Overview Research Methodologies

1
Overview Research Methodologies

• Fiona Williams
• Section of Public Health, CHS
• f.l.r.williams_at_dundee.ac.uk

2
Why?

• Natural history

3
Why?

• Describe

• Proportion of popn with disease
• Distribution of characteristics person, place
  and time
• Time trends
• Description of attitudes, behaviours and beliefs

4
Person, place and time
Age, gender, social status, smoking, BMI, blood
lipid profile
Person
Health board Urban rural Country
CHD
Time trends season
Place
Time
5
Why?

• Describe

• Proportion of popn with disease
• Distribution of characteristics person, place
  and time

Time trends Description of attitudes,
behaviours and beliefs
6
Why?

• Determinants of disease

Good health
Ill health
7
Why?

• Intervention

• Drug therapy
• Operations
• Change in ward practice
• Preventative measures
• Public health practice

8
Research question
Hypothesis
Does diet affect the incidence of bowel cancer?
Focussed
Fruit and vegetable consumption
Main outcome measure
Do 5 portions protect
9
Research methodologies

• Cross-sectional study
• Case-control study
• Cohort study

• Clinical case study
• Audit

Controlled trial RCT
10
Cross-sectional study

• Population examined once
• Distribution of characteristics in the popn at a
  given time
• Relationship between these factors
• Compare between groups to see whether data
  support an hypothesis
• PREVALENCE

11
Cross-sectional study
Disadvantages
Advantages

• Quick
• Several associations
• Standardized measurements
• Prevalence estimated

• Fairly expensive
• Interpretation of cause and effect not possible
• Recall bias
• No risk estimates possible

12
Case control study

• Rare (ish) events
• Define two groups on basis of whether disease is
  present or absent
• Look back in case notes (survey) to determine
  presence or absence of a factor
• Produce the 2x2 table odds ratio

13
Case-control study
ü
ü
ü
ü
14
Case-control study
74
48
25
52
Odds ratio 3.17
15
BIAS
Unrepresentative sample

• Interpret
• results
• incorrectly

recall and reporter
Match controls
More than one control
retrospective
Repeat in different populations
16
Case-control study
Advantages Disadvantages

• Quick
• Relatively cheap
• For rare conditions
• Odds ratio estimated

• Controls often contaminated
• Past records incomplete
• Recall bias
• Past measurements changed over time
• No absolute risk estimates possible

17
Cohort study

• Popn examined more than once
• Starts with cross-sectional study and then
  follows up prospectively
• Everyone starts with the factor but not the
  disease
• Relationship between factors measured and
  INCIDENCE of events
• 2x2 table relative risk

18
Cohort study
Relative risk
19
Cohort study
Advantages Disadvantages

• Standard measures used
• Disease process identified
• Incidence estimated

• Expensive
• Slow
• Disease must be common
• Large sample
• Drop out bias
• Drift in standardization

20
Characteristics of a researcher.

• Want to do research

Research question(s) Choose appropriate
methodology Develop a protocol Protected
time
21
Paper work.

• LREC / COREC
• Caldicott Guardian
• R D
• Clinical Governance
• RIS
• Annual monitoring

22
The researcher.
23
Help

• Research methods
• CAL http//www.personal.dundee.ac.uk/cdvflore/
• Modules research methods Public Health, CHS
• Ethics permission
• Mr Gus MacConnachie LREC /RD office, level 9
• www.dundee.ac.uk/epidemiology