Epidemiology in Action

1
Epidemiology in Action

• Unit II. Types of Epidemiology Studies
• Experimental vs. Observational
• Descriptive and Analytic Studies

• Activities
• Study the STATS
• 2. Name That Study

2
Epidemiology in Action
2
Experimental Researchers have control over
key variables in the study. Observational
Researchers do not have control over key
variables in the study.
3
EXPERIMENTAL Studies in which researchers
control key variables. These include
intervention and clinical trials that attempt to
prove cause and effect.
Epidemiology in Action
3
Vaccine trials, for example. One population
receives the vaccine (an independent variable
that the researchers control), the other doesnt
(or is given a placebo to blind subjects) and
serves as the control group. The groups are
compared in terms of how well/if the vaccine
works.
Source CDC/1963 Health workers administering
the polio vaccine to children.
4
Epidemiology in Action
4
EXPERIMENTAL This is a hypothetical example of
a study where one group of juveniles recovering
from substance abuse is treated through in-house
drug treatment while the other is not. Study
tracks rate of relapse over time.
Total who relapsed
Months After Treatment
5
Epidemiology in Action
5

• EXPERIMENTAL
• Advantages
• Gold standard for proving causation, because
  you control the variables youre studying
• Risk factors can be isolated
• Disadvantages
• Can be difficult to isolate variables
• Ethical issues

6
Epidemiology in Action
6
OBSERVATIONAL STUDIES DESCRIPTIVE Studies which
describe the FREQUENCY and PATTERNS of
occurrence of a condition or factor by time,
place, and person for example, these studies on
exercise by U.S. state. Link Behavioral Risk
Factor Surveillance System CDC SOURCE Centers
for Disease Control and Prevention (CDC).
Behavioral Risk Factor Surveillance System Survey
Data. Atlanta CDC/U.S. Department of Health and
Human Services, 2000.
7
Epidemiology in Action
7

• DESCRIPTIVE
• Advantages
• Can be conducted using widely available data
• Can be conducted on large populations
• Disadvantages
• Difficult to infer causation
• Can be expensive and time-consuming
• Susceptible to bias (surveys, for example)

8
DESCRIPTIVE Studies are used to characterize
a health problem and its risk factors. These
studies also COMPARE groups of people in terms of
changes in health conditions over time, place,
and exposure to risk factors.
Epidemiology in Action
8
For example, many organizations keep ongoing
statistics on teen pregnancy. They may DESCRIBE
the health status (pregnancy) of teen females,
and/or COMPARE the RATE of pregnancy for
POPULATIONS ( a 1975 group to a 1985 group).
They may compare rates to RISK FACTORS
(race/ethnicity, socio-economic status, etc.)
Link Teen Pregnancy.org
9
Epidemiology in Action
9
A study like the one at right DESCRIBES the
health status of a population.
Link Teen Pregnancy.org
10
Number of Teen Pregnancies, 1996 (AGI)
10
Nearly 1 million teen pregnancies occurred in
1996. To put it another way, more than 100 U.S.
teens become pregnant each hour. Forty percent
of these pregnancies were to girls under age 18,
and 60 percent were to girls aged 18-19.
Total 905,000
542,640
337,530
24,830
The Alan Guttmacher Institute. (1999). Special
report U.S. teenage pregnancy statistics with
comparative statistics for women aged 20-24. New
York Author.
Link The Alan Guttmacher Institute Home Page
This study DESCRIBES and COMPARES rates by age.
11
Teen Pregnancy Rates, Girls Aged 15-17
(AGI)(number of pregnancies per 1,000 girls)
11
After increasing 19 percent between 1972 and 1989
(including 7 percent between 1986 and 1989), the
teen pregnancy rate for girls aged 15-17
decreased 17 percent between 1989 and 1996 to its
lowest rate ever recorded.
Link Teen Pregnancy.org
The Alan Guttmacher Institute. (1999). Special
report U.S. teenage pregnancy statistics with
comparative statistics for women aged 20-24. New
York Author.
Link The Alan Guttmacher Institute Home Page
This study COMPARES two populations, girls/1972
to girls/1996.
12
Number of teen births, 1999
12
Among teens aged 15-19, more births occur to
non-Hispanic White teens than to any other
racial/ethnic group.
Data for 1999 are preliminary. Curtin, S.C.,
Martin, J.A. (2000). Births Preliminary data for
1999. National Vital Statistics Reports 48(14).
Link Teen Pregnancy.org
This study COMPARES rates to the RISK FACTOR of
Race/Ethnicity.
13
13
Teen Pregnancy Rates, Racial/Ethnic Subgroups
(AGI)(number of pregnancies per 1,000 girls aged
15-19)
Teen pregnancy rates vary substantially among the
three largest racial/ethnic subgroups. Between
1990 and 1996, the rate for African-American
teens declined 20 percent and the rate for
non-Hispanic White teens declined 24 percent.
The teen pregnancy rate for Hispanics increased
between 1990 and 1994, but then declined 6
percent between 1994 and 1996.
Non-Hispanic Black
Hispanic (any race)
Non-Hispanic White
Darroch, J.E., Singh, S. (1999). Why is teenage
pregnancy declining? The roles of abstinence,
sexual activity and contraceptive use. Occasional
Report 1. New York The Alan Guttmacher Institute.
Link The Alan Guttmacher Institute Home Page
This study COMPARES rates, in terms of
race/ethnicity, and over time.
14
14

• You may have noticed that data on teen pregnancy
  can be presented in many different ways. When
  INTERPRETING results from studies, it is
  important to pay attention to these details. For
  example
• WHAT is the study measuring, in terms of
• AGE
• TIME
• GEOGRAPHY
• RISK FACTORS?
• WHERE did the data come from?
• Who collected it, and how was it collected?
• The data on the next three screens is similar,
  but comes from different sources and tracks
  different years and ages.

Link Teen Pregnancy.org
Link The Alan Guttmacher Institute Home Page
15
Teen Pregnancy Rates, Girls Under 15
(AGI)(number of pregnancies per 1,000 girls)
15
After increasing 30 percent between 1973 and
1988, the teen pregnancy rate for girls aged 14
or younger decreased 24 percent between 1988 and
1996 to the lowest rate ever recorded.
Note denominator used is the population of girls
aged 14. The Alan Guttmacher Institute. (1999).
Special report U.S. teenage pregnancy statistics
with comparative statistics for women aged 20-24.
New York Author.
Link The Alan Guttmacher Institute Home Page
16
Teen Pregnancy Rates, Girls Under 15
(NCHS)(number of pregnancies per 1,000 girls)
16
After increasing 16 percent between 1982 and 1985
and remaining constant between 1985 and 1986, the
teen pregnancy rate for girls aged 14 or younger
decreased 22 percent between 1986 and 1996 to the
lowest rate ever recorded.
Note denominator used is the population of girls
aged 10-14. Ventura, S.J., Mosher, W.D., Curtin,
S.C., Abma, J.C., Henshaw, S. (2000). Trends in
pregnancies and pregnancy rates by outcome
Estimates for the United States, 1976-96. Vital
and Health Statistics 21(56).
Link The Alan Guttmacher Institute Home Page
17
Teen Pregnancy Rates, Girls Under 15
(NCCDPHP)(number of pregnancies per 1,000 girls)
17
The teen pregnancy rate for girls aged 14 or
younger decreased 11 percent between 1995 and
1997.
Note denominator used is the population of girls
aged 13-14. Centers for Disease Control and
Prevention. (2000). National and state-specific
pregnancy rates among adolescents United
States, 1995-1997. MMWR, 49(27), 605-11.
Link The Alan Guttmacher Institute Home Page
18
Data Sources
Epidemiology in Action
18
Teen pregnancy data are released by three
national groups

• The Alan Guttmacher Institute
• The National Center for Health Statistics (NCHS),
  Centers for Disease Control and Prevention (CDC),
  U.S. Department of Health and Human Services
  (DHHS)
• The National Center for Chronic Disease
  Prevention and Health
• Promotion (NCCDPHP)
• Each group collects different information, from
  different sources,
• in different ways. These three groups DO tend to
  operate very similarly,
• thus the similarity of their results. Other
  studies, however, will have
• marked differences in terms of collection and
  presentation of data, and
• it can affect things significantly.

Link The Alan Guttmacher Institute Home
Page Link National Center for Health
Statistics Link National Center for Chronic
Disease Prevention
19
19

• How data is gathered,
• by whom, and
• how it is analyzed
• are all factors that can
• introduce BIAS
• into research.

Library of Congress 1908 Sioux Indian smoking
Do we look at research by cigarette manufacturers
on the health dangers of smoking, for example,
the same as research done by an independent,
unbiased organization?
Were BIASED when we play favorites in terms of
choosing study subjects or in assessing exposure.
Bias is dangerous, because it can invalidate a
study.
20
International Pregnancy Rates, Teens 15-19 (AGI)
20
For example, how could a researcher make the US
statistics look even worse?
The United States has much higher pregnancy and
birth rates than other fully industrialized
countries. U.S. pregnancy rates are nearly twice
as high as rates in Canada and England and seven
to eight times as high as rates in Japan and the
Netherlands.
Singh, S., Darroch, J.E. (2000). Adolescent
pregnancy and childbearing Levels and trends in
developed countries. Family Planning Perspectives
32(1), 14-23. Pregnancy rates calculated as the
sum of births, abortions, and estimated
miscarriages (20 percent of births plus 10
percent of miscarriages).
21
21
By only presenting data on 4 countries, thats
how. (Granted, its bad, but is it that bad?)
22
Epidemiology in Action
22
This is an example of BIAS, and it happens all
the time when politicians, advertisers, and other
organizations organize their statistics in ways
that best persuade, no matter what the real
truth is. When conducting solid research, we
want to reach whats called the true value
the pure statistics, untainted by bias. For an
interesting look at how mangled research can get
when bias is allowed to reign, check out STATS, a
non-partisan, non-profit research organization in
Washington, D.C. devoted to the accurate use of
scientific and social research in public policy
debate. In their news clips section, for
example, youll see articles such as Good News!
More People are Reporting Crimes!, that tell the
real story behind the faulty use of
statistics. LINK Statistical
Assessment Service (STATS)

Source Statistical Assessment Service 2100 L.
St. NW Suite 300 Washington, D.C. 20037
23
Epidemiology in Action
23

• Problems to watch out for in epidemiological
  studies
• Selection Bias when subjects chosen are not
  representative of the target population about
  which conclusions are to be drawn.
• Information Bias errors in measuring
  exposure/risk or disease/condition
• Confounding Variable A variable that is
  related to the condition/disease under study and
  is associated with, but NOT a consequence of, the
  exposure/risk factor under investigation.

24
STUDENT ASSIGNMENT STUDY THE STATS
24

DIRECTIONS In groups of two or three, students
are to find two sets of data done on a similar
population, from two different sources. Use any
of the sources weve been introduced to so far as
starting points, or find your own sources.
Suggested study health topics include Cancer,
smoking, drugs/alcohol, teen violence,
health/nutrition issues, etc. Then.. find out
how both studies got their information how was
it collected? When was it collected? Who
collected it? What instruments (such as a
survey) did they use to collect information? How
did they analyze and present their data?
Compare/contrast the two studies, in terms of the
Who/What/When/Where/How. Determine how
the studies DIFFERED in terms of what they
measured (age/time/risk factors/etc.) Decide
how the researchers may have introduced BIAS into
their research, and discuss possible CONFOUNDING
variables that may be affecting results. Give at
least three examples of places where BIAS could
be lurking in this study, and two examples of
potential CONFOUNDING variables that may call the
results into question. Present your findings
to the class in a written report, a PowerPoint
presentation, or both.
25
25
STUDENT ASSIGNMENT STUDY THE STATS Suggested
GENERAL Resources

Link National Center for Health Statistics Link
National Center for Chronic Disease
Prevention Link NIH Health Information Link
National Center for Educational Statistics Link
Census Bureau Home Page
OTHER Resources

• Try searching by topic for research statistics.
  For example
• Teen smoking research statistics
• Cancer research statistics
• Juvenile crime research statistics
• Teen driving research statistics

26
Epidemiology in Action
26

• Observational Studies Analytic
• Three types of ANALYTIC studies
• Cohort
• Case-Control
• Cross-sectional

27
COHORT Studies compare groups
of people who have been exposed to suspected risk
factors to groups of people who have not been
exposed to determine who develops a disease or
other outcome. For example, an ongoing study of
nuns has found a link between low language skills
and development of Alzheimers disease in later
life. In this study
Epidemiology in Action
27
Exposure Disease, y/n ?
Exposedhad low language ability
Not Exposeddid not have low language ability
Link The Nun Study
Also See How to Investigate an Outbreak for
excellent lesson on cohort studies
28
Epidemiology in Action
28
COHORT Studies Summary of the
Nun Study on Linguistic Ability
SummaryThe Nun Study indicates that low
linguistic ability in early life has a strong
association with dementia and premature death in
old age. Researchers investigated the
relationship of linguistic ability in early life
to the neuropathology of Alzheimers disease and
cerebrovascular disease. They analyzed 74 Nun
Study participants handwritten autobiographies,
written between ages 19 37. Idea density of
the journals was measured, and after death their
brains were autopsied and investigated for
evidence of Alzheimers disease pathology in the
neocortex. Low idea density scores from early
life were associated with the severity of
Alzheimers disease at the time of death.
Source Linguistic ability in early life and
the neuropathology of Alzheimer's disease and
cerebrovascular disease Findings from the Nun
Study In Vascular factors in Alzheimer's
disease, Volume 903, Kalaria RN, Ince P, eds.,
pp. 34-38, New York New York Academy of
Scicences, 2000.
Link The Nun Study
Note This site is filled with interesting
statistics, study abstracts, and other materials
on a wide variety of risk factors. Examples of
more advanced statistical analyses than those
covered in this curriculum are in plentiful
supply, and easily accessible to students.
29
Epidemiology in Action
29

• COHORT
• Advantages
• Exposure precedes health outcome, necessary to
  infer causation
• Less subject to bias because exposure is
  evaluated before health outcome is known
• Can examine multiple health results to exposure
• Disadvantages
• Expensive, time consuming
• Can infer cause, not prove
• Not as useful when the total population at risk
  is not well-defined

30
Epidemiology in Action
30
CASE CONTROL Studies compare people
with a condition (cases) to a people without the
condition (controls) to study risk factors.

For example, the first studies that pointed out
the strong connection between smoking and lung
cancer were Case-Control Studies.
Source CDC/ gross pathology of lung from smoking
This link describes some interesting current
studies on teens and smoking Epidemiology of
Youth Drug Abuse - Research Findings 2/00
31
CASE CONTROL Studies compare people
with a condition (cases) to people without the
condition (controls) to study risk factors.

Epidemiology in Action
31
Smoking and Carcinoma of the Lung
Disease Status of smokers of nonsmokers P-value
Males Lung cancer 647 (99.7) 2 (0.3
Males Controls 622 (95.8) 27 (4.2) 0.00000064
Females Lung cancer 41 (46.7) 19 31.7)
Females Controls 28 (46.7) 32 53.3 0.016
Doll R. Bradford, Hill A. Smoking and carcinoma
of the lung preliminary report. British
Medical Journal 1950, 2 739-748.
32

• CASE CONTROL Studies
• Advantages
• Can study rare health outcomes quickly/less
  expensively
• Can study multiple exposures for a single outcome
• Good for cases where the total population at risk
  is not well defined
• Easier to conduct
• Disadvantages
• Greater potential for bias because of smaller
  sample size and because selection is done after
  risk and outcome have occurred

Epidemiology in Action
32
33
Epidemiology in Action
33
CROSS-SECTIONAL Studies compare current
health and exposure status of groups AT THE SAME
TIME. Uses a large sample so that inferences can
be made regarding the whole population
For example, survey a large number of college
students in terms of heavy alcohol consumption,
a risk factor. (Exposure/Non-exposure). Also
survey how many students in the Exposed/Not
Exposed populations are suffering from a stomach
complaint, and compare the rates.
Links to recent studies on college-age youth and
alcohol use Research Matters College Alcohol
Study Boozing, "binge" drinking, and violence
among college students over time Dr. Ruth Engs
34
The PROPORTION of heavy drinkers with a stomach
ailment at the time of the study is higher than
the proportion of not-heavy drinkers.
34
CROSS-SECTIONAL Studies
PROPORTION reporting stomach ailment
35
Epidemiology in Action
35

• CROSS-SECTIONAL
• Advantages
• Easy to conduct because you dont have to wait
  for outcome to occur
• Good for examining relationship between a risk
  factor and outcome
• Disadvantages
• A cause cant be inferred because only current
  health and exposure are being studied
• Not good for conditions that are long-term

36
Epidemiology in Action
36
Student Assignment Your worksheet provides
descriptions of a number of studies. In groups of
two or three, based on what youve learned about
types of studies, classify each study according
to type, discuss advantages and disadvantages of
each, and describe a hypothetical study that
could be done that would be suited to this study
design.
Name That Study
37
37
FOR EXAMPLE
Name That Study
CDC/Dr. Demetri Vacalis 1999 line graph

1. Decide if this study can be best categorized as
   EXPERIMENTAL, DESCRIPTIVE, COHORT, CASE-CONTROL,
   CROSS-SECTIONAL, or a combination.
2. Defend your choice with two reasons.
3. How is this design an ADVANTAGE/DISADVANTAGE in a
   study of THIS population?
4. What would be a hypothetical study you could do
   on a population, that would be suited to this
   type of study design? Explain WHY this design
   would be best for this type of study.

38
38
Link/Source NCADI The National Treatment
Improvement Evaluation Study (NTIES) - Changes in
Physical Mental Health
Another EXAMPLE
Name That Study

1. Decide if this study can be best categorized as
   EXPERIMENTAL, DESCRIPTIVE, COHORT, CASE-CONTROL,
   CROSS-SECTIONAL, or a combination.
2. Defend your choice with two reasons.
3. How is this design an ADVANTAGE/DISADVANTAGE in a
   study of THIS population?
4. What would be a hypothetical study you could do
   on a population, that would be suited to this
   type of study design? Explain WHY this design
   would be best for this type of study.