Evidence-Based Medicine Flashcards
P value
The probability of obtaining the observed result by chance rather than as a result of accrue effect
Alpha level
The highest risk of making false positive error that the investigator is willing to accept
Bayes theorem
If result is positive, probability that patient has the disease
If result is negative, probability that patient doesn’t have disease
attributable risk percent in there exposed
Among those exposed to y, what % of total risk for disease x is attributable to y
Population attributable risk percent
Among the population, what % of x is caused by exposure to y
Population attributable risk
Among the general population, how much of the total risk of fatal disease x is caused by exposure to y
Risk difference
A measure of absolute risk, aka attributable risk…. The risk in there exposed group minus the risk in the unexposed group
Rate difference
The rate in the exposed group minus the rate in higher unexposed group
Relative risk = risk ratio
The ratio of the risk in the exposed group to the risk in the unexposed group. If rr < or = 1, no association or negative association
Mortality ratio
Occurrence mortality in intervention group relative to controls
Strengths and weaknesses of: rct
Large numbers of participants
Less bias
Gold standard for testing hypotheses
Unethical to test harmful exposures
Expensive
Strengths and weaknesses of: cohort study
Can measure multiple outcomes for any one exposure
Can demonstrate a direction on causality
Can measure incidence and prevalence
Prone to loss of follow up bias
Confounding (non randomised)
Costly and time consuming
Want kind of study measures one outcome and many risk factors?
Case control
Want kind of study is susceptible to late look bias
Cross sectional
Want kind of study is based on exposure
Cohort
Want kind of study is based on known cases
Case control
What study is really good for rare diseases
Case control
Ecological fallacy
Inferences drawn to individuals based on populations
Want kind of study is good for interesting/new/unusual cases
Case series
What studies can test a hypothesis
Rct ( gold standard) or case control or cohort or ecological
Which is stronger and why: prospective or retrospective cohort
Prospective, due to the ability to monitor and control data collection
Provide and example of stratified allocation
Even spread across different blocks
What is the goal of case control studies
To determine differences in risk factors in participants with a particular outcome and participants without the outcome
A t-test measures how ______ the _________ is between two ______
Big the difference is between two means
A t-test takes into account the:
Variability between score and distributions
A 95% confidence interval asserts that
If there experiment were replicated 100 times, 95 of those times would contain the population parameter
Unlike probability values, confidence intervals provide information about the _____ of an estimate
Precision :)
The standard error is
A measure of precision of a SAMPLE statistic
A standard deviation is
A measure of precision of the POPULATION DISTRUBTION
Which hypothesis would you reject if t< critical value?
Alternative
Which hypothesis would you accept if t< critical value?
Null hypothesis
What is the critical level?
A number found in a table that you compare the t-test too to determine significance.
You want t to be _____ critical value for significant findings
Greater than
In a false positive error, you incorrectly accept the _____ hypothesis
Alternative
When the data represents pre-trial and post-trial results for a single group of subjects, use a _________ t test
Paired
What does a negative t score indicate?
That pretrial values were greater than post trial values…ie there has been a reduction
Risk difference
The proportion of the risk in the exposed that is due to or attributable to the exposure
A harmful exposure will yield a _______ greater than 1
Risk difference
Odds ratios are useful in ______ study designs
Case control
Bayes is used in medicine in:
Analysis of clinical decisions in Community screening program Individual patient care By Calculating ppv and posterior probabilities
In bayes theorem the numerator represents
True positive results
When applying bayes theorem to the care of an individual patient, try prior probability is analogous to….
Prevalence
To apply bayes theorem to a screening program which 3 bits of information should be known
Prevalence, specificity and sensitivity
As an equation, Probability =
Number of favourable outcomes/
Number of possible equally likely outcomes
In a normal distribution ___% fall either side 1SD, and ___% 2SD
68 and 95
Alpha can/cannot be used as the basis for ejecting the nul hypothesis
Can
Inductive reasoning
Specific to general
Generalising
Deductive reasoning
General to specific
Deducing
Binomial probability distribution
Used to describe variables with two levels
What is the advantage of an observational experimental design?
Can be used where it is unethical to withhold a treatment
What does a correlational study design measure?
Association, not causation
Nominal and ordinal are__________.
We use _________ correlation for ORDINAL data.
Non parametric
Spearman’s
Nominal data
Categories that have no relationship to each other e.g. Blood type
Ordinal data
Categories that have a relationship e.g. Young middle-aged and old
Interval and ratio data are_____. We use _________ correlation
Parametric types of data.
Pearson’s
Interval data
Related but no absolute zero e.g. Iq
Ratio data
Parametric data with absolute zero e.g. Blood pressure
Chi square test is a measure of________ that allows us to calculate ________
Difference, not association
Statistical significance
Define correlation
A measure of association between variables
When do you use a student’s t-test?
When there are two separate groups of participants
What is the difference between t tests and z tests?
T tests measure differences in means
Z tests measure differences in proportions
Critical ratio
Uses t or z test to a calculate a ratio between a parameter and the SE of that parameter
The nul hypothesis for a chi squared test of a 2x2 contingency table is that:
The two variables are independent of each other (variation in one variable is not caused in part but variation in another variable)
What category does an interventional study design fall under?
Strength and weakness?
Experimental
S: reduce ill health in entire community, increase health literacy, behaviours and environment
W: limited randomisation and difficulty in follow up
What are the phases of drug development/ public health intervention?
Basic science (discovery/1) M ethod development (1/2) E fficacy (2/3) E ffectiveness (3/4) D issemination (4/5)
Strengths and weaknesses of case control
S: less expensive than cohort, good for rare, multiple exposures
W: selection, recall and observer bias, only a single outcome, temporal relationships difficult to decipher
MR < 1 indicates that there is ______ mortality in the intervention group
Decreased
If an attributable risk is <0, factor investigated is protective or harmful?.
Protective
Degrees of freedom
N-1 in students
N-2 in paired
How do 1 way and 2 way anova differ?
Number of independent variables being tested
What kind of data do you use anova on?
Dependent is continuous
Independent is categorical
5as
Ask Access Appraise Apply Audit
Picot
People
Intervention
Comparison control
Outcome
Why can’t we calculate incidence in a case control study ?
The study population has been selected on the basis of hype heir disease
Attributable risk
Ar = risk(ex) - risk(unex)
Relative risk equation
Risk(ex) / risk(unex)
OR
Ad/bc
Ar(%)
= [risk(ex) - risk(unex)] / risk(unex) (x100)
