Diagnostic Test Characteristics
If you forget to tare (ie, reset to zero) a scale after placing a container on it, and then weigh the same thing on it multiple times, which measure will be affected: precision or accuracy?
If you forget to tare (ie, reset to zero) a scale after placing a container on it, and then weigh the same thing on it multiple times, accuracy will be affected, as all your measurements will have the same systematic error (the weight of the container), whereas precision should not be affected.
Which of the following is a required characteristic in a good screening test?
Low false negative rate => The goal of a screening test is to detect as many affected individuals as possible and minimize the number of false negatives (individuals with the disease that are missed
False positive rate
1-specificity
A decrease in ________ of the condition tested increases the test's NPV and decreases its PPV.
prevalence
_______is the proportion of individuals with the disease that are true positives (TPs). It's calculated as TP ÷ (TP + FN)
Sensitivity
SnOut:
Sensitivity (SN) rules Out (when the test is negative)
High accuracy, high precision
Darts are clustered in the same place, and are clustered in the bullseye
Low accuracy, low precision
Darts are scattered, and all around the dart board
If the diagnostic cutoff of a test is increased, which of the following changes will occur to the diagnostic test characteristics?
Increased positive predictive value => If the diagnostic cutoff is higher, fewer false positives will be seen, and the PPV will increase.
A screening test needs to be more ______
Sensitive
if a low cutoff value is chosen (eg, cutoff point A in Figure 5), the test will be more _____
Sensitive
Sensitivity Formula
TP/(TP+FN)
A positive screening test is therefore often not sufficient for diagnosis and requires a _________. This type of test needs to be highly specific to ensure that those with a positive result have a high likelihood of having the disease.
confirmatory test
The ________is the proportion of negative test results that are true negatives, or the proportion of individuals without the disease among negative tests
Negative predictive value (NPV)
A confirmatory test needs to be more ______
Specific
if a high cutoff value is chosen (cutoff point C), the test will be more ______
Specific
SpIn:
Specificity (SP) rules In (when the test is positive)
What is the most important diagnostic test characteristic for a screening test, and what is it for a confirmatory test?
A screening test needs to be sensitive, and a confirmatory test needs to be specific.
____is the ability of a diagnostic test to generate results that are true and free of systematic error or bias.
Accuracy
If the prevalence of a disease increases in a population, which of the following changes will occur to the diagnostic test characteristics?
Decreased negative predictive value => With increasing prevalence, a larger proportion of negative tests will be false negatives and the NPV will decrease.
If the positive predictive value of a test has increased in a population, does that mean the prevalence has increased or decreased?
If the PPV has increased, then disease prevalence has increased.
In summary, the following changes in diagnostic test characteristics occur when the prevalence changes:
Increase in prevalence: higher PPV, lower NPV Decrease in prevalence: lower PPV, higher NPV In both cases: no change in sensitivity or specificity
If a test is very specific, few individuals without the disease will have a positive test ______so most positive tests will be true positives, and the PPV will be high.
Low FP
_____(reliability) is a test's reproducibility, consistency, and the extent of variation in repeated results. If a test is precise, then the difference between repeated measures should be small. But that does not mean the test is accurate or valid—it could be reproducible but wrong!
Precision
Increasing a test's cutoff value will increase its ______and PPV but decreases its sensitivity and NPV.
Specificity
________ (true negative rate) is the proportion of individuals without the disease that are true negatives. It's calculated as TN ÷ (TN + FP)
Specificity
Positive predictive value (PPV)
TP / (TP + FP)
Rising the cut off
B=> C Increase in FN decrease in FP Increase in specificity=> Increase in PPV Decrease in sensitivity=> Decrease in NPV
. If a test is very sensitive, few individuals with the disease will have a negative test ________so most negative tests will be true negatives, and the NPV will be high.
Low FN
Negative predictive value
NPV = TN / (FN + TN)
Specificity gives you the ___ rate
True Negative Rate
Sensitivity gives you the ____ rate
True Positive Rate
____ Patient with the disease has a positive test result.
True positive (TP)
Decrease in prevalence
decrease PPV increase NPV
Specificity formula
TN/(TN+FP)
_____Patient with the disease has a negative test result.
False Negative (FN)
___Patient without the disease has a positive test result.
False Positive (FP)
_____is the proportion of positive tests that are true positives, or the proportion of individuals with the disease among positive tests.
Positive predictive value (PPV)
If the cutoff value of a test is changed to make it more sensitive, what will be the effect on the PPV?
The PPV will decrease; sensitivity has increased, meaning that the false positive value has increased.
Receiver Operating Curves
This compromise between sensitivity and specificity can be visualized on a receiver operating characteristic (ROC) curve
____Patient without the disease has a negative test result.
True Negative (TN)
____often is used interchangeably with accuracy; however, it more specifically refers to whether the test measures what it was intended to measure. Validity includes aspects beyond accuracy, including methodology and study design.
Validity
An ideal screening test needs to be highly sensitive to decrease the number of ____
false negatives.
An ideal confirmatory test needs to be highly specific to decrease the number of ____
false positives
Lowering a test's cutoff value will_______its sensitivity and NPV but decreases its specificity and PPV.
increase
If a patient has a positive test result and the PPV for the disease tested is 40%, what are the chances that they do not have the disease?
60% (chance of having the disease is 40%, so chance of not having the disease = 100% - 40%)
Diagnostic tests are not perfect. To calculate a new test's characteristics, you must:
=>Use the test in a patient sample that would normally require evaluation for the outcome in question =>Have everyone receive the diagnostic test as well as the gold-standard test (the best possible test for a condition, which is used as a benchmark) => Compare the results of the diagnostic test against those of the gold-standard test
When a blood sample glucose level is measured 10 times by five different glucometers, they all give the same value: 102 ± 3 mg/dL. However, a sixth glucometer gives a value of 102 ± 12 mg/dL. Compared to the first five glucometers, what is the precision and accuracy of the sixth one?
No change in accuracy, decreased precision =>The mean glucose value is the same as the other glucometers, so the result is accurate. => However, the higher standard deviation indicates more variability between the measurements, meaning that precision is lower
Lowering the cut off
B=> A Increase in FP decrease in FN Increase in sensitivity=> increase in NPV Decrease in specificity=> decrease in PPV
Increase in prevalence
increase PPV decrease NPV
low accuracy, high precision
All on the target but all over the place
Which type of test is better to rule in a disease, and which one is better to rule out?
A specific test is good to rule in a disease when its results are positive, whereas a sensitive test is good to rule out a disease when its results are negative.
High accuracy, low precision
All are close to the bullseye, but not all in the same place