HYMR Diagnostic Tests Flashcards
The diagnosis of pneumonia is based off a number of factors which include the history of present illness (HPI), vital signs, physical exam, and chest radiograph. A study was published to evaluate the ability of radiology residents to find and diagnose pneumonia on radiographic evaluation. Which of the following might be published in this type of paper to guide the reader to the amount of interrater agreement?
A. I-Squared Statistic
B. Kappa statistic
C. Beta statistic
D. Alpha statistic
B. Kappa statistic
- In studies done where someone has to interpret a physical exam finding, lab test, radiograph (i.e.. “x-ray”), or some other diagnostic test, there maybe disagreement or differences in results. In order to discern the degree of “agreeement” or “interobserver agreement” or “interrater agreement, these studies will publish the Cohen’s Kappa statistic.
- The I-Squared statistic is used when doing a meta-analysis to determine the degree of heterogeneity.
- The alpha and beta statistic are made up and there as distractors.
True or False: The sensitivity of a diagnostic test reflects its ability to rule out the presence of disease.
A. True
B. False
A. True
- The sensitivity (SN) of a diagnostic test is used when trying to rule out the presence of disease.
- The formula is SN = TP/TP + FN. As such, the more false negatives (FN) you have the lower your SN will be.
- Most clinicians desire to have tests with a high SN when screening a patient to rule out a disease or complication.
High-Yield Core Concept:
* The sensitivity of a diagnostic test is used to determine the ability of a test to detect the presence of disease when it is actually present in the patient.
Abbreviations:
* SN = sensitivity * TP = true positive * FP = false positive
Which of the following can reduce the accuracy of a diagnostic test?
A. Increased Presence of Bias
B. Low Amount of Random Error
C. High Incidence of
Disease
D. Large Sample Size
A. Increased Presence of Bias
- The greater the amount of systematic error or amount of bias present in a study the more negative of an impact it can have on the precision of a test. Also having a lot of random error can also contribute to a reduced accuracy as noted in the image (See picture)
Which of the following is most accurate about the use of a test that has a likelihood ratio (LR) of 10?
A. The pre-test probability will be larger than the post-test probability
B. The sensitivity is less than the specificity for that test
C. The post-test probability will be larger than the pre-test probability
D. The post-test odds will be less than the pre-test odds
C. The post-test probability will be larger than the pre-test probability
- A likelihood ratio (LR) greater than 1 will usually result in a post-test probability that is greater than the pre-test probability.
- Having a LARGE LR (especially if > 10) will usually increase the post-test probability that patient has the disease you are suspicious for and thus, will usually be helpful in doing the test.
The above is true because LR+= (TP%/FP%) … or… - Sensitivity/(1-Specificity).
(See figure)
Which of the following best reflects a test that can be repeated over and over again and get the same results, but those results being obtain may not reflect the true result when compared to the gold standard?
A. The test is accurate but lacks precision
B. The test is precise but has a lot of random error
C. The test is accurate and has very little random error
D. The test is precise but lacks accuracy
D. The test is precise but lacks accuracy
- A test has good precision or is precise when the results are consistent and “reproducible” of a test result. However it does not mean its accurate to true value.
- A test that is accurate reflects the “trueness” of a test and gives a measurement close to “actual” or “true” value (as compared to the gold standard measurement). Just because a test is accurate does not always mean it will be precise.
See diagram below to see how bias and random error contribute.
