RCT class #12 Flashcards
factorial design
one experiement compare 2 or more inteventions to controls by cross classifying the inteventoins. ie
a = intervention 1 + intevention 2,
c = intervention 1 + controls,
d = intervention 2 + control
Allows for interaction evaluation.
prob with factorial desing
needs a large sample size
group allocation design
commonly done, randomizes groups rather than indviduals. (ie teens in schools are gonna talk about and share shit toghter).
2 types of non-randomzied controls
concurrent and historical
why would you ever choose concurrent controls over randomized controls
- pts unwilling to be randomized
- active therpay is so entrecnhed many pts unwilling to go for an alternative.
- rare cases with limited numbe rof patients
Historical controls
non-randomized not concurrent
pros of historical controls
all current patients can receive the new intervention
cons : (3 ) of historical controls
- improvement in internvetion group may be from improved overall treatment.
- historical groups may vary based on selection’
- bias still possible (enthusias for HOT therapy ie lithium in ALS
rate limiting step of clinical trials
Database mgmt systemts
To make this work, what do you need?
john thompson, budget for the upgrades in software and time for training.
Cross over designs how are they done
most simily two cross over periods.
period 1 each participant receives A or B. Then wash out if needed then each received the alternative (order in which A and B is received is randomized
how many more patients are needed for a parallel group design than cross over
2.4x
problems with cross over
- cant use once only outcomes such as mortality
- requires strong assumption of no carry over effect
..ie educational
What are withdrawal studies
randomized to take off a med. (what are the effects of longer term use).
What does JT think about withdrawal studies
They are under done.
Why? Bc the drug co dont want to limit their use.
