This systematic review indicates that dietary intervention during pregnancy appears effective to reduce total and weekly GWG and incidence of cesarean section. However, there was no significant evidence for effects on preventing excessive GWG (GWG above IOM guidelines). In addition, dietary intervention during pregnancy seems to have a significant effect on reducing long-term postpartum weight retention, but no effect on weight retention at 6 weeks after birth. However, early postpartum weight might be less important for the prediction of maternal obesity as weight at 6 weeks postpartum may be affected by edema, morphological stage of reproductive organs and lactation compared to 6-12 months .
Total GWG was significantly reduced by almost 2 kg in all types of interventions, indicating a clinically relevant reduction. Clinical heterogeneity was prominent across trails, particularly when looking at the characteristics of the participants, but also the type, the duration and the intensity of the interventions. As for the statistical heterogeneity that was identified, a random effects model analysis was applied when comparing studies. Thornton et al  and Wolff et al  showed the strongest effect estimate. Both trials included obese women recruited around their second trimester. The interventions applied were similar in terms of food composition, giving the women individual recommendations on daily energy intake and using food records to detect unhealthy eating. The intensity of the interventions varied markedly; one  being intense with 10 sessions of one-hour dietary consultations with a trained dietitian, at each antenatal visit and the other  being more of a monitoring nature, starting after the first visit to the dietician. Using food record might be time consuming, but appears important in reducing GWG as women become more aware of the foods they eat. However, total GWG was significantly reduced by 1 kg even after exclusion of the trials by Thornton et al  and Wolff et al  from the meta-analysis.
Weekly GWG was significantly reduced by 0.26 kg/wk in studies using caloric restriction, only. Both of the Campbell et al. trials [19, 38] succeeded in reducing weekly maternal weight gain, although the magnitude of the reduction was markedly larger in the 1975 trial  where the caloric restriction was applied between 30 and 38 weeks' gestation. Women in the intervention group tended to have smaller babies than what would be expected in individual who had a high weight gain between 20 and 30 weeks' gestation. It therefore puts a question to whether dieting, late in pregnancy, might affect the weight gain of the fetus.
It is still unclear whether dietary interventions, particularly low-energy diet may increase the incidence of low birth weight since the overall effect of two caloric restriction trials was toward increased risk (RR = 1.30), but the confidence intervals were wide probably due to the limited number of studies included [24, 38]. Subgroup analysis including only 4 trials based on caloric restriction [24, 27, 38, 39] did not significantly reduce the mean birth weight, but likewise the overall effect was toward reduced mean birth weight (WMD = -127.6).
Intervention targeting the high-risk group
Polley et al  and Phelan et al  delivered the same intervention to normal weight and overweight/obese women. However, both trials, based on behavioral lifestyle intervention, managed to reduce the frequency of excessive weight gain according to the IOM recommendation only in normal weight women, while the intervention had no significant effect on overweight/obese women and the trend being in the opposite direction. On the other hand, the trials by Wolff et al  and Thornton et al , based on caloric restriction, targeting overweight/obese women were effective in reducing total GWG in the intervention group. This might indicate that a low intensity behavioral intervention aiming to decrease high-fat foods and increase PA may not be sufficient to prevent excessive GWG in high-risk women (overweight/obese). More intensive interventions involving frequent contacts (e.g., weekly nutritional counseling) and emphasis on caloric restriction (18-25 kcal/kg) seems to be more appropriate for preventing excessive GWG among the overweight/obese women.
Comparison with previous studies
A previous review , including four RCTs and five non-RCTs with either historical or concurrent controls, similarly indicated that the GWG was significantly lower in the intervention group compared to control. However, when the analysis was confined to the RCTs the results were no longer significant and only a trend to lower GWG in the intervention group was observed. This finding confirms the postulated overestimation of treatment effects by non-randomized trials. Nevertheless, the overall effect of our study was higher than that reported by Streuling et al  (reduction of 1.2 kg). This might be explained by the inclusion of new trials, particularly the ones by, Huang et al (-2.2 kg) , Thornton et al (-9.1 kg)  and Wolff et al (-6.7 kg)  in the present meta-analysis, which showed significant reduction in GWG. Another review by Dodd et al  included nine RCT, four not being specifically designed to prevent excessive GWG, but the review did not present any statistical significant data on the outcome measures.
Strengths and limitations
To date, this study is the largest systematic review including 1434 normal weight and overweight/obese women, from dietary interventional studies with available information on total GWG. This systematic review was restricted to RCTs and QCTs in order to assure comparability between interventions and control groups, and to reduce risk of bias.
It was not possible to quantify the intensity of different interventions due to lack of details provided in the articles. Furthermore, problems with confounding were detected when looking at the methodological quality. Only one out of the 13 studies  had a final classification of low risk of bias, five [18, 25, 26, 37, 39] had a moderate risk and seven [19–23, 27, 38] had a high risk of bias.
As discussed in previous reviews; comparing GWG can be problematic as there is no common standard for calculations [15, 16]. Among the reviewed studies, 10 studies calculated GWG based on self-reported pre-pregnancy weight [18, 20–27, 39] and three studies did not report the means of data collection when calculating GWG [19, 37, 38]. The final weight was taken at the day of delivery in four studies [18, 20, 21, 26], and at the last clinic visit prior to delivery in six studies [22–25, 27, 39].
Another limitation was the lack of statistical power to capture small intervention effects on some clinical outcomes. The overall estimates tended to show that dietary intervention may reduce the incidence of preeclampsia, gestational diabetes and macrosomia, but the results did not reach statistical significance. In addition, there was lack of refined information on infant outcomes. None of the trials had available information on intrauterine growth restriction or small for gestational age. It would be relevant to assess the effect of dietary interventions on subcategories of preterm birth, such as moderately preterm birth (< 37 weeks) vs. very preterm birth (< 32 weeks) or spontaneous preterm birth vs. medically induced birth. However, only 4 trials contributed data on preterm birth [24, 26, 38, 39] and none of them presented data on subcategories of preterm birth.
Implications for practice and research
Dietary intervention seems to have no adverse effect on infant birth weight and gestational duration, but we could not find strong evidence that dietary intervention significantly reduced the incidence of preeclampsia, gestational diabetes and macrosomia. Further implications for fetal, infant, or maternal health cannot be judged from the available trials. Therefore, further research, with larger sample size, is required to confirm the results. Due to low methodological quality of included studies, future trials should ensure strict and concealed randomization, intention-to-treat analysis, and adequate blinding of outcome assessment. Since adherence to weight-control programs requires considerable effort, more information is necessary on women's satisfaction and compliance with such interventions. These outcomes should be evaluated in a systematic fashion.
It is suggested that dietary interventions targeting overweight/obese women should be more intensive than interventions targeting normal weight women. However, it was not possible to systematically quantify the intensity of interventions across trials and uncertainty on optimal intensity, limits the ability to generate reliable recommendations for clinical practice. It seems that nutritional counseling based on face-to-face visits and recommendation for patient-focused caloric intake, are more likely to be successful. Nevertheless, the ability of health care systems to deliver time-intensive interventions at population level remains unknown.
Informing and educating women on appropriate weight gain before and in the beginning of pregnancy, might contribute to a better compliance. Studies should focus more on the psychological aspect to why women are overweight to begin with. Women who are heavier before pregnancy are more susceptible to increase their weight gain during pregnancy and adhere less to IOM guidelines , while women who exercise, watch their dietary intake and weight before pregnancy, might be more likely to focus on staying within the IOM guidelines. Other pregnancy related weight gains, which need to be addressed, might depend on the lifestyle changes that accompany motherhood, leaving the women more vulnerable to eating disorders . Not to forget, women who are pregnant are probably more likely to make healthier lifestyle changes, than during any other time in their life. Positive health outcomes should be lifted as well as negative outcomes, which are associated with not following guidelines in motivating women.
Previous observational studies have stated that family members might influence women on their exercise and dietary behavior during and after pregnancy and that the most normative influence was from their partners . For this reason, future studies may consider taking into account the participation of family members, such as the husband or partner, as one of the characteristics of the intervention.