Population Health and Evidence-based practice Flashcards
Epidermiology
The study of distribution and disease frequency in human populations and the application of the study to control health problems
Public Health
The science and art of preventing diseases, prolonging life and promoting health through the organised efforts of society
Randomisation
-All participants has an equal chance of being in each group
Aims to:
1) Avoid allocation bias
2) Ensure balanced baseline similarities
Can be done by:
1) Toss of a coin
2) Random number tables
Blinding
Single blinded – either assessor or participant is blinded
Double blinded – both participants and assessor are blinded
Aims to:
1) Avoid assessment bias in how assessor report participant’s health outcomes
2) Avoid potential bias in participant’s response and influence decision to withdraw from the trial
Placebo
- A control drug that is inactive and indistinguishable from treatment drug
- Helps to take into account the placebo effect when analysing treatment efficacy
- Aids blinding
Intention to treat
-All participants in a trial are analysed irrespective of their compliance with treatment
Upsides:
1) Maintains baseline similarities
2) Gives a real-world estimate of treatment efficacy
Downsides:
1) Underestimate of true efficacy
Per protocol analysis
-Only limit data analysis to participants that comply with treatment
Downsides:
1) Potential bias (characteristics of non-compliers differs from those that comply –> Biased comparison)
Relative risk
Risk of intervention group : Risk of control group
*< 1 = decreased risk, > 1 = increased risk, = 1 (no effect)
Absolute risk
1 - relative risk
*Negative no. means less risk
Numerical data
Classified into:
1) Discrete – only take certain values (e.g. number of long distance flights in a month)
2) Continuous – take any value in a range (e.g. weight, height)
Categorical data
Can be classified into:
1) Ordinal – > 2 categories with a natural order (e.g. stages of cancer)
2) Nominal – > 2 categories with no natural order (e.g. blood type)
3) Binary – can only take 2 values (e.g. yes/no)
95% Reference range
- Measure of spread of continuous numerical data only
- = Mean +/- 1.96 x SD
95% Confidence interval
- Measure of precision of sample estimate (where the ‘true’ population lies)
- = Mean +/- 1,96 x SE
SE = SD/√(n)
Run-in phase
Aims to:
1) Reject participants that are unlikely to comply with treatment
2) Check eligibility of participants
3) Ensure comparable baseline measurements
Null hypothesis
-Hypothesis that proposes there is no difference between certain characteristics in a population
E.g. there is no difference between systolic BP between normal sodium intake and reduced sodium intake
P-value
-The PROBABILITY of a result occurring by chance if null hypothesis is true
- p < 0.05 = Reject null hypothesis
- p > 0.05 = Accept null hypothesis
Type 1 error
Rejecting null hypothesis even though it is true
Type 2 error
Failing to reject null hypothesis even though it is false
