Maternal obesity: prevalence, cost, associated risk

Obesity in pregnancy is a significant contributor to maternal and perinatal morbidity and mortality with a rising global prevalence among reproductive-aged women. The global prevalence of obesity has significantly increased in the past decades, described as a ‘global epidemic’,^{(Reference James7)} with the number of people with overweight and obesity tripling between 1975 and 2016⁽⁸⁾. Less than half of pregnant women in the UK have a BMI in the normal range in the UK and 47 % of pregnant women have GWG greater than Institute of Medicine recommendations^{(Reference Relph9)}. Moreover, internationally over half of all reproductive-aged women are entering pregnancy overweight and half exceed recommended GWG^{(Reference Maxwell, Shirley and O'Higgins10)}.

Gestational diabetes (GDM), new onset of diabetes in pregnant women without a prior history, is one of the most common complications associated with obesity in pregnancy^{(Reference Adam, McIntyre and Tsoi11)}. It is defined as high blood sugar levels during pregnancy and is an increasing health problem for both mothers and babies, affecting up to 18 % of pregnancies worldwide. European GDM prevalence varies – north 8⋅9 %, east 31⋅5 %, south 12⋅3 % and west 10⋅7 %^{(Reference Paulo, Abdo and Bettencourt-Silva12)}. High BMI (>25 kg/m²) is the most important modifiable risk factor for GDM, with 6⋅8-fold higher risk for BMIs >30 kg/m² compared to normal BMIs (18⋅5–24⋅9 kg/m²)^{(Reference Paulo, Abdo and Bettencourt-Silva12)}. The effects of GWG, GDM, hyperglycaemia and obesity on adverse pregnancy outcomes and on fetal and offspring development are significant. The risk of miscarriage and congenital anomalies has been shown to increase with hyperglycaemia during organogenesis. GDM, hyperglycaemia, obesity and GWG are associated with increased odds of pregnancies affected by neural tube defects (OR, 1⋅87), spina bifida (OR, 2⋅24), cardiovascular anomalies (OR, 1⋅30), septal anomalies (OR, 1⋅20), cleft palate (OR, 1⋅23), cleft lip and palate (OR, 1⋅20), anorectal atresia (OR, 1⋅48), hydrocephaly (OR, 1⋅68), limb reduction anomalies (OR, 1⋅34)^{(Reference Stothard, Tennant and Bell13)} intellectual disability and poorer cognitive development^{(Reference Torres-Espinola, Berglund and García-Valdés14)}. GDM is also associated with a long-term tenfold increased maternal risk of type 2 diabetes (T2D)^{(Reference Vounzoulaki, Khunti and Abner15)}. Strong evidence exists that health behaviour change can reduce the development of T2D in people at risk^{(Reference Lindström, Peltonen and Eriksson16,Reference Ratner, Goldberg and Haffner17)} , but there are many demands on a new mother and weight gained during pregnancy is frequently not lost afterwards leading to increased risk of obesity, heart disease and diabetes^{(Reference Li, Song and Li18,Reference Song, Lyu and Li19)} . This risk is not confined to just the woman, early fetal programming means that her offspring also have demonstrable increased risk – her children are eight times more likely to develop diabetes or pre-diabetes by early adulthood^{(Reference Holder, Giannini and Santoro20)} with a 59 % increased risk of developing childhood obesity^{(Reference Patro Golab, Santos and Voerman21)}.

Women with GDM and those living with obesity demonstrate increased insulin resistance in pregnancy and this creates poorer metabolic health, which can even impact placental structure, maternal and cord inflammatory markers as well as endocrine and inflammatory gene expression^{(Reference Musa, Salazar-Petres and Arowolo22)}. This common footing between GDM and obesity results in women living with obesity being at increased risk of GDM, preeclampsia, gestational hypertension, fetal macrosomia, caesarean section and postpartum weight retention^{(Reference Heslehurst, Rankin and Wilkinson23)}. Furthermore, obesity and GWG in pregnancy increase the risk of complications during labour and birth^{(Reference Goldstein, Abell and Ranasinha24)}. Evidence has demonstrated that women with overweight (BMI>25 kg/m²) were more likely to have a slower labour progression, potentially due to the inadequacy of uterine contractions, and fetal distress and ultimately therefore receiving interventions such as labour induction/augmentation or operative birth^{(Reference Leddy, Power and Schulkin25)}. Women with obesity who undergo caesarean birth are additionally at increased risk for anaesthesia-related complications (epidural failure, aspiration under general anaesthesia and difficult endotracheal intubation) and post-operative wound infection^{(Reference Taylor, Dominguez and Habib26)}. Venous thromboembolism is a further serious risk in pregnant women with obesity. Evidence has demonstrated that up to 57 % of women in the UK who died from venous thromboembolism during pregnancy had BMIs in the obese category^{(Reference Knight, Bunch and Tuffnell27)}. Fetal macrosomia is a neonatal complication associated with obesity in pregnancy^{(Reference Goldstein, Abell and Ranasinha24)} and GDM and itself increase the risk for operative delivery and maternal and infant morbidity. It is associated with maternal complications such as genital tract lacerations, and postpartum haemorrhage. Infants have an increased risk of shoulder dystocia, clavicular fractures, brachial plexus injuries and nerve palsies. Importantly, GDM is associated with a high risk of neonatal intensive care unit admission, due to further associated complications such as neonatal hypoglycaemia^{(Reference Ku, Kim and Im28)}.

Maternal BMI influences maternal and neonatal morbidity, the number and duration of maternal and neonatal admissions and health service costs^{(Reference Denison, Norwood and Bhattacharya29)}. The most recent Mothers and Babies: reducing risk through audits and confidential enquiries across the UK (MBRRACE-UK) report identified maternal obesity as a significant factor in up to 30 % of maternal deaths in the UK and Ireland^{(Reference Knight, Bunch and Tuffnell27)} and this rate is seen in other countries^{(Reference Mariona30,Reference Poobalan, Marais and Aucott31)} . Rising obesity within the obstetric populations will mean that this factor will only continue to grow and impact parents, families and the wider society. The prevalence of stillbirth in the UK is above the European average, affecting almost 1 in 250–300 pregnancies after 28 weeks of pregnancy^{(Reference Flenady, Wojcieszek and Middleton32)}. The MBRRACE report concluded that up to 60 % of antepartum stillbirths could have been prevented with improved antenatal care^{(Reference Knight, Bunch and Tuffnell27)}.

Even modest increases in maternal BMI are associated with increased risk of fetal death, stillbirth and neonatal, perinatal and infant death. For BMIs of 25 and 30 kg/m², the absolute risk per 10 000 pregnancies for fetal death are 82 and 102; for stillbirth 48 and 59 and for perinatal death 73 and 86^{(Reference Aune, Saugstad and Henriksen33)}. For women who gain four or more BMI units between pregnancies, their risk is 55 % higher for stillbirth and 29 % higher for infant mortality^{(Reference Cnattingius and Villamor34)}. We also know that stillbirth risk increases linearly with increased BMI gain and that weight loss prior to a subsequent pregnancy in women with overweight will decrease neonatal mortality^{(Reference Cnattingius and Villamor34)}. The risk of late stillbirth is much greater when GDM is not diagnosed – 44 % increase in women at risk of GDM but not screened and women with raised fasting plasma glucose not diagnosed with GDM experienced a fourfold greater risk of late stillbirth than women with normal blood glucose^{(Reference Stacey, Tennant and McCowan35)}. The increased focus on detection and management was a key recommendation of the MBRRACE report to decrease stillbirths^{(Reference Knight, Bunch and Tuffnell27)}. Eleven per cent of neonatal deaths are attributed to maternal overweight and obesity (Fig. 1)⁽³⁶⁾.

Fig. 1.

Comparison of the percentages of women with a pre-pregnancy BMI ≥30 from 2010 and 2015 (risk ratios and 95 % CI). Pooled random-effects model estimate 1⋅15 (95 % CI 1⋅08, 1⋅22). Adapted from European Perinatal Health Report 2018⁽³⁶⁾.

Inappropriate GWG is a challenge that weighs heavily on this population but they are also the most nutritionally vulnerable. Suboptimal micronutrient (iron, vitamin D, folate, vitamin B6, magnesium, zinc, potassium and vitamin A) and macronutrient (fibre and carbohydrate) intakes alongside excessive sodium and dietary fat intakes are seen in these women^{(Reference Dubois, Diasparra and Bedard37)}. Dietary quality is known to continue to be poor in this population postpartum^{(Reference Morrison, Koh and Lowe38)} and yet it is known that higher dietary quality in pregnancy and lactation are associated with a healthier growth pattern in infants^{(Reference Tahir, Haapala and Foster39)} and better weight maintenance^{(Reference O'Reilly, Versace and Yelverton40)}. Maternal nutrition and interventions to reduce maternal complications from suboptimal nutrition are priorities for the WHO⁽⁴¹⁾ and the International Federation of Gynaecology and Obstetrics. The International Federation of Gynaecology and Obstetrics recently established a specific pregnancy non communicable disease (NCD) prevention committee which aims to tackle obesity and GDM as one of its cornerstone aims^{(Reference Maxwell, Shirley and O'Higgins10,Reference Adam, McIntyre and Tsoi11)} . Pregnancy and infant first year of life are a window of opportunity to future health and the long-term effects of obesity in pregnancy needs consideration. Compared to normal-weight women, women with obesity were shown to retain more weight postpartum. More specifically, postpartum weight gain was most strongly associated with weight gain during the first trimester^{(Reference Walter, Perng and Kleinman42)} and intrauterine exposure to maternal obesity can lead to adverse health outcomes in the offspring, including an increased incidence of metabolic syndrome and obesity in the child. Recent studies have shown that childhood obesity can be carried into adulthood, suggesting that fetal overnutrition can adversely affect the health of offspring throughout life^{(Reference Voerman, Santos and Patro Golab43)}. Offspring of pregnant women with obesity have 35 % increased all-cause mortality and 29 % increased rates of hospital admission with CVD^{(Reference Reynolds, Allan and Raja44)}.

Maternal obesity, GWG and GDM place a substantial economic burden on healthcare systems. Pregnant women with overweight or obesity have a significantly greater number of maternal admissions, longer admissions and higher health service costs than women of normal weight^{(Reference Denison, Norwood and Bhattacharya29)}. Excess GWG will significantly increase risks and costs in pregnancy^{(Reference Broekhuizen, Simmons and Devlieger45)}. Moreover, infants born to mothers with high BMIs also utilise significantly more health service resources in the first year of life compared to infants born to mothers of healthy weight^{(Reference Morgan, Rahman and Hill46)}; they are also at higher risk of developing childhood obesity^{(Reference Patro Golab, Santos and Voerman21)}. Pregnant women with elevated BMI are at high risk of developing GDM, which also imposes additional costs independently of BMI^{(Reference Danyliv, Gillespie and O'Neill47)}. The cost of managing GDM yielded an economic burden in the United States of $1⋅6 billion in 2017 with $5800 annual burden per case of GDM^{(Reference Dall, Yang and Gillespie48)}. In Ireland, the costs of maternity care for women with a diagnosis of GDM are 34 % greater than in women without GDM^{(Reference Gillespie, Cullinan and O'Neill49)}. Finally, women who develop GDM have a tenfold increased future risk of developing T2DM^{(Reference Vounzoulaki, Khunti and Abner15)} and an increase in health-related costs is seen postpartum compared with normoglycaemic pregnancies (€680⋅50 in annual healthcare costs 2–5 years after the index pregnancy)^{(Reference Danyliv, Gillespie and O'Neill50)}. Health services have seen an almost fourfold increase in GDM incidence caused by a widening of the diagnostic criteria, growing obesity rates and advancing maternal age^{(Reference Hanna, Duff and Shelley-Hitchen51)}. In this GDM tsunami, health services are not resourced to manage such numbers and already extended services are further diluted. Currently, health systems around the world do not have sufficient resources to manage the numbers of women at risk of developing GDM to support them reducing their health behaviour-related risks.

Adverse outcomes in pregnancy for both woman and child are socially patterned with greater risk present in low socioeconomic and education, rural and ethnically diverse and minority populations^{(Reference Knight, Bunch and Tuffnell27)}. Rates of obesity, T2DM and GDM are also higher in these populations^{(Reference Collier, Abraham and Armstrong52–Reference El-Khoury Lesueur, Sutter-Dallay and Panico54)} and intrinsically linked to the higher rates of adverse outcomes seen. Women from disadvantaged communities are currently not engaging or minimally engaging with health services to reduce their GWG and their risks of GDM, T2DM and obesity. Disadvantaged communities will typically not have access to support that fits with their needs owing to linguistic or cultural issues, competing interests of working life and raising a family and lack of financial resources to engage with provided services^{(Reference Dennison, Fox and Ward55)}. This systematic discrimination means that the most vulnerable are not able to engage with the care that will positively influence their health. While this will be for a range of reasons, unhealthy lifestyle behaviours promoting weight gain and access to universal screening for GDM is a part of this problem^{(Reference Knight, Bunch and Tuffnell27)}. Similarly in the postpartum period, women are not engaging in diabetes screening^{(Reference Herrick, Keller and Trolard56,Reference Boyle, Versace and Dunbar57)} and/or risk reduction programmes^{(Reference Dasgupta, Terkildsen Maindal and Kragelund Nielsen58,Reference Nielsen, Kapur and Damm59)} , both of which will influence their reproductive health. Guidelines for improved health outcomes in both the woman and child are in place nationally and internationally for healthy weight gain, physical activity and eating in pregnancy, health weight management and lifestyle behaviours postpartum and healthy infant feeding practices. Yet the daily challenge that presents in maternity services and public health settings is how to implement these guidelines with fidelity using the few resources available within those settings.

Implementation science approaches: frameworks, hybrid designs and health behaviour change

A key issue in the development of effective interventions to support weight management during pregnancy and postpartum is the evidence to implementation (‘know-do’) gap. It has been estimated that it takes 17 years for evidence to be adopted into practice^{(Reference Morris, Wooding and Grant60)}. The growing field of implementation science aims to use evidence and theory-informed approaches to address this gap. The UK Medical Research Council (MRC) updated its framework for the development and evaluation of randomised controlled trials for complex interventions to improve health in 2019 and 2021^{(Reference Craig, Dieppe and Macintyre61,Reference Skivington, Matthews and Simpson62)} . Intervention complexity can impact implementation and achieving effect in different settings, due to challenges relating to: standardising intervention design and delivery, sensitivity to local context, the people involved (i.e. staff and patients), organisational context and development of outcome measures and evaluation^{(Reference Datta and Petticrew63)}. Intervention complexity has been defined in a number of different ways, including the number of interacting components, groups or organisational levels targeted, variability of outcomes, the degree of tailoring or flexibility of the intervention and whether the intervention has a non-linear causal pathway^{(Reference Skivington, Matthews and Simpson62,Reference Petticrew64)} . There is increased interest in systems thinking and conceptual mapping approaches in healthcare research to understand real-world complexity^{(Reference Baugh Littlejohns, Near and McKee65)}, by focusing on ‘people, processes, activities, settings and structures and the dynamic relationships between them’^{(Reference Wutzke, Morrice and Benton66)}. A key update within the new MRC framework is a shift towards addressing how interventions interact with their context, and how the intervention interacts with systems change, to identify the conditions needed to achieve intended change mechanisms, and to ensure effectiveness in ‘real-world’ settings^{(Reference Skivington, Matthews and Simpson62)}.

The MRC framework specifies that interventions should be systematically developed and evaluated using evidence and theory. Four phases are outlined, which can be addressed iteratively and in any order: (i) development/identification of the intervention, (ii) exploration of feasibility and acceptability, (iii) evaluation, (iv) implementation. Each component should address a number of core elements, relating to contextual considerations, programme theory, stakeholder engagement, identification of key uncertainties, refinement of the intervention and economic considerations^{(Reference Skivington, Matthews and Simpson62)}. The authors highlight that early consideration of intervention implementation and investigation of it within each development and evaluation phase can increase potential for its future adoption across settings. The MRC approach aligns with effectiveness-implementation hybrid designs^{(Reference Curran, Bauer and Mittman67)}, where evaluation of implementation and effectiveness are undertaken alongside one another. There are three types of hybrid design approaches and the balance between implementation and effectiveness will vary for each type. Hybrid type 1 primarily tests the effects of an intervention while observing and gathering information on implementation. A type 2 hybrid design involves dual testing of the intervention effectiveness and either a secondary aim or co-primary aim of testing implementation strategies. A hybrid type 3 primarily tests the implementation strategy while observing and gathering information on clinical intervention effectiveness^{(Reference Curran, Bauer and Mittman67)}.

The MRC framework highlights that interventions should be systematically developed using the best available evidence and appropriate theory. Theoretical frameworks are commonly used in the development stage to develop behaviour change interventions. The behaviour change wheel is commonly used framework to ensure complete and cohesive coverage^{(Reference Michie, van Stralen and West68)}. Central to the behaviour change wheel are three core behavioural determinants (capability, opportunity and motivation), which have been organised into the COM-B model of behaviour change^{(Reference Michie, van Stralen and West68)}. Used in parallel with the COM-B model, the behaviour change technique (BCT) taxonomy is a consensus-informed transparent and systematic means to specify the content of interventions in terms of 93 distinct BCT^{(Reference Michie, Richardson and Johnston69)}. This was derived from the behaviour change wheel development^{(Reference Michie, van Stralen and West68)} and a taxonomy for physical activity and nutrition BCT^{(Reference Michie, Ashford and Sniehotta70)}. The BCT taxonomy provides a common language to describe the key ingredients in behaviour change interventions, which enables greater replicability and fidelity of intervention implementation across settings/contexts, and facilitates systematic evidence syntheses that aim to identify the most effective BCT for a given behaviour and/or context.

Implementation theories provide a framework to drive implementation strategies and to explore how or why interventions are successfully or unsuccessfully implemented^{(Reference Nilsen71)}. Current implementation theories have differing aims, including guiding the process of applying evidence in practice, exploring what influences implementation, and evaluating intervention implementation^{(Reference Nilsen71)}. The exploration, preparation, implementation, sustainment (EPIS) framework^{(Reference Aarons, Hurlburt and Horwitz72)}, and normalisation process theory (NPT)^{(Reference May, Finch and Mair73)} fall within the first and second categories, and support exploration of barriers and facilitators to implementation. The EPIS framework^{(Reference Aarons, Hurlburt and Horwitz72)} was developed from the literature on implementation in the public sector and allied health services. It has four implementation phases that describe the process through which an evidence-based practice (EBP) is adopted: exploration (consideration of the health needs of patients/communities and identification of best EBP), preparation (identification of potential contextual barriers and facilitators to implementation), implementation (EBP is adopted), sustainment (context structures, processes and supports are ongoing so that the EBP is delivered to achieve intended impacts). Within the EPIS framework, common and unique factors hypothesised to have a strong influence on implementation of EBP are described, from within the outer system context, the inner organisational context and factors related to the innovation itself^{(Reference Aarons, Hurlburt and Horwitz72)}. Outer and inner context, and innovation factors, may be more or less important in the different EPIS phases. A key component of the EPIS framework is the recognition of the interconnectedness and relationships between outer and inner contexts, which are called bridging factors. Systematic reviews show EPIS as a flexible and robust implementation framework suitable for use across low-, middle- and high-income countries^{(Reference Moullin, Dickson and Stadnick74)}. NPT^{(Reference May, Finch and Mair73)} is an explanatory model that provides a means to understand what influences implementation, and can be used alongside EPIS to understand some of the contextual- and innovation-related factors that influence it^{(Reference May, Cummings and Girling75)}. It was developed within a variety of healthcare systems and looks at individual and collective behaviour shown to be important in empirically studied implementation processes^{(Reference May, Cummings and Girling75)}. As an action theory, it describes the mechanisms of social action involved in implementing a new practice. There are four areas to NPT which are coherence (making sense of the new practice/s), cognitive participation (buy-in), collective action (resourcing) and reflexive monitoring (appraisal and feedback)^{(Reference May, Finch and Ballini76)}. A systematic review of NPT identified it as a useful theory in a wide range of interventions and that it had a particular benefit in evaluation and understanding implementation as a dynamic process^{(Reference May, Cummings and Girling75)}.

Evaluation is critical to understanding implementation and there are several frameworks to inform how evaluation is undertaken. The most commonly used are reach, effectiveness, adoption, implementation, maintenance (RE-AIM)^{(Reference Glasgow, Vogt and Boles77,Reference Glasgow, Harden and Gaglio78)} , template-for-intervention-description-and-replication checklist^{(Reference Hoffmann, Glasziou and Boutron79)} and the penetration-implementation-participation-effectiveness (PIPE) framework^{(Reference Pronk80)}. RE-AIM was originally developed to consistently report implementation of innovations, however more recently it has been used to inform programme planning^{(Reference Glasgow, Harden and Gaglio78)}. It proposes five steps to translate research into action, each of which can be evaluated to understand intervention impact or targeted with tools and strategies to achieve implementation: reach (number of people willing to participate in the intervention or programme), effectiveness (impact of intervention on key outcomes), adoption (willingness of agents (settings) willing to initiate a programme), implementation (fidelity of delivery an use of intervention strategies) and sustainment (extent to which it becomes part of routine practices and policy). Template-for-intervention-description-and-replication is the template for intervention description and replication, which has twelve items that form a checklist to improve reporting of interventions so they can be replicated and implemented^{(Reference Hoffmann, Glasziou and Boutron79)}. The items are brief name, why, what (materials), what (procedure), who provided, how, where, when and how much, tailoring, modifications, how well (planned) and how well (actual)^{(Reference Hoffmann, Glasziou and Boutron79)}. The template-for-intervention-description-and-replication checklist is commonly used in systematic review to deconstruct interventions for implementation strategy analyses and meta-analyses^{(Reference Lim, Liang and Hill81)}. Similarly, the PIPE impact metric^{(Reference Pronk80)} provides a framework to assess the implementation of health improvement programmes. The four PIPE elements are: (1) penetration of the programme into the population of interest, (2) implementation of the proposed intervention/programme/services, (3) participation in the programme and (4) effectiveness in generating expected outcomes. PIPE can be used to evaluate implementation and provide feedback about where to focus changes to improve performance of a programme^{(Reference Aziz, Absetz and Oldroyd82)}.

Interventions to reduce excessive gestational weight gain and intrapartum weight retention: what works and what needs consideration for implementation

There is a need to take a holistic view of the best time to intervene for reducing excessive GWG and intrapartum weight retention and the evidence is conflicting. On one hand, addressing maternal BMI was shown to be the only preventative strategy that reduces childhood obesity^{(Reference Patro Golab, Santos and Voerman21)} and recent systematic review found no evidence that maternal dietary and/or health behaviour change intervention during pregnancy alone modifies early childhood obesity risk^{(Reference Louise, Poprzeczny and Deussen83)}. Interpregnancy weight gain is also associated with increased rates of subsequent large for gestational age infants and higher rates of subsequent GDM, further emphasising the importance of the interpregnancy and postpartum periods in weight management^{(Reference Maxwell, Shirley and O'Higgins10,Reference Crosby, Walsh and Segurado84)} . Conversely, there is a clear need to target and achieve engagement in women with the most risk when they are most motivated to manage their health risk during pregnancy. Health behaviour change interventions during this time show improved dietary patterns, exercise habits and GWG with improved pregnancy outcomes. Level 1 evidence clearly demonstrates that antenatal health behaviour change interventions are effective in reducing GWG, adverse maternal and neonatal birth outcomes and reduced subsequent development of T2D^{(Reference Teede, Bailey and Moran85,Reference Cantor, Jungbauer and McDonagh86)} . A recent systematic review and meta-analysis of 117 randomised clinical trials with over 34 000 pregnancies showed that antenatal diet and physical activity-based interventions were associated with less GWG^{(Reference Teede, Bailey and Moran85)}. They reduced GWG (−1⋅15 kg; 95 % CI −1⋅40, −0⋅91), GDM risk (OR, 0⋅79; 95 % CI 0⋅70, 0⋅89) and total adverse maternal outcomes (OR, 0⋅89; 95 % CI 0⋅84, 0⋅94) when compared to routine antenatal care^{(Reference Teede, Bailey and Moran85)}. Interventions that combined diet and physical activity and behaviour change had the greatest impact on GWG^{(Reference Teede, Bailey and Moran85)}. Rates of women exceeding GWG recommendations (based on IOM criteria) were significantly lower after exercise-only and combined interventions during pregnancy. A separate systematic review also showed that when behavioural therapy supported combined diet and physical activity interventions, GWG was significantly reduced (standardised mean difference −0⋅16 kg; 95 % CI −0⋅28, −0⋅04, four trials, n = 2132)^{(Reference Behnam, Timmesfeld and Arabin87)}.

From an implementation perspective, the evaluation of existing pregnancy health behaviour change interventions (n = 117) using PIPE and template-for-intervention-description-and-replication frameworks leads to some interesting findings. Only 14 % interventions provided sufficient data to calculate penetration and where available the level of population penetration was low^{(Reference Bahri Khomami, Teede and Enticott88)}. Implementation was reported with moderate fidelity and participation was variable^{(Reference Bahri Khomami, Teede and Enticott88)}. The effectiveness analysis showed −1⋅15 kg GWG (95 % CI −1⋅4, −0⋅91)^{(Reference Bahri Khomami, Teede and Enticott88)}. Allied healthcare professionals were the most common and most effective delivery agents (−1⋅36 kg GWG)^{(Reference Harrison, Bahri Khomami and Enticott89)}. Structured diet health behaviour change delivered individually was the most effective format (−3⋅91 kg GWG) and having a moderate number of sessions was significant in reducing GWG (−4⋅91 kg)^{(Reference Harrison, Bahri Khomami and Enticott89)}. Further cost-effectiveness analysis of systematic review data showed that structured diet and physical activity intervention would see $4⋅75 Australian dollars returned for every dollar invested by health funders, with cost offsets from reduced perinatal morbidity and adverse events exceeding intervention costs^{(Reference Lloyd, Morton and Teede90)}. For the postpartum period, the penetration remained low and participation variable. The implementation fidelity was low and effectiveness analysis showed −2⋅3 kg weight loss with intervention^{(Reference Lim, Liang and Hill81)}. Healthcare professional delivery remained more effective (−3⋅22 kg) compared with −0⋅99 kg non-healthcare professional delivery^{(Reference Lim, Liang and Hill81)}. Diet and physical activity combined was more effective than physical activity intervention alone but the intensity (duration or number of sessions) and setting (group or individual) did not affect weight loss^{(Reference Lim, Liang and Hill81)}. The most effective BCT strategies for greater reductions in energy intake were problem solving, goal setting, feedback on behaviour, self-monitoring, credible source, behavioural substitution and reviewing goal outcomes^{(Reference Lim, Hill and Teede91)}.

While we have evidence for what works for the antenatal and postpartum periods separately, the challenge remains for bridging the gap between pregnancy and postpartum care. Women with GDM in particular report feeling like they have been dropped by the hospital system once they deliver their baby and are left to struggle on trying to manage their future chronic risk with the additional stressors of rearing a young family^{(Reference Dennison, Fox and Ward55,Reference Pennington, O'Reilly and Young92)} . Women place a large amount of trust in the hospital and their staff to support them during pregnancy^{(Reference Lie, Hayes and Lewis-Barned93)} but continuity of care is not sustained beyond the immediate postpartum period. Research suggested that interventions postpartum can be successful in management postpartum weight, however most interventions to date have been short in duration and had a relatively short follow-up period ^{(Reference Christiansen, Skjøth and Rothmann94)}. Systematic review shows that only five interventions have been conducted that cross the pregnancy to postpartum divide for GWG or postpartum weight management and that the postpartum contact was minimal^{(Reference Vincze, Rollo and Hutchesson95)}. All the identified interventions were also not integrated into routine care and as such, fail to shed any light on their implementation potential. The other aspect to this gap is that the interventions all focus on behaviour change in either the woman or the infant but not both, yet there are clear opportunities to leverage the impact of change in this instance by creating a more holistic approach to behaviour change at the family unit level. Research also indicates that the guidance available to inform translating the evidence into practice is lacking in terms of health service implementation or optimal programme delivery^{(Reference Bahri Khomami, Teede and Enticott88,Reference Makama, Skouteris and Moran96,Reference Lim, Chen and Makama97)} . There is a clear need to use implementation science methods to inform the integration of effective lifestyle interventions into routine antenatal and postpartum care if we are to reduce the burdens of overweight and obesity and its impacts. In addition to understanding how to integrate and implement effective programmes, we need to have programmes that can be delivered with fidelity across multiple contexts in a scalable, sustainable way.

Digitally delivered health behaviour change interventions can successfully influence inequality by increasing availability and access for users. Smartphones have near ubiquitous coverage in all socioeconomic groups. For example, almost 90 % of people in the UK and Spain own a smartphone^{(Reference Poushter, Bishop and Chwe98)} and mothers aged 18–49 years spend over 21 h/week on their smartphones⁽⁹⁹⁾. Apps can provide ‘around the clock’ high-quality information as well as tailored support at low cost^{(Reference Becker, Miron-Shatz and Schumacher100)}. Women with low socioeconomic status commonly use apps during pregnancy but not postpartum because of the lack of quality apps, with a postnatal app gap^{(Reference Guerra-Reyes, Christie and Prabhakar101)}. Maternal wellbeing and education level have also been identified as an enabler for engagement with mhealth-based health behaviour change interventions in pregnancy^{(Reference Roche, Rafferty and Holden102,Reference O'Brien, Segurado and Geraghty103)} . Higher mhealth engagement occurs when a health care professional (HCP) does the referral, a credible source co-designs the app with user-tailored content and the app usage starts during pregnancy^{(Reference Lim, Liang and Hill81,Reference Hussain, Smith and Yee104–Reference Taki, Lymer and Russell107)} . Reviews of nutritional information on smartphone apps for pregnancy found that although the volume of apps was high, the overall quality was low, they did not consistently provide accurate nutritional information and only a few used BCT^{(Reference Brown, Bucher and Collins105,Reference Bland, Dalrymple and White108)} . While there is evidence that mobile smart phone applications may have a role to play^{(Reference Greene, O'Brien and Kennelly109)}, the literature evaluating pregnancy or postpartum apps effectiveness and long-term engagement is sparse. Reviews indicate little to no literature exists on how to deliver digital health coaching programmes most effectively, and that digital tools to support women are ‘of low quality, had minimal behaviour change potential, and were potentially unsafe, with minimal linkage to evidence-based information or partnership with health care’^{(Reference Brammall, Garad and Boyle110)}. A qualitative systematic review identified digital health interventions were highly acceptable to pregnant women with obesity or GDM both in pregnancy and postpartum^{(Reference Kelly-Whyte, McNulty and O'Reilly111)}. Smartphone technology holds the potential to provide more advanced methods of delivering personalised health behaviour change messages, while reaching a greater number of women for significantly lower costs than would be possible with traditional methods. Implementation research around app-based interventions that cover pregnancy and postpartum periods presents a major gap that has still to be addressed.

Longer-term engagement is required for pregnancy and postpartum interventions to understand impact, yet the average duration of interventions is 6 months^{(Reference Lim, Liang and Hill81,Reference Bahri Khomami, Teede and Enticott88)} . The LIFE-Moms consortium lifestyle interventions in women with overweight or obesity (n = 1150; interventions included variable types of diet, physical activity and behaviour change; three were for pregnancy only and four continued postpartum) analysis showed lower postpartum weight retention at 12 months (2⋅2 ± 7⋅0 v. control 0⋅7 ± 6⋅2 kg, respectively; −1⋅6 kg difference (95 % CI −2⋅5, −0⋅7; P = 0⋅0003))^{(Reference Phelan, Clifton and Haire-Joshu112)}. The odds (OR = 1⋅68 (95 % CI 1⋅26, 2⋅24)) of women returning to their pre-pregnancy weight at 1-year postpartum were significantly increased but the intervention did not impact infant anthropometry and it is very difficult to extract meaningful implementation-relevant information from these different interventions^{(Reference Phelan, Clifton and Haire-Joshu112)}. However, lifestyle interventions in pregnant and postnatal women do demonstrate improved dietary patterns, exercise habits, and GWG with improved pregnancy outcomes^{(Reference Teede, Bailey and Moran85,Reference Flynn, Dalrymple and Barr113,Reference Lim, O'Reilly and Behrens114)} . The challenge remains that most are intensive, have low penetration or uptake rates, low adherence or participation rates, are not co-designed or implementation informed, require costly personnel and resources or lack long-term follow-up^{(Reference Cantor, Jungbauer and McDonagh86,Reference Bahri Khomami, Teede and Enticott88,Reference Makama, Skouteris and Moran96,Reference Lim, Chen and Makama97)} . They also do not generate evidence to address the most significant evidence practice gap – how to implement evidence-based interventions into routine care. Although evidence shows that health behaviour change can reduce GWG and intrapartum weight management, there is a large research translation gap about achieving implementable interventions with adequate population penetration and participation. Evidence to support the capacity for implementation of antenatal health behaviour change interventions in maternity care settings remains limited^{(Reference Bahri Khomami, Teede and Enticott88)}.

Implementation-focused interventions

There are several ongoing studies that are seeking to move the implementation of interventions forward into more real-world settings and whom their primary results are due in 2024. The first study is called Face-it^{(Reference Maindal, Timm and Dahl-Petersen115)} and this is a effectiveness-implementation hybrid type 1 design. The intervention seeks to change health behaviours around physical activity, diet and breastfeeding in the first postpartum year for women with previous GDM and their families. Face-it is a health system-based intervention with three core components that will be undertaken in Denmark^{(Reference Nielsen, Dahl-Petersen and Jensen116)}. The central intervention component is additional visits from the public health nurse covering the health behaviour changes required and these visits will see the public health nurse work with the family unit, rather than just the mother. The second component is mhealth coaching delivered by a healthcare professional using an app. The mhealth coaching is offered to both the woman and her partner. The final component is a structured communication system that sends the public health nurse the woman's maternity discharge letter (normally only provided to general practitioner) and provides the ability to send reminders to either the public health nurse, health coaches or the woman through the secure platform. The intervention used the MRC complex intervention design^{(Reference Skivington, Matthews and Simpson62)} to underpin the study and seeks to recruit 460 women into the randomised controlled trial in a ratio of two intervention to one usual care participants. The primary outcome is reduction in BMI from baseline to 12 months postpartum and implementation data will be collected to explore fidelity and PIPE metrics in detail.

The second implementation study is Optimal Me^{(Reference Harrison, Brammall and Garad117)}, which is a hybrid type 3 design and will be conducted in Australia. The study seeks to primarily test whether the implementation strategy of the intervention's health coaching is better delivered via telephone or online videoconferencing. Women will be randomised to either arm and both will have a mhealth app that will provide health behaviour change and text message support. Extended implementation data will be collected using the RE-AIM framework^{(Reference Glasgow, Vogt and Boles77)} to ensure the study penetration, reach, feasibility, acceptability, adoption and fidelity can be evaluated. Optimal Me aims to recruit 300 women of childbearing age who are not pregnant but wish to conceive in the next 12 months. This selective criterion is normally very challenging to recruit as one in two pregnancies is unplanned but it can be achieved because the women will be recruited through a private health insurance company and women are required to upgrade their policy if they wish to include maternity care in the following year. The study has been designed to use an alternative pathway to usual maternity care and as a result will have a higher socioeconomic position population, which will need to be accounted for in the analysis.

The final study is Bump2Baby and Me^{(Reference O'Reilly, Burden and Campoy118)}, which is a hybrid type 2 design and will recruit women during pregnancy in four countries (Ireland, England, Spain and Australia). Bump2Baby and Me will provide mhealth coaching from a healthcare professional via a smartphone app that will also support self-monitoring of diet, physical activity and weight as well as a private social media community and a tailored information library. The intervention is designed to sit alongside usual care in each country and facilitates communication between the maternity service provider and the health coaches. The study is a randomised controlled trial that seeks to recruit 800 women (200 per country) in their first trimester of pregnancy at-risk of developing GDM to either usual care or usual care and the intervention. The intervention will provide health coaching and app access for the duration of the woman's pregnancy and the first postpartum year. It will expand the health coaching provided during the first postpartum year to include healthy infant feeding and active play alongside healthy family meals. The study is underpinned by the EPIS framework and will use NPT to achieve better understanding of the factors that influenced normalisation of the intervention in both the healthcare setting and in the women^{(Reference O'Reilly, Laws and Maindal119)}. The primary outcome of the randomised controlled trial is maternal weight at 12 months. The RE-AIM evaluation framework will be used to explore implementation and fidelity aspects as secondary outcomes^{(Reference O'Reilly, Laws and Maindal119)}.