Journal of Adolescent Health 38 (2006) 444.e1– 444.

e10

Original article

Adverse childhood experiences and the association with ever using

alcohol and initiating alcohol use during adolescence
Shanta R. Dube, M.P.H.a,*, Jacqueline W. Miller, M.D.a, David W. Brown, M.S.P.H., M.S.a,
Wayne H. Giles, M.D., M.S.a, Vincent J. Felitti, M.D.b, Maxia Dong, M.D., Ph.D.a, and
Robert F. Anda, M.D., M.S.a
a
National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia
b
Department of Preventive Medicine, Southern California Permanente Medical Group (Kaiser Permanente), San Diego, California
Manuscript received March 31, 2005; manuscript accepted June 1, 2005.

Abstract Purpose: Alcohol is the most common and frequently used drug and has the potential to cause
multiple deleterious effects throughout the lifespan. Because early age at initiation of alcohol
use increases this potential and programs and laws are in place to attempt to delay the onset of
alcohol use, we studied the relationship between multiple adverse childhood experiences
(ACEs) and both the likelihood of ever drinking and the age at initiating alcohol use.
Methods: This was a retrospective cohort study of 8417 adult health maintenance organization
(HMO) members in California who completed a survey about ACEs, which included childhood
abuse and neglect, growing up with various forms of household dysfunction and alcohol use in
adolescence and adulthood. The main outcomes measured were ever drinking and age at
initiating alcohol use among ever-drinkers for four age categories: ⱕ 14 years (early adoles-
cence), 15 to 17 years (mid adolescence), and 18 to 20 years (late adolescence); age ⱖ 21 years
was the referent. The relationship between the total number of adverse childhood experiences
(ACE score) and early initiation of alcohol use (ⱕ14 years) among four birth cohorts dating
back to 1900 was also examined.
Results: Eighty-nine percent of the cohort reported ever drinking; all individual ACEs except
physical neglect increased the risk of ever using alcohol (p ⬍ .05). Among ever drinkers,
initiating alcohol use by age 14 years was increased two- to threefold by individual ACEs
(p ⬍ .05). ACEs also accounted for a 20% to 70% increased likelihood of alcohol use initiated
during mid adolescence (15–17 years). The total number of ACEs (ACE score) had a very
strong graded relationship to initiating alcohol use during early adolescence and a robust but
somewhat less strong relationship to initiation during mid adolescence. For each of the four
birth cohorts, the ACE score had a strong, graded relationship to initiating alcohol use by age
14 years (p ⬍ .05).
Conclusions: Adverse childhood experiences are strongly related to ever drinking alcohol and
to alcohol initiation in early and mid adolescence, and the ACE score had a graded or
“dose-response” relationship to these alcohol use behaviors. The persistent graded relationship
between the ACE score and initiation of alcohol use by age 14 for four successive birth cohorts
dating back to 1900 suggests that the stressful effects of ACEs transcend secular changes,

*Address correspondence to: Mrs. Shanta R. Dube, Mailstop K-67,

Centers for Disease Control and Prevention, 4770 Buford Hwy NE, At-
lanta, GA 30341-3724.
E-mail address: skd7@cdc.gov

1054-139X/06/$ – see front matter © 2006 Society for Adolescent Medicine. All rights reserved.
doi:10.1016/j.jadohealth.2005.06.006
444.e2 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10

including the increased availability of alcohol, alcohol advertising, and the recent campaigns
and health education programs to prevent alcohol use. These findings strongly suggest that
efforts to delay the age of onset of drinking must recognize the contribution of multiple
traumatic and stressful events to alcohol-seeking behavior among children and adolescents. ©
2006 Society for Adolescent Medicine. All rights reserved.
Keywords: Alcohol use; Underage drinking; Child abuse; Family violence

Alcohol and its use have been a part of American culture
dating back nearly 300 years to the emigration of Europeans
to North America [1]. Early observations that frequent al-
cohol use coincided with health and social problems led to
efforts to reduce the deleterious effects of alcohol consump-
tion. As a result, the first liquor laws focused on taxes,
license fees, and fines for drunken behavior [1]. The Tem-
perance Movement of 1773 and Prohibition in the 1920s
aimed to curb alcohol use, although their effectiveness re-
mains questionable. In fact, Prohibition was largely a fail-
ure, and in retrospect may have made matters worse [1,2]. In
the mid-1980s, the legal drinking age was increased from 18
to 21 years in some states, and this law has now been
adopted by all states. Nonetheless, underage drinking con-
tinues, and in fact youth is often the norm for persons
initiating alcohol use [3]. The success of society’s efforts
over the past 300 years to avoid problems associated with
alcohol use is debatable because alcohol use continues to be
pervasive within the United States.
Over the past century, advances in public health have
made it possible to document the varied secular and social
influences on the use of alcohol [4]. For example, the
National Longitudinal Alcohol Epidemiologic Survey found
that alcohol use in early adulthood (ages 20 –24 years) was
less than 50% for those born between 1894 and 1937 (pre-
World War II), but this figure increased to 75% of persons
born between 1968 and 1974 (Vietnam War era) [4]. Of
particular concern is the trend toward initiation of alcohol
use at earlier ages. Proposed explanations include increas-
ing availability of alcohol both inside and outside of the
home and exposure to marketing in the mass media [5]. In
fact, recent data indicate that the age at initiating alcohol use
has declined from 17.6 years in 1965 to 15.9 years in 2001
[6]. In the 2003 Youth Risk Behavior Survey, nearly 30% of
high school students reported that they had their first drink
before age 13 [7].
It is now well recognized that early initiation of alcohol
use increases the risk of alcohol-related disorders later in
life, such as alcohol dependence [8 –10]. Studies have iden-
tified several predictors of early drinking such as the influ-
ence of peers [11,12], broken family structure [12], and poor
quality of family relations [12]. Witnessing domestic vio-
lence [13] and experiencing physical, sexual, and emotional
abuse [13,14] are associated with alcohol use among ado-
lescents [13,14]. Although these studies provide evidence
that domestic violence and childhood abuse are related to
underage drinking, more information is needed to better
understand the relationship between multiple coexisting
childhood stressors in the family environment and age at
initiation of alcohol use.
The Adverse Childhood Experiences (ACE) Study is a
large epidemiologic study among adult health maintenance
organization (HMO) members that is designed to assess the
impact of childhood stressors (i.e., abuse, neglect, and other
forms of family dysfunction) on a wide range of health
behaviors and outcomes in adolescence and adulthood [15].
To date, the ACE Study has repeatedly shown that traumatic
and stressful childhood events increase the likelihood of
myriad adolescent risk behaviors, such as early age at first
intercourse [16], regular smoking by age 14 [17], attempted
suicide during adolescence [18], use of illicit drugs during
adolescence [19], male involvement in teen pregnancy [20],
and adolescent pregnancy [21].
In this study, we examined the relationship between 10
adverse childhood experiences (ACEs) with both those who
ever drank alcohol other than a few sips and initiation of
alcohol use during early adolescence (ⱕ 14 years), mid
adolescence (15–17 years) and late adolescence (18 –20
years). The ACEs included emotional, physical, and sexual
abuse; emotional and physical neglect; witnessing domestic
violence; parental separation or divorce; and living with
substance abusing, mentally ill, or criminal household mem-
bers. Because ACEs tend to co-occur, we used the total
number of ACEs (ACE score) to assess the impact of
multiple interrelated ACEs on age at alcohol initiation. We
hypothesized that multiple ACEs would increase the likeli-
hood of early initiation of alcohol use consistently in four
birth cohorts dating back to 1900, despite the influence of
social and secular trends on alcohol use over the past 100
years.
Methods
The ACE Study is an ongoing collaboration between the
Kaiser Health Plan’s Health Appraisal Center in San Diego,
California, and the Centers for Disease Control and Preven-
tion (CDC). The overall objective is to assess the impact of
numerous interrelated ACEs on a wide variety of health
behaviors and outcomes [15]. The ACE Study was ap-
proved by the institutional review boards of the Southern
California Permanente Medical Group (Kaiser Permanente),
Emory University, and the Office of Human Research Pro-
tection, Department of Health and Human Services (for-
 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10
merly Office of Protection from Research Risks, National
Institutes of Health).
Study population
The study population included adult members (aged ⱖ
19 years) of the Kaiser Health Plan who underwent a stan-
dardized medical and biopsychosocial examination at
Kaiser’s Health Appraisal Center. In any four-year period,
81% of adult members receive the examination, resulting in
more than 50,000 members being examined annually. The
primary purpose of the evaluation is to perform a complete
health assessment, rather than to provide care based on
symptoms or illness.
The ACE Study sample was drawn from the Health
Appraisal Center and consisted of two survey waves (Wave
I and Wave II). Wave I was conducted among 13,494
consecutive members attending the Health Appraisal Center
between August 1995 and March 1996, and the response
rate was 70% (n ⫽ 9508). Wave II was conducted between
June and October 1997 among 13,330 members and the
response rate was 65% (n ⫽ 8667). The overall response
rate for both waves was 68%.
After their evaluation at the Health Appraisal Center,
members were mailed the ACE Study questionnaire, which
contained detailed information about ACEs and health-
related behaviors from adolescence to adulthood. The Wave
II survey added questions to obtain more detailed informa-
tion about health topics shown to be important from Wave
I data [15,17]. For this study, we used data only from Wave
II because it included a question about age at initiation of
alcohol use that was not part of the Wave I survey.
Exclusions from the study cohort
We excluded three respondents with missing information
about race, 35 with missing information about educational
attainment, and 212 respondents who reported ever drinking
alcohol other than a few sips but did not report their age at
initiation. After these exclusions, the final study cohort
included 8417 (97%) respondents from the Kaiser Health
Plan who underwent a standardized medical and biopsycho-
social examination at the Health Appraisal Clinic. For the
analyses that examined age at initiation, we excluded 898
respondents who reported that they never drank alcohol
(n ⫽ 7519).
Characteristics of the study population
The study population (n ⫽ 8417) included 4521 women
(54 %) and 3896 men (46 %). The mean age (⫾ SD) at
interview was 55 (⫾ 15.1) years for women and 57 (⫾ 14.5)
years for men. Thirty-two percent of women and 42% of
men were college graduates; 42% of women and 39% of
men had some college education; 8% of women and 7% of
men had not graduated from high school. Seventy-four
percent of women and 76% of men were white; 4% of
444.e3
women and 4% of men were black; 12% of women and 10%
of men were Hispanic; 9% of women and 7% of men were
Asian; less than 1% of both genders were Native American.
Definition of adverse childhood experiences
All questions about ACEs pertained to the respondents’
first 18 years of life and have been described in detail
elsewhere [15]. For questions adapted from the Conflict
Tactics Scale (CTS) [22] response categories were “never”,
“once or twice”, “sometimes”, “often”, or “very often”.
Questions used to define emotional and physical neglect
were adapted from the Childhood Trauma Questionnaire
(CTQ) [23]. For both emotional and physical neglect, sets of
five CTQ items were used (Table 1). Response categories
were “never true”, “rarely true”, “sometimes true”, “often
true”, and “very often true”, and were scored on a Likert
scale ranging from 1 to 5, respectively. For emotional ne-
glect, all items were reverse scored, then added. Scores of
15 or higher (moderate to extreme on the CTQ clinical
scale) defined the respondents as having experienced emo-
tional neglect. For physical neglect, items 2 and 5 were
reverse-scored, and all 5 scores were added. Scores of 10 or
higher (moderate to extreme on the CTQ clinical scale)
defined the respondents as having experienced physical ne-
glect [23].
Definition of birth cohorts
Four birth cohort groups were defined based on respon-
dents’ age at entry into the study using 15-year increments,
with the exception of the cohort dating from 1900 to 1931.
Due to small sample sizes in the 1900 –1915 strata (n ⫽
255), we combined 1900 –1915 and 1919 –1931. The four
groups were therefore 1900 –1931, 1932–1946, 1947–1961,
and 1962–1978. These four cohorts have been used in pre-
vious reports from the ACE Study that examined the asso-
ciations of the ACE score with multiple health and behav-
ioral outcomes [18,26].
Age at initiation of alcohol use
Respondents who reported drinking any alcohol other
than a few sips (89%, n ⫽ 7519) answered the following
question to define age at initiation of alcohol use: “How old
were you when you had your first drink of alcohol other
than a few sips?” Respondents’ age was recorded as a
continuous variable. We divided age at initiation of alcohol
use into four categories: early adolescence (ⱕ 14 years),
mid adolescence (15–17 years), late adolescence (18 –20
years), and adulthood (ⱖ 21 years).
Statistical analysis
To examine associations between each of the 10 catego-
ries of ACEs and ever use of alcohol, we used logistic
regression to obtain odds ratios (OR) and 95% confidence
intervals (CI) adjusted for age at entry into the study, gen-
444.e4 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10

Table 1
Definition and prevalence (%) of each category of adverse childhood experience (ACE) by sex
Category of ACE Women Men Total
(n ⫽ 4521) (n ⫽ 3896) (n ⫽ 8417)

Abuse % % %
Emotional 12.4 7.9 10.3
(Did a parent or other adult in the household . . .)
1) Often or very often swear at you, insult you, or put you down?
2) Sometimes, often, or very often act in a way that made you afraid that you might be physically hurt?
Physical 25.5 28.1 26.7
(Did a parent or other adult in the household . . .)
1) Sometimes, often or very often push, grab, slap, or throw something at you?
2) Ever hit you so hard that you had marks or were injured?
Sexual [24] 24.8 17.1 21.3
(Did an adult or person at least 5 years older ever . . .)
1) Touch or fondle you in a sexual way?
2) Have you touch their body in a sexual way?
3) Attempt oral, anal, or vaginal intercourse with you?
4) Actually have oral, anal, or vaginal intercourse with you?
Neglect (CTQ)23
Emotional 16.9 12.5 14.8
1) There was someone in my family who helped me feel important or special
2) I felt loved
3) People in my family looked out for each other
4) People in my family felt close to each other
5) My family was a source of strength and support.
Physical 9.3 10.7 10.0
1) I didn’t have enough to eat
2) I knew there was someone there to take care of me and protect me
3) My parents were too drunk or too high to take care of me
4) I had to wear dirty clothes
5) There was someone to take me to the doctor if I needed it
Household dysfunction
Battered mother 14.0 12.1 13.1
(Was your mother (or step-mother)):
1) Sometimes, often, or very often pushed grabbed, slapped, or had something thrown at her?
2) Sometimes, often, or very often kicked, bitten, hit with a fist, or hit with something hard?
3) Ever repeatedly hit over at least a few minutes?
4) Ever threatened with or hurt by a knife or gun?
Parental discord/divorce 25.8 22.7 24.4
1) Were your parents ever separated or divorced?
Mental illness in household 25.7 14.5 20.5
1) Was a household member depressed or mentally ill? OR
2) Did a household member attempt suicide?
Household substance abuse [25] 30.9 25.7 28.4
1) Live with anyone who was a problem drinker or alcoholic? OR
2) Live with anyone who used street drugs?
Incarcerated household member 7.0 5.0 6.1
1) Did a household member go to prison?

der, race (other vs. white), and education (high school di-
ploma, some college, or college graduate vs. less than high
school). Parameter estimates were obtained by maximum
likelihood techniques and 95% CIs were based on the stan-
dard error of the model coefficients. In addition, the total
number of ACEs was added for each respondent to obtain
the ACE score, with a range of 0 to 10. Because of the small
sample sizes, we combined ACE scores of four or higher to
create an ordinal variable with five categories (0, 1, 2, 3, or
ⱖ 4). Analyses were conducted by entering the ACE score
in regression models as four dichotomous variables (yes/no)
with zero experiences.
As in the analyses for ever use of alcohol, we used
ordinal logistic regression to examine associations between
each of the 10 categories of ACEs and age at initiation. The
ordinal dependent variable was defined by the four age
categories noted earlier (ⱕ 14, 15–17, 18 –20, and ⱖ 21
years). Initiation of alcohol use during adulthood (ⱖ 21
years) was the referent. The ordinal logistic regression mod-
els were adjusted in a similar way as the logistic regression
 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10 444.e5

models described earlier, and the ACE score was entered Table 2
similarly as well. All analyses were completed using SAS Prevalence (%) and adjusted odds ratios for the relationship between
adverse childhood experiences (ACEs), the ACE score, and ever
v8.2 (SAS Institute Inc., Cary, North Carolina). drinking alcohol
Attributable risk fractions (ARFs) for early drinking among
Category of ACE Ever used alcohol
ever drinkers were calculated using adjusted ORs from ordinal
logistic regression models based upon one or more ACEs, with n % OR
zero ACEs as the referent, because a substantial increase in the (95% CI)a
risk of initiating alcohol use by age 14 was seen for persons Abuse
reporting at least one ACE. We used Levin’s formula for these Emotional
calculations: ARF ⫽ P1 (relative risk [RR] ⫺ 1) / 1 ⫹ P1 (RR No 7548 88.9 1.0 (referent)
Yes 869 92.9 1.6 (1.2–2.1)
⫺ 1), where P1 is the prevalence of an ACE score ⱖ 1 and RR
Physical
⫽ OR for early initiation of drinking for an ACE score ⱖ 1 No 6171 87.9 1.0 (referent)
[27]. The ARF is an estimate of the proportion of the health Yes 2246 93.2 1.8 (1.5–2.1)
problem (early alcohol use) that would not have occurred if no Sexual
persons had been exposed to the risk factor being assessed No 6627 88.2 1.0 (referent)
Yes 1790 93.6 2.0 (1.6–2.5)
(i.e., ACEs) [27].
Neglect
Emotional
No 7168 88.6 1.0 (referent)
Results Yes 1249 93.4 1.9 (1.5–2.4)
Physical
Prevalence of adverse childhood experiences and alcohol
No 7579 89.3 1.0 (referent)
use outcomes Yes 838 89.3 1.1 (.9–1.5)
Household dysfunction
The prevalence of each specific adverse childhood expe- Battered mother
rience (ACE) was higher for women than for men, except No 7313 88.8 1.0 (referent)
for physical abuse and physical neglect (Table 1). Sixty- Yes 1104 92.6 1.6 (1.3–2.1)
eight percent of respondents reported at least one of the 10 Parental discord/divorce
ACEs; 42% reported two or more. The prevalence of initi- No 6364 88.0 1.0 (referent)
Yes 2053 93.4 1.8 (1.5–2.2)
ation of alcohol use in early, mid, and late adolescence in Mentally ill household member
the total population was 7.5%, 26.8%, and 32.8%, respec- No 6692 88.4 1.0 (referent)
tively. Among ever drinkers (n ⫽ 7519), the mean age at Yes 1725 93.0 1.7 (1.4–2.1)
initiation (⫾ SD) was 18.7 (⫾ 4.6) years; mean age at Substance abuse in home
initiation was 12.3 (⫾ 2.3), 16.1 (⫾ 0.74), and 18.7 (⫾ No 6022 87.2 1.0 (referent)
Yes 2395 94.6 2.4 (2.0–3.0)
0.82) years for early, mid , and late adolescence, respec- Incarcerated household member
tively. Among ever drinkers, the mean age at initiation of No 7907 89.1 1.0 (referent)
alcohol use by birth cohort was 20.0 (⫾ 5.4), 18.9 (⫾ 4.5), Yes 510 92.2 1.7 (1.2–2.4)
17.8 (⫾ 3.5), and 16.7 (⫾ 3.3) for 1900 –1931, 1932–1946, ACE score
1947–1961, and 1962–1978, respectively. 0 2713 83.9 1.0 (referent)
1 2158 90.0 1.7 (1.4–2.1)
Association between ACEs and ever drinking alcohol 2 1312 91.8 2.1 (1.7–2.7)
3 839 93.2 2.6 (2.0–3.5)
The prevalence of ever drinking was 89%. Each category ⱖ4 1395 94.2 3.2 (2.5–4.1)
of ACE, with the exception of physical neglect, was asso- Total 8417 89.3 —
ciated with a 1.6- to 2.4-fold (p ⬍ .05) increased likelihood a
Odds ratios adjusted for age, gender, race, and educational attainment.
of ever drinking alcohol. The ACE score increased this risk
of ever drinking alcohol in a graded manner. Compared with
respondents who experienced no ACEs, those with four or Results for the relationship between each ACE and initia-
more ACEs were three times more likely to report ever tion of alcohol use in late adolescence (18 –20 years) as
using alcohol (p ⬍ .05) (Table 2). compared with initiation in adulthood were mixed and mod-
Association between ACEs and age at initiation of est. For example, those who had experienced contact sexual
alcohol use abuse were nearly three times (OR 2.8; 95% CI 2.3–3.5)
more likely to initiate alcohol use during early adolescence,
Among ever drinkers, each category of ACE was asso- 1.5 times (95% CI 1.3–1.7) more likely to initiate alcohol
ciated with an increased likelihood of initiating alcohol use use during mid adolescence, and 1.1 times (95% CI .98 –
during early adolescence as compared with initiation during 1.3) more likely to initiate alcohol use during late adoles-
adulthood (Table 3). The same was found for mid adoles- cence than were those who did not experience contact sex-
cence, with the exception of physical neglect (Table 3). ual abuse. The ACE score was associated in a strong graded
444.e6 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10

Table 3
Prevalence (%) and adjusted odds ratios for the relationship between adverse childhood experiences and age at initiation of alcohol use
Category of ACE Age at initiation of alcohol usea

ⱕ 14 years 15–17 years 18–20 years

n % OR % OR % OR
(95% CI)b (95% CI)b (95% CI)b

Abuse
Emotional
No 6712 7.6 1.0 (referent) 29.7 1.0 (referent) 37.3 1.0 (referent)
Yes 807 15.5 2.5 (1.9–3.3) 32.2 1.3 (1.1–1.7) 32.2 1.1 (.92–1.4)
Physical
No 5425 6.6 1.0 (referent) 28.6 1.0 (referent) 38.6 1.0 (referent)
Yes 2094 13.3 2.2 (1.8–2.7) 33.8 1.3 (1.1–1.5) 31.8 1.0 (.86–1.1)
Sexual
No 5843 7.0 1.0 (referent) 29.4 1.0 (referent) 37.8 1.0 (referent)
Yes 1676 13.5 2.8 (2.3–3.5) 32.0 1.5 (1.3–1.7) 33.0 1.1 (.98–1.3)
Neglect
Emotional
No 6352 7.5 1.0 (referent) 29.7 1.0 (referent) 37.4 1.0 (referent)
Yes 1167 13.4 2.1 (1.7–2.7) 31.7 1.2 (1.1–1.5) 33.0 1.1 (.89–1.3)
Physical
No 6771 7.7 1.0 (referent) 29.9 1.0 (referent) 37.3 1.0 (referent)
Yes 748 14.8 2.3 (1.7–3.0) 31.4 1.2 (.97–1.5) 31.4 .95 (.77–1.2)
Household dysfunction
Battered mother
No 6497 7.6 1.0 (referent) 29.7 1.0 (referent) 37.4 1.0 (referent)
Yes 1022 13.8 2.0 (1.5–2.5) 32.2 1.2 (1.0–1.5) 33.0 1.0 (.85–1.2)
Parental discord/divorce
No 5602 6.9 1.0 (referent) 28.8 1.0 (referent) 38.0 1.0 (referent)
Yes 1917 12.9 2.1 (1.7–2.6) 33.7 1.5 (1.2–1.7) 33.0 1.1 (1.0–1.3)
Mentally ill household member
No 5914 7.2 1.0 (referent) 29.8 1.0 (referent) 37.4 1.0 (referent)
Yes 1605 12.9 2.3 (1.8–2.8) 31.0 1.3 (1.1–1.5) 34.3 1.2 (1.0–1.3)
Substance abuse in home
No 5253 6.1 1.0 (referent) 27.6 1.0 (referent) 38.8 1.0 (referent)
Yes 2266 13.9 2.8 (2.3–3.5) 35.7 1.7 (1.5–2.0) 31.9 1.2 (1.1–1.4)
Incarcerated household member
No 7049 7.9 1.0 (referent) 29.9 1.0 (referent) 37.0 1.0 (referent)
Yes 470 16.4 3.0 (2.2–4.2) 32.1 1.6 (1.2–2.2) 33.2 1.4 (1.1–1.8)
Total 7519 8.4 — 30.0 — 36.7 —
a
Among ever drinkers, n ⫽ 7519; referent group for age at initiation are those who reported initiating at ⱖ 21 years.
b
Odds ratios adjusted for age, gender, race, and educational attainment.

manner (p ⬍ .05) with an increased risk of initiating alcohol
use during early and mid adolescence (Table 4). The stron-
gest graded relationship between ACE score and age at
initiation was observed for initiation during early adoles-
cence (ⱕ 14 years) (Table 4).
Association between ACEs and early initiation of alcohol
use by birth cohorts
Because the relationship between ACEs and initiation of
alcohol use by age 14 demonstrated the highest magnitude
of association, a birth cohort analysis was conducted for the
use of alcohol before age 15. For each birth cohort, a graded
relationship was found between the ACE score and initia-
tion of alcohol use by age 14 (p ⬍ .05; Table 5). The
strongest relationship was for persons born between 1947
and 1961; comparison between persons reporting no ACEs
and those reporting four or more ACEs yielded an OR of 8.8
(95% CI 4.9 –15.5) (Table 5). The test for trend between the
ACE score and the likelihood of early initiation of alcohol
use was significant in all four birth cohorts: 1962–1978, 1.3
(95% CI 1.1–1.5); 1947–1961, 1.4 (95% CI 1.3–1.5); 1932–
1946, 1.3 (95% CI 1.2–1.4); and 1900 –1931, 1.4 (95% CI
1.2–1.5) (Table 5). The estimated ARF was 55% for initi-
ating alcohol use by age 14 among ever-drinkers who had
exposure to at least one ACE.
Discussion
Every category of adverse childhood experience (ACE)
increased the likelihood of initiating alcohol use during
 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10 444.e7

Table 4
Prevalence (%) and adjusted odds ratios for the relationship between total number of adverse childhood experiences (ACE score) and age at initiation of
alcohol use
ACE score Age at initiation of alcohol usea

ⱕ 14 years 15–17 years 18–20 years

n % OR % OR % OR
(95% CI)b (95% CI)b (95% CI)b

0 2275 4.2 1.0 (referent) 25.7 1.0 (referent) 40.9 1.0 (referent)
1 1943 6.1 1.5 (1.1–2.1) 28.5 1.2 (1.1–1.4) 38.8 1.1 (0.91–1.2)
2 1205 8.5 2.4 (1.7–3.3) 33.4 1.6 (1.3–1.9) 35.1 1.1 (0.91–1.3)
3 782 12.1 3.9 (2.8–5.6) 36.2 2.0 (1.5–2.5) 31.2 1.1 (0.90–1.4)
ⱖ4 1314 16.9 6.2 (4.6–8.3) 33.0 2.0 (1.6–2.4) 31.1 1.2 (1.1–1.5)
Total 7519 8.4 — 30.0 — 36.7 —
a
Among ever drinkers, n ⫽ 7519; referent group for age at initiation are those who reported initiating at ⱖ 21 years.
b
Odds ratios adjusted for age, gender, race, and educational attainment.

early adolescence (ⱕ14 years). In particular, for each of the
10 categories of ACEs there was a two- to threefold in-
creased likelihood of initiating alcohol use by age 14. More-
over, we found a strong graded relationship between the
ACE score and the likelihood of both ever initiating alcohol
use and beginning alcohol use in early or mid adolescence.
The ARF for early initiation of alcohol use was 55%
among ever-drinkers who had exposure to at least one ACE.
These findings provide evidence that stressful and traumatic
factors in the childhood family environment are strongly
associated with the initiation of alcohol use, particularly
during early to mid adolescence. ACEs influence the use of
alcohol at an early age despite concerted efforts to delay its
use through a variety of educational and public health mes-
sages, and laws prohibiting alcohol use before age 21.
We also found that the ACE score increased the likeli-
hood of initiating alcohol use by age 14 in a graded manner
for each of 4 birth cohorts dating back to 1900. Thus, the
effects of ACEs on the initiation of alcohol use early in
adolescence appear to be a persistent rather than transient or
intermittent influence throughout successive generations.
Our results provide strong evidence that multiple early
stressors and traumatic experiences in the childhood family
environment are strong predictors of early drinking that
have transcended a variety of secular and social influences
on alcohol use over the past century.
The relationship between ACEs and initiation of alcohol
use during late adolescence (18 –20 years) was modest. We
suggest several explanations for this phenomenon. First,
during childhood and early adolescence, children are in
close proximity to the family environment; however, in late
adolescence separation from a family environment charac-
terized by violence, stress, and trauma may decrease the
likelihood of alcohol initiation during that age period. Sec-
ond, as age increases from childhood to adulthood, use of
alcohol becomes increasingly “normative” such that weaker
relationships between ACEs and alcohol use would be ex-
pected during late adolescence and early adulthood. None-
theless, the data suggest that ACEs influence the initiation
of alcohol use throughout adolescence, despite concerted
efforts to delay initiation until adulthood.
The problem of underage drinking has been well docu-

Table 5
Prevalence (%) and adjusted odds ratios for the relationship between total number of adverse childhood experiences (ACE score) and initiation of
alcohol use by age 14, stratified by birth cohort
ACE score Initiation of alcohol use by 14 years of agea

1900–1931 1932–1946 1947–1961 1962–1978

n % OR n % OR n % OR n % OR
(95% CI)b (95% CI)b (95% CI)b (95% CI)b

0 871 2.6 1.0 (referent) 796 3.9 1.0 (referent) 469 4.9 1.0 (referent) 139 13.7 1.0 (referent)
1 666 4.2 1.7 (1.0–3.1) 632 4.7 1.3 (.8–2.3) 453 6.2 1.6 (0.8–3.0) 192 17.2 1.2 (.5–2.7)
2 331 4.8 2.0 (1.0–4.0) 418 6.5 2.1 (1.4–3.3) 324 10.8 3.2 (1.7–6.1) 132 18.9 2.0 (.8–5.2)
3 180 10.0 5.8 (2.9–11.9) 254 8.7 3.1 (1.6–5.8) 244 11.9 3.8 (1.9–7.6) 104 25.0 3.1 (1.2–8.3)
ⱖ4 222 7.2 4.0 (1.9–8.1) 413 11.6 5.1 (3.0–8.7) 504 20.4 8.8 (4.9–15.5) 175 31.4 5.1 (2.1–12.1)
Totalc 2270 4.4 1.4 (1.2–1.5) 2513 6.3 1.3 (1.2–1.4) 1994 10.9 1.4 (1.3–1.5) 742 21.3 1.3 (1.1–1.5)
a
Referent group for age at initiation are those who reported initiating alcohol at ⱖ 21 years.
b
All odds ratios adjusted for sex, race, education and age at survey.
c
Odds ratio in this row represents test for trend, with ACE score as an ordinal variable.
444.e8 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10
mented in recent years. The National Youth Risk Behavior
Survey (YRBS), which is conducted by the CDC among
high school students throughout the United States, reported
that 45% of the adolescent respondents had at least one
drink of alcohol within 30 days before the survey [7].
Furthermore, negative behaviors and outcomes associated
with early onset of drinking are well documented and span
the human life course. For example, adolescents who report
using alcohol also report engaging in other risk behaviors,
such as tobacco and illicit drug use [28,29], violence and
aggression [28,30], sexual risk behaviors [29,30], and sui-
cide ideation [28,29]. Moreover, early onset of drinking is
associated with alcohol dependence later in life [8 –10].
During adolescence, developing youth become vulnera-
ble to influences and stressors in the environment (e.g.,
family and peer influences, academic pressure, media, and
advertising), and within the self (e.g., curiosity and exper-
imentation, low self-esteem). Family environment and ex-
periences, both positive and negative, influence neurodevel-
opment in ways that affect the social, emotional, and
psychological well-being of children [31]. In fact, the fam-
ily environment is the primary influence on a child’s devel-
opment. Stress and trauma, such as experiencing abuse and
witnessing domestic violence in the family, have a strong
impact on neurodevelopment that may lead to early initia-
tion of alcohol use.
Information from the neurosciences also supports the
biologic plausibility of our findings. The biologic processes
that occur when children are exposed to stressful events
such as experiencing recurrent abuse or witnessing domestic
violence can disrupt early development of the central ner-
vous system. For example, early abuse and post-traumatic
stress disorder are associated with increased levels of cor-
tisol and norepinephrine levels in children [32], and in-
creased resting heart rate [33]. These physiologic changes
may impede their ability to cope with negative or disruptive
emotions [31], and lead to problems with emotional and
behavioral self-regulation [34]. Thus, behaviors such as
substance use may emerge as a means to help regulate
emotional states and cope with stress.
Moreover, both human and animal studies have em-
pirically shown that early stress influences alcohol use in
adolescents. Two studies of school-age children found
that physical and sexual abuse increased the likelihood of
adolescent alcohol use [13,14]. In a study of adolescent
alcohol consumption among nonhuman primates, early
negative experiences increased the likelihood that peer-
raised monkeys would consume alcohol when exposed to
stress [35]. Although the latter study was conducted
among nonhuman primates, these findings are consistent
with observational studies in humans. Overall, these
studies provide evidence that stress and trauma are com-
mon factors associated with consumption of alcohol at an
early age as a means to self-regulate negative or painful
emotions.
Potential weaknesses of our study should be ad-
dressed. First, this study specifically assessed stress and
trauma within the family environment and was unable to
examine broader environmental contexts, which may in-
fluence early initiation of alcohol use. Future studies
would benefit from examining not only family level fac-
tors, but other environmental factors related to underage
drinking. Also, our study cannot provide certainty about
the temporal relationship between ACEs and alcohol use
that was initiated before 18 years of age because both the
exposure and outcome were reported as occurring at 18
years or younger. Despite these limitations, the strong
association observed between the ACE score and initia-
tion of alcohol use by age 14 and the consistent relation-
ships in all the birth cohorts dating back to 1900 merit
serious consideration.
The retrospective reporting of childhood experiences
is another potential weakness because respondents may
have difficulty recalling certain events. For example, in
longitudinal follow-up of adults whose childhood abuse
was documented, their retrospective reports of such
abuse were likely to underestimate the actual occurrence
[36,37]. Difficulty recalling childhood events likely re-
sults in misclassification (classifying persons truly ex-
posed to ACEs as unexposed) that would bias our results
toward the null [38]. This would lead to conservative
estimates of the relationships between ACEs and early
alcohol use. To assess this potential effect, we repeated
our analyses after excluding any respondent with missing
information on any one of the ACEs and found no sub-
stantial differences in the final results.
Another potential source of underestimation of the
strength of these relationships is related to the lower
number of childhood exposures and lower prevalence of
early initiation of alcohol use reported by older persons
in our study. This could be an artifact caused by potential
premature mortality in persons with multiple adverse
childhood exposures; the co-occurrence of multiple risk
factors among persons with multiple childhood exposures
is consistent with this hypothesis [15]. Thus, this poten-
tial weakness may have resulted in underestimates of the
true relationships between ACEs and early initiation of
alcohol use [38].
The prevalence of ACEs that we reported is nearly
identical to those reported in surveys of the general
population. We found that 17% of men and 25% of
women met the case definition for childhood contact
sexual abuse, similar to findings by Finkelhor et al [39] of
16% for men and 27% for women. Also, in our study 28%
of the men reported experiencing physical abuse as boys,
which closely parallels the prevalence found (31%) in a
recent population-based study of Ontario men that used
questions from the same scales [40]. The similarity of the
estimates from the ACE study to those of population-
 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10
based studies suggests that our findings are likely to be
applicable in other settings.
In conclusion, our results suggest that efforts to delay
the onset of drinking are likely to benefit from under-
standing the contribution of multiple traumatic and
stressful events in the lives of children. Programs and
interventions must go beyond the approaches that educate
youth about the detriments of substance use and include
approaches designed to help youth recognize and cope
with the all too common stressors of abuse, domestic
violence, and related adverse experiences in childhood.
Acknowledgments
The Adverse Childhood Experiences Study was sup-
ported under a cooperative agreement #TS-44-10/11 from
the Centers for Disease Control and Prevention through the
Association of Teachers of Preventive Medicine (ATPM)
and a grant from the Garfield Memorial Fund. Dr. Dong is
funded through a Career Development Award via ATPM.
References
[1] Burns E. The Spirits of America: A Social History of Alcohol.
Philadelphia, PA: Temple University Press, 2004.
[2] Barr A. Drink: A Social History of America. New York, NY: Carroll
& Graf Publishers, Inc., 1999.
[3] Warren CW, Kann L, Small ML, et al. Age of initiating selected
health-risk behaviors among high school students in the United
States. J Adolesc Health 1997;21:225–31.
[4] US Department of Health and Human Services. Tenth Special Report
to the US Congress on Alcohol and Health. Rockville, MD: US
Department of Health and Human Services, 2000.
[5] Jernigan DH, Ostroff J, Ross C, O’Hara JA. Sex differences in
adolescent exposure to alcohol advertising in magazines. Arch Pediatr
Adolesc Med 2004;158:629 –34.
[6] National Research Council and Institute of Medicine. Reducing Un-
derage Drinking: A Collective Responsibility. Committee on Devel-
oping a Strategy to Reduce and Prevent Underage Drinking. Bonnie
RJ, O’Connell ME, eds. Board on Children, Youth, and Families.
Division of Behavioral and Social Sciences, and Education. Wash-
ington, DC: The National Academics Press, 2004.
[7] Centers for Disease Control and Prevention. Youth Risk Behavior
Surveillance: United States, 2003. MMWR 2004;53(SS-02):1–96.
[8] Hingson R, Heeren T, Zakocs R, et al. Age at first intoxication, heavy
drinking, driving after drinking and risk of unintentional injury
among U.S. college students. J Stud Alcohol 2003;64:23–31.
[9] Grant BF, Stinson FS, Harford TC. Age at onset of alcohol use and
DSM-IV alcohol abuse and dependence: a 12-year follow-up. J Subst
Abuse 2001;13:493–504.
[10] Hawkins JD, Graham JW, Maguin E, et al. Exploring the effects of
age of alcohol use initiation and psychological risk factors on subse-
quent alcohol misuse. J Stud Alcohol 1997;58:280 –90.
[11] Chopak JS, Vicary JR, Crockett LJ. Predicting alcohol and tobacco use
in a sample of rural adolescents. Am J Health Behav 1998;22:334 – 41.
[12] Eaton DK, Forthofer MS, Zapata LB, et al. Factors related to alcohol
use among 6th through 10th graders: the Sarasota County demonstra-
tion project. J Sch Health 2004;74:95–104.
[13] Simantov E, Schoen C, Klein JD. Health-compromising behaviors:
why do adolescents smoke or drink? Arch Pediatr Adolesc Med
2000;154:1025–33.
444.e9
[14] Moran PB, Vuchinich S, Hall NK. Associations between types of
maltreatment and substance use during adolescence. Child Abuse
Negl 2004;28:565–74.
[15] Felitti VJ, Anda RF, Nordenberg D, et al. The relationship of adult
health status to childhood abuse and household dysfunction. Am J
Prev Med 1998;14:245–58.
[16] Hillis SD, Anda RF, Felitti VJ, Marchbanks PA. Adverse childhood
experiences and sexual risk behaviors in women: a retrospective
cohort study. Fam Plann Perspect 2001;33:206 –11.
[17] Anda RF, Croft JB, Felitti VJ, et al. Adverse childhood experiences and
smoking during adolescence and adulthood. JAMA 1999;282:1652– 8.
[18] Dube SR, Anda RF, Felitti VJ, et al. Childhood abuse, household
dysfunction and the risk of attempted suicide throughout the life span:
findings from the Adverse Childhood Experiences Study. JAMA
2001;286:3089 –96.
[19] Dube SR, Felitti VJ, Dong M, et al. Childhood abuse, neglect and
household dysfunction and the risk of illicit drug use: the Adverse
Childhood Experiences Study. Pediatrics 2003;111:564 –72.
[20] Anda RF, Chapman DP, Felitti VJ, et al. Adverse childhood experiences
and risk of paternity in teen pregnancy. Obst Gynecol 2002;100:37– 45.
[21] Hillis SD, Anda RF, Dube SR, et al. The association between ado-
lescent pregnancy, long-term psychosocial consequences, and fetal
death. Pediatrics 2004;113:320 –7.
[22] Straus M, Gelles RJ. Physical Violence in American Families:
Risk Factors and Adaptations to Violence in 8,145 Families. New
Brunswick, NJ: Transaction Press, 1990.
[23] Bernstein DP, Fink L, Handelsman L, et al. Initial reliability and
validity of a new retrospective measure of child abuse and neglect.
Am J Psychiatry 1994;151:1132– 6.
[24] Wyatt GE. The sexual abuse of Afro-American and white American
women in childhood. Child Abuse Negl 1985;9:507–19.
[25] Schoenborn CA. Exposure to alcoholism in the family: United States,
1988. Adv Data 1991;205:1–13.
[26] Dube SR, Felitti VJ, Dong M, et al. The impact of adverse childhood
experiences on health problems: evidence from four birth cohorts
dating back to 1900. Prev Med 2003;37:268 –77.
[27] Haddix AC, Teutsch SM, Shaeffer PA, Dunet DO. Prevention Effec-
tiveness: A Guide to Decision Analysis and Economic Evaluation.
New York, NY: Oxford University Press, 1996.
[28] Durant RH, Smith JA, Kreiter SR, Krowchuk DP. The relationship
between early age of onset of initial substance use and engaging in
multiple health risk behaviors among young adolescents. Arch Pediatr
Adolesc Med 1999;153:286 –91.
[29] Windle M. Alcohol use among adolescents and young adults. Alcohol
Res Health 2003;27:79 – 85.
[30] Orpinas PK, Basen-Engquist K, Grunbaum JA, Parcel GS. The co-
morbidity of violence-related behaviors with health-risk behaviors in a
population of high school students. J Adolesc Health 1995;16:216 –25.
[31] Perry BD, Pollard R. Homeostasis, stress, trauma, and adaptation-a
neurodevelopmental view of childhood trauma. Child Adolesc Psy-
chiatr Clin N Am 1998;7:33–51.
[32] De Bellis MD, Baum AS, Birmaher B, et al. A.E. Bennett Research
Award. Developmental traumatology. Part I: Biological stress sys-
tems. Biol Psychiatry 1999;45:1259 –70.
[33] Perry BD. The neurodevelopmental impact of violence in childhood.
In: Schetky D, Benedek EP, eds. Textbook of Child and Adolescent
Forensic Psychiatry. Washington, DC: American Psychiatric Press,
Inc., 2001:221–38.
[34] Van der kolk BA, Perry JC, Herman JL. Childhood origins of self-
destructive behavior. Am J Psychiatry 1991;148:1665–71.
[35] Barr CS, Schwandt ML, Newman TK, Higley JD. The use of ado-
lescent nonhuman primates to model human alcohol intake: neurobi-
ological, genetic, and psychological variables. Ann N Y Acad Sci
2004;1021:221–33.
444.e10 S.R. Dube et al. / Journal of Adolescent Health 38 (2006) 444.e1– 444.e10

[36] Della Femina D, Yeager CA, Lewis DO. Child abuse: adolescent
records vs. adult recall. Child Abuse Negl 1990;14:227–31.
[37] Williams LM. Recovered memories of abuse in women with docu-
mented child sexual victimization histories. J Trauma Stress 1995;8:
649 –73.
[38] Rothman KJ. Modern Epidemiology. Boston, MA: Little, Brown, 1986.
[39] Finkelhor D, Hotaling G, Lewis IA, Smith C. Sexual abuse in a
national survey of adult men and women: prevalence, characteristics,
and risk factors. Child Abuse Negl 1990;14:19 –28.
[40] MacMillan HL, Fleming JE, Trocme N, et al. Prevalence of child
physical and sexual abuse in the community: results from the Ontario
Health Supplement. JAMA 1997;278:131–5.