Background

GDM is defined as the presence of hyperglycaemia, first detected any time during pregnancy [1]. Globally around, 20 million or 16% of all live births are affected by hyperglycaemia during pregnancy, of which more than 90% are present in LMICs [2]. Typically, GDM is diagnosed between 24 and 28 weeks of gestation [3]. However, varying degrees of hyperglycaemia may be present before this time period [4] and adverse effects on the fetus may have already happened at the time of diagnosis of GDM [5,6,7].

Maternal metabolic health during pregnancy has critical influence on the metabolic health of the offspring and possibly even in the subsequent generations [8]. There is increasing evidence that pre-conception health of women is of critical importance in shaping the metabolic health of the next generation [9,10,11]. In addition, GDM has been shown to be associated with adverse fetal programming [12]. Even in research setting, current management strategy for GDM, at best, reduces the short-term complications by about 50% [13, 14]. Thus, the focus needs to move away from “diagnosis and treatment of GDM” to “prevention of GDM.” Prevention of GDM may provide a crucial opportunity to reduce this risk of adverse metabolic programming of the offspring and future risk of cardiometabolic disorders, in addition to benefiting the mothers [15].

Lifestyle interventions are proven to effectively prevent type-2 diabetes [16]. Hence it is conceivable that such lifestyle interventions could prevent GDM. However, studies that tested the effectiveness of lifestyle interventions provided during pregnancy on GDM have shown mixed results [14, 17,18,19,20,21]. Interestingly, interventions provided during early weeks of gestation showed promising results [22]. Data from International Weight Management in Pregnancy Collaborative (iWIP) suggest that with careful selection of subjects and the type of intervention, GDM can be prevented if the interventions are carried out in early pregnancy [23]. A recent meta-analysis shows that antenatal structured diet and physical activity-based lifestyle interventions were linked to reduced gestational weight gain and lower risk of other adverse maternal and neonatal outcomes [23] and two individual participant data (IPD) meta-analysis on prevention of GDM are ongoing [24, 25]. However, this approach may still not reduce the adverse programming that may happen at or soon after the time of conception. Therefore, the best way to abolish this excess risk is by prevention of GDM with interventions before pregnancy.

Two recent studies have shown that lifestyle interventions before pregnancy may help to prevent GDM. However, these studies are small, of varying quality and used different components of “lifestyle interventions” [26, 27]. Not many have reported in LMICs, where the biggest burden of GDM is present. This systematic review aimed to summarize the available evidence from randomized control trials (RCTs), case-control studies and cohort studies for the various components of pre-pregnancy lifestyle in reducing the risk of GDM. Where possible, we conducted meta-analysis of the studies to quantify the effect of these components.

Methods

Search strategy

This is a systematic review and meta-analysis summarizing the evidence linking “pre-pregnancy” lifestyle (diet and/or physical activity aspects) and GDM risk and reporting the results of the analysis using a robust estimate, that is, Odds Ratio (OR). The PI(E)CO framework for this study is as follows: Population: Pregnant women, Intervention/Exposure: Pre-pregnancy diet and/or physical activity-based lifestyle intervention/exposure, Comparison: No pre-pregnancy diet and/or physical activity-based lifestyle intervention/exposure, Outcomes: Prevention of GDM/reduced risk of developing GDM.

The following electronic databases were searched: MEDLINE, Web of Science, Embase and Cochrane Central Register of Controlled Trials (CENTRAL). The search term combinations were designed with appropriate MeSH, free text and word variants to capture all studies on “Pre-pregnancy lifestyle interventions/components preventing Gestational Diabetes Mellitus.”

The search carried out were “[(Pregnant women OR Pregnancy OR Pre-pregnancy) AND (Lifestyle intervention OR Diet OR Exercise OR Behavioral change intervention OR Lifestyle education) AND (Gestational Diabetes OR Gestational Diabetes Mellitus OR Hyperglycaemia during Pregnancy)]”. All studies from inception till July 2022 were searched. All database searches were limited to “human” studies in order to eliminate animal model and other irrelevant studies. No Language restrictions were applied. The protocol was registered in PROSPERO (ID: CRD42020189574).

Study selection

Study inclusion

Studies that are randomized control trials (RCTs), cohort studies and case-control studies that assessed the effect of pre-pregnancy lifestyle (Diet and/or physical activity based) in preventing GDM were included.

Study exclusion

Studies involving women with type-1 or type-2 diabetes, women aged <18 and more than 50 years, women on metformin therapy (up to 6 weeks before) for anovulation and/or infertility, women with severe anaemia defined as hemoglobin (Hb) < 80 g/L, sickle cell traits, sickle cell anaemia and other genetic Hb variants and any other serious medical illness, any lifestyle intervention started only during pregnancy, drug interventions unless they had a separate lifestyle arm to prevent GDM, lifestyle intervention that have not reported incidence of GDM in their outcomes were excluded.

Two reviewers (SS and DP) independently performed the study selection. Rayyan, a web-based tool, was used to do the screening and selection. Any conflicts arising in study selection was resolved after a discussion. A third reviewer (NS) resolved any conflict arising in study selection or during quality assessment. Reference lists and gray literature were also searched to capture any unpublished data and additional studies. Following initial title and abstract screening, full-text screening was carried out.

Data extraction

Two independent authors extracted data of the selected studies. The data extracted were author, year, type of lifestyle intervention, time of lifestyle intervention/healthy lifestyle exposure, number of participants, ethnicity, country of study, mean BMI, mean age, Odds Ratio, Relative risk and other reported results. We extracted data on the number of events (GDM and non-GDM) in women exposed and not exposed to any lifestyle factors.

Risk of bias assessment

Two independent reviewers assessed the quality of the selected studies. All the following sources of bias were addressed: blinding, sequence generation, incomplete outcome data, allocation concealment, selective outcome reporting and others. The Cochrane ROB (Risk of Bias)-2 tool was used for assessing the quality of the selected randomized control trials. The Newcastle-Ottawa Scale (NOS) for assessing the quality of non-randomized studies. Based on the NOS, (0–9 for case-control studies / 0–10 for cohort studies) studies were categorized as follows: low risk (6–9/10), some concerns (3–5), high risk (0–2). The Publication bias and small studies effect quality assessment was carried out by assessing funnel plots using Egger’s tests.

Strategy for data synthesis

The outcome assessed is the risk of GDM. Random effects model using Mantel-Haenszel method was used to calculate odds ratio (OR) with 95% confidence interval from the relevant summary estimates reported by the included studies. Studies from the same study cohort are considered only once for meta-analysis but are listed in the summary table if any useful insight are provided based on components of intervention/exposure. For cohort studies that reported the events of GDM and non-GDM for various categories of diet and/or physical activity measures, studies with similar exposure group (low carb intake/low sugar diet group, higher quality diet scores, high physical activity group) were taken into consideration for meta-analysis. Cohort studies that report association of GDM and different components of diet and physical activity are included as separate studies for the purpose of this review. Statistical heterogeneity between the studies was assessed using I2 statistics. RevMan software version 5.4.1 was used for the statistical analysis.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

The initial databases search found 7935 studies and an additional 32 studies were identified from other sources (Fig. 1). Removing duplicates and abstract screening resulted in 510 studies for full-text screening and 30 were tabulated (Tables 1–4). Five of them were RCTs (n = 2471 pregnancies) [28,29,30,31,32], 4 case-control studies (n = 19,778 pregnancies) [33,34,35,36], and 21 cohort studies (n = 235,627 pregnancies). Of these, 5 were physical activity based (n = 46197 pregnancies) [27, 37,38,39,40], and 16 were diet based exploring various components (n = 189,430 pregnancies) [27, 41,42,43,44,45,46,47,48,49,50,51,52,53,54,55]. Full text of one of the RCTs was in Russian and hence only the data from the abstract was used [28]. The study characteristics based on the type and lifestyle components were summarized in Tables 1–4. Three RCTs tested a combination of diet and physical activity factors from high income countries. Two RCTs tested diet and were from low- and middle-income countries.

Fig. 1: Study selection.
figure 1

Number of studies selected as per the inclusion and exclusion criteria; PRISMA flow chart.

Full size image
Table 1 Study characteristics of the RCTs (n = 5).
Full size table
Table 2 Study characteristics of the case-control studies (n = 4).
Full size table
Table 3 Study characteristics of other cohort studies (Physical activity- PA based n = 5).
Full size table
Table 4 Study characteristics of other cohort studies (Diet based n = 16).
Full size table

Meta-analysis were carried out to assess the pooled effect size by study type and components of lifestyle measures. Figure 2a shows the effect of pre-pregnancy RCTs of lifestyle intervention. 3 out of 5 of these studies were dietary interventions. Women in the intervention group were 24% less likely to develop GDM but this was not statistically significant (OR 0.76, 95% CI: 0.50–1.17, p = 0.21). Sub-group analysis of the studies from HICs and LMICs was also performed as shown in Fig. 2a. These studies had moderate heterogeneity. Quality assessment of these RCTs revealed that only 20% of the studies had low risk of bias. The bias was mainly due to randomization process and missing outcome data (Supplementary Fig. 1).

Fig. 2: Meta-analysis of pre-pregnancy lifestyle intervention and risk of GDM.
figure 2

The 4 panels show the different components of lifestyle interventions and risk of GDM.

Full size image

Three of the case-control studies studied dietary patterns and other case-control study focused on Dietary Inflammatory Index (DII). Meta-analysis of case-control studies was not performed as 2 studies are from the same study population and the dietary exposures are very heterogenous to combine in a meta-analysis. Quality assessment of the case-control studies revealed that low risk of bias with Newcastle-Ottawa scale score 8 (Supplementary Table 1)

Meta-analysis of cohort studies (n = 4) revealed that women who were more physically active before pregnancy were 34% less likely to develop GDM (OR 0.66, 95% CI: 0.44, 0.99, p = 0.04) (Fig. 2b). Similarly, women who had low carbohydrate/low sugar diet were 24% less likely to develop GDM (OR 0.86, 95% CI 0.68–1.09, p = 0.22) (Fig. 2c). Meta-analysis (Fig. 2d) of cohort studies linking diet score and GDM showed that women with higher quality diet scores were 28% less likely to develop GDM (OR 0.72, 95%CI: 0.60, 0.87, p = 0.0006). These studies had substantial heterogeneity among them. Quality assessment of the cohort studies revealed low risk of bias with Newcastle-Ottawa scale score 8 (Supplementary Table 1). 24 of all 30 studies were from high income countries and 6 were from low- and middle- income countries (Fig. 3). Supplementary Figs 2a-2d show the funnel plots for publication bias, which reveal symmetry and show very low risk of publications bias. Supplementary Figs 3a-3d show the sensitivity analysis based on leave-one out plots confirming the robustness of the results reported.

Fig. 3: Distribution of studies across the world.
figure 3

Worldwide distribution of included studies linking pre-pregnancy lifestyle and GDM risk.

Full size image

Discussion

Our systematic review and meta-analysis summarize crucial evidence highlighting the importance of “pre-pregnancy” lifestyle in reducing the risk of developing GDM. To our knowledge, our systematic review and meta-analysis is the first to show the evidence for pre-pregnancy lifestyle interventions and the risk of GDM. Data from RCTs were limited and showed that pre-pregnancy lifestyle interventions may reduce the risk of developing GDM by about 24%. The cohort studies explored various components of diet and physical activity measures and confirmed a strong link between higher self-reported physical activity and diets that are classified as “healthy” during the pre-pregnancy period and reduced risk of GDM. Analysis of cohort studies on low carbohydrate/low sugar diet and higher quality diet scores also showed positive effect in reducing the risk of developing GDM.

A recent systematic review during pregnancy showed that structured dietary intervention can reduce the risk of GDM by about 39% in 3029 women [23] The findings from our RCTs however did not show a significant reduction in the risk of GDM in women who received lifestyle intervention before pregnancy. This may be due to the heterogeneity of the studies using different components of the lifestyle intervention and due to the smaller number of pre-pregnancy trials conducted thus far. Nevertheless, it provides crucial evidence that lifestyle intervention before pregnancy is doable. Additionally, well-designed RCTs are required to assess the benefits of different components of dietary intervention. Interventions that focussed on physical activity during early pregnancy have been shown to reduce the development of GDM [23, 27] and type 2 diabetes in at risk individuals [16, 56]. While we did not find any RCTs that focussed only on physical activity alone interventions before pregnancy, the cohort studies summarize the available evidence and highlight the need for RCTs.

Strengths and limitations

Our study used an extensive search strategy and a robust methodology to include and evaluate the different components of dietary and physical activity interventions in the pre-pregnancy period. Inclusion of non-RCTs studies, provided a comprehensive summary and highlighted the need for future RCTs on different components of lifestyle interventions before pregnancy to reduce the risk of GDM. It also highlights the paucity of evidence from LMICs where the burden of GDM is very high. Our study however has the following limitations. Firstly, there is high heterogeneity between the studies. This is likely be due to the various different lifestyle factors that are included in these studies. Secondly, most of these cohort and case-control studies were based on food frequency and physical activity questionnaires, which would result in self-reported bias. This would need to be kept in mind and future studies should ideally use objective measures to minimize this bias. Finally, not many studies were conducted from LMICs and therefore our findings were not applicable to populations who have the highest risk of GDM.

Implications for future research

Our review also highlights the paucity of data from LMICs, which may highlight the potential difficulties of conducting RCTs, especially in the pre-pregnancy period. However, high birth rate in LMICs provide other windows of opportunities such as “inter-pregnancy” interval [26]. This may also provide opportunities for targeting women at high-risk, for example those who had GDM in their previous pregnancy, which may be more cost effective.

While our review highlights the potential benefits of various components of diet and lifestyle measures, studies that focus on reducing sedentary behavior are needed. Sedentary behavior is associated with higher metabolic and cardiovascular risk [57, 58] and break in sedentary behavior has been shown to improve glucose and metabolic profile in women post-menopause [59]. This approach may work during inter-pregnancy interventions, especially in women with previous history of GDM with young children.

Finally, future interventions should be co-developed and “personalized.” It is known that women from high-risk ethnic groups have difficulty in following “generalized” lifestyle intervention [60, 61]. Thus, it is critically important that these women receive a “personalized” lifestyle intervention that relates closely to their lifestyle. Newer technologies such as mobile phone based remote interventions offer promise that these can be achieved at a lower cost [62, 63]. Face-to-face physical activity interventions can be challenging in women with young children [64] and may increase the travel cost and time burden in LMICs. In LMICs, mobile phone connections are increasing and more accessible than clean water [65]. Thus, using mobile phone technology to deliver these interventions would be pragmatic and urgently needed.