Biostats 3- Analysis of Screening Tests

Question 1

What is the eqn for a sample size needed for comparing differences in means between 2 groups with an alpha = 0.05?

Answer 1

n = 16s^2/d^2

s= standard deviation
d= difference btwn the means

Question 2

True or False: you use the same eqn for difference among more than 2 means.

Answer 2

True

Question 3

What is the eqn needed for doing a correlation between 2 continuous variables?

Answer 3

n = 4 + 8/r

r= estimated correlation coefficient

Question 4

What does “ballparkness” give you?

Answer 4

the order of magnitude of people for a study (multiple of 10)

Question 5

How much should u multiply your calculated sample size by to reduce migration bias?

Answer 5

1.25

attrition

Question 6

Why is screening a prevetnative intervention?

Answer 6

cuz u do it on asymptomatic individuals that allows for early detection, therapy, and ↓ mortality from the disease.

Question 7

What are the 4 selection criteria for a disease that can be screened?

Answer 7

1. Disease should be an important public health problem
2. Early detection in asymptomatic individuals is possible and improves outcome
3. Screenign tests are acceptable/inexpensive/accurate
4. Treatment is available for disease

Question 8

For Bayes Theorem, what is the tool you use to describe screening test outcomes within a popualtion?

Answer 8

2x2 contigency table

Question 9

This is a proportion of true positives correctly identified by the test over those wwho do have the disease

(test + | pt present)

Answer 9

Sensitivity

Question 10

What is the SE eqn for sensitivity?

Answer 10

SE = a/(a + c)

Question 11

What is the opposite of sensitivity, where its the proportion of true negatives over those who DONT have the disease?

(test - | no disease)

Answer 11

Specificity

Question 12

What is the eqn for specificity?

Answer 12

Specificity = d/(b + d)

Question 13

True or False: the higher the sensitivity, the fewer real cases of diseases go undetected.

Answer 13

True

Question 14

True or False: True higher the specificity, the fewer cases of normal patients go undetected .

Answer 14

True

Question 15

This is the # diseased per 1,000.

Answer 15

Prevalence

Question 16

This is the probability that pts with a + test result were correctly diagnosed as having the disease.

Answer 16

Positive Predictive Value (PPV)

Question 17

What is the eqn for PPV?

Answer 17

PPV = a/(a+ b)

AKA”everyone who tested +”
use rows

Question 18

This is the proportion of pts with a negative test result that did in fact NOT have the disease.

Answer 18

Negative Predictive Value (NPV)

Question 19

What is the eqn for NPV?

Answer 19

d/(c + d)

notice we use d here cuz d is the true negative.

Question 20

Accoring to beyes theorem, waht happens to PPV and NPV as the prevalence of the disease increases?

Answer 20

PPV ↑

NPV ↓

Question 21

This is a measure of how likely it is that an event will occur.

Answer 21

Probability (P)

Question 22

In mutually exclusive probabilities, how do you do measure probabilities?

Answer 22

Additive rule

roll a 2 OR a 5…

Question 23

If 2 events are independent, how do u measure probabilities?

Answer 23

Multiplication rule

roll a 6 and THEN a 5…

Question 24

What is the % chance a chem-6 panel comes up normal for everything, if each variable has a P=0.95?

Answer 24

74%

Question 25

This is the proportion of pts who were diagnosed with disease who are alive 5 years later.

Answer 25

Survival rate

Question 26

This is the proportion of ppl who died within 1 year.

Answer 26

Mortality rate

Question 27

This is the type of bias which depends on the TIME of Dx, which can sway the survival rate.

Answer 27

Lead time bias

Question 28

This is the type of bias where screening detects less aggressive cancers with long preclinical phases but not likely to cause death. Can sway the survival rate.

Answer 28

Overdiagnosis bias

Question 29

This is the bias wher eindividuals who participate in screening trials are fundamentally different from those who do not.

Answer 29

Selection bias

Question 30

What is the goal of screening: reduce mortality rates or increase survival rates?

Answer 30

Reduce mortality rates