Outcome Measures Flashcards
5 things outcome measures need generally:
- Clinically relevant, 2. defined in protocol, 3. objective and simple to ascertain 4. used to calculate sample size 5. Clinicians and statisticians input
primary efficacy outcome measure needs to be 5
prespecified, easy to diagnose, free of errors, capable of being objective irrespective of treatment, time points specified
What is a surrogate outcome? Why use?
lab measurement or sign used as clinically meaningful outcome. decrease sample size and duration
Why not just use routine care notes for clinical trials?
Inconsistent, vague, subjective, need tighter control
What are two characteristics surrogate should have?
- effect of intervention on surrogate predicts effect on clinical outcome
- Intervention doesn’t effect clinical outcome unless through the surrogate path
4 categories of collection in a clinical trial:
- baseline assessment 2. principal criteria for patient response 3. subsidiary criteria 4. other aspects of monitoring
Clinical Effectiveness Outcome problems
Not useful if you want to compare changes in dif areas of healthcare
What to measure instead of clinical effectiveness outcomes?
Measuring utility QALY
QALY stands for
Quality adjusted life years - incorporates quantity and quality of life
QALY values
1 year in perfect health =1 dead =0
How to measure QoL
EQ-5D-5L questionnaire
What is NICE?
looks at costs and QALYs in its economic evaluation help decide what treatment is worth funding by NHS
How to calculate QALYs?
Multiply utility weight of health state by amount of time
Limitations of EQ-5D-5L
Doesn’t incoroporate mental health or social care spillovers
How to analyse continuous data:
use mean values and measure of variation, ANOVA, t test…
How to analyse binary data
Multivariable logistic regression for odds of occurence
time to events analysis 3
time between- probability/survival function
Kaplan Meier method
Hazard Definition
Probability of a person having been event free up to a certain time will have the event of interest within next infinitesimal space of time
When is hazard ratio used?
Estimate difference between 2 survival curves
HR Values 3
=1 – risk same as control
1– risk more
