Transcript 5. Prevention & Control

The Five Main Rubrics of
Epidemiology, As Applied to
Drug Dependence Syndromes
James C. (Jim) Anthony, Ph.D.
Professor and Chairman
Department of Epidemiology
College of Human Medicine
Michigan State University
[email protected]
www.epi.msu.edu
Copyright, J. Anthony, 2007
Learning Objectives
1. By the end of the session, the learner will be able to sort the
main evidence from epidemiological studies into relation to the
five main rubrics of (1) Quantity, (2) Location, (3) Causes, (4)
Mechanisms, and (5) Prevention and Control.
2. The learner also will be able to describe the main clinical
features associated with syndromes of drug dependence, and to
discuss separable relationships between drug dependence and
drug-related socially maladaptive behavior that sometimes
appears in different phases of the drug dependence process.
3. The learner will be able to discuss the 'no causation without
manipulation' theme of contemporary epidemiology, in relation to
the example of early social maladaptation and later risk of drug
use and dependence.
Is Epidemiology Just A Matte r
of Counting Up Cases and
Sorting Them Into Bins With
Labels for Static “Risk
Factors”?
No!
Epidemiology is a branch of the
biomedical sciences that uses
population-oriented research
approaches to answer specific
questions.
Does it make sense to define
epidemiology in terms of jargon
words such as ‘distribution and
dynamics of disease’?
No!
What’s the use of a definition that is
understood only by individuals who have
studied epidemiology?
What of Jeremy Morris’ Seven
Uses of Epidemiology?
Are They Useless?
No!
For anyone who can remember the seven uses, they serve as equally
useful orienting concepts to the ‘content’ of epidemiology (as opposed
to the ‘methods’ of epidemiology).
But I rarely encounter anyone who can recite all 7 uses.
Teaching epidemiology from 1978 through 1999, I learned that almost
everyone can remember and retain 5 rubrics of epidemiology, which
subsume Morris’ 7 uses.
As shown on the next page, a helpful mnemonic starts with the first
question, “How many?” and then proceeds to these questions:
“Where?” “Why?” “How” and
“What Can We Do?”
Five Main Rubrics of
Epidemiological Research
and The Associated Research Questions
1. Quantity: How many?
2. Location: Where?
3. Causes: Why?
4. Mechanisms: How?
5. Prevention & Control: What Can We Do?
These rubrics make it possible to specify an alternative definition for
epidemiologists to use when describing their research to the public.
Namely, in epidemiology, we seek increasingly definitive answers to
scientific and public health questions under five main rubrics.
1. Quantity: Within the population, how many are affected?
How many are becoming affected?
2. Location: Where, in the population experience, are cases
more or less likely to occur, and where are they more or less
likely to be found?
3. Causes: What accounts for some becoming affected while
others are spared?
4. Mechanisms: What are the linkages of states and processes
that account for people becoming cases and remaining
cases? What accounts for the observed natural history and
clinical course, comorbidities, secondary impairments,
residual disabilities?
5. Prevention & Control: How can we prevent, delay, or
ameliorate the condition?
Epidemiology is not the only branch of biomedical sciences that
seeks to answer these questions; it is distinguished by its
population orientation, consistent with an exchangeable term to
define the field of inquiry: ‘medical ecology.’
What Is One of the Unique
Skill Sets Practiced By
Epidemiologists Under the
First Rubric (Quantity)?
MEASURING
DISEASE
OCCURRENCE AND
FREQUENCY IN
POPULATIONS
Epidemiological
Population Estimates
for the
United States,
1992-1998
Tobacco, 1 in 3
Heroin, 1 in 4-5
Inhalant
drugs, 1 in 20
Psychedelic
drugs, 1 in 20
Analgesic
Drugs, 1 in 11
Anxiolytic,
sedative, &
hypnotic
drugs, 1 in 11
Estimated
fraction
of drug users
who have
developed a
syndrome of
dependence
Crack + HCl, 1 in 5
Cocaine HCl, 1 in 6
Alcohol, 1 in 7-8
Stimulants other
than cocaine, 1 in 9
Cannabis,
1 in 9-11
(Adapted from Anthony et al., 1994; Chen & Anthony, 2002)
General Concept of Drug Dependence
(generally adapted to each drug)
We can define and measure a drug dependence syndrome in
relation to:
(i) disturbances of the mental life (e.g.,obsession-like
ruminations and recurrent thoughts or cravings);
(ii) disturbances of behavior (e.g., sometimes expressed in the
form of compulsion-like repetitions of drug-involved
behavior);
(ii) manifestations of neuroadaptation secondary to drug
exposures (e.g., experienced as a subjective feeling that the
same dose of drug is less efficacious, or observable in the form
of clinical features of a withdrawal syndrome after longsustained daily or near-daily consumption).
For a Scientific Reason, This General Concept of Drug
Dependence Does Not Include a Criterion for Legal or
Social Maladaptation, Which Needs to Be Defined,
Conceptualized, and Measured Quite Separately.
• Drug-induced social maladaptation (e.g., getting into trouble with the law) is
not necessary under this concept of drug dependence. We think of the social
maladaptation as a distinguishable phenomenon that has as much to do with
the legal and social responses to a person’s drug use.
• Many examples of drug dependence without social maladaptation exist.
These cases are not differentiable from other cases unless we allow drug
dependence to be defined, conceptualized, and measured without reference to
drug-induced or drug-related socially maladaptive behavior.
• Moreover, due to time and place variation in drug laws, and to variations in
enforcement, prosecution, and judicial response to drug law violations, a case
definition that appeals to concepts of legal and social maladaptation will
induce epidemiological variations from place to place and time to time.
(Anthony et al.., Clinical Neuroscience Research, 2005).
The Concepts of Drug Dependence Were
Derived By Studying Cases In Contact
With Specialty Treatment Services.
But What Do We Know About The
Fraction of Community Cases Who Seek or
Enter Treatment?
Another Epidemiological Parameter Estimated Under
the Rubric of 'Quantity':
What Proportion Make First Treatment Contact
Within 5-10 Years After Onset of Dependence?
Within 5 Years:
Alcohol
Controlled Substances
12-18%
40-45%
Within 10 Years:
25-30%
Alcohol
Controlled Substances 55-60%
Estimates from National Comorbidity Survey Replication, USA (Wang et al., 2005)
What Is One of the Unique
Skill Sets Practiced By
Epidemiologists Under the
Second Rubric (Location)?
Identifying when,
where, and within
which population
subgroups the cases
are more or less likely
to arise.
TIME TREND IN THE ESTIMATED ANNUAL INCIDENCE RATE
FOR NEW CANNABIS USE IN THE U.S., BY AGE
Data from NHSDA, Late 1990s, reported by Gfroerer et al., 2002
Adapted by Anthony, 2006
Locational parameters of interest:
When: The Time Dimension, 1965-1999
Who: Facets of Person, Ages 12-17, 18-25
Where: Facets of Place, USA
UPPER BOUND
LOWER
BOUND
Based upon these estimates, during the late 1990s,
the US was accruing as many new cases of cannabis
use as were being accrued during the last interval
of peak incidence during the Vietnam and
immediate post-Vietnam era.
From: Anthony, 2006
Under the Location Rubric: Are Crack-Cocaine Smokers More
Likely To Experience Rapid-Onset of Cocaine Dependence?
Locational Parameter of Interest:
Facets of Person, What Form of Cocaine Use?
Cocaine Hydrochloride Powder only, typically intranasally, versus Recent Cocaine HCl powder use plus
crack-smoking, versus Past Cocaine HCl powder use plus crack smoking.
35
Estimated
risk of
experiencing
each clinical
features
associated 20
with rapidonset of
cocaine 10
dependence
5
Should we think
of this
as a 'causal
association'?
Here, we shift to
the third rubric
“Of Causes”
Cocaine HCl
(n=572)
Crack-smoking
(Recent HCL)
(n=190)
Crack-smoking
(Past HCL)
(n=93)
0
Inability
Tolerance Salience
Health
To Cut
Problems
Reduced Emotional
Used Down
Activities Problems
More
Than
Intended
Chen & Anthony, Psychopharmacology, 2004
What Are Unique Skill Sets
Practiced By Epidemiologists
Under the Third Rubric
(Causes)?
Population-Level
Research on the
Suspected Causes of
Health Conditions
Question Under the Rubric of Causes: Is there increased risk of rapidonset cocaine dependence in response to crack-cocaine smoking?
We do not yet have definitive evidence to answer this question. The
government has dug in its heels and will not release the data required to push
forward on this research front.
Why don’t we have the definitive evidence?
Crack-cocaine smoking is not distributed at random in the population; some
subgroups are more likely than others to smoke crack-cocaine – e.g., Males age
20-29, of African-American heritage.
This subgroup also is more likely to progress rapidly to cocaine dependence, once
cocaine use starts – even when crack-smoking is not in the picture (O’Brien &
Anthony, Neuropsychopharmacology, 2005).
Is this an example of underlying gene-based or cultural vulnerability
inherent to African-American males?
Probably not. When we hold constant neighborhood conditions, White nonHispanic males and African-American males are equally likely to have started
smoking crack-cocaine? (Lillie-Blanton, Anthony, Schuster, JAMA, 1993)
So we are left with an unanswered epidemiological research question.
NSDUH data have been gathered so that we might to press forward toward
increasingly definitive evidence, but these data are not being released for analyses,
due to concerns about potential violation of federal law that protects privacy of
research participants.
What Are Unique Skill Sets
Practiced By Epidemiologists
Under the Fourth Rubric
(Mechanisms)?
Population-Level
Research on Suspected
Mechanisms, Natural
History, and Clinical
Course
A Causal Model for Estimating Suspected Long-Term Effects
of Prenatal Cocaine Exposure on Neurodevelopmental Functioning
at the Time of Entry Into Schooling
Parameters of this
model are estimated
after epidemiological
sampling and longterm followup of an
epidemiologically
derived sample of
newborns
(See Bandstra et al., 1999-2006)
An Elaboration of the Causal Model to Pose a Specific Question:
Does the evidence tend to support a specific causal effect of
prenatal cocaine exposure upon sensorimotor functioning, over and
above more general impact upon levels of neurodevelopmental
functioning?
To date, our evidence
tends to support this
specific causal path,
but the jury still is
out.
(See Bandstra et al., 1999-2006)
Rubric: Mechanisms (Natural History):Population-averaged drug-specific estimates
for risk of becoming drug dependent for the first time, plotted since first use.
Wagner & Anthony,
Neuropsychopharmacology,
2002
What Are Unique Skill Sets
Practiced By Epidemiologists
Under the Fifth Rubric
(Prevention & Control)?
Population-Level
Prevention and Public
Health Experiments
Current Theme We Are
Debating in Epidemiology
No causation without manipulation?
The debate is pushing epidemiology in the
direction of intersections with experimental
sciences and toward greater mastery of
randomized controlled trial designs and
randomized instrumental variable designs.
Exemplar Problem 1
As already discussed, Crack Smokers seem to be more likely to develop a
rapid-onset syndrome of cocaine dependence.
Is this because crack 'causes' increased risk of the syndrome?
Is it because crack-smokers were different to begin with, before cracksmoking (e.g., increased levels of a susceptibility trait)?
Is it because crack 'recruits' highly susceptible individuals?
We can’t experiment by assigning crack at random, but we CAN randomly
assign crack-prevention programs!
To the extent that these programs actually reduce crack cocaine smoking,
there ought to be concomitantly reduced risk of rapid-onset cocaine
dependence in that experimental subgroup – if crack smoking is causing
rapid-onset cocaine dependence.
Exemplar Problem 2
We have seen evidence that prenatal cocaine exposure might be causing
long-term impairments in neurodevelopmental functioning, and perhaps a
specific impairment in sensorimotor functioning.
But were all potentially confounding variables brought under control?
Isn't it possible that some background characteristic, condition, or
process is accounting for what now is observed as a spurious
relationship between prenatal cocaine exposure and later sensorimotor
deficits?
Here again, we cannot experimentally induce prenatal cocaine exposure at
random, but we can randomly assign programs that seek to reduce
prenatal cocaine exposure, with followup to check sensorimotor
functioning deficits in the newborns whose mothers did not have the
benefit of that program.
Exemplar Problem 3
Youths, especially boys, with a history of conduct
problems and other socially maladaptive behavior, are
more likely to (a) engage in early-onset drug use, and
(b) experience drug dependence syndromes, once drug
use starts.
Years of observational studies confirmed the
robustness of this aspect of 'location of cases' but
never could resolve the causal question:
Was the early social maladapation a 'cause' of the later
drug involvement?
Adopting the 'no causation without manipulation'
perspective, we undertook a population-level
randomized controlled trial to probe this causal
question.
RCT: Good Behavior Game universal intervention in Grades 1-2:
Estimated ~50% reduction in risk of starting to smoke tobacco
by age 14 among males.
60
Usual & Customary Control
50
49%
40
Mastery
Learning
30
20
Good Behavior Game
10
0
22%
6
8
AGE
12
Kellam & Anthony, Am J Public Health, 1998
14
REPLICATIONS & EXTENSION OF THIS WORK
Storr and colleagues (Drug & Alcohol Dependence, 2002) have
reported results from an independent replication, also
consistent with an enhanced Good Behavior Game impact on
risk of early onset tobacco smoking.
Via follow-up of these young people to the early years of young
adulthood, Storr and colleagues (under review) now are
reporting that the long-term impact of the Good Behavior
Game appears to include reduced risk of tobacco dependence
syndromes. That is, delay of onset of tobacco smoking was
followed by reduced risk of becoming tobacco dependent.
NEXT STEPS
Working with collaborators in the Colleges of Education and
with community, school, and parent leaders, members of our
research group hope to learn whether this type of primary
school intervention might yield beneficial impact in other local
communities.
Via secondary effects on others, these preventive
interventions directed toward youthful tobacco smoking should
have lasting public health benefits in addition to benefit in
relation to the health of individual participants.
Recap: Learning Objectives
1. By the end of the session, the learner will be able to sort the
main evidence from epidemiological studies into relation to the
five main rubrics of (1) Quantity, (2) Location, (3) Causes, (4)
Mechanisms, and (5) Prevention and Control.
2. The learner also will be able to describe the main clinical
features associated with syndromes of drug dependence, and to
discuss separable relationships between drug dependence and
drug-related socially maladaptive behavior that sometimes
appears in different phases of the drug dependence process.
3. The learner will be able to discuss the 'no causation without
manipulation' theme of contemporary epidemiology, in relation to
the example of early social maladaptation and later risk of drug
use and dependence.
The End!
Thank you for your kind attention.
Photo credit: Pan American Health Organization
http://www.paho.org/English/DPI/categ02.htm