1/14/11 Health Behavior and
Cancer
Richard O’Connor, Ph.D.
Associate Professor of Oncology
Department of Health Behavior
Roswell Park Cancer Institute
Objectives
  Explain role of behaviors in etiology and treatment of
cancer
  Explain theories of health behavior and behavior change
  Explain issues of measurement in health behavior
Relevance of Behavior
  Risk factors
  Protective/preventative factors
  Screening
  Information seeking
1 1/14/11 Role of Behaviors in Cancer
  Tobacco use accounts for about 1/3 of cancer deaths
  Physical inactivity and poor diet account for up to 30%
of cancer deaths
  Compliance with screening recommendations, which can
help prevent or mitigate cancer, is a behavioral issue
  Mammography, colonoscopy, PSA/DRE, PAP
Behavioral Risk Factors
  Smoking
  Lung, oral, trachea, bladder,
esophagus, kidney, pancreas, cervix,
colon, leukemia, stomach
  Smokeless tobacco use
  Oral, pancreas
  Physical inactivity
  Colon, breast
  Alcohol use
  Oral, esophagus, liver
  Sexual activity
  Cervix, Oral
  Low fruit and vegetable
consumption
  Breast, colorectal, oral, larynx,
esophagus, stomach
  Obesity
  Breast, endometrium, kidney,
esophagus, colon
  Tanning/Excessive sun exposure
  Melanoma
Worldwide deaths from sitespecific cancers attributable to
selected risk factors by sex.
For every cancer site, solid
blocks of color represent deaths
not attributable to risks
assessed and broken blocks of
color represent deaths
attributable to selected risk
factors
Danaei et al. Lancet 2005.
2 1/14/11 Conceptual model depicting the relations between discrete aspects of cancer fear/anxiety,
their function, and impact on screening behavior.
Consedine N S et al. Cancer Epidemiol Biomarkers Prev
2004;13:501-510
©2004 by American Association for Cancer Research
Purpose of a Theoretical
Framework
  Focus attention on certain factors, allowing
you to ignore others;
  Models force the investigator to make
causal assumptions explicit.
  To make predictions (allows hypothesis
testing)
  For practitioners models allow one to
understand why interventions work or fail to
work and help guide improvements in
programs.
Health Behavior Models
  Stimulus response theory (rewards &
punishments)
  Cognitive theories (tell me what I need to know)
  Health belief model (HBM)
  Theory of Reasoned Action
  Social Learning Theory (social influences and
expectations)
  Diffusion of Innovations (macro social influences)
3 1/14/11 Theory of Reasoned Action
  Behavior is best predicted from a person’s intention to
perform the behavior.
  Intention to perform the behavior is the result of two
factors:
  Attitude about the behavior
  Social norms related to the behavior
Theory of Planned Behavior
  Extends TRA to include perceived behavioral control
  Belief that one has, and can exercise, control over
performing the behavior
  People may try harder to perform a behavior if they feel
they have a lot of control over it
  Behavioral skills
4 1/14/11 Attitude toward the behavior
  Attitude toward the behavior is a function of one’s
beliefs about the following:
  Belief that doing the behavior will lead to a particular
outcome;
  The individual’s evaluation of the outcome (rating of good
or bad)
Social Norms
  behavioral expectations and cues within a
society or group
  customary rules of behavior that coordinate
our interactions with others
  Deference to the social norms maintains one's
acceptance and popularity within a particular
group
  ignoring the social norms risks one becoming
unacceptable, unpopular or even an outcast from a
group
Social Norms
  Norms are a special category of beliefs
  perceived to be socially shared regarding prevalent or
prescribed behaviors
  behavioral (descriptive) norms refer to the most
common actions or behaviors actually exhibited
in a social group.
  what most individuals of a social group actually do.
  attitudinal (injunctive) norms refer to the most
widely shared beliefs or expectations in a social
group about how people in general or members
of the group ought to behave in various
circumstances.
5 1/14/11 Social Norms
  A way to shape individual behavior is to alter the social
norms around that particular behavior
  if you alter the injunctive norm, the descriptive norm may
follow
Social Norms
  a key part of social-cognitive theories of health
behavior
  but we rarely stop to think about what the norms are
and how they came to be
Important people smoke…like doctors!
6 1/14/11 Cancer for the Holidays…
SMOKING ON FILM:
Case study in social norming
• In 1937-8 alone, tobacco companies agreed to pay stars at least $3.2 million
(2008 dollars) for their advertising services.
• The tobacco companies spent more to advertise Hollywood than Hollywood
spent to advertise itself.
• This pattern of intense cross-promotion changed only when tobacco
companies began shifting their advertising dollars to the new medium of
television and the "Studio System" of talent contracts ended in the early
1950s.
7 1/14/11 Precaution Adoption Process
Model
  Stage theory describing movement from lack of
awareness to maintenance of a behavior
  People who are unaware or unengaged in an issue
are different from those who have simply decided
not to act
Precaution Adoption Process
Model
  People move through stages in linear fashion, but
can recycle through stages after ‘deciding.’
  Interventions may differ for people in different
stages
  Raising awareness
  Increasing engagement
  Motivating to act
8 1/14/11 Transtheoretical Model
  Behavior change as process, not event
  5(6) distinct stages (Stages of Change)
  Precontemplation, contemplation, preparation,
action, maintenance, (termination)
  Circular rather than linear (people can
move between stages readily in any
sequence)
Processes of Change
  Transitions between the stages of change are effected
by processes of change.
 
consciousness raising, counterconditioning,
dramatic relief, environmental reevaluation,
helping relationships, reinforcement
management, self-liberation, selfreevaluation, social liberation, and stimulus
control. Transtheoretical Model
  Extended TTM includes decisional balance and self
efficacy
  TTM originally developed out of smoking cessation, but
is now widely applied to hard-to-change behaviors,
particularly relapsing behaviors.
9 1/14/11 Behavioral Diagnosis:
ASK: What is the behavioral problem in whom,
when, where? (Identify and describe the health
behavior)
ASK: Why? (What are the correlates of this
behavioral problem? – i.e. knowledge, beliefs,
attitudes, values, social norms, behavioral skills,
situational determinants)
ASK: What can be done to change the problem? (Is
there an intervention or a way to correct or change
the behavioral problem? – i.e. how can you bring
about behavioral change?)
Measurement of Behavior
Measuring Behavior:
  How do you measure behavior?
  Levels of measurement:
(nominal, ordinal, interval, ratio)
Measurement Error
The combined error that results from
inevitable imperfections and variability in
the process of measurement
  Random error
  Systematic Error (BIAS)
Reliability and Validity in measurement
10 1/14/11 Reliability
Concerns the extent to which measurement is
repeatable and consistent (free from random
errors)
If the random error in your measurements is so
large that there is almost no stability in your
measures, you can't explain anything!
Reliability
Reliable measures: measure a variable precisely
and consistently.
Factors to consider when determining reliability
of a measure:
Precision
Sensitivity
Resolution
Consistency
Reliability also serves as a limiting factor on
the correlation one can see between two
measures
Types of Reliability
  Test-retest reliability
  Inter-rater reliability
  Intra-observer reliability
  Parallel forms reliability
  Internal consistency
Measuring Reliability
Reliability is often expressed as a correlation
coefficient (i.e. Pearson or Spearman rank)
There are also other measures of reliability
(i.e. index of concordance, Kappa, Cronbach’s alpha,
etc.)
11 1/14/11 Validity
The extent to which a measurement actually
measures what it is intended to measure
Factors to consider when determining validity
of a measure:
  Accuracy
  Specificity
Types of Validity:
  Face validity
  Criterion validity
(concurrent, predictive, discriminant)
  Construct validity
Internal validity
External validity (generalizability)
Biochemical validation
Threats to Internal Validity:
Threats to External Validity:
  Selection
  Learning
  Attrition
 Experimenter effects
  History
 Reactance
  Regression to the mean
 Diffusion
  Instrumentation
 Poor operationalization
 Maturation
12 1/14/11 Sources of Invalidity of Measures
  Reactive measurement effects
  Awareness of being tested
  Role selection
  Measurement as change agent
  Response sets
  Error from the Investigator
  Interviewer effects
  Change in research instrument
  WORDING MATTERS!!!
VALIDITY IS ALSO AFFECTED BY BIAS !
Bias is the result of systematic error in the design or
conduct of a study
  Selection Bias
  Information Bias
Bias has a preferred direction and won’t
average out over participants
  Day effects
  Order effects
  Practice effects
We spend a lot of time designing
studies to avoid bias
  There’s a difference between bias and
imprecision
BIAS
IMPRECISION
13 1/14/11 Feasibility
Concerns whether the proposed measurement
procedure is possible, practical, and worthwhile
 
Consider:
  Time
  Cost
  Cost-benefit analysis: ASK…
Does the information obtained justify the
time and cost required?
Developing Measures
Development of measures or identification of
suitable existing measures is crucial to any
evaluation or study
  Direct vs. Indirect
  Multiple measures
  Proximate measures
Careful attention needs to be paid to how the
measure is taken
  Wording makes a difference
  Questions should be age and culturally
appropriate
  Context matters
14 1/14/11 Types of Measures:
  Measuring effects on persons served
  Measuring effects on agencies
  Measuring effects on larger systems
  Measuring effects on the public
  Measuring microbehaviors
Proximate measures
  Measuring one behavior in order to assess the
behavior of true interest
  Sometimes, the best you can do is a proxy
measure
  Also can use proxy measures when the behavior
of interest is distal
  Measuring urine carcinogen levels as index of disease risk
from smoking in currently healthy smokers
Unobtrusive measures
  Measuring without the subject’s knowledge
  Don’t necessarily have to even have the
subject present
  Less risk of reactance; may be ‘truer’
measure of behavior
15 1/14/11 Examples of unobtrusive measures
  Differential erosion of floor tiles in museum as index
of popularity of exhibits
  Clustering by race/ethnicity in a college dining hall as
index of racial attitudes
  Number of condoms taken from a tray as index of
community interest in safer sex
  Counting number of disposed alcoholic beverage
bottles outside a house as index of alcohol intake
  Using spent cigarette butts to measure smoking
behaviors
  Hidden cameras and other methods of observation
Measurement of correlates of Behavior
 
Attitudes (Likert scales, Semantic differential,
rating scales, Q-sorts)
 
Beliefs (perceived risks vs. absolute risks)
 
Knowledge
 
Values
 
Behavioral skills
 
Social influences
 
Situational influences
Scales: Multiple measures
  Because most behavioral measures are fallible,
we need more than one to measure a construct
The more items you
endorse, the stronger the
attitude
GUTTMAN
SCALE
ITEM 1
DOMAIN
SAMPLING
SCALE
ITEM 2
ITEM 3
ITEM 4
EVENT 1
RISK
SCALE
EVENT 2
EVENT 3
Risk
Multiple items measure
the same underlying
construct in different
ways; items correlated
The more events you
identify, the greater
your risk; items not
necessarily correlated
16 1/14/11 Methods of Data Collection
Focus Groups
In-person
Telephone-based
Computer web-based
Surveys
Interviews
Questionnaires
Observation
In-person
Hidden
Selecting the Survey Method
  Population issues
  Sampling issues
  Question issues
  Content issues
  Bias issues
  Administrative issues
Advantages & disadvantages of survey
methods:
  Interviews
VS. Questionnaires
What are the COSTS vs. BENEFITS of the available
methods . . .???
Surveys: Interviews, Questionnaires
 Examples:
  BRFSS: http://www.cdc.gov/brfss/
  YRBSS: http://www.cdc.gov/nccdphp/dash/yrbs/
  NHIS: http://www.cdc.gov/nchs/nhis.htm
  NHANES: http://www.cdc.gov/nchs/nhanes.htm
  NSDUH: http://www.samhsa.gov/oas/nhsda.htm
  MTF: http://www.monitoringthefuture.org/
17 1/14/11 Direct Observation
  Types of observation:
  Clinical Exam
  Physical measurement
  Observation in a defined setting
  Observation in an undefined setting
  Intra-observer vs. inter-observer reliability
  Factors affecting observer reliability
Other Data Sources
Some examples of other data sources:
  Medical records
  Government statistics
  Self-report diary or journal records
  Expert panel meetings
  Consensus meetings
  Scientific meetings/conferences
Shameless Plug
  Space for 1-2 students in my laboratory for research
experience
  Tobacco product characterization
  Consumer perception research
  Exposure studies with novel products
  Could lead to final MS project
  Contact me: [email protected]
18