We set a broad remit for the review with pre-defined study inclusion criteria ( Table 1 ). We included studies with any measure or estimate of maternal obesity set as the exposure of interest, and the outcome as any measures of offspring cardiovascular health at any age. We included studies of both adult and paediatric populations in the English language.

The searches were carried out by two independent investigators (LK, MA) using the Ovid Medline (1946-October 2020) and EMBASE electronic databases. In Ovid Medline, we used the advanced search option to select relevant MeSH (Medical Subject Headings) terms. To capture broad results, we used the explode function, included all subheadings, and included a keyword search for each term using the multi-purpose (mp) function. Search terms were combined using Boolean operators. The results were limited to English language articles and human studies ( Table 2 ). The titles and abstracts of the search results were visually screened for eligibility according to pre-defined criteria by LK and MA ( Table 1 ). A similar approach was taken to searching the Embase database to identify further unique hits. Additional studies were sought through cross-referencing, author searches, and grey-literature searches. The full text of selected papers was then examined to confirm eligibility. Papers that did not meet these criteria were excluded. In cases of ambiguity, eligibility for inclusion was determined after full text review and discussion with project supervisors (ZRE, SEP).

The studies were assessed for their risk of bias using the ROBINS-I (Risk of Bias in Non-randomised Studies – of Interventions) tool (21). This tool assesses the risk of bias in seven domains by asking a series of questions, before coming to an overall judgement about risk of bias (low, moderate, serious, critical). The seven domains comprise potential bias from confounding, participant selection, intervention (exposure) classification, deviations from intended intervention (exposure), missing data, outcome measurement, and selective reporting of results. Further explanations of these domains and criteria for assessment of overall risk of bias are available in a dedicated publication (21).

The 27 included studies were published between 2011 and 2020 in 11 different countries. All were observational studies. All used Maternal pre-pregnancy body mass index (MppBMI) or maternal body mass index (BMI) during pregnancy as the main exposure of interest. Six studies used gestational weight gain (GWG) as an additional exposure. One study also used skinfold thickness and fat percentage by bioimpedance. The studies measured cardiovascular health outcomes in offspring at various ages: 10 papers studied neonates, 11 studied children (0-18years), and eight studied adults (17-86 years). A diverse range of outcomes were considered, including CHD, cardiometabolic parameters (such as, BMI, blood pressure (BP) measurement, serum lipids, and serum insulin levels), and incident/prevalent CVD.

We organised the identified studies according to the age range studied and the outcome measured: neonatal CHD (n=7), neonatal cardiometabolic parameters (n=1), childhood cardiometabolic parameters (n=11), adult cardiometabolic parameters (n=5) and adult CVD (n=3). The characteristics, outcomes, and exposures of studies are summarised in Figure 2 , and in Tables 3 – 7 .

The studies were evaluated for their risk of bias using the ROBINS-I tool. 16 studies were appraised to be of moderate risk of bias, while 5 were of severe risk and 5 of critical risk. One study did not have enough information available to come to a bias assessment. No studies were at a low risk of bias. The results of the bias assessment are displayed in Supplementary Table 2 and summarised in Tables 3 – 7 .

Several studies examined the link between maternal obesity and neonatal CHD. Brite et al. (24) conducted a retrospective electronic records review of maternal and offspring records of 121,815 deliveries across 19 hospitals included in the Consortium on Safe Labour cohort study. Odds of any offspring CHD ascertained from hospital discharge records are reported for different pre-pregnancy BMI categories (overweight, obese, and morbidly obese) compared to normal weight mothers as defined by WHO categories. Additionally, odds of disease-specific CHDs were considered in mothers with obesity (BMI ≥30kg/m²) compared to mothers with BMI<30kg/m². Models were adjusted for study site, age, race, insurance status, and maternal smoking. A further analysis on a subset of 5,131 pairs was also adjusted for maternal glucose tolerance. The sample included 1,388 (1%) cases of CHD. The authors report a dose-dependent increase in odds of any CHD across increasing BMI categories. Compared to normal weight mothers, mothers with morbid obesity (BMI ≥40kg/m²) were at highest risk of having a child with CHD [OR 1.36 (1.03-1.78)], followed by mothers with obesity [OR 1.25 (1.08-1.45)], and mothers who were overweight [OR 1.18 (1.03-1.34)]. Compared to mothers without obesity, those with BMI ≥30kg/m² had significantly greater odds of having an infant with conotruncal defects [OR 1.33 (1.03–1.72)], atrial septal defects [OR 1.22 (1.04–1.43)], or ventricular septal defects [OR 1.38 (1.06–1.79)]. In the subset of 5,131 pairs adjusted for maternal glucose tolerance, these results remained consistent.

Madsen et al. (26) report a large population-based case-control study evaluating the relationship between maternal obesity and offspring CHD using linked birth-hospital discharge records between 1992-2007. The authors considered any CHD recorded within three years of birth. The sample comprised 14,142 infants with CHD and 141,420 comparators without CHD selected at random and matched on year of delivery. In models adjusted for gestational diabetes, infants with CHD were more likely to have a mother with obesity [OR 1.22, (1.15–1.30)]. Consistent with the observations by Brite et al. (24), Madsen et al. (26) also report a dose-response relationship between increasing BMI and higher risk of offspring CHD. They also report association of higher BMI with greater risk of specific CHDs, namely left [OR 1.27 (1.02–1.59)] and right [OR 1.43 (1.20–1.69)] ventricular outflow tract defects, and hypoplastic left heart syndrome [OR 1.86, 95% (1.13–3.05)]. In contrast to Brite et al. (24), they report no statistically significant association with conotruncal defects [OR 1.04 (0.82–1.33)]. These studies have several strengths, both making use of large, standardised health records to ascertain maternal BMI and offspring health outcomes. However, notably, Brite et al. (24) report missing BMI data for 77,000 women, and noted a higher prevalence of CHD in women with missing BMI data. This suggests not-at-random missingness in a manner relevant to the outcome of interest and may thus be a source of systematic bias. Similarly, Madsen et al. (26) used self-reported BMI for women who had missing physical measurements of BMI, potentially introducing recall/reporting bias in ascertainment of the exposure. Furthermore, neither study considered key modifying variables, such as exposure to teratogens, folic acid intake, and maternal co-morbidities other than gestational diabetes, which may represent important sources of confounding.

A small case control study by Alvarado-Terrones et al. (22), investigating maternal factors associated with incidence of total anomalous pulmonary venous connection (TAPVC) presented findings consistent with these observations. This study included 55 mother-child cases with isolated TAPVC and 152 healthy mother-child cases. Both groups had no maternal history of addiction, pre-eclampsia, type 1 diabetes, type 2 diabetes, or GDM. They report no difference in rates of family history, addiction, folic acid supplementation, or alcoholism in cases vs controls. The authors report that compared to normal weight mothers, those with pre-pregnancy BMI in the overweight and obese categories had significantly higher risk of infant TAPVC [OR 1.9 (0.9-3.8); OR 3.7 (1.5-9.5), respectively].

Complementary to these findings, Tang et al. (23) report association of elevated MppBMI with increased risk of obstructive heart defects (OHDs). This study identified several infant and maternal single nucleotide polymorphisms (SNPs) that interacted with maternal obesity to increase the risk of OHDs, some of which were protective in normal weight women (23) but not in those with obesity. These findings suggest important genetic evidence supporting the association between maternal BMI and infant CHD.

Several studies have failed to reproduce the observed associations between maternal BMI and infant CHD risk. Ghaderian et al. (25) present a retrospective case-control study of 164 CHD consecutive cases of infants referred to their paediatric unit and 158 infants without major malformations and their mothers. Infants with syndromic CHD, genetic syndrome, or chromosomal defects were excluded, as were mothers with gestational or pre-pregnancy diabetes. Maternal BMI was extracted from medical records and categorised into WHO obesity groups. In contrast with previously mentioned studied (22, 24, 26), Ghaderian et al. (25) report no statistically significant association between maternal obesity and infant CHD. It is possible that by more careful exclusion of genetic causes of CHD and mothers with history of diabetes, Ghaderian et al. (25) were able to elucidate the true nature of the maternal obesity-CHD relationship isolated from these confounding factors. However, the authors may also have been underpowered to detect an association with obesity due to the small number of mothers in the obesity categories (n=23 cases, 14%) as indicated by the wide confidence intervals reported by these authors ( Table 3 ). There is also ambiguity around the timing of BMI measurement, without explicit reporting that the measures were taken before or during pregnancy.

Dolk et al. (27) also present a case control study of babies diagnosed with CHD before 6 months of age, excluding those with genetic syndromes and children born to mothers with pre-gestational diabetes. In a different approach to other studies, they also included stillborn babies with prenatal diagnosis of CHD, mitigating the effect of survivor bias. This is an important consideration, as CHD may result in intra-uterine death and maternal obesity is also associated with an increased risk of stillbirth. Therefore, in excluding stillborn babies, cases of CHD may be missed, underestimating the effect of MppBMI on CHD. In the study by Dolk et al. (27) cases were ascertained by paediatric cardiologists based on clinical records and International Paediatric Cardiology Code criteria. Controls were babies without CHD recruited from routine maternity outpatient clinics. BMI data was extracted from maternity health records. Mothers with pre-gestational diabetes were excluded. Similar to Ghaderian et al. (25), Dolk et al. (27) report no significant association between maternal BMI category and infant CHD in crude or multivariable models (adjusted for: maternal age, previous pregnancy, maternal education, socioeconomic deprivation, diet).

In a further study, Kaplinski et al. (28) investigated the association between genetically determined maternal obesity and offspring conotruncal defects in 466 patients identified via Children’s Hospital of Philadelphia (1992-2010) and the Paediatric Cardiac Genomic Consortium (2010-2012). This study reported no significant association (in contrast to Brite et al. (24) and Madsen et al. (26)), suggesting that the relationship may not be causal (28). However, only the genetic risk of obesity was assessed, not the actual presence of maternal pre-pregnancy obesity phenotype. Since obesity develops through a complex interaction of genetic and lifestyle factors, genetic non-compliance may be an issue in this study. Nevertheless, these studies highlight the need to consider the role of the genetic risk of obesity as well as BMI itself.

Lemas et al. (29) recruited 753 mothers and infants from the Healthy Heart Study and 1,012 controls ( Table 4 ). The exposure of interest was MppBMI. The outcomes considered were neonatal cardio-metabolic markers including cord blood glucose, insulin, glucose-insulin ratio, total and high-density lipoprotein cholesterol (HDL-c), TG, free fatty acids, and leptin (29). The study found that MppBMI was positively associated with cord-blood insulin and glucose, and negatively associated with HDL-c (29). Adjustment for neonatal adiposity and foetal growth rendered the association with insulin levels non-significant, suggesting the effects may be partially mediated by neonatal adiposity. This study adjusted for multiple potential confounders relating to health and environmental status of mother and child. A potential limitation is measurement of MppBMI, which was obtained from medical records in 90% and self-report in 10%; this approach may introduce erroneous and non-standardised reports as well as recall and information bias.

A total of 12 studies investigated the association between maternal BMI and cardiometabolic parameters during childhood ( Table 5 ). The age range was 2-18 years. The papers studied a range of different outcomes split into six domains: childhood anthropometrics, BP, insulin levels, lipids, cardiovascular phenotypic measures.

Nine studies report positive associations between MppBMI and offspring childhood anthropometrics (30, 31, 33, 35–38, 40, 41). Of these, three studies had critical and two had severe risk of bias, predominantly due to confounding and selection bias ( Table 5 and Supplementary Table 2 ).

Filler et al. (34) conducted a study in children from emergency departments and clinical hospital appointments, and found that in confounder-adjusted models, higher MppBMI was associated with greater BMI, SBP, HDL-c, insulin, and all body fat measures in the offspring at 6 years. However, recruitment from hospital clinics may have introduced selection bias (34). As the children included in the study were in hospital for a medical problem, they are unlikely to be representative of the general population. Furthermore, disease processes for which the children are hospitalised can have independent effects on BMI and cardiometabolic markers which may importantly influence associations in this sample. Therefore, associations observed by Filler et al. (34) may not be widely generalisable. Techur-Pedro’s doctoral thesis (35), includes a retrospective longitudinal cohort study of live births registered on the Palau Ministry of Health Maternal and Child Health Birth Registry. This study similarly reports that among children aged 4-6 years, the risk of overweight or obesity was 2.57-times higher among those born to mothers with pre-pregnancy obesity compared to children born to mothers with normal MppBMI. However, this study had a 72% attrition rate at offspring follow-up, severely limiting the internal validity of the study due to attrition bias (35).

Despite shortcomings in several of the studies in this category, there is broad consistency of findings with other, more methodologically robust studies. Perng et al. (37) report a prospective cohort of pregnant women and their children in the USA. They studied 1,090 mother-child pairs and measured offspring overall and central adiposity at 6-10 years, along with blood metabolic parameters. The study found that every 5kg/m² increase in MppBMI was associated with 0.92kg higher total fat, 0.27 BMI z-score and 0.39kg trunk fat in the offspring (37). An important strength of this study was its adjustment for offspring lifestyle factors during childhood, including diet and television watching habits, as well as paternal BMI. Having a parent with obesity means that, beyond genetic factors, the child is more likely to grow up in an obesogenic environment, with unhealthy diet and exercise habits. Offspring lifestyle must be considered in studies in this area to ensure any association is not due to the lifestyle impact of having a parent with obesity. No other studies adjusted for these variables, making confounding bias due to these factors a major issue in all other studies.

Across these studies, a range of different methods were used to assess offspring anthropometrics. BMI is a useful measure of obesity because it is easy to calculate and correlates well with subcutaneous fat mass. However, this correlation is not true of all body types and some studies suggest that measures of visceral adiposity are more predictive of adverse cardiovascular complications of obesity. It was therefore valuable that several studies measured additional markers of offspring obesity and visceral adiposity ( Table 5 ). In summary, published studies provide consistent evidence that MppBMI is positively associated with offspring BMI during childhood, with higher BMI categories conferring an increased risk.

Six studies measured the impact of MppBMI on offspring BP. Five studies found that MppBMI was significantly associated with higher BP in the offspring (30, 33, 34, 37, 38) ( Table 5 ). Perng et al. (37) found a positive association between MppBMI and higher offspring systolic BP. A similar association was found by Gaillard et al. (33) who used a prospective cohort study to assess the BMI of 5,908 mothers throughout pregnancy and followed up their offspring at 6 years. Higher weight in early pregnancy was associated with higher childhood systolic BP, although this was attenuated by adjustment for the child’s BMI, suggesting partial mediation by this variable (33). However, this study is limited by the fact that MppBMI was ascertained from self-report (33).

Veena et al. (31) recruited 504 pregnant women without gestational diabetes at their booking appointment and followed-up their offspring at 9.5 years, measuring offspring BMI, BP and insulin levels, as well as paternal BMI at five years. They report no association between maternal obesity during pregnancy and offspring BP after accounting for childhood BMI ( Table 5 ). However, in this study maternal BMI was measured at 30 weeks’ gestation, which could reflect GWG, as opposed to MppBMI. Indeed, Gaillard et al. (33) found only early, not mid or late, pregnancy BMI to be positively associated with offspring BP. Therefore, the measurement of maternal BMI at such a late stage in pregnancy may have missed any correlation with offspring BP and could have contributed to the contradictory result to other studies in this section. Although gestational weight gain does correlate with maternal BMI, it is a proxy estimate and introduces measurement error and noise into the measurement, potentially diluting the effects of MppBMI. Finally, Wang et al. (39) investigated the interaction of maternal serum homocysteine levels and MppBMI with offspring SBP. Elevated blood homocysteine levels are known to be associated with increase CVD risk, and this study found a significant combined effect of maternal homocysteine concentration and pre-pregnancy obesity on offspring SBP, giving an example of how cardiovascular risk factors may compound offspring risk (39).

Currently available evidence on the association between MppBMI and childhood BP is high in quality and consistent, indicating a likely underlying causal effect. Gaillard et al. (33) further demonstrate attenuation (but not complete disappearance) of this relationship with adjustment for childhood BMI, suggesting a partial mediating effect of this variable. Thus overall, the current evidence suggests that maternal obesity is linked to higher offspring BP, but that this is almost completely mediated through greater obesity in childhood.

Five studies assessed the impact of MppBMI on offspring insulin levels. All studies used fasting insulin levels as the outcome measure, and consistently found a positive correlation between MppBMI and offspring insulin levels (31, 33, 37, 38, 40). Generally, the studies were of good quality, with only the study by Gaillard et al. (33) at a serious risk of bias due to the use of self-report MppBMI.

However, it is insulin resistance, rather than absolute insulin levels, that underlies the pathophysiology of type 2 diabetes (T2DM). To assess this, Perng et al. (37) and Tan et al. (38) used the Homeostasis Model Assessment of Insulin Resistance (HOMA-IR), an accurate tool for assessing insulin resistance (42). Both studies found a positive association between MppBMI and HOMA-IR. However, in non-diabetic subjects, the HOMA-IR is less useful and fasting insulin levels are an acceptable surrogate for insulin resistance (42). Therefore, the additional use of HOMA-IR by these studies may be of limited benefit.

Tan et al. (38) also used the Whole Body Insulin Sensitivity Index (WBISI), which is of greater benefit than HOMA-IR as the use of an Oral Glucose Tolerance Test integrates both insulin sensitivity and secretion (43). When comparing the offspring of normoglycaemic mothers with overweight/obesity to the offspring of normoglycaemic normal weight mothers, Tan et al. (38) found that MppBMI was positively associated with WBISI. However, confounders such as sociodemographic factors were not considered, and self-reported MppBMI was used to measure the exposure for some participants.

Adiposity has been strongly linked to insulin resistance. Therefore, it was important that studies adjusted for current offspring anthropometrics: in the four studies that adjusted for this, the association was partially attenuated (31, 33, 38, 40). Brandt et al. (40) also showed individuals with elevated insulin levels at age eight years had a significantly higher BMI trajectory from one year onwards. In summary, MppBMI has been consistently associated with both higher levels of insulin and insulin resistance in child offspring, supporting a potential underlying causal relationship, and this association is only partially explained (and therefore mediated) by the offspring BMI.

Four studies investigated the role of MppBMI on offspring lipid levels, although results were mixed. Gaillard et al. (33) and Tan et al. (38) found a significant association between higher MppBMI and adverse offspring lipid levels: lower HDL-c and higher low density lipoprotein-cholesterol (LDL-c) levels, though this analysis was not adjusted for by child body fat percentage or BMI and sociodemographic confounders, and consequently, the study by Gaillard et al. is at serious risk of bias from unadjusted confounders. Additionally, these findings were not supported by the study by Veena et al. (31), which found no association, although maternal BMI in this study was measured late in pregnancy at 30 weeks’ gestation. Finally, Perng et al. (37) investigated only the impact of MppBMI on triglyceride levels, reporting no significant relationship. Considering current evidence, no definitive conclusions can be made regarding the association between MppBMI and offspring lipid levels as the results are contradictory and the quality of the studies is mixed.

Two studies used echocardiography to measure offspring cardiovascular structure and function in childhood (32, 36). In a study of 4,852 parents and children, Toemen et al. (32) found that MppBMI (self-reported) and early pregnancy BMI (measured) were associated with higher left ventricular mass and left ventricular mass index in the offspring at six years. However, these associations were attenuated to non-significance when adjusted for offspring BMI. Also, no significant association was found with relative wall thickness or fractional shortening. Similarly, Litwin et al. (36) found no association with offspring LV diastolic and systolic function, this study was at serious risk of bias as GDM was the only confounder considered.

There is a paucity of evidence in this area, which limits the conclusions that can be drawn. The results available suggest that higher MppBMI is associated with adverse cardiovascular phenotypes in childhood, but that this may be wholly related to childhood BMI. Further studies are required before definitive conclusions can be drawn in this field.

Two studies assessed the association between MppBMI on the offspring vasculature (30, 32). Toemen et al. (32) found MppBMI was positively associated with greater aortic root diameter of offspring measured on echocardiography. Cox et al. (30) used fundal image analysis to calculate the retinal tortuosity index, an indirect measure of hypertension, coronary heart disease and stroke risk (44). Higher MppBMI was significantly associated with an increased retinal tortuosity index, independent of BP, in the offspring at age 4-6 years (30). However, only 56% of participants were followed-up, placing the study at serious risk of bias (30). The low number of studies and heterogenous designs mean conclusions cannot be drawn about the impact of MppBMI on offspring vasculature. However, these results suggest a potential association that warrants further investigation.

Two studies measured offspring anthropometrics in adulthood (15, 45). Both studies found a positive association between MppBMI and offspring BMI ( Table 6 ). Hochner et al. (15) prospectively followed-up 1,400 offspring at 32 years, using a self-report questionnaire to measure MppBMI and GWG. For every one standard deviation (SD) increase in MppBMI, offspring BMI increased by 1.8 kg/m² and waist circumference (WC) increased by 3.5 cm, and similarly, greater GWG was associated with higher offspring BMI in adulthood (15). Eriksson et al. (45) followed up 2,003 offspring at 62 years and found MppBMI was positively associated with lean body mass, fat mass, and BMI of the offspring. However, this study failed to adjust for confounders, except birthweight. Maternal BMI was also measured before delivery, so measurements may reflect gestational weight gain, as opposed to MppBMI. This study is likely impacted by survival bias as those who survived until 62 years may have had a more favourable body composition than those who died. Lawrence et al. (47) measured the change in offspring BMI from 17-32 years and found a significant association between higher MppBMI and a greater interval increase in offspring BMI, even after adjustment for genetic risk scores. Overall, the results are suggestive of a positive association between MppBMI and offspring obesity in adulthood. However, the extent to which this is caused by genetic factors, environmental factors or pregnancy-related factors is difficult to isolate.

Three studies measured the impact of MppBMI on offspring BP in adulthood (15, 16, 45). Only Hochner et al. (15) found a significant positive association between MppBMI and offspring systolic and diastolic BP. However, the association was attenuated after adjustment for the current offspring BMI, indicating that offspring BMI is a full mediator of this association. Kaseva et al. (16) found no significant association between MppBMI and offspring BP at 24 years. Eriksson et al. (45) found no significant association at 62 years. In summary, MppBMI may be associated with offspring BP during adulthood, though this relationship, similar to relationships with childhood BP, may be mediated by offspring BMI.

Three studies addressed the role of MppBMI on offspring insulin levels during adulthood (15, 16, 45). Hochner et al. (15) and Kaseva et al. (16) found a significant association between MppBMI and offspring insulin levels at 32 and 24 years respectively. Kaseva et al. (16) also found a significant association with fasting glucose levels, although there is insufficient information available in the study to assess bias. The associations were attenuated to non-significance in both studies when adjusted for current offspring BMI. Eriksson et al. (45) found no association between maternal obesity and neither insulin or fasting glucose levels in offspring at 62 years but was at a serious risk of bias. While the results suggest there may be an association between MppBMI and offspring adult insulin and glucose levels, it is likely to be mediated by the offspring’s current BMI.

Two studies assessed the role of MppBMI in the risk of offspring type 2 diabetes. Lahti-Pulkkinen et al. (46) and Eriksson et al. (48) found a significant association between MppBMI and offspring diabetes risk. Lahti-Pulkkinen et al. (46) linked the birth records of 118,201 children born in Aberdeen between 1950-2011 to the Scottish register for diabetes (up to 2012). Offspring of mothers who were overweight or obese had an increased hazard of T2DM compared to normal BMI mothers (hazard ratio 3.48 and 1.39 respectively) (46).

Eriksson et al. (48) used the Helsinki Birth Cohort study to follow disease outcomes in 13,345 individuals born in Helsinki between 1934 to 1944. Maternal BMI was associated with an increased risk of diabetes (48). However, maternal BMI was measured at the end of pregnancy and did not adjust for the gestation at which the BMI was measured, failing to distinguish between GWG and MppBMI. Furthermore, Eriksson et al. identified diabetes cases through prescriptions of medication, missing diet-controlled and undiagnosed cases. This method also relies on the completeness and accuracy of medication records. Some diabetes medications can be used for other diseases, such as metformin, so there is also a risk of misclassification.

The results suggest an association between MppBMI and offspring type 2 diabetes. However, the studies have several key limitations and neither adjusted for current offspring BMI, limiting definitive conclusions.

Three studies investigated the impact on adult offspring lipid levels (15, 16, 45). Only Hochner et al. (15) found a positive association between MppBMI and offspring triglyceride levels at 32 years, but this was wholly mediated by offspring current BMI, suggesting no direct association between MppBMI and triglyceride levels. Kaseva et al. (16) and Eriksson et al. (45) similarly found no association between MppBMI and offspring lipid measures (Total Cholesterol, LDL-c, HDL-c and triglycerides), although both have key issues. Overall, it remains unclear whether there is an association between MppBMI and offspring lipid levels as existing evidence is limited to a few studies with high risk of bias.

Three studies investigated the effect of maternal BMI during pregnancy on offspring CVD. All found that an elevated MppBMI was associated with increased risk of offspring hospital admission and death due to CVD (14, 48, 49) ( Table 7 ).

Reynolds et al. (14) linked the Aberdeen Maternity and Neonatal Databank to national death and morbidity records for 37,709 individuals aged 34-61years. The study found that the children of mothers with obesity were more likely to be admitted to hospital for a cardiovascular event than children of mothers with a normal BMI (14).

Razaz et al. (49) used the Swedish Birth Register to follow up offspring between 1-25 years. The children of mothers who were overweight or obese were more likely to develop CVD compared to those of normal BMI mothers (49). Interestingly, Razaz et al. (49) conducted a sibling analysis, where CVD cases were matched to their siblings, in order to remove the impact of childhood environment, which the siblings likely shared. The study found that increase in maternal BMI between pregnancies was associated with an increase in offspring CVD risk (49). This finding is suggestive of a causal relationship between MppBMI and offspring CVD. However, it is important to note that despite the sibling study design, there may still be a degree of persistent environmental confounding.

Importantly, none of the studies exploring the role of MppBMI in CVD measured or adjusted for the current offspring BMI, lifestyle factors or sociodemographic factors, so residual confounding is very likely in these studies. Overall, the results suggest there is a positive association between MppBMI and risk of hospital admission and death from CVD, but this may be partly or wholly explained by confounding, and further investigations are needed to address this question.

A main limitation of existing studies is inadequate consideration of confounding. Residual confounding reduces specificity and severely limits causal conclusions. Our study question cannot ethically or practically be answered in human randomised controlled trial. Therefore, controlling for confounders needs to be a key focus of future observational studies. While controlling for all variables may not be feasible, we have identified key areas where confounding should be addressed ( Supplementary Table 3 ). Furthermore, experimental studies using animal models may have a role in isolating causal relationships in this context.

Accurate and non-biased measurement of maternal obesity is clearly highly important, given that it is the exposure of interest. However, very few studies performed dedicated prospective measurement of maternal BMI as part of a research protocol. In most studies, MppBMI was ascertained from self-report questionnaires or medical records. Using self-report methods, women generally underestimate MppBMI in higher BMI categories (50). This approach may also introduce recall bias as women may forget their pre-pregnancy BMI over time. Importantly such reporting biases may differently influence self-reports of BMI in mothers of children with cardiovascular illnesses compared to those with healthy children thereby introducing systematic bias. Other studies obtained BMI from medical records. Although this is a more objective method for ascertainment of BMI, there is also potential to introduce biases given that recording of BMI is not performed using pre-defined protocols or standardised equipment.

Another important consideration is timing of maternal obesity measurement. This is clearly important given that women gain weight during pregnancy and that exposure to maternal obesity may have different consequences for the offspring depending on gestational age of exposure. Furthermore, variation in time of measurement of maternal obesity limits between study comparisons. Measuring MppBMI at a later stage in pregnancy may reflect GWG, as opposed to MppBMI. Studies should therefore aim to measure the BMI as soon as possible at a standard gestation. However, this is difficult to achieve as women may not always have a reason to attend a medical facility to have their BMI checked. Women gain minimal weight during the first trimester, indicating BMI at the 10 week booking appointment is a reliable substitute for MppBMI (51). In our review, many studies measured maternal BMI at an unknown, late, or inconsistent time.

In studies investigating the offspring during childhood or adulthood, offspring were followed up for 2-18 years and 24-72 years respectively. The risk of obesity, hypertension, insulin resistance, and hypercholesterolaemia increase with age; Lawrence et al. (47) found that the children of mothers with obesity had a higher increase in BMI between 17-32years. This makes it difficult to draw conclusions about the impact of MppBMI on cardiometabolic parameters because the effects are potentially only manifest at certain ages. To understand how the risk changes with time, measurements should be repeated at different ages. However, drop-out, time, and cost limit the feasibility of this recommendation.

A further consideration is the potential time-varying nature of exposure to maternal obesity on offspring cardiovascular risk. It is possible, for example, that the importance of maternal obesity as a driver of adverse cardiovascular health may be more important at younger rather than older ages. As an individual transitions to adulthood other environmental exposures may contribute more to the overall cardiovascular risk than maternal obesity. However, it is also reasonable to suggest, based on available evidence, that early life exposure to maternal obesity is both a direct and indirect driver of adverse cardiometabolic profile and lifestyle behaviours throughout life. Indeed, the importance of early life exposures as determinants of health through to older age is the key tenet of lifecourse epidemiology theory (52). Further research, using different approaches, is needed to determine with greater certainty the mechanisms through which maternal obesity impacts offspring cardiovascular health and to isolate its independent effect over environmental exposures in later life. Animal studies and genetic studies are likely to provide greatest insight into such mechanism as the separation of such tightly intertwined factors (maternal obesity, environment/lifestyle) using observational approaches is extremely challenging and may not be possible.

Our results were suggestive of a positive association between MppBMI and CHD. Previous studies have proposed several underlying mechanisms, although no definitive cause has been found. Firstly, impaired glycaemic control increases CHD risk and individuals with obesity are more likely to experience type 2 and gestational diabetes. However, most studies in this review excluded mothers with diabetes, reducing the likelihood of this as a potential explanation of the relationship.

Folic acid supplementation decreases the risk of neural tube defects and CHD (53–55). Most studies in this review controlled for folic acid supplementation, which, in theory, would indicate folic acid deficiency does not explain the elevated risk of CHD. However, women with obesity are more likely to be deficient in folic acid despite supplementation and may require higher doses (54, 56). Therefore, future studies should measure serum folate levels, as well as supplementation. Furthermore, other teratogenic exposures were not adequately considered in existing studies.

Finally, previous studies have suggested women with obesity may be less likely to comply with antenatal screening. Similarly, increased abdominal fat may make defects less visible via ultrasound scan. Without a prenatal diagnosis, women with obesity may be less likely to terminate an affected pregnancy, increasing CHD rates compared to normal weight women.

We conclude that maternal obesity during pregnancy is associated with an adverse offspring cardiometabolic profile and higher CVD risk. The next question is whether these associations are simply due to the genetic and environmental impact of having a parent with obesity, or if exposure to maternal obesity in-utero independently programs the foetus to have worse cardiovascular health. The first refers to the widely accepted view that obesity and CVD occur as a consequence of unhealthy lifestyle habits and genetics. A mother with obesity is more likely to eat an unhealthy diet and exercise infrequently (57). Her children are more likely to adopt these habits, increasing their risk of obesity, hypercholesterolaemia, hypertension, and insulin resistance. These risk factors would increase the offspring’s risk of CVD later in life. Furthermore, a mother with obesity may have several genetic predispositions to obesity that could further increase the offspring’s obesity risk, should these be inherited. It is possible that any associations between MppBMI and offspring cardiovascular health are simply due to the lifestyle effect of having a mother with obesity and the inheritance of obesogenic genes.

However, we found evidence that adverse cardiometabolic changes may be present in neonates of mothers with obesity. The neonatal period is not yet strongly influenced by environmental and lifestyle factors, suggesting the conditions in-utero may influence offspring cardiovascular health at very early ages. Furthermore, if the elevated risk was only due to lifestyle and genetic factors, paternal obesity would be expected to confer an equal risk to offspring health. Three studies in this review adjusted for paternal BMI, of these two found a significant association with an adverse offspring cardiometabolic profile even after adjustment for paternal BMI. Furthermore, Perng et al. (37) found a significant association between higher MppBMI and an adverse offspring BMI and cardiometabolic profile after adjustment for offspring television watching habits and diet during childhood. These studies suggest that offspring exposure to maternal obesity during pregnancy confers an additional risk, over that expected from lifestyle and genetic factors.

The second mechanism involves the Developmental Origins of Health and Disease (DOHaD). This was described by Barker following the observation that death rates distributions from coronary heart disease mirrored infant mortality and low birth-weight rates 70 years earlier (7). The theory postulates that exposure to less favourable conditions in-utero triggers physiological adaptations that help the foetus to survive (6, 7). Since these adaptations occur during critical periods of organ development, the changes are likely irreversible and impact foetal long-term development (6, 7). While these compensatory adaptations may be helpful for in-utero survival, the infant may be left with increased risk of developing certain non-communicable diseases later in life (6, 7). For example, the changes that help a foetus to survive conditions of poor nutrient availability in-utero could be maladaptive in a post-natal environment where there is sufficient food availability, increasing the offspring’s later risk of coronary heart disease. Whilst direct causal evidence in this context is lacking, several lines of evidence provide support for the hypothesis (58, 59).

The Foetal Overnutrition Hypothesis suggests that mothers with obesity are more likely to have a diet high in calories and fat, which exposes the foetus to excess nutrients in-utero ( 12). Several observational studies have shown an association between excessive maternal fat and sugar intake and the development of childhood obesity in the offspring (12). This hypothesis is further supported by research into diabetes during pregnancy. Here, the foetus is exposed to excessive sugar, lipids and amino acids, which contribute to foetal macrosomia and an increased risk of childhood obesity (60). Furthermore, high calorie diets during pregnancy have been associated with changes in hypothalamic gene expression in animal models, which could mediate appetite and satiety in the offspring and therefore offspring obesity levels and cardiovascular risk (61).

Epigenetic changes are the second proposed underlying mechanism for the DOHaD (62). Epigenetic changes are heritable alterations in gene expression, but not the underlying genetic sequence, through DNA methylation, histone modification and RNA processes (63). The environment in-utero is a key period for the development of epigenetic markers; and nutrition influences this process (63). For example, offspring born during the Dutch Hunger Winter were exposed to famine conditions in utero. Exposed offspring were more likely to develop coronary heart disease in adulthood and at a younger age (64). Such individuals were also found to have less DNA methylation of the IGF-2 gene 60 years later compared to unexposed siblings (63). Similar epigenetic changes have been reported in animal studies for maternal obesity. In primates, a high calorie maternal diet during pregnancy was shown to cause covalent modification of histone proteins at specific sites, which altered gene expression in the offspring (65). A mouse-model found that a high-fat maternal diet during pregnancy decreased global and gene-specific DNA methylation in offspring, which altered gene expression of dopamine and opioids (66). These offspring showed an increased preference for sucrose and fat (66). Therefore, animal studies have shown exposure to pre-natal maternal obesity causes epigenetic changes that adversely influence offspring diet. However, the impact on disease risk is unclear and human studies are limited. A systematic review by Van Dijk et al. (67) found that obesity was associated with different levels of methylation at several specific sites, and some evidence that this may be reversible with weight loss programs. Liu et al. (68) measured the impact of MppBMI on CpG methylation in neonatal cord blood DNA. While one significant site was identified, the impact on gene expression and offspring long term health were not investigated, so the clinical relevance of this observation is unknown.

The rising prevalence of obesity worldwide is a major public health concern. Obesity is a key risk factor for numerous conditions, and its rising prevalence is causing an increase in the incidence of CVD, type 2 diabetes, and several cancers. We found several associations between an elevated MppBMI and adverse offspring cardiovascular health, including an increased risk of CHD, obesity, worse cardiometabolic parameters, and greater CVD burden. The children of mothers with obesity are more likely to be obese themselves and have worse cardiovascular health. These observations highlight the importance of early life factors on later health outcomes (52). These findings are concerning for many reasons. 30% of reproductive age women are obese, which is likely to impact the health of the next generation, further perpetuating the increasing trend in the prevalence of obesity and CVD. Secondly, our results suggest the associations with offspring cardiovascular health may not be adequately explained by sharing lifestyle and genetic factors with an obese parent. Exposure to maternal obesity in-utero could program offspring to have worse cardiovascular health. These offspring would be high risk for developing CVD, given the likely presence of genetic, environmental, and developmental risk factors for obesity. These children could then grow up to also be mother with obesity, creating a vicious cycle, where the obesity crisis worsens with each generation.

To protect the offspring, we should address maternal BMI during and before pregnancy (59). If developmental programming is occurring, obesity prevention strategies could hypothetically reverse such programming. Meanwhile, if developmental programming is not occurring, MppBMI could be used as a screening tool to identify offspring at a high risk of cardiovascular complications. Pre- and post-natal interventions could be targeted towards such groups, reinforcing good lifestyle habits in the mother that she can pass on to her child. However, it is not enough to only address diet and lifestyle through educational interventions (69–71). Obesity is strongly influenced by health inequalities such as socioeconomic or educational status. Access to fresh healthy food is more difficult in some, predominantly poorer, areas and healthy food is often more expensive than unhealthy options, limiting its accessibility. Therefore, poverty must be addressed, and health education must have a particular focus on individuals in areas with greater socio-economic deprivation.

The broad framing of the remit of this review, in terms of definition of the exposure, offspring cardiovascular health indicators, and age range covered permitted a wide-reaching assessment of the associations of maternal obesity with offspring cardiovascular health. We used a systematic and standardised approach validated by two independent investigators. However, despite a thorough and broad review of the topic, we were unable to draw definitive conclusions in several areas due to limitations of the source literature. We cannot exclude the possibility of publication bias. Finally, the heterogenous design of the studies precluded meta-analysis of the results.