Intro to Evidence Based Medicine Flashcards
EBM triad domains
- clinical judgement
- best relevant evidence
- pt’s preferences & values
types of empirical evidence
observations = data grounded in physical reality quantitative = chemical & biological processes that involve the use of stats
function of stats
- Summarize a set of data
- Measure the degree of association between 2 variables
- Control potential confounding variables
- Predict future health outcomes
- Distinguish in sample data between genuine trends and random variability
descriptive statistics
frequency & percentage for categorical variables
ex) pie charts or bar graphs of BMI category
continuous variables
mean
standard deviations
measure of effect size
quantifies effect of X on Y
effect can be:
-causal
-statistical relationship
purpose of clinical research
investigates the causes of disease, factors that prevent disease, and treatments that restore health
-empirical investication
categorical variables
BMI
box plot
graphical technique which is useful for determining if the data are skewed and for detecting outliers
whiskers in a box plot
drawn from each end of the box to a point
-indicates outliers by the IQR
interquartile range
distance between the 75th and 25th percentiles in a box plot
top whisker in box plot
1.5 IQR above the 75th percentile
bottom whisker in box plot
below the 25th percentile
mild outliers
more than 1.5 IQR
-represented as circles in a box plot
extreme outliers
more than 3.0 IQR
-represented as asterisks in a box plot
effect
refers to either a causal or a statistical relationship
Risk Reduction Measures
- The more a treatment reduces risk,
- The greater will be the ARR
- The greater will be the RRR
- The smaller will be the NNT
confounding variables
methods of statistical control
-referred to as Z
statistical control
quantify the degree of relationship between X and Y after taking into account a third variable (Z) that is related to both
regression analysis
popular method for statistically controlling confounding variables
prognostic statistics
Predict future health-related outcomes
-based on various characteristics of that patient
inferential statistics
Distinguish in sample data between genuine trends and random variability
-help researchers to decide whether what was found to be true of the sample
*
Is likely to be true of the population?
or
Was just due to random sampling variability (chance)?
test statistics
generate 95% confidence intervals and P values
types of inferential statistics (test statistics)
- Z (generated by the Z-test)
- t (generated by the t-test)
- F (generated by the F-test)
- Χ2 (generated by the chi-square test)
- Wald (generated by the Wald test)
* All are used to compute 95% CIs and P values.
Evidence-based medicine is the intersection of 3 sets of considerations. One of these is Clinical Judgment. What are the other two?
- Patient’s Preferences and Values
3. Best Relevant Evidence
Identify type of statistic: The probability of recurrence of breast cancer over the next 10 years for a postmenopausal woman who does not smoke.
Prognostic statistic
Identify type of statistic: The proportion in a sample of Americans who are hypertensive.
Descriptive statistic
Identify type of statistic: A 95% confidence interval for the proportion of Americans in the population who are hypertensive.
Inferential statistic
Identify type of statistic: The correlation between the body mass index (BMI) of anorexics and their preferred BMI.
Measure of effect size
Identify type of statistic: Using regression to take into account gender in a study of the relationship between forced expiratory volume and age in a sample of children.
Statistical control
the 3 conditions required to establish causality in a clinical study
- Covariation
- Correct time order
- Absence of plausible alternative explanations
Covariation
To show that one variable (X) is a cause of another (Y), researchers must first show that the 2 variables covary or are correlated
> If Y is a disease, and X is a suspected cause, then the presence of X must be shown to be associated with the presence OR the absence of Y
ex) lung CA and smoking
X (prevention) - no Y ( disease)
ex) flu vaccine and absence of flu
how clinical researchers document associations between cause & effect
in terms of likelihood, tendencies or what is true on average
ex) People who smoke should be more likely to contract lung cancer.
ex) Patients who receive acupuncture should on average experience less severe headaches.
likelihood
measured in terms of probability or odds
probability
number of times an outcome will occur divided by the number of times the outcome could occur
* varies from 0 to 1
Odds
The probability an outcome will occur divided by the probability the outcome will not occur
* vary from 0 to infinity
