DI Flashcards
What is Validity
Degree measurement represents true value
What is Reliability
Reproducibility of measurement
Subjective Outcomes
Likert Scales
Self-reported symptoms
Objective Outcomes –
Continuous Data
- Lab values
- BP
- Weight
Objective Outcomes –
Nominal Data
- Death (yes/no)
- ED admit count
- MI
what is Minimal Clinically Important Difference (MCID)
Amount of improvement that is important to patients
Methods to determine MCID
- Expert consensus
- Anchor to independent categorical assessment scores
- Distribution-based method relies on the statistical properties of the distribution of outcome
Outcomes Important to Patients
Pain scales, depression scores
Delayed or reduced rates of death
Reduced hospitalizations or disease incidence
not lab valves
What is a surrogate endpoint
”A laboratory measurement or physical sign that is used in therapeutic trials as a substitute for a clinically meaningful end point that is a direct measure of how a patient feels, functions, or survives, and that is expected to predict the effect
of therapy.
common Surrogate
Osteoporosis • BMD Cardiovascular • LDL cholesterol • LV Ejection fraction • Blood pressure • Premature ventricular contractions (PVCs)
Clinically Meaningful Endpoints
Osteoporosis •Fractures Cardiovascular • CV death, MI, stroke • HF survival • CV death, MI, stroke • Fatal arrhythmia, sudden death
Surrogate Endpoints Disadvantages
Trial too short to evaluate long-term side-effects • Patients don’t relate their clinical experience to surrogate • Intervention may affect surrogate endpoint but not clinical outcome
Surrogate Endpoints Advantages
• Smaller sample size;
expose fewer patients
• Shorter trials / faster to market
• Easier to measure • Less invasive for patients
Surrogates are justified for
- Rare diseases (familial hypercholesterolemia)
* Slow developing but fatal diseases with no effective therapy (AML)
Composite Outcome Definition
• Combination of two or more related “component” outcomes
• Patients are
counted once if any of the
“component” outcomes occurs
Composite Outcome Disadvantages
• Can be misleading if the
treatment effect varies across the components and components have unequal importance
• Can be difficult to interpret
Composite Outcome Advantages
• Improves statistical power (i.e. smaller samples) • Allows use of multiple, similar outcomes (don’t have to choose)
validity of composite outcomes
depends on similarity
in patient importance, treatment effect and incidence across the components.
P-values
• Indicates precision in the abstract (i.e. rejecting the null). • Does NOT provide good indication of variability. • Does NOT provide any indication of clinical significance. • Less desirable
Confidence Intervals (CI)
• Indicates precision in the abstract (i.e. rejecting the null). • Indicates variability (i.e. wideness of interval) • Provides information about clinical significance • More desirable
Alpha (Type I) error
Inconsistent with other findings and wide variation
OR
• Inconsistent with other findings and close to null
Type 2 (beta) error
Narrow CIs (little variation) but “touches” the null
AND Effect size within MCID
• Look for consistency with other findings
Superiority Trial
New drug (e.g. Prilosec™) better than old drug (H2A) or placebo
Non-inferiority Trial
New drug (e.g. Nexium™) is no worse than old drug (e.g. Prilosec™) • No worse than a preset margin
Equivalency Trial
New drug (e.g. generic omeprazole)
“equivalent to old drug (e.g. Prilosec™)
• No worse AND no better than a preset
margin
delta (D)
clinically acceptable difference
D usually established from
pooled placebo controlled trials
• Treatments with CIs within the D are said to be “non-inferior” or “equivalent”
