Cognitive and Mental Health of Young Mothers’ Offspring: A Meta-analysis

The protocol guiding this review was registered with PROSPERO (August 3, 2020; CRD42020152026). A systematic search of 5 databases (Medline, Embase, Web of Science, PsycINFO, and CINAHL) from their inceptions until February 2022 was conducted, including abstracts and gray literature. This was created in collaboration with a health sciences librarian and included terms for adolescent pregnancy combined with cognitive and mental health outcomes (Supplemental Figs 3A–3E).

Studies were selected if they included offspring born to a young mother (<21 years at delivery), included a control group of offspring born to mothers ≥21,²³  and assessed cognitive, externalizing, and/or internalizing mental health outcomes. Twenty one years of age was set as the cut point to capture the oldest definition of young or teenage childbearing found during a preliminary scan of the literature. Studies were excluded if they involved an intervention or if they were not available in English, as resource constraints did not permit translation. Studies were further restricted to those originating from Western countries, as young childbearing differs in non-Western and Western countries.²⁴  Cognitive outcomes included all those relating to cognitive functioning, learning and educational attainment. Externalizing was defined as any of conduct disorder, delinquency, and/or other behavioral problems; attention deficit/hyperactivity disorder (ADHD) or hyperactivity; substance misuse; and violent and nonviolent criminal convictions. Internalizing grouped depression, anxiety, suicidal ideation, psychotic disorders, and any other general measures of psychological functioning and mental health. Four independent reviewers (L.C., M.F., C.L., Z.T.) screened titles and abstracts and conducted full-text review. The ancestry approach identified studies not located by our search.

A data extraction template was pilot tested by 4 independent reviewers (L.C., M.F., C.L., Z.T.) on 5 randomly selected studies. The result was a standardized extraction form that included information on study design, duration, measurement occasion(s), and study eligibility criteria; study objectives; sample size; and maternal and offspring characteristics, outcomes and measures, covariates, and study outcomes. If any data were missing, we contacted its lead, second, or senior authors by e-mail. The extraction form also extracted risk of bias using the Newcastle-Ottawa scales for cohort, case control, and cross-sectional studies.^25,26  These scales rank selection bias, between-group comparability, and exposure in outcome ascertainment, and score cohort and case control studies out of 9 stars and cross-sectional studies out of 10. A lower number of stars indicates a higher risk of bias. Disagreements that arose during the data extraction and bias assessments were again reviewed and resolved by the senior author (R.V.).

Offspring outcomes were examined in 4 separate developmental stages: infancy (0–3 years), childhood (4–9 years), adolescence (10–19 years), and adulthood (20+ years).^27,28  Studies were included in meta-analyses if they assessed an outcome in the same category as another study (judged according to our a priori criteria) within the same offspring age group. The minimum number of studies per meta-analysis was set as 2, provided they were meaningfully similar with respect to sample characteristics and outcome measurement. This threshold was chosen in the interest of providing the most complete age-specific estimates within all outcomes, acknowledging that our age-stratified approach led to a small number of studies in some categories.

Meta-analyses were conducted using a random-effects model in RevMan 5.3 to produce standardized mean differences (SMD) with 95% confidence intervals (CIs) for each outcome. Where dichotomous data were reported, odds ratios were converted to SMDs with CIs.²⁹  95% CIs were converted to standard errors, so the generic inverse variance outcome type could be used to generate forest plots.²⁹  I² tests were used to assess heterogeneity. Egger’s publication bias test was planned for meta-analyses with >10 studies. For analyses with fewer studies, funnel plots were visually inspected.³⁰ 

Planned subgroup analyses examined offspring sex, single parent status, and first-born offspring.^31,32  Predefined sensitivity analyses examined studies with a low risk of bias or studies that assessed outcomes with methods other than self-report.

After removing duplicate records, 4318 titles and abstracts were screened (κ = .96). 140 full-text articles were assessed and 28 were included.^{17–20,23,31–53}  The manuscript from Zimmerman et al contained 2 studies and both were included separately.⁵⁴  The ancestry approach identified 6 additional studies,^54–59  bringing the total to 35 (Fig 1).

Seventy-five outcomes were extracted and 51 were eligible for meta-analysis. Twenty-four were ineligible because they were the lone study to assess an outcome within an age group,^{34,36,49,51,52,54}  they used the same sample as another study,⁴⁵  or because data were insufficient to calculate SMDs.^{19,32,33,42,43,46,47,51}  These studies were included in a narrative synthesis.

Studies contained outcomes on 6 648 357 offspring, 375 561 of which were born to a young mother. Sample sizes varied from 42 to 1 826 992 participants.^38,50  Offspring ages ranged from 6 months to 40 years.^35,37  Across 19 studies, 19 cognitive outcomes were reported. Twenty-seven different studies assessed a total of 56 mental health outcomes, and these were further divided into 2 subcategories. The first (n = 38 outcomes) included externalizing, substance misuse, and antisocial behavior, and the second (n = 18 outcomes) included internalizing and general mental health or illness outcomes. A detailed breakdown of included studies’ characteristics, along with their methods of assessment, can be found in Tables 1 and 2.

Twenty-one studies were rated at low risk of bias (defined as >7 stars on a Newcastle-Ottawa scale) and eleven moderate risk (6–7 stars). Three were deemed at high risk of bias (0–5 stars) because case and control groups were not representative of the broader population (eg, recruited only from a psychiatric unit) (Supplemental Tables 3A–3C). Other common bias sources included small or inadequate sample sizes and self-reported outcomes. Visual inspection of funnel plots revealed no asymmetry, suggesting a low risk of publication bias for all meta-analyses.

Of the 19 cognitive outcomes, 17 reported that young mothers’ offspring experience greater cognitive or learning problems and/or achieve fewer years of education. Fourteen cognitive outcomes were pooled in 4 meta-analyses: infant cognitive problems (n = 4 outcomes), child cognitive and learning problems (n = 3), adolescent cognitive, and learning problems (n = 5) (Fig 2), and adult educational attainment (n = 2). The offspring of young mothers had poorer cognitive function as infants (SMD = 0.30 [0.06 to 0.55] I² = 76%),^35,38–40  children (SMD = 0.43 [0.07 to 0.79] I² = 99%),^31,41,59  and as adolescents (SMD = 0.43 [0.24 to 0.62] I² = 82%).^17,43,44,53  Our meta-analyses showed no difference in adult educational attainment (SMD = −0.37 [−0.81 to 0.07] I² = 97%) (Supplemental Fig 4A).^23,50  No subgroup analyses were conducted because of a lack of uniform reporting of separate data, and sensitivity analyses did not reveal any significant differences in our findings (Supplemental Table 4).

The 5 cognitive outcomes that could not be meta-analyzed suggested that the offspring of young mothers have poorer cognitive functioning. Morinis et al and Cornelius and colleagues suggested that the cognitive functioning of children born to young mothers was worse than their counterparts at 5 (MD = −8.6 [−10.6 to −6.5]) and 6 years using the BAS-II Naming Vocabulary scale and the Stanford-Binet IQ score, respectively.^33,42  Dahinten and colleagues also showed 10 to 11-year-olds had poorer math scores on the CAT/2.¹⁹  This was consistent with Levine et al’s finding that young mothers’ offspring were more likely to repeat a grade by 14–15 years (SMD = 0.63),⁴⁵  and Khatun et al’s report that 21-year-olds born to young mothers had lower IQs (SMD = −0.32 [−0.43 to −0.21]).³⁷ 

Thirty-eight externalizing, substance misuse, and criminal behavior outcomes were assessed and 28 reported increased risk in the offspring of young mothers. Seven meta-analyses were generated: child externalizing (n = 2 outcomes), child hyperactivity (n = 2), adolescent delinquency or conduct disorder (n = 7) (Fig 3), adolescent substance misuse (n = 5), adolescent hyperactivity and ADHD (n = 4), adult violent criminal convictions (n = 2), and adult substance misuse (n = 3).

As children, the offspring of young mothers exhibit somewhat higher levels of externalizing than their peers (SMD = 0.36 [0.03 to 0.68] I² = 89%)^57,59  and are more hyperactive (SMD = 0.28 [0.14 to 0.42] I² = 48%).^57,59  These trends persist into adolescence, where they have more symptoms of delinquency and conduct disorder (SMD = 0.24 [0.12 to 0.36] I² = 77%),^{17,18,20,44,48,53}  substance misuse (SMD = 0.21 [0.03 to 0.40] I² = 72%),^17,44,48,53  and hyperactivity (SMD = 0.42 [0.34 to 0.50] I² = 0%).^18,48,55,56  As adults, they are more likely to be convicted of violent crimes (SMD = 0.36 [0.22 to 0.50] I² = 100%),^37,50  and to misuse substances (SMD = 0.44 [0.23 to 0.65] I² = 99%) (Supplemental Fig 4B).^37,49,50  No subgroup analyses were conducted because of a lack of uniform reporting of separate data on offspring sex, birth order, and paternal presence from each study, and sensitivity analyses did not reveal any significant differences in our findings (Supplemental Table 4).

Of the 13 outcomes that could not be meta-analyzed, Agnafors et al showed young mothers reported more externalizing symptoms in their 3-year-olds on the Child Behavior Checklist (CBCL) (SMD = 0.45 [0.01 to 0.89]),⁵⁴  and Christ et al found these offspring were more likely to self-report DSM-III symptoms of conduct disorder at ages 6 to 13.⁴⁷  Guèvremont and colleagues found young parents reported more conduct problems and hyperactivity in their children on the Strengths and Difficulties Questionnaire (SDQ).⁵⁸  Chang et al reported these children had 78% higher odds of ADHD diagnosis or being treated for it (hazard ratio [HR] = 1.78 [1.72 to 1.84]).⁴⁶  Dahinten et al also reported no difference between young mothers’ offspring and controls in self-reported delinquency or hyperactivity at 10 to 11 years on the CBCL.¹⁹  East et al also found no difference in behavioral problems in 12-year-olds as reported by their teachers and mothers on the BPC.⁴³  However, Levine et al showed the 14 to 15-year-old offspring of young mothers were more likely to report fighting at work or school (SMD = 0.59) and using cannabis (SMD = 0.13).⁴⁵  Pogarsky et al further reported that, in a group of adolescents, 47% of those born to young mothers admitted to engaging in delinquent acts, compared with 21% of controls.³²  Harden et al further found 14 to 39-year-old offspring self-reported more symptoms of conduct disorder.⁴⁹  Finally, Coyne et al used crime registry linkage to show these offspring were more likely to be convicted of a crime as adults (HR = 1.64).⁵¹ 

Of the 18 internalizing and general mental health outcomes in the included studies, 11 reported that the offspring of young mothers experienced more problems. Seven showed no difference, and a lone study found levels of anxiety were lower in young mothers’ offspring. Three meta-analyses were conducted: child internalizing (n = 3 outcomes), adolescent anxiety (n = 3), adolescent depression (n = 3), and adult general mental illness (n = 3).

The offspring of young mothers have higher levels of overall internalizing symptoms as children (SMD = 0.29 [0.14 to 0.45] I² = 79%),^31,57,59  but not anxiety (SMD = −0.11 [−0.31 to 0.10] I² = 34%),^18,53  or depression (SMD = 0.12 [−0.01 to 0.26] I² = 5%) as adolescents.^17,53  As adults, they are more likely to experience mental illness (SMD = 0.35 [0.34 to 0.36] I² = 0%) (Fig 4) (Supplemental Fig 4C).^23,37,49  No subgroup analyses were conducted because of a lack of uniform reporting of separate data, and sensitivity analyses did not reveal any significant differences in our findings (Supplemental Table 4).

Six internalizing and general mental health outcomes could not be meta-analyzed. Andreozzi et al examined maternal-infant attachment of 18-month-olds using Ainsworth’s strange situation task and found no difference between these groups in secure versus insecure attachment (χ² = 0.003, P = .99).³⁴  Agnafors and colleagues showed young mothers reported higher levels of internalizing in their 3-year-olds on the CBCL (SMD = 0.48 [0.04 to 0.92]),⁵⁴  and Guèvremont et al found young parents reported more emotional symptoms in their 2 to 5-year-olds on the SDQ.⁵⁸ 

Among adolescents, Dahinten et al showed the offspring of young mothers self-reported being no more anxious than their peers,¹⁹  and East and colleagues found no difference in overall self-reported general psychological functioning.⁴³  One case-control study by Zilikis suggested the adolescents of young mothers were more often admitted to psychiatric units for suicidal ideation (SMD = 0.71).⁵² 

Across 35 studies, we extracted data on 19 cognitive and 56 mental health outcomes in over 6 million offspring from 6 months to 40 years of age. Our meta-analyses suggest that the offspring of young mothers have delayed cognitive development in infancy, struggle more with school during childhood and adolescence, and may attain fewer years of education by adulthood. More symptoms of hyperactivity are seen in these children, and externalizing, substance misuse, and criminal behavior are more common during their adolescence and adulthood. Findings for internalizing psychopathology are mixed, with higher levels in childhood but not adolescence and elevated rates in adulthood. The effect sizes of these findings are small but largely consistent with those in our narrative synthesis and suggest being born to a young mother has adverse effects on cognitive function and mental health across the lifespan.

Disparities in cognitive functioning are present as early as 6 months and may not resolve as offspring age. The magnitude of this effect may be even larger than the impact of socioeconomic disadvantage,⁶⁰  though we caution comparing these outcomes without acknowledging the likely confounding relationship between them. Nonetheless, our findings illustrate the negative cognitive impact of being born to a young mother and suggest that this exposure alone is useful for identifying offspring who may be at risk. Still, the mechanisms by which this risk develops and persists are unclear. Socioeconomic challenges may contribute along with higher rates of pregnancy and delivery complications.⁶¹  Later postnatal environmental factors likely also play a role. Young mothers experience more stressors and receive less support, which may compromise their focus on their offspring⁶²  and impair their ability to provide as much stimulation as mothers ≥21.⁶³ 

Findings of increased levels of externalizing, substance misuse, and criminal behavior were similarly persistent. These offspring are more prone to symptoms of externalizing and hyperactivity as children, and increased levels of ADHD, substance misuse, and delinquency are seen in adolescence. This is consistent with Lee et al’s meta-analysis,²¹  and suggests being born to a young mother is also a useful exposure for estimating risk of externalizing and criminal behaviors. Confounding factors notwithstanding, it may be an even stronger predictor of externalizing than being a victim of abuse or experiencing socioeconomic disadvantages.⁶⁴ 

Although the mechanisms underlying these findings are not entirely understood, young mothers may have higher levels of externalizing problems themselves, suggesting that genetics and the early environment play a key role. Young mothers may also apply discipline less consistently and/or use more coercive parenting,⁶⁵  which can negatively influence their ability to regulate emotions.⁶⁷  Externalizing problems generally exhibit homotypic continuity, and the home environments provided by young mothers may continue to contribute to a higher risk of delinquency in offspring as they mature.⁶⁷ 

Young mothers’ offspring also appear more prone to internalizing problems. Being born to a young mother was associated with higher levels of internalizing symptoms in children and increased the likelihood of mental illness diagnoses in adulthood. These increased risks could be contributed to by genetic inheritance along with parental stress and relationship challenges.^68,69  It should also be noted that two-thirds of the samples in the adolescent anxiety and depression meta-analyses sampled Black offspring exclusively, a group in which past studies have suggested young motherhood may have a less detrimental effect.⁷⁰  Regardless, more high-quality research is needed in this area to definitively determine relative risks of internalizing difficulties.

This review should be taken within the context of its limitations. First, whereas grouping analyses by age provides a summary, the analyses include studies with different samples and do not necessarily indicate the expected development of a single individual born to a young mother. Many studies also suffer from attrition, as the offspring of young mothers were more likely to be lost to follow-up than controls. Our search strategy did not capture gray literature, and several included studies examined unique samples that may not be fully generalizable (eg, Morris et al³³ ). In addition to sampling constraints, there were limitations in outcome measurement. Thirty of the 73 outcomes (including 27 of the 55 mental health outcomes) were assessed via offspring self-report. Despite having restricted our search to Western countries, there may still be differences between these countries in terms of cultural norms and the resources available to young mothers, and our analysis does not allow us to consider how these differences affect offspring. Furthermore, the breadth of our meta-analyses comes at the expense of having relatively few studies within each. Thus, any conclusions drawn about effect sizes should be considered preliminary.

The relationship between being born to a young mother and experiencing problems with mental health and cognition is likely confounded by several factors, including maternal mental health, socioeconomic background, and the degree to which offspring have access to family support. It is important to elucidate these factors to better understand why the offspring of young mothers are worse off than their peers and to establish a causal link, however this review is limited in its ability to do either given a lack of uniform reporting data on our identified subgroup and sensitivity analyses. Regardless, our review establishes absolute magnitudes of risk in mental health and cognitive functioning by age group and furthers the notion that being born to a young mother is a useful screening criterion for identifying offspring in need of support. We present our hypotheses as to why this effect occurs not to hint at a causal interpretation of our results, but instead to provide avenues for future investigations.

Our understanding of the cognitive and mental health outcomes of the offspring of young mothers could be improved by following up cohorts longitudinally over multiple developmental periods to better understand their trajectories and identify specific periods and targets for prevention and treatment. Studies should also attempt to obtain outcome data from interviews and linking to national registries and other existing large databases. It is crucial for more of this work to include developing countries, considering greater than 90% of adolescent pregnancies occur in these regions.⁷¹ 

Despite its limitations, the present systematic review is the largest synthesis of the outcomes of young mothers’ offspring to date and the first to provide estimates of risk by developmental stage. Although rates of adolescent pregnancy in developed countries have fallen over the past 30 years, young motherhood costs an estimated $9.4 billion annually in the United States alone,⁷²  and early childbearing remains a considerable barrier to the equity of health and opportunity of young mothers and their offspring everywhere.⁷³  Our findings reinforce the importance of prevention efforts but more importantly emphasize the need to monitor these offspring and provide targeted assistance to them and their families. Indeed, it may be beneficial to intervene early, given that this population can fall behind their peers as young as 3 years of age. Supporting the development of young mothers and their offspring is a critical public health goal with potential benefits for these individuals, their families, and society.

We thank Jo-Anne Petropoulos for her assistance in developing and refining our search strategies.

ADHD

attention deficit hyperactivity disorder

CI

confidence interval

HR

hazard ratio

SMD

standardized mean difference