statistics 1 Flashcards
what is a randomised control trial?
when a study population is chosen and sampled and then the groups are randomly split in two one groupie given the treatment and another is given the control both groups are followed up and the results are compared
why are trials conducted?
- ascertain the safety of a new drug
- demonstrate tolerability
- monitor adverse effects
- demonstrate efficacy
- to see if it is superior to an equivalent treatment
- to demonstrate cost effectiveness
what is the benefit of randomising in a trial?
- allows random confounding factors to be held constant between 2 groups
- all potential confounding variables (known and unknown) will be distributed by chance between two groups
- it is the only way to infer causality between groups
what is a confounder?
a variable that affects both the dependent and independent variable
e.g. when investigating how maternal weight gain affects infant birthweight- the length of gestation also influences both of these factors, therefore it is a confounding factor
what is equipoise?
Clinical equipoise is the assumption that there is not one ‘better’ intervention present (for either the control or experimental group) during the design of a randomized controlled trial (RCT).
It is not ethically acceptable to randomize patients to an arm of a trial which is known to be inferior or potentially harmful.
what is internal vs external validity
internal validity relates to how well a study is conducted (its structure) and whether the study establishes trustworthy cause and effect
external validity relates to how applicable the findings are to the real world…. to what extent can these findings be generalised to other people, situations and times?
what is the definition of bias?
any departure of results from the truth
- a partiality that prevents objective consideration of the issue or situation
in RCT:
systematic error is introduced into sampling or testing by selecting or encouraging one outcome or answer over others
bias is independent of both sample size and statistical significance
what are all the different types of bias
- selection bias
- performance bias
- attrition bias
- observer/detection bias
what is selection bias?
- Study sample does not represent target population
- Can happen when participants are asked to volunteer for a study
* can also happen when participants are recruited from a particular area….e.g. patients being trailed for a new diabetes drug recruited from a hospital setting may be different from the diabetic population at large
what is observer/detection bias?
- can occur in 2 ways
1. not adequately capturing the outcome f interest - e.g. when measuring disability relating to schizophrenia, how do you account for death? it at all?
2. if there is a systematic difference in the way that information is being collected for the different groups being studied - e.g. using manual and electric blood pressure measurements
what is attrition bias?
Attrition bias is a systematic error caused by unequal loss of participants from a randomized controlled trial (RCT)
intention to treat analysis?
method for analyzing results in a prospective randomized study
- all participants who are randomized are included in the statistical analysis and analyzed according to the group they were originally assigned
regardless of what treatment (if any) they received. this helps to erradicate attrition bias
per-protocol analysis?
only analyse patients who finish the treatment according to the study protocol.
so it does not include people who have dropped out of the study but this can lead to bias
as When some participants in either or both the groups are excluded, the remaining participants in the two groups can no longer be considered as balanced
what is a type 1 error?
false positive- rejection of a true null hypothesis
what is a type 2 error?
false negative-non-rejection of a false null hypothesis
what is a P value?
This is the probability of the null hypothesis being true p = or < 0.05 is often taken as the threshold to reject the null hypothesis (5 in 100 / 1 in 20 chance of the result being due to chance)
what is the relationship between sample size for trials and error in estimation?
error in estimation decreases as sample size increases
what are the different categorical catagories?
- yes/ no questions 2. answers with several catagories that don’t have an order…e.g. sexuality-> straight, bisexual, gay, trans 3. an answer with several ordered levels such as economic status-> rich, okay, poor
what is a variable scale?
continuous scale such as height
what is parametric data?
it makes the assumption that the data is evenly distributed
what is the problem when the data is skewed?
- the mean is different from the median - the mean is greatly influenced by the skew - so the median is a better representation of the data than the mean, and you should give the inter quarter rang instead of the standard deviation
how do we descriptively analyse scale data?
mean with standard deviation or median with interquartile range
how do you descriptively analyse categorical data?
use percentages
when to use T- test or ANOVA?
For randomised control trials that are comparing a continuous variable between groups
e.g. if asking is one group gets better scores on a SCALE than another= does drug A reduce depression symptoms compared to drug B?
the population should be randomly sampled
the observations should be made independently
and variances in each group should be equal
variables should be normally distrbuted
if more than 2 groups are being compared use the ANOVA test
