Lecture 4 : Ecological Studies, CrossSectional Surveys

1

Epidemiology

• Lecture 4 Ecological Studies, Cross-Sectional
  Surveys
• and Before and After Studies

2
Learning outcomes

• Describe the key features of ecological studies
• Describe the key features of cross sectional
  surveys
• Describe the key features of before and after
  studies
• Interpret epidemiologic data related to
  ecological studies, cross sectional surveys and
  before and after studies

3
Ecological Studies

• Determine whether populations with high exposure
  levels also have high levels of outcome
• Often the initial approach to examining a
  potential association between an exposure and an
  outcome
• Focus on the occurrence of disease and patterns
  of risk factors, disease/outcomes among
  populations
• Rely usually on administrative data often
  collected for other purposes

4
Ecological Studies

• Search for associations between disease
  occurrence and the presence or absence of
  possible etiologic agents in populations
• across several geographical areas at one
  particular time (cross sectional ecologic study)
• over time (longitudinal ecologic study)

5
Rate of decayed/missing/filled teeth regional
differences
6
Rate of decayed/missing/filled teeth change
over time
7
CDC, MMWR 1999 48621  
8
Incidence of poliomyelitis, US, 1951-90
CDC, 1991 quoted in Jekel
9
Correlation between dietary fat intake and breast
cancer incidence by country
Prentice,J Natl Cancer Inst, 1988 quoted in Gordis
10
Ecologic fallacy

• Attributing to members of a group, group
  characteristics that they do not possess as
  individuals

11
Interpreting Ecological Studies

• Changes in trends in certain data sources may be
  artifactual due to
• changes in coding practices
• change in ICD versions or revisions
• training of physicians

An inaccurate observation, effect, or result,
due to the technology used, assumptions made or
other error.
12
Interpreting Ecological Studies

• Changes in trends in certain data sources may be
  artifactual due to
• changes in diagnosis
• new tests, procedures or
• new definitions for assigning diagnosis

An inaccurate observation, effect, or result,
due to the technology used, assumptions made or
other error.
13
Interpreting Ecological Studies

• Changes in trends in certain data sources may be
  artifactual due to
• changing denominator population
• boundary changes eg postcodes
• changing geographic coverage or census
  inaccuracies eg Changing definition for rural and
  remote

14
Cross Sectional Survey
15
Descriptive Cross-sectional survey

• are used to measure the prevalence of disease in
  a population
• Example
• Repetition of these surveys can be very useful in
  finding out whether changes are occurring over
  time called monitoring or surveillance

16
Regular health surveys

• In Australia there are a number of regular health
  surveys conducted on a random sample of the
  population
• Focus on personal and demographic
  characteristics, illness and health related
  habits
• These surveys may also examine frequencies of
  disease and other characteristics may be examined
  in relation to age, sex, location, education, etc

17
Regular health surveys

• Data from such surveys are helpful in needs
  assessments and health

18
ABS National Health Survey

• Conducted by ABS every 3 years
• Data on around 40,000 people randomly sampled
  across Australia
• Questionnaire self-administered sometimes
  interviewer assisted
• Focus Health states and health actions
• Data has been used to analyse the relationship
  between certain health states and risk factors

19
NSW Health Survey

• First conducted in 1997 and then 1998, now a
  continuous program
• Random sampling of telephone numbers
• Uses CATI to administer questionnaire
• Data on around 34,000 people (1997-1998)
• Focus Health states, health actions,
  satisfaction with services, evaluate management
  of asthma and diabetes
• Data has been used to analyse the relationship
  between certain health states and risk factors

20
(No Transcript)
21
Analytical Cross Sectional Surveys

• Obtain data on prevalence of outcome and exposure
  to compare differences in outcome between exposed
  and unexposed
• Use a defined population at a particular time.
• Exposure and outcomes are determined at the same
  point in time

22
Analytical Cross Sectional Surveys

• Four possible groups can be identified from the
  study
• Group a exposed and has the outcome
• Group b exposed and has no outcome
• Group c not exposed and has the outcome
• Group d not exposed and has no outcome
• Can calculate the RR or OR

23
Example

• Is milk drinking associated with peptic ulcer?

24
Issues interpreting cross sectional survey data

• Non Response bias
• Prevalent cases
• Prevalence influenced by incidence and duration
  of disease
• Long term survivors more likely to be counted
• Temporal relationship between exposure and
  outcome is difficult to ascertain
• antecedent-consequent bias

25
Advantages Disadvantages

• Cheap
• Fast
• Good for hypothesis generating

• Prevalent cases
• antecedent-consequent bias
• Poor at hypothesis testing

26
Before and after studies (BAS)

• Also called pre and post tests
• Often used in the testing of public health and
  health promotion interventions
• Assess what is happening before and after an
  intervention is in place
• Where multiple measures are made before and after
  the intervention is implemented, the study is
  called a time series.
• The design of both time series and before and
  after studies may be strengthen by the use of a
  control group.

27
Time Series
Intervention
Time
A
Outcome measures
Time
B
Outcome measures
28
Issues interpreting before and after studies

• Other factors that may affect outcomes
  (confounding)
• Selection of population bias (e.g. asthma in
  animal handlers)
• Control group subjects serve as their own
  control
• Temporal relationship attributed to the
  intervention

29
QUESTION 1

• Which of the following is not a feature of a
  cross-sectional study?
• Fairly quick and easy to perform
• Useful for determining prevalence of disease and
  risk factors
• Useful for determining temporal relationship
  between cause and effect
• Not useful for determining temporal relationship
  between cause and effect

30
QUESTION 2

• Which of the following is a feature of an
  ecological study?
• Not useful for suggesting hypotheses
• Useful for testing hypotheses
• Useful for measuring trends in disease rates
• Is an experimental study design/type

31
QUESTION 3

• Which of the following statements is not a
  feature of analytical cross-sectional studies?
• Measures of association (RR, OR) cannot be
  calculated
• Measures of association (RR, OR) can be
  calculated
• Exposure and outcomes are determined at the same
  point in time
• Four possible groups can be identified from the
  study

32
QUESTION 4

• A time series study is
• Where multiple measures are made before and after
  the intervention is implemented
• Where multiple measures are only made before the
  intervention is implemented
• Where multiple measures are only made after the
  intervention is implemented
• None of the above

33
QUESTION 5

• Cross-sectional surveys
• Preferentially identifies chronic cases
• Preferentially identifies acute cases
• Equally identifies acute chronic cases
• None of the above

34
ANSWER 1

• Which of the following is not a feature of a
  cross-sectional study?
• Fairly quick and easy to perform
• Useful for determining prevalence of disease and
  risk factors
• Useful for determining temporal relationship
  between cause and effect
• Not useful for determining temporal relationship
  between cause and effect

35
ANSWER 2

• Which of the following is a feature of an
  ecological study?
• Not useful for suggesting hypotheses
• Useful for testing hypotheses
• Useful for measuring trends in disease rates
• Is an experimental study design/type

36
ANSWER 3

• Which of the following statements is not a
  feature of analytical cross-sectional studies?
• Measures of association (RR, OR) cannot be
  calculated
• Measures of association (RR, OR) can be
  calculated
• Exposure and outcomes are determined at the same
  point in time
• Four possible groups can be identified from the
  study

37
ANSWER 4

• A time series study is
• Where multiple measures are made before and after
  the intervention is implemented
• Where multiple measures are only made before the
  intervention is implemented
• Where multiple measures are only made after the
  intervention is implemented
• None of the above

38
ANSWER 5

• Cross-sectional surveys
• Preferentially identifies chronic cases
• Preferentially identifies acute cases
• Equally identifies acute chronic cases
• None of the above