Yonkadonk Flashcards
3 prong approach to EBM?
- Critical problem solving
- Medical informatics
- Critical appraisal of med. literature
What are foreground questions?
Ask for specific knowledge about managing Pts with a disorder
EBM prescription
- Formulate & ask a question
- Access the evidence
- Critically appraise the evidence
- Apply the evidence
- Assess the use of info. in practice
Where do we find info. for caring for Pts?
- Texts
- Pharmaceutical texts
- Journals
- Drug company info.
- Self made info
- Other people
What is the best source of information?
Systematic reviews
Advantages of EBM?
- Improves confidence w/ decision-making
- Assists communication w/ Pts/other providers
- Dec. time wading through literature
- Fosters focused & productive reading habits
- Dovetails w/ technology (PDAs, electronic databases)
Disadvantages of EBM?
- Requires commitment in time & effort
- Not everyone is skilled at database searches
- Not everyone can afford resources
- Not everyone is skilled in appraising the literature
- Better know & choose reliable filters
- Good evidence not always out there
- Risks misinterpretation
Kinds of clinical questions?
- Etiology
- Diagnosis
- Therapy
- Prognosis
Best type of study for etiology & diagnosis?
- Cohort
- Cross-sectional
- Case-control
Best type of study for therapy?
RCT
Best type of study for prognosis?
- Cohort
2. RCT
How do you ask a PICO question?
- Pt
- Intervention
- Comparison intervention
- Outcome
What does EBM serve to do?
- Standardize practice while maintaining Pt centered core
- Promote life-long learning
- Response to practice variability
- Provide granularity on complex questions & gray areas
What are the components of a causal relationship?
Hill’s guidelines
- Strength of assoc.
- Consistency
- Specificity
- Time
- Biological gradient
- Biological plausibility
- Coherence w/ other data
- Analogy
What are the 3 maxims of clinical decision making?
- Diseases commonly occur
- Uncommon manifestations of common diseases are more common that common manifestations of uncommon diseases
- No disease is rare to the Pt that has it
What are the styles of clinical reasoning?
- DDx model
- Hypotheticodeductive model
- Exhaustive model
- Algorithmic model
- Heuristic model
What is the best style of clinical reasoning?
Hypotheticodeductive model
Based on probability, comes w/ experience
What are internal influences of medical decision making?
- Assumptions of objective findings
- Jumping to conclusions
- Personal biases
- What ?s you ask
- Your own risk taking nature
What are external influences of medical decision making?
- Anatomical differences
- Diff. therapeutic responses
- Pt biases & barriers
- Co-worker biases
Protocol vs. Guideline
P - Must follow
G - recommended to follow
When should you refer Pts?
- Know what you know
- Know what you don’t know
- Understand your scope
- Understand your pt (ex. insurance status)
What are the sections of a research paper?
- Intro
- Review of Related Medical Literature
- Methodology
- Results
- Summary/discussion
What is basic research?
- Understand, explore & develop theory
- Describe & provide foundation
- Process of collecting/analyzing info to develop theory
More focused on developing theory
What is applied research?
- Apply & test theory
- Predict, compare & explain cause
- Results either support or don’t support thoery
- Action research
More focused on testing theory
Qualitative research
- Analyzing non-numerical data to answer questions
- Narrative data
- Less structured
- Fairly flexible
- Design can evolve during study
Quantitative research
- Analyzing numerical data to answer questions
- Highly structured
- Very specific
- Strict rules/principles studies must adhere to
What is a discrete variable?
Whole numbers
What is a continuous variable?
Any number
What is a nominal level of measurement?
Numbers only used to differentiate subjects
Can’t do math on them
What is an ordinal level of measurement?
Numbers are categories but they have an order
ex. 1st, 2nd, 3rd
What is an interval level of measurement?
Ordered set of values w/ no absolute zero
Differences btwn values is equal
ex. IQ scores
What is a ratio level of measurement?
Ordered set of values w/ an absolute zero
What is dichotomous data?
Either you have it or you don’t
ex. STD - yes or no
What is validity?
aka accuracy
Does study measure what it’s supposed to measure
What is reliability?
Consistency of results to each other
Can you eliminate random error?
NO ma’am
What is sampling error?
Sample may not be representative of population due to chance
What is sampling bias?
Sample not representative cuz you messed up boiii
What is a parameter?
A numerical value that describes a population
What is a statistic?
A numerical value that describes a sample
What is descriptive statistics?
- Central tendency - mean, median, mode
- Variation - range, STDev, variance
- Relative position - %ranks, standard scores
Where did research ethics come from?
Belmont Report
If p>alpha, what do you do?
Accept null hypothesis
No significant difference
p value shows strength of relationships - not cause
Type 1 error
Yelling fire when there isn’t fire
Reject null hypothesis when it’s true
Dec. by lowering the critical value
Type 2 error
Yelling fire when there IS a fire
Accept null hypothesis when it’s false
More serious error
Dec. by inc. sample size
If a CI for continuous data contains 0, what does that mean?
Not statistically significant
If a CI for ratios contain 1, what does that mean?
Not statistically significant
Why is an adequate sample size important?
- Dec. Type 2 error
- Anticipate compliance & dropout
- Stratify data
What type of study is best for rare exposures?
cohort
What are confounding variables?
Variables that obscures the effect of another variable
Form of bias
What is the purpose of multivariate analysis?
Look at multiple variables/factors/parameters at once & adjust for their effects
Best test for descriptive studies?
Cross-sectional
What is the best test for rare diseases?
Case control studies
What is an odds ratio?
The probability of one event occurring over another
OR = 1 no effect
OR > 1 inc. odds
OR < 1 dec. odds, possible protective effect
When to use case-controlled studies & weaknesses
- Rare diseases
- Explore multiple exposures
- Info. needed ASAP
- Exposure data hard to obtain
- Little known about disease
- Long latency period
- Underlying population is dynamic
Weaknesses:
- Bad for rare exposures
- Bias may be introduced b/c it’s retrospective
- Temporal relationships hard to determine
Grading criteria
- Risk/benefit
- Evidence quality
- Values & preferences
- Cost
Central tendency
- Mean
- Median
- Mode
When do you use paired t-test?
Compare 2 sets of observations in a single sample
Tests the hypothesis that the mean diff. btwn 2 msmts is 0
What does the Pearson Correlation Coefficient tell you?
The linear relationship btwn 2 pairs of variables for quantitative data
What does chi square tell you?
Test null hypthesis btwn 2 categorical vaiables
compare expected vs observed results
When is ANOVA used?
To compare more than 2 grous
How do you compare 2 means?
Tukey’s test
What is Mann-Whitney U test?
Compare 2 independent samples from the same population when the data are not normally distributed
When is Kruskall-Wallace used?
When dealing w/ 1 nominal variable & 1 measureable variable & the data are not normally distributed
Non-parametric analog to ANOCA
Risk ratios
= 1 no diff. in risk btwn groups
> 1 inc. risks
< 1 dec. risk
Advantages of cohort studies
- Only direct means to establish an absolute risk
- Unbiased measure of exposure
- Can assess relationship btwn single exposure & many diseases
Disadvantages of cohort studies
- Impractical for rare diseases where thousands required to be enrolled to get a few cases
- $$$ & time
- Can only assess 1 or few exposures at a time
When do you use cohort studies?
- Prognosis
- Etiology
- Prevention
What is sensitivity?
Proportion of people w/ disease who have a positive test
Few false -
What is specificity?
Proportion of people w/o a disease who have a negative test
Few false +
Likelihood ratios
+ How good a test is at ruling in disease
Larger the number, better the test
10 is excelente
1 is useless
- How good a test is at ruling out a disease
What tests can you calculate with prevalence of a disease?
Sensitivity, specifity, PPV
3 rules for good screening tests
- Sensitivity & specificity must be high
- Prevalence matters
- A low cost conf. test must be avail.
What does parallel testing inc?
Sensitivity
less false -
What does serial testing inc?
Specificity