Critical Numbers 1 Flashcards
risk ratio =
A/B
risk difference =
A - B
NUMBER NEEDED TO TREAT
1/|RD|
NUMBER NEEDED TO HARM
1/RD if RD > .0
RELATIVE RISK DIFFERENCE
A - B/B x100%
Pyramid of evidence
- Systematic review/ Meta-analysis of RCTs
- RCTs
- Controlled cohort study
- Case control study
- Case series
- Case study
- Anecdote
- Wishful thinking/blind hope/delusional thinking
Confounding:
A factor that independently influeneces the outcome of a situation but doesn’t lie on the causal pathway
2 approaches to stats:
Confidence Intervals
PI values
If a CI goes over 0…
it is not statistically significant
What is a confidence interval?
A range of values, so defined hat the true value probably lies within
P values are always between…
0 and 1
When your p value is small:
It is very unlikely that your results are down to chance.
Reject Ho
When your p value is large:
It is likely that your results are down to chance
Accept/do not reject Ho
P values are statistically significant if they are…
less than 0.05
Cross-sectional studies:
Non-experimental
Observational
Individual based
Analytic
Case-control study:
Non-experimental
Observational
Individual based
Analytic
Cohort-study:
Non-experimental
Observational
Individual based
Analytic
Ecological study:
Non-experimental
Observational
Population based
Analytic
Descriptive study (health survey):
Non experimental
Observational
Population based
Descriptive
Case reports/series:
Non experimental
Observational
Individual based
Descriptive
RCT:
Experimental
Interventional study
Randomized
Quasi-experimental/ Field trial/ Community trial:
Experimental
Interventional study
Non-randomized
RCT strengths:
Minimise bias and confounding
Can look at multiple outcomes
Strong evidence of causal relationship
RCT weaknesses:
Huge cost Huge teams Dropouts Ethical concerns Complex to manage
What does a case-control study do?
Looks at individual cases, past exposure is measured, prevalence of past exposure among cases and controls are compared.
What does a cross-sectional study do?
Measures outcomes and exposure variables simultaneously in a given population at a given time
What does an observational study do?
Record outcomes without intervention
What does PICO stand for?
Patient/population
Intervention
Comparison/Control
Outcome
What does PPI stand for?
Patient Public Involvement
What does a descriptive study do?
Describes the occurrence of a disease, generates a hypothesis based on aetiology
What is cluster sampling?
A population is partitioned into groups or clusters and a sample of clusters is selected by simple random sampling
What is stratified random sampling?
A population is partitioned into groups and a sample is selected by simple random sampling in each group.
What is systemic random sampling?
Members of the population are selected at equal intervals, first member is selected at random
What is standard deviation?
A degree of variability around the mean value
What is the clinical use of standard deviation?
The mean plus or minus two s.d. is considered the normal reference range.
E.g. normal Hb levels
When is standard error low?
When sample size is large and/or variability in the data is low
Equation for standard error:
SE = SD/ square root n
The width of a CI depends on:
SE
level of confidence
The wider a CI..
the less precise