Flashcards in Intro to Epidemiology Deck (45):
What does a meaningful statistic need?
1. A denominator population
2. A time frame
What are examples of denominators?
+ Health board
+ Disease register
+ Recruited to a study
Denominator must correspond to numerator - without denominator pop. and time dath rate are meaningless
What is timeframe?
+ n-year follow-up
What is incidence?
+ Number of new cases
+ A rate or proportion
+ Useful for identifying causes of diseases
+ Occurs, by definition, only in people without the disease
What is prevalence?
+ Proportion of population that has disease:
+ Identifies disease burden
+ Useful for planning services
+ Depends partly on incidence
Occasional cases occuring irregularly
Persistent background levels of occurence (low to moderate levels)
Occurence in excess of the expected level for a given time period
Epidemic occuring in or spreading over more than one continent
What are non-modifiable exposures?
What are modifiable exposures?
+ Alcohol consumption
What is risk?
(No. outcomes in group / No. people in group) x 100
What is relative risk (RR/risk ratio)?
Risk in exposed / Risk in unexposed
What is the relative risk reduction (RRR)?
(1 - Relative risk) x 100
What is the absolute risk reduction (ARR/risk difference)?
Risk in unexposed - Risk in exposed
What is number needed to treat (NNT)
1 / Absolute risk reduction
Odds ratio (OR)?
Commonly used estimate of risk ratio (non-RCT study designs)
Rate ratio (RR)?
Ratio between two mortality rates, hospitalisation rates stc.
Hazard ratio (HR)?
A special kind of rate ratio (survival analysis)
What are confidence intervals?
+ Represents range of plausible values
+ Can be presented for any statistic/effect measure
+ Values near the limits/more extreme values less plausible than those in the middle
+ The wider the interval the greater the uncertainty
+ Very useful in appraising published research
Steps in cross-sectional study?
+ Sample population
+ Estimate the population:
- different exposures
- different signs/symptoms
- different outcomes
+ Use data:
- to describe prevalence/burden
- to explore associations
Steps in case-control study?
+ Select cases with an outcome
+ Select controls without the outcome
+ Explore EXPOSURES in cases and controls
+ Compare exposures in cases and controls
+ Identify association
Steps in a cohort study?
+ Select people without an outcome
+ Classify according to an exposure
+ Compare RISK od disease in exposed and unexposed
Steps in randomised controlled trial (RCT)?
+ Random allocation:
+ Compare RISK of outcome in interventon and control groups
Study design? Objective: Treatment effect
Study design? Objective: Cause
Study design? Objective: Prognosis
Study design? Objective: Incidence
Study design? Objective: Prevalence
Study design? Time-frame: Future
+ Cohort: prospective
Study design? Time-frame: Past
+ Cohort: retrospective
What is confounding?
When a true relationship gets "confused" by another factor
What is bias?
- what data are collected
- how data are collected
- how data are analysed
- how data are interpreted
- how data are reported
What does bias lead to?
Wrong conclusions concerning:
What is the hierarchy of evidence, from least from to confounding and bias to most?
1. Systematic reviews and meta-analyses
2. Experimental designs; RCTs; pseudo-RCTs
3. Quasi-experimental designs
4. Observational-analytic designs: cohort study, case-controlled study
5. Observational-descriptive designs: Cross-sectional studies, case series, case study
6. Background information/expert opinion
What are the 9 criteria for causality?
1. Strength (effect size)
2. Consistency (reproducibility)
5. Biological gradient
Describe criteria (strength) for inferring causality:
A causal link is more likely with STRONG associations (RR or OR)
However, small association does not mean that there is not causal effect
Describe criteria (consistency) for inferring causality:
A causal link is more likely if the association is observed in DIFFERENT STUDIES and different sub-groups
Describe criteria (specificity) for inferring causality:
A causal link is more liekly when a disease is associated with ONE SPECIFIC FACTOR
The more specific an association between a factor and an effect is, the bigger the probability of a causal relationship
Describe criteria (temporality) for inferring causality:
A causal link is more likely if EXPOSURE to the putative vause has been shown to PRECEDE THE OUTCOME
(i.e RCT, prospective cohort)
Describe criteria (biological gradient) for inferring causality:
A causal link is more likely if DIFFERENT LEVELS OF EXPOSURE to the putative factor lead to DIFFERENT RISK of acquiring the
- Greater exposure should generally lead to greater
incidence of the effect.
- In some cases, the mere presence of the factor can
trigger the effect.
- In other cases, an inverse proportion is observed:
greater exposure leads to lower incidence
Describe criteria (plausability) for inferring causality:
A causal link is more likely if a BIOLOGICALLY PLAUSIBLE MECHANISM is likely or demonstrated
However, knowledge of the mechanism is limited by current knowledge
Describe criteria (coherence) for inferring causality:
A causal link is more likely if the observed association CONFORMS WITH CURRENT KNOWLEDGE
- epidemiological and laboratory findings
- But, lack of laboratory evidence cannot
invalidate an epidemiological association
Describe criteria (experiment) for inferring causality:
A causal link is very likely if REMOVAL or prevention of the putative factor leads to a REDUCED OR NON-EXISTENT RISK of acquiring the outcome
- Experimental evidence