Effectiveness of Lifestyle Interventions during Pregnancy on Preventing Gestational Diabetes Mellitus in High-Risk Women: A Systematic Review and Meta-Analyses of Published RCTs

Background: Until now, it is uncertain whether lifestyle interventions during pregnancy can prevent gestational diabetes mellites (GDM) in high-risk pregnant women. Objective: This study aims at investigating the effectiveness of dietary interventions and/or exercise interventions during pregnancy for preventing GDM in high-risk pregnant women. Materials and Methods: Eligible randomized controlled trials (RCTs) were selected after a search in CENTRAL, Scopus, and PubMed. Synthesis was performed for the outcome of GDM in women with any identified GDM risk factor. Separate meta-analyses (MA) were performed to assess the efficacy of either nutrition or physical activity (PA) interventions or both combined compared with standard prenatal care for preventing GDM. Subgroup and sensitivity analyses, as well as meta-regressions against OR, were performed to assess potentional heterogeneity. Overall quality, the quality of RCTs, and publication bias were also evaluated. Results: A total of 13,524 participants comprising high-risk pregnant women in 41 eligible RCTs were analyzed for GDM. Women receiving only a nutrition intervention during pregnancy were less likely to experience GDM compared with women following standard prenatal care. Among 3109 high-risk pregnant women undergoing only dietary intervention for preventing GDM, 553 (17.8%) developed GDM; however, the result of the MA was marginally not significant (OR 0.73, 95%CI 0.51, 1.03; p-value 0.07), (Q 21.29, p-value 0.01; I2 58% (95%CI 10, 78%)). Subgroup analyses demonstrated an effect for studies that were conducted in Great Britain (OR 0.65, 95%CI 0.49, 0.81; p-value 0.003), and in Spain (OR 0.50, 95%CI 0.27, 0.94; p-value 0.03), for studies with forms of the Mediterranean diet as the intervention’s component (OR 0.61; 95%CI 0.46, 0.81; p-value 0.0005), and for studies including a motivation arm in the intervention (OR 0.71, 95%CI 0.58, 0.87; p-value 0.0008). Among 2742 high-risk pregnant women being analyzed for GDM outcome after receiving only an exercise intervention, 461 (16.8%) were diagnosed with GDM. Women after receiving PA intervention were less likely to develop GDM (OR 0.64, 95%CI 0.51, 0.80; p-value < 0.0001), (Q 11.27, p-value 0.51; I2 0% (95%CI 0, 99%)). Finally, 1308 (17%) cases of GDM were diagnosed among 7673 high-risk pregnant women undergoing both diet and PA intervention. Women in the group of mixed lifestyle intervention had a significant reduction in incidence of GDM (OR 0.70, 95%CI 0.55, 0.90; p-value 0.005), (Q 50.32, p-value < 0.0001, I2 66%, (95% CI 44, 79%)). Conclusions: The results of this study support the efficacy of lifestyle interventions during pregnancy for preventing GDM in high-risk women if an exercise component is included in the intervention arm, either alone, or combined with diet. A combined lifestyle intervention including physical exercise and a Mediterranean diet accompanied by motivation support may be considered the most effective way to prevent GDM among high-risk women during pregnancy. Future research is needed to strengthen these findings.


Introduction
According to the latest guidelines of the American Diabetes Association (ADA), gestational diabetes mellites (GDM) is a type of diabetes mellitus (DM) that is recognized after the first trimester of pregnancy [1,2].It is the most common gestational and childbirth complication [3,4].Independently of the applied diagnostic criteria, the incidence of GDM is increasing in worldwide [5,6].Therefore, it is an important disease that affects pregnancies [7], increasing the incidence of both short-term and long-term unfavorable health circumstances [8].
Several interventions to mitigate hyperglycemia have been suggested, the main ones being pharmacological, PA and lifestyle [21].However, the research question of whether GDM could be prevented by interventions during pregnancy or before pregnancy remains unanswered [22,23].Physical activity (PA) may be protective against T2DM, although the data regarding PA and GDM are less extensive and less convincing [24].Yet, there is insufficient evidence that lifestyle interventions are effective in preventing GDM [25].
According to a recent systematic review (SR) and meta-analysis (MA), any intervention based on diet or PA or both of them, resulted in significantly less occurrence of GDM [26].Additionally, a recent network MA demonstrated that GDM could be prevented with the implementation of exercise plus probiotic interventions, whereas dietary only, or dietary plus PA interventions did not alter the GMS incidence [1].However, a previous SR and MA did not find any benefit with diet, or exercise, or a mixed approach for reducing GDM risk [27].Similarly, three SRs and MAs assessing lifestyle interventions during pregnancy including diet and PA reported no effect on decreasing the outcome of GDM in obese or overweight women [23,28,29].
Preventing GDM is a priority in pregnancy [21].Moreover, pregnancy may represent a good time period for lifestyle changes as pregnant women have strong willingness to improve the health benefits for them and for their offsprings [28].In this SR and MA, we aim to summarize the most recent evidence regarding the efficacy for preventing GDM among high-risk women of any lifestyle interventions, including either nutrition or PA interventions, or combined diet-plus-exercise interventions that are implemented during the gestational period.

Search Strategy
PubMed, Cochrane Library Central Register of Controlled Trials (CENTRAL), and Scopus were searched (from inception of data to August 2022).The search strategy for Pubmed included keywords related to diet, nutrition, exercise, PA, and GDM combined with the Cochrane Collaboration search algorithm for randomized controlled trials (RCTs).CENTRAL and Scopus were searched systematically using the same keywords (Table S1).Based on title and/or abstract, the full text was retrieved for unclear items or potential eligibility.One investigator (GIT) screened all the databases.A second investigator (AB) checked the items for which the first investigator (GIT) could not reach a decision.Discrepancies were resolved through consensus.

Eligibility Criteria
The PICO (population, intervention, comparator, and outcome) approach was used for selecting eligible trials.RCTs in English language, including high-risk pregnant women for GDM with any identified risk factor were accepted.RCTs evaluating the risk factor of overweight and/or obesity among participants via BMI undivided and as a total group without participants' stratification were considered as eligible.RCTs appraising any type of active lifestyle intervention of diet alone, exercise alone, or both during pregnancy compared with standard antenatal care were included.Trials involving the outcome of GDM being diagnosed by any recommended modality were chosen.
Protocols of RCTs, pilot RCTs, secondary analyses, and abstracts from conference proceedings were excluded.

Data Extraction
Two independent researchers (GIT and PP) extracted the data.Discrepancies were resolved by consensus.When necessary, a third arbitrator (AB) contributed to the final decision.The extracted items were the first author's name, publication year, country where performed, type of RCT, number of centers for multicentered trials, study duration, dropout rate, sample size, women's mean age, women with low education level according to authors' assessment of any eligible trial, GDM risk factors, the type of intervention and the care for provided in the control group, as well as any potential reported side-effects for both the experimental, and comparator groups.The diagnostic methods and time period of screening for GDM were also recorded.The Consensus on Exercise Reporting Template (CERT) tool for complex interventions was used to evaluate exercise programs in RCTs applying only PA in the intervention arm [31].Finally, the number of patients with GDM being diagnosed by any method was captured as an outcome, separately in any group.

Quality Assessment of the Studies and Rating of Overall Evidence
To evaluate the quality of the eligible RCTs, the risk of bias tool proposed by the Cochrane Collaboration was used [32].In addition, overall evidence was rated through the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) framework (GRADEpro, Version 3.6.1.McMaster University, 2011) [33].

Statistical Analysis
The main analyses included all available data.The significance level for Cochran's Q statistic was set at p-value < 0.1, and for the rest of the analyses, at p-value < 0.05 [34].SPSS 22.0 (SPSS, Inc., Chicago, IL, USA), Stata Statistical Software 10.1 (Stata, College Station, TX, USA), and Review Manager 5.4.1 (Cochrane Collaboration, London, UK) were used for the analyses.
Separate analyses [36] were performed for studies based on the studies' performance country, for studies with more than 10% of participants with a low education level, and for studies evaluating or not overweight or obesity as a GDM risk factor.Moreover, separate analyses were conducted for studies with an intervention duration of more than 20 weeks and less than 20 weeks, and for trials including a Mediterranean diet as a component of dietary interventions or including any dietary intervention.Finally, subgroup analyses included trials assessing both a motivation component in the intervention, and trials not including motivation.The effect of the RCTs with the largest sample size was also estimated by their exclusion in sensitivity analyses [36].Additionally, sensitivity analyses were also performed for studies with low attrition bias [36].Meta-regression analyses on GDM OR were conducted with the effect of baseline risk, and study duration as covariates [36,37].Publication bias was assessed via the visual analysis of a funnel plot [38].The statistical test of Egger was also performed for publication bias assessment [39].

Eligible Studies
The search yielded 10,086 items (2181 in CENTRAL, 6419 in Scopus, and 1486 in PubMed).A total of 1598 items were excluded as duplicated.Out of the 8488 remaining items, 8374 were excluded as non-relevant based on the title, or abstract.Thus, 114 papers were retrieved in full text.Out of the 114 articles, 73 were excluded: 4 studies were published in a non-English language; 32 studies did not include an eligible population; 6 studies included a non-eligible intervention; 20 trials did not report the onset of GDM as an outcome; 2 papers reported a pilot RCT; 8 papers included secondary analyses of RCTs; and 1 paper was retrieved from a conference.Finally, this study included 41 eligible published RCTs.Specifically, 10 of them reported diet-only interventions, 13 exercise-only interventions, and 18 diet-plus-exercise interventions (Figure 1).
One trial did not report any equipment type used during exercise [52] (Table S2).One other trial did not mention the provider [49] (Table S2).Two studies did not describe whether PA was performed individually or in a group [55,62], and one of them did not mention the existence of supervision or not [55] (Table S2).No data regarding the motivation component of intervention were found in five RCTs [48,49,54,56,62] (Table S2).One trial did not introduce decision rules for progressing exercise program [52].The location of the exercise was not clarified in one trial [54] (Table S2).Seven out of thirteen studies did not determine decision rules for the starting level of exercise [46,[48][49][50]52,56,57] (Table S2).

Effectiveness of Lifestyle Interventions during Pregnancy
After drop-out, a total of 3109 (1530 in intervention, and 1579 in control groups) pregnant women with high risk for GDM receiving dietary intervention were analyzed for GDM outcome.Among them, 553 (17.8%) developed GDM (241 (15.7%) in the intervention group, and 312 (19.7%) in the control group).Combining studies, heterogeneity across them was significant (Q 21.29, p-value 0.01), and variability was large (I 2 58% (95%CI 10, 78%)).Therefore, an RE model was selected for synthesizing the effect of dietary interventions.Women receiving nutrition intervention during pregnancy were less likely to develop GDM compared with women following standard prenatal care; however, the result of the MA was marginally not significant (OR 0.73, 95%CI 0.51, 1.03; p-value 0.07) (Figure 2).tion group, and 312 (19.7%) in the control group).Combining studies, heterogeneity across them was significant (Q 21.29, p-value 0.01), and variability was large (I 2 58% (95%CI 10, 78%)).Therefore, an RE model was selected for synthesizing the effect of dietary interventions.Women receiving nutrition intervention during pregnancy were less likely to develop GDM compared with women following standard prenatal care; however, the result of the MA was marginally not significant (OR 0.73, 95%CI 0.51, 1.03; p-value 0.07) (Figure 2).Studies presented by weight contributing to the meta-analysis-blue dots, in odds ratio (OR) estimates corresponding to their 95% confidence interval (95% CI)-black blocks.Summary of OR by random effect model demonstration, as well as metrics of heterogeneity [3,4,16,20,[40][41][42][43][44][45].
In contrast to the summary OR, a significant effect was shown when analyses were limited to (a) studies that were performed in Great Britain (OR 0.65, 95%CI 0.49, 0.81; pvalue 0.003), and in Spain (OR 0.50, 95%CI 0.27, 0.94; p-value 0.03); (b) to studies including overweight or obesity as a GDM risk factor (OR 0.78, 95%CI 0.64, 0.96; p-value 0.02), or studies where BMI was not included among GDM risk factors (OR 0.57, 95%CI 0.36, 0.93; p-value 0.02); (c) to studies including forms of the Mediterranean diet as an intervention component (OR 0.61 95%CI 0.46, 0.81; p-value 0.0005); and finally, (d) to studies with a motivation component in the intervention (OR 0.71, 95%CI 0.58, 0.87; p-value 0.0008).The test of difference was not statistically significant for any subgroup analysis (Table S6).In the sensitivity analysis evaluating the effect of the RCT with the largest sample size [20], the OR remained non-significant (OR 0.73; 95%CI 0.48, 1.13; p-value 0.16) (Table S6).Due to the small number of trials and potentional variability among studies in subgroup analyses, significant results should be interpreted cautiously.Meta-regression analyses with Figure 2. Dietary intervention and the risk of gestational diabetes mellitus in high-risk pregnant women.Studies presented by weight contributing to the meta-analysis-blue dots, in odds ratio (OR) estimates corresponding to their 95% confidence interval (95% CI)-black blocks.Summary of OR by random effect model demonstration, as well as metrics of heterogeneity [3,4,16,20,[40][41][42][43][44][45].
In contrast to the summary OR, a significant effect was shown when analyses were limited to (a) studies that were performed in Great Britain (OR 0.65, 95%CI 0.49, 0.81; pvalue 0.003), and in Spain (OR 0.50, 95%CI 0.27, 0.94; p-value 0.03); (b) to studies including overweight or obesity as a GDM risk factor (OR 0.78, 95%CI 0.64, 0.96; p-value 0.02), or studies where BMI was not included among GDM risk factors (OR 0.57, 95%CI 0.36, 0.93; p-value 0.02); (c) to studies including forms of the Mediterranean diet as an intervention component (OR 0.61 95%CI 0.46, 0.81; p-value 0.0005); and finally, (d) to studies with a motivation component in the intervention (OR 0.71, 95%CI 0.58, 0.87; p-value 0.0008).The test of difference was not statistically significant for any subgroup analysis (Table S6).In the sensitivity analysis evaluating the effect of the RCT with the largest sample size [20], the OR remained non-significant (OR 0.73; 95%CI 0.48, 1.13; p-value 0.16) (Table S6).Due to the small number of trials and potentional variability among studies in subgroup analyses, significant results should be interpreted cautiously.Meta-regression analyses with baseline risk, and study duration as covariates did not show a statistically significant effect on the summary OR (Table S7).
461 reports represent a 16.8% level of occurrence of diagnosed GDM (including 192 (14%) in the intervention group, and 269 in the control group (19.7%)) among 2742 high-risk pregnant women (1375 in the intervention group, and 1367 in control group) analyzed for gestational diabetes outcome, after drop-out.Heterogeneity across studies was nonsignificant (Q 11.27, p-value 0.51); however, variability could not be excluded as the upper limit for I 2 is greater than 75% (I 2 0% (95%CI 0, 99%)).Therefore, the results of the synthesis are presented through the FE model.The result of the MA revealed a significant reduction in GDM outcomes for women participating in the exercise intervention group compared with those participating in the standard care group (OR 0.64, 95%CI 0.51, 0.80; p-value < 0.0001) (Figure 3).
Finally, among 7673 (3863 in the experimental group, and 3810 in the comparator group) high-risk pregnant women analyzed for GDM undergoing both diet and exercise intervention, 1308 cases (17%) were diagnosed with diabetes of pregnancy (622 (16.1%) in the intervention group, and 686 (18%) in the control group).The combined studies were analyzed by an MA with RE in the presence of significant heterogeneity (Q 50.32, p-value < 0.0001, and large variability (I 2 66%, (95% CI 44, 79%)).Women in the group of mixed lifestyle intervention had a significant reduction in GDM incidence (OR 0.70, 95%CI 0.55, 0.90; p-value 0.005).(Figure 4).group) high-risk pregnant women analyzed for GDM undergoing both diet and exercise intervention, 1308 cases (17%) were diagnosed with diabetes of pregnancy (622 (16.1%) in the intervention group, and 686 (18%) in the control group).The combined studies were analyzed by an MA with RE in the presence of significant heterogeneity (Q 50.32, p-value < 0.0001, and large variability (I 2 66%, (95% CI 44, 79%)).Women in the group of mixed lifestyle intervention had a significant reduction in GDM incidence (OR 0.70, 95%CI 0.55, 0.90; p-value 0.005).(Figure 4).S10).
Based on the sensitivity analyses, the summary OR remained statistically significant when evaluating the effect of the RCT with the largest sample size [49] (OR 0.67, 95%CI 0.52, 0.86; p-value 0.002), and the RCT with a high attrition bias (drop-out ratio 31.4%)[62] (OR 0.73, 95%CI 0.57, 0.93; p-value 0.01) (Table S10).The meta-regression analyses with covariates of baseline risk and study duration did not show an effect on the summary OR (Table S11).

Quality of Reporting, Potential Bias, and Quality of Evidence
Assessing the quality of RCTs with dietary intervention, some of them were considered dubious according to specific quality reports.Particularly, three trials did not provide information on blinding of either participants or personnel [3,14,41]; in addition, one of them did not declare allocation concealment [3].Also, another three trials did not mention participant blinding [40,42,43].Four studies were considered high-risk regarding blinding of participants and personnel [4,20,44,45]; specifically, three of them were unblinded for participants [4,20,44], and one of them was unblinded for personnel [45] (Table 10).Based on the funnel plot assessment via visualization, the studies did not have a close distribution around the summary effect estimate [38] (Figure 5).Egger's test of small study effects had a p-value of 0.726.There was potential performance bias in four out of ten RCTs, and it was rated un clear in the remaining six.The overall quality of evidence was moderate for dietary inte ventions during pregnancy compared with standard prenatal care for preventing GDM high-risk pregnant women.(Table S12).
Evaluating the quality of RCTs with exercise intervention, some concerns were als evident in relation to performance bias.Three studies did not report blinding of eith participants, or personnel [49,52,54].Furthermore, performance bias was considered un clear in five more RCTs [50,51,53,55,56]: three of them did not report information on pa ticipant blinding [50,51,53], and two of them did not report on personnel blinding [55,56 A detection bias could not be excluded in two trials with unclear performance bias [54,56 as information on recording data were not provided.Performance bias was judged as hig in five studies [46][47][48]57,62]: one study was unblinded for both participants and personn [62], and four were unblinded for participants only [30,47,48,57].In addition, one trial r There was potential performance bias in four out of ten RCTs, and it was rated unclear in the remaining six.The overall quality of evidence was moderate for dietary interventions during pregnancy compared with standard prenatal care for preventing GDM in high-risk pregnant women.(Table S12).
Evaluating the quality of RCTs with exercise intervention, some concerns were also evident in relation to performance bias.Three studies did not report blinding of either participants, or personnel [49,52,54].Furthermore, performance bias was considered unclear in five more RCTs [50,51,53,55,56]: three of them did not report information on participant blinding [50,51,53], and two of them did not report on personnel blinding [55,56].A detection bias could not be excluded in two trials with unclear performance bias [54,56], as information on recording data were not provided.Performance bias was judged as high in five studies [46][47][48]57,62]: one study was unblinded for both participants and personnel [62], and four were unblinded for participants only [30,47,48,57].In addition, one trial reported a participant drop-out rate of 41% [48] (Table 11).When examining the funnel plot, variation in the standard error of the studies with exercise intervention is evident.Small studies are positioned closer to the summary effect estimate [38] (Figure 6).However, Egger's test of small study effects had a p-value of 0.399.When examining the funnel plot, variation in the standard error of the studies w exercise intervention is evident.Small studies are positioned closer to the summary eff estimate [38] (Figure 6).However, Egger's test of small study effects had a p-value of 0.3  S13).
Similar to the quality assessment of RCTs with only dietary, and only exercise in In total, five out of thirteen RCTs had potentional performance, or attrition bias.The other eight RCTs were unclear in relation to blinding.The overall quality of evidence was moderate for exercise interventions during pregnancy compared with standard prenatal care for preventing GDM in high-risk pregnant women.(Table S13).
Similar to the quality assessment of RCTs with only dietary, and only exercise intervention, some uncertainty was arose for RCTs with a combined intervention of diet and exercise.Performance bias was assessed as unclear in half of them; in particular, three trials did not give information on either participant or personnel blinding [18,59,64], and six trials only on participant blinding [6,11,58,61,65,66].In addition, one RCT had a participant drop-out rate of 31.4% [55].The performance bias was judged as high in the other half of the studies; specifically, seven were unblinded for participants and personnel [5,7,8,17,60,63,67], and two trials were unblinded for participants [10,13] (Table 12).Studies with exercise intervention form an asymmetry when the funnel plot is visually evaluated [38] (Figure 7).Similarly, Egger's test of small study effects had a p-value of 0.001.
In summary, ten out of eighteen RCTs had a potential performance or attrition bias, and nine were unclear about blinding.The overall quality of evidence was low for dietary plus exercise interventions during pregnancy compared with standard prenatal care for preventing GDM in high-risk pregnant women.(Table S14).In summary, ten out of eighteen RCTs had a potential performance or attrition bias, and nine were unclear about blinding.The overall quality of evidence was low for dietary plus exercise interventions during pregnancy compared with standard prenatal care for preventing GDM in high-risk pregnant women.(Table S14).

Discussion
For pregnant women at high risk of GDM, undergoing lifestyle interventions during pregnancy may have benefit on preventing GDM.PA interventions in pregnancy were significant in reducing the incidence of GDM in high-risk women.Combined interventions of diet plus exercise during pregnancy in high-risk pregnant women may also lead to reduced GDM occurrence.However, the protective effect of dietary interventions on GDM was marginally non-significant when they were not combined with exercise during pregnancy in high-risk women.In the subgroup analyses, it was found that the PA interventions also produced significant results for GDM prevention when separate analyses where limited to studies that were conducted in Spain, and in China, to studies with more than 10% of participants having a low education level, independently of overweight or obesity as a GDM risk factor, of intervention duration more than 20 weeks or of emotional support.The effect of a mixed intervention approach on preventing GDM also remained significant in subgroup analyses based on studies that were performed in China and in Italy, in studies involving a percentage of women with low education level more than 10%, independently of increased BMI as a GDM risk factor or duration of intervention below 20 weeks.Dietary interventions in pregnant women at high-risk for GDM that were conducted in Great Britain, and Spain, regardless of participants being overweight or obese or interventions including a Mediterranean diet or motivation arm may have benefits in preventing GDM.Heterogeneity across studies was significant.
Pregnant women, regardless of the risk for GDM, did not benefit from lifestyle or pharmacological interventions during pregnancy in terms of GDM prevention, according to an overview SR [27].In this study we focused on high-risk pregnant women, assessing the effect of lifestyle interventions on preventing GDM.Previous studies evaluating the preventive role of lifestyle interventions in pregnancy on preventing GDM in high-risk

Discussion
For pregnant women at high risk of GDM, undergoing lifestyle interventions during pregnancy may have benefit on preventing GDM.PA interventions in pregnancy were significant in reducing the incidence of GDM in high-risk women.Combined interventions of diet plus exercise during pregnancy in high-risk pregnant women may also lead to reduced GDM occurrence.However, the protective effect of dietary interventions on GDM was marginally non-significant when they were not combined with exercise during pregnancy in high-risk women.In the subgroup analyses, it was found that the PA interventions also produced significant results for GDM prevention when separate analyses where limited to studies that were conducted in Spain, and in China, to studies with more than 10% of participants having a low education level, independently of overweight or obesity as a GDM risk factor, of intervention duration more than 20 weeks or of emotional support.The effect of a mixed intervention approach on preventing GDM also remained significant in subgroup analyses based on studies that were performed in China and in Italy, in studies involving a percentage of women with low education level more than 10%, independently of increased BMI as a GDM risk factor or duration of intervention below 20 weeks.Dietary interventions in pregnant women at high-risk for GDM that were conducted in Great Britain, and Spain, regardless of participants being overweight or obese or interventions including a Mediterranean diet or motivation arm may have benefits in preventing GDM.Heterogeneity across studies was significant.
Pregnant women, regardless of the risk for GDM, did not benefit from lifestyle or pharmacological interventions during pregnancy in terms of GDM prevention, according to an overview SR [27].In this study we focused on high-risk pregnant women, assessing the effect of lifestyle interventions on preventing GDM.Previous studies evaluating the preventive role of lifestyle interventions in pregnancy on preventing GDM in high-risk women demonstrated contradictory results.An SR and MA evaluating dietary interventions for preventing GDM reported a possible effect in obese pregnant women [68], and marginally non-significant results in general population [68].According to three SRs and MAs, overweight and obese pregnant women may also have benefits in prevention of GDM from PA interventions during pregnancy [69][70][71].Moreover, evaluating overall high-risk GDM pregnant women, independently of obesity, two MAs reported significant results for preventing GDM with exercise during pregnancy [12,17].However, three SRs and MAs evaluating lifestyle interventions during pregnancy, including diet and PA interventions, among overweight and obese pregnant women found no prevention of GDM [18,28,29].In another study, the SR and MA did not demonstrate a significant reduction in GDM with PA during pregnancy in overweight and obese women [14].In this trial, evaluating the effect of lifestyle interventions on preventing GDM in high-risk women with any identified GDM risk factor, we expanded high-risk population under examination.As far as is known, this is the first study appraising systematically the effect of lifestyle interventions during pregnancy on preventing GDM among pregnant women with any risk factor for GDM.
An SR identified intrapersonal themes as the most frequently reported barriers and enablers to PA during pregnancy [72].Moreover, obese women were less active compared with normal weight women [53], and their adherence to exercise programs was low [53].Overweight and obese pregnant women changed their lifestyle habits with difficulty, probably due to a lack of motivation [17], or believing they were not at risk [11].Furthermore, the efficacy of exercise interventions during pregnancy addressing this population was not often measured [53].The optimal amount of PA in obesity is unknown, although most guidelines suggest exercising through obese pregnancy [53].The lack of success of some RCTs in preventing GDM could be attributed to the lack of statistical power and poor adherence to study protocols [73].In addition, the main issue of performing RCTs with lifestyle interventions concerns performance bias, as blinding either participants or personnel might be difficult due to the nature of the interventions [5,17].
Pre-pregnancy increased BMI is a modified and important GDM risk factor [74].The preconception stage is the critical time for implementing public health strategies attempting to reserve overweight and obesity [74].Women of childbearing age have been highlighted as a target group for prevention of obesity, according to the WHO [75].These strategies may target a limitation to gained weight in pregnancy and postpartum [75].The pre-pregnancy time period would be ideal regarding the implementation of interventions aiming at weight loss [28].However, such interventions are difficult to achieve because many pregnancies are unplanned (i.e., 50% of pregnancies in the UK), and only a small percentage of women planning pregnancy follow nutrition and lifestyle recommendations [28].Thus, a pre-pregnancy intervention may reach only a small proportion of the targeted group of women [28].
The antenatal period is considered an ideal time to intervene as mothers are motivated to make changes that could optimise their outcome and that of the baby, with opportunities for regular contact with health care professionals [76].Trials of lifestyle interventions demonstrating safety and efficacy for both the mother and her offspring, could be performed before gestational metabolic changes during the first trimester of pregnancy [73].Social support also may have an enabling role [72].Person-centred strategies using behaviour change techniques should be used to address intrapersonal and social factors to translate pregnant women's positive attitudes into increased PA participation [72].Targeted and well-designed exercise programs are needed for obese pregnant women [53].Furthermore, studies on bariatric surgery prior to pregnancy suggest that interventions targeting maternal obesity can also improve short-and long-term offspring health [57].
GDM increases short-term and long-term maternal and offsprings' complications [7,8].Major unfavorable short-term conditions include macrosomia [8], pre-eclampsia, [28], preterm birth [8], and polyhydramnios [8].Furthermore, GDM is a well-established predictor of future DM [6].Women with GDM, as well as the next generations, have an increased risk of developing Type 2 (T2) DM [16,59].Particularly, up to 70% of women with GDM may develop T2DM in later life [17].T2DM is diagnosed soon after delivery in up to 10% of women with previous GDM [6], and its prevalence may approach 38% within the first postpartum year [7].The percentage of T2DM has been observed to increase by up to 70% in a 10-year duration of follow-up [6], and by up to 60% in the 16th postpartum year [7].Additionally, women who experience GDM may have an increased risk of developing cancer during their lifetime [2].GDM pregnancies may increase the overall risk of breast cancer (BC), and especially oestrogen receptor-positive BC [77].GDM has also been corelated with pancreatic cancer in pregnancy and in later life [78].
GDM represents a heavy economic burden on patients and societies due to its diagnosis and treatment [8].In China, it has been found that GDM diagnosis and treatment, delivery, and the management of a mother's neonatal complications in the last trimester of pregnancy cost USD 1929.87 on average per woman more than a non-GDM pregnant woman [8].In Australia, the cost for women with GDM not needing insulin is AUD 2026, AUD 2534 for women needing insulin, and AUD 3826 if fetal heart rate is being monitored via cardiotocography [79].Also in Australia, according to previous estimations, AUD 53,985 additional direct costs are incurred for every 100 women with a singleton pregnancy and receiving treatment for mild GDM in addition to routine obstetric care, while AUD 6521 additional charges are incurred by women and their families; in addition, AUD 27,503 was the incremental cost per additional serious perinatal complication prevented, AUD 60,506 was the cost per perinatal death prevented, and AUD 2988 was the cost of each discounted life-year gained [80].
Preventing GDM could have greater economic and health benefits than treatment [1].It is likely that additional health service and personal financial spending on GDM prevention in the general population of high-income countries (HIC) would lead to reductions in perinatal mortality and in serious perinatal complications sufficient to justify the spending [80].Lifestyle interventions aiming to prevent excessive GWG and GDM in the general population were found to be neutral and possibly cost-effective for the health care system of Australia [81].The cost-effectiveness evaluation of the multicenter RCT conducted in Europe including diet only or exercise only or a combined approach for pregnant women at increased risk of GDM found that promotion of healthy eating and PA was the preferred strategy for limiting GWG, and for quality of life (QUALY) after delivery; however, the mixed approach of lifestyle interventions was not cost-effective for fasting blood glucose (FBG) [82].Another assessment of the cost efficacy of a lifestyle intervention for preventing GDM in high-risk pregnant women in Finland showed that interventions targeting neonatal birth weight were not cost-effective [83].In addition, an analysis of the cost of exercise intervention in high-risk pregnant women in Holland reported no significant results for GDM prevention, FBG, insulin sensitivity, infant birth weight and maternal QUALY [84].
Community and personal-based strategies are needed to support social and psychological challenges that pregnant women experience with physical exercise.The time during pregnancy at which lifestyle interventions are initiated varies.However, the majority of eligible RCTs initiated lifestyle interventions before the 20th gestational week.Additionally, the lack of estimation of overall activity level and resting time of participating pregnant women in the RCTs may contribute to conflicting results.The RCTs do not report total physical activity or the resting time of the pregnant women.Potentially, lifestyle variations including the type and time of occupation and the associated leisure time of participant pregnant women may contribute to differences in the efficacy of lifestyle interventions in preventing GDM that are observed across countries.
This study has several limitations.The results cannot be generalized to the general population.In addition to overweight or obesity, women with any risk factor for GDM including ethnicity, medical and family history, and sedentary lifestyle were evaluated, aiming to broaden the criteria of the population under assessment.Therefore, the risk for GDM was not similar.Furthermore, this study included only trials that evaluated lifestyle interventions initiated during pregnancy; therefore, their findings cannot be generalized to lifestyle interventions that may begin in the preconception period.Heterogeneity across studies could not be excluded.Procedures for blinding personnel and/or participants may face some limitations due to the nature of interventions.

Conclusions
The findings of this study support the effectiveness of lifestyle interventions during pregnancy, including a separate exercise arm or exercise combined with diet, compared with usual prenatal care, for preventing GDM in high-risk pregnant women.Nutrition interventions alone were marginally insufficient in preventing GDM.However, dietary interventions during pregnancy in the form of Mediterranean dietary patterns that are combined with exercise and are motivationally enhanced may be the preferred strategy for preventing GDM among high-risk women.Future, large, high-quality RTCs and SRs are necessary to confirm the protective effect of lifestyle interventions for GDM.

Figure 1 .
Figure 1.Flow chart of procedures for selecting studies.

Figure 1 .
Figure 1.Flow chart of procedures for selecting studies.

Figure 4 .
Figure 4. Diet-plus-exercise intervention and the risk of gestational diabetes mellitus in high-risk pregnant women.Metrics of RCT in odds ratio (OR) with 95% confidence interval (95% CI)-black blocks.Performance of meta-analysis with random effect model.Studies are presented in ascending order by weight-blue dots [5-8,10,11,13,17,18,59-67].The summary OR remained significant when separate analyses were limited to studies including (a) performance in China (OR 0.55, 95%CI 0.33, 0.90; p-value 0.02), and in Italy (OR 0.32, 95%CI 0.16, 0.60; p-value 0.0005); (b) to studies with more than 10% of participants having a low education level (OR 0.73, 95%CI 0.55, 0.95; p-value 0.02); (c) to trials without considering increased BMI as a GDM risk factor (OR 0.17, 95%CI 0.06, 0.51; pvalue 0.001); and (d) to trials with interventions lasting less than 20 weeks (OR 0.81, 95%CI 0.67, 0.97; p-value 0.02).The test of difference was significant for subgroup analyses based on the countries where studies were performed (p-value 0.02), based on increased BMI as

Figure 4 .
Figure 4. Diet-plus-exercise intervention and the risk of gestational diabetes mellitus in high-risk pregnant women.Metrics of RCT in odds ratio (OR) with 95% confidence interval (95% CI)-black blocks.Performance of meta-analysis with random effect model.Studies are presented in ascending order by weight-blue dots [5-8,10,11,13,17,18,59-67].The summary OR remained significant when separate analyses were limited to studies including (a) performance in China (OR 0.55, 95%CI 0.33, 0.90; p-value 0.02), and in Italy (OR 0.32, 95%CI 0.16, 0.60; p-value 0.0005); (b) to studies with more than 10% of participants having a low education level (OR 0.73, 95%CI 0.55, 0.95; p-value 0.02); (c) to trials without considering increased BMI as a GDM risk factor (OR 0.17, 95%CI 0.06, 0.51; p-value 0.001); and (d) to trials with interventions lasting less than 20 weeks (OR 0.81, 95%CI 0.67, 0.97; p-value 0.02).The test of difference was significant for subgroup analyses based on the countries where studies were performed (p-value 0.02), based on increased BMI as a GDM risk factor (p-value 0.003), and based on intervention duration (p-value 0.03).In contrast, the test of difference for subgroup analyses based on participant's low education level was non-significant (p-value 0.34) (TableS10).Based on the sensitivity analyses, the summary OR remained statistically significant when evaluating the effect of the RCT with the largest sample size[49] (OR 0.67, 95%CI 0.52, 0.86; p-value 0.002), and the RCT with a high attrition bias (drop-out ratio 31.4%)[62](OR 0.73, 95%CI 0.57, 0.93; p-value 0.01) (TableS10).The meta-regression analyses with covariates of baseline risk and study duration did not show an effect on the summary OR (TableS11).

Figure 5 .
Figure 5. Funnel plot including all studies comparing diet intervention vs. standard prenatal ca for gestational diabetes prevention among high-risk pregnant women.

Figure 5 .
Figure 5. Funnel plot including all studies comparing diet intervention vs. standard prenatal care for gestational diabetes prevention among high-risk pregnant women.

Figure 6 .
Figure 6.Funnel plot including all studies comparing exercise intervention vs. standard pren care for gestational diabetes prevention among high-risk pregnant women.In total, five out of thirteen RCTs had potentional performance, or attrition bias.T other eight RCTs were unclear in relation to blinding.The overall quality of evidence w moderate for exercise interventions during pregnancy compared with standard pren care for preventing GDM in high-risk pregnant women.(TableS13).Similar to the quality assessment of RCTs with only dietary, and only exercise in

Figure 6 .
Figure 6.Funnel plot including all studies comparing exercise intervention vs. standard prenatal care for gestational diabetes prevention among high-risk pregnant women.

Figure 7 .
Figure 7. Funnel plot including all studies comparing diet-plus-exercise intervention vs. standard prenatal care for gestational diabetes prevention among high-risk pregnant women.

Figure 7 .
Figure 7. Funnel plot including all studies comparing diet-plus-exercise intervention vs. standard prenatal care for gestational diabetes prevention among high-risk pregnant women.

Table 1 .
Characteristics of eligible trials with dietary intervention.

Name of First Author, Year of Publication Country Number of Participating Centers Study Design Study Duration (mo) Drop-Out Rate n (%)
ND, no data; UK, United Kingdom; USA, United States of America.

Table 2 .
Characteristics of eligible trials with exercise intervention.
USA, United States of America.

Table 3 .
Characteristics of eligible trials with dietary plus exercise intervention.
ND, no data; UK, United Kingdom; USA, United States of America; UAE, United Arab Emirates.

Table 4 .
Characteristics of participants in studies with dietary intervention.

Table 5 .
Characteristics of participants in studies with exercise intervention.

Table 6 .
Characteristics of participants in studies with dietary plus exercise intervention.

Table 6 . Cont. Name of First Author, Year of Publication Sample Size (Inter- vention/Control) Mean Age (SD), yr Intervention/Control Low Education Level, n (%) Intervention/Control Risk Factors for GDM
ND, no data; BMI, body mass index; DM, diabetes mellitus; GDM, gestational diabetes mellitus; PCOS, polycystic ovary syndrome; FPG, fasting plasma glucose.

Table 7 .
Eligible studies with dietary intervention.

Table 7 .
Cont.Four clinic-based visits of participants.Components: Counseling on Mediterranean diet, and increased consumption of EVOO and nuts; daily consumption of EVOO ≥ 40 mL, and pistachios 25-30 g for at least 3 days weekly.Free provision of 10 L of EVOO, and 2 kg of pistachios during the first and second visit.Advice against consumption of alcohol and juices.Motivation: Free of cost supplies Assessment: Questionnaires, MEDAS for assessment adoption of Mediterranean Diet; MEDAS score 0-12 points, without consideration of alcohol and juice.

Table 8 .
Eligible studies with exercise interventions.

Table 8 .
Cont.Exercise of low intensity in the beginning of the intervention, cycling for 30 min.Continuous increase in the duration until 45-60 min by adding 5 min to the cycling phases of moderate intensity or to the intervals of cycling.A warm-up and cool-down period for 5 min of each one, in the beginning, and in the end of sessions, respectively.In a group, supervised exercise program, on a choice of performance day and h.Components: Exercise program for 50-60 min with a warm-up period for 10 min, aerobic resistance exercise both for 15-20 min, and a cool-down period for 10 min.

Table 9 .
Eligible studies with dietary plus exercise interventions.

Table 10 .
Quality of reporting for eligible studies with dietary interventions.

Table 11 .
Quality of reporting for eligible studies with exercise intervention.

Table 12 .
Quality of reporting for eligible studies with diet-plus-exercise interventions.