Negative childhood experiences and health inequalities among adults over 45: Evidence from China

Background : Growing health inequalities have become an important challenge for Chinese society. This study analyses the impact of negative childhood experiences on health among Chinese adults over 45. Methods : This study utilized data from the 2014 Life Course Survey under the China Health and Retirement Longitudinal Study, with a total sample of 5,842 adults over the age of 45. This paper used a linear regression model to analyse the factors affecting the self-rated health of respondents. The concentration index was then used to measure the degree of health inequality, and finally, we decomposed the concentration index. Results : The regression results showed that negative childhood experiences had a significant negative impact on the health of adults over 45 ( β = -0.021, p < .000). The concentration index decomposition found that negative childhood experiences was associated with more health inequalities. The elasticity coefficient for negative childhood experiences was -0.119, the concentration index of negative childhood experiences was -0.019, and the contribution to health inequalities was 12.86%. The mechanism test and analysis found that negative childhood experiences have a greater negative impact on the health of the low-income group than on that of the high-income group. In fact, negative childhood experiences have reduced the socioeconomic status of respondents, leading to a decline in healthcare utilization, which has increased health inequalities in China. Conclusion : This study found childhood was associated with health 45 and to reduce health inequalities, more help and health interventions should be offered to groups with negative childhood experiences.


Theoretical Analysis and Literature Review
Research on health inequalities focused on how to measure health inequalities and the causes of health inequalities. There are many ways to measure health inequalities, but the most common is the concentration index. Because health inequalities can be calculated, further mathematical methods can be used to study the causes of health inequalities [9,10]. Moreover, the factors affecting health inequalities can be divided into two categories. The first category is social policy, such as social medical insurance policy and social welfare policy [11,12]; the other category includes individual socioeconomic characteristics, and income is the factor that draws the most attention.
The general conclusion of this research is that high-income groups have better health status and low-income groups have worse health status, leading to health inequalities among the "pro-rich" in many countries [13][14][15].Previous literature has focused on what 5 factors affect health inequalities, and can provide a reference for seeking controlled variables in this paper, but the above literature ignored the impact of negative childhood experiences on residents' health and health inequalities.
Since the 1960s, the life course theory has been applied in social science research. This theory focuses on the impact of events in the full life course on individual development [16]. In recent decades, the life course theory has gradually become an important analytical theory for studying health, health behaviour and death. In the field of health research, the life course theory holds that childhood is the foundation for the development of people's entire life cycle; in other words, the events experienced in childhood are the "starting point" and basis of many diseases of adulthood [17].
However, Ben-Shlomo and Kuh [18] found that socioeconomic status in adulthood can compensate for this negative effect. Thus, this section mainly discusses health impacts from two perspectives: negative childhood experiences and socioeconomic status in adulthood based on this life course theory.
Extensive studies have been conducted to test the relationship between negative childhood experiences and health, and the results were mixed. On the one hand, some researchers have found that negative childhood experiences may lead to health issues in adulthood. Acquired diseases such as heart disease, diabetes, and high blood pressure may be attributed to hypogenesis [19] and malnutrition [20,21] during the foetal period or during childhood experiences of famine [22]. With randomized experiments in developing countries, some economists have even discovered that negative childhood experiences have led to cognitive dysfunction, which in turn has brought about health 6 issues in adulthood [23][24][25]. In addition, some studies found that famine or malnutrition early in life had no significant effect on health [26,27]. However, on the other hands, some researchers have showed opposite results. For example, some scholars have analysed government-implemented child policies from the opposite perspective and their impact on health and income in adulthood, such as the malaria eradication campaign in South America [28] and the US food stamp programme [29]. They found that improvements in nutrition in childhood could help to improve health and income in adulthood.
Regarding research on socioeconomic status and health, some literature has found that improved socioeconomic status in adulthood can have a protective effect, which can compensate for the negative effect on health brought by negative childhood experiences [30]. However, some scholars found that negative experiences in the early years were highly related to socioeconomic status in adulthood [1,31]. Montez and Hayward [32] found that negative childhood experiences increased the incidence of disability and mortality in adulthood, which leaded to low socioeconomic status. Shi [7] found that childhood experiences of poverty and abuse not only leaded to health issues but also declines in socioeconomic status and unhealthy lifestyles in adulthood.
Regarding such research in China, many scholars paid attention to famine of "threeyear hard period" when the Chinese suffered through a very severe famine from 1959 to 1961and people born in this period of time generally have experiences with "starving" and "fleeing". Studies found that individuals born or growing up in the "three-year hard period" generally have shorter heights, worse health and lower income in adulthood 7 [27].
Though existing health research indicates growing interest in childhood experiences, the following research gaps still exist: 1) existing studies focus on the impact of negative childhood experiences on health, but none of them have studied its impact on health inequalities; and 2) most of the current research focuses on single experience, and overestimates its impact on health, but ignores the cumulative effects of different experiences on health or even health inequalities [33]. This paper will fill in such research gaps, and enrich existing research by contributing literature on health inequalities in the following aspects. First, CHARLS Life Course Survey study is used to provide retrospective data on childhood experiences from a large number of respondents. The survey comprises individuals over 45 years old, and most of the respondents experienced negative events, such as famine in their early years. Second, we construct an indicator that can measure the extent of an individual's suffering from negative childhood experiences. This indicator aggregates multiple types of negative experiences and can help capture the cumulative stress of multiple traumas [34]. Third, we focus on the mechanism of the impact of negative childhood experiences on health inequalities, thus providing references for health inequality studies.

Data and sampling
This study utilized data from the China Health and Retirement Longitudinal Survey

Dependent variable
Self-rated health (SRH) was applied to assess the health level of adults over 45 in this paper. SRH has several outstanding advantages. First, unlike single indicators such as prevalence and disability rate, SRH is a comprehensive indicator that can fully reflect the individual's health status. Second, although SRH is subjective and has certain measurement errors, many studies have shown that SRH can effectively predict 9 mortality and functional damage [35,36]. Third, SRH is simple and easy to obtain. In this study, SRH is coded as "very poor, poor, fair, good, very good", respectively. 5.23%, 20.11%, 52.72%, 13.43% and 8.51% of the participants reported their SRH as "very poor, poor, fair, good, very good", respectively.

Independent variable
The independent variable is negative childhood experiences. Drawing from the selection of indicators from previous studies [38], we focus on exogenous and irreversible experiences. The independent variable in the present study were: parents' issues, violence, and famine and fleeing. Parents' issues were measured by these questions: "Did your parents have a long time be sick on bed when you were young?", "Did your parents have abnormality of mind when you were young?", "Did your parents often quarrel?", "Were your biological parents divorced?";Violence is determined by these questions: "When you were a child, had you often picked on or bullied by kids in your neighbourhood or your classmates?", "When you were growing up, did your female guardian often hit you? ";Famine and fleeing were measured by this question: "When you were a child before age 17, was there ever a time when your family (including your grandparents, parents, siblings, children and so on) did not have enough food to eat and even moved away from the famine-stricken area?".
For each of the questions above, "yes" indicated "1" and "no" was set to "0". Sum scores were calculated for the above questions to measure the numbers of negative childhood experiences to indicate the extent of an individual's suffering from negative childhood experiences. The sum scores ranged from 0 to 7, and 12.79% of the participants had never experienced negative events in their childhood.  [40]. Because there were no information of genetic inheritance and personality in the CHARLS Life Course Survey data, we use characteristics of the respondents' native family in their childhood as proxy variables, including both father and mother's education levels (in years, illiterate = 0, primary = 6, secondary = 9, high school =12, university=16), parent's family origin (poor peasant = 1, otherwise = 0), care received from parents (yes =1, otherwise = 0), and family economic status in respondents' childhood (good =1, bad = 0). Specific variable definitions and basic information are shown in Table 2.

Statistical analysis
Depending on the study purpose, previous studies have proposed two measures of inequality in health, namely, pure inequalities and socioeconomic inequalities. This article uses socioeconomic inequalities in health, specifically, inequalities in health related to income. The most common measure of health inequalities related to income is the health concentration index and its decomposition, where the concentration index is based on a concentration curve.
The health concentration curve contains two key elements: one is the health indication, and the other is socioeconomic status, which must be a continuous variable and was characterized by income in this study. The health concentration curve is combined, given both the cumulative population ranked by income (horizontal axis) and their corresponding health status (vertical axis). The concentration index is calculated based 13 on the concentration curve, which is twice the area between the concentration curve and the fair line [13]. The mathematical expression for the concentration index is as follows: where C is the concentration index and L is the concentration curve. When the concentration curve coincides with the fair line, C is equal to 0; when the concentration curve is above the fair line, C is negative; and when the concentration curve is below the fair line, C is positive.
The concentration curve and concentration index can only indicate the existence and extent of inequality. To investigate which factors contribute to health inequality, the CI decomposition method is needed. Through CI decomposition, the contribution of each influencing factor can be decomposed. This paper used the Wagstaff linear method, the basic idea of which is similar to that of linear regression, which decomposes CI into parts that can be interpreted by observable variables and parts that cannot be explained by data (similar to the residuals in the regression).
Assuming the health indicator is y , the health influencing factors are k x , and the linear regression equation is established as follows: where is the coefficient and is the residual. Assuming that the coefficient vector of each observation is the same, the difference in health comes from influencing factor k x . The concentration index can be expressed as follows: where is the mean of y , is the mean of k

Decomposition of health inequality
On the basis of the regression results, the abovementioned health inequality CI is decomposed, and the regression coefficient is used as the basis to calculate the elasticity coefficient and each variable's contribution to health inequalities. The results are shown in  Next, we explored the mechanism of negative childhood experiences in affecting health inequalities. The impact of negative childhood experiences on health inequalities may result from the different effects of negative childhood experiences on the health of different income groups, which in turn increases health inequalities. We aggregate the income of all respondents and find a median that can divide the respondents into 2 groups. The first group accounts for 50% of the total income, and the second group accounts for the remaining 50% of the total income. Furthermore, all respondents in the first group are poorer than those in the second group, through which we can examine the heterogeneity in different income groups (see Table 5). In Model 2 and Model 4, negative childhood experiences have negative effects on both the low-income group (β =-0.0209, p<0.001) and the high-income group (β =-0.0195, p<0.001). By comparing the regression coefficients, we find that negative childhood experiences have a larger negative impact on the low-income group and a smaller negative impact on the highincome group, which leads to an increase in health inequalities.

Discussion
This paper found that negative childhood experiences have a negative impact on an individual's health, and has a greater negative impact on the low-income groups, thus expanding the health inequality of the residents. This means that both the high-income group and the low-income group have lower socioeconomic status, but the negative impact on the low-income group is greater, while the negative impact on the highincome group is smaller [41].
Many studies have demonstrated the relationship between socioeconomic status and individual health [41,42]. Link and Phelan [43] found that socioeconomic status was the most influential factor on an individual's health and life expectancy and that the health of those with higher socioeconomic status was significantly better than the health of those with lower socioeconomic status [44]. Socioeconomic status not only has an impact on people's health but can also have a cumulative effect; that is, people who have long had higher socioeconomic status (or lower status) have better (or worse) health [45].
Education, occupation and income had been thought to be important indicators of socioeconomic status [46]. In health research, education reflects a person's ability to actively access social, psychological and economic resources; occupation reflects a person's social status, sense of responsibility, physical activity status and health risk; income level reflects a person's consumption ability, housing conditions, nutrition status and access to medical and health care resources [47]. Therefore, the unfortunate 20 experience of childhood may affect the access to medical service resources and expand health inequality by reducing the residents' education level, professional status and income level.
Negative childhood experiences reduce the possibility of residents' access to education [48]. However, there is a positive correlation between education and health.
For example, Winkleby et al. [42] found that education affects people's health through multiple mechanisms and that high levels of education usually translate to better jobs, higher incomes, and greater budget constraint sets [49]. With more economic resources invested in health, such as medical insurance, medical services, health care equipment and facilities, individuals are more likely to have better health [50]. Therefore, negative childhood experiences may change the investment and utilization of medical services of residents through the level of education, and then expand the health inequality.
Negative childhood experiences reduce the possibility of people getting good jobs [7].
Worse jobs increase the possibility of risk exposure in the future, resulting in people in worse jobs facing higher health risks and lacking in health insurance, which makes it difficult to obtain better medical service resources and health care [51]. In China, people in different groups (such as urban workers, urban residents, rural residents, rural migrant workers, etc.) enjoy different medical insurance. The financing level and subsidy standard of the basic medical insurance for urban workers are far higher than that of the basic medical insurance for urban residents and the new rural cooperative medical insurance; the level of medical services utilization (prevalence rate, medical treatment rate, average cost per visit, etc.) of urban employees' basic medical insurance 21 participants is higher than that of other two social medical insurance participants [52]. Therefore, negative childhood experiences may affect the utilization of medical services and expand health inequality through occupational status.
Based on the data of the US census, Kitagawa and Hauser [53] found that the health level of low-income people was lower regardless of gender (male, female) or race (white, black). Lei & Lin [11] got a similar conclusion. He found that income level significantly affected people's health, low-income people have poor health, highincome people have good health, and there was a significant positive correlation between income level and health. Thus, the key factor that the income level can improve the health level is that patients with higher income have stronger ability to pay, then the more medical services are used, the higher the level of improving the health ability. However, negative childhood experiences reduced the income level of residents and further expanded the health inequality.

Conclusion
Regression results found that negative childhood experiences have a negative impact on an individual's health. Concentration index decomposition found that the contribution of negative childhood experiences to health inequality was 12.86%.
Through the mechanism test, negative childhood experiences may lead to lower socioeconomic status (fewer years of schooling, worse occupation and lower income), which was negatively related to medical consumption constraints. Further studies have shown a positive correlation between socioeconomic status and healthcare utilization, which can also cause negative health issues and more likelihood in health inequalities. 22 Research on health in adulthood has long emphasized recent risk factors. If the origin of adult diseases can be adequately addressed early in life, both scientific research and medical practice will benefit from this shift [30], as many social problems, including health inequality issues, could be prevented during this early period. The most effective and operative approach to avoid negative childhood issues may be family support aimed at prevention and early intervention, which can help families improve relationships and achieve a balance between work and family responsibilities, thus ensuring family stability and function.

CHARLS: China Health and Retirement Longitudinal Study;
CI: Concentration Index.

Declarations Ethics approval and consent to participate
The study was exempt from human subjects' approval (non-identifiable data; not human subjects).