5 main steps of EBM
Define a clinical question, find evidence, appraise evidence, apply evidence to patient, communicate evidence to patient
Validity questions for therapy (6)
- Was the assignment of pxs to treatment randomized?
- Was follow-up sufficiently long and complete?
- Were pxs analyzed in groups to which they were randomized?
- Were pxs and clinicians both blind?
- Were groups treated equally?
- Were groups similar at start of the trial?
Intention to treat
Does it over / underestimate the effect of the experimental intervention?
- Intention to treat (ITT) - Pxs are analyzed according to the group to which they were randomized. Preserves the value of randomization, providing for an unbiased comparison of groups
- ITT often underestimates the effect of the experimental intervention.
- Not using ITT will often overestimate the effect of the experimental intervention.
Importance questions for therapy (2)
- What was the magnitude of the treatment effect?
* How precise is the estimate of the treatment effect? How representative is the sample compared to the population?
Relative risk reduction
Same formula as what?
- Relative risk reduction (RRR; aka Efficacy) = (CER – EER) / CER = 1 – RR
- Same formula for relative benefit increase (RBI)
Formula for number needed to treat
Number needed to treat (NTT) = 1 / ARR
What does a confidence interval tell you?
Tells the range w/in which the true effect of the intervention likely lies.
For which values do you not want the confidence interval to include zero? (3)
RRR, ARR, ABI
How is pre-test probability determined? (4)
Determine pre-test prob w/ clinical experience, epidemiology literature, consultation, or prevalence of disease
Validity questions for diagnosis (4)
- Was there an independent blind comparison to a reference (gold) standard?
- Was the test evaluated in an appropriate spectrum of patients?
- Was the reference standard applied regardless of the test result?
- Was the test validated in a second, independent group of patients?
Importance question for diagnosis
Can this test distinguish pas w/ and w/o the diagnosis?
Does sensitivity / specificity rule in or rule out?
Sensitivity rules out
Specificity rules in
Likelihood ratio definition
Positive formula
Negative formula
- Likelihood ratio (LR) = prob of a clinical features in px w/ disease / prob of same clinical feature in px w/o the disease
- Positive LR = sens / (1 – spec)
- Neg LR = (1 – sens) / spec
Conclusive positive / negative LR
Not helpful positive / negative LR
Conclusive positive is >10
Conclusive negative is less than 0.1
Not helpful positive is 1-2
Not helpful negative is 0.5-1
