2. Refresh... why is epidemiology
important?
What is epidemiology?
Inform decisions
Monitor change
Equity
3. Why do we collect data?
Monitoring and surveillance
Identification of outbreaks
Identification of areas of public health significance
Effectiveness of programs
Linking exposure and disease
To identify causes of disease/ states of health
Provides indication of risk of disease based on
exposure
5. Epidemiological approaches are
applied in situations in which we
are interested in upstream factors
i.e. Personal, community or
environmental risk factors known to
affect disease rates
6. What are we measuring?
Need to have clear idea about what we are trying to determine
Prevalence vs Incidence
Prevalence - proportion of population that has disease at given time (i.e. Existing
cases)
Number of people with disease at given point in time
___________________________________________
Total number of people in population
Incidence – rate at which people develop disease/ health outcome (i.e. New
cases)
Number of people who develop disease in given time period
_________________________________________________
8. Case reports and case series
Beginning point for scientific study – identify
potential health outcomes, stimulate interest in
area, potential to advance knowledge
Detailed descriptions of one/ more cases that are
unusual
Selective nature and limited amount of information
provided
provide little evidence of causality
cannot provide much information about patterns of
disease
9. Ecological study design
First step in determining whether association
exists
Studies of differences in health status (death
rates, disease patterns, health-risk
behaviours) between countries and regions,
or within the same region at different times
Use existing data sources
Aim to generate hypotheses about
environmental or behavioural exposures and
disease
10. The group, rather than the individual, is the
unit of analysis
Advantages
1. Data easily available - studies inexpensive
2. Hypothesis-generating – suggest avenues for
research
Disadvantages
1. Ecological fallacy: might not accurately represent
the exposure – disease relationship at the
individual level
2. Cannot directly link “exposure” with the
“outcome”
11.
12. Extending beyond surveys of
individuals
Food supply data
Nutrient composition of foods
Australian Household Expenditure Survey
International comparisons
13. Food supply data
To calculate „available food for consumption‟ – Food balance
sheets
Information on amounts of food (raw commodities)
available for consumption per year
Food available for use =
production + imports – exports
Food available for consumption =
production + imports – exports – industrial use – animal use
Important – food available for consumption does NOT tell us
how much food is actually eaten!
14. Food supply data
To compare food trends within Australia Apparent
food consumption data
Apparent consumption =
(commercial production + estimated home production +
imports + opening stocks)
(exports + usage for processed foods + non-food usage +
wastage + closing stocks)
MINUS
15. Cross-sectional study design
Begin with a defined population
Exposed
+ have
disease
Exposed
+ do not
have
disease
Not exposed
+ have
disease
Not exposed
+ do not
have
disease
Gather data on exposure and disease
16. Advantages of cross-sectional study design
1. Relatively inexpensive
2. No follow-up required
3. Less participant burden
Disadvantages of cross-sectional study design
1. Cannot assess temprality
2. Can establish associations, but not causation
3. Considerable potential for confounding
4. Neyman (prevalence) bias
(diseases or exposures that are longer-lasting will be
over-represented relative to those of shorter
duration)
17.
18. Australian Health Survey
http://www.abs.gov.au/websitedbs/D3310114.n
sf/home/Australian+Health+Survey+-
+Frequently+Asked+Questions
19. Case-control study design
Observational, retrospective & focused on
individuals
Comparison of exposure frequencies
among diseased (case) and non-diseased
(Control) groups
20. NOT EXPOSED
EXPOSED
NOT EXPOSED
EXPOSED
Population
TIME
Adapted From: Fletcher et al (1996) Clinical Epidemiology: the essentials.
Baltimore: Williams and Wilkins.
CASES
(people with
disease)
TIME
Direction of inquiry
CONTROLS
(people without
disease)
Design of case-control study
21. Advantages of case-control study design
1. Suitable for rare diseases
2. Possible to evaluate a large number of
potential causes/risk factors
3. Efficient design, requiring less time and more
modest costs than prospective studies
4. Higher on hierarchy of study designs
22. Disadvantages of case-control study design
1. Potentially difficult to distinguish exposures that precede
disease (antecedent causes) from concurrent associated
factors
2. Requires representative samples of cases and controls
3. Selection of appropriate controls can be difficult
4. Generalisability of findings
Refusals
Representation of original populations
5. Results potentially flawed due to
Reliance upon accurate historical information about exposure, in
particular recall may differ between cases and controls (recall bias)
Unsuitability of controls (lack of comparability with cases or
overmatching)
6. Still cannot be CERTAIN that if exposure preceded the
disease, then it is in fact causal
23.
24. Cohort study
Observational, prospective & focused on
individuals
Essence is comparison of disease
frequencies among exposed and non-
exposed groups
25. Population
People
without the
disease
TIME & Direction of inquiry
Disease
Disease
No disease
No disease
Adapted From: Fletcher et al (1996) Clinical Epidemiology:
the essentials. Baltimore: Williams and Wilkins.
Exposed
Not Exposed
Design of cohort study
26. Advantages of a cohort study
1. Possible to establish temporality
2. Possible to study several outcomes from
exposure to same hazard
3. Bias - less of a problem than case-control or
cross-sectional studies
27. Disadvantages of a cohort study
1. Expensive in terms of time and money
2. Large, representative sample required
3. Higher risk of attrition bias
4. Measures used in early steps of study may
change (eg, due to technological
improvements) and so results at different
points in time may not be directly
comparable
28.
29. Randomised controlled trials
(RCT)
Experimental, prospective
Types
Clinical trials – patients
Prevention trials – healthy people
Community trials – community level
30. The structure of a clinical trial
Population of
patients with
the condition
SAMPLE TIME
Improved
Improved
Not Improved
Not Improved
Experimental intervention
Comparison intervention
32. Advantages of RCT
Advantages of RCT
1. If randomisation is successful,
considered strongest design for causal
inference
2. Researchers control the exposure –
specific and accurate
Disadvantages of RCT
1. Expensive and time consuming
2. Attrition
3. Compliance
34. Hierarchy of study designs
Randomised control trial
Cohort study
Case-control study
Cross-sectional study
Ecological study
Experimental
Observational
Increasingcostandevidence
Descriptive – describes occurrences of disease or exposure. Most likely to be used to look for patterns of disease and to measure the occurrence of disease or risk factors.Analytic studies – involves planned comparisons between people with/ without disease or exposed/ not exposed groups.
Time point for prevalence should always be reported.Long lasting disease may have high prevalence, but low incidence. Disease that rapidly resolves may have low prevalence, but high incidence. To measure incidence need to start with cohort of people who are disease free.
Dietary behaviours are complex and are the culmination of a range of factors including access, availability and affordability to foods that, arguably, are beyond the realm of the individual and are characteristics of the food supply. This contrasts with physical activity, where there may be fewer contextual factors that may influence an individual’s behaviour. In understanding the epidemiology of dietary behaviours, nutrition-related health outcomes and trends among populations, an understanding of the food supply with which the population interacts is important. In bringing about better health among populations, the scope of public health nutrition extends beyond health promotion and policy strategies that encourage individuals to make more healthy dietary choices to ensuring that the available food supply also promotes these choices.
In Australia, information about household expenditure on food is collected periodically as part of the Australian Bureau of Statistics’ Household Expenditure Survey. Limitations of this data are that it does not collect information on the quantity of foods purchased and the descriptions of foods purchased are not sufficient to allow for analyses of nutrient contents. Information on food supply can take a variety of factors into account, including production, imports, export, industrial and animal use of food, and waste.Food Balance Sheets Provide information about the amount of food available for human consumption in a country in a given year from the FAOtakes into account production, changes in stocks, imports & exports, agricultural & industrial use of foodstuffsrelates only to primary producegive trends in food supply over time both within & between countries Relates to national average, provides estimates per kg of food per head per year OR per grams of food per head per day.Important– food availability data is not the same as food consumption data (i.e. Information on the availability of food does not tell us how much food is actually eaten).
To compare trends in ‘apparent’ food consumption over time: Still an estimate as not measuring actual intake at the household levelApparent Consumption Data Derived by ABS in way similar to food balance sheets, however reporting is not limited to primary produce Not used for all food products e.g. if there was a better way of deriving consumption or for foods for which all components of the equation were not available (some milk products, beer, eggs, wine, cheese etc)Gives overall trends & correlationsData is useful in nutrition planningLong term data minimizes daily & seasonal variationsLarge sample populations increase validity of conclusionsNo placebo effectNo participant error