Literature search and study selection

A comprehensive search of the relevant literature was performed in electronic databases (MEDLINE, EMBASE and the Cochrane Library) from inception to March 2014. The search strategy was designed by combining the search terms: “diet,” “vitamins,” “probiotics,” “gestational,” “diabetes” and “pregnancy” using their word variants and Boolean operators AND and OR as appropriate. No language restrictions were applied. We contacted the authors of primary studies to obtain any relevant unpublished data. Additionally, we searched the reference lists of the included studies for relevant literature.

Studies were selected in a two-stage process. We screened the titles and abstracts against the pre-specified inclusion criteria for relevant citations. This was followed by assessment of the full texts of the selected abstracts. We included randomised studies that evaluated the effect of nutritional interventions in pregnancy: diet based advice, mixed approach (combination of diet and lifestyle including physical activity) and nutritional supplements that have the potential to reduce the risk of GDM such as vitamin-D, myo-inositol and probiotics. We did not include studies that evaluated only physical activity. The comparator was standard antenatal care. Primary prevention of GDM was the outcome of interest. The secondary outcomes were maternal and fetal complications such as pre-eclampsia, mode of delivery, gestation at delivery, birth weight of the fetus, neonatal death and neonatal intensive care unit admissions. We included studies on low risk and high-risk women. Women were classified as high risk if they have at least one of the following characteristics: obesity, previous history of GDM or fetal macrosomia, advanced maternal age, and family history of diabetes. [15] Two independent reviewers (ER and MC) selected the studies. Any discrepancy between them was resolved by a third reviewer (ST).

Study quality assessment and data extraction

Two independent reviewers (ER and MC) assessed the study quality and extracted the data using pre-designed forms. We assessed the risk of bias [16] in the following domains: sequence generation, allocation concealment, blinding, incomplete outcome data and selective reporting. Data were extracted in 2x2 tables for dichotomous outcomes. For continuous outcomes we extracted data on mean and standard deviation for both the groups. When more than one definition was used for GDM in a study, we extracted data for outcomes using the most recent diagnostic criteria. Any discrepancies were resolved by discussion with the third reviewer (ST).

Data synthesis

Data were summarised as risk ratio (RR) with 95% confidence interval (CI) for dichotomous outcomes, and standardised or weighted mean difference with 95% CI for continuous outcomes using the random effects model. We assessed statistical heterogeneity between trials by using the I² statistic. We undertook subgroup analysis planned a priori to explore whether the effect on the outcome would vary according to the type of intervention, and Body Mass Index (BMI), and risk status of the participants for GDM. The subgroup difference was evaluated using Chi squared test. When more than one intervention was compared to standard care in a study, we chose the combined intervention over the individual diet or nutritional supplement for the pooled analysis. We used random effects model for meta-analysis. Sensitivity analysis was undertaken by substituting the individual intervention instead of the combined method to assess for any change in the summary estimates of effects. We used Harbord’s modified test to assess for publication bias [17] and potential small study effect. All analyses were performed with Review Manager (RevMan version 5.2) and Stata software (version 11).

Study selection

Our initial search in electronic databases yielded 1761 citations. Further eight studies were identified from the reference lists of the selected studies. Twenty RCTs with 6,444 women were included in the review. [18–37] The process of study identification and study selection is provided in Fig. 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow diagram outlining study selection in the systematic review on nutritional manipulation in the prevention of gestational diabetes.

https://doi.org/10.1371/journal.pone.0115526.g001

Characteristics of the included studies

The following nutrition based methods were evaluated: diet based interventions (5 RCTs; 1,309 women) [25,31,33,36,37], mixed (diet and lifestyle) approach (13 RCTs; 4,745 women) [18,20–24,27–30,32,34,35] and nutritional supplements: myo-inositol (1 RCT; 220 women) [19] and diet with probiotics (1 RCT; 170 women) [26] .

Of the 20 included studies, 13 RCTs were on women at risk of developing gestational diabetes [18,19,21,22,25,30–37] and seven included any risk women [20,23,24,26–29]. Twelve RCTs examined the effect of the intervention in either obese or/and overweight women [18,20,22,27,30–35,37] and eight included women of any BMI. [19,23–26,28,29,36]

The interventions varied in their composition, especially those based on diet. The diet based strategy included low glycaemic index diet [36] , and restricted energy intake according to the individual requirements. [37] The mixed approach group delivered a combination of diet and lifestyle including physical activity. The nutritional supplement myo-inositol was provided as a 2 g dose twice a day with 200 ug folic acid. The probiotics Lactobacillus Rhamnosus GG and Bifidobacterium lactis Bb12 in dose of 10¹⁰ colony forming units were taken every day in addition to intensive dietary counselling. [26]

The interventions were delivered in groups or as a one-to-one contact session, and often in more than one step. The nutritional advice was accompanied by psychological input in some studies. [23,31] Participants were provided with food diaries to record their food intake and the dieticians tailored the intervention according to the caloric requirements. Most of the interventions from the mixed approach category were based on the national recommendations on healthy eating in pregnancy, with additional input provided by trained healthcare personnel. The mixed approach category aggregates studies with complex interventions targeting weight gain from different angles; ranging from change in a type of consumed foods and daily physical activity pattern [23,27] , behavioural change [18,22] to weight gain monitoring only [24] . All the interventions were commenced before 28 weeks at varied time points in the first or second trimester.

The definition of GDM varied between the studies (Table 1). The following maternal outcomes were evaluated: preterm delivery, caesarean section, induction of labour, pre-eclampsia and pregnancy induced hypertension. The fetal outcomes included birth weight, shoulder dystocia and admission to neonatal intensive care unit (NICU). The follow up period varied from 6 weeks after delivery to the end of exclusive breastfeeding.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Clinical characteristics of included studies evaluating the effectiveness of nutrition manipulation in primary prevention of gestational diabetes mellitus (GDM).

https://doi.org/10.1371/journal.pone.0115526.t001

Quality of the included studies

All studies with diet based interventions had low risk of bias for adequate randomisation and attrition [25,26,31,33,36,37] ; half of them (3/6) had low risk for allocation concealment [26,31,36] . Only one study reported adequate blinding of researchers and participants [26] and 33% were considered as high risk in this domain (2/6) [25,36] . The blinding of outcomes assessors was adequate in one diet study [31] and inadequate in one [25] . The risk of bias due to selective reporting was low in 33% of the included diet based studies (2/6) [26,31] and inadequate in other two trials [25,36] .

Twelve trials on mixed (diet and lifestyle) approach had a low risk of bias for adequate randomisation (92%, 12/13). [18,20–24,27,28,30,32,34,35] Allocation concealment was well described in six [20,22,27,28,30,32] out of 13 mixed approach interventions studies. Blinding of staff and participants was adequate in one trial [22] and inadequate in two; and blinding of outcomes assessment was correct in 31% of included trials (4/13) [20,22,28,34]. Attrition bias was low for majority of studies except one study that was at high risk [23]. Selective reporting bias was low in 54% of mixed approach trials [18,20,22–24,28,29] and was considered to be high in four trials (31%) [21,27,32,35].

Of the two trials on nutritional supplements, the trial on myo-inositol supplementation [19] was assessed to have low risk of bias for adequate randomisation, attrition and selective reporting and high risk for blinding of staff and participants. The trial on probiotics had low risk of bias for all domains except for detection bias. [26] Fig. 2 provides the quality assessment of the included studies for diet based and mixed approach groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Quality assessment of the meta-analysed RCTs with a) diet based interventions, and b) mixed approach (diet and lifestyle) reporting rates of gestation diabetes (GDM).

https://doi.org/10.1371/journal.pone.0115526.g002

Effect of nutritional manipulation on GDM

Interventions that were mainly based on diet reduced the rates of GDM by 33% (RR 0.67; 95% CI 0.39, 1.15I² = 52%) (Fig. 3a). There were no differences between the two groups for mixed (diet and lifestyle) approach (RR 0.95; 95% CI 0.89, 1.22; I² = 23%) (Fig. 3b). The risk of GDM was reduced by 60% for probiotics (with diet) in comparison to standard care (RR 0.40; 95% CI 0.20, 0.78; p<0.01). A similar reduction was observed for the nutritional supplement myo-inositol (RR 0.40; 95% CI 0.16, 0.99; p = 0.05) (Fig. 3c).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Forest plot of the meta-analysed RCTs with a) diet interventions, b) mixed approach (diet and lifestyle), and c) nutritional supplements reporting rates of gestation diabetes (GDM).

https://doi.org/10.1371/journal.pone.0115526.g003

Small study effect

Due to insufficient number of studies for myo-inositol and diet with probiotic supplementation we examined funnel plot asymmetry only for diet based and mixed approach groups. For both groups Harbord’s statistical tests for small study effect was insignificant (p = 0.87 and p = 0.21, respectively),

Subgroup analysis

Subgroup comparison was possible to conduct only for two intervention categories: diet based and mixed approach. There was no subgroup differences based on maternal risk for GDM for both intervention groups and for BMI category comparison for the mixed approach group (Table 2). There was a significant difference according to the BMI for diet based intervention (p = 0.04). A significant reduction in GDM was observed in the subgroup comprised of obese and overweight women (RR 0.40; 95% CI 0.18, 0.86, I² = 29%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Subgroup analyses for intervention types and clinical characteristics for gestational diabetes mellitus (GDM) in evaluation of nutritional manipulation in pregnancy.

https://doi.org/10.1371/journal.pone.0115526.t002

Effect of nutritional interventions on other maternal and neonatal outcomes

Maternal outcomes.

Eleven trials [19,20,22,24,27–29,32–34,36] evaluated the role of interventions in preventing preterm delivery before 37 weeks of gestation None of the interventions significantly reduced the rate of preterm deliveries; however, the risk reduction was the highest (51%) for diet based group (RR 0.49, 95% CI 0.19, 1.29).

Fourteen RCTs [18–20,23,24,26–29,32–36] reported the effect of interventions on the caesarean section rate and six studies [18,20,27,32,33,36] on the rate of the induction of labour. There was no significant effect of the intervention on the rates of evaluated outcomes (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of findings for maternal and neonatal outcomes from trials with nutritional manipulation in pregnancy.

https://doi.org/10.1371/journal.pone.0115526.t003

Ten studies reported on the effects of the intervention on gestational hypertension [18–20,24,27–29,32,33,37] and nine on pre-eclampsia [18,20,24,28,29,32,33,35,37]. The risk of gestational hypertension and pre-eclampsia was reduced by 84% and 34%, respectively, in diet-based group (Table 3). There was no noticeable effect of mixed approach on the occurrence of discussed outcomes.

Summary of findings

Nutritional manipulation based on diet or mixed approach does not appear to prevent GDM. There was a trend towards beneficial effect in women on mainly diet-based intervention, with a potential for significant reduction in GDM risk when limited to obese and overweight women. Nutritional supplements such as probiotics and myo-inositol show promising role in the strategy for primary prevention of GDM.

Relevance to current evidence

Until now, there has been no robust evidence to provide guidance on the primary prevention of GDM due to the small number of studies limited to few interventions in published reviews. [38] The number of eligible studies has doubled since our previous review that evaluated the effect of mixed approach (diet and lifestyle modification) on GDM. [11] By evaluating all the relevant interventions, our review is the first to systematically assess the effects of nutritional manipulation in pregnancy on GDM. We complied with current guidelines and used a comprehensive search strategy without any language restrictions. By including only randomised trials, we avoided some of the pitfalls encountered by earlier reviews that included quasi-randomised studies [10] and women with GDM [9].

Effects of interventions on GDM

Amongst evaluated interventions, diet based interventions appear to show potential for preventing GDM. This could be due to the following reasons: individual dietary and components; change in gestational weight gain and effect of nutritional supplements.

The interventions promoted the uptake of healthy components such as fibre, probiotics and food rich in vitamins such as myo-inositol that may have an additive effect in reducing the concentrations of maternal glucose. [19,26] The women in the intervention group had reduced total energy intake and glycaemic load compared to the controlled group. [30,36] Low glycaemic index diet attenuates the increase in insulin resistance observed in pregnancy, thereby reducing the risk of GDM. [39] The risk of GDM is known to be reduced by a quarter with each 10-g/day increment in total fibre intake. [40] The largest benefit with diet was observed where there was a multidisciplinary input into the intervention, with the use of food diaries [31] and feedback methods.

Diet based interventions have also shown the greatest reduction in gestational weight gain compared to other methods. [11] The reduction in gestational weight gain may have influenced the fall in the rates of GDM. [11] Serum leptin, a known factor associated with GDM [41], was lowered by 20% with reduced gestational weight gain. [37] Cord leptin concentrations were also increased in newborns born to mothers with diabetes. [42]

We did not observe the beneficial effect in the subgroup with mixed approach that combined diet and physical activity. This is consistent with previously published reviews that did not show beneficial effect of physical activity in pregnancy on pregnancy outcomes. [11,43] Rather than physical activity failing to have an expected impact on GDM, it is likely that women in the intervention group had poor compliance with the intervention. Objective assessments with methods such as accelerometry have shown no difference in the physical activity between the two groups. [30] The largest trial on mixed approach (diet and lifestyle) in pregnancy, the LIMIT study failed to show a benefit with the intervention for GDM and other maternal outcomes including gestational weight gain.[20] Non-compliance with the intervention, with a quarter of women not attending the required two sessions with the dietician could have contributed to the lack of benefit.

Simple interventions based on nutritional supplements such as myo-inositol and probiotics appear to have significant potential in preventing GDM. Inositol is available in cereals, meat, fresh fruit and vegetables, corn and legumes. The average dietary intake contains 1g of inositol/day. Myo-inositol is known to increase the sensitivity to insulin, [44] a possible mechanism for the observed reduction in GDM.

Other supplements such as the probiotics, consisting of microorganisms of beneficial nature, appear to reduce the risk of GDM when combined with a dietary intervention. [26] By altering the gut microbiome, and by modifying the concentration of plasma lipopolysaccharides, probiotics alter the inflammatory pathways and sensitivity to insulin. It is possible that the benefit observed in the Luoto trial in reducing GDM [26] was due to a synergistic action between a diet rich in probiotics in addition to probiotics supplements.

Safety of the interventions

Any intervention evaluated in pregnancy needs to pass a rigorous evaluation of its safety to the mother and baby. Our previous detailed evaluation of diet and mixed interventions in pregnancy did not find adverse effects to the mother or baby, except in extreme conditions such as starvation. [11] Although theoretical concerns have been raised regarding the risk of preterm delivery with inositol, this was not observed in both randomised and observational studies on inositol in pregnancy. Inositol use in early pregnancy may in an additional beneficial role, by preventing the risk of neural tube defect in folate resistant mothers. [45]

Limitations

Our findings were limited by differences in the inclusion criteria of the studies, variation in the components of the intervention such as duration, intensity and frequency, non-standardised care in the control group and non-uniform definitions of GDM. Furthermore, women in the intervention group had more than one intervention, such as diet and probiotics, making it difficult to delineate the beneficial effect of an individual intervention. It is possible that a different criterion for the diagnosis of gestational diabetes may have yielded changed estimates of effect. [46] Women in the control group may have accessed these interventions resulting in Hawthorne effect for the following reasons: interventions were easily accessible, including over the counter nutritional supplement; and absence of blinding of the women or health care provider, in any of the included studies. None of the studies evaluated GDM as a primary outcome. Hence it is possible that the different arms could have been treated differently, such as additional screening for GDM, and close follow-up in the intervention group, thereby influencing the outcome. Studies were limited in their reporting on proportion of women who complied with the intervention, which could have a major influence on the effect size observed.

Clinical applicability

Since women with GDM are mostly seen in the secondary care, with frequent follow ups including ultrasound assessment of fetal growth, any effective intervention that prevents GDM is likely to be cost effective in the long run. Dietary interventions are complex, and require a change in the behaviour of mothers, to have a positive impact on the outcomes. Furthermore, they require reinforcement and feedback with food diaries, and regular visits with healthcare professionals such as dieticians, midwives and clinicians. The diet based intervention may have a role in primary prevention of GDM, especially in obese and overweight pregnant women.

With a projected increase in the National Health Service (NHS) spend from £8.8 billion to £13 billion per year in the next 25 years on Type 2 diabetes and its complications, [47] primary prevention of GDM has significant societal and economic benefits. Interventions based on diet and nutritional supplements show potential to prevent GDM, with the possibility of promoting the health of subsequent generations, by reducing the risks of obesity and adult onset diabetes in children born to mothers with GDM.

Research recommendations

The role of diet-based interventions in obese and overweight pregnant women, the population most likely to develop GDM, needs further evaluation. The beneficial effects of simple interventions such as probiotics and myo-inositol on GDM appear promising. There is a need to evaluate the effects of supplements by large multicentre randomised trials, involving wider group of individuals such as non-Caucasians and obese women. The optimal dose, frequency and type of inositol isomer need to be identified. Similarly the effects of different genera or strains of probiotics and their varied dose on GDM need to be identified. Given the considerable resources required to deliver the complex interventions based on diet, it is possible that nutritional supplements will also be cost effective. Furthermore, they are an attractive option as they are easily available as over the counter supplements.