Intervention content

Most dietary interventions^{(Reference Fair and Soltani49,Reference Griffith, Alsweiler and Moore50)} promoted general healthy eating advice according to national recommendations, with emphasis on restriction of sugar intake and increased fibre intake. In women with overweight or obesity, some, but not all interventions also encouraged reduced energy intake. The majority of dietary interventions have been combined with some physical activity advice or programme. The behavioural basis for the interventions is often not described, but where it is, interventions were informed by elements from the control and/or social cognitive theories, and involving goal-setting, self-monitoring of goals and weight, problem solving and in some cases, social support and motivational strategies.

In the UK pregnancies better eating and activity trial^{(Reference Poston, Bell and Croker41)}, 1555 women with obesity, of multiple ethnicities residing in the UK, were offered from approximately 17 weeks' gestation, a complex intervention on behaviour change and self-monitoring of diet and physical activity, or standard antenatal care of lifestyle advice at the booking appointment. To improve glucose tolerance, women in the intervention group were encouraged to follow a healthy eating pattern, but to not necessarily restrict their energy intake. The dietary advice emphasised swapping carbohydrate-rich foods of medium-high glycaemic index with lower glycaemic index foods, to reduce glycaemic load of the diet, as well as reducing foods high in saturated fat. Physical activity advice focused on gradual increases in walking to reach moderate intensity, with options for additional physical activity/exercise. In the Finnish gestational diabetes prevention (RADIEL) study^{(Reference Koivusalo, Rönö and Klemetti51)}, women with a history of GDM and/or pre-pregnancy obesity were randomised to either individualised diet, physical activity and weight management counselling, or to receive leaflets with general information on diet, physical activity and weight control during pregnancy, as well as health education according to standard practice. In this study, intervention participants with obesity were advised to not gain weight during the first two trimesters. The dietary advice was based on the Nordic Nutrition Recommendations^{(Reference Becker, Lyhne and Pedersen52)}, and aimed to optimise participants' consumption of fruit and vegetables, berries, high-fibre foods, low-fat dairy products, unsaturated fats, fish and low-fat meat products, and to minimise the intake of sugary foods and drinks. Women were encouraged to follow the ‘plate model’, filling half of the plate with vegetables, one-quarter with complex carbohydrates (e.g. potato, rice, pasta) and one-quarter with protein foods (e.g. meat, fish, eggs, beans). The diet provided 6694–7531 kJ/day, with 40–50 % of energy coming from carbohydrates, 30–40 % from fat and 20–25 % from protein. Physical activity recommendations included 150 min of moderate-intensity activity per week and reduction in sedentary lifestyle. Both the diet and the activity plan in this study were monitored, and tailored to each woman's abilities, progress and preferences. For example, if a woman was struggling with physical activity, the focus would shift towards diet.

A few trials promoted a Mediterranean-style diet^{(Reference Zhang, Xia and Weng53)}. For example, ESTEEM (effect of simple, targeted diet in pregnant women with metabolic risk factors on pregnancy outcomes) was a multicentre RCT across five maternity units in the UK, in 1252 women with metabolic risk factors (obesity, chronic hypertension or hypertriglyceridaemia), which assessed whether a Mediterranean-style diet reduces adverse pregnancy outcomes compared to usual antenatal advice on diet, physical activity, weight management and hypertension in pregnancy^{(Reference Al Wattar, Dodds and Placzek54)}. The components of the diet included high intake of fruit and vegetables, non-refined grains and legumes, moderate-to-high consumption of fish, small-to-moderate intake of poultry and dairy products, low consumption of red and processed meat and minimal intake of sugary drinks, fast food and foods rich in animal fat. ESTEEM particularly emphasised increased intake of unsaturated fat from nuts (30 g/day) and extra virgin olive oil (0⋅5 l/week). The intervention package included dietary education sessions, advice on grocery shopping, ideas for recipes and healthy eating advice when eating out.

Intervention characteristics

Dietary interventions have been delivered mainly by experienced dietitians and nutritionists, or trained study nurses, and occasionally by trained health coaches or research staff. Most interventions commenced before 20 weeks' gestation, and were delivered until 34–37 weeks' gestation, or until delivery. The number of face-to-face visits has ranged from two to –eight per participant, either as part of, or in addition to routine antenatal appointments, usually supplemented with remote contacts (range two to eleven). The frequency of intervention delivery has been almost equally distributed among the studies, between every 2–3 months, monthly and weekly. Intervention sessions lasted roughly 60 min on average^{(Reference Fair and Soltani49,Reference Griffith, Alsweiler and Moore50)} .

Interventions were delivered mostly through individual face-to-face sessions, but some studies delivered remote interventions, through telephone or mobile applications. For example, Herring et al.^{(Reference Herring, Cruice and Bennett55)} delivered a telephone-based behavioural lifestyle intervention to run alongside routine antenatal appointments, including advice to reduce sugar intake, limit fast foods and foods high in fat, and control portion sizes, incorporate daily walking, and to weekly monitor weight gain to meet Institute of Medicine recommendations, supplemented with daily text messages for encouragement with the programme and enhancing self-efficacy. In the gestational weight gain and optimal wellness trial by Ferrara et al.^{(Reference Ferrara, Hedderson and Brown56)}, women in the intervention group received an adapted version of the National Diabetes Prevention Programme designed to be delivered as a mix of in-person and telephone sessions. More recently, an intervention by Ding et al.^{(Reference Ding, Gou and Guan57)}, using a mobile social media application, offered a free instant messaging platform for women during pregnancy, to interact with their HCP at any time regarding nutrition and weight management topics.

Beyond counselling, intervention packages included several materials, such as leaflets, fact sheets, newsletters, handbooks, recipe ideas, goal logbooks, diet, physical activity and weight monitoring charts and notebooks, text messages for self-monitoring and encouragement, pedometers, online websites for extra information and digital versatile discs with physical activity regimens. In the RADIEL study, participants also had access free of charge to public swimming pools, and/or guided exercise groups^{(Reference Koivusalo, Rönö and Klemetti51)}. In the ESTEEM and St Carlos Mediterranean-style interventions, key foods, such as nuts and/or extra virgin olive oil, were provided to participants in addition to counselling^{(Reference Al Wattar, Dodds and Placzek54,Reference Assaf-Balut, Garcia de la Torre and Duran58)} .

Intervention effects on gestational diabetes mellitus

An individual participant data (IPD) meta-analysis of trials of diet and physical activity interventions in women with overweight or obesity in pregnancy⁽⁵⁹⁾ showed that the intervention groups gained less weight than usual care groups (from booking appointment until before delivery), with mean difference of −0⋅70 kg and 95 % CI −0⋅92, −0⋅48 kg, and relatively low heterogeneity (I ² = 14⋅1 %), consistent across various subgroups of women based on age, parity, BMI, ethnicity and pre-pregnancy medical conditions. However, there was considerable variability in the intensity of the interventions and definitions of outcomes. The majority of trials failed to reduce GDM risk (OR 0⋅89, 95 % CI 0⋅72, 1⋅10). A potential benefit from lifestyle interventions in preventing GDM was shown only when aggregated and IPD data were combined, and not in the IPD meta-analysis alone, but the accuracy of this approach is doubtful^{(59,Reference Rogozinska, Marlin and Jackson60)} .

In a recent Cochrane overview of systematic reviews and meta-analyses, quality of evidence from preventative lifestyle RCT for GDM was reported as low, due to variations in the definitions of GDM, type and intensity of interventions and controls, sparse data on compliance, heterogeneity in patient and study characteristics and selection criteria^{(Reference Griffith, Alsweiler and Moore50)}. Overall, low quality of evidence from systematic reviews shows no benefit or harm from physical activity or exercise interventions alone, against routine antenatal care, on the risk of GDM (risk ratio 1⋅10, 95 % CI 0⋅66, 1⋅84)^{(Reference Griffith, Alsweiler and Moore50)}, although higher quality systematic reviews have demonstrated significant reductions in incidence of GDM^{(Reference Bennett, Walker and Blumfield61,Reference Davenport, Ruchat and Poitras62)} . The effect of dietary advice alone on GDM risk is also unclear. A small number of Mediterranean-style diets have shown a significant benefit compared to routine antenatal care (OR 0⋅66, 95 % CI 0⋅52, 0⋅82), with no heterogeneity^{(Reference Zhang, Xia and Weng53)}. However, overall, low quality of evidence again suggested that dietary interventions alone do not seem to confer a benefit or harm regarding GDM risk (risk ratio 0⋅60, 95 % CI 0⋅35, 1⋅04)^{(Reference Griffith, Alsweiler and Moore50)}. Low-quality evidence has also shown unknown benefit or harm of low glycaemic index diets v. moderate-high glycaemic index diets on risk of GDM (risk ratio 0⋅91, 95 % CI 0⋅63, 1⋅31)^{(Reference Griffith, Alsweiler and Moore50)}. Conversely, moderate-quality evidence suggested that a combination of diet and exercise can possibly reduce the risk of GDM (risk ratio 0⋅85, 95 % CI 0⋅71, 1⋅01), but it is unclear if these beneficial effects apply to all women or only to high-risk women (e.g. high BMI, age, high-risk ethnicity)^{(Reference Griffith, Alsweiler and Moore50)}. In addition, many lifestyle interventions lack detailed reporting about the type, intensity, content, theoretical basis and setting of the intervention, as well as cost-effectiveness analysis, which prevents their delivery in a scalable way.

The mechanism behind the success of some dietary interventions over others in preventing GDM is unknown. A meta-analysis and meta-regression suggested that key aspects which confer benefit among successful interventions are: targeting a high-risk population based on risk evaluation models (including important GDM risk factors and not just BMI), early implementation of the intervention (e.g. before 20 weeks' gestation, or ideally, in the first trimester) and attenuation of GWG through intense diet and exercise programmes^{(Reference Guo, Shu and Fu63)}. However, although pregnancy is a unique time when women are in frequent contact with their clinical care team, and are motivated to make lifestyle changes to protect the health of their baby, intensive lifestyle modifications, especially those requiring face-to-face counselling and multiple visits, may be hard to follow, can decrease treatment efficacy and may not be feasible in routine care^{(Reference Yeo, Samuel-Hodge and Smith64)}. The UK pregnancies better eating and activity trial with good attendance rates (seven of eight sessions) led to only a small difference in GWG between the groups (−0⋅42 kg from before pregnancy up to 28 weeks' gestation, intervention v. control), and an increase in exercise by 7⋅5 metabolic equivalent of task hours/week, with no differences in moderate or vigorous activity between the groups^{(Reference Poston, Bell and Croker41)}, while data have shown that an increase of 16 metabolic equivalent of task hours/week of exercise may be required as a minimum to reduce GDM, with increase in vigorous exercise when appropriate^{(Reference Zhang, Solomon and Manson65)}. Trials that have shown effectiveness in GDM prevention so far have included intensive face-to-face dietetic support and/or provision of key foods to study participants, but they provide no indication of how these strategies could be delivered in routine antenatal care^{(Reference Koivusalo, Rönö and Klemetti51,Reference Al Wattar, Dodds and Placzek54,Reference Assaf-Balut, Garcia de la Torre and Duran58,Reference Bruno, Petrella and Bertarini66,Reference Petrella, Malavolti and Bertarini67)} . The RADIEL study reduced GDM incidence by 39 % in women at high risk, but required extra face-to-face visits and a group session to deliver standard healthy eating and physical activity advice. This may not be manageable in routine practice, given the amount of appointments women are already required to attend during pregnancy, and competing priorities they face (e.g. childcare, work)^{(Reference Koivusalo, Rönö and Klemetti51)}. The ESTEEM and St Carlos trials implemented Mediterranean-style dietary advice supplemented with extra virgin olive oil and nuts, provided as part of the trial, from early or mid-pregnancy, and showed a similar reduction in GDM risk^{(Reference Al Wattar, Dodds and Placzek54,Reference Assaf-Balut, Garcia de la Torre and Duran58)} . Nevertheless, wide scale implementation of a Mediterranean-style diet in certain countries such as the UK could be challenging. Studies outside pregnancy show that people have a favourable attitude towards this type of diet, but there are a number of barriers to adoption. The diet exists in many different versions, is perceived in various ways and knowledge of its composition is limited^{(Reference Huo, Du and Xu68,Reference Bloomfield, Koeller and Greer69)} . This diet may also not be acceptable to people of all ethnic backgrounds, and the cost of foods such as extra virgin olive oil is higher^{(Reference Bush H, Bradby and Anderson70–Reference Moore, McEvoy and Prior72)}.

Data also support the effectiveness of technology-assisted interventions for weight control, offering convenience, and facilitating engagement, retention and delivery at high intensity, but at low cost^{(Reference Bennett, Herring and Puleo73–Reference Bennett, Foley and Levine75)}. For example, in women with established GDM, telehealth has been shown to increase efficiency and to improve perinatal outcomes^{(Reference Ferrara, Hedderson and Brown56)}. Herring et al.^{(Reference Herring, Cruice and Bennett55)}, in a high-intensity preventative trial in women of African descent, with overweight and obesity, tested an inexpensive technology-assisted behavioural intervention requiring little health coaching, involving regular contact with HCP via text messaging and phone calls, as well as social support through a social media forum, and showed that the intervention significantly reduced GWG and prevalence of excess GWG, compared to usual care (also involving guidance on optimal GWG). The benefits in GWG did not translate into reduction in GDM incidence; however, this study was not statistically powered to detect such difference. Ding et al.^{(Reference Ding, Gou and Guan57)} used social software to assist with delivery of the intervention, but this was still intense, as it involved interaction with HCP at any time for personalised counselling and information, in addition to general education, daily exercise plan and weekly messaging from HCP. In this study, the incidence of GDM was significantly lower in the intervention group (24 %) than in the control group (37⋅8 %). However, reported differences in rates of GDM between the two groups in this study were unadjusted for known confounders, which could have changed the direction or significance of the results.

In view of the aforementioned, lack of clarity about effective components and mechanisms of complex diet and/or physical activity interventions remains. What the results of some successful interventions suggest however could be that dietary modification with or without physical activity can have an effect on GDM risk, independent of GWG control, since the weight gain difference between the two trial arms was only −1⋅2 kg in the ESTEEM^{(Reference Al Wattar, Dodds and Placzek54)}, and −0⋅5 kg in the RADIEL study^{(Reference Koivusalo, Rönö and Klemetti51)}, in favour of the intervention. But the optimum dietary composition is unknown. In the RADIEL study, there was only a small improvement reported in adherence to the diet recommendations in the intervention group compared with the control group, as indicated by a diet quality score^{(Reference Koivusalo, Rönö and Klemetti51)}. In the Mediterranean dietary interventions, it was speculated that increased unsaturated fat and polyphenol consumption might have been beneficial with regards to insulin sensitivity and inflammation^{(Reference Al Wattar, Dodds and Placzek54,Reference Assaf-Balut, Garcia de la Torre and Duran58)} . However, a reduction in dietary carbohydrate as a natural result of higher fat and protein consumption, under conditions when energy intake remains similar, may have also played a role. Indeed, successful interventions to date included some advice for reducing sugar consumption and/or moderating starchy foods, but detailed dietary data are not reported in most of them, so as to explore this. In the recent remote intervention by Ding et al.^{(Reference Ding, Gou and Guan57)}, both groups were consuming a diet high in carbohydrates (average of 283 and 277 g/day in the intervention and control groups, respectively), from baseline until the point of testing for GDM, with no difference between the groups, throwing doubt on the need for carbohydrate reduction, though again notably, analyses were not adjusted for confounders. The authors attributed the benefit in GDM risk to early implementation before 12 weeks' gestation, and to the higher than usual frequency of contact of women with their HCP, although they do not report engagement data. One RCT specifically aimed at reducing total carbohydrate intake as part of an intensive healthy eating intervention^{(Reference Simmons, Devlieger and van Assche76)}. In this study, women were randomised to receive healthy eating alone, or physical activity advice alone, both, or usual care, and as part of the dietary advice, they were encouraged to eat more non-starchy vegetables, and reduce intake of both starchy carbohydrates and sugary foods and drinks. Women in any of the intervention arms did not significantly limit their GWG compared to usual care, before the time point of GDM diagnosis (<20 weeks' gestation until 24–28 weeks' gestation). This study was underpowered to detect difference in GDM, and although it resulted in significant reduction in total carbohydrate intake in favour of the healthy eating only group compared to usual care at 24–28 weeks' gestation, the degree of reduction was small (−4⋅8 portions/week, <1 portion/day). Thus, the effect of a reduced-carbohydrate diet as a preventative strategy for GDM remains to be investigated.