PPHC 11: Evidence Evaluation – How do we identify disease in populations?

Question 1

What is a test?

Answer 1

lab test, x-ray, ultrasound, answer to a patient question, identification of a symptom

Question 2

What is a screening?

Answer 2

identifying the possible presence of an as-yet-undiagnosed disease in a person without signs/symptoms

Question 3

What is a diagnosis?

Answer 3

determining which disease or condition explains a person’s signs or symptoms

Question 4

What is sensitivity?

Answer 4

probability that an individual with the disease has a positive test result

• ranges from 0-100%
• true positive / (true positive + false negative)

Question 5

What is specificity?

Answer 5

probability that an individual without the disease has a negative test result

• ranges from 0-100%
• true negative / (true negative + false positive)

Question 6

What is positive predictive value?

Answer 6

probability that an individual with a positive test result has the disease

• true positive / (true positive + false positive)

Question 7

What is negative predictive value?

Answer 7

probability that an individual with a negative test result does not have the disease

• true negative / (true negative + false negative)

Question 8

What are likelihood ratios?

Answer 8

ratio of 2 likelihoods

• calculated from knowing the sensitivity and specificity of the test
• combine prevalence and sensitivity

Question 9

What is LR < 1?

Answer 9

the less likely the disease/condition

Question 10

What is LR = 1?

Answer 10

no change in likelihood of disease/condition

Question 11

What is LR > 1?

Answer 11

the more likely the disease/condition

Question 12

What is positive likelihood ratio?

Answer 12

probability that an individual with the disease has a positive test result, divided by the probability that an individual without the disease has a positive test result

• sensitivity / (1−specificity)
• > 1

Question 13

What is negative likelihood ratio?

Answer 13

probability that an individual with the disease has a negative test result, divided by the probability that an individual without the disease has a negative test result

• (1−sensitivity) / specificity
• < 1

Question 14

Which characteristic of diagnostic tests are all you really need in practice?

Answer 14

likelihood ratios

Question 15

What is the area under the receiver operating characteristic curve (AUROC)?

Answer 15

probability that a classifier will correctly rank a randomly chosen person with the disease higher than a randomly chosen person without the disease

• area under a plot of sensitivity against (1−specificity)

Question 16

What is prevalence?

Answer 16

proportion of individuals in a population having a disease at a particular time point = pre-test probability

• ranges from 0-100%

Question 17

What characteristics of diagnostic tests take prevalence into account?

Answer 17

• predictive value varies with the prevalence of the condition
• likelihood ratios can take into account different prevalences

Question 18

What characteristics of diagnostic tests do NOT take prevalence into account?

Answer 18

• sensitivity
• specificity

Question 19

What is the basis for sensitivity?

Answer 19

true positive

• false positive: not specific enough

Question 20

What is the basis for specificity?

Answer 20

true negative

• false negative: not sensitive enough

Question 21

What are the 3 ways you can get a ‘feel’ for the pre-test probability?

Answer 21

• by experience
• by looking at the literature
• by discussing with a colleague

Question 22

What is the impact that prevalence has on false positives and false negatives?

Answer 22

• false positives are the problem in low-prevalence settings
• false negatives are the problem in high-prevalence settings

Question 23

Given a LR and a baseline estimate describe generally how one would make an estimate of a post-test probability.

Answer 23

1. what is the baseline estimate of the problem/diagnosis

• aka pre-test probability
• rough estimate such as ~1%, ~10%, ~20%, ~30%
• informed by the prevalence of the condition and your assessment of patient-specific factors – somewhat intuition

2. what is the treatment threshold – is baseline estimate above or below threshold

• severity of the condition
• potential benefits/harms/inconveniences of the treatment
• utilizing a patient’s values and preferences in a shared decision-making way
• only do a test if it will change what you would do

3. is there a test that could increase or decrease the baseline estimate enough to change your or your patient’s treatment decision
4. make an informed decision

Question 24

Two Different Scenarios with Somewhat Different Approaches

Answer 24

• more common/acute symptomatic conditions – baseline chances > ~10%
• less common conditions and screening decisions – baseline chances < ~2%