Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
30 Cards in this Set
- Front
- Back
cross over trial design (3)
|
• Specialized type of controlled trial
• Each patient receives each treatment • Compare treatments within the same patient rather than between patients |
|
how randomization is used in crossovers
|
If randomization used, then randomize order of treatment, rather than to which treatment
|
|
typical issues with RCT/crossover (not specific to crossover)( 4)
|
– Sampling, inclusion/exclusion criteria
– Blinding- as in, is everyone blinded? – Compliance – Outcome measurement- always look at this; is this an established measure/approach |
|
Schematic of 2 x 2 crossover: describe
|
Top line is period 1 and period 2
Group A: Treatment X--->then Y Group B: Treatment Y--->Then X |
|
Key Factors that must be addressed for Effectiveness of Crossover Design (5)
|
• Carryover and period effects on the outcome
• Treatment sequencing and patient assignment (is there carryover effect?) • Crossover rules and timing of measurements • Dropout, faulty data and other data problems • Statistical analysis and sample size |
|
Order Effects in Crossover Trial (2)
|
carryover effect
period effect |
|
carryover effect
|
therapeutic effect carries over from one treatment to the next
|
|
Things to avoid to minimize carryover effect (3)
|
-avoid drugs with long half life (inadequate washout)
-pharmacological effects persisting after drug is eliminated -conditioning or learned behavior that is part of the treatment? Maybe it's about some type of counseling/pt education-->can't wash this out as people don't forget what they learned (so not good design for that kind of thing) |
|
period effect
|
the disease may progress, regress, or fluctuate in severity during the study- like tumors; just don't use this for crossovers
|
|
Treatment Sequencing and Patient Assignment methods (4)
|
1) Fixed sequence for all patients: A-B
2) Random assignment to sequence: A-B or B-A assigned by randomization 3) Deterministically balanced: A-B then B-A repeating to make sure equal numbers 4) Uncontrolled or haphazard |
|
fixed sequence properties (2)
|
not ideal due to limited info on order effect- so you can't tell if there is an order effect i think, is what it's saying
–Simplest to carry out |
|
random assignment to a sequence- preference
|
Most preferred since will take advantage of benefits of randomization
|
|
When order effects are a risk , what is a way you can go about assigning sequences in a way that is non-universally accepted, but that might work?
|
may use ABB, AAB, ABA, BAB or even 4 periods (use a bunch of periods (have to take treatment more than once)- too much burden, not liked)
|
|
Timing of Measurements options (2)
blinding? |
time dependent cross over (e.g. after specific amt of time, 2 weeks, etc)- timing should be blinded when possible
disease state dependent crossover (e.g. after control of sx) |
|
Crossover Rules (2) regarding timing between measurements
|
measurements may be delayed to allow for washout (may take days/weeks)
return to baseline before next treatment is desirable- sometimes not achieved |
|
dropouts and crossover trials (2)
|
•Effects of dropouts on data analysis are magnified compared to parallel design
•Dropouts may be more common because of longer duration of trial |
|
how to avoid dropouts
|
•Try to limit burden on subjects, while getting useful data
|
|
sample size calculations in crossover trials vs other trials
|
Sample size calculations similar to other trials with advantage of lower intra-subject variability since repeated observations made in same patient tend to be more similar than those made in different patients
|
|
crossover sample size required vs. regular RCT
why? |
Even if no reduction in variability, the number of patients needed is less than parallel design because patients provide data for both groups
|
|
statistical analysis of a crossover - how is it done (2)
indicate the power of first step |
•first test for treatment-time interactions (order effects), this is between groups and has relatively low power
–E.g. Group A Trtmnt X vs. Group B Trtmnt X •Second, compare data from same patients over time: Trtmnt X vs. Trtmnt Y |
|
ways of comparing treatment X vs. T statistically (3)
|
paired t-test, multivariate regression, analysis of variance
|
|
Primary Advantages of Crossover Design (3)
|
May use fewer patients, yet maintain power as you get more data from each subject (response to more than one treatment)
less variability greater power with paired samples |
|
disadvantages of crossover design (3)
|
•Sensitive to patient drop out and missing data and may decrease power if have dropouts – fewer cases for statistical tests
•Trial longer for each patient –Can contribute to missing data Carryover effects or period effects may be difficult to assess – Need some type of washout period |
|
If carryover effect is likely, then is crossover design a good choice?
|
no
|
|
properties of appropriate Conditions and Treatments for Crossover (5)
|
•Condition must be relatively stable
–More confident that difference are due to different treatments no disease variability •Condition must not be “curable” •Response to treatment should be rapid –Will help limit subject burden •Treatment effect should not be long lasting –Will give confidence that washout period will eliminate carryover effects •Drop out rate should not be likely to be high (due to amplification of dropouts since you are missing twice the data per dropout) |
|
good diseases/types of studies for crossovers (4 disease states, 3 study types)
|
•Pharmacokinetic studies
•Drug interaction studies •Short term pharmacologic effects, e.g. acid production in the stomach •Chronic pain •Hypertension •Diabetes •Glaucoma (chronic conditions that can be stabilized) |
|
poor examples for crossovers (5)
|
•Peptic ulcer disease
•Depression •Seasonal allergies •Most infectious diseases •Most types of cancer |
|
double dummy definition
|
idea that one of the treatments needs 2 sprays and one doesn't- so they put a dummy second spray for drug that was one spray
|
|
Don't have to worry about power if...
|
findings were significant
|
|
go over migraine study questions
|
---
|