Outcomes exam 2; Lu material

OR =

ad/bc

RR =

[a/(a+b)]/[c/(c+d)]

One way to handle confounding is to consider the associations.....

within strata of the confounders

What is stratified analysis

looks at effect each independent variable has on outcome separately to account for confounding

When RR are different but both in the same direction when doing a stratified analysis, what are your options

1. report both RR

2. report an average of the RR

When RR are different but both in different directions when doing a stratified analysis, what are your options

you must report them seperately

stratifying confounders helps provide insight into...

relationships of interest

If the stratum-specific effects are not equal then we say

there is an interaction between the stratifying variable and the exposure

often we will assume that the true stratum specific risk ratios are

equal

When we assume that the true stratum specific risk ratios are equal, we can say refer to it as

1. common stratum specific RR

2. effect of the exposure controlling for the confounder

3. adjusted effect of the exposure

If you say that the stratifying variable is an effect modifier of the relationship between the exposure and the outcome, then it means that

the stratum specific effects are not equal

What is the weighted average of the stratum-specific effects

mantel-haeszel RR or OR

Why is stratified analysis not usually used as the final analysis

we want to control for many confounders, and the strata would become too small to be informative

when do you use mantel-haenszel

1. data is categorical

2. want to measure OR or RR

3. need to control for one confounder by stratification

When you want to control for more than one continuous variable you need to use what

1. two way ANOVA

2. multiple variable regression

when two random variables are related so that when one tends to be high the other tends to be high they are considered

correlated

the correlation coefficient takes values between

-1 and 1

When the correlation coefficient is 0, then

the two variables are uncorrelated

When the correlation coefficient is +, that means

that when one variable is large, the other tends to be large

when the correlation coefficient is -, then

when one variable is large, the other tends to be small

r represents what

correlation coefficient

The estimate of the correlation is called

pearson's correlation coefficient

it is vital that you always inspect a simple graph of the data for a ________, before ________

linear relationship, proceeding to a statistical analysis

what is the Y intercept in simple linear regression

B0

Mean (Y) =

B0 + B1X

in simple linear regression, each unit change in X leads to a change in Mean (Y) by

B1 units

How do we estimate parameters using the simple linear regression model

we need a data set available which contains observations of the outcome and exposure level for sample of n independent subjects

ordinary least squares estimation

a method for finding the best-fitting line in simple linear regression by minimizing the sum of the squared differences between the observed values and the values predicted by the line.

What are the three purposes of multivariable regression models

1. developing models to predict the value of an outcome variable from multiple predictor variables

2. estimating the effect of one variable, controlling for other variables

3. look at the association between a quantitative predictor and an outcome

in the general multiple regression model, the predictors can be

categorical or continuous

in the general multiple regression model, it can be shown that the ____ are interpretable as the effect of the corresponding predictor, controlling for _______

Bs, controlling for all other variables in the model

it is important to realize multiple regression models make what two assumptions

1. the relationship between the E(Y) and the quantitative predictors in linear

2. the effects are simply additive

Logistic regression is used for

binary outcomes

which is more widely seen in medical research

logistic regression

in logistic regression, the a(n) ________ is generated

odds ratio

1 multiple choice option

Phase I characteristics

1. 20 to 100 healthy volunteers

2. lasts several months

3. 70% move on to next phase

Purpose of Phase I

safety and dosage

Phase II characteristics

1. up to several hundred people with the disease/condition

2. lasts several months to 2 years

3. 33% of drugs move to next phase

What is the purpose of phase II

efficacy and side effects

Phase III characterisitcs

1. 300-3,000 volunteers with the disease/condition

2. lasts 1-4 years

3. 25-30% move to next phase

What is the purpose of phase III

efficacy and monitoring of adverse reactions

which phase trials are considered pivotal studies

phase III

which phase includes several thousand volunteers who have the disease/condition

phase IV

________ are carried out once the drug or device has been approved by the FDA during the post-market safety monitoring

phase IV

what are the Hill's criteria to assess causation

1. strength of association

2. consistency

3. specificity

4. temporality

5. biologic gradient

6. plausibility

7. coherence

8. experimentation

9. analogy

strength of association

larger association = more likely the exposure is causing the disease

weak associations may be causal, but

it is harder to rule out bias and confounding

consistency

the association is observed repeatedly in difference persons, places, times, and circumstances

Specificity

a single exposure should cause a single disease

what is the weakest of all hills criterial

specificity

what is the most agreed upon hills criteria

temporality

temporality

causal factor must precede the disease in time

biological gradient

dose-response relationship between exposure and disease where higher exposure = increasingly higher risks of disease

plausibility

1. biological or social model exists to explain the association

2. association does not conflict with current knowledge of natural history and biology of disease

coherence

observed association is consistent with the natural course of teh disease or outcome

experimentation

intervention that modifies the exposure through prevention, treatment, or removal should result in less disease

analogy

has a similar relationship been observed with another exposure and/or disease