Cocaine
Use
and
Characteristics
of
Young
Adult
Users
from
1987
to
1992:
The
CARDIA
Study
Barbara
L.
Braun,
PhD,
David
Murray,
PhD,
Peter
Hannan,
MStat,
Stephen
Sidney,
MD,
MPH,
and
Chap
Le,
PhD
Introduction
Despite
a
decline
in
cocaine
use
prevalence
in
the
United
States
from
1987
to
1992,1-5
the
problems
related
to
cocaine
use
are
significant
and
appear
to
be
increasing
in
magnitude.46
For
example,
cocaine
has
continued
to
be
strongly
implicated
in
medical
emergencies,7
in
drug
abuse
deaths,8'9
and
in
violent
behav-
ior
and
violent
crime.''4
Previous
studies
have
identified
fac-
tors
associated
with
cocaine
use
and
abuse
in
young
adults
(e.g.,
use
of
other
illicit
drugs
in
both
adulthood'5-7
and
adolescence'8).
Cocaine
use
among
young
adults
is
negatively
associated
with
educa-
tional
level,
is
higher
among
those
who
are
unemployed,
and
has
been
shown
to
be
higher
among
Blacks
than
Whites
more
than 35
years
of
age
but
higher
among
Whites
than
Blacks
at
younger
ages.6
Young
adult
cocaine
users
are
more
likely
to
have
unstable
societal
roles,
reflected
in
higher
divorce
rates,
less
job
stability,
and
more
criminal
convictions.15
They
are
less
likely
to
have
been
married
and
more
likely
to
have
lived
or
to
currently
live
with
a
nonmarital
partner.'5
Although
cocaine
use
has
been
stud-
ied,
there
are
methodological
limitations
in
the
published
data
and
understudied
aspects
of
use.
Previous
studies
have
been
characterized
by
lack
of
ethnic
diversity,
limited
age
groupings,
few
women,
and
an
inability
to
assess
the
simultaneous
effect
of
factors
such
as
ethnicity,
sex,
and
age.
Data
collection
has
often
involved
face-to-
face
or
telephone
interviews;
such
meth-
ods
have
been
reported
to
increase
the
likelihood
of
underreporting
in
compari-
son
with
more
anonymous
methods
such
as
self-administered
questionnaires.'9
Adult
risk
factors
for
current
cocaine
use
may
differ
from
adolescent
predictors,
and
adults
26
years
of
age
or
older
may
have
different
characteristics
than
the
younger
adults
surveyed
in
previous
stud-
ies.
In
addition,
characteristics
of
cocaine
users
may
have
changed
in
recent
years.
The
Coronary
Artery
Risk
Develop-
ment
in
Young
Adults
study
(CARDIA)
provided
an
opportunity
to
address
co-
caine
use
without
some
of
the
limitations
of
previous
studies.
The
goal
of
CARDIA
was
to
identify
and
monitor
the
distribu-
tion
of
coronary
heart
disease
risk
factors
(including
illicit
drug
use)
in
a
randomly
sampled
biethnic
cohort
of
men
and
women
of
varied
socioeconomic
status
who
were
20
to
32
years
of
age
in
1987.
The
CARDIA
data
set
also
includes
unique
data
on
lifestyle
and
behavioral
characteristics
that
may
influence
cocaine
use.
This
unique
longitudinal
cohort
study
allowed
us
to
assess
cocaine
use
by
sociodemographic
characteristics
and
to
assess
the
simultaneous
effect
of
these
and
other
factors
in
relation
to
cocaine
use.
The
purpose
of
the
present
study
was
to
determine
the
proportion
of
cocaine
users
in
this
cohort
and
to
evaluate
the
relationship
of
cocaine
use
to
sociodemo-
graphic
factors
and
other
substance
use
At
the
time
of
the
study,
Barbara
L.
Braun,
David
Murray,
and
Peter
Hannan
were
with
the
Division
of
Epidemiology,
University
of
Minnesota
School
of
Public
Health,
Minneapo-
lis;
Dr
Braun
is
now
with
Duluth
Clinic,
Duluth,
Minn.
Stephen
Sidney
is
with
the
Kaiser-Permanente
Division
of
Research,
Oak-
land,
Calif.
Chap
Le
is
with
the
Division
of
Biostatistics,
University
of
Minnesota
School
of
Public
Health.
Requests
for
reprints
should
be
sent
to
Barbara
L.
Braun,
PhD,
Duluth
Clinic,
Divi-
sion
of
Education
and
Research,
400
E
Third
St,
Duluth,
MN
55805.
This
paper
was
accepted
April
4,
1996.
December
1996,
Vol.
86,
No.
12
Young
Adult
Cocaine
Use
behavior
over
the
5-year
period
1987
to
1992.
Methods
The
data
used
in
this
study
were
collected
at
the
second
and
fourth
exami-
nations
of
the
CARDIA
study.
From
March
1985
through
June
1986,
5115
Black
and
White
men
and
women
18
to
30
years
of
age
were
recruited
and
examined
at
one
of
four
clinical
sites:
Birmingham,
Ala;
Chicago,
Ill;
Minneapolis,
Minn;
and
Oakland,
Calif.
In
Birmingham,
Chicago,
and
Minneapolis,
participants
were
ran-
domly
recruited
from
the
community
or
from
selected
census
tracts.
At
the
Oak-
land
center,
participants
were
randomly
recruited
from
the
Kaiser-Permanente
health
plan
membership.
Recruitment
methods
effectively
achieved
the
study
aim
of
a
cohort
approximately
balanced
on
two
levels
of
age
(24
years
old
or
less
in
1985
and
more
than
24
years
old
in
1985),
ethnicity
(Black
and
White),
sex,
and
education
level
(high
school
or
less
or
greater
than
high
school
in
1985).
Details
on
sampling
methods
and
clinic
protocol
have
been
published
previously.2022
Par-
ticipants
were
reexamined
in
1987/88
(retention
rate:
90.4%),
1990/91
(reten-
tion
rate:
85.1%),
and
1992/93
(retention
rate:
80.6%).
Study
Sample
The
4624
subjects
who
attended
the
second
CARDIA
examination
are
the
focus
of
this
report.
Subjects
were
ex-
cluded
from
all
analyses
if
they
did
not
attend
the
fourth
examination
or
if
they
had
missing
cocaine
use
data
at
either
examination.
Cocaine
use
data
were
miss-
ing
for
52
participants
in
1987.
Thirty-six
subjects
died
from
1987
to
1992,
and
662
participants
did
not
attend
the
1992
examination.
Cocaine
use
data
were
miss-
ing
for
26
participants
in
1992.
Thus,
the
final
sample
included
3848
participants.
Sample
sizes
in
the
analyses
reported
here
may
be
slightly
smaller
as
a
result
of
other
occasional
missing
data.
Ascertainment
of
Cocaine
Use
Beginning
in
1987,
the
self-adminis-
tered
illicit
drug
use
questionnaire,
mod-
eled
after
the
National
Household
Survey
of
Drug
Abuse,23
included
questions
about
cocaine
use.
Current
cocaine
users
were
those
reporting
at
least
1
day
of
cocaine
use
in
the
previous
month,
a
generally
accepted
method
for
determin-
ing
current
use
of
an
illicit
substance.
Sociodemographic
Characteristics
Sociodemographic
information
was
collected
at
each
examination.
Level
of
education
was
calculated
as
a
semicontinu-
ous
variable
ranging
from
1
year
to
20
or
more
years.
Marital
status
at
each
exami-
nation
was
classified
into
four
mutually
exclusive
categories:
married,
divorced,
living
in
a
marriage-like
relationship,
and
single.
Employment
was
evaluated
in
terms
of
(1)
the
prevalence
of
full-
or
part-time
employment,
(2)
the
occupation
of
partici-
pants
employed
full
or
part
time,
and
(3)
a
variable
reflecting
adoption
of
more
tradi-
tional
adult
social
roles.
Occupation
was
coded
in
accordance
with
theAlphabetical
Index
of
Industries
and
Occupations,
1980
Census
ofPopulation,
Final
Edition,
Novem-
ber
198224
and
the
Classified
Index
of
Industnes
and
Occupations.25
The
participant's
self-reported
main
daily
activities
hierarchically
determined
three
mutually
exclusive
categories
reflect-
ing
adoption
of
more
traditional
adult
social
roles.
Full-time
employment
was
the
first
employment
consideration.
The
second
level
required
a
report
of
no
full-time
employment
and
one
or
more
of
the
following
activities:
in
school,
working
part
time,
raising
children,
and
homemak-
ing.
The
third
level
required
self-reported
unemployment
and
activities.
none
of
the
other
Substance
Use
Behavior
Licit
and
illicit
substance
use
data
were
collected.
Tobacco
smoking,
a
yes-no
variable,
was
defined
as
smoking
at
least
five
cigarettes
a
week
almost
every
week.
Alcohol
consumption
was
classified
as
(1)
prevalence
of
past-year
drinking,
(2)
mean
ethanol
intake
per
week,
and
(3)
a
dichotomous
variable
reflecting
high
and
low
alcohol
intake.
Past-year
drinkers
were
those
who
reported
consuming
any
alcoholic
beverage
within
the
previous
year.
Ethanol
intake
was
calculated
with
the
following
formula25:
mean
mL
ethanol/
day
=
(usual
number
of
12-oz
beers
per
week/7
x
14.2
mL
ethanol/beer)
+
(usual
number
of
5-oz
glasses
of
wine
per
week/7
x
17.2
mL
ethanol/wine)
+
(usual
number
of
1.5-oz
hard
liquor
drinks
per
week/7
x
19.09
mL
ethanol/drink).
High
alcohol
intake
was
defined,
for
each
ethnicity-sex
group,
as
greater
than
the
75th
percentile
of
alcohol
consumed
per
day
among
those
consuming
at
least
one
drink
per
week
(approximately
two
beers
per
day
for
men
and
approximately
one
beer
per
day
for
women).
Current
mari-
juana
use
was
classified
as
at
least
1
day
of
use
in
the
past
month;
at
least
1
day
of
amphetamine
or
heroin
use
in
the
past
American
Journal
of
Public
Health
1737
TABLE
1
-Prevalence
of
Cocaine
Use
in
1987
and
1992
by
Ethnicity-Sex
Group:
The
Coronary
Artery
Risk
Development
In
Young
Adults
(CARDIA)
Study
Difference
in
Examination
2:1987
Examination
4:1992
Prevalence
of
Ethnicity
Current
Use,
and
Use
Status
No.
%
No.
%
%
(%
Change)
Men
Black
Current
95
12.5
64
8.4
-4.10
(32.8)
Ever
172
22.6
251
33.0
Never
494
64.9
446
58.6
White
Current
87
9.0
20
2.1
-6.90
(76.7)
Ever
387
40.1
487
50.4
...
Never
492
50.9
459
47.5
Women
Black
Current
63
6.0
39
3.7
-2.30
(38.3)
Ever
188
17.9
248
23.6
...
Never
799
76.1
763
72.7
...
White
Current
54
5.0
13
1.2
-3.80
(76.0)
Ever
425
39.7
489
45.7
...
Never
592
55.3
569
53.1
...
December
1996,
Vol.
86,
No.
12
Braun
et
al.
month
defined
current
use
of
other
illicit
drugs.
Data
Analysis
The
proportion
of
self-reported
co-
caine
users
in
the
cohort
was
examined
in
1987
and
in
1992.
The
consistency
be-
tween
self-reported
cocaine
use
status
in
1987
and
1992
was
evaluated
with
a
phi
coefficient
(SAS
PROC
CORR27).
The
bivariate
relationship
between
each
socio-
demographic
characteristic
or
substance
use
behavior
and
cocaine
use
status
was
assessed
cross
sectionally
in
1987
and
in
1992
by
means
of
chi-square
(SAS
PROC
FREQ27)
or
linear
regression
(SAS
PROC
GLM27)
analysis
techniques.
Cross-sec-
tional
log
linear
modeling
(SAS
PROC
CATMOD27)
was
used
to
determine
two-
and
three-way
interactions
between
the
sociodemographic
variables,
the
sub-
stance
use
variables,
and
cocaine
use
status.
The
multivariate
relationship
of
the
sociodemographic
characteristics
and
the
substance
use
behaviors
to
current
cocaine
use
from
1987
to
1992
was
analyzed
longitudinally
through
repeated
measures
analysis
of
variance
tech-
niques28
via
the
GLIMMIX
macro
for
SAS
PROC
MIXED29;
this
procedure
allows
differentiation
of
within-person
effects
and
between-person
effects.
A
final
repeated
measures
model
composed
of
those
literature-based
factors
and
inter-
actions
related
to
current
cocaine
use
was
identified.
The
type
I
error
rate
was
established
a
priori
at
.05;
however,
each
of
the
eight
terms
in
the
final
model
had
a
P
value
below
.006.
Thus,
with
the
exception
of
the
Time
x
Employment
interaction
(P
<
.007),
each
of
the
terms
would
have
been
significant
even
under
the
conservative
Bonferroni
criterion
of
.05/8
=
.0063.30
As
a
result
of
the
number
of
time-related
interactions,
odds
ratios
(ORs)
and
confidence
intervals
(CIs)
were
calculated
from
logistic
regression
models
stratified
by
examination
date
(SAS
PROC
LOGISTIC27).
Results
Self-reported
current
cocaine
use
among
CARDIA
participants
declined
between
the
second
and
fourth
examina-
tions
across
all
ethnicity-sex
groups
(Table
1).
Although
the
percentage
decline
was
greater
among
Whites
than
among
Blacks,
higher
never-use
rates
were
noted
in
Blacks
relative
to
Whites
and
in
women
relative
to
men
in
both
1987
and
1992.
The
phi
coefficient
between
self-
reported
cocaine
use
in
1987
and
1992
was
.37,
suggesting
a
moderate
level
of
indi-
vidual
change
in
cocaine
use
status
over
the
5
years.
Eighty
participants
reported
current
cocaine
use
at
both
points;
3493
reported
nonuse
at
both
points.
Fifty-six
participants
reported
a
change
from
non-
use
to
current
use,
and
219
reported
a
change
from
current
use
to
nonuse.
Among
current
cocaine
users
in
1992,
58.8%
were
current
cocaine
users
in
1987
(80/136).
The
overlap
between
current
cocaine
use
in
1987
and
1992
varied
across
ethnicity-sex
groups.
Among
Black
men,
56.3%
of
current
users
in
1992
were
also
current
users
in
1987.
The
corresponding
overlap
rates
among
Black
women,
White
men,
and
White
women
were
46.2%,
80.0%,
and
76.9%.
Bivariate
relationships
between
cur-
rent
cocaine
use
and
sociodemographic
characteristics
or
licit
and
illicit
substance
use
in
1987
and
1992
are
presented
in
Table
2.
Results
were
similar
across
ethnicity-sex
groups;
therefore,
data
are
presented
in
aggregate
form.
In
1987
and
1992,
higher
proportions
of
current
co-
caine
users
than
nonusers
were
Black
and
male.
Cocaine
users
at
both
time
points
were
less
educated
than
nonusers,
al-
though
a
mean
educational
level
of
at
least
12
years
was
noted.
No
age
differ-
ences
were
noted
by
cocaine
use
status.
Lower
proportions
of
cocaine
users
than
nonusers
were
employed
full
time
and
married;
higher
proportions
were
unemployed
and
single
(Table
2).
Among
cocaine
users,
76.9%
and
62.5%
were
employed
either
full
or
part
time
in
1987
and
1992,
respectively;
among
nonusers,
the
corresponding
proportions
were
89.3%
and
92.4%
(data
not
shown
in
Table
2).
The
relationship
between
professional
1738
American
Journal
of
Public
Health
TABLE
2-Sociodemographic
Characteristics
and
Substance
Use
in
1987
and
1992,
by
Cocaine
Use
Status:
The
Coronary
Artery
Risk
Development
in
Young
Adults
(CARDIA)
Study
1987
1992
Nonuse
Current
Use
Nonuse
Current
Use
(n
=
3549)
(n
=
299)
(n
=
3712)
(n
=
136)
Ethnicity,
%
White
53.4
47.2
54.0
24.3
Black
46.6
52.8
46.0
75.7
Sex,
%
Male
43.5
60.9
44.3
61.8
Female
56.5
39.1
55.7
38.2
Age,
%
Youngera
42.1
41.8
42.2
39.7
Olderb
57.9
58.2
57.8
60.3
Mean
education
level,
y
14.3
13.3
14.7
12.9
Employment,
%
Employed
full
time
68.9
60.9
72.1
48.9
Employed
part
time,
stu-
28.9
32.0
26.0
37.0
dent,
or
homemaker
Unemployed
2.2
7.1
1.9
14.1
Marital
status,
%
Married
32.7
18.5
45.6
18.4
Marriage-like
relation-
12.8
19.1
12.9
22.1
ship
Divorced
4.8
3.4
7.1
8.8
Single
49.8
59.1
34.4
50.7
Substance
use
Current
drinkers,
%
84.0
97.6
81.2
94.1
Mean
alcohol
intake,
10.1
33.2
10.2
34.9
mL/d
Smoking
prevalence,
%
25.5
58.9
24.8
71.3
Past-month
illicit
drug
use,
%
Marijuana
19.4
79.6
13.3
71.3
Heroin
or
speed
0.93
7.0
0.54
8.1
Note.
All
differences
by
cocaine
use
status,
except
for
age,
were
significant
(P
<
.05).
a2Q.26
years
old
in
1987.
b27-32
years
old
in
1987.
December
1996,
Vol.
86,
No.
12
Young
Adult
Cocaine
Use
occupational
status
and
cocaine
use
dif-
fered
by
ethnicity-sex
group.
Among
Blacks,
cocaine
use
was
associated
with
nonprofessional
occupations
in
both
1987
and
1992.
Among
Black
men,
11.0%
of
cocaine
users
and
17.3%
of
nonusers
reported
professional
status
in
1987.
In
1992,17.5%
of
current
users
and
23.1%
of
nonusers
reported
professional
occupa-
tions.
In
1987,
17.2%
of
Black
female
nonusers
and
9.4%
of
users
reported
professional
occupations;
the
correspond-
ing
percentages
in
1992
were
25.2%
and
0.0%.
Among
White
men,
occupational
status
was
not
related
to
cocaine
use;
40.3%
of
nonusers
and
20.0%
of
users
in
1987,
and
48.3%
of
nonusers
and
31.6%
of
users
in
1992,
reported
professional
occupations.
Among
White
women,
how-
ever,
cocaine
use
was
associated
with
nonprofessional
occupational
status.
Among
nonusers,
40.0%
in
1987
and
46.9%
in
1992
reported
professional
occu-
pations;
35.6%
and
28.6%
of
users
in
1987
and
1992,
respectively,
reported
such
occupations.
The
proportions
reporting
current
drinking,
smoking,
or
other
illicit
drug
use
were
higher
in
current
cocaine
users
than
in
nonusers.
Cocaine
users
also
reported
a
higher
mean
daily
alcohol
intake
than
nonusers.
The
multivariate,
longitudinal
rela-
tionship
between
current
cocaine
use
and
sociodemographic
characteristics
and
sub-
stance
use
behavior
was
also
examined.
Interactions
in
the
prediction
of
current
cocaine
use
were
found
between
ethnicity
and
age
and
between
sex
and
age.
No
other
two-
or
three-way
interactions
were
identified.
In
1987,
Blacks
27
to
32
years
old
were
more
likely
to
be
cocaine
users
than
Whites
or
younger
Blacks
(Table
3).
In
1992,
however,
Blacks
were
3.46
(95%
CI
=
2.21,
5.41)
times
as
likely
as
Whites
to
be
cocaine
users,
regardless
of
age
(data
were
derived
from
a
simpler
model).
From
1987
to
1992,
cocaine
use
increased
by
10%
among
Blacks
(OR
=
1.10,
95%
CI
=
0.68,
1.77)
but
declined
by
63%
among
Whites
(OR
=
0.37,
95%
CI
=
0.22, 0.63).
Men
in
the
older
half
of
the
cohort
(i.e.,
those
27
to
32
years
of
age
in
1987)
were
more
likely
to
be
cocaine
users
than
women
or
men
in
the
younger
half
of
the
cohort
(i.e.,
those
20
to
26
years
of
age
in
1987)
at
each
examination.
Employment
was
related
to
current
cocaine
use
at
both
examination
points
(Table
3).
The
odds
ratios
for
current
cocaine
use
among
the
unemployed,
in
comparison
with
those
employed
full
time,
were
2.42
in
1987
and
4.00
in
1992.
Homemakers,
those
employed
part
time,
and
those
in
school
were
no
more
likely
to
be
cocaine
users
than
those
employed
full
time.
Current
cocaine
use
increased
3.23-
fold
(95%
CI
=
1.38,
7.57)
among
the
unemployed.
Single
people
were
more
likely
to
be
current
cocaine
users
than
married
people
(ORs
=
1.73
in
1987
and
2.11
in
1992).
Those
divorced
or
living
in
a
marriage-like
relationship
in
1987
were
no
more
likely
to
be
cocaine
users
than
married
people,
but
living
in
a
marriage-like
relationship
was
associated
with
increased
cocaine
use
in
1992
(Table
3).
Longitudinal
analysis,
however,
suggested
no
change
in
the
overall
relationship
between
marital
sta-
tus
and
cocaine
use
from
1987
to
1992.
The
licit
and
other
illicit
substance
use
behaviors
were
strongly
and
consis-
tently
related
to
current
cocaine
use
in
both
1987
and
1992
(Table
3).
Smoking
status
and
past-month
use
of
amphet-
amines
or
heroin
were
related
to
in-
creased
odds
of
current
cocaine
use.
Past-
month
marijuana
use,
the
strongest
predic-
tor
of
current
cocaine
use
status,
inter-
acted
with
alcohol
intake
and
increased
the
odds
of
current
cocaine
use.
The
strength
of
the
relationship
between
alco-
hol
and
marijuana
use
and
cocaine
use
declined
between
1987
and
1992,
al-
American
Journal
of
Public
Health
1739
TABLE
3-Odds
Ratios
for
Current
Cocaine
Use
In
1987
and
1992
from
Multivariate
Logistic
Regression
Models:
The
Coronary
Artery
Risk
Development
In
Young
Adults
(CARDIA)
Study
1987
1992
Odds
95%
Confidence
Odds
95%
Confidence
Ratio
Interval
Ratio
Interval
Ethnicity
Black
Youngera
1.00
...
1.00
...
Oldera
2.43
1.54,
3.85
3.13
1.70,
5.76
White
Youngera
1.32
0.88,1.99
0.41
0.21,
0.82
Oldera
1.47
0.93,
2.33
0.71
0.35,1.43
Sex
Male
Younger
1.00
...
1.00
...
Older
2.43
1.54,
3.85
3.13
1.70,
5.76
Female
Younger
0.98
0.65,1.49
1.25
0.67,
2.32
Older
1.27
0.78,
2.06
1.40
0.73,
2.72
Employment
status
Employed
full
timea
1.00
...
1.00
...
Employed
part
time,
1.21
0.87,1.69
1.10
0.64,1.89
student,
or
homemakera
Unemployeda
2.42
1.27,
4.64
4.00
1.77,
9.03
Marital
status
Married
1.00
...
1.00
...
Divorced
0.95
0.45,
2.04
1.64
0.74,
3.65
Marriage-like
relationship
1.37
0.89,
2.11
2.11
1.16,
3.87
Single
1.73
1.20,
2.48
2.11
1.24,
3.60
Smoking
status
(yes
vs
no)
1.74
1.32,
2.30
2.80
1.83,
4.28
Marijuana
and
alcohol
use
No
past-month
marijuana
use
Low
alcohol
intakea
1.00
...
1.00
...
High
alcohol
intakea
6.35
3.72,10.84
2.18
1.10,
4.32
Past-month
marijuana
use
Low
alcohol
intakea
18.06
11.54,
28.28
9.79
5.64,17.00
High
alcohol
intakea
34.88
21.89,
55.62
18.67
10.87,
32.07
Past-month
heroin
or
speed
2.41
1.28,
4.52
8.10
3.03,
21.70
use
(yes
vs
no)a
Note.
The
younger
age
group
comprised
participants
who
were
20-26
years
old
in
1987;
the
older
age
group
comprised
those
who
were
27-32
years
old
in
1987.
Alcohol
intake
was
divided
at
the
75th
percentile
among
drinkers
consuming
at
least
one
drink
per
week
(ethnicity-sex
specific).
aObserved
relationship
with
cocaine
use
changed
from
1987
to
1992.
December
1996,
Vol.
86,
No.
12
Braun
et
al.
though
alcohol
use
and
marijuana
use
both
remained
strong
predictors.
Discussion
Previous
studies
have
suggested
that
cocaine
use
is
declining
among
young
adults.'-
Such
studies
have
also
shown
that
use
of
cocaine
is
associated
with
other
licit
and
illicit
drug
use,
lower
educational
levels,
unemployment,
ethnic-
ity,
being
divorced
or
living
in
a
marriage-
like
relationship,
and
criminal
activ-
ity.6,10,11,13-18
These
investigations
have
suffered
from
methodological
issues
such
as
limited
representation
by
sex,
ethnicity,
and
age
or
an
inability
to
simultaneously
evaluate
the
effects
of
sociodemographic
characteristics
and
substance
use
behav-
ior.
The
present
study
provided
an
oppor-
tunity
to
simultaneously
assess
these
fac-
tors
in
a
cohort
of
Black
and
White
men
and
women
20
to
32
years
of
age
in
1987,
sampled
to
provide
adequate
representa-
tion
by
sex
and
ethnicity.
Consistent
with
the
extant
adoles-
cent
and
young
adult
literature
,6,15-18
bivariate
analyses
in
this
study
showed
that
a
higher
proportion
of
current
co-
caine
users
than
nonusers
were
Black,
unemployed,
single,
and
had
a
low
educa-
tional
level.
Also,
higher
proportions
reported
licit
and
other
illicit
substance
use.
These
data
suggest
that
cocaine
use
in
adults
20
to
37
years
old
continues
to
be
associated
with
less
adherence
to
societal
norms
of
lowered
levels
of
substance
use
among
adults
and
less
adoption
of
adult
roles
such
as
marriage
and
employment.
Our
multivariate
analyses,
however,
revealed
some
important
differences
sug-
gesting
that
limited
ethnic
or
sex
represen-
tation
and
the
lack
of
simultaneous
analyses
have
provided
a
somewhat
dis-
torted
picture
of
cocaine
use
in
young
adults.
After
adjustment
for
the
other
terms
in
the
model,
no
decline
in
cocaine
use
was
observed
among
Blacks,
while
a
marked
decline
was
noted
in
Whites;
similar
levels
of
cocaine
use
were
seen
in
men
and
women
born
from
1962
to
1967,
while,
in
those
born
from
1955
to
1961,
higher
levels
of
use
were
reported
by
men
than
women.
Increased
cocaine
use
contin-
ued
to
be
associated
with
single
marital
status
and,
in
1992,
with
living
in
a
marriage-like
relationship;
higher
levels
of
alcohol,
marijuana,
and
other
illicit
drug
use;
smoking;
and
unemployment.
The
relationship
between
unemployment
and
cocaine
use
had
strengthened
signifi-
cantly
by
1992.
The
results
of
this
study
must
be
interpreted
in
light
of
three
design
fea-
tures:
the
analysis
plan,
the
subjects
included,
and
the
reliance
on
self-
reported
data.
A
multisequential
analysis
plan
was
established
a
priori.
The
likeli-
hood
of
a
type
I
error
increases
with
multiple
analyses,
and
a
Bonferroni
correc-
tion,
while
considered
conservative,30
may
be
applied
to
reduce
the
likelihood
of
a
type
I
error.
All
terms
included
in
the
final
model
met
the
criteria
for
inclusion
under
this
stringent
correction
except
for
the
Time
x
Employment
interaction
(P
<
.007),
suggesting
low
probability
of
a
type
I
error.
The
CARDIA
participants
were
more
likely
to
be
stable
individuals
than
the
general
population
since
they
were
followed
for
7
years
and
were
able
to
comprehend
and
comply
with
a
4-hour
examination.
No
differential
loss
to
fol-
low-up
was
noted
in
1992
by
cocaine
use
status
in
1987.
Thus,
these
participants
were
not
likely
to
have
suffered
the
extreme
effects
of
chronic,
heavy
drug
use.
Finally,
the
validity
and
reliability
of
self-reported
substance
use
data
have
been
questioned.
Self-reported
illicit
drug
data
have
been
viewed
as
underrepresent-
ing
the
prevalence
of
drug
use
as
a
result
of
hesitancy
in
admitting
illegal,
socially
unacceptable
behaviors
and
poor
recall
of
sporadic
drug
use.31'32
Several
features
of
the
CARDIA
study
design
helped
to
minimize
underreporting
of
illicit
drug
use.
For
example,
a
self-administered
questionnaire,
rather
than
a
face-to-face
or
telephone
interview,
was
used
to
assess
substance
use,
and
the
examination
site
was
an
isolated
clinic
not
associated
with
any
law
enforcement
agency,
governmen-
tal
agency,
employer,
or
medical
care
facility.
In
addition,
only
8.0%
of
the
sample
had
discrepant
drug
use
reports
from
1987
to
1992.
Of
these
individuals,
82.9%
reported
10
or
fewer
lifetime
experiences
with
cocaine
as
of
1987,
suggesting
that
discrepant
reporting
was
the
result
of
inaccurate
recall
of
low
usage.
Nevertheless,
findings
in
this
study
may
be
attenuated
as
a
result
of
underre-
porting
of
cocaine
use.
The
results
of
this
study
provide
important
data
for
the
prevention
and
control
of
cocaine
use
among
young
adults.
Cocaine
initiation
and
cessation
occurred
at
a
moderate
level,
suggesting
that
prevention
and
control
efforts
may
be
warranted
in
this
age
group.
The
high
never-use
and
ever-use
rates
in
both
1987
and
1992,
however,
suggest
that
preven-
tion
efforts
may
be
most
efficient
if
directed
toward
those
at
high
risk
of
cocaine
use.
Identification
of
adults
likely
to
use
cocaine
is
possible;
in
the
present
study,
factors
associated
with
current
cocaine
use
were
essentially
stable
from
1987
to
1992.
In
1992,
more
traditional
adult
roles
such
as
marriage,
employment,
and
homemaking
were
associated
with
a
reduced
likelihood
of
cocaine
use,
while
higher
levels
of
licit
and
illicit
substance
use
and
unemployment
were
associated
with
an
increased
likelihood
of
use.
These
findings
suggest
that
cocaine
use
is
part
of
a
complex
set
of
social
problems
found
in
poorer
young
adults.
Unemployment,
mul-
tiple
drug
use,
drinking,
and
smoking
were
all
associated
with
cocaine
use.
Even
after
adjustment
for
these
factors,
Blacks
and
older
men
were
at
greater
risk
than
Whites
and
younger
men,
suggesting
that
cocaine
use
is
a
more
common
part
of
this
syndrome
in
these
subgroups
than
in
other
subgroups.
Programs
designed
to
reduce
cocaine
use
may
have
limited
success
in
the
presence
of
other,
contin-
ued
social
problems.
O
Acknowledgments
This
research
was
supported
by
contracts
N01-HC-48047,
NO1-HC-48048,
N01-HC-
48049,
and
N01-HC-48050
from
the
National
Heart,
Lung
and
Blood
Institute.
References
1.
O'Malley
PM,
Johnson
LD,
Bachman
JG.
Quantitative
and
qualitative
changes
in
cocaine
use
among
American
high
school
seniors,
college
students
and
young
adults.
In:
Schober
S,
Schade
C,
eds.
The
Epidemi-
ology
of
Cocaine
Use
and
Abuse.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1991:19-43.
Research
monograph
110.
2.
Rouse
BA.
Changes
in
cocaine
use-The
United
States
1972-1990.
Addict
Recovery.
May/June
1991:19-22.
3.
National
Household
Survey
on
Drug
Abuse:
Main
Findings
1988.
Rockville,
Md:
Na-
tional
Institute
on
Drug
Abuse;
1990.
DHHS
publication
ADM
90-1682.
4.
National
Household
Survey
on
Drug
Abuse:
Population
Estimates
1992.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1993.
DHHS
publication
SMA
93-2053.
5.
National
Survey
Results
on
Drug
Use
from
the
Monitoring
the
Future
Study,
1975-1992.
Volume
2:
College
Students
and
Young
Adults.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1993.
DHHS
publication
ADM
93-3598.
6.
Kandel
DB.
The
social
demography
of
drug
use.
Milbank
Q.
1991;69:365-414.
7.
Kopstein
A.
Drug
abuse
related
emergency
room
episodes
in
the
United
States.
Br
J
Addict.
1992;87:1071-1075.
8.
Annual
Emergency
Room
Data:
1991.
Rock-
ville,
Md:
National
Institute
on
Drug
Abuse;
1992.
DHHS
publication
ADM
92-1955.
1740
American
Journal
of
Public
Health
December
1996,
Vol.
86,
No.
12
9.
Annual
Medical
ExaminerData:
1991.
Rock-
ville,
Md:
National
Institute
on
Drug
Abuse;
1992.
DHHS
publication
ADM
92-1955.
10.
Wish
ED.
US
drug
policy
in
the
1990's:
insights
from
new
data
from
arrestees.
IntJ
Addict.
1990-1991;25:377-409.
11.
Wish
ED,
O'Neill
J.
Cocaine
use
in
arrestees:
refining
measures
of
national
trends
by
sampling
the
criminal
popula-
tion.
In:
Schober
S,
Schade
C,
eds.
RTe
Epidemiology
of
Cocaine
Use
and
Abuse.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1991:57-70.
Research
monograph
110.
12.
Drugs
and
Crime
in
America
1988
Annual
Report.
Washington,
DC:
US
Dept
of
Justice,
Office
of
Justice
Programs;
1990.
13.
Goldstein
PJ,
Brownstein
HH,
Ryan
PJ,
Bellucci
PA.
Crack
and
homicide
in
New
York
City,
1988:
a
conceptually
based
event
analysis.
ContempDrugProbl.
1989;16:
651-687.
14.
Goldstein
PJ,
Bellucci
PA,
Spunt
BJ,
Miller
T.
Frequency
of
cocaine
use
and
violence:
a
comparison
between
men
and
women.
In:
Schober
S,
Schade
C,
eds.
7he
Epidemiology
of
Cocaine
Use
and
Abuse.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1991:113-138.
Research
mono-
graph
110.
15.
Kandel
DB,
Murphy
D,
Karus
D.
Cocaine
use
in
young
adulthood:
patterns
of
use
and
psychosocial
correlates.
In:
Kozel
NJ,
Adams
EH,
eds.
Cocaine
Use
in
America:
Epidemiologic
and
Clinical
Perspectives.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1985:76-110.
Research
monograph
61.
16.
Ritter
C,
Anthony
JC.
Factors
influencing
initiation
of
cocaine
use
among
adults:
findings
from
the
Epidemiologic
Catch-
ment
Area
Program.
In:
Schober
S,
Schade
C,
eds.
The
Epidemiology
of
Cocaine
Use
and
Abuse.
Rockville,
Md:
National
Insti-
tute
on
Drug
Abuse;
1991:189-210.
Re-
search
monograph
110.
17.
Windle
M,
Miller-Tutzauer
C.
Anteced-
ents
and
correlates
of
alcohol,
cocaine,
and
alcohol-cocaine
abuse
in
early
adulthood.
J
DrugEduc.
1991;21:133-148.
18.
Kandel
DB,
Simcha-Fagan
0,
Davies
M.
Risk
factors
for
delinquency
and
illicit
drug
use
from
adolescence
to
young
adulthood.
JDrugIssues.
1986;16:67-90.
19.
Gfroerer
JC,
Hughes
AL.
Collecting
data
on
illicit
drug
use
by
phone.
In:
Turner
CF,
Lessler
JT,
Gfroerer
JC,
eds.
Survey
Mea-
surement
of
Drug
Use.
Methodological
Stud-
ies
Volume.
Washington,
DC:
US
Dept
of
Health
and
Human
Services,
Alcohol,
Drug
Abuse,
and
Mental
Health
Adminis-
tration;
1992.
20.
Hughes
GH,
Cutter
G,
Donahue
R.
Re-
cruitment
in
the
Coronary
Artery
Disease
Risk
Development
in
YoungAdults
(CAR-
DIA)
study.
Controlled
Clin
Trials.
1987;8:
68S-73S.
21.
Freidman
GD,
Cutter
GR,
Donahue
RP,
et
al.
CARDIA:
study
design,
recruitment,
and
some
characteristics
of
the
examined
subjects.
J
Clin
Epidemiol.
1988;41:1105-
1116.
22.
Cutter
GR,
Burke
GL,
Dyer
AR,
et
al.
Cardiovascular
risk
factors
in
young
adults:
the
CARDIA
baseline
monograph.
Con-
trolled
Clin
Trials.
1991;12(suppl
1).
Young
Adult
Cocaine
Use
23.
Survey
Measurement
of
Drug
Use:
Method-
ological
Issues.
Rockville,
Md:
National
Institute
on
Drug
Abuse;
1992.
DHHS
publication
ADM
92-1929.
24.
1980
Census
of
the
Population-Alphabeti-
cal
Index
of
Industries
and
Occupations.
Washington,
DC:
US
Dept
of
Commerce,
Bureau
of
the
Census;
1982.
25.
1980
Census
of
the
Population-Classified
Index
of
Industries
and
Occupations.
Wash-
ington,
DC:
US
Dept
of
Commerce,
Bu-
reau
of
the
Census;
1982.
26.
Dyer
AR,
Stamler
J,
Oglesby
P,
et
al.
Alcohol
consumption,
cardiovascular
risk
factors,
and
mortality
in
two
Chicago
epidemiological
studies.
Circulation.
1977;
56:1067-1074.
27.
SAS/STAT
Users
Guide.
Version
6.
Fourth
Edition.
Volume
1
and
2.
Cary,
NC:
SAS
Institute
Inc;
1989.
28.
McCullagh
P,
Nelder
JA.
Generalized
Lin-
ear
Models.
New
York,
NY:
Chapman
&
Hall
Ltd;
1989.
29.
Beta-Test
version.
Cary,
NC:
SAS
Institute
Inc;
1994.
30.
Kleinbaum
DG,
Kupper
LL,
Muller
KE.
Applied
RegressionAnalysis
and
Other
Multi-
variable
Methods.
Boston,
Mass:
PWS-Kent
Publishing
Co;
1988.
31.
Single
E,
Kandel
D,
Johnson
BD.
The
reliability
and
validity
of
drug
use
re-
sponses
in
a
large
scale
epidemiological
survey.
JDrug
Issues.
Fall
1975;426-443.
32.
Harrison
ER,
Haaga
J,
Richards
T.
Self-
reported
drug
use
data:
what
do
they
reveal?Am
JDrugAkcoholAbuse.
1993;19:
423-441.
December
1996,
Vol.
86,
No.
12
American
Journal
of
Public
Health
1741
