Pre-pregnancy and pregnancy obesity and neurodevelopmental outcomes in offspring: a systematic review

Authors

  • R. J. Van Lieshout,

    Corresponding author
    1. Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada
    2. The Offord Centre for Child Studies, McMaster University, Hamilton, Ontario, Canada
      RJ Van Lieshout, Department of Psychaitry and Behavioural Neurosciences, McMaster University, Hamilton Health Sciences, Chedoke Division, Box 2000, Central Building Room 304, 1200 Main Street West, Hamilton, Ontario, Canada L8N 3Z5. E-mail: vanlierj@mcmaster.ca
    Search for more papers by this author
  • V. H. Taylor,

    1. Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada
    Search for more papers by this author
  • M. H. Boyle

    1. Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada
    2. The Offord Centre for Child Studies, McMaster University, Hamilton, Ontario, Canada
    Search for more papers by this author

RJ Van Lieshout, Department of Psychaitry and Behavioural Neurosciences, McMaster University, Hamilton Health Sciences, Chedoke Division, Box 2000, Central Building Room 304, 1200 Main Street West, Hamilton, Ontario, Canada L8N 3Z5. E-mail: vanlierj@mcmaster.ca

Summary

Maternal obesity in pregnancy is associated with a number of adverse outcomes for mother and her offspring both perinatally and later in life. This includes recent evidence that suggests that obesity in pregnancy may be associated with central nervous system problems in the foetus and newborn. Here, we systematically review studies that have explored associations between maternal overweight and obesity in pregnancy and cognitive, behavioural and emotional problems in offspring. The 12 studies eligible for this review examined a wide range of outcomes across the lifespan and eight provided evidence of a link. These data suggest that the offspring of obese pregnancies may be at increased risk of cognitive problems and symptoms of attention deficit hyperactivity disorder in childhood, eating disorders in adolescence and psychotic disorders in adulthood. Given the limitations of existing data, these findings warrant further study, particularly in light of the current worldwide obesity epidemic.

Introduction

The last 20 years has seen an exponential increase in rates of overweight and obesity around the world (1,2). Women of childbearing age have not been spared in this epidemic and approximately one-third of US (2) and one-fifth of UK (3) women are obese. Obesity prevalence among women aged 12–44 years in the US doubled between 1976 and 2004 and the number having a body mass index (BMI) ≥40 tripled (4). In the decade following 1993, a 70% increase in pre-pregnancy obesity was reported in a study of nine US states (5).

Coinciding temporally with this epidemic has been an increasing interest in how exposures in pregnancy can affect the long-term health and development of the offspring of these gestations. The process by which persistent physiological alterations in offspring result from intrauterine conditions is referred to as foetal programming (6). Indeed, a large number of studies that attest to the importance of the intrauterine environment suggest that it can ‘programme’ or affect later risk of obesity (7), hypertension (8) and even depression (9) in offspring independent of genetic and socioeconomic confounders.

The presence of maternal obesity in pregnancy creates an intrauterine environment that is suboptimal for both the mother and her foetus and is a major modifiable contributor to adverse maternal and child health outcomes. Pre-pregnancy obesity is a risk factor for diabetes mellitus (DM), hypertension, thromboembolic disease and asthma in pregnancy (10). At delivery, obese women are 20% more likely than normal weight women to suffer perinatal haemorrhage and three times more likely to develop an infection (11). They are less likely to breastfeed (12) and are more prone to lactation delay and failure than normal weight women (13). Obese mothers are also twice as likely to give birth to macrosomic infants who themselves have higher rates of birth trauma and perinatal asphyxia (14). In the longer term, pre-pregnancy obesity is associated with a number of adverse health outcomes for the mother including cardiovascular and metabolic disease (11) and the offspring of these pregnancies appear more likely to be obese (7) and suffer from the metabolic syndrome (15).

The cardiovascular and endocrine systems may not be the only ones whose functioning is affected or ‘programmed’ by obesity in pregnancy. In fact, a recent systematic review suggests that infants born to mothers who were obese were at increased risk of central nervous system (CNS) developmental problems including neural tube defects, spina bifida and hydrocephaly (16).

If maternal obesity does adversely affect neurodevelopmental outcomes in offspring later in life, it is not clear if this occurs via a direct programming effect or if obesity simply serves as a marker of the actual pre- or post-natal causal factors. The foetal programming hypothesis would posit that the intrauterine exposures accompanying maternal obesity would alone be sufficient to produce long-term adverse effects on offspring neurodevelopment. Such an effect could occur via the dys-regulation of maternal hormonal or immune systems, or even nutrient excesses. Alternatively, pregnancy obesity could be linked to neurodevelopmental outcomes via a number of indirect or non-causal pathways. For example, obesity in pregnancy might simply be a marker of problems known to affect overweight mothers and that also influence foetal neurodevelopment such as folate (17) or vitamin D deficiency (18). Pregnancy exposures common to obese women including increased rates of DM (10), exposure to traumatic events in pregnancy and/or perinatal depression could also be responsible (19). Moreover, the effects of pregnancy obesity on the offspring's cognitive, emotional and behavioural development might not be because of pre-natal exposures but rather mediated post-natally by obstetric complications or child health problems such as obesity or DM. Finally, genetic and/or environmental confounders might be responsible for observed associations. Maternal cognitive problems (20), maladaptive personality traits (21), frank psychiatric disorder (22) or even exposure to poverty (23) might not only increase a mother's risk of obesity, but also affect the risk of foetal neurodevelopmental problems independent of the intrauterine state associated with maternal obesity.

Given the potential for maternal obesity during pregnancy to affect CNS development, coupled with the fact that it may be a modifiable risk factor for CNS problems makes this area particularly worthy of study. As a result, we systematically reviewed extant studies to determine if the offspring of women who were overweight or obese prior to or during pregnancy had higher rates of neurodevelopmental problems in childhood, adolescence and adulthood than those born to mothers with a BMI in the normal range. We anticipated that the offspring of women who were overweight and obese would manifest higher levels of problems than those born to women with normal BMIs.

Methods

We searched MEDLINE and EMBASE from their respective inceptions to 22 September 2010 for studies in all languages that have examined the association between maternal overweight and obesity, weight gain during pregnancy and neurodevelopmental outcomes in the offspring of these gestations. While we examined the predictive value of maternal overweight, obesity (BMI ≥ 30) was our main predictor of interest. The following search strategy was utilized for MEDLINE: [(pregnan* OR pre-pregnancy body mass index) AND (exp body mass index OR exp obesity OR exp overweight)] AND (exp mental disorders OR behav* OR exp brain diseases OR cognit* OR neuropsych* OR executive function* OR development* OR exp human development OR exp child development OR exp developmental disabilities OR exp personality development OR exp learning disorders OR math* OR arithmetic OR exp reading OR spelling OR exp dyslexia OR exp schools OR academic OR exp communication disorders OR exp psychomotor disorders OR exp perceptual disorders). This search was limited to studies on humans. Our EMBASE search utilized the same strategy, using the same keywords and equivalent medical subject heading terms in that database. This is available upon request.

Only studies that examined BMI or weight gain as a predictor of its main outcomes were eligible. These studies could examine any cognitive, psychological, behavioural, emotional or psychiatric outcome assessed after 1 month of age.

All reviewers evaluated the titles and abstracts yielded by our search to determine whether articles met eligibility criteria. They also hand searched the reference section of each relevant article from the initial screening for relevant studies. Studies were chosen for full text review if it was deemed that there was any possibility that the article was eligible. All reviewers independently assessed all studies identified from this step and planned to resolve disagreements through discussion, although this was not required.

Adjusted estimates of the association between maternal obesity and the study outcome were the main outcomes considered. We also extracted data on sample size, study design and the obesity definition used as well as study limitations. Study quality was assessed using the Newcastle-Ottawa Scale (24). This scale is scored on a points system and case–control and cohort studies can earn a maximum score of 9 based on selection, comparability and exposure criteria.

We classified study findings in terms of whether or not they provided evidence in support of our a priori hypothesis of an association between maternal obesity and an increased risk of neurodevelopmental problems. We used the following headings to summarize study findings with respect to this hypothesis: SUPPORT: clear evidence of an association between maternal obesity and offspring outcome; MIXED: some but inconsistent evidence of an association between maternal obesity and offspring outcome, or a clear association limited by major methodological limitations and NO SUPPORT: no evidence of an association between maternal obesity and neurodevelopmental outcome. Given the heterogeneity of the eligible study outcomes and the relatively low number of studies for each of these categories, a decision was made not to meta-analyze the results of eligible studies.

Results

Our initial search identified 1713 potentially eligible studies in MEDLINE and 2931 in EMBASE (Fig. 1). Forty-two full text articles were examined. Upon review, 12 articles remained (25–36). Of these 12, two examined associations between increased maternal BMI and childhood intelligence quotient (IQ) (25,26), two reported on foetal alcohol syndrome (27,28), two examining attention deficit hyperactivity disorder (ADHD) in children (29,30) (one of which also contained data on negative emotionality (30)), four contained data on schizophrenia and related psychoses in adulthood (31–34) and two examined eating disorders (35,36). One of the eating disorders studies also contained data on depressive/anxiety problems in adolescents diagnosed by a healthcare professional (36). The other used maternal weight gain during pregnancy as a predictor of anorexia and bulimia nervosa (35). A summary of both the methodological issues and the substantive results of these studies can be found in Table 1. Of note, four additional studies either only examined BMI either as a confounder or covariate and did not provide adjusted estimates of its effect on neurodevelopmental outcomes (37–40) and were not eligible.

Figure 1.

Maternal obesity in pregnancy and neurodevelopmental outcomes in offspring: flow diagram.

Table 1.  Studies reporting data on maternal obesity or weight gain during pregnancy and offspring cognitive and psychiatric outcomes
ReferenceSampleDesignPredictor/exposureStudy objectives/outcomesFindingsLimitations (sampling, measurement, control of error)
Newcastle-Ottawa Scale (NOS) Score
  1. ADHD, attention deficit hyperactivity disorder; anova, analysis of variance; BMI, body mass index; CI, confidence interval; DM, diabetes mellitus; GA, gestational age; GDM, gestational diabetes mellitus; IQ, intelligence quotient; LMP, last menstrual period; N, number of mother–offspring pairs; NS, not significant; OR, odds ratio; PGDM, pregestational diabetes mellitus; RR, relative risk; SES, socioeconomic status.

IQ
 Neggers et al. 2003 (25)Mother–children pairs (average age ∼5 years) (N = 355), born in 1985–1989Sample from a prospective cohort of low income African–American children whose mothers were selected because of low plasma zinc levels (USA)Pre-pregnancy BMI self-reported by mother at ∼23 weeks after LMP.To assess the association between maternal pre-pregnancy BMI and maternal weight gain in pregnancy and psychomotor developmentSupportSampling: highly selected, small and disadvantaged sample (low SES, low mean maternal and child IQ, low zinc level). High rate of attrition.
BMI analyzed as a continuous and categorical measure.Overall, verbal and non-verbal IQEach increase of 1 unit in maternal BMI associated with significantly reduced IQ and nonverbal IQ.Measurement: retrospective maternal self-report of weight at ∼23 weeks GA.
Reference group for categorical analyses: BMI = 19.8–26Overall IQ scores of offspring of obese women were 4.7 points lower and nonverbal scores 5.6 points lower than those whose mothers had normal BMI.Control of error: no adjustment for maternal PGDM or GDM.
Overweight: 26.1–29NOS = 7
Obese >29
Pregnancy weight gain also analyzed as a continuous variable
 Heikura et al. 2008 (26)Two Finnish birth cohorts (1966, N = 12 058 and 1986, N = 9032) at ∼11.5 years oldCohort study (Finland)Pre-pregnancy BMI self-reported retrospectively at 25 weeks after LMP.To investigate maternal sociodemographic and health factors assessed during pregnancy for their association with intellectual disabilityMixedMeasurement: retrospective maternal self-report of weight at ∼25 weeks GA.
Height also self-reported by some mothers.
Control of error:
 Did not adjust for maternal or paternal IQ.
 No adjustment for maternal PGDM or GDM.
NOS = 7
Reference group: BMI: 20–25.IQ < 70 (intellectual disability)1966 obese: OR = 1.3, 95%CI 0.5–3.1
Thin: <20.
Overweight: 25–30.
Obese: ≥30
1986 obese: OR = 3.6, 95%CI 2.0–6.6
Foetal alcohol syndrome
 May et al. 2005 (27)6-year-olds (53 cases and 116 controls). Born in 1993Case–control (South Africa's Western Cape Province)Pre-pregnancy BMI (self-reported retrospectively 7 years after pregnancy)To describe risk factors for FASNo supportSampling: case and control mothers differed in a number of important ways including SES, education etc.
Foetal alcohol syndrome (diagnosed by dysmorphologists)Unadjusted t-test showed that BMI was significantly lower in mothers of Foetal alcohol syndrome children (24.9) than control mothers (27.5), P = 0.019.Generalizability of results may be limited.
Measurement: retrospective maternal self-report of weight and height 7 years after delivery, alcohol intake etc. Possible social desirability bias.
Control of error: did not appear to control for important confounders (e.g. maternal foetal alcohol spectrum disorder, GDM or PGDM).
NOS = 5
 May et al. 2008 (28)6-year-olds (72 cases and 134 controls). Born in 1996Case–control (South Africa's Western Cape Province)Pre-pregnancy BMI (self-reported retrospectively 7 years after pregnancy)To describe risk factors for FASNo supportSame as May et al. 2005.
Foetal alcohol syndrome (diagnosed by dysmorphologists)Unadjusted anova showed that BMI was significantly lower in mothers of foetal alcohol syndrome children (22.5) than control mothers (27.4), P = 0.001.NOS = 5
ADHD (Children)
 Rodriguez et al. 2008 (29)7- to 12-year olds (N = 12 556). Born 1978–19873 Prospective birth cohorts (Scandinavia)BMI from medical records at ∼10 weeks after LMPTo examine the association between BMI and/or weight gain and core symptoms of ADHD in school-age offspringSupportMeasurement: weight assessed at 10 weeks after LMP was called ‘pre-pregnancy weight’.
Only single informant; unknown clinical relevance of scale of scale cut-off selected.
Reference group: BMI = 19–26Top 10%ile of ADHD Symptom Scores (teacher rated)Overweight (BMI > 26): OR = 1.43, 95%CI 1.12–1.83)
Overweight: BMI > 26Each unit increase in BMI: OR = 1.04, 95%CI 1.02–1.07)Control of error:
 No adjustment for familial risk ofADHD.
 No adjustment for maternalPGDM or GDM.
Pregnancy weight gainFor women with high BMI, weight gain further increased odds (OR = 1.24, 95%CI 1.07–1.44).NOS = 8
 Rodriguez, 2010 (30)5-year-olds (N = 1714). Born 1999–2000Prospective birth cohort (Sweden)Pre-pregnancy BMI from the Swedish Medical Birth RegisterTo assess the association between maternal pre-pregnancy obesity and risk for ADHD symptoms in childrenMixedSampling: high rate of attrition.
Reference group: BMI = 20–25Top 15%ile of ADHD, Negative emotionality symptom scores (mother and teacher rated)Parent report: no increased odds for any outcome.Measurement:
 Unclear if Height and Weight are self-reported or assessed by medical professionals.
 Unclear when this was assessed.
 Unknown clinical relevance of scale cut-off selected.
Overweight (BMI 25–30)Teacher report:
 Overweight:
  Inattention: OR = 2.00, 95%CI 1.20–3.35.
  Hyperactivity: NS.
  Negative emotionality: OR = 1.81, 95%CI 1.22–2.69.
 Obese:
  Inattention: OR = 2.09, 95%CI 1.19–4.82.
  Hyperactivity: NS.
  Negative emotionality: NS.
Obese (BMI ≥ 30)Outcomes:
 Overweight at risk but not obese, results positive only for certain symptoms with specific informants.
 No adjustment for maternal PGDM or GDM.
NOS = 7
Psychoses
 Jones et al. 1998 (31)Participants up to 28 years old (N = 10 578). Born in 1966Prospective birth cohort (Finland)Pre-pregnancy BMI reported retrospectively by mother at 24–28 weeks GATo determine if abnormalities of pregnancy, delivery and the neonatal period are associated with adult-onset schizophreniaMixed
BMI > 29 OR = 2.1, 95%CI 0.9–4.6
Measurement: retrospective maternal self-report of weight and height at 24–28 weeks GA.
Control of error:
 No adjustment for familial risk of schizophrenia.
 No adjustment for maternal PGDM or GDM.
NOS = 6
BMI > 29
Reference group: BMI = 19–29
Schizophrenia (psychiatrist diagnosed)
 Schaefer et al. 2000 (32)30- to 38-year olds (63 cases and 6570 controls). Born 1959–1967Case–control study of participants in a Prospective US birth cohort of those born between 1959 and 1967BMI measured at study enrolment by healthcare personnelTo examine the relation between maternal pre-pregnant BMI and schizophrenia spectrum disordersSupportSampling: high rate of attrition.
Measurement: unclear GA at study enrolment.
Control of error: no adjustment for maternal PGDM or GDM.
NOS = 9
Pre-pregnancy obesity (BMI > 30)Schizophrenia and Spectrum Disorders (psychiatrist diagnosed)Schizophrenia and Spectrum
Reference group: BMI = 20–27Obese: RR = 2.9, 95%CI 1.3–6.6
Schizophrenia alone: RR = 2.7, 95%CI 0.95–7.8
 Wahlbeck et al. 2001 (33)Individuals born between 1924 and 1933 (N = 7086)Prospective birth cohort (Finland)Late pregnancy BMI from birth recordsTo assess the influence of maternal body size, infant size at birth and childhood growth on schizophrenia riskNo support
BMI of obese mothers used as reference group.
Sampling: high rate of attrition.
Did not assess if prediction improved by inclusion of quadratic term (i.e. did not rule out a J-shaped relation between BMI and schizophrenia).
Reference group: BMI > 30Schizophrenia (inpatient discharge diagnosis)Those with BMI less than 30 were at increased odds (OR = 3.06, 95%CI 1.12–8.38) relative to those born >30.
In continuous analyses, each decrease of 1 unit in maternal BMI increased odds of schizophrenia by 9% (OR = 1.09, 95%CI 1.02–1.17).
Measurement: timing of weight and height measures of women not defined.
Control of error: no adjustment for maternal PGDM or GDM, SES or familial risk of schizophrenia.
NOS = 6
 Kawai et al. 2004 (34)∼20-year-olds (cases: N = 52, controls: N = 6570). Born on or after 1966Case–control study (Japan)BMI measured at first and last antenatal care visits by clinic personnelTo assess associations between maternal antenatal factors and schizophrenia in offspringSupportSampling: case–control study, relatively young sample.
Control of error:
 No adjustment for SES, maternal mental illness during pregnancy other than psychoses.
 No adjustment for maternal PGDM or GDM (no mothers in study had PGDM or GDM).
NOS = 4
BMI as a continuous measure in early (GA = ∼18.5 weeks) and late (GA = ∼38.3 weeks) pregnancySchizophrenia (psychiatrist diagnosed)For every 1 unit increase in early pregnancy BMI, odds of schizophrenia increased 24% (OR = 1.24, 95%CI 1.02–1.50).
For every one unit increase in late pregnancy BMI, odd of schizophrenia increased 19% (OR = 1.19; P < 0.05).
Eating disorders
 Favaro et al. 2006 (35)Females born between 1971 and 1979 (187 cases and 554 controls)Case–control study of females born on 2 wards in Italy. Cases supplemented by patients from eating disorder clinics (Italy)Pregnancy weight gain from hospital records.To explore the role of obstetric complications in the development of eating disordersNo supportSampling: some ascertainment bias in the collection of cases. Weight gain standards applied to all women regardless of their pre-pregnancy BMI.
Reference group:
 7–15 kg
 >15 kg
Anorexia and Bulimia Nervosa (psychiatrist diagnosed)Anorexia: OR = 0.8, 95%CI 0.4–1.7
Bulimia: OR = 0.9, 95%CI 0.4–2.3
Control of error:  No examination of familial risk of eating disorders.
 No adjustment for maternal PGDM or GDM.
NOS = 5
 Allen et al. 2009 (36)14-year-old males and females born between 1989 and 1991. N = 1597Cohort (Australia)Maternal BMI self-reported at 16 weeks after LMPEating disorder cases (assessed using a 24-item scale adapted from the Child Eating Disorder Examination and Eating Disorder Examination Questionnaire)Support
Each increase of 1 unit in maternal BMI increased odds of eating disorder caseness by 11% (OR = 1.10, 95%CI 1.05–1.16).
Each increase of one unit in maternal BMI increased odds of a diagnosed depressive/anxiety disorder by 7% (OR = 1.07, 95%CI 1.03–1.11).
Sampling: high sample attrition.
Measurement: maternal weight was retrospectively self-reported.
Control of error:
 No adjustment for parental eating disorders.
 No adjustment for maternal PGDM or GDM.
NOS = 6

Seven studies used pre-pregnancy BMI as their predictor (25–28,30–32), one examined maternal BMI in the first trimester (29), two in the second (34,36) and one in the third (33). Only three studies examined pregnancy weight gain (25,29,35). Sample sizes in these studies ranged from 169 to 21 090 mother–offspring pairs. Five were case–control studies and seven contained data from prospective birth cohorts.

Of the 12 studies examined, five provided clear support for an association between maternal obesity and neurodevelomental problems. One of these examined childhood IQ (33), one looked at ADHD in children (29), two focused on schizophrenia in adults (32,34) and one examined eating disorders in adolescence (36). Three studies provided mixed support of a link according to our classification system: one for childhood IQ (27), one for ADHD (30; which also suggested that negative emotionality was also increased in the offspring of mothers with pre-pregnancy obesity) and one of schizophrenia (31). Of the remaining four studies: two on foetal alcohol syndrome (27,28), one on schizophrenia (33) and one examining eating disorders in adults (35) did not support the presence of a link.

Of the 12 studies reviewed, six attempted to control for a familial or genetic risk of the outcome in mothers (25,30–34) and seven attempted to adjust for potential post-natal socioeconomic confounders that are likely to have been present pre-natally (25,26,30,31,35,36). All three of the studies that adjusted for both types of confounders (25,30,31) provided some support for an association between maternal obesity and poorer neurodevelopmental outcome. No study adjusted for the presence of maternal DM or other perinatal factors such as macrosomia or birth complications. None of the reviewed studies examined the role post-natal factors might have played in mediating the association between pregnancy obesity and offspring outcomes. The sampling frames of eligible studies were extremely variable, ranging from geographically limited high-risk groups to multinational birth cohorts.

Five out of the seven studies that examined pre-pregnancy obesity, one that assessed BMI in the first trimester and both studies that examined it in the second trimester found BMI to be a significant predictor of neurodevelopmental problems in offspring. However, despite the fact that the only study that assessed BMI in the third trimester failed to support a link, and Kawai and colleagues' data (34) suggested that increased second trimester BMI was more deleterious than third, heterogeneity in the studies examined precludes firm conclusions from being made regarding the presence of a sensitive period of exposure to maternal obesity.

Overall, no particular study design, sample characteristics or obesity definitions seemed to predict significant associations. Moreover, no conclusions can be made with respect to the existence of a dose–response effect for maternal obesity on neurodevelompental outcomes.

Discussion

Despite the fact that relatively few studies have examined associations between maternal obesity prior to and during pregnancy and neurodevelopmental outcomes in offspring, some evidence suggests that it may be associated with an increased risk of certain cognitive and psychiatric problems across the lifespan. The results of the studies that comprise this body of evidence are inconsistent however, and it is not possible to definitively conclude that maternal obesity negatively impacts neurodevelopment in humans at present time. More work needs to be done to examine if obesity pre-natally programmes or obesity is causally related to adverse neurodevelopmental outcomes in offspring.

The finding that the offspring of these gestations may be at increased risk of reduced IQ and symptoms of ADHD in childhood, eating disorders in adolescence and perhaps even non-affective psychotic disorders as adults warrants further study, particularly in light of the current worldwide obesity epidemic. If maternal obesity is causally related to these outcomes, given the ever-increasing number of obese women of childbearing age seen around the world, the prevention and treatment of obesity might lead to significant health gains for these women and their offspring.

While methodologically limited, two studies supported the existence of an association between maternal pre-pregnancy and early pregnancy obesity and intellectual disability in their children. In one of these studies, Heikura and colleagues (26) examined associations between pre-pregnancy BMI and intellectual disability in youth. They found that maternal obesity was a predictor of intellectual disability in those born in 1986 [odds ration (OR) = 3.6, 95% confidence interval (CI) 2.0–6.6] but not those born in 1966 (OR = 1.3, 95%CI 0.5–3.1). This finding raises the possibility that pre-pregnancy obesity does not directly programme low IQ, but that the observed association is due to unmeasured confounders. Unfortunately, these authors were not able to determine if a linear association existed between increasing BMI and decreasing IQ. Moreover, while the percentage of women with BMI > 30 was similar in both 1966 and 1986, the fact that all of these women were combined into a single group raises the possibility that there may have been a greater proportion at the more extreme high end of the BMI spectrum in 1986 than in 1966. It is also not clear if such era-specific effects are present for psychiatric outcomes.

Two studies by Rodriguez and colleagues also suggest that a link may exist between pre-pregnancy overweight and obesity and symptoms of ADHD in children (29,30). In both studies, teacher ratings of the offspring of obese mothers had increased levels of ADHD symptoms in childhood. However, in the latter study, parent reports of childhood ADHD symptoms and negative emotionality (i.e. sadness, fear and anger) failed to support a link. While this could mean that the effect is small or even absent, in light of the low to moderate associations seen between parent and teacher ratings of children's ADHD symptoms (41), one cannot conclusively rule out a link. That parental ADHD was adjusted for in their most recent study that provides tantalizing preliminary evidence that maternal obesity could pre-natally programme offspring ADHD. Future research should attempt to replicate these findings and extend the follow-up of these individuals to determine the adolescent and adult impact of this exposure.

The association between increasing maternal BMI and the risk of eating disorders in 14-year-old Australian youth was another notable finding, but it was limited by the fact that the authors were not able to adjust for eating pathology in the parents of these individuals (36). Therefore, while it is possible that there may be a direct intrauterine effect of obesity on this outcome, the effects could also be mediated via the impact of weight on the interactions between mothers, fathers and their children around eating and weight perception.

The studies contained in this review also suggest that there may be an effect of maternal obesity on the later risk of schizophrenia. Evidence of an association was seen in three of four studies that have examined this outcome, including one that adjusted for both maternal risk of psychosis and socioeconomic status (32). In another study, the authors reported that early gestation obesity had a larger effect than obesity that occurred later in pregnancy on risk, raising the possibility that either the timing or cumulative amount of exposure to obesity in pregnancy is relevant (34).

Even though all three studies that adjusted for both potential familial and environmental confounders reported evidence of associations between maternal obesity and neurodevleopmental problems, it does not necessarily follow that these effects are because of intrauterine programming. Confounders such as a familial risk of cognitive/psychiatric problems and low socioeconomic status may not only act individually to increase risk but are also likely to interact with one another or even be correlated. This complex interplay complicates the interpretations of studies in this area, even when both are measured and considered in statistical analyses. This complexity is increased further by the elevated risk observed among obese individuals for certain maladaptive personality characteristics (21), high anxiety and problems with decision making (42). These problems could contribute to the development of emotional and behavioural problems in offspring via their impact on rearing practices and in order to provide adequate control over error, they should be considered in these studies in addition to clinical diagnoses of specific psychiatric syndromes. Future studies would benefit from the utilization of genetically sensitive designs or the examination of subsequent pregnancies within the same mother–partner pair to attempt to separate the relative contributions of these genetic and environmental influences.

Seven of the 12 eligible studies (25–29,31,36) utilized retrospective maternal self-reports of weight and four (26–28,31) of height. In general, when self-reported, height tends to be overestimated and weight and BMI underestimated compared with direct measurement (43). Previous studies suggest that this discrepancy in BMI may be particularly pronounced for obese individuals (44). While the true impact of this bias is not known, it is possible that some of the eligible studies have underestimated the size of the association between maternal BMI and neurodevelopmental outcomes in offspring.

To date, relatively little attention has been paid to the biochemical mechanisms that might underlie observed associations. The intrauterine milieu of obese pregnancy is complex but increased levels of oestrogen (45), cortisol (46), free fatty acids (47), pro-inflammatory cytokines (48) and oxidative stress (49) might play a role. Post-natal mediating factors such as obstetric complications, later childhood weight and health problems, or even sympathetic nervous system overactivity (50) could also be involved. Decrements in IQ and emotion regulation problems in children might also be mediators of the link between maternal obesity and later risk of psychopathology.

Potential post-natal maternal mediators of the association between obesity in pregnancy and offspring neurodevelopmental problems could include the increased risk of depression seen in the post-partum period (19) and beyond (51), the maladaptive personality characteristics (21) of some obese women, problems with breastfeeding (12), and perhaps even increased levels of stress (52) and stress sensitivity (53). The children of obese women could also be affected by the increased levels of discrimination their mothers face both in the community and in the workplace (54).

Future studies in the area of maternal obesity in pregnancy and neurodevelopmental problems in offspring should focus on obtaining unbiased risk estimates by using direct measurements of height and weight if possible, examining a broad range of outcomes across the lifespan, utilizing generalizable samples and validated assessments of outcome and adjusting for putative confounders. Such work would also benefit from an examination of putative sensitive periods and the potential biochemical mediators or causes of the association. These could include the assessment of genetic variants germane to risk pathways and their interaction with maternal obesity and its associated biochemical sequelae. These data would provide us with important clues as to the pathophysiology of the neurodevelopmental outcomes of interest and potentially guide the development of novel therapeutic agents. Case–control studies and data from existing pregnancy and birth cohorts could be used to rapidly test the above hypotheses.

A clear demonstration of causal effects requires the use of experimental approaches. Researchers conducting experimental studies in humans aimed at examining strategies designed to reduce pre-pregnancy obesity and to help women reach and maintain healthy weights during pregnancy could consider monitoring cognitive and temperamental outcomes in infants and toddlers and, if resources exist, emotional and behavioural functioning in the youth offspring of treatment and control groups. However, these studies would need to be adequately powered to examine outcomes in offspring if they are to advance our current knowledge. If human data accumulate and support a link, animal studies could be used to develop a better understanding of the mechanisms underlying this association.

Given the increasing prevalence of maternal obesity, demonstrating that causal associations exist between it and neurodevelopmental outcomes in offspring is an important goal and could provide realizable targets for the primary prevention of cognitive and psychiatric problems. Certainly, more data are needed before obesity treatments can be touted as beneficial to offspring neurodevelopment.

Conflicts of Interest Statement

The authors have no conflicts relevant to this manuscript to disclose.

Acknowledgements

Dr Van Lieshout is supported by a Canadian Institutes for Health Research Fellowship and Dr Taylor by an Early Researcher Award from the Province of Ontario. Dr Boyle is supported by a Canada Research Chair in the Social Determinants of Child Health.

Ancillary