Li et al BMC Public Health (2022) 22:1958 https://doi.org/10.1186/s12889-022-14268-5 BMC Public Health Open Access RESEARCH The effect of ECD program on the caregiver’s parenting knowledge, attitudes, and practices: based on a cluster-randomized controlled trial in economically vulnerable areas of China Ying Li1, Shanshan Li2,5, Lei Tang3* and Yu Bai4 Abstract Background The first three years of life are the critical and sensitive periods for the formation of individual abilities However, existing data indicates that early childhood development (ECD) in economically vulnerable areas of China is lagging, which is closely related to the lack of parenting knowledge and poor parenting practices Methods We conducted a non-masked cluster-randomized controlled trial in a former nationally designated poverty county of China All 6–36-month-old children and their caregivers living in 18 communities/clusters (10 towns and districts of the county seat) were enrolled in a 9-month parenting training program In the treatmentgroup communities, ECD centers were installed where community workers provided parenting training sessions If caregivers were unable to visit the center, home-based parenting training was offered No intervention was provided to the control group Furthermore, we assigned half of the treatment group to receive monthly developmental feedback in addition to the parenting training Based on the baseline and follow-up data, we investigated the treatment effects on parenting knowledge, attitudes, and practices through Intention-to-Treat (ITT) and Treatment-onthe-Treated (TOT) analyses Results We found no effects on the parenting knowledge and attitudes of the caregivers but significant effects on the parenting practices The effects were heterogeneous among families with different characteristics Specifically, on average, the program had the largest effect on internally oriented caregivers, mothers with higher education, and mothers who are primary caregivers We want to emphasize that, although the ITT effect on parenting practices (the average treatment effect) were stronger for mothers with higher education, the TOT effect on parenting practices (the local average treatment effect, LATE) were stronger for mothers with less education That is, even though on average the program helped mothers with higher education, but among complier families, the program benefited mothers with less education *Correspondence: Lei Tang tangleiceee@163.com Full list of author information is available at the end of the article © The Author(s) 2022 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data Li et al BMC Public Health (2022) 22:1958 Page of 17 Conclusion The findings indicate that, at least in the short run, the program can directly change caregivers’ parenting practices without changing their knowledge and attitudes Future studies are needed to investigate whether parenting knowledge and attitudes can change in the long run Keywords Early childhood development, Parenting knowledge, Parenting attitudes, Parenting practices, Economically vulnerable areas, Randomized controlled trial Introduction The human brain and nervous system undergo rapid changes during early childhood Early childhood development, such as, motor, language, cognition, social emotion, and other fields (Aboud and Yousafzai 2015) can predict children’s future academic performance, human capital accumulation, and their adulthood income levels (Engle et al 2007; Black et al 2008; Currie and Almond 2011) However, existing studies show that approximately 249 million children under the age of five in lowand middle-income countries (hereinafter referred to as “LMICs”) are at risk of poor development, of which 17.43 million (roughly 8%) are in China, ranking second in the world (Lu et al 2016) The ECD problem is particularly severe in economically vulnerable areas in China According to an empirical study, half of the children living in rural China are at risk of cognitive developmental delay, and 52% are at risk of language developmental delay These risks will increase over time if no measures are taken (Yue et al 2019) Children can make significant progress in motor, language, cognitive, and socio-emotional development if properly cared for (Grover 2005) Caregivers are the primary providers of stimulating and supportive experiences related to ECD (Britto et al 2002; Richter 2004; Bradley and Corwyn 2005) However, in LMICs, children are exposed to a variety of psychosocial risk factors for not receiving appropriate care, including insufficient stimulation, ineffective parenting practices, unresponsive care, and the inability of parents/caregivers to understand infant behaviors (Wallander et al 2014; Yue et al 2017; Li et al 2019), all of which can have a detrimental impact on a child’s development The above risk factors can be classified as lack of parenting knowledge, attitude, and practices Parenting knowledge is defined as an understanding of child development norms, milestones, developmental processes, and familiarity with parenting skills (Benasich and Brooks-Gunn 1996), and it is a predictor of child development (Huang et al 2005) Existing studies have not only found a direct link between parenting knowledge and children’s outcomes, including reduced behavioral problems and improved cognitive and motor performance (Benasich and Brooks-Gunn 1996; Dichtelmiller et al 1992; Rowe et al 2015), but have also indicated that parenting knowledge and beliefs have a continuous and stable influence on children’s outcomes (Zigler 1992) The mechanisms are that parenting knowledge influences a child’s outcomes by shaping the family environment and improving the quality of parenting practices (Huang et al 2005; Sajedi et al 2016) Knowledgeable mothers are more likely to provide their children with books and learning materials suitable for their children’s interests and age, and knowledgeable mothers are more likely to read, talk, and tell stories more to their children (Grusec 2011; Gardner-Neblett et al 2012) Moreover, knowledgeable mothers may provide a warm and positive environment that promotes children’s social-emotional development (Smith 2002) Parenting attitudes are the product of parenting knowledge, parenting values, and goals (or expectations) for their children’s development These values and goals are, in turn, influenced by cultural, social, and parental experiences and their overall values and goals (Rogoff 2003; Okagaki and Bingham 2005; Iruka et al 2015) To a certain extent, the formation of attitudes is determined by parental self-efficacy, that is, the ability of parents to perceive their influence on the child’s development Parental self-efficacy affects parenting abilities, parenting practices, and children’s abilities (Jones and Prinz 2005) Parenting practices refer to parent-child activities, such as reading, singing, playing with children, and cultivating a child’s sense of discipline Previous studies have shown that parenting practices play an important role in individual development (Evens et al 2000; Black et al 2017; Yue et al 2017) For example, Bai et al (2019) found that the better the parent-child interactions and discipline behaviors, the lower the risk of delayed child development In the short-term, positive parenting practices can stimulate and promote children’s cognitive and language development as well as stimulate and maintain children’s enthusiasm and interest in learning In the long term, it has a positive impact on children’s early literacy, academic performance, and future life happiness (Darling and Steinberg 1993; Mulvaney et al 2006; Keels 2010; Page et al 2010) Due to the positive role of parenting knowledge, attitudes, and quality practices in child development (NASEM, 2016), the Lancet ECD series has embraced nurturing care as the basis for successful ECD strategies (Britto et al 2017) Most ECD programs in LMICs focus on psychosocial stimulus interventions that are typically directed at encouraging parents to provide children with opportunities to explore their surroundings, Li et al BMC Public Health (2022) 22:1958 solve problems, and interact with others (Jeong et al 2018) These projects are effective in improving child development (Yousafzai and Aboud 2014; Yousafzai et al., 2014; Aboud and Yousafzai 2015; Britto et al 2015, 2017; Rao et al 2017) However, it should be noted that the key to the success of these programs is their ability to change the knowledge, beliefs, attitudes, and practices of caregivers Some studies using randomized controlled trial methods have found that ECD interventions not only have a positive impact on parents’ knowledge of child development but also have an impact on caregivers’ practices in caring and feeding their children (Alkon et al 2014; Yousafzai et al 2015) Significant benefits for mothers’ parenting practices have also been found in some ECD interventions aimed at improving parent-child interactions by promoting mothers’ sensitivity and responsiveness to infants (Cooper et al 2002, 2009) Previous research has also shown that providing caregivers with opportunities to learn how to observe and respond to their children through games and communication interactions can improve parenting knowledge, the quality of the family environment, parenting involvement in child development, and parent-child interactions (Yousafzai et al 2015) Some studies on disadvantaged children and their families in developing countries have found significant impacts on mothers’ knowledge of child development (Rahman et al 2008; Powell et al 2004; Aboud 2007; Jin et al 2007) An analysis of 34 home visits for high-risk infants revealed that home-based parenting interventions continuously improved the family environment and parenting skills (Kendrick et al 2000) Similarly, a review of six large-scale home visit programs in the United States found that home-based parenting interventions have a positive impact on parenting attitudes and practices, especially for those who need them the most (Gomby et al 1999) Another study of preterm infants found that home-based interventions can promote more sensitive and responsive parenting skills, thereby improving parent-child interactions (Goyal et al 2013) A systematic review also found that ECD interventions in LMICs targeting children under two years of age have positive impacts on parenting outcomes Specifically, the interventions had medium-to-large positive effects on the home care environment, parenting knowledge, and mother-child interactions (Jeong et al 2018) While previous studies have investigated the effect of ECD interventions on parenting knowledge, attitudes, and practices, to our knowledge, there has been no comprehensive evaluation of the effect of such ECD intervention programs on the improvement of parenting skills or capacities in China, which is the key mechanism for the success of ECD programs (Jeong et al 2018) Moreover, Page of 17 parents from various cultural backgrounds have different expectations of their children’s socialization, parenting attitudes, and practices (NASEM, 2016), which may result in different effects on children’s development In addition, few studies have examined such effects of government-led, multi-delivery models, all-inclusive, and large-scale ECD programs In recent years, the Chinese government has released several policy documents to emphasize the importance of early development of children under the age of three However, the implementation of this policy remains to be explored This study used a non-masked cluster-randomized controlled trial to assess the causal effects of a government-led ECD intervention project implemented in economically vulnerable areas in China on caregivers’ parenting knowledge, attitudes, and practices The rationale of assigning treatments on cluster levels is to avoid contamination We believe that the results of this study will be of great significance for not only the Chinese government but also other countries in a similar situation to formulate comprehensive intervention policies to promote ECD in rural areas Data and methods Trial design This cluster-randomized controlled trial was conducted in a former, nationally designated poverty county in Shaanxi Province, China To minimize the risk of contamination across the treatment and control groups, we randomized the treatment (i.e., delivery of weekly parenting training) at the community level among 18 communities/clusters (10 towns and districts of the county seat) The treatment group was randomly divided into two treatment arms: “feedback” and “no feedback.” These two treatment arms were enrolled for a weekly parenting training program, but the feedback arm received monthly feedback on the child’s developmental progress on the top of the parenting training program The control group did not receive any treatment The intervention was implemented for nine months, from July 2018 to March 2019 A baseline survey was conducted in June 2018, followed by a follow-up survey in April 2019 Randomization and masking In January 2018, the research team generated a random allocation sequence using the STATA program for the random assignment of the communities into control and treatment groups Furthermore, half of the treatment group were randomly assigned to the two (i.e., “feedback” and “no feedback”) treatment arms at the individual level and stratified by the children’s levels of development (i.e., the baseline scores of Ages and Stages Questionnaires, Third Edition (ASQ-3)) The reason of stratification is to ensure balance of the treatment arms and to increase Li et al BMC Public Health (2022) 22:1958 Page of 17 Fig 1 Trial profile efficiency In June 2018, the research team enrolled participants and assigned them to interventions at the community level Although complete masking was not possible in this study, the caregivers and parenting trainers were unaware that they were involved in the experiment Furthermore, the survey teams were blinded to the group assignments Sample size determination The sample size was determined using a power calculation with a detectable effect size on the main outcome variable of interest at 80% power, given a two-sided significance level of 0.05 Allowing for an attrition rate of 10%, based on evidence from earlier field experiments in rural China, we assumed an adjusted intraclass correlation coefficient (ICC) of 0.01 and that baseline scores account for 50% of the variation in scores at follow-up Based on these parameters, we calculated that nine clusters of 35 children per treatment arm would allow us to detect an effect size of 0.24 SD at 80% power, given a twosided significance level of 0.05 Participants We obtained the birth registration of the study county from the county-level office of the National Health Commission (NFC) and recruited all children aged 6–36 months and their primary caregivers Birth registration was confirmed by local councilors and the research team On average, 14 children were enrolled in each cluster or community As of June 2018, the baseline sample included 995 child-caregiver dyads from 18 communities Due to migration, illnesses, or short-term leave of residence (such as visiting relatives in other places during the survey period), the sample attrition rate was approximately 15% The resulting follow-up sample included 845 child-caregiver dyads as of April 2019 Furthermore, we excluded samples with inconsistent types of caregivers in the baseline and follow-up surveys, and the final sample for analysis included 643 child-caregiver dyads The trial profile of this study was as follows (see Fig. 1): All study protocols were approved by the institutional review boards (IRBs) All subjects provided written informed consent to participate in the experiment and data collection before the commencement of the project There was no known harm or risks to the participants The development of research questions and outcome measures was not influenced by participants’ priorities, experiences, and preferences Research reports were available to participants upon request Li et al BMC Public Health (2022) 22:1958 Page of 17 Table 1 Baseline characteristics of the participants in the treatment and control group Control Individual characteristics Gender Months Premature Minority The child was firstborn Low birth weight ( 9 years of formal education) Age of the mother (years) Hukou of parents Whether is Dibao (low income family) Distance from the home to township seat N Treatment 0.484 (0.025) 20.594 (0.513) 0.041 (0.008) 0.085 (0.038) 0.745 (0.026) 0.043 (0.007) 0.026 0.489 (0.018) 20.139 (0.370) 0.044 (0.012) 0.008 (0.004) 0.780 (0.021) 0.045 (0.010) 0.004 (0.066) 0.111 Table 2 Baseline parenting knowledge, attitudes, and practices in the treatment and control group p-value 0.888 0.455 Knowledge KIDI total score Attitude PLOC overall 0.878 PLOC parental efficacy 0.046 PLOC parental responsibility 0.276 0.863 PLOC child control of parent’s life 0.824 PLOC parental control of child’s behavior (0.073) -0.082 0.073 (0.081) 0.072 (0.084) -0.046 0.269 Practices Parent-child interactions Total parent-child interactions Reading books (0.070) (0.081) 0.718 0.667 (0.022) 0.450 (0.031) 0.443 (0.080) 29.232 (0.511) 1.397 (0.126) 0.112 (0.063) 29.267 (0.375) 1.354 (0.114) 0.124 Telling stories 0.172 Singing songs (0.029) 1.516 (0.021) 1.561 (0.156) 483 (0.205) 481 0.933 Taking child outside the home for play 0.957 Playing with the child with toys 0.784 Naming things, counting, drawing 0.738 Disciplining practices Total discipline practices 0.859 Taking away children’s things Notes for variable Minority: China is a unified multi-ethnic country Due to the large population of Han ethnicity (over 90% of the country’s total population), it is customary to refer to the other 55 ethnic groups as ethnic minorities In terms of differences among the ethnicities, Han ethnicity speak Chinese, and some ethnic minorities have their own languages and scripts In addition, there are differences between Han and some ethnic minorities in life customs, values, folk customs, and etiquette In terms of geographical area of residence, most ethnic minorities are distributed in frontier provinces, and some of them migrate to live in Han-dominated regions In the sampled area of this study, there is a very small number of Hui ethnicity These few Hui people have been migrated to the sampled area and integrated into local cultures for a long time, thus, there is no significant differences in terms of language, values, and customs between these minorities and the Han ethnicity The Chinese government does not discriminate in preferential policies neither Therefore, in terms of the intervention, the two groups received the identical treatments Intervention and implementation Limiting time Explanation N Control Treatment p-value 0.518 (0.004) 0.504 (0.005) 0.031 65.209 (0.245) 15.505 (0.133) 16.111 (0.155) 15.215 65.630 (0.201) 15.814 (0.117) 16.280 (0.145) 15.236 0.139 (0.109) 18.378 (0.093) 18.312 (0.123) (0.166) 3.476 (0.167) 0.414 (0.045) 0.365 (0.039) 0.646 (0.044) 0.842 3.348 (0.162) 0.457 (0.035) 0.346 (0.030) 0.566 (0.035) 0.809 (0.022) 0.783 (0.029) 0.425 (0.024) 0.717 (0.026) 0.453 (0.025) (0.034) 7.068 (0.123) 2.688 (0.036) 2.660 (0.090) 1.719 (0.052) 483 7.069 (0.111) 2.730 (0.039) 2.679 (0.073) 1.660 (0.044) 481 0.081 0.439 0.879 0.754 0.555 0.436 0.676 0.139 0.317 0.065 0.502 0.992 0.418 0.874 0.388 All children and their caregivers in the treatment group were enrolled in a government-led ECD intervention project The intervention focused on child psychosocial stimulation (hereafter referred as “weekly parenting training”) These services were delivered by locally employed and project-trained parenting trainers who had educational attainments at the level of the mothers that they provided the services The local government was in Li et al BMC Public Health (2022) 22:1958 Page of 17 Table 3 ITT analysis of treatment on parenting knowledge, attitudes, and practices Treatment effect Coefficients SE Panel A Parenting Knowledge(N = 563) KIDI total score Panel B Parenting Attitude (N = 563) PLOC overall PLOC parental efficacy PLOC parental responsibility PLOC child control of parent’s life PLOC parental control of child’s behavior Panel C Parenting Practices(N = 564) Parent-child interactions Reading books Telling stories Singing songs Taking child outside the home for play Playing with the child with toys Naming things, counting, drawing Disciplining practices Taking away children’s things Limiting time Explanation Romano-Wolf P-value 0.031 0.067 0.951 0.012 0.113 0.015 0.004 -0.088 0.084 0.079 0.082 0.062 0.103 1.000 0.406 1.000 1.000 0.762 0.126 0.136 0.064 -0.037 0.050 0.046 0.040 0.036 0.030 0.010 0.319 0.762 0.049 0.146 0.033 0.038 0.406 0.010 -0.060 -0.101 -0.051 0.060 0.089 0.066 0.762 0.703 0.772 Notes: In all regressions, we controlled for baseline parental knowledge, attitude and practices, child characteristics, and family characteristics The scores for parental knowledge and attitudes were internally standardized All standard errors were clustered at the village level We also adjusted the multiple hypotheses testing problem using the step-down procedure of Romano and Wolf (2005) to control for the family wise error rate (FWER) The significance levels are as follows: *p