Introduction
Methods
Search procedures
Inclusion/exclusion criteria
Search results
Assessment of the quality of evidence
Barker et al. [41] | de Brujin et al. [38] | Davis and Sandman [49] | Garthus-Niegel et al. [27] | Leis et al. [42] | Loomans et al. [43] | Murray et al. [40] | O’Connor et al. [46] | O’Connor et al. [48] | O’Donnell et al. [50] | Pickles et al. [45] | Prenoveau et al. [47] | Sharp et al. [39] | Van den Bergh & Marcoen [44] | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Did the study address a clearly focused issue? | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 |
Is the population clear? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Are the factors studied clear? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Are the outcomes clear? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Is it clear whether the study tried to detect a beneficial or harmful effect? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Was the sample recruited in an acceptable way? | 1 | 0 | 1 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 2 | 1 | 2 | 1 |
Was the cohort representative of a defined population? | y | n | n | n | n | n | n | n | n | n | y | n | y | n |
Was everybody included who should have been included? | n | n | y | n | y | n | n | n | n | y | y | y | y | y |
Was the exposure accurately measured to minimise bias? | 1 | 1 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Did they use objective measurements? | n | n | n | n | n | n | y | n | n | n | n | n | n | n |
Are they valid measures? | y | y | y | Partial | y | y | y | y | y | y | y | y | y | y |
Were all the subjects classified into exposure groups using the same procedure? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Was the outcome accurately measured to minimise bias? | 2 | 1 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 2 | 2 | 2 | 1 |
Did they use objective measurements? | Partial | Partial | n | n | Partial | Partial | y | n | n | n | n | Partial | n | Partial |
Are they valid measures? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Were the measurements methods similar in the different groups? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Have the authors identified and taken into account of all important confounding factors? | y | y | y | y | y | y | n | y | y | y | y | y | y | n |
Was the follow-up of subjects complete enough? | y | partial | y | y | y | y | y | y | y | y | y | y | y | Partial |
Was the follow-up of subjects long enough? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Results | 1 | 1 | 2 | 1 | 1 | 1 | 2 | 2 | 2 | 1 | 1 | 1 | 1 | 1 |
Were confidence intervals given? | n | n | y | n | n | n | y | y | y | n | n | n | n | n |
Are conclusions adequately supported by the results? | y | y | y | y | y | y | y | y | y | y | y | y | y | y |
Can the results be applied at population level? | 2 | 0 | 0 | 0 | 2 | 0 | 0 | 0 | 0 | 0 | 2 | 0 | 2 | 0 |
Total quality score out of 12 | 9 | 5 | 8 | 6 | 9 | 6 | 7 | 7 | 7 | 7 | 10 | 7 | 10 | 6 |
Study | Location | Design | Participants and study | Mean maternal age | Measures | Informants on child outcomes | Psychotropic drugs use in pregnancy | Tests of gender differences | Quality rating |
---|---|---|---|---|---|---|---|---|---|
Barker et al. [41] | England | Prospective cohort study | 3298 mother–child pairs ALSPAC | During pregnancy = 28 years | Mothers: CCEI Prenatal assessments at 32 weeks pregnancy Anxiety not measured in the postnatal period Children: DAWBA aged 7-8 years | Mothers Fathers | Not reported | No | Q = 9 |
de Bruijn et al. [38] | Netherlands | Prospective cohort study | 444 women and children | At delivery = 30.7 years | Mothers: STAI (State subscale) and SCL-90 (anxiety scale) Prenatal assessments at 12, 24 and 36 weeks pregnancy Children: CBCL aged 14-54 months | Mothers Fathers | Not reported | Yes | Q = 5 |
Davis and Sandman [49] | Southern California | Prospective cohort study | 178 mother–child pairs | At time of assessment = 38.1 years | Mothers: STAI (State subscale) and Pregnancy Related Anxiety Scale Prenatal assessment at 19, 25 and 31 weeks pregnancy Children: CBCL aged 6-9 years | Mothers | Not reported | No | Q = 8 |
Garthus-Niegal et al. [27] | Norway | Prospective cohort study | 1472 women and children The Akershus Birth Cohort (ABC) | Mean maternal age not reported | Mothers: Hopkins Symptom Checklist and The Impact of Events Scale Postnatal assessment at 8 weeks postpartum Children: ASQ:SE aged 2 years | Mothers | Not reported | Yes | Q = 6 |
Leis et al. [42] | England | Prospective cohort study | 2891 mother–infant pairs ALSPAC | During pregnancy = 29.1 years | Mothers: CCEI Prenatal assessment measured at 18 and 32 weeks pregnancy Postnatal assessment measured at 8 weeks and 8 months postpartum; 21, 33, 61, and 73 months and 11 years of childhood Children: SDQ aged 10–11 years | Mothers Teachers | Not reported | No | Q = 7 |
Loomans et al. [43] | The Netherlands | Prospective cohort study | 3446 mothers 3520 teachers 3758 children ABCD longitudinal study | During pregnancy = 31.8 years | Mothers: STAI (State subscale) Prenatal assessment measured at 16 weeks pregnancy Children: SDQ aged 5 years | Mothers Teachers | No –psychotropic drug use excluded from sample | Yes | Q = 6 |
Murray et al. [40] | England | Nested case control study | 73 mothers diagnosed with social anxiety disorder and 63 non-anxious controls | Age at follow-up Index group = 36.2 years Control group = 36.6 years | Mothers: SCID; Social Interaction and Anxiety Scale; Social Phobia Scale Prenatal assessment measured at 20–30 weeks pregnancy Children: ADIS-P; CBCL aged 4–5 years | Mothers Teachers | Not reported | No | Q = 7 |
O’Connor et al. [46] | England | Prospective cohort study | 7,448 women and children ALSPAC | During pregnancy = 28 years | Mothers: CCEI Prenatal assessment measured at 18 and 32 weeks pregnancy Postnatal assessment measured at 8 weeks and 8 21, 33 months Children: SDQ aged 47 months | Mothers | Not reported | Yes | Q = 7 |
O’Connor et al. [48] | England | Prospective cohort study | 6,493 women and children ALSPAC | During pregnancy = 28 years | Mothers: CCEI Prenatal assessment measured at 18 and 32 weeks pregnancy Postnatal assessment measured at 8 weeks, 8, 21 and 33 months Children: SDQ aged 81 months | Mothers | Not reported | Yes | Q = 7 |
O’Donnell et al. [50] | England | Prospective cohort study | 7944 women and children ALSPAC | During pregnancy = 28 years | Mothers: CCEI Prenatal assessment measured at 18 and 32 weeks pregnancy Postnatal assessment measured at 33 months. Children: SDQ aged 4, 7, 9, 11 and 13 years | Mothers | Not reported | Yes | Q = 7 |
Pickles et al. [45] | England | Prospective cohort study | 813 mothers and infants Wirral Child Health and Development Study | At 20 weeks pregnancy = 26.9 years | Mothers: STAI (State scale); The Pregnancy-Specific Anxiety Scale Prenatal assessment measured at 20 weeks of pregnancy Postnatal assessment measured at 9 weeks, 14 months and 3.5 years Children: CBCL aged 3.5 years | Mothers | Not reported | Yes | Q = 10 |
Prenoveau et al. [47] | England | Prospective cohort study | 296 mothers and infants Oxford Parent Project | At 3 months postpartum = 32.3 years | Mothers: GAD-Q; SCID Postnatal assessment measured at 9 weeks and 2, 3, 6, 10, 14 and 24 months. Children: CBCL aged 2 years | Mothers | Not reported | No | Q = 7 |
Sharp et al. [39] | England | Prospective cohort study | 316 mothers and infants Wirral Child Health and Development Study | During pregnancy = 26.8 years | Mothers: STAI (State scale) Prenatal assessment measured at 32 weeks of pregnancy Postnatal assessment measured at 5, 9 and 29 weeks, 14 months and 2.5 years. Children: CBCL aged 2.5 years | Mothers | Not reported | Yes | Q = 10 |
Van den Bergh and Marcoen [44] | Belgium | Prospective cohort study | 71 mothers and children | During pregnancy between 18 and 30 (mean age not reported) | Mothers: STAI Prenatal assessment measured at 12–22 weeks and 32–40 weeks pregnancy Children: CBCL; STAIC aged 8–9 years | Mothers Teachers | Not reported | No | Q = 6 |
Study | Covariates | Data analysis | Primary results | Main limitations | |||
---|---|---|---|---|---|---|---|
Prenatal anxiety | Effect size | Postnatal anxiety | Effect size | ||||
Barker et al. [41] | Low SES No partner Teen pregnancy Criminal behaviour Substance misuse Cigarette smoking Prenatal depression Postnatal depression Prenatal anxiety Postnatal anxiety | Single path analytic model | Prenatal anxiety predicted increases in child internalizing difficulties after controlling for postnatal anxiety, pre- and postnatal depression and other risk factors | Small | Postnatal anxiety not reported (0 to 12 months postpartum) | N/A | Sample: low rates of ethnic minorities in sample. High attrition rates Measures: did not examine role of prenatal timing of exposure to maternal anxiety or severity of anxiety. All measures based on maternal reports, raises possibility of shared method variance Mechanisms: no measurement of parenting quality as potential mediator or moderator |
de Bruijn et al. [38] | Educational level of parents Prenatal smoking Women’s parity Child’s age Postnatal anxiety (mothers and fathers) Postnatal depression (mothers and fathers) | Linear Multivariate Regression | After controlling for confounding factors, significant effects were found for mean prenatal STAI scores on internalizing problems in girls, as reported by fathers No significant effects found for mothers reports | Small | Unobtainable | N/A | Sample: no ethnic diversity (Caucasian only). Participants had higher education level and smoked less during pregnancy Measures: large age range reported on child outcomes Mechanisms: no measurement of parenting quality as potential mediator or moderator |
Davis and Sandman [49] | Gestational age at birth Mother’s emotional distress at follow-up Maternal education Child sex | Linear and logistic multivariate regression | Elevated prenatal cortisol and elevated pregnancy-specific anxiety associated with increased anxiety in children following adjustment for covariates. No significant effects of generalised anxiety during pregnancy | Small | Not investigated | N/A | Sample: low rates of ethnic minorities. Low rates of low SES Measures: all measures based on maternal reports, raises possibility of shared method variance Mechanisms: did not account for exposure to extreme stress. No measurement of parenting quality as potential mediator or moderator |
Garthus-Niegal et al. [27] | Postnatal depression Postnatal anxiety Age at delivery Maternal education Gestational age Current child health problems Child gender | Linear multivariate regression | Not investigated | N/A | After adjustment for covariates, postnatal PTSD symptoms predicted increased socio-emotional problems, particularly for boys. No effects for symptoms of generalised postnatal anxiety | Small | Sample: selective attrition. Education and symptoms of depression were more likely to drop out. Measures: all data based on mother reports. Relationships could have occurred due to common method bias Mechanisms: No measurement of parenting quality as potential mediator or moderator |
Leis et al. [42] | Marital status Maternal age at birth Child birth weight Child gender Maternal educational attainment Cigarette smoking Alcohol use Postnatal Anxiety Pre- and postnatal Depression | Linear multivariate regression | After adjustment for covariates, elevated symptoms of prenatal anxiety predicted increased offspring emotional problems. Finding significant for maternal, but not teacher reports on child outcome | Small | Unobtainable | N/A | Sample: high attrition rates limits confidence in generalizability of results Measures: did not examine role of prenatal timing of exposure to maternal anxiety Mechanisms: did not account for family functioning, paternal mental health. No measurement of parenting quality |
Loomans et al. [43] | Birth weight Gestational age Parity Ethnicity Maternal educational level Prenatal smoking Prenatal alcohol use Current anxiety, stress & depression Parental history of psychopathology | Linear multivariate regression | After controlling for covariates, elevated prenatal anxiety symptoms predicted increased child emotional problems, as reported by mothers. No effects on teacher reports after controlling for covariates | Small | Not investigated | N/A | Sample: non-random sample attrition. Women who were younger, less educated, did not have a western background, and were more anxious were less likely to participate Measures: significant findings only for maternal reports, raises the possibility of shared method variance Mechanisms: no measurement of parenting quality as potential mediator or moderator |
Murray et al. [40] | Child gender Mother’s IQ Birth order Maternal age SES | Linear or logistic multiple regression | Not investigated | N/A | Children of socially anxious mothers had significantly increased emotional problems as reported by mothers. This effect was moderated by infant attachment style and only significant for securely attached children. No significant findings For teacher reports | Small | Sample: low rates of ethnic minorities and low SES Measures: CBCL and ADIS-P relied on maternal reports—potentially confounding effects of the maternal disorder and shared method variance Mechanisms: Did not examine father or family members who might compensate for difficulties experienced by mothers. Did not control for prenatal anxiety |
O’Connor et al. [46] | Gestational age Birth weight Mode of delivery Parity Smoking Alcohol consumption SES Maternal education Maternal age Prenatal and Postnatal depression Postnatal anxiety | Hierarchical logistic regression | After controlling for covariates, prenatal anxiety at 18 weeks predicted elevated emotional problems in girls Prenatal anxiety at 32 weeks gestation predicted elevated emotional problems in boys and girls | Small | Postnatal anxiety at 8 weeks predicted elevated emotional problems in boys and girls. Postnatal anxiety at 8 weeks was also a predictor of emotional problems in girls aged 4 years | Small | Sample: sample attrition was not random. Attrition was more likely in those with higher anxiety scores at earlier assessments. This could result in a diminished effect of prenatal anxiety Measures: all data based on mother reports, therefore, shared method variance could explain the effects Mechanisms: physiological mechanisms not accounted for. No measurement of parenting quality as potential mediator or moderator |
O’Connor et al. [48] | Maternal depression Postnatal Anxiety Gestational age Birth weight Mode of delivery Parity Prenatal smoking Prenatal alcohol use SES Maternal education Maternal age | Logistic regression | After controlling for covariates, elevated prenatal anxiety at 32 weeks predicted increased emotional problems in boys but not girls | Small | After controlling for covariates, elevated postnatal anxiety at 8 weeks predicted increased emotional problems in boys but not girls | Small | Sample: attrition was not completely random. Those who dropped out were more likely to be initially anxious and at greater psychological disadvantage Measures: all data based on mother reports, therefore, shared method variance could explain the effects Mechanisms: no measurement of parenting quality as potential mediator or moderator |
O’Donnell et al. [50] | Maternal SES Parenting behaviour Maternal age Smoking Alcohol/substance use Birth weight Gestational age Maternal postnatal depression & anxiety Paternal pre- and postnatal anxiety | Longitudinal growth model | Elevated prenatal anxiety predicted persistently higher emotional problems across childhood with no diminishment of effect into adolescence | Small | Unobtainable | N/A | Sample: low rates of ethnic minorities Measures: all data based on mother reports, therefore, shared method variance could explain the effects Mechanisms: did not have direct measures of biological mechanisms that might mediate the effect. No measurement of parenting quality as potential mediator or moderator |
Pickles et al. [45] | Maternal age Marital status Education SES Smoking Alcohol consumption Sex of infant Birth weight Postnatal depression and anxiety | Regression | Frequency of infant stroking modified the association between pregnancy-specific anxiety and maternal ratings of offspring emotional problems. The effect of prenatal generalised state anxiety no longer significant following adjustment for confounders | Small | Unobtainable | N/A | Sample: low rates of ethnic minorities Measures: all data based on mother reports. Relationships could have occurred due to common method bias Mechanisms: the data do not allow determination of what physiological mechanisms account for the observed associations |
Prenoveau et al. [47] | Mothers age Infant age Infant sex Infant birth order Mother marital status Postnatal depression Maternal anxiety Maternal depression | Latent trait-state occasion modelling | Not investigated | N/A | Maternal anxiety trait or state factors did not predict child emotional problems after controlling for the depression trait factor | Unobtainable | Sample: low rates of ethnic minorities Measures: all data based on mother reports, therefore, shared method variance could explain the effects Mechanisms: did not examine physiological mechanisms that could account for the observed associations |
Sharp et al. [39] | Psychological abuse Breast-feeding Mother’s age Marital status Education & SES Prenatal Smoking Alcohol consumption Infant sex Birth weight Maternal sensitivity Postnatal anxiety Postnatal depression | Generalised Linear Latent and Mixed Models | After controlling for covariates, prenatal anxiety predicted elevated child emotional problems for female offspring. High maternal stroking eliminated the association between prenatal anxiety and child emotional problems | Unobtainable | Unobtainable | N/A | Sample: low rates of ethnic minorities Measures: all data based on mother reports, therefore, shared method variance could explain the effects Mechanisms: did not examine physiological mechanisms that could account for the observed associations |
Van den Bergh and Marcoen [44] | Parental education Prenatal Smoking Birth weight Child gender Postnatal anxiety | Hierarchical regression | After controlling for covariates, prenatal anxiety at 12–22 weeks gestation was significantly associated with child self-reported anxiety No significant associations for mother or teacher-reported emotional problems | Small | Unobtainable | N/A | Sample: small sample, all Caucasian. Limited generalisability Measures: inconsistent findings across informants and measures Mechanisms: no measurement of parenting quality as potential mediator or moderator |