critical numbers Flashcards
what is prevalence probability?
probability of having a disease at a given point in time
what is incidence probability?
probability of getting a disease during a specified poit in time
what is the length-time bias?
conditions with a longer duratoon are more likely to be captured in prevalence
if risk of X given: drug A = 0.6 and drug B = 0.2 what is the risk difference?
0.6-0.2 = 0.4 (40%)
if risk of X given: drug A = 0.6 and drug B = 0.2 what is the risk ratio?
=3
3 times the risk
if risk of X given: drug A = 0.6 and drug B = 0.2 what is the relative risk difference?
200%
((a-b)/b) *100 = 200%
relative risk ____ the risk
relative risk exaggerates the risk, they do not indicate a baseline
what is number needed to treat/harm?
indicates the potential benefit of a clinical intervention
what type of study uses routinely collected data to show trends and generate hypotheses?
ecological
what type of study divides a population into those with the disease and those without and collects data at a single point in time?
cross-sectional study
what type of study matches people with a disease to those without for age, sex, habits, class etc.?
case-control study -retrospective
what type of study studies a population to see if they’re exposed to the agent in question and if they develop the disease?
cohort study - prospective
give 3 advantages of an ecological study?
1) uses routine date and so is quick and cheap
2) few ethical issues
3) useful for generating hypotheses
give 3 disadvantages of an ecological study
1) cannot show causation
2) inconsistency in data presentation
3) bias - variation in diagnostic criteria
give 4 advantages of a cross-sectional study
1) can give rapid insight into events within a population
2) few ethical issues
3) good for generating hypotheses
4) quick and cheap
give 3 disadvantages of a cross-sectional study?
1) prone to bias
2) no time bias
3) could be reporting medical oddities
give 3 advantages of a case-control study
1) quick - results can be obtained quickly (retrospective)
2) cheap
3) usually a small number of people required to produce statistically significant results
give 3 disadvantages of a case-control study?
1) retrospective date may be unreliable - selective memory
2) shows association but not causation
3) prone to selection and information bias
4) cannot calculate incidence
give 3 advantages of a cohort study?
1) can calculate incidence and so find relative and abs risk
2) reduced chance of bias - exposure measured before disease develops
3) can distinguish causes from associated factors
give 3 disadvantages of a cohort study?
1) expensive - long time and large population
2) causation cannot be calculated - control study is needed for this
3) often difficulties with follow-up
give 3 advantages of a RCT?
1) confounders are equally balanced
2) blinding minimises bias
3) statistical tests are easier when confounders are minimised
give 4 disadvantages of a RCT
1) expensive - large populations
2) poor compliance can mean statistical tests lose their power
3) volunteer bias
4) ethical difficulties in withholding treatment from control groups
what does presence of bias in a set of results imply?
the results of the study are not accurate
will the standard error in a set of results be big or small if the precision is good?
there will be small standard error
what are confounding factors?
factors that can affect the validity of a study
they may be responsible for the results seen
what should a good study have?
randomisation of participants to interventions (reduces bias/confounding)
show causation rather than association
have outcome measures (results) for atleast 80% of the population
define systematic review
systematic methods are used to identify, select and analyse relevant research in order to answer a specific question
give 2 examples of non-random sampling?
convenience sampling
purposive sampling
what is number needed to treat?
the number of patients that need to be treated in order to have an impact on one person
is a test significant if the 95% confidence interval contains zero?
no it is not significant
how can you calculate the NNT?
1/risk difference
what is nominal data?
a qualitative classification of data - a naming system
what is ordinal data?
when data has numerical scores existing in order
what is binary data?
data that can only take one of 2 possible states e.g. daed or alive
what is discrete data?
finite number of values e.g. number of people living in a house
what is continuous data?
data that can take any value e.g. height
what is linear regression used for?
estimating mean differences between groups
what is logistic regression used for?
binary outcomes, it models and odds ratio
what is poisson regression used for?
it models rate ratios
what is cox regression used for?
models hazard ratios
define morbidity
suffering from a disease
if the prevalence of a disease is low, what affect does this have on false negatives ?
number of false negatives will decrease
NPV will increase and PPV will decrease
what is association?
a statistical link between exposure and disease
what is causation?
a statistical link where a disease is directly caused by the exposure
what percentage of all data in a sample will fall into one SD limit?
68%
what percentage of all data will fall into two SD limits?
95%
what does a low standard error mean?
low variability from sample to sample
better precision in estimating mean
low in a large sample size
estimates the variability
what does a 95% confidence interval mean?
the range of values that you can be 95% confident contains the true mean of the population
what is a null hypothesis?
a statement to be tested of no difference or association
typically aim to prove this hypothesis is not true
statistical significance _____ mean clinical significance
statistical significance doesn’t mean clinical significance
what does crude mean in statistics?
where confounding variables aren’t accounted for
what does adjusted mean?
where confounding variables are accounted for
how can chance be reduced?
ensuring a sufficiently large sample
using confidence intervals and p-values
how can confounders be prevented at design stage?
randomisation
restriction
matching - where control group matches the confounders in the case group
how can bias be prevented?
ensuring an appropriate selection of participants
ensuring data is collected and measured correctly
what are the 2 types of bias?
selection bias - problem with study population
information bias - problem with information provided
how can information bias occur?
researcher - observer bias
participant - recall/respondent bias
instruments - wrongly calibrated instruments
