Maternal Age at Childbirth and Parental Reports of Attention-Decit/Hyperactivity Disorder and Learning Disability, Ages 4 to 15 Years

Background: Studies have shown that young maternal age at childbirth can increase the risk of attention-decit/hyperactivity disorder (ADHD) in the offspring, but a study of the U.S. population has not been reported. Moreover, there is no reported research on young and advanced maternal age at childbirth and on whether it can contribute to learning disability (LD) risk for offspring. Methods: This study evaluated the association between young and advanced maternal age at childbirth and offspring ADHD and LD risk in the U.S. population. Using data from 8,098 participants from the National Health and Nutrition Examination Survey (NHANES) conducted in 1999–2004, we analyzed the association between maternal age at childbirth and ADHD and LD risks in offspring. Odd ratios (ORs) and 95% condence intervals (CIs) for maternal age at childbirth in association with ADHD and LD risks in offspring were estimated using multivariate logistic regression models after adjustment for age, gender, race, body mass index (BMI), family income, smoking during pregnancy, and NHANES cycles. Restricted cubic spline (RCS) models were used to evaluate potential nonlinear relationships. Sensitivity analysis were performed to ensure the reliability of the results. Results: Among all participants, the offspring of subjects with a maternal age at childbirth of 18–24 years had an increased risk of ADHD (OR=1.34, 95% CI: 1.01, 1.79), LD (OR=1.36, 95% CI: 1.06, 1.79) or both ADHD and LD (OR=1.48, 95% CI: 1.20, 1.81). Additionally, compared with subjects with a maternal age of 25–29 years, subjects with a maternal age of 35–39 years had lower odds of having offspring with ADHD (OR=0.60, 95% CI: 0.36, 1.00) and higher odds of having offspring with LD (OR=1.34, 95% CI: 1.01, 1.78). The relationship between maternal age at childbirth and LD risk presents a U-shaped curve. Additionally, the risk of ADHD and LD caused by maternal age at childbirth appears to be affected by gender. Conclusions: These results provide epidemiological evidence showing that young and advanced maternal age at childbirth are associated with ADHD and LD risks.


Introduction
Maternal age at childbirth is considered as the age of the mother at the time of delivery. Generally, young maternal age at childbirth is de ned as being 19 years old or younger, and advanced maternal age is de ned as being at least 35 years old [1]. The effects of young and advanced maternal age at childbirth may affect the health outcomes of mothers and offspring. Previous studies have suggested that young maternal age at childbirth is associated with maternal cardiovascular disease risk [2] and low birth weight, preterm birth, and neonatal mortality [3]. Several adverse health effects are associated with advanced maternal age at childbirth, for example, cerebral palsy, neurocognitive disorders, psychiatric disorders [4], and stillbirth risk [5][6][7] in offspring and maternal cancer risk [8,9]; however, some studies have observed that advanced maternal age did not seem to be associated with the long-term morbidity of the offspring [10].
Neurological disorders in children manifest as abnormalities in neurocognition, self-regulation, and adaptive functioning [11]. Abnormal neurobehavioral function in children can cause health problems and other issues, such as emotional disorders [12], lower education levels [13], anti-social behavior [14], and premature mortality [15]. Among these disorders, attention-de cit/hyperactivity disorder (ADHD) and learning disability (LD) are common neurological disorders in children. ADHD and LD are de ned as persistent inattention and hyperactivity and as a reduced ability to learn, respectively. The global prevalence of ADHD in children is approximately 5% [16], and the prevalence was reported as 18.1% in Tunisian adolescents [17], 8.8% in Nigerian residents [18], and 6.26% in Chinese children and teenagers [19]. The prevalence rates of LD in the U.S. and in a city in India were found to be 9.7% [20] and 3.08% [21], respectively. Previous studies on the risk factors for ADHD and LD mainly focused on genetic factors [15,[22][23][24], while approximately 10-40% of the risk may be related to environmental factors, such as premature birth and smoking during pregnancy for ADHD [25,26] and low birth weight [27], maternal BMI before pregnancy [28], and maternal anemia [29] for LD. Although many previous studies have suggested that ADHD is highly related to hereditary factors, no gene with a de nite pathogenic effect has been found [30]. Although Thapar et al found that both prenatal and perinatal factors may be risk factors of ADHD and LD [15], these risk factors are not fully understood. Therefore, fully understanding these risk factors may effectively prevent the occurrence of ADHD and LD.
A recent study found that young maternal age at childbirth is positively correlated with the risk of ADHD in offspring in Denmark [31]. However, there is a lack of relevant research in the U.S. Because the U.S. has immigrants from multiple countries, the genetic background and cultural differences of the population are different from those of Denmark. Additionally, the prevalence of ADHD is much higher in the U.S. than the average level globally [32]. The average maternal age at rst childbirth increased both from 1970 to 2006 and from 2011 and 2012 in the U.S. [33]. This study was performed due to the lack of U.S. research on the association between maternal age at childbirth and ADHD and LD risk for offspring. This study is the rst to explore the association between maternal age at childbirth and ADHD and LD in offspring and to evaluate its nonlinear relationship in the U.S.

Study population
The National Health and Nutrition Examination Survey (NHANES) is conducted by the National Center for Health Statistics (NCHS) at the Centers for Disease Control and Prevention every two years, and the main purpose is to assess the health and nutritional status of children and adults in the U.S.. The user data agreement is available online (https://www.cdc.gov/nchs/data_access/restrictions.htm). The NCHS Research Ethics Review Board approved these NHANES cycles. All subjects provided written informed consent. Data on ADHD and LD outcomes are available only in 1999-2004; thus, we obtained publicly available NHANES data generated through surveys conducted in 1999-2004, including data from a total of 31,126 participants. By using unique survey participant identi ers, we combine information on their characteristics with the questionnaire information. For maternal age at childbirth, data are available only for offspring aged 0 to 15 years, and the age range for ADHD and LD evaluation were 4-19 years and 4-15 years, respectively. We therefore excluded participants under age 4 and over age 15. Although pregnancy may affect the outcomes of analysis, we did not exclude pregnant subjects (n = 7) in our analysis. Finally, 8,098 subjects were included in our study: 3,977 boys and 4,121 girls. The ow chart for inclusion and exclusion is listed in Fig. 1.

Assessment of maternal age at childbirth and ADHD and LD
This information was based on parental/guardian responses to the following NHANES interview questions, asked twice for both LD and ADHD (for children ≤ 15 years of age): "How old was your biological mother when you were born?". Age was recorded as a continuous variable. The outcomes of ADHD and LD were obtained from parental reports of LD and/or ADHD. Speci cally, the data were based on parental/guardian responses to two NHANES interview questions: "Has a doctor or health professional ever told you that you had attention de cit disorder?" and "Has a representative from a school or a health professional ever told you that you had a learning disability?" For children ≥ 4 years old, the same questions were asked for both LD and ADHD. The answer to both of these questions was "yes" or "no". The answer was included as a binary variable for subsequent analysis. We divided children into those with parental/guardian reports of ADHD, those with LD, and those with LD and ADHD.

Covariates
Several factors related to outcomes were adjusted in our logistic regression analysis, including age (continuous variable), gender (categorical variable), race (categorical variable), body mass index (BMI, continuous variable), family income (continuous variable), and NHANES cycle (categorical variable). Information regarding age, gender, race, and family income was obtained by questionnaires in face-toface interviews. Height and body weight were recorded in a physical examination and were used to calculate the BMI. Information related to smoking during pregnancy was based on parental/guardian responses to a questionnaire. The following question was asked: "Did your biological mother smoke at any time while she was pregnant?".

Statistical method
Continuous variables are presented as the mean and standard deviation, and categorical variables are presented as the frequency and percentage. Because the variables BMI and family income were missing for 429 and 686 participants, respectively, in order to ensure that the sample size was not reduced, we used the mean value imputation method. We divided maternal age at childbirth into the following groups: ≤17 years, 18-24 years, 25-29 years, 30-34 years, 35-39 pairs, and ≥ 40 years. Considering that both young and advanced maternal age at childbirth may be risk factors for ADHD or LD, we regarded 25-29 years as the reference age [34]. Logistic regression was used to assess the association between maternal age at childbirth and ADHD, LD and ADHD + LD after adjusting for age, gender, race, BMI, family income, and NHANES cycle. We also used strati cation analysis to explore the differences in this association between different genders and races. In addition, we used restricted cubic spline (RCS) regression to investigate the nonlinear trend between maternal age at childbirth and the risk of ADHD, LD and ADHD + LD. For the sensitivity analysis, in the selection of covariates of logistic regression, we excluded BMI and NHANES cycle. P < 0.05 was considered a statistically signi cant difference. In multiple comparisons, to avoid false positives, the Bonferroni correction was used in our analysis, with P values less than (0.05/5 [group] = 0.01) considered statistically signi cant.

Results
The average maternal age at childbirth in our study was 25.6 ± 6.1 years old (Table 1). Maternal age at childbirth differs among races; this age was lowest in the non-Hispanic Black population (24.5 ± 6.1 years old) and highest in the non-Hispanic White population (27.3 ± 5.8 years old). Subjects with high family income had a higher maternal age at childbirth on average than those with low family income (27.4 ± 5.8 vs 24.0 ± 6.0, P < 0.001). In addition, maternal age at childbirth did not differ signi cantly on the basis of smoking during pregnancy (25.6 ± 6.0 vs 25.6 ± 6.1, P = 0.961).  Table 2 suggests that the percentages of offspring with ADHD, LD and ADHD + LD in our study are 6.4%, 10.4% and 13.5%, respectively. The maternal age at childbirth for offspring who had ADHD, LD, or ADHD + LD was younger than that for offspring who did not have ADHD, LD, or ADHD + LD (25.2 vs 26.7, P = 0.001 for ADHD; 25.7 vs 26.6, P < 0.001 for LD; 25.5 vs 26.7, P < 0.001 for ADHD + LD). ADHD, attention-de cit/hyperactivity disorder; LD, learning disability.
The logistic regression results showed that compared to subjects aged 25-29 years, the risk of ADHD, LD, and both ADHD and LD in offspring was increased in subjects younger than 25 years old and older than 39 years old at childbirth (Table 3) after adjusting for age, gender, race, BMI, family income, smoking during pregnancy, and NHANES cycle. Among these groups, the adjusted OR for subjects aged 18-24 years was 1.48 (95% CI: 1.20, 1.81) for the risk of both ADHD and LD in offspring, with a P value less than 0.01 (Bonferroni adjustment of the P value). In addition, we also observed that compared to subjects aged 25-29 years at childbirth, the risk of ADHD was reduced, while the LD risk and the risk of both ADHD and LD in offspring was increased among people aged 30-39 years. However, these results were not statistically signi cant, which may be related to our small sample size. The sensitivity analysis showed that the results are robust (Table S1). * the P value was less than the Bonferroni adjustment (0.010).
Our analysis was further subjected to strati cation by gender and race. We found that subjects aged 35-39 years had a higher risk of LD in female offspring than did subjects aged 25-29 years (OR = 1.67; 95% CI 1.14, 2.45, P = 0.009), and the risk of both ADHD and LD for boys was higher for subjects aged 18-24 years than for those aged 25-29 years at childbirth (OR = 1.60; 95% CI 1.26, 2.04, P < 0.001) (Fig. 2).
When maternal age at childbirth was considered as a continuous variable using RCS regression models (Fig. 3), a U-shaped association of maternal age at childbirth with the risk of ADHD + LD was observed (P value = 0.007). The risk of ADHD + LD was lowest among subjects aged 25-29 years. When performing RCS regression models of the LD risk outcome, we found a similar but more pronounced U-shaped curve with a greater slope (P value = 0.005). However, the nonlinear association between maternal age at childbirth and ADHD was not signi cant (P value = 0.842).

Discussion
This was one of the rst studies nding that young maternal age at childbirth in the U.S. population may increase the risk of ADHD and LD in offspring and that advanced maternal age at childbirth seems likely to increase the risk of LD in offspring. In addition, we also found that these associations differ by gender. Furthermore, the relationship between maternal age at childbirth and the risk of LD in offspring presents a U-shaped curve.
Previous epidemiological studies have shown an increased risk of ADHD in the offspring of mothers with young maternal age at childbirth in the Swedish population [35]; similarly, the risk of ADHD in offspring is also increased for mothers with young maternal age at childbirth in the Danish population [31]. The results of the above two studies are similar to ours and may be explained because the development of ADHD is related to neurotransmission pathways. Therefore, we speculated that the education levels of the subjects with advanced maternal age at childbirth were higher, and these individuals may have greater health knowledge during pregnancy. Therefore, compared to the 25-29-year-old group, participants with advanced maternal age may pay more attention to nutritional supplements and additives during pregnancy, such as folic acid, which are bene cial for neural development, thus reducing the risk of ADHD in their offspring. Other potential reasons are that there are complex ethnic backgrounds in the U.S.; the genetic backgrounds of different races are quite different, and genetic factors are considered to be related to the development of ADHD [36,37]. In contrast to the results of the present study, a study of the Danish population revealed that advanced maternal age at childbirth (over 35 years old) is a protective factor against ADHD in offspring. While our results showed that the risk of ADHD in offspring of individuals older than 39 is likely to be increased, the difference was not signi cant, which may be related to the small sample size of our study. Another reason for the difference in study results is that the Danish study did not group and analyze the data of people over 40 years old.
At present, few studies have been conducted on maternal age at childbirth and LD risk in offspring, which may be related to the diagnosis of LD without International Classi cation of Disease (ICD) standards. Although LD evaluation is subjective, its evaluation is easy to perform and can re ect certain neural functions [38,39]. At present, the etiology of LD is still unclear. Some scholars have suggested that genetic factors may play a role, and the symptoms of LD are also broader. Adolescents who exhibit abnormalities in reading, writing, listening, speaking, and performing math in will be suggested to have LD. However, the cause of LD in offspring is complex and may not be related to a neural pathway. In this study, we observed a U-shaped curve of maternal age at childbirth and offspring LD risk. It is suggested that the individuals with younger and more advanced maternal ages at childbirth will have a greater risk of having offspring with LD. At present, most of the existing studies suggest that young maternal age at childbirth increases the risk of health effects on offspring. However, controversy exists for subjects with advanced maternal age at childbirth. Appropriate older age of mother is bene cial for the offspring with regard to nancial status and education [1]. The social economic position and resources of older mothers are greater than those of younger mothers, and over time, the offspring of older mothers experience the effects of improvements in public health conditions and educational expansion. Although negative effects occur during the perinatal period, there is no difference between these offspring and offspring of mothers with a younger age at childbirth during adulthood [40]. However, it should be noted that the oocytes of older mothers are aging, and accumulated DNA damage [41] and the imbalance in mitochondrial homeostasis [42] in aging oocytes have a negative impact on the development of fertilized eggs [43]. DNA damage and mitochondrial function are closely related to nerve development [44,45]. These characteristics of aging oocytes may explain the decreased learning ability of offspring. However, some researchers suggest that some patients are classi ed as having LD because of their parents' misunderstanding of the education process and the parents have higher expectations for their offspring [46]. The intelligence or neurodevelopment of offspring may not be abnormal, but the speci c molecular mechanism of the relationship between age and LD still requires further research.
Our results indicate that maternal age at childbirth is related to the risk of offspring LD. Due to the recall bias in retrospective cohorts, animal studies may clarify the cause and effect relationship. Some existing mechanistic studies have con rmed that the spatial learning capacity of mice at 32-35 weeks is lower than that of mice at 9-12 weeks, which may be related to the decrease in the expression of vitamin D receptor (VDR) in the early embryogenesis process [47]. In addition, compared with 3-month-old mice, 15-18-month-old mice showed anxiety-like behavior, and the gene expression pattern of the hippocampus also changed [48]. We also speculated that late production age may be related to the egg quality [49].
These possible mechanisms may explain the link between maternal age at childbirth and ADHD and LD risks in offspring. However, there are few studies on the mechanism of the relationship between young maternal age at childbirth and ADHD and LD risk, possibly because few animal models exist. We speculated that the link between young maternal age at childbirth and ADHD and LD risk may arise because the family income and education levels of young mothers are lower than those of older mothers [50]. Additionally, Fall et al found that young maternal age at childbirth may be associated with poor birth outcomes and nutrition for offspring in low-and middle-income countries [1].
Our results also showed a sex difference in the association of young maternal age at childbirth and the risk of ADHD + LD among male and female offspring. The reason for this phenomenon is that the ADHD incidence rate of boys is higher than that girls [15,30]. Additionally, sex differences in gonadal steroid hormone levels during perinatal development result in gender differences in neurodevelopment. Furthermore, compared to early-adolescent and middle-aged mice, male offspring have less developed sexual organs [51], which suggests that androgen may also be low in early-adolescent mice, and changes in hormones in adolescence are related to brain volume and neurodevelopment [52]. Therefore, we speculate that hormonal disorders in the male offspring of young age mothers may cause an neurodevelopmental abnormality that leads to the increased risk of both ADHD and LD in boys rather than girls.
As an advantage, this study is the rst to report the association between maternal age at childbirth and offspring ADHD and LD risks in the U.S. population. We found that young or advanced maternal age at childbirth may lead to ADHD and LD risks in offspring. Through the results of subgroup analysis, we identi ed the susceptible population, which may provide a basis for research on the underlying mechanism or prevention. Applying the RCS model is helpful to determine the association between maternal age at childbirth and the ADHD and LD risks of offspring.
Although the present study results have some scienti c signi cance, there are still several limitations of our study as a whole. First, genetic factors have a great in uence on ADHD and LD risks. Even if we consider race as a covariate for adjustment and further analysis, we still cannot exclude the bias of genetic factors. Second, some maternal lifestyles during pregnancy also affect offspring ADHD and LD, although we adjusted for the variable of smoking during pregnancy. In fact, there are additional factors, such as gestational hypertension and gestational diabetes, but we cannot adjust for them because no such data are contained in the NHANES database. Third, in this study, some covariates were missing. To avoid reducing the sample size, we imputed data for the missing variables, but strictly speaking, this approach may have biased the actual results. Fourth, the outcome variables in this study based on selfreported data, leading to potential subjective bias. It is thus necessary for future research to use objective indicators.
In conclusion, the present study results indicated a U-shaped association between maternal age at childbirth and the ADHD and LD risk of offspring in a nationally representative U.S. survey. Future studies with large sample sizes and mechanistic studies are needed to con rm the risks of ADHD and LD.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The datasets generated and/or analyzed during the current study are publicly available from the Centers for Disease Control and Prevention (https://wwwn.cdc.gov/nchs/nhanes/Default.aspx).

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by funding from the National Science Foundation of China (81903409).
Authors' contributions YC, WWD and HBH designed and parametrized the mathematical model, conducted the analyses, and wrote the rst draft of the article. SYM, and YC supported the model parameterization, conducted the statistical analyses, and participated in the drafting of the article. SYM and HBH conceived and led the design of the study and model and performed the analyses and drafting of the article. XMM revised the manuscript. All authors have read and approved the nal article.