HSPH scientist Flashcards
Bias in case control trials
Recall
Reverse causality
Selection of cases
Selection of controls
What is a nested case control study?
Cohort study
Then go back and do case control
Bias in cohort studies
Loss to follow up
Cohort selection
Exposure usually measured at just one point in time
Kaplan Meier curves
Plot survival over time
Can be used to generate Hazard ratios
Controls for RCTs
Geographical
Historical
Randomised
Parallel RCT
Only receive one treatment
For irreversible treatments
Crossover RCT
Receive both eventually Sequence randomised For reversible treatments Person acts as own control But takes longer Carry over effects --> need washout period
Cluster RCT
Randomise pre-existing groups
E.g. schools, villages
Avoids contamination
Improves compliance
Factorial RCT
Tests >1 intervention
Sample power minimum
85%
Forest plot
Shows results of meta analysis
Problems with meta-analysis
Heterogeneity of data
- variation in results due to different methods, intervention
Publication bias
- data finding significant results favours publication
- inconclusive data less likely to be published
SD and % of data
1 SD = 68%
- 64 SDs = 90%
- 96 SDs = 96%
- 58 SDs = 99%
Two theories of ontology
Objectivism/realism = world is independent of us Subjectivism/contructivism = world is constructed by us
Two theories of epistemology
Positivism = things can be studies systematically like nature Interpretivism = humans are different so must be studied differently