Author Affiliations: Crimes Against Children Research Center, University of New Hampshire, Durham (Drs Finkelhor and Turner and Ms Shattuck); and Psychology Department, Sewanee, the University of the South, Sewanee, Tennessee (Dr Hamby).
Objective To test and improve upon the list of adverse childhood experiences from the Adverse Childhood Experiences (ACE) Study scale by examining the ability of a broader range to correlate with mental health symptoms.
Design Nationally representative sample of children and adolescents.
Setting and Participants Telephone interviews with a nationally representative sample of 2030 youth aged 10 to 17 years who were asked about lifetime adversities and current distress symptoms.
Main Outcome Measures Lifetime adversities and current distress symptoms.
Results The adversities from the original ACE scale items were associated with mental health symptoms among the participants, but the association was significantly improved (from R2 = 0.21 to R2 = 0.34) by removing some of the original ACE scale items and adding others in the domains of peer rejection, peer victimization, community violence exposure, school performance, and socioeconomic status.
Conclusions Our understanding of the most harmful childhood adversities is still incomplete because of complex interrelationships among them, but we know enough to proceed to interventional studies to determine whether prevention and remediation can improve long-term outcomes.
The Adverse Childhood Experiences (ACE) Study1 has attracted considerable scientific and policy attention in recent years, in part because it suggests that potentially preventable childhood experiences, particularly physical and sexual abuse and neglect, may increase a person's risk for serious health problems and higher mortality rates much later in life.
The study has demonstrated relationships between adverse childhood experiences and many adult health risks.1- 10 These results, which have been published widely in the health sciences, are based on a survey and medical records of more than 17 000 members of the Kaiser Health Plan in San Diego, California.1,11
Nonetheless, research using the ACE Study model has some important limitations, in part because of the retrospective way in which data on childhood adversities have been gathered. The average age of respondents when they supplied information about their childhood experiences was 55 to 57 years. As a result, it is hard to be certain, particularly from such a remote vantage, whether it is these particular childhood experiences or unmeasured covariates that are the most important predictors. In addition, the ACE Study list of preventable childhood adversities omits certain domains judged by many developmental researchers to be important in predicting long-term health and well-being outcomes. Among the predictors missing from the ACE Study model are peer rejection, exposure to violence outside the family, low socioeconomic status, and poor academic performance.
For example, longitudinal studies show that growing up in poverty increases lifelong risk for various negative life events and negative health outcomes.12- 14 Peer rejection and lack of friends are associated with the development of many disorders.15- 17 Poor school performance in childhood is associated with poor outcomes in adulthood, such as unemployment.18 Witnessing community violence has been shown to be a mental health hazard for adults and children.19,20 These major childhood adversities are not currently measured by the ACE scale.
In addition, measuring childhood adversities during childhood, rather than later, may offer other improvements to the ACE Study's early life predictors of health outcomes.21 It allows the possibility of obtaining a more accurate and comprehensive assessment of childhood events than one would be able to obtain after many years. It also would allow a more sensitive untangling of the relationship among various adversities in ways that better explain causal sequences.
Although an obvious disadvantage is the inability to assess the long-term effects of childhood adversity on the negative life events and health conditions posited in the ACE Study model, examining more short-term effects in childhood is consistent with the logic of the model. Specifically, the ACE Study model relies strongly on the idea that adverse childhood experiences create a burden of psychological stress that changes behavior, cognitions, emotions, and physical functions in ways that promote subsequent health problems and illness.22 Among the hypothesized pathways, adverse childhood experiences lead to depression and posttraumatic stress disorder, which in turn can lead to substance abuse, sleep disorders, inactivity, immunosuppression, inflammatory responses, and inconsistent health care use, possibly leading to other medical conditions later in life.23,24 Therefore, childhood behavioral and emotional symptoms very likely represent a crucial mediator linking adverse childhood experiences and the longer term health-related problems found in the ACE substudies.
Thus, in the present study, we tried to replicate the ACE Study findings in a cohort of youth, using psychological distress as an outcome measure, and to explore whether the adversities enumerated by the ACE Study could be improved upon by considering a more comprehensive range of possible adversities, including some of the domains not considered in the ACE Study.
These analyses use data from the National Survey of Children's Exposure to Violence (NatSCEV),25 a representative sample of US children and adolescents. The NatSCEV was designed to obtain incidence and prevalence estimates for a wide range of childhood victimizations and other adversities. The survey was conducted between January 2008 and May 2008 with a nationally representative sample of 4549 children aged 0 to 17 years living in the contiguous United States. Interviews with parents and youth were conducted over the telephone by the employees of an experienced survey research firm.
The foundation of the design was a nationwide sampling frame of residential telephone numbers from which a sample of telephone households was drawn by random digit dialing. This nationally representative cross section yielded 3053 of the 4549 completed interviews. To ensure that the study included a sizable proportion of racial/ethnic minorities and low-income respondents for more accurate subgroup analyses, there was also an oversampling of US telephone exchanges that had a population of 70% or more of African American, Hispanic, or low-income households. This oversample yielded the remaining 1496 of the completed interviews. Sample weights were calculated to adjust for differential probability of selection associated with (1) study design, (2) demographic variations in nonresponse, and (3) variations in within-household eligibility. For this study, we analyzed a subsample of the entire sample of 4549 respondents. This subsample consisted of 2030 youth who were aged 10 to 17 years at the time of the interview and for whom complete data were available on the variables of interest. Analyses in this study are weighted by the sample weights.
A short interview was conducted with an adult caregiver (usually a parent) in each household to obtain family demographic information. One child was randomly selected from all eligible children living in a household by choosing the child with the most recent birthday. If the selected child was aged 10 to 17 years, the main telephone interview was conducted with the child. If the child was younger than 10 years, the interview was completed with the caregiver. However, the current analysis is based only on the 2030 youth aged 10 to 17 years who provided self-report information. Respondents were paid $20 for their participation. The interviews, averaging 45 minutes in both waves, were conducted in either English or Spanish. All procedures were approved by the institutional review board at the University of New Hampshire.
The cooperation rate for the random digit dialing cross-section portion of the survey was 71%, and the response rate was 54%. The cooperation and response rates associated with the smaller oversample were somewhat lower at 63% and 43%, respectively. These are good rates by current survey research standards.26- 30 Although the potential for response bias remains an important consideration, several recent studies and our own analysis25 have shown no meaningful association between response rates and response bias.31- 34
This survey used an enhanced version of the Juvenile Victimization Questionnaire, an inventory of childhood victimization.35- 37 The Juvenile Victimization Questionnaire obtains reports on 48 forms of youth victimization covering 5 general areas of interest: conventional crime, maltreatment, victimization by peer and siblings, sexual victimization, and witnessing and exposure to violence.38 The survey also contains questions about adverse life events in the parent interview section and in a separate section on adversity.
For the present study, which was not originally designed to test the ACE Study model, we selected victimization and adversity items in 2 steps. First, we used screener items and their associated follow-up questions to construct victimization types that most closely matched the abuse and neglect items in the original ACE Study, and we chose family background and adversity items to match the household dysfunction items of the original ACE Study. Using these items, we constructed a replication of the original ACE Study. In the second step, we selected additional types of victimization and adversity items not included in the original ACE Study but that are known to be important correlates of health and well-being outcomes. The measures selected in these 2 steps are described in the next section of this article. Important differences from the ACE Study items are noted in eTable 1.
The following measures were coded 0 for no and 1 for yes so that they could be summed to create the replicated ACE Study items. All are lifetime measures.
Emotional abuse: One item asked respondents, “At any time in your life, did you get scared or feel really bad because grown-ups in your life called you names, said mean things to you, or said they didn't want you?”
Physical abuse: Several screeners assessed the child's experience of physical assault. Children who answered yes to any of these assault screeners were coded as having experienced physical abuse if the incident was perpetrated by parent, an adult relative, or another adult caregiver.
Sexual abuse: Four screeners asked about the child's experience of sexual assault or attempted rape by a known adult, an adult stranger, or a peer or sibling.
Emotional neglect: Four questions about family social support were used to construct an indicator of emotional neglect. These items are shown in eTable 1. Total scores ranged from 4 to 16. Children whose family support score was 10 or lower were coded as having experienced emotional neglect.
Physical neglect: A single item asked whether the child had ever experienced a time when adults in his or her life “didn't take care of them the way they should,” including not providing enough food, not taking them to the doctor when they were sick, or not making sure they had a safe place to stay. Children who answered yes were coded as having experienced physical neglect.
Mother treated violently: Twelve screeners asked children whether they had witnessed specific kinds of violence and abuse. Children who answered yes to any of these questions and who reported that their mother was the victim were coded 1 on this item.
Household substance abuse: A single item assessed whether the child had a family member who “drank or used drugs so often that it caused problems.”
Household mental illness: Children who had a parent or sibling with depression, bipolar disorder, anxiety, or “other psychiatric disorder” (information obtained from the parent interview) or children who had “someone close” attempt suicide were coded 1 on household mental illness.
Parental separation or divorce: We coded any respondent who was not currently living with 2 biological or adoptive parents as having experienced parental separation or divorce.
Incarcerated household member: One adversity item asks whether a parent or guardian had ever been sent to prison.
The measures listed herein, not included in the ACE Study, were examined as additional correlates of children's distress. A summary of these items is reported in eTable 2. Unless otherwise specified, questions regarding these items were asked in the child's portion of the interview:
Peer victimization (assault, physical intimidation, or emotional victimization by a nonsibling peer)
Parents always arguing (respondents were asked whether there was a time in their lives when their parents were always arguing)
Property victimization (experience of a robbery, theft, or vandalism by a nonsibling perpetrator)
Someone close to the child had a bad accident or illness
Exposure to community violence (6 screeners asked whether the child had been exposed to certain types of crime and violence, including witnessing an assault, experiencing a household theft, having someone close murdered, witnessing a murder, experiencing a riot, or being in a war zone)
No good friends (child had no “really good friends at school” at the time of the interview)
Below-average grades (parent reported that the child had “below-average” grades in school)
Someone close to the child died because of an accident or illness
Parent lost job (children reported that there was a time when their “mother, father, or guardian lost a job or couldn't find work”)
Parent deployed to war zone (parent had to leave the country to fight in a war and was gone for several months or longer)
Disaster (child had experienced a “very bad fire, flood, tornado, hurricane, earthquake, or other disaster”)
Removed from family (child was “sent or taken away from his or her family for any reason”)
Very overweight (parent reported that the child was “quite a bit overweight” compared with other boys/girls his or her age)
Physical disability (parent reported that the child had been diagnosed with a “physical health or medical problem that affects the kinds of activities that he or she can do”)
Ever involved in a bad accident
Neighborhood violence is a “big problem” (asked in the parent interview)
Homelessness (a time when the child's family “had to live on a street or in a shelter because they had no other place to stay”)
Repeated a grade
Less masculine or feminine than other boys or girls his or her age (asked in the parent interview)
Distress symptoms were measured using shortened versions of the anger, depression, anxiety, dissociation, and posttraumatic stress scales of the Trauma Symptoms Checklist for Children (TSCC).39 Respondents were asked how often they had experienced each symptom within the past month. Response options were on a 4-point scale from 1 (not at all) to 4 (very often), and responses from the items of all 5 scales were summed to create a total distress score consisting of 28 items. The Cronbach α value for total distress score in this study was 0.93.
Demographic information was obtained in the initial parent interview, including the child's sex, age (in years), race/ethnicity (coded into 4 groups: white non-Hispanic, black non-Hispanic, other non-Hispanic, and Hispanic any race), socioeconomic status (SES), and place size of the child's town or city of residence. Socioeconomic status is a continuous composite score based on the sum of the standardized household income and standardized parental educational level (for the parent with the highest educational level) scores, which was then restandardized. For our revised version of the ACE scale, we created a dummy indicator for low SES that flags children whose continuous SES value fell in the bottom, roughly 20%.
The ACE scale constructed with variables from NatSCEV that mimic the original items is associated with distress levels among youth aged 10 to 17 years, as measured by the Trauma Symptom Checklist for Children. Model 1 in Table 1 reports the regression of distress scores on the items from the replicated ACE scale. The cumulative items were strongly associated with distress, and there was a clear dose-response relationship between the adversities and distress, as has been demonstrated in previous research.1
However, the original ACE scale items did not each make an independent contribution to distress as illustrated in model 1 of Table 1. Two items, parental separation or divorce and incarceration of a household member, were not significant in the regression model of the whole scale. In addition, when other childhood adversities (not considered in the ACE studies) were added to the model (model 2 of Table 1), several ACE scale items dropped below significance. Moreover, several of the added childhood adversities showed strong associations with distress. These included peer victimization, property victimization, parents always arguing, having no good friends, having someone close with a bad illness or accident, SES, and exposure to community violence.
A revised ACE scale was then constructed, removing the original items that were no longer significant in the extended model. Significant new items were added to the scale, including parents always arguing, having no good friends, having someone close with a bad illness or accident, peer victimization, property victimization, and exposure to community violence. The old and new scales are contrasted in Table 2. Regression with the new scale determined R2 = 0.34 vs R2 = 0.21 for the original version of the scale.
In this study, it was possible to improve the value of the original ACE scale considerably by adding some childhood adversities not included in the original scale and excluding others that were in the scale. The value of adding several items not considered in the ACE studies is consistent with several publications showing their harmful effect on child development. In fact, there are likely even more domains of childhood adversity that might be measured and added that could further improve its predictive ability, for example, low IQ,40 parental death, and food scarcity. The present study illustrates that the original ACE scale could likely be improved even more with additional developmental research.
However, this analysis also confirms that some of the key ACE scale items, particularly the child maltreatment exposures, remain very important and make discrete independent contributions, even when many other adversities are considered. Moreover, several of the new adversities identified in this study are additional forms of interpersonal victimization—property crime, peer victimization, and exposure to community violence—which reinforce findings from other studies41,42 highlighting the cumulative harm of different forms of childhood victimization.
There are several limitations of the current study that bear emphasis. First, this study did not operationalize the adverse childhood events in the same way that the original ACE instrument did. Second, the dependent variable, the TSCC, used in this exercise was not an outcome used in the original ACE Study. The TSCC may be better associated with the impact of some childhood events, such as violence exposure, than others and may not necessarily be reflective of what would best predict long-term health effects. In fact, some childhood adversities may affect later health not through psychological processes, such as distress symptoms, but through other mechanisms, for example, failure to receive proper early health care. Moreover, unlike the ACE Study, the outcome measure was short term and the causal sequence between adversities and outcome cannot be assumed. All the variables in this study come from self-report and, in most cases, from children, which may be inaccurate and introduce method associations.
Before additional work on the ACE scale is undertaken, some important issues are worth discussing, even beyond the findings of the current study. One issue concerns what the goal or best use of this or related scales should be. One possible use for this kind of scale is as a risk assessment tool with older adolescents or adults to help health care providers better understand who is most likely to require services and treatment for health problems. However, the goal for which the scale has been most widely used to date is to advocate for and influence prevention policies by highlighting crucial developmental factors that prevention programs should target to improve general health and reduce medical costs and social service expenditures.22,43,44
In many ways the first goal, risk assessment, is a much easier one to accomplish than the second, selection of prevention targets. To successfully satisfy the first goal, research has to find strong associations between risk indicators and later outcomes. The ACE scale seems clearly successful at this. For the second goal, however, a good risk indicator is not sufficient. The indicator has to be a proven causal contributor, which modified would make a difference. Much of the discussion about the ACE scale assumes that its items are causal contributors to the numerous negative adult outcomes, but this may not be the case.
Without detailed longitudinal studies and the measurement of many additional variables, it may be very difficult to tease out whether, for example, it is household substance abuse that affects later outcomes or some unmeasured underlying parental emotional problem or lack of self-control. Moreover, a very important, but difficult to test, alternative explanation for many of the ACE Study findings is that inherited genes for health problems or some temperamental qualities create a spurious connection between abuse and neglect by parents or other family context variables and mental and physical health conditions in their offspring. If this were to be the case, it is possible, although not likely, that even preventing child abuse would make modest differences on health outcomes.
There are other problems with using an ACE scale even as a long-term risk assessment tool. One is that risk assessment has to factor in social changes regarding the frequency, norms, and impact of different experiences. For older respondents who answered the original ACE Study questionnaire, parental divorce may have been an unusual and stigmatizing event and sexual abuse a hidden experience that one never talked or heard anything about. Among a younger cohort, more cultural awareness and the increased availability of support, including professional intervention, may mean that the experience of sexual abuse or parental divorce might have different consequences. This may be why parental divorce was not a significant predictor in the current study.
Another problem is the possibility of reverse causation in which bad later life outcomes induce reports of more negative early childhood experiences. There is some evidence that people recall more negative historical adversity when they have poor adult outcomes, mental health, and physical problems.45 To the degree that this is true, variables identified in later life, such as in the ACE Study, will not prove as predictive of ultimate health outcomes when assessed in earlier life stages.
An additional philosophical problem worth considering in discussions about the implications of ACE-type research is whether advocates should use a list of childhood features that are associated with long-term health effects as the primary criterion of what childhood adversities to prioritize for prevention. For example, if sexual abuse were demonstrated to be minimally associated with long-term health effects, would that disqualify it as a priority for primary prevention? No. Many childhood adversities are candidates for prevention not because they create long-term health risks but because they violate the rights of children or cause pain and suffering at the moment. Their contributions to long-term health can be additional evidence to consider but may not be primary. Such adversities illustrate the tension between a utilitarian and human rights perspective in child welfare policy.
This research suggests that the goal of identifying childhood adversities that are precursors to long-term health and behavioral outcomes may be improved by considering a wider range of adversities measured in a more contemporaneous way. Such an approach might be well advanced by using longitudinal studies that have been monitoring children into adulthood.12
However, more discussion is needed about the goals and usefulness of such efforts. Although additional efforts to refine an adverse childhood experience checklist that predicts later health outcomes has scientific merit, an argument can be made that enough is known about certain harmful childhood experiences22 that more testing of parts of this model should be carried out through experiment rather than correlation. There is enough consensus that exposure to violence, sexual abuse, and emotional mistreatment are harmful and likely have long-term health effects; therefore, the next generation of studies should probably focus on preventing and remediating these exposures and following up to determine whether health outcomes improve.
Correspondence: David Finkelhor, PhD, Crimes Against Children Research Center, University of New Hampshire, 126 Horton Social Science Center, 20 Academic Way, Durham, NH 03824 (email@example.com).
Accepted for Publication: June 7, 2012.
Published Online: November 26, 2012. doi:10.1001/jamapediatrics.2013.420
Author Contributions:Study concept and design: Finkelhor, Turner, and Hamby. Analysis and interpretation of data: Finkelhor, Shattuck, Turner, and Hamby. Drafting of the manuscript: Finkelhor and Shattuck. Critical revision of the manuscript for important intellectual content: Finkelhor, Turner, and Hamby. Statistical analysis: Shattuck. Obtained funding: Finkelhor and Turner. Administrative, technical, and material support: Finkelhor and Turner. Study supervision: Finkelhor.
Conflict of Interest Disclosures: None reported.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Pediatrics editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Finkelhor D, Shattuck A, Turner H, Hamby S. Improving the Adverse Childhood Experiences Study scale. JAMA Pediatr.. Published online November 26, 2012. doi:10.1001/jamapediatrics.2013.420
eTable 1. Items Included in Replicated ACE Scale
eTable 2. Additional Victimization and Adversity Items
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 8
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.