Research Article

Relationship of Childhood Abuse and

Household Dysfunction to Many of the
Leading Causes of Death in Adults
The Adverse Childhood Experiences (ACE) Study
Vincent J. Felitti, MD, FACP, Robert F. Anda, MD, MS, Dale Nordenberg, MD, David F. Williamson, MS, PhD,
Alison M. Spitz, MS, MPH, Valerie Edwards, BA, Mary P. Koss, PhD, James S. Marks, MD, MPH

Background: The relationship of health risk behavior and disease in adulthood to the breadth of
exposure to childhood emotional, physical, or sexual abuse, and household dysfunction
during childhood has not previously been described.
Methods: A questionnaire about adverse childhood experiences was mailed to 13,494 adults who had
completed a standardized medical evaluation at a large HMO; 9,508 (70.5%) responded.
Seven categories of adverse childhood experiences were studied: psychological, physical, or
sexual abuse; violence against mother; or living with household members who were
substance abusers, mentally ill or suicidal, or ever imprisoned. The number of categories
of these adverse childhood experiences was then compared to measures of adult risk
behavior, health status, and disease. Logistic regression was used to adjust for effects of
demographic factors on the association between the cumulative number of categories of
childhood exposures (range: 0 –7) and risk factors for the leading causes of death in adult
life.
Results: More than half of respondents reported at least one, and one-fourth reported $2
categories of childhood exposures. We found a graded relationship between the number
of categories of childhood exposure and each of the adult health risk behaviors and
diseases that were studied (P , .001). Persons who had experienced four or more
categories of childhood exposure, compared to those who had experienced none, had 4-
to 12-fold increased health risks for alcoholism, drug abuse, depression, and suicide
attempt; a 2- to 4-fold increase in smoking, poor self-rated health, $50 sexual intercourse
partners, and sexually transmitted disease; and a 1.4- to 1.6-fold increase in physical
inactivity and severe obesity. The number of categories of adverse childhood exposures
showed a graded relationship to the presence of adult diseases including ischemic heart
disease, cancer, chronic lung disease, skeletal fractures, and liver disease. The seven
categories of adverse childhood experiences were strongly interrelated and persons with
multiple categories of childhood exposure were likely to have multiple health risk factors
later in life.
Conclusions: We found a strong graded relationship between the breadth of exposure to abuse or
household dysfunction during childhood and multiple risk factors for several of the
leading causes of death in adults.
Medical Subject Headings (MeSH): child abuse, sexual, domestic violence, spouse abuse,
children of impaired parents, substance abuse, alcoholism, smoking, obesity, physical
activity, depression, suicide, sexual behavior, sexually transmitted diseases, chronic obstruc-
tive pulmonary disease, ischemic heart disease. (Am J Prev Med 1998;14:245–258) © 1998
American Journal of Preventive Medicine

Department of Preventive Medicine, Southern California Perma- denberg) Atlanta, Georgia 30333. Department of Family and Com-
nente Medical Group (Kaiser Permanente), (Felitti) San Diego, munity Medicine, University of Arizona Health Sciences Center,
California 92111. National Center for Chronic Disease Prevention (Koss) Tucson, Arizona 85727.
and Health Promotion, Centers for Disease Control and Prevention, Address correspondence to: Vincent J. Felitti, MD, Kaiser Perma-
(Anda, Williamson, Spitz, Edwards, Marks) Atlanta, Georgia 30333. nente, Department of Preventive Medicine, 7060 Clairemont Mesa
Department of Pediatrics, Emory University School Medicine, (Nor-- Boulevard, San Diego, California 92111.

Am J Prev Med 1998;14(4) 0749-3797/98/$19.00 245

© 1998 American Journal of Preventive Medicine PII S0749-3797(98)00017-8
Introduction underlie many of the leading causes of death in
adults.6,7,36,37

O
nly recently have medical investigators in pri-
mary care settings begun to examine associa-
tions between childhood abuse and adult Methods
health risk behaviors and disease.1–5 These associations Study Setting
are important because it is now clear that the leading
The ACE Study is based at Kaiser Permanente’s San
causes of morbidity and mortality in the United States6
Diego Health Appraisal Clinic. More than 45,000 adults
are related to health behaviors and lifestyle factors;
undergo standardized examinations there each year,
these factors have been called the “actual” causes of
7 making this clinic one of the nation’s largest free-
death. Insofar as abuse and other potentially
standing medical evaluation centers. All en-
damaging childhood experiences contribute
rollees in the Kaiser Health Plan in San
to the development of these risk factors, then See
Diego are advised through sales literature
these childhood exposures should be recog- related
about the services (free for members) at the
nized as the basic causes of morbidity and Commentary clinic; after enrollment, members are ad-
mortality in adult life. on pages 354, vised again of its availability through new-
Although sociologists and psychologists 356, 361. member literature. Most members obtain
have published numerous articles about the
appointments by self-referral; 20% are re-
frequency8 –12 and long-term consequenc-
ferred by their health care provider. A recent review of
es13–15 of childhood abuse, understanding their rele-
membership and utilization records among Kaiser
vance to adult medical problems is rudimentary. Fur-
members in San Diego continuously enrolled between
thermore, medical research in this field has limited 1992 and 1995 showed that 81% of those 25 years and
relevance to most primary care physicians because it is older had been evaluated in the Health Appraisal
focused on adolescent health,16 –20 mental health in Clinic.
adults,20 or on symptoms among patients in specialty Health appraisals include completion of a standard-
clinics.22,23 Studies of the long-term effects of child- ized medical questionnaire that requests demographic
hood abuse have usually examined single types of abuse, and biopsychosocial information, review of organ sys-
particularly sexual abuse, and few have assessed the im- tems, previous medical diagnoses, and family medical
pact of more than one type of abuse.5,24 –28 Conditions history. A health care provider completes the medical
such as drug abuse, spousal violence, and criminal activity history, performs a physical examination, and reviews
in the household may co-occur with specific forms of the results of laboratory tests with the patient.
abuse that involve children. Without measuring these
household factors as well, long-term influence might be Survey Methods
wrongly attributed solely to single types of abuse and
the cumulative influence of multiple categories of The ACE Study protocol was approved by the Institu-
adverse childhood experiences would not be assessed. tional Review Boards of the Southern California Per-
To our knowledge, the relationship of adult health risk manente Medical Group (Kaiser Permanente), the
behaviors, health status, and disease states to childhood Emory University School of Medicine, and the Office of
abuse and household dysfunction 29 –35
has not been Protection from Research Risks, National Institutes of
Health. All 13,494 Kaiser Health Plan members who
described.
completed standardized medical evaluations at the
We undertook the Adverse Childhood Experiences
Health Appraisal Clinic between August–November of
(ACE) Study in a primary care setting to describe the
1995 and January–March of 1996 were eligible to
long-term relationship of childhood experiences to
participate in the ACE Study. Those seen at the clinic
important medical and public health problems. The
during December were not included because survey
ACE Study is assessing, retrospectively and prospec- response rates are known to be lower during the
tively, the long-term impact of abuse and household holiday period.38
dysfunction during childhood on the following out- In the week after visiting the clinic, and hence
comes in adults: disease risk factors and incidence, having their standardized medical history already
quality of life, health care utilization, and mortality. In completed, members were mailed the ACE Study
this initial paper we use baseline data from the study to questionnaire that included questions about child-
provide an overview of the prevalence and interrelation hood abuse and exposure to forms of household
of exposures to childhood abuse and household dys- dysfunction while growing up. After second mailings
function. We then describe the relationship between of the questionnaire to persons who did not respond
the number of categories of these deleterious child- to the first mailing, the response rate for the survey
hood exposures and risk factors and those diseases that was 70.5% (9,508/13,494).

246 American Journal of Preventive Medicine, Volume 14, Number 4

Figure 1. ACE Study design. *After exclusions, 59.7% of the original wave I sample (8,056/13,494) were included in this analysis.

A second survey wave of approximately the same tains a single question about childhood sexual abuse
number of patients as the first wave was conducted that reads “As a child were you ever raped or sexually
between June and October of 1997. The data for the molested?” Respondents were slightly more likely to
second survey wave is currently being compiled for answer affirmatively than nonrespondents (6.1% vs.
analysis. The methods for the second mail survey wave 5.4%, respectively).
were identical to the first survey wave as described
above. The second wave was done to enhance the Questionnaire Design
precision of future detailed analyses on special topics
We used questions from published surveys to construct
and to reduce the time necessary to obtain precise
the ACE Study questionnaire. Questions from the Con-
statistics on follow-up health events. An overview of the
flicts Tactics Scale39 were used to define psychological
total ACE Study design is provided in Figure 1.
and physical abuse during childhood and to define
violence against the respondent’s mother. We adapted
Comparison of four questions from Wyatt40 to define contact sexual
Respondents and Nonrespondents abuse during childhood. Questions about exposure to
We abstracted the completed medical evaluation for alcohol or drug abuse during childhood were adapted
every person eligible for the study; this included their from the 1988 National Health Interview Survey.41 All
medical history, laboratory results, and physical find- of the questions we used in this study to determine
ings. Respondent (n 5 9,508) and nonrespondent childhood experiences were introduced with the
(n 5 3,986) groups were similar in their percentages phrase “While you were growing up during your first 18
of women (53.7% and 51.0%, respectively) and in their years of life . . .”
mean years of education (14.0 years and 13.6 years, Questions about health-related behaviors and health
respectively). Respondents were older than nonrespon- problems were taken from health surveys such as the
dents (means 56.1 years and 49.3 years) and more likely Behavioral Risk Factor Surveys42 and the Third Na-
to be white (83.9% vs. 75.3%) although the actual tional Health and Nutrition Examination Survey,43
magnitude of the differences was small. both of which are directed by the Centers for Disease
Respondents and nonrespondents did not differ with Control and Prevention. Questions about depression
regard to their self-rated health, smoking, other sub- came from the Diagnostic Interview Schedule of the
stance abuse, or the presence of common medical National Institute of Mental Health (NIMH).44 Other
conditions such as a history of heart attack or stroke, information for this analysis such as disease history was
chronic obstructive lung disease, hypertension, or dia- obtained from the standardized questionnaire used in
betes, or with regard to marital status or current family, the Health Appraisal Clinic. (A copy of the question-
marital, or job-related problems (data not shown). The naires used in this study may be found at www.elsevier.
health appraisal questionnaire used in the clinic con- com/locate/amepre.)

Am J Prev Med 1998;14(4) 247

Table 1. Prevalence of childhood exposure to abuse and household dysfunction
Category of childhood exposurea Prevalence (%) Prevalence (%)

Abuse by category
Psychological 11.1
(Did a parent or other adult in the household . . .)
Often or very often swear at, insult, or put you down? 10.0
Often or very often act in a way that made you afraid that 4.8
you would be physically hurt?
Physical 10.8
(Did a parent or other adult in the household . . .)
Often or very often push, grab, shove, or slap you? 4.9
Often or very often hit you so hard that you had marks or 9.6
were injured?
Sexual 22.0
(Did an adult or person at least 5 years older ever . . .)
Touch or fondle you in a sexual way? 19.3
Have you touch their body in a sexual way? 8.7
Attempt oral, anal, or vaginal intercourse with you? 8.9
Actually have oral, anal, or vaginal intercourse with you? 6.9
Household dysfunction by category
Substance abuse 25.6
Live with anyone who was a problem drinker or alcoholic? 23.5
Live with anyone who used street drugs? 4.9
Mental illness 18.8
Was a household member depressed or mentally ill? 17.5
Did a household member attempt suicide? 4.0
Mother treated violently 12.5
Was your mother (or stepmother)
Sometimes, often, or very often pushed, grabbed, slapped, 11.9
or had something thrown at her?
Sometimes, often, or very often kicked, bitten, hit with a 6.3
fist, or hit with something hard?
Ever repeatedly hit over at least a few minutes? 6.6
Ever threatened with, or hurt by, a knife or gun? 3.0
Criminal behavior in household
Did a household member go to prison? 3.4 3.4
Any category reported 52.1%
a
An exposure to one or more items listed under the set of questions for each category.

Defining Childhood Exposures Risk Factors and Disease Conditions Assessed

We used three categories of childhood abuse: psycho- Using information from both the study questionnaire
logical abuse (2 questions), physical abuse (2 ques- and the Health Appraisal Clinic’s questionnaire, we
tions), or contact sexual abuse (4 questions). There chose 10 risk factors that contribute to the leading
were four categories of exposure to household dysfunc- causes of morbidity and mortality in the United
tion during childhood: exposure to substance abuse States.6,7,36,37 The risk factors included smoking, severe
(defined by 2 questions), mental illness (2 questions), obesity, physical inactivity, depressed mood, suicide
violent treatment of mother or stepmother (4 ques- attempts, alcoholism, any drug abuse, parenteral drug
tions), and criminal behavior (1 question) in the house- abuse, a high lifetime number of sexual partners
hold. Respondents were defined as exposed to a cate- ($50), and a history of having a sexually transmitted
gory if they responded “yes” to 1 or more of the disease.
questions in that category. The prevalence of positive We also assessed the relationship between childhood
responses to the individual questions and the category exposures and disease conditions that are among the
prevalences are shown in Table 1. leading causes of mortality in the United States.6 The
We used these 7 categories of childhood exposures to presence of these disease conditions was based upon
abuse and household dysfunction for our analysis. The medical histories that patients provided in response to
measure of childhood exposure that we used was simply the clinic questionnaire. We included a history of
the sum of the categories with an exposure; thus the ischemic heart disease (including heart attack or use of
possible number of exposures ranged from 0 (unex- nitroglycerin for exertional chest pain), any cancer,
posed) to 7 (exposed to all categories). stroke, chronic bronchitis, or emphysema (COPD),

248 American Journal of Preventive Medicine, Volume 14, Number 4

diabetes, hepatitis or jaundice, and any skeletal frac- Assessing the Possible Influence of Exclusions
tures (as a proxy for risk of unintentional injuries). We
To determine whether our results were influenced by
also included responses to the following question about
excluding persons with incomplete information on any
self-rated health: “Do you consider your physical health
of the categories of childhood exposure, we performed
to be excellent, very good, good, fair, or poor?” because
a separate sensitivity analysis in which we included all
it is strongly predictive of mortality.45
persons with complete demographic information but
assumed that persons with missing information for a
Definition of Risk Factors category of childhood exposure did not have an expo-
sure in that category.
We defined severe obesity as a body mass index (kg/
meter2) $35 based on measured height and weight;
physical inactivity as no participation in recreational Results
physical activity in the past month; and alcoholism as a Adverse Childhood Exposures
“Yes” response to the question “Have you ever consid-
ered yourself to be an alcoholic?” The other risk factors The level of positive responses for the 17 questions
that we assessed are self-explanatory. included in the seven categories of childhood exposure
ranged from 3.0% for a respondent’s mother (or
stepmother) having been threatened with or hurt by a
Exclusions from Analysis gun or knife to 23.5% for having lived with a problem
Of the 9,508 survey respondents, we excluded 51 drinker or alcoholic (Table 1). The most prevalent of
(0.5%) whose race was unstated and 34 (0.4%) whose the 7 categories of childhood exposure was substance
educational attainment was not reported. We also ex- abuse in the household (25.6%); the least prevalent
cluded persons who did not respond to certain ques- exposure category was evidence of criminal behavior in
tions about adverse childhood experiences. This in- the household (3.4%). More than half of respondents
volved the following exclusions: 125 (1.3%) for (52%) experienced $1 category of adverse childhood
household substance abuse, 181 (1.9%) for mental exposure; 6.2% reported $4 exposures.
illness in the home, 148 (1.6%) for violence against
mother, 7 (0.1%) for imprisonment of a household Relationships between
member, 109 (1.1%) for childhood psychological Categories of Childhood Exposure
abuse, 44 (0.5%) for childhood physical abuse, and 753 The probability that persons who were exposed to any
(7.9%) for childhood sexual abuse. After these exclu- single category of exposure were also exposed to an-
sions, 8,056 of the original 9,508 survey respondents other category is shown in Table 2. The relationship
(59.7% of the original sample of 13,494) remained and between single categories of exposure was significant
were included in the analysis. Procedures for insuring for all comparisons (P , .001; chi-square). For persons
that the findings based on complete data were gener- reporting any single category of exposure, the proba-
alizable to the entire sample are described below. bility of exposure to any additional category ranged
The mean age of the 8,506 persons included in this from 65%–93% (median: 80%); similarly, the probabil-
analysis was 56.1 years (range: 19 –92 years); 52.1% were ity of $2 additional exposures ranged from 40%–74%
women; 79.4% were white. Forty-three percent had (median: 54.5%).
graduated from college; only 6.0% had not graduated The number of categories of childhood exposures by
from high school. demographic characteristics is shown in Table 3. Statis-
tically, significantly fewer categories of exposure were
Statistical Analysis found among older persons, white or Asian persons,
and college graduates (P , .001). Because age is
We used the Statistical Analysis System (SAS)46 for our associated with both the childhood exposures as well as
analyses. We used the direct method to age-adjust the many of the health risk factors and disease outcomes,
prevalence estimates. Logistic regression analysis was all prevalence estimates in the tables are adjusted for
employed to adjust for the potential confounding ef- age.
fects of age, sex, race, and educational attainment on
the relationship between the number of childhood Relationship between
exposures and health problems.
Childhood Exposures and Health Risk Factors
To test for a dose-response relationship to health
problems, we entered the number of childhood expo- Both the prevalence and risk (adjusted odds ratio)
sures as a single ordinal variable (0, 1, 2, 3, 4, 5, 6, 7) increased for smoking, severe obesity, physical inactiv-
into a separate logistic regression model for each risk ity, depressed mood, and suicide attempts as the num-
factor or disease condition. ber of childhood exposures increased (Table 4). When

Am J Prev Med 1998;14(4) 249

 *Number exposed to first category. For example, among persons who were psychologically abused, 52% were also physically abused. More persons were a second category than would be
Categories
persons with 4 categories of exposure were compared

Additional
Any Two

(40–74)
to those with none, the odds ratios ranged from 1.3 for
physical inactivity to 12.2 for suicide attempts (Table 4).

54.5
74

64
64
41

40
47
62
Similarly, the prevalence and risk (adjusted odds
ratio) of alcoholism, use of illicit drugs, injection of
illicit drugs, $50 intercourse partners, and history of a
Additional
Category
Any One

(65–93)
sexually transmitted disease increased as the number of
childhood exposures increased (Table 5). In compar-

93

86
86
65

69
74
86

80
ing persons with $4 childhood exposures to those with
none, odds ratios ranged from 2.5 for sexually trans-
Imprisoned

mitted diseases to 7.4 for alcoholism and 10.3 for

Member

injected drug use.

(6–10)
8.5
9
9
6

8
7
10
—
Childhood Exposures and
Clustering of Health Risk Factors
Violently

(23–39)
Treated

We found a strong relationship between the number of

39
35
23

29
26
—
37
32

childhood exposures and the number of health risk

factors for leading causes of death in adults (Table 6).
(31–50)

For example, among persons with no childhood expo-

Mental
Illness
Percent (%) Exposed to Another Category

sures, 56% had none of the 10 risk factors whereas only

50
38
31

34
—
38
42
38

14% of persons with $4 categories of childhood expo-

sure had no risk factors. By contrast, only 1% of persons
Substance

with no childhood exposures had four or more risk

(39–62)

factors, whereas 7% of persons with $4 childhood

Abuse

48.5

exposures had four or more risk factors (Table 6).

51
45
39

—
46
59
62

Relationship between
(34–47)
Sexual
Abuse

Childhood Exposures and Disease Conditions

40.5
47
44
—

34
37
41
40
Table 2. Relationships between categories of adverse childhood exposure

When persons with 4 or more categories of childhood

exposure were compared to those with none, the
Physical

(19–52)

odds ratios for the presence of studied disease con-

Abuse

25.4
52*

ditions ranged from 1.6 for diabetes to 3.9 for

—
22

19
22
31
29

chronic bronchitis or emphysema (Table 7). Simi-

larly, the odds ratios for skeletal fractures, hepatitis
Psychological

or jaundice, and poor self-rated health were 1.6, 2.3,

and 2.2, respectively (Table 8).
(22–54)
Abuse

29.5
—
54
24

22
30
34
29

Significance of Dose-Response Relationships

In logistic regression models (which included age,
expected by chance (P , .001; chi-square).
898
874
1770

2064
1512
1010
271
median
Sample

range
Size*

gender, race, and educational attainment as covariates)

we found a strong, dose-response relationship between
the number of childhood exposures and each of the 10
Mother treated violently

risk factors for the leading causes of death that we

studied (P , .001). We also found a significant (P ,
Household dysfunction:

Member imprisoned
Childhood Exposure

.05) dose-response relationship between the number

Substance abuse
Childhood Abuse:

of childhood exposures and the following disease con-

First Category of

Physical abuse

Mental illness
Psychological

Sexual abuse

ditions: ischemic heart disease, cancer, chronic bron-

chitis or emphysema, history of hepatitis or jaundice,
skeletal fractures, and poor self-rated health. There was
no statistically significant dose-response relationship
for a history of stroke or diabetes.

250 American Journal of Preventive Medicine, Volume 14, Number 4

Table 3. Prevalence of categories of adverse childhood exposures by demographic characteristics
Number of categories (%)a
Sample size
Characteristic (N) 0 1 2 3 4

Age group (years)

19–34 807 35.4 25.4 17.2 11.0 10.9
35–49 2,063 39.3 25.1 15.6 9.1 10.9
50–64 2,577 46.5 25.2 13.9 7.9 6.6
$65 2,610 60.0 24.5 8.9 4.2 2.4
Genderb
Women 4,197 45.4 24.0 13.4 8.7 8.5
Men 3,859 53.7 25.8 11.6 5.0 3.9
Raceb
White 6,432 49.7 25.3 12.4 6.7 6.0
Black 385 38.8 25.7 16.3 12.3 7.0
Hispanic 431 42.9 24.9 13.7 7.4 11.2
Asian 508 66.0 19.0 9.9 3.4 1.7
Other 300 41.0 23.5 13.9 9.5 12.1
Educationb
No HS diploma 480 56.5 21.5 8.4 6.5 7.2
HS graduate 1,536 51.6 24.5 11.3 7.4 5.2
Any college 2,541 44.1 25.5 14.8 7.8 7.8
College graduate 3,499 51.4 25.1 12.1 6.1 5.3
All participants 8,056 49.5 24.9 12.5 6.9 6.2
a
The number of categories of exposure was simply the sum of each of the seven individual categories that were assessed (see Table 1).
b
Prevalence estimates adjusted for age.

Assessment of the Influence of Exclusions comes may be more strongly related to unique combi-
nations or the intensity of adverse childhood exposures
In the sensitivity analysis where missing information for
than to the total breadth of exposure that we used for
a category of childhood exposure was considered as no
our analysis. However, the analysis we present illustrates
exposure, the direction and strength of the associations
the need for an overview of the net effects of a group of
between the number of childhood exposures and the
complex interactions on a wide range of health risk
risk factors and disease conditions were nearly identical
behaviors and diseases.
(data not shown). Thus, the results we present appear
to be unaffected by our decision to exclude persons for Several potential limitations need to be considered
whom information on any category of childhood expo- when interpreting the results of this study. The data
sure was incomplete. about adverse childhood experiences are based on
self-report, retrospective, and can only demonstrate
associations between childhood exposures and health
Discussion risk behaviors, health status, and diseases in adulthood.
We found a strong dose response relationship between Second, some persons with health risk behaviors or
the breadth of exposure to abuse or household dysfunc- diseases may have been either more, or less, likely to
tion during childhood and multiple risk factors for report adverse childhood experiences. Each of these
several of the leading causes of death in adults. Disease issues potentially limits inferences about causality. Fur-
conditions including ischemic heart disease, cancer, thermore, disease conditions could be either over- or
chronic lung disease, skeletal fractures, and liver dis- under-reported by patients when they complete the
ease, as well as poor self-rated health also showed a medical questionnaire. In addition, there may be me-
graded relationship to the breadth of childhood expo- diators of the relationship between childhood experi-
sures. The findings suggest that the impact of these ences and adult health status other than the risk factors
adverse childhood experiences on adult health status is we examined. For example, adverse childhood experi-
strong and cumulative. ences may affect attitudes and behaviors toward health
The clear majority of patients in our study who were and health care, sensitivity to internal sensations, or
exposed to one category of childhood abuse or house- physiologic functioning in brain centers and neuro-
hold dysfunction were also exposed to at least one transmitter systems. A more complete understanding
other. Therefore, researchers trying to understand the of these issues is likely to lead to more effective ways
long-term health implications of childhood abuse may to address the long-term health problems associated
benefit from considering a wide range of related ad- with childhood abuse and household dysfunction.
verse childhood exposures. Certain adult health out- However, our estimates of the prevalence of child-

Am J Prev Med 1998;14(4) 251

Table 4. Number of categories of adverse childhood exposure and the adjusted odds of risk factors including current
smoking, severe obesity, physical inactivity, depressed mood, and suicide attempt
Number Sample Adjusted 95%
of size Prevalence odds confidence
Health problem categories (N)a (%)b ratioc interval

Current smokerd 0 3,836 6.8 1.0 Referent

1 2,005 7.9 1.1 ( 0.9–1.4)
2 1,046 10.3 1.5 ( 1.1–1.8)
3 587 13.9 2.0 ( 1.5–2.6)
4 or more 544 16.5 2.2 ( 1.7–2.9)
Total 8,018 8.6 — —
Severe obesityd 0 3,850 5.4 1.0 Referent
(BMI $ 35) 1 2,004 7.0 1.1 ( 0.9–1.4)
2 1,041 9.5 1.4 ( 1.1–1.9)
3 590 10.3 1.4 ( 1.0–1.9)
4 or more 543 12.0 1.6 ( 1.2–2.1)
Total 8,028 7.1 — —
No leisure-time 0 3,634 18.4 1.0 Referent
physical activity 1 1,917 22.8 1.2 ( 1.1–1.4)
2 1,006 22.0 1.2 ( 1.0–1.4)
3 559 26.6 1.4 ( 1.1–1.7)
4 or more 523 26.6 1.3 ( 1.1–1.6)
Total 7,639 21.0 — —
Two or more weeks of 0 3,799 14.2 1.0 Referent
depressed mood in
the past year
1 1,984 21.4 1.5 ( 1.3–1.7)
2 1,036 31.5 2.4 ( 2.0–2.8)
3 584 36.2 2.6 ( 2.1–3.2)
4 or more 542 50.7 4.6 ( 3.8–5.6)
Total 7,945 22.0 — —
Ever attempted suicide 0 3,852 1.2 1.0 Referent
1 1,997 2.4 1.8 ( 1.2–2.6)
2 1,048 4.3 3.0 ( 2.0–4.6)
3 587 9.5 6.6 ( 4.5–9.8)
4 or more 544 18.3 12.2 (8.5–17.5)
Total 8,028 3.5 — —
a
Sample sizes will vary due to incomplete or missing information about health problems.
b
Prevalence estimates are adjusted for age.
c
Odds ratios adjusted for age, gender, race, and educational attainment.
d
Indicates information recorded in the patient’s chart before the study questionnaire was mailed.

hood exposures are similar to estimates from nationally estimate actual occurrence.47,48 Underestimates of
representative surveys, indicating that the experiences childhood exposures would result in downwardly bi-
of our study participants are comparable to the larger ased estimates of the relationships between childhood
population of U.S. adults. In our study, 23.5% of exposures and adult health risk behaviors and dis-
participants reported having grown up with an alcohol eases. Another potential source of underestimation
abuser; the 1988 National Health Interview Survey of the strength of these relationships is the lower
estimated that 18.1% of adults had lived with an alcohol number of childhood exposures reported by older
abuser during childhood.41 Contact sexual abuse was persons in our study. This may be an artifact caused
reported by 22% of respondents (28% of women and by premature mortality in persons with multiple
16% of men) in our study. A national telephone survey adverse childhood exposures; the clustering of mul-
of adults in 1990 using similar criteria for sexual abuse tiple risk factors among persons with multiple child-
estimated that 27% of women and 16% of men had hood exposures is consistent with this hypothesis.
been sexually abused.12 Thus, the true relationships between adverse child-
There are several reasons to believe that our esti- hood exposures and adult health risk behaviors,
mates of the long-term relationship between adverse health status, and diseases may be even stronger than
childhood experiences and adult health are conserva- those we report.
tive. Longitudinal follow-up of adults whose childhood An essential question posed by our observations is,
abuse was well documented has shown that their retro- “Exactly how are adverse childhood experiences linked
spective reports of childhood abuse are likely to under- to health risk behaviors and adult diseases?” The link-

252 American Journal of Preventive Medicine, Volume 14, Number 4

Table 5. Number of categories of adverse childhood exposure and the prevalence and risk (adjusted odds ratio) of health
risk factors including alcohol or drug abuse, high lifetime number of sexual partners, or history of sexually
transmitted disease
Number Sample Adjusted 95%
of size Prevalence odds confidence
Health problem categories (N)a (%)b ratioc interval

Considers self an 0 3,841 2.9 1.0 Referent

alcoholic 1 1,993 5.7 2.0 (1.6–2.7)
2 1,042 10.3 4.0 (3.0–5.3)
3 586 11.3 4.9 (3.5–6.8)
4 or more 540 16.1 7.4 (5.4–10.2)
Total 8,002 5.9 — —
Ever used illicit drugs 0 3,856 6.4 1.0 Referent
1 1,998 11.4 1.7 (1.4–2.0)
2 1,045 19.2 2.9 (2.4–3.6)
3 589 21.5 3.6 (2.8–4.6)
4 or more 541 28.4 4.7 (3.7–6.0)
Total 8,029 11.6 — —
Ever injected drugs 0 3,855 0.3 1.0 Referent
1 1,996 0.5 1.3 (0.6–3.1)
2 1,044 1.4 3.8 (1.8–8.2)
3 587 2.3 7.1 (3.3–15.5)
4 or more 540 3.4 10.3 (4.9–21.4)
Total 8,022 0.8 — —
Had 50 or more 0 3,400 3.0 1.0 Referent
intercourse partners 1 1,812 5.1 1.7 (1.3–2.3)
2 926 6.1 2.3 (1.6–3.2)
3 526 6.3 3.1 (2.0–4.7)
4 or more 474 6.8 3.2 (2.1–5.1)
Total 7,138 4.4 — —
Ever had a sexually 0 3,848 5.6 1.0 Referent
transmitted diseased 1 2,001 8.6 1.4 (1.1–1.7)
2 1,044 10.4 1.5 (1.2–1.9)
3 588 13.1 1.9 (1.4–2.5)
4 or more 542 16.7 2.5 (1.9–3.2)
Total 8023 8.2 — —
a
Sample sizes will vary due to incomplete or missing information about health problems.
b
Prevalence estimates are adjusted for age.
c
Odds ratios adjusted for age, gender, race, and educational attainment.
d
Indicates information recorded in the patient’s chart before the study questionnaire was mailed.

ing mechanisms appear to center on behaviors such as family and household dysfunction. High levels of expo-
smoking, alcohol or drug abuse, overeating, or sexual sure to adverse childhood experiences would expect-
behaviors that may be consciously or unconsciously edly produce anxiety, anger, and depression in chil-
used because they have immediate pharmacological or dren. To the degree that behaviors such as smoking,
psychological benefit as coping devices in the face of alcohol, or drug use are found to be effective as coping
the stress of abuse, domestic violence, or other forms of devices, they would tend to be used chronically. For

Table 6. Relationship between number of categories of childhood exposure and number of risk factors for the leading
causes of deatha
% with number of risk factors
Sample
Number of categories size 0 1 2 3 4

0 3,861 56 29 10 4 1
1 2,009 42 33 16 6 2
2 1,051 31 33 20 10 4
3 590 24 33 20 13 7
$4 545 14 26 28 17 7
Total 8,056 44 31 15 7 3
a
Risk factors include: smoking, severe obesity, physical inactivity, depressed mood, suicide attempt, alcoholism, any drug use, injected drug use,
$50 lifetime sexual partners, and history of a sexually transmitted disease.

Am J Prev Med 1998;14(4) 253

Table 7. Number of categories of adverse childhood exposure and the prevalence and risk (adjusted odds ratio) of heart
attack, cancer, stroke, COPD, and diabetes
Number Sample Adjusted 95%
of size Prevalence odds confidence
Disease conditiond categories (N)a (%)b ratioc interval

Ischemic heart disease 0 3,859 3.7 1.0 Referent

1 2,009 3.5 0.9 (0.7–1.3)
2 1,050 3.4 0.9 (0.6–1.4)
3 590 4.6 1.4 (0.8–2.4)
4 or more 545 5.6 2.2 (1.3–3.7)
Total 8,022 3.8 — —
Any cancer 0 3,842 1.9 1.0 Referent
1 1,995 1.9 1.2 (1.0–1.5)
2 1,043 1.9 1.2 (1.0–1.5)
3 588 1.9 1.0 (0.7–1.5)
4 or more 543 1.9 1.9 (1.3–2.7)
Total 8,011 1.9 — —
Stroke 0 3,832 2.6 1.0 Referent
1 1,993 2.4 0.9 (0.7–1.3)
2 1,042 2.0 0.7 (0.4–1.3)
3 588 2.9 1.3 (0.7–2.4)
4 or more 543 4.1 2.4 (1.3–4.3)
Total 7,998 2.6 — —
Chronic bronchitis or 0 3,758 2.8 1.0 Referent
emphysema 1 1,939 4.4 1.6 (1.2–2.1)
2 1,009 4.4 1.6 (1.1–2.3)
3 565 5.7 2.2 (1.4–3.3)
4 or more 512 8.7 3.9 (2.6–5.8)
Total 7,783 4.0 — —
Diabetes 0 3,850 4.3 1.0 Referent
1 2,002 4.1 1.0 (0.7–1.3)
2 1,046 3.9 0.9 (0.6–1.3)
3 587 5.0 1.2 (0.8–1.9)
4 or more 542 5.8 1.6 (1.0–2.5)
Total 8,027 4.3 — —
a
Sample sizes will vary due to incomplete or missing information about health problems.
b
Prevalence estimates are adjusted for age.
c
Odds ratios adjusted for age, gender, race, and educational attainment.
d
Indicates information recorded in the patient’s chart before the study questionnaire was mailed.

example, nicotine is recognized as having beneficial tendency to ignore psychological issues in the manage-
psychoactive effects in terms of regulating affect49 and ment of organic disease make improbable any full
persons who are depressed are more likely to understanding of the original causes of adult disease
smoke.50,51 Thus, persons exposed to adverse child- (Figure 2). Thus, incomplete understanding of the
hood experiences may benefit from using drugs such as possible benefits of health risk behaviors leads them to
nicotine to regulate their mood.49,52 be viewed as irrational and having solely negative
Consideration of the positive neuroregulatory effects consequences.
of health-risk behaviors such as smoking may provide Because adverse childhood experiences are common
biobehavioral explanations53 for the link between ad- and they have strong long-term associations with adult
verse childhood experiences and health risk behaviors health risk behaviors, health status, and diseases, in-
and diseases in adults. In fact, we found that exposure creased attention to primary, secondary, and tertiary
to higher numbers of categories of adverse childhood prevention strategies is needed. These strategies in-
experiences increased the likelihood of smoking by the clude prevention of the occurrence of adverse child-
age of 14, chronic smoking as adults, and the presence hood experiences, preventing the adoption of health
of smoking-related diseases. Thus, smoking, which is risk behaviors as responses to adverse experiences
medically and socially viewed as a “problem” may, from during childhood and adolescence, and, finally, help-
the perspective of the user, represent an effective ing change the health risk behaviors and ameliorating
immediate solution that leads to chronic use. Decades the disease burden among adults whose health prob-
later, when this “solution” manifests as emphysema, lems may represent a long-term consequence of ad-
cardiovascular disease, or malignancy, time and the verse childhood experiences.

254 American Journal of Preventive Medicine, Volume 14, Number 4

Table 8. Number of categories of adverse childhood exposure and the prevalence and risk (adjusted odds ratio) of skeletal
fracture, hepatitis or jaundice, and poor self-rated health
Number Sample Adjusted 95%
of size Prevalence odds confidence
Disease condition categories (N)a (%)b ratioc interval

Ever had a skeletal 0 3,843 3.6 1.0 Referent

fracture 1 1,998 4.0 1.1 (1.0–1.2)
2 1,048 4.5 1.4 (1.2–1.6)
3 587 4.0 1.2 (1.0–1.4)
4 or more 544 4.8 1.6 (1.3–2.0)
Total 8,020 3.9 — —
Ever had hepatitis or 0 3,846 5.3 1.0 Referent
jaundice 1 2,006 5.5 1.1 (0.9–1.4)
2 1,045 7.7 1.8 (1.4–2.3)
3 590 10.2 1.6 (1.2–2.3)
4 or more 543 10.7 2.4 (1.8–3.3)
Total 8,030 6.5 — —
Fair or poor self-rated 0 3,762 16.3 1.0 Referent
health 1 1,957 17.8 1.2 (1.0–1.4)
2 1,029 19.9 1.4 (1.2–1.7)
3 584 20.3 1.4 (1.1–1.7)
4 or more 527 28.7 2.2 (1.8–2.7)
Total 7,859 18.2 — —
a
Sample sizes will vary due to incomplete or missing information about health problems.
b
Prevalence estimates are adjusted for age and gender.
c
Odds ratios adjusted for age, gender, race, and educational attainment.
d
Indicates information recorded in the patient’s chart before the study questionnaire was mailed.

Primary prevention of adverse childhood experi- derstanding of the behavioral coping devices that
ences has proven difficult54,55 and will ultimately re- commonly are adopted to reduce the emotional impact
quire societal changes that improve the quality of family of these experiences. The improbability of giving up an
and household environments during childhood. Recent immediate “solution” in return for a nebulous long-
research on the long-term benefit of early home visitation term health benefit has thwarted many well-intended
on reducing the prevalence of adverse childhood experi- preventive efforts. Although articles in the general
ences is promising.56 In fact, preliminary data from the medical literature are alerting the medical community
ACE Study provided the impetus for the Kaiser Health to the fact that childhood abuse is common,59 adoles-
Plan to provide funding to participate at 4 locations cent health care is often inadequate in terms of psycho-
(including San Diego County, California) in the Com- social assessment and anticipatory guidance.60 Clearly,
monwealth Fund’s “Healthy Steps” program. This pro- comprehensive strategies are needed to identify and
gram extends the traditional practice of pediatrics by intervene with children and families who are at risk for
adding one or more specialists in the developmental and these adverse experiences and their related out-
psychosocial dimensions of both childhood and parent- comes.61 Such strategies should include increased com-
hood. Through a series of office visits, home visits, and a munication between and among those involved in
telephone advice line for parents, these specialists develop family practice, internal medicine, nursing, social work,
close relationships between children and their families pediatrics, emergency medicine, and preventive medi-
from birth to 3 years of age. This approach is consistent cine and public health. Improved understanding is also
with the recommendation of the U.S. Advisory Board on needed of the effects of childhood exposure to domes-
Child Abuse and Neglect that a universal home visitation tic violence.19,62 Additionally, increased physician train-
program for new parents be developed57,58 and provides ing63 is needed to recognize and coordinate the man-
an example of a family-based primary prevention effort agement of all persons affected by child abuse,
that is being explored in a managed care setting. If these domestic violence, and other forms of family adversity
types of approaches can be replicated and implemented such as alcohol abuse or mental illness.
on a large scale, the long-term benefits may include, In the meantime, tertiary care of adults whose health
somewhat unexpectedly, substantial improvements in problems are related to experiences such as childhood
overall adult health. abuse5 will continue to be a difficult challenge. The
Secondary prevention of the effects of adverse child- relationship between childhood experiences and adult
hood experiences will first require increased recogni- health status is likely to be overlooked in medical
tion of their occurrence and second, an effective un- practice because the time delay between exposure

Am J Prev Med 1998;14(4) 255

Figure 2. Potential influences throughout the lifespan of adverse childhood experiences.

during childhood and recognition of health problems medical and public health research, education, and
in adult medical practice is lengthy. Moreover, these practice can be offset only by the magnitude of the
childhood exposures include emotionally sensitive top- implications that these changes have for improving the
ics such as family alcoholism29,30 and sexual abuse.64 health of the nation.
Many physicians may fear that discussions of sexual
violence and other sensitive issues are too personal We thank Naomi Howard for her dedication to the ACE Study.
even for the doctor-patient relationship.65 For example, This research is supported by the Centers for Disease Control
the American Medical Association recommends screen- and Prevention via cooperative agreement TS-44-10/12 with the
ing of women for exposure to violence at every en- Association of Teachers of Preventive Medicine.
trance to the health system;66 however, such screening
appears to be rare.67 By contrast, women who are asked
about exposure to sexual violence say they consider
such questions to be welcome and germane to routine References
1. Springs F, Friedrich WN. Health risk behaviors and
medical care,68 which suggests that physicians’ fears
medical sequelae of childhood sexual abuse. Mayo Clin
about patient reactions are largely unfounded.
Proc 1992;67:527–32.
Clearly, further research and training are needed to
2. Felitti VJ. Long-term medical consequences of incest,
help medical and public health practitioners under- rape, and molestation. South Med J 1991;84:328 –31.
stand how social, emotional, and medical problems are
3. Felitti VJ. Childhood sexual abuse, depression and family
linked throughout the lifespan (Figure 2). Such re- dysfunction in adult obese patients: a case control study.
search and training would provide physicians with the South Med J 1993;86:732– 6.
confidence and skills to inquire and respond to patients 4. Gould DA, Stevens NG, Ward NG, Carlin AS, Sowell HE,
who acknowledge these types of childhood exposures. Gustafson B. Self-reported childhood abuse in an adult
Increased awareness of the frequency and long-term population in a primary care setting. Arch Fam Med
consequences of adverse childhood experiences may 1994;3:252– 6.
also lead to improvements in health promotion and 5. McCauley J, Kern DE, Kolodner K, Schroeder AF, et al.
disease prevention programs. The magnitude of the Clinical characteristics of women with a history of child-
difficulty of introducing the requisite changes into hood abuse. JAMA 1997;277:1362– 8.

256 American Journal of Preventive Medicine, Volume 14, Number 4

 6. Mortality patterns: United States, 1993. Morb Mortal 26. Claussen AH, Crittenden PM. Physical and psychological
Wkly Rep 1996;45:161– 4. maltreatment: relations among types of maltreatment.
7. McGinnis JM, Foege WH. Actual causes of death in the Child Abuse Negl 1991;15:5–18.
United States. JAMA 1993;270:2207–12. 27. Moeller TP, Bachman GA, Moeller JR. The combined
8. Landis J. Experiences of 500 children with adult sexual effects of physical, sexual, and emotional abuse during
deviation. Psychiatr Q 1956;30(Suppl):91–109. childhood: long-term health consequences for women.
9. Straus MA, Gelles RJ. Societal change and change in Child Abuse Negl 1993;17:623– 40.
family violence from 1975 to 1985 as revealed by two 28. Bryant SL, Range LM. Suicidality in college women who
national surveys. J Marriage Family 1986;48:465–79. were sexually and physically punished by parents. Vio-
10. Wyatt GE, Peters SD. Methodological considerations in lence Vict 1995;10:195–201.
research on the prevalence of child sexual abuse. Child 29. Zeitlen H. Children with alcohol misusing parents. Br
Abuse Negl 1986;10:241–51. Med Bull 1994;50:139 –51.
11. Berger AM, Knutson JF, Mehm JG, Perkins KA. The 30. Dore MM, Doris JM, Wright P. Identifying substance
self-report of punitive childhood experiences of young abuse in maltreating families: a child welfare challenge.
adults and adolescents. Child Abuse Negl 1988;12:251– Child Abuse Negl 1995;19:531– 43.
62. 31. Ethier LS, Lacharite C, Couture G. Childhood adversity,
12. Finkelhor D, Hotaling G, Lewis IA, Smith C. Sexual abuse parental stress, and depression of negligent mothers.
in a national survey of adult men and women: prevalence, Child Abuse Negl 1995;19:619 –32.
characteristics, and risk factors. Child Abuse Negl 1990; 32. Spaccarelli S, Coatsworth JD, Bowden BS. Exposure to
14:19 –28. family violence among incarcerated boys; its association
13. Egelend B, Sroufe LA, Erickson M. The developmental with violent offending and potential mediating variables.
consequence of different patterns of maltreatment. Child Violence Vict 1995;10:163– 82.
Abuse Negl 1983;7:459 – 69.
33. McCloskey LA, Figueredo AJ, Koss MP. The effects of
14. Finkelhor D, Browne A. The traumatic impact of child
systemic family violence on children’s mental health.
sexual abuse Am J Orthopsychiatry. 1985;55:530 – 41.
Child Dev 1995;66:1239 – 61.
15. Beitchman JH, Zucker KJ, Hood JE, DaCosta GA, Akman
34. Brent DA, Perper JA, Moritz G, Schweers J, Balach L,
D, Cassavia E. A review of the long-term effects of sexual
Roth C. Familial risk factors for adolescent suicide: a
abuse. Child Abuse Negl 1992;16:101–18.
case-control study. Acta Psychiatr Scand 1994;89:52– 8.
16. Hibbard RA, Ingersoll GM, Orr DP. Behavioral risk,
35. Shaw DS, Vondra JI, Hommerding KD, Keenan K, Dunn
emotional risk, and child abuse among adolescents in a
M. Chronic family adversity and early child behavior
nonclinical setting. Pediatrics 1990;86:896 –901.
problems: a longitudinal study of low income families.
17. Nagy S, Adcock AG, Nagy MC. A comparison of risky
J Child Psychol Psychiatry 1994;35:1109 –22.
health behaviors of sexually active, sexually abused, and
36. U.S. Department of Health and Human Services. Physical
abstaining adolescents. Pediatrics 1994;93:570 –5.
activity and health: A report of the Surgeon General.
18. Cunningham RM, Stiffman AR, Dore P. The association
Atlanta, Georgia. U.S. Department of Health and Human
of physical and sexual abuse with HIV risk behaviors in
Services, Centers for Disease Control and Prevention,
adolescence and young adulthood: implications for pub-
National Center for Chronic Disease Prevention and
lic health. Child Abuse Negl 1994;18:233– 45.
Health Promotion; 1996.
19. Council on Scientific Affairs. Adolescents as victims of
family violence. JAMA 1993;270:1850 – 6. 37. Rivara FP, Mueller BA, Somes G, Mendoza CT, Rushforth
20. Nelson DE, Higginson GK, Grant-Worley JA. Physical NB, Kellerman AL. Alcohol and illicit drug abuse and the
abuse among high school students. Prevalence and cor- risk of violent death in the home. JAMA 1997;278:569 –
relation with other health behaviors. Arch Pediatr Ado- 75.
lesc Med 1995;149:1254 – 8. 38. Dillman DA. Mail and telephone surveys: the total design
21. Mullen PE, Roman-Clarkson SE, Walton VA, Herbison method. New York: John Wiley & Sons; 1978.
GP. Impact of sexual and physical abuse on women’s 39. Straus M, Gelles RJ. Physical violence in American fami-
mental health. Lancet 1988;1:841–5. lies: risk factors and adaptations to violence in 8,145
22. Drossman DA, Leserman J, Nachman G, Li Z, et al. Sexual families. New Brunswick: Transaction Press; 1990.
and physical abuse in women with functional or organic 40. Wyatt GE. The sexual abuse of Afro-American and White-
gastrointestinal disorders. Ann Intern Med 1990;113: American women in childhood. Child Abuse Negl 1985;
828 –33. 9:507–19.
23. Harrop-Griffiths J, Katon W, Walker E, Holm L, Russo J, 41. National Center for Health Statistics. Exposure to alco-
Hickok L. The association between chronic pelvic pain, holism in the family: United States, 1988. Advance Data,
psychiatric diagnoses, and childhood sexual abuse. Ob- No. 205. U.S. Department of Health and Human Services,
stet Gynecol 1988;71:589 –94. Washington, DC; September 30, 1991.
24. Briere J, Runtz M. Multivariate correlates of childhood 42. Siegel PZ, Frazier EL, Mariolis P, et al. Behaviorial risk
psychological and physical maltreatment among univer- factor surveillance, 1991; Monitoring progress toward the
sity women. Child Abuse Negl 1988;12:331– 41. Nation’s Year 2000 Health Objectives. Morb Mortal Wkly
25. Briere J, Runtz M. Differential adult symptomatology Rep 1992;42(SS-4).1–15.
associated with three types of child abuse histories. Child 43. Crespo CJ, Keteyian SJ, Heath GW, Sempos CT. Leisure-
Abuse Negl 1990;14:357– 64. time physical activity among US adults: Results from the

Am J Prev Med 1998;14(4) 257

 Third National Health and Nutrition Examination Sur- 57. U.S. Advisory Board on Child Abuse and Neglect. Child
vey. Arch Intern Med 1996;156:93– 8. abuse and neglect: critical first steps in response to a
44. Robins LN, Helzer JE, Groughan J, Ratliff K. National national emergency. Washington, DC: U.S. Government
Institute of Mental Health Diagnostic Interview Schedule: Printing Office; August 1990; publication no. 017-092-
its history, characteristics, and validity. Arch Gen Psychi- 00104-5.
atry 1981;38:381–9. 58. U.S. Advisory Board on Child Abuse and Neglect. Creat-
45. Idler E, Angel RJ. Self-rated health and mortality in the ing caring communities: blueprint for an effective federal
NHANES I Epidemiologic Follow-up Study. Am J Pub policy on child abuse and neglect. Washington, DC: U.S.
Health 1990;80:446 –52. Government Printing Office; September 1991.
46. SAS Procedures Guide. SAS Institute Inc. Version 6, 3rd
59. MacMillan HL, Fleming JE, Trocme N, Boyle MH, et al.
edition, Cary, NC: SAS Institute; 1990.
Prevalence of child physical and sexual abuse in the
47. Femina DD, Yeager CA, Lewis DO. Child abuse: adoles-
community. Results from the Ontario Health Supple-
cent records vs. adult recall. Child Abuse Negl 1990;14:
ment. JAMA 1997;278:131–5.
227–31.
48. Williams LM. Recovered memories of abuse in women 60. Rixey S. Family violence and the adolescent. Maryland
with documented child sexual victimization histories. Med J 1994;43:351–3.
J Traumatic Stress 1995;8:649 –73. 61. Chamberlin RW. Preventing low birth weight, child
49. Carmody TP. Affect regulation, nicotine addiction, and abuse, and school failure: the need for comprehensive,
smoking cessation. J Psychoactive Drugs 1989;21:331– 42. community-wide approaches. Pediatr Rev 1992;13:64 –71.
50. Anda RF, Williamson DF, Escobedo LG, Mast EE, Giovino 62. Kashani JH, Daniel AE, Dandoy AC, Holcomb WR. Family
GA, Remington PL. Depression and the dynamics of violence: impact on children. J Am Acad Child Adolesc
smoking. A national perspective. JAMA 1990;264:1541–5. Psychiatry 1992;31:181–9.
51. Glassman AH, Helzer JE, Covey LS, Cottler LB, Stetner F, 63. Dubowitz H. Child abuse programs and pediatric resi-
Tipp JE, Johnson J. Smoking, smoking cessation, and dency training. Pediatrics 1988;82:477– 80.
major depression. JAMA 1990;264:1546 –9. 64. Tabachnick J, Henry F, Denny L. Perceptions of child
52. Hughes JR. Clonidine, depression, and smoking cessa- sexual abuse as a public health problem. Vermont, Sep-
tion. JAMA 1988;259:2901–2. tember 1995. Morb Mortal Wkly Rep 1997;46:801–3.
53. Pomerlau OF, Pomerlau CS. Neuroregulators and the
65. Sugg NK, Inui T. Primary care physicians’ response to
reinforcement of smoking: towards a biobehavioral expla-
domestic violence. Opening Pandora’s box. JAMA 1992;
nation. Neurosci Biobehav Rev 1984;8:503–13.
267:3157– 60.
54. Hardy JB, Street R. Family support and parenting educa-
66. Council on Scientific Affairs. American Medical Associa-
tion in the home: an effective extension of clinic-based
preventive health care services for poor children. J Pedi- tion Diagnostic and Treatment Guidelines on Domestic
atr 1989;115:927–31. Violence. Arch Fam Med 1992;1:38 – 47.
55. Olds DL, Henderson CR, Chamberlin R, Tatelbaum R. 67. Hamberger LK, Saunders DG, Hovey M. Prevalence of
Preventing child abuse and neglect: a randomized trial of domestic violence in community practice and rate of
nurse home visitation. Pediatrics 1986;78:65–78. physician inquiry. Fam Med 1992;24:283–7.
56. Olds DL, Eckenrode J, Henderson CR, Kitzman H, et al. 68. Friedman LS, Samet JH, Roberts MS, Hans P. Inquiry
Long-term effects of home visitation on maternal life about victimization experiences. A survey of patient pref-
course and child abuse and neglect: Fifteen-year fol- erences and physician practices. Arch Int Med 1992;152:
low-up of a randomized trial. JAMA 1997;278:637– 43. 1186 –90.

258 American Journal of Preventive Medicine, Volume 14, Number 4