We found a strong graded relationship between the ACE score and the risk of premature death of a family member. This relationship was consistent for each of four successive birth cohorts suggesting that the finding is not related to secular trends or to the amount of time a respondent had to "observe" premature death in the family. In addition, each of the individual categories of adverse childhood experiences was also associated with an increased risk for premature death of a family member. These findings are consistent with related studies that linked risks for premature death, such as heart diseases, substance abuse, and attempted suicide among individuals who reported these ACEs [9
The mechanisms for association observed between ACEs and premature death in the family merit serious consideration. First, the impact of ACEs on subsequent family member premature mortality may take place through the combined effects of social and biological risk occurring at different life stages, such as neurological development during childhood, and health behaviors inculcated during adolescence and adulthood [34
]. Thus, the entire family (i.e., members of all ages) would be affected. Moreover, family relationships may provide social control of health behaviors indirectly by affecting the internalization of norms for healthy behavior, and sanctions for deviating from behavior conducive to health [15
]. Multiple ACEs indicate a disordered, stressful social environment that can decrease, eliminate, or reverse these favorable impacts of social control. Family members living in such environments are more likely to have been exposed to persons engaging in undesirable health practices such as smoking, illicit substance use, excessive drinking, or unsafe sexual practices, which may have resulted from responses to traumatic stress (ACEs) [15
Second, ACEs are likely to contribute to an intergenerational cycle of risks [38
]. Evidence suggests that certain ACE related conditions may be "transmitted" intergenerationally. For example, early maternal and paternal age has been found among the second generation of teenage mothers [40
]. Several studies also reported association of mental illness between generations [41
]. Thus, persons with high ACE scores may be more likely to have parents, or even children with high ACE scores. If such is the case, the relationship of ACEs to many of the leading causes of death may account for the graded relationship of the ACE score to the premature death of a family member. In addition, other changes in family status that are associated with ACEs, such as a suicide or homicide of a family member or changes in residence [14
] may be precipitants of behaviors or physiologic changes that may result in early death.
Finally, some ACE-related risk factors appear to be interrelated in increasing the early death of the family member. For example, depressed people are more likely to commit suicide, illicit drug users were more likely to die a violent death, and alcoholics were more likely to be involved in fatal accidents. This was further evidenced by our findings that parental smoking, family member suffering a heart attack before age 60, being murdered, or having attempted suicide were strongly associated both with the number of ACEs and with the increased risk of family member premature death.
Although economic well-being is considered to be a basic explanation for mortality overall [33
] the adjustment for respondents' present or childhood socioeconomic position had little influence on the findings in either the current study on family member premature death or the previous investigation of other health risks for the leading causes of death [9
], rendering a simple explanation in terms of socioeconomic confounding less likely.
Several aspects of our study strengthen our findings. First, participants reported their family history during a routine clinical examination that preceded the survey about ACEs, reducing the possibility of bias toward attributing premature death of family member to ACEs. Second, population-based studies found levels of exposure to certain ACEs nearly identical to ours [46
], suggesting our findings are likely to be applicable in other study settings. Third, a wide range of interrelated ACEs was studied, enabling us to assess the relationship of each ACE to the risk of premature death.
The prevalence of childhood exposures we reported is nearly identical to those reported in population-based surveys in North America. We found that 16 percent of the men and 25% of the women met the case definition for childhood sexual abuse, similar to findings by Finkelhor et al. that 16% of men and 27%of women had been sexually abused [48
]. As for physical abuse, 28% of the men from our study reported experiencing this as boys, which closely parallels the percentage found (31%) in a recent population-based study of Ontario men that used questions from the same scales [49
]. The similarity of the estimates from the ACE study to those of population-based studies suggests that our findings are likely to be applicable in other settings.
The graded relationship of the ACE score to premature death of a family member is particularly strong for younger respondents and diminishes with increasing age of the respondents (Figure ). The decreasing strength of this relationship is likely due to the fact that the prevalence of premature death in the family increased with increasing age of the respondent (Table ), which was expected because older respondents have had more time to experience the premature death of a family member and because they have lived through periods of time wherein life expectancy was lower relative to the periods of life experience for younger respondents.
Data from the National Mortality Followback Survey indicate that reporting of the age of death of a family member is quite accurate. In fact, the agreement between age at death based on death certificate information and age at death based upon reports from next of kin show 92.7% agreement for ages within 1 year of actual age at death [50
Several potential limitations need to be considered when interpreting our results. First, our estimates of prevalence of family member premature death have limitations since certain information is unavailable regarding the family members, such as sex, age at death, employment, education, and differences in life circumstances. Second, since our question about family member premature death did not gather information regarding how many premature deaths occurred in a family, and our analysis treated any number of deaths as one single event, we could not estimate actual rates of premature death in the family; thus the prevalences we report almost certainly represent underestimates of the actual risk for premature death. Third, our findings were based on a retrospective survey which made it difficult to distinguish the causal impact of ACEs on premature mortality. In theory, prospective studies related to child maltreatment would avoid these potential biases. In practice, given the social and legal implications, this type of studies are nonetheless difficult to conduct [51
It is possible that ACEs pose increased risk not only for respondents' health but also the premature death of their family members. This possibility has significant policy implications and suggests that ACEs may play an important role as a marker for hidden risk factors existing within the family that place individuals at risk of early death. These findings may also improve our understanding of the etiology of certain diseases and thus help in the formation of future preventive initiatives. The problem of ACEs in our society is a complex one and to adequately address this problem will require a multifaceted solution that may ultimately lead to decompression of premature mortality. Advocacy for the needs of one of our most vulnerable population, children, and subsequent generations will, no doubt, become increasingly important.