Evaluating Tests and Other Medical Evidence
1. To evaluate
the performance of a new diagnostic test, the developer checks it out on 100
known cases of the disease for which the test was designed, and on 200 controls
known to be free of the disease. Ninety of the cases yield positive tests, as
do 30 of the controls. Based on these data, the specificity and false positive
error rates of the test are, respectively,
a. 75% and 10%
b. 85% and 10%
c. 85% and 15%
d. 85% and 25%
e. 90% and 15%
Based on all the information currently available, you estimate that the patient in your office has a one in four chance of having a serious disease. You order a diagnostic test with sensitivity of 95% and specificity of 90%.
2. The result
comes back positive. Based on all the information now available, the chance
your patient really has the disease is closest to
f. roughly 30%
3. The result
comes back negative. Based on all the information now available, the chance your patient really has the disease is closest to
4. Test A has 70%
sensitivity and 90% specificity. Test B has 90% sensitivity and 80%
specificity. Assuming that responses to Tests A and B are statistically
independent, and assuming pre-test probability of disease of 50%, determine the
sensitivity and specificity, as well as the positive and negative predictive
a. using Test A and Test B in parallel.
b. using Test A and Test B in series, Test A first.
c. using Test A and Test B in series, Test B first.
5. A test with
99.9% sensitivity and 99% specificity is used to screen a population for a
disease with 1% prevalence. The proportion of test positives in the screen who
actually have the disease will be roughly
6. The American Disease X Foundation reports that 6% of the population over 50 years of age has Disease X. You inquire as to the source of their information, and they cite disease population screening data in the literature which reports that 6% of that population was positive when screened. Referring to the literature, you discover that the screening test used had sensitivity of 95% and specificity of 98%. What proportion of the population over 50 years of age do you think really has the disease?
7. Three tests
exist that might be used in screening a population for Disease D, which you
believe is present in about 1% of the population. Test A has sensitivity of 99%
and specificity of 99.9%. Test B has sensitivity of 95% and specificity of 99%.
Test C has sensitivity of 99.9% and specificity of 98%. Using the percentage of
test positives to estimate the population prevalence of the disease would most
help you support the corresponding institutional position, if you were
a. CEO of the American Disease D Foundation
b. Newt Gingrich
c. Director of the UI's Institute for Truth in Medicine
8. In an article in the Archives of Your Specialty, researchers report that screening Test T is enormously effective, and should be used routinely. As evidence, they report that the five-year survival rate of patients whose disease was discovered by screening with Test T is 90%, whereas the five-year survival rate of patients whose disease was discovered by other means is only 60%. However, in an article that same month in the Journal of the College of Your Specialty, another group of researchers report that screening with Test T has no effect. As evidence, they report that a study which randomized subjects to a program of routine screening with Test T or to no special screening regimen found an 80% five-year survival rate in both groups. The issue seems even more confusing because this second group of investigators acknowledges that if their data are reanalysed to compare the survival rate of subjects who actually followed the screening regimen to those who did not, whatever group they were assigned to, the cohort of those who actually were screened had five year survival of 85% whereas the cohort of those who were not had survival of only 75%. Despite this, these investigators maintain the position that no effect of screening has been demonstrated. Write a brief note for the General Practitioner's Quick Office Practice Digest, explaining why these various sets of results might be expected to in the way they do and advising whether or not to screen with Test T.
9. An article in The New York Times last year, in connection with the prostate cancer treatment of Buffalo Bills head football coach Marv Levy and Washington, D.C. mayor Marion Barry, pointed out that annual prostate cancer diagnoses have almost doubled since 1990, and that most experts attribute this to increased detection of existing cancer via the prostatic specific antigen (PSA) test. The Times article suggested that early detection via the PSA test was leading to early and inevitably more successful treatment of prostate cancer, with a cure rate in the 80-90% range for men with early diagnosis. The Times gave no specific data pertaining to the beneficial effect of the increase in early diagnosis produced by use of the PSA.
Among the data
that might be used to support claims that use of the PSA is improving the
health outcomes of prostate cancer victims are
a. the sensitivity and specificity of the PSA test;
b. the positive and negative predictive values of the PSA test in the specific population of interest;
c. the overall five-year post-dx survival rate (sometimes called the "cure rate" in cancer studies);
d. the difference in such rates between men diagnosed by PSA tests as compared to those diagnosed by other means;
e. the difference in such rates between men who've previously had one or more PSA tests as compared to those who've not, whatever the method of diagnosis;
f. the mean overall post-dx survival;
g. the difference in these means between men who've been diagnosed by PSA tests as compared to those diagnosed by other methods;
h. the difference in these means between men who've previously had one or more PSA tests as compared to those who've not;
i. the age-adjusted mortality rate for prostate cancer;
j. the difference in rates of diagnosis between men randomly assigned to receive periodic PSA screening and men randomly assigned to periodic digital rectal examination;
k. the difference in five year post-dx survival rates between men randomly assigned to receive periodic PSA screening and men randomly assigned to periodic digital rectal examination;
l. the difference in age-adjusted mortality rates for prostate cancer between men randomly assigned to receive periodic PSA screening and men randomly assigned to periodic digital rectal examination.
With reference as appropriate to particular biases, explain why each of the above types of data does or does not provide reliable information on the impact of PSA on the health of men with prostatic cancer.
10.Which of the
following is not conducive to success of a medical screening program?
a. high incidence
b. long duration of inapparent disease
c. excellent therapeutic results at all stages
d. screening inexpensive
e. high specificity of the test
being misled by stage-migration bias in comparing outcomes over different
periods of time between which diagnostic acuity has changed, we should
a. compare stage-specific outcome measures
b. compare directly stage-adjusted outcome measures
c. evaluate outcomes in the later period using the SMR based on reference outcome rates from the earlier period
d. compare outcome measures combining and unadjusted for disease stage