Pregnancy

Women entering pregnancy with obesity are at increased risk of multiple adverse outcomes, including gestational diabetes (GDM), pre-eclampsia, emergency caesarean delivery, induction of labour, antenatal depression and maternal mortality^{(Reference Fakhraei, Denize and Simon11,Reference Ellis, Brown and Barger12)} . Infants also have increased risks, such as large-for-gestational-age (LGA) or macrosomia, neonatal hypoglycaemia and admission to neonatal intensive care units^{(Reference Marchi, Berg and Dencker13)}. Postnatally, there is reduced breastfeeding, and increased depression^{(Reference Achike and Akpinar-Elci14–Reference Dachew, Ayano and Betts16)} and weight retention^{(Reference Rong, Yu and Han17)}. The intrauterine environment can impact on fetal development, the results of which persist across the life course through fetal programming^{(Reference Strain, Spaans and Serhan18)}. Children born to women with obesity and/or excessive GWG are significantly more likely to develop overweight or obesity themselves^{(Reference Heslehurst, Vieira and Akhter19,Reference Voerman, Santos and Patro Golab20)} . The associations between maternal obesity and child health persist into adulthood, particularly relating to diabetes^{(Reference Lahti-Pulkkinen, Bhattacharya and Wild21)}. The increased risks may be explained by the fetal overnutrition hypothesis or due to metabolic, inflammatory and neuroendocrine changes that occur from an altered fetal hormone and nutrient exposure when mothers have obesity during pregnancy^{(Reference Taylor and Poston22,Reference Forbes, Barr and Reynolds23)} .

Preconception and interpregnancy

Following pregnancy, the postnatal period is also a preconception period for women who have future pregnancies, providing an opportunity for preconception dietary and weight management interventions^{(Reference Barker, Dombrowski and Colbourn24)}. A recent systematic review explored associations between weight change in the prepregnancy period (including both preconception and interpregnancy) and outcomes in a subsequent pregnancy^{(Reference Nagpal, Souza and Moffat25)}. Compared with weight stability, weight gain significantly increased the odds of developing subsequent GDM, hypertensive disorders, preeclampsia, LGA, caesarean delivery, stillbirth and low Apgar score, and reduced the odds of having a vaginal birth following previous caesarean^{(Reference Nagpal, Souza and Moffat25)}. Interestingly, prepregnancy weight loss significantly reduced the odds of developing GDM compared with weight stability, but increased the odds of SGA, stillbirth and preterm delivery. There were also different patterns when analysis was carried out based on maternal BMI. For example, preterm delivery was significantly increased for women with a BMI≥25 kg/m² but not for BMI<25 kg/m². Weight loss reduced the odds of LGA for women with a BMI≥25 kg/m² but increased the odds of SGA in women with a BMI<25 kg/m². Other systematic reviews exploring the interpregnancy period have found similar results^{(Reference Oteng-Ntim, Mononen and Sawicki26–Reference Timmermans, van de Kant and Oosterman28)}. The associations between prepregnancy weight loss (which is the current UK guideline recommendation) and some potential for increased risks in subsequent pregnancies require further exploration. For example, how important is the amount of weight loss, reason for weight loss (e.g. intentional or unintentional) and timescale for weight loss? These data would help inform evidence-based guidelines to optimise maternal and infant health.

Healthcare costs

Healthcare services have increased costs relating to maternal obesity due to the additional care required to minimise risks and maximise healthy outcomes for women and their babies. A study in Wales identified that women with overweight and obesity in pregnancy have increased service usage and costs of 23 % and 37 %, respectively^{(Reference Morgan, Rahman and Hill29)}. However, this was restricted to just direct costs rather than complete health costs. Other economic studies show similar results^{(Reference Watson, Howell and Johnston30–Reference Lenoir-Wijnkoop, Van der Beek and Garssen32)}. Additionally, a study in Canada reported that children born to mothers with overweight or obesity had more physician visits, hospital admissions and longer hospital stays with estimated costs in the first 18 years of life of $1415 (95 % CI 590, 2285), compared to $231 (95 % CI −403, 847) for maternal recommended weight^{(Reference Kuhle, Muir and Woolcott33)}. The additional costs required for routine care for women with maternal overweight and obesity support the need for interventions to reduce risks from both a maternal and infant health and well-being perspective, but also from an economic perspective.

Deprivation

UK data show a deprivation gradient with maternal weight status. Women have increased odds of their BMI being above or below the recommended range as the level of deprivation increases (Table 2). These data demonstrate wider determinants of maternal obesity than individual behaviour. There is potential for interventions that focus on individual agency and behaviour change alone to increase the inequality gap for maternal obesity. Alongside increasing levels of deprivation and poverty in the UK, food insecurity is also increasing. Women are particularly vulnerable to food insecurity due to working in low-income or part-time jobs, as well as societal expectations for them to be the main carers for children and family members, carry out unpaid housework and be the main providers of food^{(Reference Dolin, Compher and Oh34–Reference Ivers and Cullen36)}. When finances are scarce, women report restricting their own food intake in favour of their children and other household members^{(Reference McIntyre, Glanville and Raine37–Reference Armstrong, Hepworth and Black39)}. Food insecurity is associated with a nutritionally poor diet, consumption of energy-dense food and living in an obesogenic environment, thus increasing the risk of developing obesity^{(Reference Watson, Booth and Velardo40–Reference Nettle, Andrews and Bateson43)}. The preconception, pregnancy and postnatal periods present increased nutrient requirements, and food insecurity in pregnancy increases the risk of adverse health outcomes for both mother and child^{(Reference Augusto, de Abreu Rodrigues and Domingos44,Reference Kazemi, Masoumi and Shayan45)} . In the UK, food insecurity has increased since the implementation of austerity measures following the global financial crisis in 2008^{(Reference Dowler, Kneafsey and Lambie46–Reference Fuller, Bankiewicz and Davies48)}, further exacerbated by Covid-19^{(Reference Fuller, Bankiewicz and Davies48,Reference Loopstra49)} , and most recently in 2022 by increases in the cost of living (largely influenced by increased costs in fuel and food)^{(Reference Fuller, Bankiewicz and Davies48,50,Reference Food Standards Agency51)} (Fig. 1). The implications of continuing increases in food insecurity on maternal nutritional health and well-being, and the potential impact on fetal development, are severe.

Fig. 1.

Household food insecurity in the UK.

Note: Moderate or severe food insecurity indicates answering yes to one or more of having smaller/skipping meals, been hungry but not eaten and/or not eaten for a whole day. Reproduced using data from the Food and You survey 2018^{(Reference Fuller, Bankiewicz and Davies48)}, Food Foundation 2020, 2021, 2022^{(Reference Loopstra49,50,Reference Food Standards Agency51)} .

Table 2.

Odds of maternal BMI category and area of residence deprivation

NHS, National Health Service; NIMATS, Northern Ireland Maternity System; NMPA, National Maternal and Perinatal Audit; IMD, index of multiple deprivation.

* OR calculated from data from the NMPA for births in England, Scotland and Wales NHS maternity services between 1st April 2015 and 31st March 2017, and from the NIMATS for Northern Ireland 2011–2017^{(Reference Relph, Coe and Carroll2,Reference Kent, Cardwell and Young3)} . Bold data are statistically significant.

^† IMD quintile 1, least deprived = deprivation reference group.

^‡ Minoritised ethnic groups include the categories defined by the NMPA report as South Asian, black, mixed and other ethnic group^{(Reference Relph, Coe and Carroll2)}; and by the NIMATS as other (i.e. not white)^{(Reference Kent, Cardwell and Young3)}.

Ethnic groups

Data for England between 1995 and 2007 showed a significantly increased prevalence of maternal obesity among Black and South Asian (compared with White) ethnic groups (adjusted odds ratio (AOR) 1⋅70, 95 % CI 1⋅62, 1⋅78; AOR 1⋅72, 95 % CI 1⋅66, 1⋅79 respectively)^{(Reference Heslehurst, Sattar and Rajasingam52)}. Within the South Asian group, Pakistani women had the highest odds of maternal obesity (AOR 2⋅19, 95 % CI 2⋅08, 2⋅31), followed by Indian (AOR 1⋅49, 95 % CI 1⋅39, 1⋅60) and Bangladeshi (AOR 1⋅15, 95 % CI 1⋅06, 1⋅24). This study applied the WHO Asian-specific BMI criteria for obesity (BMI≥27⋅5 kg/m²) for South Asian women^{(Reference Consultation53)}. More recent data have been published for by the NMPA and NIMATS (Table 2). Due to the categorisation used by NIMATS (White and Other), UK-wide data can only estimate odds of BMI categories for minoritised ethnic groups compared with White. These data show a significantly increased odds of underweight, overweight and obesity class I, but significantly decreased odds of obesity classes II and III among minoritised ethnic groups (Table 2). However, data reported by the NMPA show that the pooling of heterogeneous ethnic groups masks the significantly increased odds for Black women for all obesity classes (OR ranging from 2⋅06; 95 % CI 1⋅96, 2⋅16 for class III, to 2⋅54; 95 % CI 2⋅48, 2⋅61 for class I) (Table 2). A limitation of the NMPA data is that it did not apply the WHO Asian-specific BMI criteria, which is likely to explain the difference in results compared with previously published data^{(Reference Heslehurst, Sattar and Rajasingam52)}.

Intersectionality

It is important to consider the implications of multiple inequalities through an intersectionality lens: how do the different inequalities reported in this paper (i.e., deprivation, food insecurity, ethnicity, obesity) as well as others not discussed (e.g. stigma and discrimination) collide and have a cumulative impact on pregnancy health? Recent data show the cumulative increased risks for ethnic groups and deprivation. These data relate to stillbirth, preterm delivery and fetal growth restriction, where the highest risks were seen for Black and South Asian women residing in the highest deprivation quintile^{(Reference Jardine, Walker and Gurol-Urganci54)}. However, there is limited evidence relating to maternal obesity and intersectionality, which warrants further exploration.

Preconception interventions

The preconception period could be the ideal opportunity for interventions to address risk factors for pregnancy prior to conception, based on the life course approach and embryo developmental programming^{(Reference Stephenson, Heslehurst and Hall55,Reference Hanson, Barker and Dodd56)} . However, the preconception phase is not straight-forward to define and the exact time of conceiving is often unknown as half of pregnancies in high-income countries are unplanned^{(Reference Public Health England57,Reference Bearak, Popinchalk and Ganatra58)} . While observational evidence demonstrates strong relationships between health behaviours before pregnancy and positive pregnancy outcomes, limited intervention evidence exists^{(Reference Schummers, Hutcheon and Bodnar59–Reference Aune, Saugstad and Henriksen61)}. Preconception interventions to date tend to be universal (e.g. folic acid fortification) or focus on fertility outcomes, with limited evidence specific to maternal weight^{(Reference Toivonen, Lacroix and Flynn62–Reference Price, Sumithran and Nankervis64)}. Systematic reviews published during the past decade have attempted to identify preconception intervention evidence, but little progress has been made in relation to obesity. In 2014, a systematic review explored the effectiveness of preconception interventions on maternal, new-born and child health outcomes; however, no data for maternal obesity were reported amongst the included studies^{(Reference Dean, Lassi and Imam65)}. In 2015, a Cochrane review aimed to identify preconception weight loss interventions targeting overweight or obesity, but found no eligible studies for inclusion^{(Reference Opray, Grivell and Deussen66)}. A systematic review in 2017 assessed the effect of preconception interventions on maternal and child health outcomes but again, no data on maternal obesity were reported from the included studies^{(Reference Brown, Mueller and Edwards67)}. Preconception interventions for overweight and obesity were identified in systematic reviews published in 2017 and 2018^{(Reference Lan, Harrison and Misso63,Reference Price, Sumithran and Nankervis64)} ; however, these were specific to women seeking fertility treatment. Most recently in 2021, an integrative review identified evidence on preconception care^{(Reference Cha, Smart and Braxter68)}. However, the outcomes related to women's knowledge of preconception care in relation to overweight and obesity, rather than preconception nutrition or obesity interventions^{(Reference Cha, Smart and Braxter68)}. Across all of these reviews, the evidence for preconception interventions identified tended to be poor or moderate quality, highlighting the need for further high-quality research in the preconception period. An on-going trial, Get Ready!, uses an online digital tool to identify South Asian women with a BMI≥23 kg/m² or women with a BMI≥25 kg/m² for all other ethnicities who are planning pregnancies, and provides a personalised intervention to improve behaviours and biomarkers of metabolic health^{(Reference Flynn, Pryke and Wadhera69)}. In summary, the preconception periods present theoretical promise, but there is currently a lack of high-quality interventions to inform evidence-based care.

Pregnancy interventions

There is a very different picture in relation to interventions delivered during pregnancy. Two recent systematic reviews of systematic reviews reported data for intervention effects on maternal diet and physical activity behaviours during pregnancy (n 16 reviews reporting data from 311 unique papers)^{(Reference Heslehurst, Hayes and Jones70)}, and on health-related outcomes (n 63 reviews reporting data from 675 unique papers)^{(Reference Hayes, McParlin and Azevedo71)}. For maternal behaviour outcomes, the data showed overall consistent evidence for interventions increasing fruit and vegetable consumption and decreasing carbohydrate and fat intake, and for beneficial effects on physical activity. However, there was a lack of results from meta-analysis included in the reviews due to the heterogeneity in how studies reported behavioural outcomes^{(Reference Heslehurst, Hayes and Jones70)}.

For the health-related outcomes, the data showed a consistent pattern in interventions significantly reducing GWG in intervention arms (among 80 % of meta-analyses)^{(Reference Hayes, McParlin and Azevedo71)}. When data were grouped according to intervention type, results were proportionately similar regardless of whether the interventions were diet only, physical activity only or combined. However, there was a noticeable difference in effect sizes with the largest reduction in GWG seen among diet-only interventions compared with physical activity-only or combined interventions (Table 3). Patterns in meta-analyses of postnatal weight retention were similar, with a majority showing a significant reduction. However, there were limited data to explore patterns according to intervention type. Results for most other health outcomes (e.g. GDM, pregnancy-induced hypertension, LGA) were mixed^{(Reference Hayes, McParlin and Azevedo71)}. Overall, the meta-analyses tended to show a negative direction of effect among the intervention arms, although there was minimal or inconsistent statistical significance. An example is in Table 3 for GDM, where overall only 37 % of meta-analyses showed a significant reduction in GDM among intervention arms. However, when the meta-analyses were grouped according to intervention type, there was an increase in the proportion of statistically significant results in diet or physical activity-only interventions compared with combined interventions. Similar to GWG, the largest effect sizes were seen with diet-only interventions^{(Reference Hayes, McParlin and Azevedo71)}.

Table 3.

Summary of meta-analysis results for interventions delivered during pregnancy

GDM, gestational diabetes; GWG, gestational weight gain; PA, physical activity; RR, relative risk; MD, mean difference.

* Results ranged from MD −0⋅21 kg (95% CI −0⋅34, −0⋅08) to MD −4⋅65 kg (95% CI −8⋅14, −0⋅56).

^† Results ranged from MD −1⋅56 kg (95% CI −2⋅94, −0⋅99) to MD −5⋅77 kg (95% CI −9⋅34, −2⋅21).

^‡ Results ranged from MD −0⋅36 kg (95% CI −0⋅64, −0⋅09) to MD −2⋅22 kg (95% CI −3⋅13, −1⋅30).

^§ Results ranged from MD −0⋅68 kg (95% CI −1⋅28, −0⋅09) at 12 months to MD −1⋅90 kg (95% CI −1⋅69, −1⋅12) at 6 months.

^‖ Results ranged from RR 0⋅61 (95% CI 0⋅41, 0⋅90) to RR 0⋅83 (95% CI 0⋅69, 1⋅00).

^¶ Results ranged from OR 0⋅33 (95% CI 0⋅14, 0⋅76) to RR 0⋅56 (95% CI 0⋅36, 0⋅87).

** Results ranged from RR 0⋅51 (95% CI 0⋅31, 0⋅82) to RR 0⋅74 (95% CI 0⋅57, 0⋅97).

Data from: Hayes et al.^{(Reference Hayes, McParlin and Azevedo71)}.

These data suggest that we need to understand more about why some pregnancy interventions are effective and others are not to move the evidence-base forward and identify the best strategies for improving the health of women and their babies. For example, what can we learn from the content, mode of delivery, timing and intensity of interventions that are effective compared with those that are not effective? The review data suggest that diet-only interventions may have the greatest effect, although the benefits of physical activity during pregnancy should not be dismissed. Two types of diet interventions, the Mediterranean and low glycaemic index (GI) diets, are used as examples to demonstrate the need to move our knowledge forward in this area.

Mediterranean diet during pregnancy

Mediterranean diet interventions focus on implementing a new dietary pattern rather than altering specific components. This may be beneficial as food and nutrients are consumed together rather than in isolation, which may provide a more comprehensive insight to the understanding of diet–disease relationships^{(Reference Hu72)}. Within pregnancy, two key trials have highlighted the potential benefit of the Mediterranean diet: ESTEEM and St Carlos^{(Reference Wattar, Dodds and Placzek73,Reference Assaf-Balut, García de la Torre and Durán74)} .

The ESTEEM multicentre trial in the UK recruited 1252 women with metabolic risk factors (obesity, chronic hypertension or hypertriglyceridemia) and aimed to address inequalities by targeting inner-city women^{(Reference Wattar, Dodds and Placzek73)}: 60 % of participants were Black or Asian and 69 % had obesity. The intervention consisted of a Mediterranean-style diet where women were encouraged to have a high intake of nuts, extra virgin olive oil, fruits, vegetables, non-refined grains and legumes; moderate to high consumption of fish; low intake of poultry, dairy products, red and processed meats; and avoidance of sugary drinks, fast food and food rich in animal fat. The results suggested that women in the intervention group had a 35 % reduction in odds of GDM (AOR 0⋅65, 95 % CI 0⋅47, 0⋅91), with a greater reduction among women with obesity (AOR 0⋅58, 95 % CI 0⋅40, 0⋅86). Women receiving the dietary intervention also gained less weight throughout pregnancy (mean difference [MD] −1⋅2 kg, 95 % CI −2⋅2, –0⋅2 kg), and improved dietary behaviours. These included a significantly increased consumption of olive oil, nuts, fish, pulses and preferential consumption of chicken and turkey, and significantly reduced consumption of red/processed meat and butter/margarine. However, there was no significant difference between intervention and control arms for additional health outcomes such as pre-eclampsia, SGA or stillbirth. The St Carlos trial was based in Spain and recruited 874 women^{(Reference Assaf-Balut, García de la Torre and Durán74)}. The population was different to the ESTEEM trial: two-thirds of women were White, with a mean BMI of 22⋅9 kg/m² (SD 3⋅6) in the intervention arm and 23⋅3 kg/m² (SD 4⋅0) in the control. However, the results relating to GDM were almost identical to ESTEEM (AOR 0⋅67, 95 % CI 0⋅53, 0⋅84). Additionally, this study identified improved glucose tolerance among all women, and a significant reduction in some pregnancy outcomes including reduced LGA, SGA, caesarean delivery and preterm birth.

Taken together, these two trials suggest that the Mediterranean diet may be an effective strategy for the prevention of GDM. The similar effect sizes demonstrate that the interventions are effective in very different populations and may go some way to addressing inequalities. Both trials also reported high adherence, suggesting that the Mediterranean diet is an acceptable intervention for pregnant women. The potential mechanisms for beneficial effect include the increased intake of polyphenol-rich foods, which act to improve glycaemic control through a variety of biological pathways^{(Reference Kim, Keogh and Clifton75)}. The increased intake of MUFA and n-3 PUFA may exert an anti-inflammatory response, particularly in adipose tissue, and counteract the effect of SFA which decrease insulin sensitivity^{(Reference Martín-Peláez, Fito and Castaner76)}. The increased intake of fruits, vegetables and wholegrains reduces GI and increases fibre, which both act to promote euglycaemia^{(Reference Riccardi, Rivellese and Giacco77)}. As the Mediterranean diet changes the whole dietary pattern, the beneficial effect from different aspects may compound towards a positive effect. Further research could focus on applying this dietary strategy in the context of the preconception and postnatal/interpregnancy periods.

Low GI diet during pregnancy

The UPBEAT multicentre randomised-controlled trial in the UK encouraged switching from high GI foods to low/moderate GI foods^{(Reference Poston, Bell and Croker78)}. The intervention targeted women with obesity and maternity services in deprived areas with high prevalence of Black and South Asian populations. Results showed no intervention effect on GDM or LGA; however, there was a significant reduction in GWG (MD −0⋅55 kg, 95 % CI −1⋅08, −0⋅02) and there were improvements in diet and physical activity. There was a significantly reduced dietary glycaemic load (MD −21, 95 % CI −26, −21), energy intake/day (MD −0⋅70 MJ/day, 95 % CI −0⋅96, −0⋅45), carbohydrate intake (MD −1⋅4 %, 95 % CI −2⋅2, −0⋅58), total fat (MD −0⋅88 %, 95 % CI −1⋅49, −0⋅26) and saturated fat intake (MD 0⋅85 %, 95 % CI −1⋅2, −0⋅51), and significantly increased median physical activity at 28 weeks' gestation (intervention 1836 metabolic equivalents [METS]/per week (interquartile range [IQR] 792–4158); control 1386 METS/week (IQR 639–3363), P = 0⋅001). There was also a significant difference in maternal sum of skinfolds at 36 weeks' gestation (control 125 mm (SD 27); intervention 122 mm (SD 26), P = 0⋅008) suggesting reduced adiposity among women receiving the intervention.

Despite not observing a significant intervention effect for the primary outcome, the women and children were followed up. Analysis of these data investigated the effect of maternal dysglycaemia on the infant epigenome^{(Reference Antoun, Kitaba and Titcombe79)}. A sub-set of 557 women were included in the analysis of the relationship between maternal GDM and genome-wide DNA methylation in the infant. Results from the analysis of cord blood samples indicated that maternal GDM, fasting plasma glucose and 1 and 2 h plasma glucose levels were associated with significant changes to the cord blood methylome in sites associated with cell signalling and transcriptional regulation. Women who had received the intervention appeared to display an attenuation of epigenetic changes associated with GDM, 1 and 2 h plasma glucose-associated methylation. These results suggest that low GI diet may have had an effect at epigenetic level, therefore, having a life course impact beyond pregnancy. Further analysis of 514 mothers and infants at 3 years also indicated sustained benefits of the low GI intervention. Infants born from mothers in the intervention arm had a significantly lower pulse rate (−5 bpm, 95 % CI −8⋅41, −1⋅07), which may imply a reduced CVD risk, although odds of childhood overweight or obesity were not significant at this time point (OR 0⋅73, 95 % CI 0⋅50, 1⋅08)^{(Reference Dalrymple, Tydeman and Taylor80)}. There was evidence that the positive intervention effect on maternal diet persisted 3 years postnatal with lower glycaemic load, total energy and SFA intake, and increased protein intake. These data highlight the importance of long-term follow-up to fully understand the potential for wider benefits of interventions beyond pregnancy for both maternal and child health.

Postnatal and interpregnancy interventions

As previously described, the postnatal period can be a preconception or interpregnancy period for diet interventions, and women have shown motivation for weight loss during this time^{(Reference Phelan81)}. Additionally, approximately half of women have excessive GWG with associated postnatal weight retention^{(Reference Deputy, Sharma and Kim82)}. Interventions in this period could support longer-term weight management as well as preconception weight management for women who have subsequent pregnancies. A systematic review^{(Reference Dalrymple, Flynn and Relph83)} highlighted that commencement of postnatal interventions varies, with some beginning in pregnancy while others commence at varying postnatal time points. There have been several systematic reviews and meta-analyses exploring postnatal interventions to date, which overwhelmingly related to postnatal weight loss^{(Reference Dalrymple, Flynn and Relph83–Reference Lim, Liang and Hill86)} with a lack of data on other maternal health outcomes or subsequent pregnancies.

Meta-analyses demonstrate that interventions appear to be effective at reducing weight in the postnatal period^{(Reference Dalrymple, Flynn and Relph83–Reference Lim, Liang and Hill86)}. In contrast to the evidence-base for pregnancy interventions, postnatal intervention data suggest that combined diet and physical activity interventions result in greater effect, and more consistent statistical significance, than either behaviours on their own. For example, Dodd et al.^{(Reference Dodd, Deussen and O'Brien85)} reported greater postnatal weight loss for combined interventions (MD −2⋅49 kg, 95 % CI −3⋅34, −1⋅63) than for diet (MD −1⋅82 kg, 95 % CI −2⋅19, −1⋅44) or physical activity (MD −1⋅45 kg, 95 % CI −2⋅41, −0⋅50). Similarly, meta-analyses from Lim et al.^{(Reference Lim, O'Reilly and Behrens84,Reference Lim, Liang and Hill86)} reported significantly greater weight loss from combined interventions (MD −3⋅15 kg, 95 % CI −4⋅34, −1⋅96 and −3⋅15 kg, 95 % −4⋅34, −1⋅96) than physical activity (MD −1⋅63 kg, 95 % −2⋅16, −1⋅10 and −0⋅78 kg, 95 % CI −1⋅73, 0⋅16) or diet-only interventions (MD −2⋅30 kg, 95 %CI −5⋅27, 0⋅67). When reported, postnatal weight loss was also significantly greater in interventions which targeted women with overweight or obesity compared to those targeting all women (MD −3⋅17 kg, 95 % CI −4⋅45, −1⋅89 and −1⋅04 kg, 95 % CI −1⋅74, −0⋅34 respectively)^{(Reference Lim, Liang and Hill86)}. Unlike pregnancy, the potential duration of the postnatal period is highly variable, and this is reflected in the duration of postnatal interventions. One review suggested that short/medium duration (≤6 months) interventions tended to show greater weight loss than longer durations (>12 months)^{(Reference Dalrymple, Flynn and Relph83)}. Whereas, a meta-analysis showed similar results regardless of whether weight was measured at the end of the intervention with variable durations (MD −2⋅49 kg, 95 % CI −3⋅34, −1⋅63) or at 12 months postnatal (MD −2⋅41 kg, 95 % CI −3⋅89, −0⋅93)^{(Reference Dodd, Deussen and O'Brien85)}. Thus, future interventions should further explore the optimal duration of the intervention for weight reduction.

Evidence from general populations suggests greater intervention intensity results in greater weight loss^{(Reference Perri, Nezu and Patti87,88)} , whereas a systematic review^{(Reference Lim, Liang and Hill86)} of postnatal interventions to date shows greater weight loss with low intensity and flexible interventions. Further research exploring postnatal intervention intensity and duration is required to inform the best approach for this population. Of note, we identified few postnatal weight management interventions that focused solely on women with obesity within the existing literature. As women with obesity are at greater risk of postnatal weight retention (>5 kg) and are more likely to gain two or more interpregnancy BMI units compared to women with recommended weight^{(Reference Sha, Cheng and Li89)}, there is a need for effective strategies targeting women with obesity.

Cost-effectiveness evidence for interventions

The cost-effectiveness evidence for interventions across the reproductive cycle is limited, and focused on the pregnancy and post-partum periods. An economic evaluation of diet and/or physical activity interventions in pregnancy using individual participant data from thirty trials (n 17 727 women) found no evidence that interventions were cost-effective compared to usual care^{(Reference Rogozinska, Marlin and Jackson90)}. This result remained when secondary analysis was performed based on maternal BMI. The heterogeneity between studies due to different intervention types and intensities may have reduced the accuracy of the costs imputed into the model. As discussed in the pregnancy intervention section of this paper, there appears to be differences in effect size of interventions for some outcomes depending on whether they are combined, or diet or physical activity only. Further work extending the economic evaluation to look at specific intervention components and modes of delivery, as well as expanding data on costs to allow for calculation of quality-adjusted life years (QALY), would allow for a full cost-utility analysis. A further limitation with current intervention cost-effectiveness analyses is that they tend to focus only on outcomes that have been significantly different in intervention and control arms. For example, a cost-effectiveness analysis of the LIMIT trial (an intervention targeting women with a BMI≥25 kg/m² in pregnancy) focused on respiratory distress syndrome and infant macrosomia, even though data were collected on a range of maternal and infant outcomes^{(Reference Dodd, Turnbull and McPhee91,Reference Dodd, Ahmed and Karnon92)} .

A cost-utility analysis was conducted for a postnatal weight loss intervention that was delivered to sixty-eight Swedish women for 12 weeks^{(Reference Hagberg, Brekke and Bertz93)}. The intervention targeted women with a BMI>25 kg/m² and focused on energy intake reduction. The cost-utility analysis suggested the intervention was effective; the intervention group had a cost per QALY of 8643–9758 USD and the willingness to pay per QALY was set at 50 000 USD. Another Swedish postnatal dietary intervention in 110 women found similar results^{(Reference Hagberg, Winkvist and Brekke94)}. The 12-week intervention with 2-year follow-up showed a significant improvement in quality of life compared to standard care (P < 0⋅050). The cost per QALY gained was 1704–7889 USD, and the likelihood of cost-effectiveness was 0⋅77–1⋅00 based upon a willingness to pay of 50 000 USD per QALY. These two studies suggest a beneficial effect of interventions in the postnatal period on quality of life, as well as being cost-effective. Given the small sample sizes of these trials, cost-effectiveness data from larger studies with a more diverse demographic would provide a better understanding of cost-effectiveness. A gap in the existing evidence-base is for economic evaluations to also incorporate longer-term outcomes that may be improved by interventions, such as those reported by the UPBEAT trial, and outcomes more closely aligned with public health, such as improvements in maternal behaviours. Future research could also incorporate patient and public involvement to help guide the design of economic evaluations with a focus on more holistic patient benefit beyond clinical outcomes.

Interventions and inequalities

Given the significant associations between maternal weight and inequalities, particularly deprivation, food insecurity and ethnicity, we must take a critical look at the potential for interventions to reduce the inequality gap, or to widen it. There are some existing interventions, such as ESTEEM and UPBEAT, which have specifically targeted inequalities in their study design and could have potential for addressing these inequalities. However, on the whole, the evidence-base to date does not address inequalities through targeting. The focus of interventions to date is also on individual responsibility (albeit supported) for women to improve diet, physical activity, minimise GWG, prevent adverse pregnancy outcomes and lose weight postnatally. This focus on individuals has the potential to increase the inequality gap given that individual responsibility places resource demands (e.g. financial and time) on the individuals, and can also limit potential for scale-up and implementation^{(Reference Cobiac, Vos and Veerman95,Reference White, Adams and Heywood96)} . Evidence suggests that the transition from pregnancy to postnatal may be associated with a poorer diet, and it can be challenging to identify ways to increase physical activity, restrict intakes of food and drink and cope with child rearing at the same time, especially amongst low-income women^{(Reference Craigie, Macleod and Barton97,Reference George, Hanss-Nuss and Milani98)} . As women are disproportionately affected by food insecurity, particularly those living in poverty, access to healthy food and individual agency for healthy diets is more difficult^{(Reference Ivers and Cullen36)}. Interventions which use technology-based strategies have shown potential impact on postnatal weight management, but the effectiveness of those interventions in disadvantaged groups of women is limited^{(Reference Evans, Harrington and Patchen99–Reference Colleran and Lovelady101)}. Thus, interventions also need to consider ways to reduce health inequalities more holistically, considering the wider determinants of health, and any focus on individual responsibility needs to include access to the resources required for women to make behaviour changes. Further research is required to identify how to optimise diet interventions across reproductive stages in a way that is equally accessible to all women in order to tackle inequalities. More involvement of patients and public, who have lived experience of these inequalities, in the design of interventions, possibly using co-production methods, may be one way to move this agenda forward.