborn outcomes (12–17). Therefore, the aim of this study was to analyze the efficacy of dietary intervention in patients with GDM in maternal and newborn outcomes by systematic review with meta-analysis of randomized clinical trials (RCTs).
Research Design and Methods
This systematic review was carried out using a protocol constructed according to the Cochrane Handbook recommendations (18) and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (19).
Data Sources and Searches
We searched databases from MEDLINE, Embase, ClinicalTrials.gov register, Cochrane, and Scopus to identify RCTs that reported dietary intervention and hyperglycemia in pregnancy or GDM and reported maternal and newborn outcomes, through March 2014. In addition, we also searched all published abstracts from the American Diabetes Association (ADA) and European Association for the Study of Diabetes annual meetings for the last 5 years.
The initial search comprised the MeSH terms “Diet” [MeSH], “Pregnancy”[MeSH], “Diabetes, Gestational” [MeSH], or glucose intolerance and related entry terms associated with a high-sensitivity strategy for the search of RCTs available at http://www.sign.ac.uk/methodology/filters.html#random. All potentially eligible studies were considered for review, limited to the English, Spanish, or Portuguese language. A manual search was also performed in the reference lists of included articles and from recent reviews about the topic (20,21).
Study Selection
We included RCTs that evaluated the effect of dietary intervention on patients diagnosed with GDM or glucose intolerance or hyperglycemia during pregnancy with reported maternal and newborn outcomes. Dietary intervention must have lasted for at least 4 weeks and continued until delivery. We excluded studies if they were not randomized, they included patients with type 1 or type 2 diabetes, the dietary intervention was the same in all studied groups, the dietary characteristics were not available, or they did not report any of the outcomes of interest.
The maternal outcomes evaluated were as follows: weight gain, rate of cesarean sections, labor induction, gestational age at delivery, and number of patients who start using insulin. The newborn outcomes were as follows: weight, frequency of newborn with macrosomia (birth weight >4 kg), large for gestational age (birth weight >90th centile), or small for gestational age (SGA; birth weight <10th centile), prematurity, and birth trauma or hypoglycemia.
Data Extraction and Quality Assessment
All citations retrieved from electronic databases were imported to the EndNote program. Two reviewers (M.J.A. and L.V.V.) independently analyzed the titles and abstracts of every paper retrieved from the literature search to identify potentially eligible studies. All studies that did not meet the inclusion criteria were excluded. The full text of the remaining papers was obtained for further examination. Disagreements were solved by a third reviewer (J.L.G.).
The data of included studies were independently extracted by the same two reviewers using a standardized data extraction form. Extracted data included the following: first author’s name, year of publication, number of participants, details of the study design (i.e., randomization method), trial duration, and patient characteristics (age, BMI, gestational age at the diagnoses of GDM, and ethnicity [white or non-Caucasian]). Diet characteristics (total energy, macronutrients, and fiber content) and evaluation of dietary compliance were extracted from intervention and control diet descriptions. The dietary prescription or recommendation was extracted and, when available, we chose to present the actual intake data. Authors were contacted in order to obtain any relevant missing information, and personal contact allowed us to include additional data from two studies (12,14).
We divided the RCTs into four categories according to the type of dietary intervention: low glycemic index (GI), total energy restriction, low carbohydrate content, and others. The GI is an in vivo measure of the blood glucose response to a standard amount of carbohydrate from a food, relative to a reference food (glucose or white bread.) The GI value ranks foods on a scale from 0 to 100 according to the extent to which they raise blood glucose after eating. Foods with GI <55 have been considered low GI foods (22–24). The daily dietary GI is calculated based on the carbohydrate content per serving of each ingested food. Restriction of total energy was defined according to ADA recommendations for medical nutrition therapy in GDM: modest caloric restriction consisting of 1,600–1,800 kcal or ∼33% reduction in caloric intake (10). Consumption of carbohydrates represents 45–65% of total daily calories according to the dietary reference intakes for the general population (25); low carbohydrate diet was characterized by a carbohydrate intake lower than 45% of energy, without restricting the total energy. The fourth category was created to include any study identified in the database search that fulfilled our inclusion criteria of this systematic review but did not meet traditional dietary interventions, such as low GI, energy restriction, or low carbohydrate diets.
In the evaluation of quality assessment, two reviewers (M.J.A. and L.V.V.) independently assessed the methodological quality of studies. We used a score based on the Cochrane Collaboration tool for assessing risk of bias of every included study and GRADE score (Cochrane, GRADE) for each meta-analysis.
According to the Cochrane Collaboration, biases were classified into six domains: selection, performance, detection, attrition, reporting, and other (18,26). The “other” chosen domain was the assessment of dietary compliance. The risk of bias for each domain was classified as high, low, or unclear. Regarding dietary compliance, the risk was classified as “low” if the study described the method of dietary compliance evaluation.
The quality of the body of evidence of each meta-analysis was assessed by the GRADE approach (27), including factors that may decrease (e.g., methodological quality, directness of evidence, heterogeneity, precision of effect estimates, and risk of publication bias) or increase (e.g., large magnitude of effect, reduction or spurious effect due to plausible confounding factors, and dose-response gradient) the quality of evidence. Each evaluated factor was rated as high, moderate, low, or very low (27,28). Using this approach, we considered a serious risk of bias when an individual study had more than three unclear or high risk of bias, and imprecision was defined as a meta-analysis confidence interval >0.5.
Data Synthesis and Analysis
Descriptive data from the systematic review were presented as mean and/or range, when available. Changes in continuous data were reported as absolute differences between arithmetic means at baseline and end of study, and weight mean differences were used in the analyses. Relative risk (RR) was evaluated for binary data.
The heterogeneity between the studies was evaluated by Cochran Q test, and a P for trend ≤0.10 was considered statistically significant. The I2 test was also performed to evaluate the magnitude of heterogeneity. Possible publication bias was assessed using a contour-enhanced funnel plot of each trial’s effect size against the standard error. Funnel plot asymmetry was evaluated by Begg and Egger tests, and a significant publication bias was considered if the P value was <0.10.
All statistical analyses were performed using Stata 11.0 software (Stata, College Station, TX). Significance was set at P < 0.05, and 95% CIs are quoted throughout.
Results
Literature Search
We identified 1,170 studies in database searches (Fig. 1). Of these, 1,121 were excluded based on title and abstract, leaving 49 articles for further full text evaluation. Forty of these studies were excluded, most of them due to a nonrandomized design, not reporting the outcomes of interest, diet not being the primary intervention, being a subanalysis of a major study, not providing the dietary characteristics, or if the patients were not followed until delivery. Therefore, we selected nine trials to be included in the current systematic review.
Flow diagram of literature search to identify RCTs evaluating the effect of diet in patients with GDM.
Study Characteristics
The characteristics of the nine included trials are summarized in Table 1. A total of 884 pregnant women aged 28.7–33.2 years (mean 31.5 years) who had a diagnosis of GDM between 24.1 and 30.3 weeks of pregnancy (mean 27.4 weeks) were included. The diagnosis of GDM was established by different criteria (29–37) in the RCTs included in this systematic review (Supplementary Table 1).
Characteristics of the studies according to dietary intervention in GDM
The available data from the reviewed RCTs allowed us to perform meta-analyses of low GI, total energy restriction, and low carbohydrate diets (Fig. 2). Only one study was included in the fourth category of dietary intervention.
Forest plots of dietary intervention for patients with GDM: low GI diet (A), energy restriction diet (B), and low carbohydrate diet (C).
Low GI Diet
Low GI diet analyses included four studies (12–15) and 257 patients, aged 32.9 years, in whom the diagnosis of GDM was established at 26.8 weeks (24.1–30.1). Regarding diet characteristics, mean energy intake was similar in the intervention (1,675 kcal/day, range 1,477–1,834) and control (1,694 kcal/day, range 1,485–1,932) groups. The GI score in the low GI dietary intervention ranged from 47 to 49 (mean 48.9), whereas in the control group the range was 47 to 58 (mean 53.5). Dietary GI in all trials was calculated based on food composition tables and published GI values using the glucose = 100 scale. The daily total fiber intake was 27.5 g/day (25.6–30) in the intervention and 23.6 g/day (22.9–25) in the control groups.
Data from low GI trials allowed us to evaluate the maternal weight gain in the last visit of the study, frequency of cesarean section, insulin use, SGA and macrosomia, and newborn weight. Patients in the low GI diet group used insulin less frequently (RR 0.767 [95% CI 0.597, 0.986]; P = 0.039), and the newborn weight (MWD −161.9 g [95% CI −246.4, −77.4]; P = 0.000) was lower than those in the control group. However, there was no significant change in maternal weight gain (MWD −0.412 kg [−1.842, 1.017]; P = 0.428) or cesarean section rates (RR 1.045 [0.736, 1.483]; P = 0.286) or an increase in the number SGA newborns (RR 1.588 [0.603, 4.182]; P = 0.349) or with macrosomia (RR 0.479 [0.147, 1.561]; P = 0.222). The less frequent use of insulin means that 13 out of 100 patients with GDM will not need to use insulin if they adopt a low GI diet during pregnancy.
No heterogeneity was found for the analyses of maternal weight gain (I2 = 0%; P = 0.428), cesarean section rates (I2 = 20.2%; P = 0.286), insulin use (I2 = 34.0%; P = 0.208), newborn weight (I2 = 0%; P = 0.404), number of SGA newborns (I2 = 0%; P = 0.745), and macrosomia (I2 = 0%; P = 0.967) and also no publication bias as assessed by funnel plot and Egger test, except a possible publication bias for macrosomia analyses (Egger test, P = 0.043) (Supplementary Fig. 1).
Total Energy Restriction Diet
Total energy restriction diet analyses included two RCTs (38,39) with 425 patients aged 30.6 years (30.4–30.7) in whom GDM diagnosis was established at between 24 and 30 weeks of pregnancy. Energy restriction intervention was a calorie-restricted diet of 35 kcal/ideal body weight (kg)/day in one of the studies (38). In the other (39), a moderate restriction diet representing 70% of recommended dietary intake for women with GDM was adopted. This means a reduction of ∼30% total of energy intake.
In the total energy restriction category, data for cesarean rates, frequency of macrosomia, and neonatal hypoglycemia were available. Incomplete data about other newborn outcomes and maternal weight prevented us from further analyses. Total energy restriction intervention did not increase the number of cesarean sections (RR 1.091 [95% CI 0.769, 1.496]; P = 0.588), frequency of macrosomia (1.002 [0.649, 1.547]; P = 0.992), or neonatal hypoglycemia (1.014 [0.718, 1.434]; P = 0.936).
High heterogeneity in the studies was observed for rates of both cesarean section (I2 = 73.7%; P = 0.051) and neonatal hypoglycemia (I2 = 75.3%; P = 0.045) without publication bias as assessed by funnel plot. For the analyses of frequency of macrosomia, there was no heterogeneity (I2 = 26%; P = 0.245) or publication bias by funnel plot. Egger tests were not calculated due to the small number of included studies (Supplementary Table 1).
Low Carbohydrate Diet
Two studies of low carbohydrate diet (16,17) evaluated 182 patients aged 30.8 years (28.7–32.8) at 29.8 weeks (29.2–30.3) of pregnancy at GDM diagnoses. Total daily energy from carbohydrates ranged from 40 to 45% in the intervention group and from 55 to 60% in the control groups.
No difference was found regarding cesarean section rates (RR 1.182 [95% CI 0.764, 1.829]; P = 0.588) and frequency of insulin use (1.061 [0.796, 1.415]; P = 0.685) and macrosomia (0.346 [0.063, 1.912]; P = 0.992) in the low carbohydrate diet. Data on maternal weight gain and newborn outcomes were incomplete in those studies.
No heterogeneity was found in the analyses of the rate of cesarean section (I2 = 0%; P = 0.619) and the frequency of insulin use (I2 = 57.7%; P = 0.124) and macrosomia (I2 = 0%; P = 0.590). However, visual asymmetry was present for the frequency of insulin use and macrosomia as evaluated by funnel plot. Egger tests were not calculated due to the small number of included studies (Supplementary Fig. 1).
Others: Ethnic Diet
A small Italian study conducted in 20 pregnant women with GDM (40) compared an ethnic-based food choice diet (“ethnic meal plan”) with a diet recommended by ADA, both with the same energy and macronutrient composition. The intervention diets consisted of the adoption of food plans based on six ethnic groups (Chinese, Filipino, Moroccan, Nigerian, Romanian, and Bangladeshi). Typical dishes from the foreign women’s home country were chosen. The study did not demonstrate differences in cesarean section rates, maternal and fetal weight, insulin use, and macrosomia.
Meta-analyses Quality Evaluation
Individual quality of studies revealed a low risk of bias for most evaluated domains, except for concealment allocation in two studies (10,12). Most RCTs described some type of assessment of diet compliance (Supplementary Table 2).
The GRADE quality of evidence was from moderate to high for low GI meta-analyses, except for macrosomia. However, we graded the maternal weight gain meta-analyses as low quality. Actually, the individual data for pregnancy weight gain could have been inaccurate because this information could have been collected in different follow-up periods. GRADE quality was low for all evaluated maternal and newborn outcomes of energy restriction and low carbohydrate meta-analyses (Supplementary Table 3).
Conclusions
This systematic review of dietary intervention in GDM included 884 pregnant women from nine RCTs. We were able to perform three meta-analyses according to the type of dietary interventions used: low GI, total energy restriction, and low carbohydrate diets. Remarkably, only the low GI diet was associated with beneficial outcomes: less frequent insulin use and lower newborn weight than the control dietary interventions, without increasing the number of SGA newborns and macrosomia. These meta-analyses rated a moderate to high quality according to the GRADE profile. On the other hand, meta-analyses of energy restriction and low carbohydrate diets did not influence any of the studied outcomes.
Diet is the main treatment for GDM, and as far as we know, this is the first systematic review with meta-analyses that shows benefits of a specific dietary intervention in GDM management. Moreover, we evaluated these benefits based on hard outcomes, instead of on results related to glucose parameters only.
The last published systematic review from Cochrane on different dietary advice for women with GDM (20) could not find any significant dietary benefit. Unlike our results, the low GI did not influence insulin use and birth weight. It is worth noting that most of those results (birth weight, cesarean section, and SGA) were based on only one study analysis. Furthermore, as compared with ours, the authors included only one RCT in type 2 diabetes and GDM (41) and did not include two other studies (12,15) in their low GI meta-analyses. Although the results of energy restriction and low carbohydrate diets were similar to ours, the Cochrane review included only one RCT in each of these diet categories. Another recent systematic review about nutritional strategies for women with GDM (21) did not perform any meta-analyses.
The low GI diet was the only confirmed advantageous dietary intervention to be followed during pregnancy by women with GDM. According to our results, 13 out of 100 patients with GDM will not need to use insulin if they adopt a low GI diet during pregnancy and birth weight was lower but without increasing the SGA rates than the control diets. Beneficial effect on birth weight was already observed in non-GDM pregnant patients after a low GI diet (42). It has been shown that a low GI diet improves insulin sensitivity (43) and reduces the insulin requirement possibly due to its ability to reduce postprandial glucose excursions (44). Noteworthy, the GI in intervention groups was really low (47–49), which could be an indirect measure of dietary compliance in GDM patients.
Regarding macrosomia and maternal weight, the quality of our meta-analyses prevents us from arriving at a definitive conclusion about the effect of low GI diet in these outcomes. This could be related to a relatively small number of macrosomic newborns, both in the intervention (4/128) and in the control groups (8/122). On the other hand, in a trial that evaluated the effect of low GI in 800 pregnancies in non-GDM women at high risk for macrosomia (45), the adoption of a low GI diet also did not reduce fetal macrosomia. There is no reason to assume that patients with GDM would behave differently.
Another promising dietary intervention for GDM patients could be the Dietary Approaches to Stop Hypertension (DASH) diet. In a small trial, the DASH diet adopted for 4 weeks during pregnancy was able to reduce insulin use, cesarean rates, and birth weight in patients with GDM. We did not include that study in our systematic review because the dietary intervention was not continued until delivery (46). However it is important to note that the DASH diet can be considered a low GI dietary approach. This observation reinforces the role of a low GI diet for GDM women.
A possible limitation of our systematic review could be related to different diagnostic criteria for GDM used in the included studies. However, in our search, the terms hyperglycemia and pregnancy were both included. This strategy should have prevented missing any study on abnormal glucose homeostasis in pregnancy. This is an important aspect of our systematic review since it is well known that it is worthwhile to treat even mild glucose abnormalities in pregnancy (47). Regarding our main result, the low GI diet, a dose-response analysis would provide valuable information. However, a narrow GI interval (47–49) did not allow us to conduct it. Another potential weakness of our systematic review was the small number of studies included in each dietary intervention category analysis. It would be interesting to compare all included trials through a network meta-analysis. However, different dietary strategies could not be connected and we were unable to perform this analysis. Furthermore, in the included studies, the outcomes evaluated were not always the same, were not uniformly standardized, or were even not available. A low GRADE quality score for most performed meta-analyses outcomes (low GI: macrosomia and maternal weight; energy restriction diet: cesarean section, macrosomia, and neonatal hypoglycemia; low carbohydrate diet: cesarean section, insulin use, and macrosomia) could explain the lack of difference between intervention and control diets. Finally, the adherence of dietary interventions was not assessed or reported in three trials. All these aspects reinforce the need to perform well-designed RCTs on the effects of dietary intervention in patients with GDM.
In conclusion, we demonstrated that in patients with GDM, the use of a low GI diet was associated with less frequent insulin use and lower birth weight than the control diets, without any detected adverse effects. Therefore, the present available evidence suggests that a low GI diet is the most appropriate dietary intervention to be prescribed to patients with GDM.
Article Information
Funding. This study was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, and Fundo de Incentivo à Pesquisa Hospital de Clínicas de Porto Alegre.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. L.V.V. was engaged in conception and design, data extraction, statistical analyses, interpretation of data, and drafting of the manuscript. J.L.G. was responsible for conception, design, and critical revision of the manuscript. M.J.A. was responsible for conception, design, data extraction, statistical analyses, interpretation of data, drafting of the manuscript, and administrative and technical support.
Footnotes
This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc14-1530/-/DC1.
- Received June 23, 2014.
- Accepted August 21, 2014.
- © 2014 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.
References
In this Issue
