Adachi Child Health Impact of Living Difficulty (A-CHILD) Study: Research Protocol and Profiles of Participants

Background The Adachi Child Health Impact of Living Difficulty (A-CHILD) study has been conducted since 2015 to clarify the associations between socioeconomic factors and child health, as well as to accumulate data for political evaluation of the child-poverty agenda. This paper describes the purpose and research design of the A-CHILD study and the baseline profiles of participants, together with the future framework for implementing this cohort study. Methods We have conducted two types of continuous survey: a complete-sample survey started in 2015 as a first wave study to target first-grade children in all public elementary schools in Adachi City, Tokyo, and a biennial fixed grade observation survey started in 2016 in selected elementary and junior high schools. Questionnaires were answered by caregivers of all targeted children and also by the children themselves for those in the fourth grade and higher. The data of A-CHILD also combined information obtained from school health checkups of all school-grade children, as well as the results from blood test and measurement of blood pressure of eight-grade children since 2016. Results The valid responses in the first wave were 4,291 (80.1%). The number of households in “living difficulties”, such as low household income or material deprivation, stood at 1,047 (24.5%). Conclusions The A-CHILD study will contribute to the clarification of the impact of poverty on children’s health disparities and paves the way to managing this issue in the community.

A c c e p t e d V e r s i o n 5

Child poverty in Japan
In Japan, child poverty has recently become a major political issue. A UNICEF study showed that the financial and material deprivation level of Japanese children was relatively higher than that of children in other developed countries. 1 The proportion of children who lived in households below the poverty line, that is, households with less than half of the median household-size-adjusted income of the population, was 16.3 % as of 2012, 2 rendering Japan to be ranked 12 th among 35 OECD countries. 3 Confronted with child poverty in Japan, an act to accelerate policies for disadvantaged children was implemented in 2013, with all levels of local governments to tackle the various problems due to child's poverty. In such circumstances, while many related literature reviews and policy descriptions were published in Japan, there is a lack of studies which use quantitative data to study the effects of the poverty on child health and development.
Moreover, in spite of the presence of longitudinal studies in some developed countries, 4,5 such well-designed studies are still scarce in the Japanese context. 6 Cross-sectional study about recent conditions among children does not enough to help our knowledge about developmental processes that could be a clue for the prevention and early intervention efforts. Therefore, it is of practical and academic value to conduct a A c c e p t e d V e r s i o n 6 longitudinal study that not only studies the effect of poverty on child health and development in Japan, but also evaluates each policy for disadvantaged children in poverty.
We herein describe the aim, conceptual content, study design, and the characteristics at baseline and following waves of a cohort study, the Adachi Child Health Impact of Living Difficulty (A-CHILD), which started in 2015 with the purpose of monitoring and revealing factors related to the health and development of the child in poverty in Japan.

Project for child poverty in Adachi-city
Adachi City is one of the 23 special wards in the Tokyo Prefecture, and is located in northernmost area of Tokyo. As of 2015, the population in Adachi City was approximately 690,000. The average income in Adachi City is JPY3.3 million (USD1 ≈ JPY122) in 2015, 7 which is similar to the average in Japan (JPY3.1 million) but lower than that of the whole Tokyo (4.5 million JPY). 8 The proportion of population that received public aid was 3.7 % in 2015, which was the highest rate in the Tokyo Prefecture (mean 2.2 %). The healthy life expectancy in Adachi City is two years shorter than that of Tokyo Prefecture, 9 which suggests a health disparity between the people in Adachi City and those in other areas in Tokyo. One possible reason for this could be that there are more people in Adachi City who suffered from lifestyle chronic diseases such as diabetes complicated with kidney diseases, 9 which could have stemmed from poor dietary habits or overweight in childhood. It has been shown that the origins of adult health disparities were consequential of the physical and developmental damages due to poverty in the early periods of childhood. 10,11 Under the leadership of the mayor of Adachi City, Yayoi Kondo, the city government initiated the "Adachi Project Connected to the Future (child poverty implementation plan of Adachi City)" in fiscal year 2015 with aim of enabling children to adopt a desirable lifestyle and enjoy good health without being affected by household conditions or the environment where they were born and raised in. The political framework and goals for disadvantaged children were viewed by the Japanese Cabinet Office as a model case by a local government. 12 In the same year, in collaboration with the Adachi City government, we started A-CHILD study. Before the project in Adachi City, i.e. "Adachi Project Connected to the Future" was disseminated, we were able to obtain the data of the first wave, which would show the baseline status of children and their families. We also follow up with targeted children for the implementation period of the project in Adachi City.

The overall concepts and aims of A-CHILD study
A sizeable number of studies in medical, psychology and social epidemiology have demonstrated that both early-childhood experiences of poverty have extended impacts on their lifelong physical and psychological development. [13][14][15][16] Childhood poverty is also related to an increased risk of poor academic achievement and which may consequently lead to lower income 5,17 , resulting in "the cycle of poverty" through generations.
Recent studies about child poverty have focused on not only monetary but also multidimensional aspects of poverty, 18-21 whose evidence would provide a broad understanding of the actual deprivations with which children are confronted. Monetary poverty indicators such as low household income have been widely used for studies of child poverty, however, monetary indicators alone would not be sufficient to capture the state of poverty because of the lack of capturing specific values, such as properties, debts and benefits in-kind. 22, 23 Therefore, alternative multidimensional poverty approaches have been developed, 21, 24 including Townsend's relative deprivation, 25 basic needs, 26 or social exclusion approach. 19,27 By viewing child poverty from multiple perspectives, it will be possible to examine the path that poverty could have on children over the long A c c e p t e d V e r s i o n 9 term. But so far in Japan, few studies have identified both multiple aspects of poverty and child's health outcomes.
Based on the recent policy and research situation regarding child's poverty in Japan, our aims of A-CHILD study are as follows. First, we describe the current situation of children in poverty and their health status in a local-governmental area. In collaboration with the Adachi City in Tokyo, we conducted a longitudinal study targeting children and their families from the fiscal year that this city had started taking action for disadvantaged children. Second, we focus on the effect of not only the monetary aspects but also the multidimensional aspects of child poverty to explore the impact of low income, material deprivation, or payment difficulty in family on the child's health and development. We will examine some specific concerns, for example, whether material deprivation such as not the preparation of appropriate books/toys for children's age would affect children's psychological development, even if children's household is not low-income. Third, we aim to reveal the mechanisms and impact of local governmental policies on the reduction of health disparities in children. For example, food education emphasizing eating vegetables has been implemented at a public nursery school in Adachi City. Therefore, we will investigate whether the dietary habits cultivated before attending school would affect the dietary habits and physical growth of children after schooling, regardless of the family's economic situation. An important aim of the A-CHILD study is to elucidate the modifiable mediator of the link between poverty and child's health, which would break the cycle of intergenerational poverty.

Study design
A-CHILD study is broadly divided into "original survey" using questionnaires developed for A-CHILD study itself, and another part using data of the survey and medical examination that Adachi City has carried out for elementary and junior high school students. The original survey also included two types of surveys: longitudinal surveys that began in the first grade of elementary school, and biennial cross-sectional surveys of fourth grade or older. "Wave" indicates the year of the survey since the first year of A-CHILD study. The details and purpose of each survey are shown below.

Original questionnaire survey
We have conducted two types of continuous survey. The first was a complete-sample survey started in 2015 as a first wave study to target first-grade children (aged 6-7 years old) in all public elementary schools in Adachi City. In order to create a panel data that was started from the first grade, we conducted a follow-up survey on the children who The questionnaire was answered by caregivers and their children. Adachi City is divided into five administrative areas, each with different geographical characteristics. Therefore, we asked for cooperation in the survey so that we could get answers from at least one elementary and a junior high school from each administrative area. As a result, nine elementary schools and seven junior high schools were selected, which in major urban areas with hub stations or in residential areas away from the station. That is, the second wave comprised both a complete sample survey of second-grade children in all grades. Participants were informed that participation in the study was voluntary and that participants could indicate in the questionnaire their disagreement if when they did not agree to participate in the survey. All children were assigned a unique ID by an administrative officer of Adachi City, and we can access only the data without personal information such as a child's name and class number. Since the second and subsequent waves, the annual data of each survey were linked to this ID.

Data from Study Attitude survey and school checkups
In the second wave in 2016, we also used data from the Study Attitude survey, which was  The data of the A-CHILD study features objective child health information obtained from mandatory school health checkups including physical measurement and dental checkup. Further, with caregiver's consent, the eighth-grade children were invited to undergo a "child lifestyle-related health checkup", which included a blood test and the measurement of blood pressure.
Among the original questionnaires returned, we adopted as valid responses only the answers that could be merged with the data of the Study Attitude survey and school health checkup via a unique ID, excluding the questionnaires from caregivers who did not agree to participate or those that were submitted blank. The questionnaire answered by the child was validated only when the child him/herself agreed to participate, and the caregivers also agreed that their children answered. The flowchart of the participant recruitment is shown in Figure 1 to 3. We received the endorsement for each survey from the Adachi City government, which contributed credibility and higher participation rates of this study.

Main outcome measures
We prepared a set of questionnaire items relevant to the circumstances of poverty and other social determinants of health and those underlying mechanisms based on currently were excluded from the question items because these data had already been acquired in the first wave. On the other hand, questions were added to the second wave items to obtain more details than those in the first wave, such as parents' membership of community groups and children's participation for events in the community (Table 1).
The question items for the caregivers of the children in fourth, sixth and eighth A c c e p t e d V e r s i o n 15 grades in the second wave were almost the same as those for caregivers of first-grade children in the first wave, but did not include the same items in the questionnaire for children who would be responding themselves. We also modified several sentences in the questionnaire to capture what we would like to know based on the qualitative and quantitative review of results in the first wave. Table 2 summarizes the items answered by children since the second wave. These items included children's lifestyles such as eating habits, wake-up time and bedtime, toothbrushing, frequency of reading books, places they spent time in after school, and so on. It also included subjective questionnaires such as school social capital and their selfesteem. According to child's grades, we also added questions on whether they could do simple cooking on their own (for the children in sixth and eighth grades only) and whether they were interested in dieting (for children in the eighth grade only).
Physical measurement and dental checkup in the school health checkup were assessed according to a national-standardized guideline. 28 Details of how each checkup was conducted are described in the online supplementary material.  Table 3.
In the A-CHILD study, poverty of children should be considered not only in terms of household economic hardship but also in the aspect of whole family environment. Therefore, we defined "households in living difficulty" as households that fell under any one of the following: (1) annual household income below JPY3 million; (2) presence of material deprivation; and (3) experience of payment difficulty. For the following reasons, the threshold for economic difficulty was set to less than JPY3 million per household.
First, assuming maternal households receiving public assistance (for example, a 30s mother and a 1st-grade child in elementary school), the annual income will be equivalent to JPY2.72 million when calculated based on welfare standards. In addition, there was little difference in the proportion of households with material deprivation and experience A c c e p t e d V e r s i o n 17 of payment difficulty between households with an annual income of JPY2 million and households with JPY3 million. To avoid missing the households in living difficulty, it seemed reasonable to consider that income groups with less than JPY3 million lived in economic difficulty. In the questionnaire, we asked about things that respondents could not possess for economic reasons; not only home appliances for daily necessities and savings that can be dealt with when needed, but also places and goods which are considered necessary for child well-being (Table 4). We defined the existence of material deprivation when the respondent did not have at least one of the necessities. We also asked if they had any experience of being unable to pay for something in the past year (Table   4). In the first wave, the number of households with income of less than JPY3 million was 489 (11.6%), the number of households with material deprivation was 670 (15.8%), and the number of households with payment difficulty was 389 (9.2%). The number of households with at least one of the above three aspects of poverty, that is, households in living difficulty, was 1,047 (24.5%) ( Table 5). Among households receiving/received public aid, which was one of the indications of financial difficulty, the percentage of households in living difficulty was 84%. This high percentage indicates the validity of the definition of living difficulty.
The second wave Figure 3 shows the flow chart of subjects in the target grades of the second survey. Among the collected questionnaires, we removed the responses without consent for participation in the survey, those who did not answer the Study Attitude survey, those who did not have school health checkup data and those whose child refused to answer the child survey for the fourth, sixth and eighth grades. We identified as valid responses 4,358 in second grade The selected results of the questionnaire for the caregivers who participated in the second wave are shown in Table 6. Furthermore, the households of the second wave were also counted according to the aspects of poverty as the first wave (Table 4). Table 7 shows selected results of the questionnaires answered by children in the fourth, sixth and eighth grades.
The results of physical measurement and dental checkup in school health checkup are shown in Table 8. Among the target children of in the eighth grade in the second survey in 2016, 502 (85.4%) underwent additional blood test and a measurement of blood pressure. The averages of the results are also shown in Table 8.
Missing rates for some questions were very low (e.g. 0.1% for child sex in first-graders in 2015). On the other hand, questions related to the finance in households, such as household income (5.5% in 2015) and receipt of some kind of public aid (3.5 -14.8 % in 2015), tended to be higher missing rate (results not shown in Tables). We assume those missing values would be occurred in not at random, that is, systematic differences should remain between the missing values and the observed values. We did not address any adjustments for those missing and remained them in the data, because it should be left to the researchers how to handle incomplete data, in accordance with each research question.

Strengths and limitations
The Japanese act to accelerate the development of policies for disadvantaged children declares that children in poverty should grow up in a desirable environment in terms of education, living conditions, working conditions, and economic perspectives, so that they are not affected by the environment in which they were born and raised in. The principle of this anti-child poverty act in Japan aligns with the recent public health goal that we should tackle the social determinants of health to reduce the health gap, which is outlined  interactions that determine health and its disparities in a child's growth process at the individual, household and school, and regional level, by adopting an appropriate analytic strategy such as multi-level analysis.
However, it should also be noted that the A-CHILD study does have its limitations. First, the study was conducted in Adachi City, which is one of the 23 special wards in Tokyo, and the targeted schools for biennial observational surveys of children in the fourth grade and higher make up only a portion of the public schools in the city. In addition, while the data of the "child lifestyle-related health checkup" with blood test and blood pressure measurements are valuable, only applicants in eighth grade underwent this checkup. For these reasons, the generalizability of the result from the A-CHILD study may be limited. Our target population is the children who lived in Adachi City, however, Although the response rates for A-CHILD were high enough for any grade, some questionnaires remained unanswered. For example, if the parents were rarely at home, or were seriously sick, or if familial relationships were not working, the questionnaire might not be answered by their caregivers. Also, caregivers who were not native Japanese may find it difficult to answer because the questionnaire was in Japanese. In A-CHILD, we A c c e p t e d V e r s i o n 22 indicated contact information not only in Japanese but also English, Chinese, Korean, and Tagalog, but there was no actual inquiry. Lastly, given that drop-outers from the cohort study would differ in several ways from the respondents, any conclusions drawn from the study may be biased. Previous studies suggested that dropout from the cohort study was not at random but tended to be higher among lower socio-economic populations. [30][31][32] Therefore, any additional treatment for non-response in the follow-up, e.g. inverse probability weighting or multiple imputation, 33-39 could be a benefit for the researcher using the longitudinal data of A-CHILD for handling potential non-random dropouts.
In spite of these limitations, almost all local municipalities in Japan have now begun to tackle the issue of child poverty, with the results obtained in Adachi City being very meaningful and helpful for these municipalities. In the future, it is necessary to clarify whether experiences of living difficulty can influence physical and psychological development, behaviors, and the health status of children as they get older. For that purpose, we plan to continue to follow the children participating in the A-CHILD longitudinal survey until at least the eighth grade and beyond. With these advantages, the A-CHILD will continue to contribute important information for the development of    The questionnaires in Study Attitude survey in Japanese conducted by Adachi city are available if required.