Diagnostics Quiz

Question 1

what is sensitivity? What does it mean if it is high or low?

Answer 1

propotion of diseased animals that will test positive
high: you’ll be able to “catch” the diseased animals and there will be low false negatives, your positives will be more likely to be true positives
low: you’ll miss a lot of the diseased animals so you may have false negatives and you may not trust a positive result very much

Question 2

what is specificity? What does it mean if it is high or low?

Answer 2

specificity is the proportion of the non diseased animals that will test negative
high: it means you’ll catch all the healthy or undiseased animals and there is a low chance of false positives
low: a lot of healthy animals will appear diseased, you’ll miss some of the healthy ones, and there is a high chance for false positives

Question 3

explain spin & snout

Answer 3

SpIN: you need a high specificity to rule IN disease because if specificity is high there is a low chance for false positives which means you can trust the positive test result
SnOUT: you need a high sensitivity to rule OUT disease because if sensitivity is high there is a low chance for false negatives which means you can trust a negative test result

Question 4

What is analytic sensitivity and specificity?

Answer 4

analytic sensitivty refers to the minimum amount/concentration an assay/test can detect
analytic specificity: the ability of the assay to only react to one compound and not cross react

Question 5

in a select point or cuttoff test, like the lactate example, what happens if you maximize sensitivity? what happens if you maximize specificity?

Answer 5

maximizing sensitivity means there will be few false negatives but there WILL be false positives, so you’ll think there’s more diseased animals than there is
maximizing specificity means there will be few false positives but there WILL be false negatives, meaning you could “miss” some of the diseased animals

Question 6

what is positive predicted value? what is it influenced by?

Answer 6

PPV: true positives/test positives
influenced by pretest probability and specificity (low false positives)

Question 7

what is negative predicted value? what is it influenced by?

Answer 7

NPV: true negatives/test negatives
influenced by pretest probability and sensitivity (low false negatives)

Question 8

what happens to our faith in a positive or negative test result when the pretest probability or true prevalence increases?

Answer 8

our faith in a negative result decreases

Question 9

what happens to our faith in a positive or negative test result when the pretest probability or true prevalence decreases?

Answer 9

our faith in a positive test result decreases

Question 10

what is a likelihood ratio of a positive test?

Answer 10

true positives/false positives , aka, how likely is this animal to be diseased if it has tested positive?

Question 11

what is a likelihood ratio of a negative test?

Answer 11

true negaties/false negatives , aka, how likely is this animal to be not diseased after testing negative?

Question 12

for series testing, who is considered positive? how are series tests usually done?

Answer 12

only those that are positive for ALL tests
these tests are usually done sequentially

Question 13

for series parallel testing, who is considered positive? how are series tests usually done?

Answer 13

those who test positive for only ONE of the tests
usually done at the same time, simultaneously

Question 14

what are we trying to increase when we use series testing?

Answer 14

we are increasing diagnotic specificity because we want to reduce false positives and we want our negatives to be true negatives

Question 15

what are we trying to increase when we use parallel testing?

Answer 15

increase diagnostic sensitivity because we want to reduce false negatives and we want our positives to be true positives

Question 16

if we are doing surveillence testing, what do we want to be high, and what kind of testing should we use?

Answer 16

we want a positive test to truly be positive (low false negatives), so we want a sensitive test, and we should use parallel testing (even if they test positive to one test they are considered disease positive). You’re being more cautious here

Question 17

if we are doing a confirmatory test to make a diagnosis, what do we want to be high, and what kind of testing should we use?

Answer 17

you want a negative test to truly be negative (low false positives), so we want a specific test, and we should use series testing (they are only considered disease positive if they test positive for ALL tests). You don’t want to treat a serious condition that they may not actually have.

Question 18

what is the difference between census and sampling testing? pros and cons for each?

Answer 18

census is sampling the entire population. It is expensive and not feasible. sample testing is a subgroup of the population selected to represent the entire population. It is prone to biases and errors

Question 19

if you want to detect disease prescence in a population, what information do you need to know (6 things)?

Answer 19

-populations size you are going to sample (you can figure this out with epitools based on the level of prevalence you want to be able to detect)
- the sensitivity and specificity of the test you’re using
- expected number of affected animals (approximate)
- acceptable error level of confidence level
- desired precision or decimal places
- the upper limit of sample size (how many could we ACTUALLY sample?)

Question 20

describe the difference between a type I and type II error

Answer 20

type I: concludes an association when there isn’t one (conclude a herd diseased when they arent)
type II: concludes there is no association when there is one (condlude a herd isn’t diseased when they are)

Question 21

describe the difference between random sampling, systemic random sampling, and stratified random sampling

Answer 21

random sampling: use numbers from a hat or software to randomly select a portion
systemic random sampling: selecting at an interval, like selecting every 3rd animal that comes through the chute, etc
stratified random sampling: assign srata groups then randomly select animals within those groups (age, sex, BCS, etc)

Question 22

what is herd sensitivity? what is it influenced by?

Answer 22

probability that an affected herd will yield a positive herd level test (have at least one positive). Influenced by indivudual Se and Sp, herd prevalence, and number of animals sampled

Question 23

what happens to herd sensitivity when you increase the number of animals sampled? What happens to herd sensitivity when you increase herd prevalence?

Answer 23

it increases for both (it’s easier to find a diseased animal if you sample more of them, and it’s easier to find a diseased animal if there are more diseased animals in your population to begin with)

Question 24

what is herd specificity? what is it infleunced by?

Answer 24

probability that an unaffected herd will yield a negative level test. influenced by the individual sp level and number of animals sampled

Question 25

what happens to herd specificity when you increase the number of animals sampled? what happens when you decrease the test specificity? What if there is a change in prevalence?

Answer 25

it decreases if you increase number of sampled animals (if you test more animals, chances are you’ll eventually find a diseased one) it decreases if you decrease test specificity (if your test isn’t as specific there will be more false positives) and it has NO effect on prevalence (because if they are unaffected the prevalence should be zero anyway)