Final Flashcards
(24 cards)
What does p value tell you?
How probable a result was due to luck
What is a .15 p value?
what is considered statistically significant?
.15= 15% chance result was due to luck
.05 is considered statistically significant
> 5%= results were negative
Where do you look in a research paper for MCID?
Results or Discussion
What is the Suggest MCID for:
OPS= PSFS= Oswestry= Roland Morris= AROM=
OPS= 2-3 PSFS= 3 for 1, 2 for 3 Oswestry= 4-6 Roland Morris= 2-5 AROM= 20% improvement
Whats considered a _ benefit for MCID calculation in Systematic Reviews?
Small
Medium
Large
Small= 0.2
Medium=0.5
Large=0.8
What is a perfect NNT?
1
What is considered a successful NNT?
single digits
How is agreement between docs measured?
Kappa
What is considered to be a statistically significant Confidence interval?
One that does not cross the line of no effect.
A study that does not compair dropouts is called a _
As Per Protocol
What does Intent to treat decrease the chances of
Over-estimation of treatment results
Is it okay to not have phase “intent to treat” ever?
Yes, if they’re no dropouts or crossovers.
What 3 pieces of information can you gain from a forest plot?
1) Statistical Significance
2) Clinical significance
3) Precision
What are the big 3 questions to ask about diagnostic tests?
1) How good is the TEST/sign/symptom?
2) How STRONG is the evidence behind the claim.
3) How APPLICABLE is it to my patients?
3 questions to ask for Therapy?
1) How likely/ how much is it going to help?
2) How STRONG is the evidence
3) How APPLICABLE is it to my patients?
Quality of an RCT (ABCDFIX)
Allocation Blinding Comparability Drop Outs (5-20%) Follow Up Intention to treat analysis X-factor (biases)
Calculation for Positive LR=
Sensitivity/1-specificity
Calculation for Negative LR=
1=sensitivity/specificity
2/5/10 rule=
+LR ratio of = increase in post test probability
2=15%
5=30%
10=45%
Rule for -LR=
A test with a specificity of 95% means that if tested on 100 people, _% would cross react with condition free subjects
5%
A positive test with high specificity is most helpful at ruling in a condition, but only if these 3 guidlines are followed
1) Specificity and Sensitivity dont add up to be 100
2) There is at least a 50% probability that the patient has the condition pre-test
3) Subjects that the test is based on are similar to your patients
(T/F) The actual predictive value of a test is always lower than the isolated specificity number?
TRUE
What is a perfect Kappa number?
Poor=
Perfect= 1
Poor=<0.25
