CVB MDM

Question 1

What is a derivation study, as it relates to a clinical prediction rule?

Answer 1

Usually a retrospective study

Used to develop a clinical prediction rule. Must be validated by another study

Question 2

What is specificity?

How do you calculate specificity?

Answer 2

Given the absence of disease, the probability that a test will be negative

Specificity = TN / (FP+TN)

Question 3

What is sensitivity?

How do you calculate sensitivity?

Answer 3

Given the presence of disease, the probability that a test will be positive

Sensitivity = TP / (TP+FN)

Question 4

What is length time bias?

How can it be eliminated?

Answer 4

Length time bias occurs when screening measures uncover less aggressive/less severe forms of disease, such as tumors with a better prognosis. The misconception that arises is that catching the tumor early leads to better prognosis, when in reality these tumors had a better prognosis to begin with

This occurs because more severe tumors are likely to progress and present as disease before they have a chance to be caught with screening

To eliminate: Study the screening test using an RCT with intention to treat analysis

Question 5

How can you convert probability into odds?

Answer 5

Odds = probability / (1 – probability)

Question 6

What is a training set, as it referes to clinical prediction rules?

Answer 6

The data used to develop a prediction rule in a derivation study

(usually retrospective)

Question 7

What is the formula for LR+?

Answer 7

LR+ = sensitivity / (1 - specificity)

Question 8

What is the best negative LR?

Answer 8

0

If a test is perfect, a negative result = 0% likelihood that you have disease

Question 9

What is lead time bias?

How can it be eliminated?

Answer 9

Lead time bias occurs when the time between disease diagnosis and death is increased only because screening caught the disease early, and not because screening helped to initiate treatment that prolonged life.

In other words, when screening offers no benefit in terms of prognosis/treatment: it only tells people they have the disease earlier.

To eliminate: compare age specific mortality instead of survival

Question 10

What is secondary prevention?

Give some examples

Answer 10

Secondary prevention detects and treats disease early in the disease process, while it is still asymptomatic and “treatable”

Involves screening and follow-up diagnosis + treatment

• HIV testing
• PAP smears

Question 11

What is the relationship between negative post-test probability and negative predictive value?

Answer 11

Negative post-test probability = 1 – negative predictive value

Negative predictive value = 1 - Negative post-test probability

(use bottom row)

Question 12

Is sensitivity used to rule a disease in or out?

Answer 12

SNOUT - sensitivity, rules a disease out

Question 13

What kind of prevention does surveillance fall under?

Answer 13

Tertiatry prevention

Question 14

What is chemoprevention?

Give some examples

Answer 14

The use of drugs to prevent disease

Examples:

• Folate during pregnancy to prevent neural tube defect
• Antibiotic prophylaxis
• Aspirin to prevent clotting
• Statins

Question 15

What is the relationship between positive predictive value and positive post-test probability?

Answer 15

Positive predictive value = positive post-test probability = TP / (TP+FP)

(use top row)

Question 16

What is tertiatry prevention?

Give some examples

Answer 16

Tertiary prevention prevents the deterioration/progression of the disease. The goal is to reduce complications of the disease, given that it is already present

Basically, long term treatment. Surveillane is necessary

• Beta blocers to reduce death in patients with past myocardial infarction
• Insulin/blood glucose control in diabetics

Question 17

To determine whether a screening test is effective in altering disease progression, why would you analyze age-specific mortality (as opposed to survival rates?)

Answer 17

Analyzing age-specific mortality eliminates lead-time bias

It can also help to control for length-time bias

Question 18

What is primary prevention?

Give some examples

Answer 18

Primary prevention prevents disease from occuring by removing the cause of disease

May be carried out in community or clinical settings

• Immunizations
• Behavioral counseling
• Prophylactic surgery
• Seatbelt campaigns

Question 19

What is compliance bias?

How can it be eliminated?

Answer 19

Compliance bias occurs when patients who are more compliant with treatment have a beter prognosis for reasons other than their compliance with the treatment

For example, patients who comply with treatment are more likley have better baseline health or more investment in healthy lifestyle in general

Eliminate by studying the treatmeng with a RCT with intention to treat analysis

Question 20

What is the formula for negative post-test probability?

Answer 20

Negative post-test probability = FN / (TN+FN)

Question 21

What is the formula for LR-?

Answer 21

LR- = (1 - sensitivity) / specificity

Question 22

What is a completely unhelpful LR?

Answer 22

1

With a + or - result, you are equally likely to have the disease

Question 23

What must be true of a disease for screening to be beneficial?

Answer 23

The disease that is screened for should have…

• A significant burden of suffering
• A latent period where the disease is present but not yet clinically significant
• An intervention that would improve outcomes if the disease is found early
• A good screening tool

Question 24

What is the relationship between pre-test odds, LR, and post-test odds?

Answer 24

Pre-test odds x LR = Post-test Odds

Question 25

Is specificity used to rule a disease in or out?

Answer 25

SPIN - specificity, rules a disease in

Question 26

What is the formula for positive post-test probability?

Answer 26

Positive post-test probability = TP / (TP+FP)

OR

Pre-test odds x LR = Post-test odds, and then convert post-test odds to probability

Question 27

What is the best positive LR?

Answer 27

Infinity

In a perfect test, a positive result means that you always have the disease

If a test has a high LR+, like 100, then a positive result means you are 100x more likely to have the disease

Question 28

Describe a validation study, as it relates to a clinical prediction rule

Answer 28

Usually prospective studies in a clinical setting with close follow-up and a broad-spectrum definition of disease

Must use a patient population different from the derivation study

Question 29

What is a test set, as it referes to clinical prediction rules?

Answer 29

The data used to validate a clinical prediction role in a validation study

Must use a population different than the derivation study

(Usually prospective studies in a clinical setting with close follow-up and a broad-spectrum definition of disease)

Question 30

How can you convert odds into probability?

Answer 30

Probability = odds / (1 + odds)

Question 31

How does sensitivity relate to the true positive rate of a clinical prediction rule?

Answer 31

Sensitivity = True positive rate

Question 32

How does specificity relate to the false positive rate of a clinical prediction rule?

Answer 32

Specificity = 1 - false positive rate