Quiz week 30
| Disease +ve | Disease -ve | |
| Test +ve | A |
B |
| Test -ve | C |
D |
Q1. All of the following formulae are correct and represent sensitivity, specificity, positive predictive value and negative predictive value EXCEPT:
The answer is choice 2. PLEASE see the explanation below.
Q2. Now we think of a machine which opens oysters that have pearls inside. We feed it 20 oysters. It opens 8 of them. 2 out of the 8 did not have pearls. We then went on to open all those that it did not open and found 3 pearls but 9 were without pearls. What is the specificity of the machine?
The answer is choice 3
The 5 x 5 table version of the 2 x 2
I suggest that you practice making this table by yourself after you have seen my version of it. I think that the 2x2 table is a misnomer because it is actually a 3x3 table. One has to know each of the boxes in it to correctly calculate. My version of it has 2 more columns and rows and it is the answer to all the questions from a "3x3" table.
D on topÞ +ve before -ve T belowß |
Disease +ve | Disease -ve | Values ß | |
| Test +ve | A |
B |
Þ
The general direction is from Left to Right |
Positive predictive value = a/a+b (left/left+right) |
| Test -ve | C |
D |
Ü The general direction is from Right to Left |
Negative predictive value = d/d+c (right/right+left) |
ß The general direction of the equation is from top to bottom |
Ý The general direction of the equation is from bottom to top |
Odds ratio is criss- cross direction WITH RISK FACTOR TAKING THE PLACE OF TEST (upper leftxlower right)/(upper rightxlower left) AxD/BxC | ||
| S & S Þ | Sensitivity = a/a+c (top/top+bottom) | Specificity = d/d+b (bottom/bottom+top) | Published 11/28/99 superscore.com |
Now I hope you see how systematic this table is and how it clarifies the whole scenario for the following 5 things.
Sensitivity = I am going to explain the sensitivity of a radio. There are 10 stations in the air. If it can only pick up 7 of them, it is 70% sensitive. Therefore if there are 10 cases of a disease in a population, and only 7 can be picked up by the test, it is 70% sensitive.
Specificity = Of all that are free of the disease, how many actually test negative. Now we think of the machine which opens oysters that have pearls inside. We feed it 20 oysters. It opens 8 of them. 2 out of the 8 did not have pearls. We then went on to open all those that it did not open and found 3 pearls but 9 were without pearls. Therefore the specificity is 9/11 - i.e. of the total without the pearls (11), it correctly identified 9.
Positive predictive value: We developed a test that can tell us with fair accuracy about a problem. It is - lets say for learning purposes sensitivity is 80% and specificity is 90%. Therefore if the test is done in a population of 20 people where the prevalence is 50%, 10 are without disease and 10 are with disease.
Of the 10 who have the disease, only 8 will turn positive on the test (because the sensitivity is 80%).
Of the 10 who do not have the disease, only 9 will be negative (because the specificity is 90%).
Therefore we have a total of 8 true positives and 1 false positive = 9 positives. Positive predictive value is used to tell us that if a person tests positive, what are the chances that he truly has the disease. In this case, it is 8/9.
Now to understand negative predictive value, of all those who tested negative, how many are truly negative? Easy?!!
2 out of the 10 who have the disease actually came out negative on the test (because it was only 80% sensitive) and 9 of those who do not have the disease turned out negative (because the specificity is 90%). Therefore a total of 11. Now how many of those who turned negative on the test are truly negative? This is the negative predictive value.
The answer is 9/11. So the prevalence, sensitivity and specificity are all key in this calculation. The 2 x 2 chart gives us the sensitivity and specificity but where examiners trick us is by not giving the prevalence and making us waste time in the test because without that, no prediction can be made.