Term
| participants chosen how in cross-sectional? |
|
Definition
| without regard to disease or exposure status |
|
|
Term
| how are participants chosen for case control? |
|
Definition
| chosen based on their disease status |
|
|
Term
| how are participants chosen for cohort? |
|
Definition
| chosen based on their exopsure status |
|
|
Term
| what do we do if we can't find expsoed individuals for or study? |
|
Definition
intervene experimentally.
- create two or more groups
- follow groups over time and count cases of disease |
|
|
Term
| give discription of laboratory experiment (3) |
|
Definition
-individuals randomly assigned to groups (Exposed, not exposed)
- introduce challenge (pathogen)
- control enviroment (ie temp, moisture , ect). any confounding factors) |
|
|
Term
| RBCTs aka Field trials. briefly describe (3) |
|
Definition
- individuals randomly assigned to groups (exposure pos or neg)
- no control of the challenge
- no control of environment |
|
|
Term
| randomized blind control trials, (experimental studies in the "real world") exert how much control over environment and challenge? |
|
Definition
|
|
Term
| randomized (blind) control trials are useful for testing which two things? |
|
Definition
| preventative studies (vaccines, feed additives) and some therapeutic studies (anti-cancer drugs) |
|
|
Term
| do randomized (blind) control trials reuire identical groups? if so, how? |
|
Definition
|
|
Term
| list the 7 stpes in an RBCT. |
|
Definition
1. hypothesis
2. determine the reference population
3. RANDOMLY select participants
4. groups RANDOMLY assigned
5. follow up all with equal scrutiny (blinding)
6. statistical analysis
7. interpretation and conclusions |
|
|
Term
| the 9 important questions for RBCTs should be asked when? |
|
Definition
| before the study when they're much easier to fix |
|
|
Term
| For RBCTs, how do you answer if you have a clearly defined question? |
|
Definition
ask if treatment works
- is it better than NOTHING (control group)
[test through one expsure group getting treatment, other get nothing]
- is new treatment better than whats CURRENTLY AVAILABLE
[test by one exposure group getting new treatment, other getting old treatment] |
|
|
Term
| For RBCTs, how do you know if its ethical to intervene? |
|
Definition
for treatment group:
- new treatment vs. safe avialble treatment?
- dose given? (dose response studies - think about the high and low ends of dose spectrum)
for control group:
- withholding treatment?
- best available, when new (better) treatment available? |
|
|
Term
| For RBCTs, how do you determine the reference population? |
|
Definition
- what population do you want to make comments about? (ie cattle in canada)
- where are you going to sample from? (ie cattle in ontario)
- is your study group representative of population? (ie are ranomly selected from ontario rep of all of canada?) |
|
|
Term
| For RBCTs, how do you determine which groups are needed? |
|
Definition
if asking if new treatment works
- need NEGATIVE control group (placebo)
if asking if new treatment is better than standard
- use POSTIVE control group (best current Rx)
** be careful when using historical controls when looking at populatins groups as things change over time |
|
|
Term
| For RBCTs, how are the groups created? |
|
Definition
randomization.
- reduces possiblity that unidentified expsoures (confounders) will bias results of study
done through:
- fliping coin for every participants
- random number generator (list) |
|
|
Term
| how do you know if a randomization process worked? |
|
Definition
-individuals in treatment and control groups should be alike in evey other way
- everyone has EQUAL chance of being in either group (no matter disease severity!) |
|
|
Term
| For RBCTs, name the 5 types of study designs. in all of these cases, we want to ensure what? |
|
Definition
- completely radnomized (indivuals randomized to treatments)
- randomized block (groups randomized to treatments)
- cross-over (individuals are their own controls [both placebo and exposure])
- factorial (2 or more treatments)
- split-plot (treatments at different levels)
want to ensure equal numbes of participants in groups with respect to time and location. |
|
|
Term
|
Definition
| the length of time to ensure that drug has completely drained out of system before we introduce placebo or new drug. the amount of time depends on time that treatment had effect for. |
|
|
Term
| For RBCTs, how do you determine outcomes of interest? |
|
Definition
think about:
- ease of measurement (blood test, questionarie)
- biological importance (pos serology test might not necessarily = immune to pathogen, but mortally/morbidiy are meaningful)
- economical importance (mostly in animal health, ie milk production, # of offspring)
- quantativve vs. qaulatiave measures |
|
|
Term
| For RBCTs, how do you determine how big a study should be, ie how many per treatment group is enough? |
|
Definition
- want to ensure we have enough participants to show statisical difference if one exisits (POWER)
-SAME factors to consider as in observational studies, and same calculations.
- number required increases with:
- differnece in treatment effects expected
- increasing number of covariates (any other variable expect expsore and outcome)
- increasing number of groups (split plot design = lots of participants) |
|
|
Term
| For RBCTs, how do you avoid bias? |
|
Definition
selection bias
- choose partiicpants carefully (ie only participants at certain stage of disease)
- loss to follow up may be an issue
confounding bias
- measure potential confounders. stratify groups by these potential confounders, control at analysis stage, or exclude groups to reduce potential confoudning.
information bias
- misclassification of outcome due to WISH BIAS
- has potential to create differential misclassification issue in the OUTCOME meausremet, so use BLINDING to reduce probabily of wish bias leading to misclassification. |
|
|
Term
| For RBCTs, what statisciatl anaylsis should be used? |
|
Definition
chi square: proportions
t-test: means
if p <0.05 difference sign
only extrapolate to target population. |
|
|
Term
| where do randomized controlled studies and randomized contolled doubld blind fit on triangle of evidence? |
|
Definition
| RCDB very top, RCS second highest. |
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
| randomized block desgin aka stratified randomization |
|
Definition
|
|
