Make your own free website on Tripod.com

Epidemiology

Data types

Rate
= frequency/population.
Prevalence
Proportion of defined population with a condition at any one point in time.
Incidence
Number of new cases arising over a defined period of time in a defined population.
Risk
Probablity that people with certain exposures will acquire a disease.
Relative risk
= Risk in exposed/Risk in nonexposed
Excess risk
= Risk in exposed - Risk in nonexposed
Population excess risk
= excess risk × prevalence of exposure
How much the incidence would drop if exposure was eliminated.
Population excess fraction
= population excess risk/incidence of disease
Fraction of disease attributable to exposure.
Numbers needed to treat (NNT)
1/absolute risk reduction
Comparison of prevalence and incidence
PrevalenceIncidence
NumeratorTotal cases countedNew cases
DenominatorAll individuals examinedPopulation at risk
TimePoint in timeTime period
How measuredCross sectional studyCohort study
Comparison of risk measures
Relative riskExcess riskPopulation excess risk
MeasuresForce of effectSize of effect for individualSize of effect for population
GeneralisableYesNoNo
UseResearchPracticePractice

Study types

Ecological study
Observational study examining the relation between disease rates and exposure; e.g. migrant study showing 2nd generation Japanese migrants to USA have lower rates of gastric carcinoma (implies environmental rather than genetic cause); cervical cancer linked to sexual promiscuity.
Advantages: cheap, simple (often uses published data).
Disadvantages: very poor evidence, cannot determine temporal relationships.
Used to direct future research.
Case-control study
Cases identified and matched for controls. Exposures are then determined retrospectively. Odds ratios may be calculated.
Advantages: able to study disease with low incidence.
Disadvantages: difficult to determine temporal relationship between exposure and disease, bias in recalling exposure.
Cohort study
Sample population with known risk factors, followed up for a period of time. May or may not be controlled. E.g. The Longitudinal study (1% population defined at 1971 census), British Doctors Study (smoking).
Advantages: temporal relationship between exposure and disease is clearer; abloe to ethically study exposures which are thought to be harmful; better evidence than case-control study.
Disadvantages: cannot study diseases with a small incidence, people may drop out of study, lag between exposure and disease may be very long.
Randomised-controlled trial
Disadvantages: may only ethically study exposures which are thought to be benficial.
Support for causality:
  1. Biologically plausible
  2. Association is strong
  3. Evidence is good (RCT better than CCT)
  4. Dose-response relationship
  5. Time sequence compatible (exposure always before disease)
Wilson and Jungner criteria