Early stimulation and responsive care: a mediator of caregivers’ depression on the suspected developmental delay of left-behind children in China

Purpose: Due to multiple factors, children left behind in rural areas suffer from delayed neurodevelopment (suspected developmental delay, SDD). This study aimed to analyze the effect of caregivers’ depression on left-behind children’s SDD, with early stimulation and responsive care mediating. Methods: A cross-sectional survey was conducted in villages of �ve Chinese counties. A total of 904 left-behind children and their primary family caregivers were enrolled. Information on the children and their caregivers were collected through face-to-face interviews. The Zung Self-rating Depression Scale (ZSDS) was used to measure caregivers’ depressive symptoms. The Ages and Stages questionnaires, third edition (ASQ-3) was used to screen children for SDD, which contains �ve domains: communication (CM), gross motor (GM), �ne motor (FM), problem-solving (CG), and personal social (PS). Results: Of left-behind children’s caregivers, 39.7% experienced varying symptoms of depression. The prevalence of total SDD among left-behind children under three years was 31.4%. Caregivers’ ZSDS scores were positively correlated with the FM, GM, CG, and PS of the left-behind children. Caregivers’ early stimulation & responsive care was positively correlated with the CM, FM, CG, and PS of the left-behind children. Conclusion: Left-behind children under three years in rural China were at high risk of SDD and their caregivers’ mental health was not good. Caregivers’ depressive symptoms may negatively affect the suspected developmental delay of left-behind children through caregivers providing less early stimulation and responsive care. A comprehensive intervention should be conducted to promote the health of left-behind children and their caregivers.


Introduction
With economic development and social change, an increasing number of laborers migrate from lower or middle-income regions for employment opportunities in China. When parents migrate, they commonly leave their children behind in their rural hometowns, which resulted in a growing number of children being left behind in the care of other family members or caregivers [1]. In 2014, there were 40.51 million Chinese left-behind children in the countryside, accounting for 29.4% of the national rural children and 15.0% of the national children [2].
Early childhood, especially the rst 3 years of life is a critical developmental stage for an individual's cognitive development, social and emotional functions or disorders throughout their life span [3].
Evidence showed that left-behind children's physical, psychological, and social development at an early age were facing a great challenge because of China's tidal wave of migrants [2] [4]. According to socialecological systems theory, parenting behavior from parents and connections between family members will have a signi cant impact on the early development of children [5]. Because of parenting migration, the family structure and function changed, and the parenting behavior of left-behind children was taken over by caregivers of other family members such as grandparents. When the infant is deprived of a wellstructured family environment or a transition of parenting behavior, it may constitute a risk factor for the child's early development.
Parenting behavior can be in uenced by many factors, one important factor is caregiver depression. Depression is common among caregivers in rural China. Based our previous cross-sectional study in rural areas of ve Chinese provinces reported that the prevalence of depressive symptoms among left-behind children's caregivers was 40.02% [6]. The results of another study showed that among female caregivers in China's Qinling Mountainous region, 18.9% of mothers and 34.7% of grandmothers were depressed [7].
A meta-analysis found that depressed caregivers display more negative interactions with their infants, toddlers, and young children which may produce long-term adverse effects on children's development [8].
Consistent ndings in developing countries have also indicated that young children of depressed mothers have lower levels of cognitive function [9] [3]. Our previous study on left-behind children found a correlation between caregivers' depressive symptoms and children's social-emotional development [6] [10].
The relationship between caregiver depression, its impact on parenting practices, and child outcomes is complex and remains to be fully elucidated. One focused point is the effect of maternal and paternal depression on parenting behavior, the parent-child relationship, and general family functioning [11].
Recent studies have linked the poor cognitive development of children in rural China to an absence of positive parenting practices [12] [7]. This is because early age-appropriate stimulation and responsive care bene t the brain architecture by forming, strengthening, and pruning synaptic connections [13].
Therefore, a positive parenting style in which caregivers give age-appropriate stimulation to the child contributes to the child's early development. A lack of early stimulation, on the other hand, can have a detrimental effect on the child.
Our previous study found that for left-behind children, their development may be adversely affected by interrupted parent-child relationships and weakened nurturing care due to parent-child separation [10] A multi-stage strati ed sampling method was applied in the rst four counties (Yudu County, Sansui County, Tongjiang County, and Lushi County). The villages in each county were classi ed into three layers according to the number of left-behind children in the village. In each layer, three villages were sampled at random. For Pingshan County, the sampling strategies applied in other villages were not appropriate due to its limited number of left-behind children, so all villages in Pingshan County were selected instead. At the same time, a matching control village from the same county was selected for each sampled village, according to the number of children under 3 years old, per capita income, and distance from the county hospital.
Overall, a total of 113 villages were then selected for the baseline survey. In the selected villages, all the left-behind children aged 0 to 3 years old and their caregivers were targeted. The ethical approval for the project was granted by the Ethics Committee of the Peking University Health Science Center.

Data collection
Information was collected through face-to-face interviews with primary caregivers by trained investigators. Sociodemographic information of children included their age, gender, low birth weight, and parental divorced/passed away. For primary caregivers, information was collected on their age, gender, ethnicity, educational level, and identity (mother/father, grandparent, others). The Household Economic Index was measured by the number of household appliances and vehicles, including televisions, rice cookers, washing machines, refrigerators, air conditioners, computers, motorcycles/tricycles, and automobiles.

Children's suspected developmental delay
The third edition of the Ages and Stages Questionnaires (ASQ-3), a validated tool to detect developmental delays in children aged 1 to 66 months, consists of a series of 21 age-appropriate questionnaires [14].
There are six questions in each of the ve developmental domains (communication, gross motor, ne motor, problem-solving, and personal-social). Caregivers scored the child according to how well he or she had the skill. A score of 0 indicates "Never or rarely", meaning that the child had not developed the skill; a score of 5 indicates "sometimes", meaning that the skill was developing; a score of 5 indicates "most of the time", meaning that the child demonstrates the skill regularly or most of the time [15]. Domain scores are used to classify children as "development on schedule" (scores above the cutoff score in all domains), or "at risk of developmental delay in the domain" (below the cutoff score in any domain), refer for further assessment. ASQ-3 has been empirically studied using over 15,000 children with proven robust sensitivity and speci city results [15].
Caregivers' depressive symptoms The Zung Self-Rating Depression Scale (ZSDS) [16], a widely used scale for assessing depressive symptoms, was utilized to evaluate the depressive symptoms of caregivers. The ZSDS consists of 20 self-report items that were identi ed in factor analytic studies of the syndrome of depression. Items explored psychological and physical symptoms: 10 expressed negative experiences such as "I feel depressed and blue", and 10 expressed positive experiences with a reverse rating such as "I eat as much as I used to". Participants rated each item as it applied to them during the past week ranging from 1 (none, or very little) to 4 (most, or all of the time).
The overall raw score ranges from 20 to 80. The crude score was multiplied by 1.25 to obtain the standard score [17]. A cut-off point of clinical signi cance using a standard score of 50 is recommended, meaning that a standard score of 50 or above is considered indicative of the depression symptom. The degree of depressive symptoms was divided into 3 levels: light (50-59 points), moderate (60-69 points), and severe (≥ 70 points) [18].

Early stimulation & responsive care
Based on the Multiple Indicator Cluster Surveys-version 6 (MICS-6), we formulated a questionnaire to assess child nurturing care by whether children are given early stimulation & responsive care, which was estimated by the diversity of caregiver-child interaction. In particular, caregiver-child interaction includes the following activities: [A] reading to the child or looking at picture books together; [B] telling the child stories; [C] singing to the child or together with him/her (including lullabies); [D] taking the child outside; [E] playing with the child; [F] recognizing things, counting or drawing with the child. If there have been no less than four types of interaction in the past three days, the caregiver could be considered to have provided early stimulation & responsive care to the child.

Statistical analysis
Continuous variables are described by mean and standard deviation, whereas categorical variables are described by frequency and composition ratio. Univariate logistic regression was carried out to estimate the association between each factor and the suspected developmental delay. A chi-square test was used to compare the prevalence of suspected developmental delay in children of different gender.
For further con rmation of the association between caregivers' depressive symptoms and suspected developmental delay of left-behind children, multivariate logistic regression was conducted based on the univariate logistic regression results. According to previous studies, sociodemographic variables with a pvalue below 0.2 (caregivers' gender, caregivers' age, caregivers' education, caregivers' type, number of vehicles and appliances, children's gender, children's age, low birth weight, parental divorce or passed away) were considered as potential confounders. Due to the low correlation coe cients between caregivers' depressive symptoms and the other independent variables, the issue of covariance could be disregarded for treatment. Multivariate logistic regression analysis was conducted with backward methods. Caregivers' depressive symptoms were categorized as a dichotomous variable according to the severity of the symptom (no or light; moderate or severe). Each suspected developmental delay was considered as the dependent variable, with the caregiver's depression score as the independent variable, while confounders were also included.
Based on the results of the regressions, a hypothetical mediation model was used to assess the factors that in uence SDD and the relationship between them. The ZSDS score was employed as a standard for assessing caregivers' depressive symptoms. The structural equation modeling (SEM) was used to estimate the direct and indirect effects of caregivers' depressive symptoms on SDD. We estimated the direct and indirect effects of the ZSDS score on SDD and then conducted separate moderating analyzes to analyze the moderating effect of early stimulation & responsive care. Caregivers' depressive symptoms were used as the predictor, early stimulation & responsive care as the mediator, and SDD outcomes as the dependent variable. The model was adjusted for caregivers' gender, age, type, and education; the number of vehicles and appliances; children's gender and age; low birth weight; and whether the parental divorce or passed away. The coe cients were standardized to allow for direct comparison.
Descriptive analyses and regressions were computed using SPSS 23, while structural equation modeling was calculated by Mplus 7.4.
For all two-tailed tests, P < 0.05 was considered statistically signi cant.

Characteristics of left-behind children and their caregivers
The social demographic characteristics of the participants are shown in Table 1. Children were generally balanced in terms of gender composition, consisting of 479 boys (53.0%) and 425 girls (47.0%). The median age of them was 20 months. Caregivers were mostly women (89.8%), with a median age of 50 years old. Children's grandparents were the major caregivers (64.7%), and 49.6% of caregivers had a middle school or higher education level.

Prevalence of suspected developmental delay among left-behind children
Among the 904 children aged 0 ~ 3 years, the prevalence of total SDD was 31.4% with the highest rates of SDD in problem-solving (13.5%), followed by ne motor (13.4%), communication (12.6%), and personal social (11.4%) (shown in Table 2). The prevalence of total SDD and the ve developmental domains (CM, GM, FM, CG, PS) are presented respectively and compared by gender. SDD was more commonly observed in boys (34.9%) than in girls (27.5%) (χ 2 = 5.623, P = 0.018). The prevalence of the developmental domain of FM and PS were also statistically signi cantly higher in boys than in girls (P < 0.001).

Risk and protective factors for suspected developmental delay in left-behind children
According to the results from univariate logistic regression (Table 1), The result after categorizing depressive symptoms showed that children whose caregivers had moderate or severe depressive symptoms were more likely with SDD comparing with children whose caregivers with no depressive symptoms (OR = 2.99, 95%CI: 1.88 ~ 4.76). Similarly, the ZSDS score was positively related to SDD in leftbehind children.

Mediation role of early stimulation & responsive care between caregivers' depressive symptoms and suspected developmental delay of left-behind children
The estimated coe cients of the structural equation model were illustrated in Fig. 1, and Table 4 and   In terms of direct effects, the positive association between caregivers' ZSDS scores and children's SDD was statistically signi cant (P = 0.003). For indirect effects, ZSDS scores were negatively associated with early stimulation & responsive care (-0.098, P = 0.016), while early stimulation & responsive care in turn was negatively associated with SDD (-0.301, P < 0.001). With an estimated coe cient of 0.128 (P = 0.003), the direct effect of the ZSDS score and SDD explained 81.01% of the total effect (0.128/0.158); while the indirect effect was 0.030 (P = 0.031), explaining 18.99% of the total effect.
Early stimulation & responsive care may mediate the association between caregivers' depressive symptoms and certain dimensions ( ne motor, problem solving, and personal social) of the SDD in leftbehind children.

Discussion
In this study, we explored the risk and protective factors of suspected developmental delay among leftbehind children under 3 years in rural China. We found that caregivers' depressive symptom was positively associated with SDD in the surveyed children, and the effect was mediated by early stimulation & responsive care.
Our study used ASQ, a primary screening measurement to re ects overall child neurodevelopment. The prevalence of overall SDD among the 904 left-behind children in our study was 31.4%. Up to now, research on the SDD of left-behind children under 3 years old in China was limited. The current research was mainly on rural children including left-behind and non-left-behind children. One study reported that 39.7% of children under 3 years old in poor areas of China had SDD problems [12]. Another study found that 37.6% of surveyed children aged 6-24 months had a signi cant delay of cognitive development in China's Qinling Mountains [7]. As we know, "the rst 1000 days" is thought to be a critical period for children to reach their developmental potential. Parental absence during early childhood may have a large detrimental on the developmental outcomes of the left-behind children in China.
Previous studies indicate that male gender, prematurity, low birth weight, hospital history, mother's education level, and lacking family support or stimulation (such as no books at home or no children's stories in a week) are risk factors for SDD [19] [20] [21]. Our study provided further con rmation that a child of the male gender or lack of early stimulation would be at higher risk of SDD. Meanwhile, caregivers' depressive symptoms, especially moderate or severe depressive symptoms, as a potential in uencing factor for SDD was founded, which was consistent with research in developing countries that young children of depressed mothers have lower levels of cognitive function [9] [3].
In rural China, the mental health among caregivers of children is not optimistic. According to Wei et al.,39.8% of caregivers in poor areas had depressive symptoms [12]. Another survey in the Qinling Mountains of China found that 23.5% of female caregivers had depressive symptoms [7]. In our study, 39.7% of the left-behind caregivers suffered from depressive symptoms, with 9.4% of them scoring within a moderate or higher severity of depression. The higher prevalence of depression among caregivers of children may be associated with lacking social support from family and friends, the burden of caregiving, within-family con icts, poverty, and prolonged separation of elderly people from their children, etc [7] [22].
Considering that rural-urban migration will be persistent in China, it is therefore imperative to understand the underlying features of depression among caregivers of left-behind children to better prevent and deal with depression in the future.
The depressive states of caregivers can have adverse effects on parenting behaviors [23] [8]. Children reach developmental potential can be in uenced by multiple factors, in which responsive caring and early learning are important domains [3]. When a young child lacks a well-structured home environment and appropriate early stimulation, the family may also be a risk factor for a child's development [24]. Our previous study has found that the transition of family structure and functioning, as well as consequently weakened home environment, resulted in social-emotional problems for left-behind children under 3 years [10]. This study reveals that in left-behind children, the association between caregivers' depressive symptoms and overall child neurodevelopment (SDD) is mediated by early stimulation and responsive care. The early stimulation and responsive care explained 18.99% of the relationship between caregivers' depressive symptoms and children's SDD, especially in three domains of ne motor, problem solving, and personal social development. The reactive stimulus was signi cantly associated with improved quality of mother-child interactions [25] Our study provides evidence on the prevalence of SDD among left-behind children under 3 years old in rural China and their risk factors. Some limitations should also be noted. Firstly, being a cross-sectional design, this study is unable to illustrate possible cause and effect relationships, as well as unable to track the possible effects of the outcome variables on the independent variables. Secondly, left-behind children and their caregivers are the study population, so we can not analyze any differences between children left behind and not left behind population. Thirdly, children's overall suspected developmental delay was measured by ASQ-C, which is a screening tool but not a diagnosis tool. Therefore the results should be interpreted carefully, but it is cost-effective and suitable for wide-scale screening in rural areas of China.