Introduction to Epidemiology

1
Introduction to Epidemiology

• Cohort Studies
• Jan Risser, PhD

2
Research Design

• no comparison
• Aggregate data
• Prevalence cases
• Prevalence cases
• Incidence
• Incidence

• Case Reports / Case Series
• Ecologic Studies
• Case control
• Cross sectional
• Cohort Studies
• Experimental (Clinical Trials)

3
Research Design

• descriptive studies
• case report/case series
• cross-sectional
• analytic studies
• observational
• cross sectional
• case-control (retrospective)
• cohort (prospective, longitudinal)
• intervention
• experimental (clinical trials)

4
Research Design
5

• Numerator
• all those with the attribute at a particular time
• Denominator
• the population during that same time period

6
Prevalence

• denominator may be
• population - prevalence of syphilis in Houston
• sample - as with prevalence of iron deficiency
  among female athletes
• depends on an inflow of new cases and an outflow
  of old cases
• prevalence of influenza by month (old cases out,
  new cases in)

7

• Cumulative Incidence the probability (risk) of
  an individual developing the disease (outcome)
  during a specific period of time.

8

• Incidence density the rate of an individual
  developing the disease (outcome) during a
  specific period of time. Person-time is used as
  the denominator.
• One subject followed one year contributes one
  person-year of follow-up time.

9
Incidence

• Dynamic population
• total number of cases is fairly small
• population in flux
• no way of tabulating persons at risk
• Closed population
• population is fixed by study design
• individuals are followed periodically
• denominator includes only persons at risk

10
Incidence

• Closed population
• Development of iron deficiency anemia among
  female varsity athletes at Rice University.
• Dynamic population
• Incidence of influenza in Houston/Harris County,
  2002-2003 influenza season.

11
Prevalence and Incidence
12
Cross-sectional or Prevalence Studies

• Commonly used
• Survey of group to identify prevalence of disease
  or exposure
• Exposure and disease status assessed at the same
  time
• Individual is unit of observation and analysis
• Typically descriptive in nature to quantify
  magnitude of the problem

13
Cross-sectional studies

• also called survey or prevalence study
• involves disease prevalence
• usually involves random sampling and
  questionnaire measurement
• cannot distinguish whether hypothesized cause
  preceded the outcome
• used to generate hypotheses

14
Does smoking cause depression? Or does
depression cause smoking?
School Sample
Identify mood disturbed students
Identify Smokers
Findings strong association between
mood disturbance, especially depression and
smoking
15
Cross-sectional studies

• Focus is on some real population (target
  population) defined geographically and temporally
• Participants (study population) are selected
  because they represent that population

16
Cross-sectional studies

• National Health and Nutrition Examination Survey
  (NHANES)
• Single interview
• Prevalence of disease measured
• Prevalence of exposures measured
• Assessment of association between exposure and
  disease
• Only survivors can be surveyed

17
Cross-sectional studies
Disease Status
Yes
No
Total
b
a
Yes
a b
Exposure Status
No
d
c
c d
b d
a c
N
18
Cross-sectional studies

• Research questions usually have the following
  general form
• Is Factor A associated with Factor B?
• "What is the prevalence of Disease X or Factor Y
  in a given population?"
• "How many people in the population have Disease X
  or Factor Y?"

19
Cross-sectional studies

• From Corpus Christi Heart Project
• Question
• Among women hospitalized for a definite or
  possible myocardial infarction is there a
  difference in coronary heart disease risk factors
  between Mexican-Americans and non-Hispanic
  Whites?

20
Cross-sectional studies

• Variables - coronary risk factors
• Previous myocardial infarction (MI)
• Diabetes
• Hypertension
• Hypercholesterolemia
• Smoking

21
Cross-sectional studies

• Target population
• Mexican-American, non-Hispanic White women with
  MI, aged 25-74, residing in Nueces County, Texas
• Study population
• Mexican-American, non-Hispanic White women, 25-74
  yrs, from Nueces County, Texas and hospitalized
  MI

22
Cross-sectional studies

• Age-adjusted prevalence of selected risk
  factors
• Mexican- Non-Hispanic
• American White
• No. No.
• Previous MI 14 20.3 20 31.1
• Diabetes 36 53.0 23 34.1
• Hypertension 41 61.9 49 73.1
• High Cholesterol 24 35.7 21 32.7
• Current Smoker 16 24.7 26 42.7
• Ever smoked 35 53.1 45 66.4

23
Cross-sectional studies

• Sampling schemes
• Probability sample every element in the
  population has a similar probability of being
  included in the sample (a simple random sample)
• Non-probability sample not a simple random
  sample

24
Cross-sectional studies

• Measure of association
• Usual method of analysis
• Prevalence of disease in exposed compared to
  prevalence of disease in non-exposed and
  Prevalence Ratios calculated
• (a/(ab) / c(cd)
• prevalence odds ratios may also be calculated (ad
  / bc)

25
Cross sectional studies

• Strengths
• Program planning and justification
• Can differentiate population
• May generate new etiologic hypotheses

• Weaknesses
• No cause-effect
• No temporality
• Prevalent cases are survivors
• Rare events a problem
• Quickly emerging diseases a problem

26
Prevalence Studies
Incident cases All new cases arising in a defined
population
Time
Prevalent Cases Present at a point in time
Early Deaths
Leave Population Severe disease, Mild disease,
Move
Cures
27
Cohort studies

• longitudinal or prospective studies
• starts with people free of disease
• assesses exposure at baseline
• assesses disease status at follow-up
• must be at least two measurement
• baseline and follow-up
• individual is unit of observation and analysis

28
Cohort Study - Essence

• Examine the incidence of disease or mortality in
  relation to an exposure of interest in a subset
  of a define population over a specified time
  period
• Individuals free of disease
• Exposure known
• Follow-up
• Temporal sequence of exposure and disease

29
Selection of comparison group

• General population data
• Use of pre-existing data from the general
  population as the basis for comparison
• A study of asbestos and lung cancer used the US
  male population as the comparison group
• Caution healthy worker effect

30
Selection of comparison group

• Internal comparison
• Unexposed members of same cohort
• Stratify by level of exposure
• Framingham Heart Study
• Comparison or Separate cohort
• A cohort who is not exposed from another similar
  population
• Asbestos textile vs. Cotton textile workers

31
Cohort Study internal comparison group
Study Population or Defined Population
Exposed
Not Exposed
32
Cohort Study
Study Population
Target Population
External Population
Exposed
Not Exposed
33
Comparison Group

• Should to be as similar as possible to the
  exposed group with respect to all other factors
  except the exposure.
• If the exposure has no effect on disease
  occurrence, then the rate of disease in the
  exposed and comparison group will be the same

34
Diagram of a cohort study
Randomization
Control
Intervention
Lost to follow-up
Measure outcome
Measure outcome
35
Cross-sectional studies leading to cohort studies
Disease Status
Yes
No
Total
b
a
Yes
a b
Exposure Status
No
d
c
c d
b d
a c
N
36
Cross-sectional studiesselect the non-diseased
Disease Status
Yes
No
Total
b
Yes
Exposure Status
No
d
b d
N
37
Cohort studies
Disease Status
Yes
No
Total
b
a
Yes
a b
Exposure Status
bd from previous table
No
d
c
c d
b d
a c
N
38
Cohort study and Relative Risk
Disease
Y
N
a
b
ab
Y
Exposure
cd
d
c
N
ac
N
bd
Incidence exposed a/(ab) Incidence not
exposed c/(cd) Relative Risk a / (ab)
/ c / (cd)
39
Cohort study and Rate Ratio
E
E-
B
A
Diseased
AB
PTE
PTE-
Person-time
Incidence Density E A / PTE Incidence
Density E- B / PTE- Rate Ratio A / PTE
/ B / PTE-
40
Cohort Study

• Rate or Risk Differences
• Attributable Risk the absolute effect of the
  exposure or the excess risk of disease in those
  exposed compared with those non-exposed
• AR I(exposed) I(non-exposed)
• AR ID(exposed) ID(non-exposed)

41
Cohort Study

• Subject selection - exposure status must be
  accurate
• Outcome may be biased by exposure
• Groups should have similar risk of disease
  development other that exposure of interest
• Ascertainment of information should be similar
  among groups

42
Cohort Study

• DES study
• cohort of females exposed to DES and females born
  at the same time not exposed
• Exposed / non exposed similar
• age at first intercourse, numbers of sexual
  partners, history of sexually transmitted
  diseases, percent virgins
• RR 2.0 for any degree of squamous neoplasia

43
1951-1957 study of CHD by Cholesterol level -
Framingham
44
Prospective or Retrospective
1981
2001
Self Selection
Exposed
Not Exposed
2001
2021
Measure outcome
Measure outcome
45
Retrospective Cohort Study

• Nurses Health Study
• 1976-1979
• 14,607 married female participants reported
  vasectomy as the couples method of birth control
• 1989
• 14,607 original women and 14,607 age-matched
  controls whose husbands had not had a vasectomy
  prior to 1978

46
Retrospective Cohort Study

• 96 new cases of prostate cancer were diagnosed
• Vasectomy was associated with an increased risk
  of prostate cancer
• Age-adjusted RR 1.56 (95 Confidence Interval
  1.03 2.37)

47
Body Fat and 5-year risk of death

• 41,837 women (55-69)
• Base line information about height, weight, waist
  and hip measurements
• Followed 5 years. 1,504 deaths
• BMI associated with Death in J shaped fashion
• Waist/hip circumference ration associated with
  mortality
• Each 0.15 unit increase 60 greater risk of
  death.

48
BMI compared to WH
49
Doll Hill, 1970

• Cohort study of smoking and lung cancer among
  male physicians
• Target population - male physicians
• Exposure categories
• Non-smokers
• Light smokers
• Moderate smokers
• Heavy smokers

50
Cohort Study
51
Cohort Study
52
Cohort Study - Counfounding

• severity of co-morbid conditions confounds
  association between ACE inhibitors and ESRD
• smoking confounds association between coffee
  intake and pancreatic cancer
• obesity probably confounds association between
  serum cholesterol levels and CVD risk
• loss to follow-up confounds relationship between
  very low calorie diets and weight loss

53
Confounding

• Control confounding by
• Design stage
• restriction (limit eligibility)
• matching (as with DES)
• Analysis stage
• stratification (analyze by strata)
• multivariate analysis (Cox proportional hazard
  model, logistic regression)

54
Cohort studies

• Strengths
• determination of incidence and risk
• multiple exposures and multiple outcomes
• can establish cause - effect
• good when exposure is rare
• temporal relationship clear
• minimizes selection and information bias

• Weaknesses
• validity affected by losses to follow-up
• often requires large sample
• ineffective for rare diseases
• long time to complete
• expensive
• subject attrition
• confounding by other factors possible

55
Comparison of Studies
56
Comparison of Studies

• Incident Cases
• all new cases in a defined population
• Over time
• early deaths
• cures
• leave population
• Prevalent Cases
• all cases present at point in time

57
Study Design

• Major study designs differ based on
• Number of observations made
• Directionality of exposure
• Data collection methods
• Timing of data collection
• Unit of observation
• Availability of subjects

58
Ecologic Studies

• Also called correlational studies
• Focus is on multiple groups that are defined
  geographically or temporally
• Purposes include
• generating hypotheses for further
• testing hypotheses about an association

59
Ecologic Studies

• In an ecological study, variables may be measured
  on individuals, although comparisons are always
  between groups, never between individuals
• Results are extremely prone to confounding (bias)

60
Ecologic Studies

• Research questions usually have the following
  general form
• Is the prevalence (or average level) of Factor X
  associated with the prevalence or incidence of
  Disease (or Factor) Y?

61
Ecologic Studies

• General question
• Is the mineral content of consumed water
  associated with coronary heart disease?
• Specific question
• Among men in England and Wales is the calcium
  content of a community's water supply associated
  with mortality due to coronary heart disease?

62
Ecologic Studies

• Independent variable
• Calcium content of the municipal water supply
• Dependent variable
• Cardiovascular mortality rate of the community
• Target population
• Not stated, industrialized nations is implied
• Study population
• 61 towns in England and Wales

63
Ecologic Studies

• Soft Water Towns Hard Water Towns
• Number examined 245 244
• Age 51.5 51.7
• Height, cm 173.6 174.0
• Weight, kg 73.7 74.5
• never smoked 19.4 16.9
• ex-smokers 30.0 36.3
• smokers 32.9 27.0
• Mean cigarettes/day 15.9 14.0
• Plasma cholesterol (mg/100 ml) 245.0 237.4
• Heart rate 77.8 74.0
• BP-systolic 141.7 139.3
• BP-diastolic 87.6 85.5

pLancet 20 Jan 1973
64
Ecologic Studies

• Interpretation
• Men residing in communities with hard water have
  a lower rate of cardiovascular mortality, lower
  plasma cholesterol, lower heart rate,and lower
  diastolic blood pressure than men residing in
  communities with soft water.

65
Ecologic Studies

• Correlation Coefficient
• r the measure of association
• r quantifies the extent to which there is a
  linear relationship between exposure and disease.
  
• r ranges between 1 and 1

66
Ecologic Studies

• Distinguishing characteristic
• Determines an association between two variables
  by comparing groups
• Groups are formed based on a variable which is
  not being studied, such as a geographic or
  temporal variable

67
Ecologic Studies

• Strengths
• Quick and relatively inexpensive
• Simple to conduct
• Availability of data from surveillance programs
  and disease registries

• Weaknesses
• Cant link exposure with disease
• Cant control for potential confounding factors
• Ecological fallacy making a causal inference
  on the basis of group observations

68
(No Transcript)
69
(No Transcript)
70
(No Transcript)
71
Study Designs

• What is the prevalence of illicit drug or alcohol
  use during pregnancy in Houston, TX?
• Survey of pregnant women

72
Study Designs

• Is the per capital revenue from cigarette taxes
  in the counties of California associated with
  lung cancer mortality rates for those counties?
• No individual data

73
Study Designs

• Do women who have more than 10 years of oral
  contraceptive use have a higher incidence of
  developing breast cancer than women who have
  never taken oral contraceptives?
• At risk followed for development of disease

74
Study Designs

• A total of 10,000 Vietnam veterans, half of whom
  are known by combat records to have been in areas
  where Agent Orange was used and half of whom are
  known to have been in areas where no Agent Orange
  was used, are asked to give a history of cancer
  since discharge.

75
Study Design

• Is the incidence of perinatal infection among
  infants born to mothers chronically infected with
  hepatitis B associated with the maternal viral
  titers?
• Exposure maternal viral titers
• Outcome incidence of perinatal infection

76
Study Design

• Infants born to hepatitis B chronically infected
  mothers were randomized and assigned to one of
  two groups
• one received hepatitis B vaccine and
• the other received placebo
• All infants were followed for two years to
  determine the incidence of hepatitis B infection

77
Study Design

• Smoking histories are obtained from all patients
  entering a hospital who have lip cancer and are
  compared with smoking histories of patients
  admitted to the same hospital for cosmetic
  surgery of the face
• Disease lip cancer
• Exposure smoking history

78
Study Design

• The entire population of a given community is
  examined, and all are questioned extensively
  about their diet, and history of colon cancer.
• These people free of colon cancer (by
  self-reported history) are then followed for
  several years to see whether their eating habits
  will predict their risk of developing bowel
  cancer.

79
Study Design

• To test the efficacy of vitamin C in preventing
  colds, Army recruits are randomly assigned to two
  groups one given 500 mg of vitamin C daily, and
  one given a placebo.
• Both groups are followed to determine the number
  and severity of subsequent colds.

80
Study Design

• The physical examination records of the incoming
  first-year class of 1935 at the University of
  Minnesota are examined in 1980 to see whether the
  freshmens recorded height and weight at the time
  of admission to the university were related to
  their chance of developing CHD by 1981. 

81
Study Design

• 1500 adult men who worked for Lockheed Aircraft
  were initially examined in 1951 and were
  classified screened for coronary artery disease
  risk factors (including weight, diet,
  hypertension, family history)
• Every 3 years they have been examined for new
  cases of the disease attack rates in different
  subgroups have been computed annually.

82
Study Design

• A random sample of middle-aged sedentary women
  was selected from four census tracts, and each
  subject was examined for evidence of
  osteoporosis.
• Those found to have the disease were referred to
  their primary medical provider.
• All others were randomly assigned to either an
  exercise group, which followed a 2-year 3 day a
  week program of systematic exercise, or a control
  group, which had no exercise program, but met 3
  days per week to play cards and do crafts.
• Both groups were observed semi-annually for
  incidence of osteoporosis.

83
Study Design

• Questionnaires were mailed to ever 10th person
  listed in the city telephone directory. Each
  person was asked to provide his or her age, sex
  and smoking habits and to describe the presence
  of any respiratory symptoms during the preceding
  7 days.

84
Cohort Study - Groups similar?

• ACE inhibitors and risk of ESRD
• are patients who are prescribed ACE inhibitors
  different than those prescribed other drugs
• Serum cholesterol and risk of CVD
• what other life style differences might there be
• Coffee and risk of pancreatic cancer
• what are potential confounders
• loss to follow-up
• what if loss to follow-up associated with exposure