Study Design and Interpretation Flashcards
Define cohort study
Looking at a group who don’t have the disease, following their exposures, and seeing who develops the disease.
Define case control study
Looking at ‘cases’ with the disease, and ‘controls’ without the disease, then looking back to see what they were exposed to.
Define randomised controlled trial
Exposing one group to an intervention, and one group to a placebo and comparing outcomes.
Define cross sectional study
A snapshot in time of people with the disease.
Define ecological study
Geographical - comparing different populations.
Temporal - comparing groups over time.
What can associations be due to?
Chance, bias, confounding, reverse causality, true causal association.
Describe the 6 Bradford Hill Criteria for Causation
Temporality - exposure precedes outcome.
Dose-response (increased exposure = increased risk.)
Strength (size of effect.)
Reversibility (removal or exposure = reduction of risk.)
Consistency (same results from different studies etc.)
Biological plausibility.
Describe 5 types of bias and what they mean
Selection: systematic error in selection of participants or allocation of groups.
Information bias: error in measuring or classifying exposure or outcome such as observer bias, participant bias, instrument bias.
Publication bias: only publishing positive results.
Lead time bias: early identification appearing to increase survival.
Length time bias: diseases progressing more slowly will be picked up earlier, making it look like the survival time is longer.
Define confounding
An apparent association between exposure and outcome which is actually the result of another factor.
Define incidence
The number of new cases during a time period.
Define prevalence
The number of existing cases at a specific time point.
Define relative risk and how to calculate it
Risk of one group compared to another.
Absolute risk in exposed/absolute risk in non-exposed
Define absolute risk and how to calculate it
Number of people with the disease/total exposed population
Define attributable risk (and it’s other name) and how to calculate it
Absolute risk reduction.
Amount of disease specifically due to the exposure.
Risk in exposed - risk in unexposed
Define NNT and how to calculate it
Numbers needed to treat - how many Px need Tx to prevent one bad outcome.
1/attributable risk