Assessment of Additional Medical Costs Among Older Adults in Japan With a History of Childhood Maltreatment

Key Points Question Are medical costs higher among older adults with a history of childhood maltreatment than among those without? Findings In this cross-sectional study of 978 adults, using national health insurance claims data in Japan, mean annual medical costs among those who experienced any childhood maltreatment were significantly higher than costs among those who did not, even after adjusting for age and sex. Meaning In this study, childhood maltreatment was associated with additional medical costs among older adults living in Japan, which highlights the importance of primary and secondary prevention of child maltreatment.

Furthermore, recent research using a cohort study of independent adults older than 65 years in Japan 25,26 has shown that childhood maltreatment can lead to deteriorating health in late adulthood, including greater higher-level functional limitations and greater risk of diseases such as diabetes and hypertension.
Therefore, experiencing childhood maltreatment may lead to additional health care costs in later life. 27,28Previous studies in the United States using health care system data 29,30 have shown that childhood sexual abuse and/or physical abuse were associated with an increase in annual health care costs among women aged 18 to 65 years after controlling for demographic characteristics.
Similar results were obtained in a Canadian study using ambulatory self-reported health care costs, 31 although the amount of annual health care costs was smaller than in studies that used health care system data.Notably, these previous studies exclusively targeted women who were younger than 65 years.To our knowledge, the long-term association of childhood maltreatment with health care costs among individuals aged 65 years or older, including men, has not been examined.Moreover, no previous studies on health care costs have included the associations of childhood emotional abuse and neglect with health care costs.
In Japan, the social costs of childhood maltreatment in 2012 were reported to be ¥1.6 trillion (US$14.7 billion), including direct and indirect costs. 32However, the actual medical costs of those who experienced childhood maltreatment are still unknown because medical costs were estimated indirectly in the study.As the Japanese population is aging, elucidating the association of childhood maltreatment with health care costs among older adults is critical.
This study aimed to assess additional medical costs associated with childhood maltreatment, using the database of health insurance claims that is linked with data from the Japan Gerontological Evaluation Study (JAGES), 2013, a population-based cohort of independent adults aged 65 years or older across Japan.We also calculated additional annual medical costs for different types of childhood maltreatment (ie, witnessing intimate partner violence [IPV] and experiencing physical abuse, emotional neglect, and emotional abuse).

Data and Sample
Our sample was based on participants of JAGES 2013, which was designed to investigate the social determinants of health among noninstitutionalized, functionally independent individuals aged 65 years or older.Started in 2003, JAGES is the largest cohort study of older adults in Japan, with samples from 30 municipalities in 15 prefectures.Data from JAGES 2013 were linked with the database of national health insurance claims for a city with more than 1.5 million residents, of whom 7257 were aged 75 years or younger in 2013 (ie, those who were Յ74 years in fiscal year [FY] 2012 [ie, April 2012 to March 2013]).Individuals with employer-sponsored health insurance or those who receive public assistance are not eligible for national health insurance.The database of health well care visits for those with national health insurance was used to distinguish between individuals who did not use national health insurance during FY2012 and FY2013 (66 individuals) and individuals who may not have been eligible for national health insurance.Those who may not have been eligible for national health insurance, including those who have employer-sponsored health insurance and those who receive public assistance, were excluded from the sample (2343 in FY2012; 2325 in FY2013).Medical and pharmacy claims of 5155 individuals were available for either FY2012 or FY2013.A sample of 978 individuals was used for this analysis because a randomly selected fifth of the JAGES 2013 sample was asked questions regarding adverse childhood experiences.An additional 34 participants who did not answer all 4 questions regarding childhood maltreatment were also excluded (Figure).This study was approved by the institutional review board at Chiba University in Chiba, Japan, and informed consent was waived because this study used secondary data.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Measurement Medical Costs
Mean annual medical costs, including publicly funded health care costs, for FY2012 and FY2013 were calculated from medical fee points (1 point indicates ¥10, ie, the unit of cost is ¥10), which were recorded in the database of national health insurance claims for city residents.These medical costs included outpatient, inpatient, and pharmaceutical costs but not dental care costs.When medical costs were available for both FY2012 and FY2013, the mean annual medical cost was used, and if only 1 FY was available, the available annual data were used as the mean annual cost for the 2 FYs.

History of Childhood Maltreatment
History of maltreatment before the age of 18 years was assessed as part of the adverse childhood experiences questions in the JAGES 2013 survey, based on previous studies 1 and modified to suit to older adults living in Japan. 25Questions regarding the 4 following types of childhood maltreatment were asked: (1) witnessing IPV, (2) experiencing physical abuse, (3) experiencing emotional neglect, and (4) experiencing emotional abuse.A participant was classified as having witnessed IPV if they answered yes to the statement, "Your father was violent with your mother."Witnessing IPV was included as a type of childhood maltreatment in this study because IPV in front of children has been defined as child maltreatment in Japan since the Child Abuse Prevention Act was passed in 2004.A participant was classified as having experienced physical abuse if they answered yes to the statement, "(You) were hit hard by your mother/father, causing an injury."A participant was classified as having experienced emotional neglect if they answered no to the statement "(You) felt loved by your parents."A participant was classified as having experienced emotional abuse if they answered yes to the statement "(You) were told hurtful things or were insulted by your mother/father."We aggregated the responses for any type of child maltreatment.In addition, the total number of maltreatment types experienced in childhood was used to assess the cumulative association of multiple types of child maltreatment with health care costs.

Statistical Analysis
Mean annual medical costs were calculated and compared using t tests between those who experienced any type of childhood maltreatment and those who experienced none.Considering that the distribution of medical costs was highly skewed (eFigure in the Supplement), a generalized linear model (GLM) was used to examine the association of childhood maltreatment history with medical costs.Model 1 was adjusted for age and sex because these characteristics were assumed to be preexposure variables that would be associated with a history of childhood maltreatment and medical costs in late adulthood (model 1). 33,34We used a χ 2 test to determine which self-reported diseases were associated with a history of childhood maltreatment (eTable 1 in the Supplement).
These diseases were added in model 2 to explore whether medical history mediated the association of childhood maltreatment and additional medical costs.In both GLMs, log-link function and gamma distribution for the mean-variance association were specified according to the Box-Cox test and modified Park test. 35The marginal effects of the exposure variable and covariates in Japanese yen were computed from the GLMs.
We conducted t tests and GLMs for each type of childhood maltreatment and the total number of childhood maltreatment experiences.Model 2 in the GLM was only performed on types of childhood maltreatment that were found to be significant in t tests.The GLMs were conducted with complete data (ie, missing values for each type of childhood maltreatment and covariates were excluded); the largest percentage of missing values was 1.6%, for emotional neglect.Statistical significance was set at P < .05,and all tests were 2-tailed.All analyses were conducted using Stata version 14.1 (StataCorp) from October 2017 to February 2019.

Results
Table  in our sample (ie, 18.0%), and the total population of individuals aged 65 to 74 years in FY2012 to FY2013 in Japan (ie, 16 million), 38,39 we can estimate that the additional annual medical costs associated with childhood maltreatment were more than ¥333 billion (US$3.1 billion) (95% CI, ¥32

Discussion
To our knowledge, this is the first study that examined additional annual medical costs associated with childhood maltreatment among independent older adults aged 65 to 74 years.Using the database of health insurance claims that is linked with data from a population-based cohort study in Japan, we found that mean annual medical costs among those with any history of childhood maltreatment were significantly higher than among those with no history of maltreatment.
Older adults with a history of childhood physical abuse had ¥295 148 higher medical costs than those without; however, the increase was not statistically significant when their age and sex were considered.This was not consistent with a previous study, 30 which reported that total annual health care costs calculated from health care system records were significantly higher among women with a history of physical abuse compared with women without, even after adjusting for calendar year, age, and education.However, the previous study only included women aged 18 to 64 years, while the target population of our study was men and women aged 65 to 74 years.Although we had smaller sample size, our study examined longer-term consequences of physical abuse among both women and men than previous studies.Findings on the other types of maltreatment cannot be compared directly with previous studies because, to our knowledge, no studies have investigated the long-term associations of witnessing IPV or experiencing emotional neglect and abuse in childhood with medical costs.
Having more than 1 type of childhood maltreatment or adverse childhood experience has been reported to increase the risk of health problems 1,4,18,20,[40][41][42][43] and health care costs later in life. 28,30,31wever, our study found no significant difference in the long-term medical costs between those with more than 2 types of maltreatment and those with no history of child maltreatment.This can be

JAMA Network Open | Public Health
Additional Medical Costs Among Older Adults in Japan With a History of Childhood Maltreatment explained by a selective survival effect, 44,45 ie, those with a history of multiple childhood maltreatment types were more likely to have been excluded from our sample because they may have died or become functionally dependent before they reached the target age of our study (ie, 65-74 years), given that having more adverse childhood experiences has been found to be associated with premature mortality. 25,46,47r study also explored whether disease history mediated the association of childhood maltreatment with medical costs in older adulthood.Our findings are consistent with previous studies, in that the long-term association of child maltreatment with health care costs seemed to be mediated by physical and mental health concerns. 31The magnitude of the association of childhood maltreatment with medical costs was largely reduced when diseases associated with childhood maltreatment were added to our analysis model.Furthermore, our study implied that childhood physical abuse was associated with an increased risk of stroke and cancer, while emotional neglect was associated with an increased risk of kidney or prostate gland disease, which may have led to additional medical costs among independent older adults aged 65 to 74 years.This finding is in line with a previous study using the 2012 Canadian Community Health Survey data that indicated childhood physical abuse was associated with stroke, cancer, and other diseases. 48Emotional neglect has also been found to be associated with an increased risk of such diseases, 49 but its association with kidney or prostate gland diseases had not been found in previous studies.Because there are a limited number of studies that examine the association of specific types of child maltreatment, especially emotional neglect, with adult diseases and different measures of diseases were used in these studies, future research using objective measures of diseases, including a wide range of diseases associated with child maltreatment, such as substance use disorders and mood disorders, is warranted to elucidate what kinds of diseases could explain the association of childhood maltreatment with increased medical costs in later life.
Furthermore, we estimated that additional medical costs associated with childhood maltreatment were more than ¥333 billion (US$3.1 billion) per year nationwide, which is approximately 0.9% of total annual medical spending (ie, the sum of outpatient, inpatient, and pharmaceutical costs) in Japan in FY2012 and FY2013 and 8.4% of total annual medical spending among those aged 65 to 74 years in FY2012 and FY2013. 36,37These figures suggest tremendous long-term consequences of childhood maltreatment, considering that this estimate is only for those aged 65 to 74 years who are functionally independent.Additionally, the estimate from this study is more than twice that of indirect medical costs (¥136 billion [US$1.3billion]) estimated in the previous study in Japan. 32Our findings could contribute to more accurate estimations of the costs of child maltreatment using real health insurance claims data in Japan.

Limitations
There are several limitations of this study.First, we might have overexcluded those who did not use national health insurance (ie, those whose medical costs were ¥0) because we could not completely distinguish between participants who were eligible for national health insurance but did not use it and those who were not eligible for national health insurance.However, a comparison of demographic information between those whose medical costs in both FY2012 and FY2013 were not available and those whose medical costs were available implies that those who were excluded from the sample seemed likely to be ineligible for national health insurance (eTable 2 in the Supplement).
In addition, the prevalence of childhood maltreatment was not significantly different between these groups (eTable 2 in the Supplement), and similar results on the association between childhood maltreatment and medical costs were obtained from GLM analysis including those whose medical costs were ¥0 (data not shown).Second, retrospective self-reports of childhood maltreatment were used in our study, which can be subject to recall bias. 50,51Given that another study 52 suggested retrospective measures are more likely to produce false-negatives than false-positives, our findings on the association of childhood maltreatment with medical costs may be underestimated.Third, childhood sexual abuse was not assessed in our study because of the feasibility of the survey in the

JAMA Network Open | Public Health
Additional Medical Costs Among Older Adults in Japan With a History of Childhood Maltreatment cultural context of Japan.However, a population-based study in Japan reported that the prevalence of childhood sexual abuse was quite low (ie, 0.5%), 53 which contributed little to the underestimation of prevalence of childhood maltreatment.Furthermore, even if childhood sexual abuse were queried in surveys, previous studies in Japan 54,55 have reported low response rates (ie, 19.1% and 25.6%, respectively).9][30][31] Fourth, survival selection bias could underestimate the association of childhood maltreatment with medical costs.Fifth, the generalizability of our findings may be limited because our study only included independent adults aged 65 to 74 years in a particular city.When available, future research should include such older adults to better understand the long-term association of childhood maltreatment with medical costs using a representative sample.

Conclusions
Our study provided evidence that history of childhood maltreatment, particularly emotional neglect, was associated with increases in medical costs among older men and women.The findings demonstrated significant long-term consequences of childhood maltreatment and underlined the importance of primary prevention (ie, home visitation programs during pregnancy for high-risk populations) and early intervention (ie, screening of adverse childhood events and intervention in pediatric clinics) for child maltreatment.

Figure .
Figure.Flowchart of Study Participants

Table 1 .
Characteristics of the 978 Participants (continued)

Table 3 .
Unadjusted Annual Medical Costs With and Without History of Child Maltreatment, by Type of Childhood Maltreatment

Table 4 )
. Based on these values, the prevalence of any childhood maltreatment

Table 4 .
Average Marginal Effects of Childhood Maltreatment on Annual Medical Costs Abbreviations: AME, average marginal effect; IPV, intimate partner violence; NA, not applicable.
billion to ¥637 billion [US$294.4 million to US$5.9 billion]) nationwide.When kidney or prostate gland disease and musculoskeletal disease were added in model 2, any childhood maltreatment was not associated with annual medical costs (AME, ¥101 214 [US$931]; SE, ¥53 961 [US$496]; 95% CI, Finally, the association of total number of childhood maltreatment types with mean annual medical costs among older adults was also examined in the adjusted model.Those who experienced 1 type of childhood maltreatment had significantly higher total annual medical costs than those who experienced none (AME, ¥122 173 [US$1124]; SE, ¥58 373 [US$537]; 95% CI, ¥7763-¥236 582 [US$71-US$2177]; P = .04).However, mean annual medical costs among those who experienced more than 2 types of childhood maltreatment did not significantly differ from medical costs among those who experienced none.