Trajectories of Maternal Depressive Symptoms From Pregnancy to First Six Months Postpartum and Child Neurodevelopment at Eight Months

Maternal depression was an increasing risk of child neurodevelopment diculties. However, maternal depression was always chronic and recurrent, and few studies have explored the association between the severity and duration of maternal depression and child development. We aimed to explore the relationship between trajectories of maternal depressive symptoms from pregnancy to the rst six months postpartum and child development at eight months. Methods: Data was collected from 988 mother-child pairs who participated in Shenzhen Birth Cohort Study. Maternal depressive symptoms were evaluated by the Edinburgh Postnatal Depression Scale (EPDS) at four points: late pregnancy, 1 month, 3 months, and 6 months postpartum. Child emotional and behavioral developments were assessed by Ages and Stages Questionnaires: Social-Emotional (ASQ-SE) and Ages and Stages Questionnaires-Third Edition (ASQ-3) at 8 months postpartum. Latent prole class analysis (LPA) was used to identify the trajectories of maternal depressive symptoms. Univariate and multivariate linear regression were conducted to explore the association between the depressive symptoms trajectories and child development. Four trajectories of maternal depressive symptoms were identied subclinical(n=93), and in ASQ-3 total, a “depressed/not depressed” approach to screening for depression in the perinatal period. score and 3 ASQ-3 domains (ne motor, problem-solving, and personal-society), even after adjusting for confounding. Furthermore, the nal multivariable model showed that children of mothers assigned to class2 were more likely to have lower scores in ASQ-3 total, 2 ASQ-3 domains (communication motor and gross motor), and ASQ-se, compared with children of class1, after adjusting for covariates. Risk factors identied in the multi-model included low birth weight, sleep quality at late pregnancy, low education, and breastfeeding at 3month. maternal passive smoking, preterm, gestational hypertension, fetus gender, subclinical hypothyroidism, or hypothyroidism, breastfeeding at 3months, low birth weight, sleep quality at late-pregnancy and residence type. this study provides novel evidence of the negative impact of persistently high and subclinical depressive symptoms on delayed child development at eight months. symptoms, increase and persistent ant low-level symptoms, subclinical depressive symptoms, and persistent high symptoms. Even after accounting for demographic, obstetric, birth outcomes, and feeding factors, the association between depression symptom trajectories and child emotional and behavioral development remained signicant. Especially sub-clinical levels of depressive symptoms are associated with adverse emotional-behavioral outcomes for children. These ndings draw attention to the need to move beyond a “depressed/not depressed” approach to screening for depression in the perinatal period. And the ndings of the present study have important implications for policy and practice efforts focused on improving maternal and child health in the perinatal and early childhood period.

from routine medical records. For cohort children, Mothers are interviewed and children are examined in child health care clinics at age 1, 3, 6, 12 months, or by telephone interview for those mothers who are unable to bring their child to the pediatric clinic. The study plan to follow up all cohort children annually until age 18 years. Procedures and methods were approved by ethics committees at Nanshan Maternity & Child Healthcare Hospital of Shenzhen (NSFYEC-KY-2020031), and informed consent was obtained from all participants.
This study included 1387 pregnant women enrolled in SZBCS who completed the prenatal questionnaires in the rst, second, and third trimesters respectively. 1299 pregnant women agreed to participate after delivery. Of the 1299 mother-child pairs, 28 were excluded because they delivered twins, and one child because of congenital heart disease. Of the remaining 1271, 988 mother-child pairs nished the questionnaires at child one month, 3months, 6months, and 8months, respectively. The main reasons for dropout were personal reasons, such as moving away, lack of interest, lack of time, except part of the child have not yet reached the survey months.

Depression symptoms
Depressive symptoms were accessed by the Edinburgh Postnatal Depression Scale (EPDS) [20]. The EPDS is a 10-item self-report instrument designed to measure the last 7days emotional expression during the postpartum period. Responses are scored from 0-3, total scores with a range of 0-30. A higher score suggested severer depression symptoms. Generally, a score of greater than 12 was recommended for identifying as depression. Good reliability and validity were demonstrated in the Chinese translation of EPDS, with internal consistency (Cronbach's a) of 0.87 [21]. The effectiveness of the scale in prenatal and postnatal has been recognized [22,23]. In this study, the EPDS was asked to complete for mothers at 4-time points: late pregnancy, 1month, 3month, and 6month postpartum.

Emotional and behavioral development
Child behavioral development was measured by the Ages and Stages Questionnaires-Third Edition (ASQ-3) [24] and social-emotional development was measured by the Ages and Stages Questionnaires: Social-Emotional (ASQ-SE) at ages 8 months. The ASQ-3/ASQ-SE is a parent-reported screening tool to capture the risk of developmental delays. ASQ-3 consists of 30 items, which were divided into 5 domains, namely communication, gross, ne, problemsolving, and personal-social. Each domain contains 6 items, and possible responses for each item were "yes" (10 points), "sometimes" (5 points), and "no" (0points). A total score of each domain is calculated as a sum of the 6 items ranging from 0 to 60. The higher the score, the better the skills in the given domain. The ASQ-3 is a reliable and valid instrument to screen and monitor the development of Children in China, and the Cronbach's alpha coe cient was 0.8 [25].
ASQ-SE consists of 8 versions depending on the different age groups, and the number of items ranges from 19 to 33. There are three kinds of the answer: rarely or never, sometimes, most of the times, and 0, 5, and 10 were scored respectively. Additional 5 points were added If worrying about the child's performance of each item. The higher the total score, the worse social-emotional development. The internal consistency of the Chinese version's ASQ-se was measured by Cronbach's a, which ranged from 0.56-0.77 across various age intervals, and the item reliability ranged from 94-96% [26].

Covariates
Information on Socio-demographic, obstetric conditions, and child feeding were collected by the follow-up questionnaires during pregnancy and postpartum.

Data statistics
Statistical analysis was conducted using SPSS (23.0) and Mplus (7th ed) [28], and the P-value of less than 0.05 was considered to be statistically signi cant. Continuous variables were described as means (X ± S), and categorical variables as proportions(n%). Longitudinal pro le class (LPA) was used to identify trajectories of women's depressive symptoms across four-time points (late pregnancy, 1, 3, and 6 months). To t successive models, the number of classes was increased from 1, until the most parsimonious models with the smallest number of classes were identi ed. Model solutions were evaluated by comparing Likelihood ratio statistic (L2), Bayesian Information Criterion (BIC), and Akaike Information Criterion (AIC) across the successive models, Entropy is an index for assessing the precision of assigning latent class membership, with higher probability values indicating greater precisions of classi cation. The Vuong-Lo-Mendall-Rubin likelihood ratio test was also used to test for signi cant differences between the models [19]. The class membership of the most parsimonious model was saved and used in subsequent analysis. Then a bivariate linear regression analysis was used to measure the association between trajectories of maternal depression symptoms and child development (5 ASQ-3 domains, total scores, and social-emotional scores) at 8 months.
A multivariable linear regression model was used by adjusting confounders, including maternal age, pre-pregnancy BMI, maternal education level, household residence type, marital status, parity, sleep quality at 3rd trimester, maternal passive smoking, gestational diabetes mellitus, subclinical hypothyroidism, or hypothyroidism, gestational hypertension, breastfeeding at 3 month, LBW and PTB. Regression results were presented as adjusted b and 95% con dence intervals.

Results
Page 4/12 3.1 Sample characteristics A total of 988 mother-child pairs met all the study criteria and were included in this study nally. Table 1 describes the characteristics of the sample included in this study. Most of these women were between 25 ~ 34 years of age at recruitment, were married, had a tertiary-level education, and had a temporary residence in Shenzhen in China. 526 (51%) participants were primigravida. The proportion of women who reported EPDS scores ≥ 13 (indicative of a probable diagnosis of major depression) in late pregnancy and at 1, 3, 6 months postpartum was 6.0, 7.8, 12.9, and 12.0%, respectively.

Trajectories of maternal depressive symptoms
The nal model with four trajectories of maternal depressive symptoms, at four-time points (late pregnancy, 1, 3, and 6 months postpartum) were accepted, as its t indexes(entropy = 0.822) were higher than the 2-and 3-class models (supplemental table). Additionally, the entropy value for the 4-class model was highest(entropy = 0.822), suggesting the best precision in assigning the individual to their classes. The 3-class model was not chosen, despite a signi cant result from the Vuong-Lo-Mendell-Rubin likelihood ratio test and LMR due to its lower entropy than the 4-class model [29]. Figure 1 shows the 4-class trajectories of maternal depression symptoms. The rst trajectory characterized by persistent low levels of depressive symptoms, consisted of 60.1% (n = 605) of mothers. The second trajectory presented a pattern of subclinical depressive symptoms with a high depressive symptom score at pregnancy, and had a decrease at postpartum, comprising 9.2%(n = 93) of mothers. The third trajectory consists of 25.3%(n = 255) mothers, who presented a pattern of moderate depressive symptoms during pregnancy with increasing symptoms in postpartum. The fourth group included 5.4%(n = 54) of the mothers, who were characterized by persistently high levels of depressive symptoms. Therefore, the 1-to 4-class trajectories of depressive symptoms were named orderly as "low depressive", "subclinical depressive", "moderate-low and increasing" and "persistent high depressive".

Associations between trajectories of maternal depressive symptom and child emotional and behavioral development
The associations between maternal depression symptom trajectories and 5 ASQ-3 domains (communication motor, gross motor, ne motor, problem-solving motor, and personal-society motor), ASQ-3 total score, and ASQ-se score are presented in Table2-4. Covariates were adjusted in multivariable analysis, including maternal age, pre-pregnancy BMI, maternal education, sleep quality at late-pregnancy, passive smoking antenatal, parity, GDM, subclinical hypothyroidism or hypothyroidism, low birth weight, preterm birth, and breastfeeding at 3month. At eight months, the ASQ-3 total score of children of persistent high depressive symptom group had an increased risk of low score of ASQ-3 than children of mother in other groups, as were their ASQ-se score and 3 ASQ-3 domains ( ne motor, problem-solving, and personal-society), even after adjusting for confounding. Furthermore, the nal multivariable model showed that children of mothers assigned to class2 were more likely to have lower scores in ASQ-3 total, 2 ASQ-3 domains (communication motor and gross motor), and ASQ-se, compared with children of class1, after adjusting for covariates. Risk factors identi ed in the multi-model included low birth weight, sleep quality at late pregnancy, low education, and breastfeeding at 3month.

Discussion
The present study identi ed four groups of maternal depressive symptoms trajectories, from late pregnancy to 6months after birth: "low depressive", "subclinical depressive", "moderate-low and increasing" and "persistent high depressive". The in uence of severity and chronicity of maternal depressive symptoms on child development has been examined. We found that mothers experiencing persistent high symptoms are more likely to report that their children have a lower score on the ASQ-3 total, ASQ-se, and 3 ASQ-3 domains ( ne, problem-solving, and personal-society motor). Similarly, children of mothers assigned to subclinical depressive symptoms were more likely to have lower scores in ASQ-3 total, 2 ASQ-3 domains (communication motor and gross motor), and ASQ-se, compared with children of low depressive symptoms. This study investigated the effect of maternal depression symptoms over time and found that "subclinical symptoms" and "persistent high symptoms" had an increased risk for poor emotional and behavioral development. As such, this study provides novel evidence of the negative impact of persistently high and subclinical depressive symptoms on delayed child development at eight months.
Some previous research supports our view. An association between maternal depressive symptoms at prenatal or postnatal, and child poor neurodevelopment was found [10,11]. Kingston et al. explored the depression symptoms across four points from mid-pregnancy to one year postpartum and identi ed four distinct trajectories of maternal depressive symptoms: low level (64.7%), early postpartum (10.9%), subclinical (18.8%), and persistent high (5.6%). In multivariable models, the children born to women with persistent high depressive symptoms had the highest proportion of elevated behavior symptoms, followed by mothers with moderate symptoms (early postpartum and subclinical trajectories) and lowest for minimal symptoms [30]. Another study(N = 1983) also supporting mothers experiencing high anxiety symptoms in the perinatal period have an adverse impact on child developmental delays.
They identi ed three distinct trajectories of maternal stress and anxiety symptoms from mid-pregnancy to 3 years postpartum. Multivariate analysis showed mothers belonging to the high anxiety symptoms class had an increased risk (adjusted OR 2.80, 95% CI 2.80 (1.42 ─ 5.51), p = 0.003) of having a child with developmental delays at 3 years [31]. But their explorations are limited to one aspect of child development, caused by maternal depression. As a result, they are unable to capture the simultaneous in uence of maternal depression on multiple areas of child development, such as communication, motor, problemsolving and social-emotional development. It is important to note that, maternal with high depressive symptoms at prenatal had a signi cant effect on poor emotional and behavioral development compared with increased and persistent even at low-level symptoms. Researchers found that children of mothers who had elevated depressive symptoms during pregnancy had worse mental health in adolescence [32,33]. The effect of prenatal maternal depressive symptoms on child outcomes is noteworthy. Further studies are warranted to explore how the course of prenatal and postnatal depression affects child development.
In the nal multivariate model, we explored the relationships between maternal depression symptom trajectories and children's emotional and behavioral development after accounting for demographic, obstetrics, birth outcomes, and feeding factors. sleep quality at late pregnancy, low birth weight, breastfeeding at 3 month and lower education level of mothers were also associated with poorer emotional and behavioral for development children. Some of the characteristics might be pathways by which maternal mental health di culties contribute to children's emotional and behavioral di culties. For instance, depression symptom might lead to chronic sleep loss and low birth weight. The former might contribute to adverse pregnancy outcomes and maternal and child stress 'overload', and the latter might increase the gross and ne motors risk of a diagnosis of neurodevelopment delay [34]. Depression symptom can also affect children by affecting mothers' behavior. Studies showed positive associations between breastfeeding and child neurodevelopment [35,36].
Our ndings have implications for physicians who provide care during pregnancy. Clinically, if we only considered women's scores using the recommended screening cut-off of 13 or greater on the EPDS, those assigned to the subclinical depressive symptom trajectory may not be identi ed as their mean scores are below the clinical cut-off points. However, the proportion of subclinical women is high, and the signi cant effect on the child neurodevelopment, this raises questions for current health care that rely on cut-offs to evaluate women's mental health requiring intervention or follow up, and provide evidence of early identi cation and support.
This study bene ts from using longitudinal statistical modeling of data drawn from a prospective birth cohort study. However, there are several limitations to note. First, The EPDS test was self-written by the mother, which might lead self-report bias. Second, the EPDS was only assessed at late pregnancy for all women, not at rst and second trimester, preventing classifying the trajectory of depressive symptoms during pregnancy by trimester. Third, we couldn't include all covariates associated with children's outcomes n the study. Finally, we supposed data were missing at random in the present analysis, this may result in an underestimate of maternal depressive symptoms as more depressed women may be more likely to miss a study assessment.

Conclusion
Women's depressive symptoms from late pregnancy through postpartum at six months followed four distinct trajectories de ned by low-level symptoms, increase and persistent ant low-level symptoms, subclinical depressive symptoms, and persistent high symptoms. Even after accounting for demographic, obstetric, birth outcomes, and feeding factors, the association between depression symptom trajectories and child emotional and behavioral development remained signi cant. Especially sub-clinical levels of depressive symptoms are associated with adverse emotional-behavioral outcomes for children. These ndings draw attention to the need to move beyond a "depressed/not depressed" approach to screening for depression in the perinatal period. And the ndings of the present study have important implications for policy and practice efforts focused on improving maternal and child health in the perinatal and early childhood period.

Declarations
Ethics approval and consent to participate Procedures and methods were approved by ethics committees at Nanshan Maternity & Child Healthcare Hospital of Shenzhen (NSFYEC-KY-2020031), and informed consent was obtained from all participants or their legal guardians.

Methods
We con rm that all methods were performed in accordance with the relevant guidelines and regulations.

Consent for publication
Not applicable.
Availability of data and materials