The reason why you get a high percentage (16%) is that the HIV negative population which is the population susceptible to a false positive result is so much larger (19 times) than the HIV positive population in your calculation. On an individual bases the test is as advertise. The chance of false positive test is 1 in 100. If the test is truly independent than the chance doing the test twice and getting a positive result both times is 1 in 10,000 Doing the test 3 times and getting a positive result all 3 times is 1 in a million and so on and so on. No medical professional would base a diagnosis on a single test.
I bought one of those HIV test-packs, where you put some blood on a piece of paper and mail it in. The ad says its 95% accurate. Now 95% might sound pretty good if we were talking about scoring from the line, but in this case it means that one out of twenty are getting a false test result. Not nearly good enough in my books. Has anyone had any experience with these packs, knows anything about how they operate, etc.?
The only one that I know of that is truely FDA approved is from Home Access Corporation, and it is over 99.9% accurate (which is the same accuracy as an office test).
Here is the URL for that one
http://www.homeaccess.com/02/01/
Which one did you purchase?
I wouldn't rely too much on this pack with that much of an error. Did it say whether that's more false negative or positive? I would think it would give you false positives more, just from the point of view of being safe about this. A false negative could be very tragic.
It doesn't replace testing at a clinic, but if it's false positives that they are talking about, you might consider getting the expensive clinic test only half as much, as long as this home test is negative, of course.
BTW, it's not whether STD questions were legit or not, it's whether they've been asked before in, say, the last two weeks or month. I've tried to answer the same question again and again. Now I've made a notepad copy of my last text, and I'll just cut & paste it from now on.
/Zin
There are two kinds of errors that a laboratory test can make: False Positives and False Negatives.
The "accuracy" of a test is usually described by its Sensitivity and Specificity. Sensitivity is the number of True Positives divided by the sum of True Positives plus False Negatives. Sensitivity is a measure of the test's ability to detect a condition if it is present. Specificity is the number of True Negatives divided by the sum of True Negatives and False Positives. Specificity is a measure of the test's ability to rule out a condition if it is not present.
Sensitivity and Specificity tend to be mutually exclusive. That is if the Sensitivity is high, the Specificity tends to be low and vice versa.
In the case of a home HIV test, you want a test that has high Specificity. That is, if the test says you are negative, you can be reasonably sure that you are truly negative. If you get a positive result, it only means you need to confirm the result with a higher Sensitivity test.
When the test states that its is 95% accurate, it is probably giving the total number of true results (negative and positive) divided by the total number of results. As a previous reply notes, this isn't very helpful since it doesn't tell you if the test is Sensitive, Specific, both, or neither!
I could be making math mistakes.
Suppose 1 person out of 20 who purchases an HIV test has HIV (I am guessing higher than the population. The population HIV rate is less than 1% in the US. The rate will be HIGHER in the group of people that purchase the test). Suppose the test is 99% accurate (1 chance in a hundred it is wrong).
You either have HIV (1/20) or you don't (19/20).
Chances of not having HIV and positive (19/20)*(1/100) =19/2000
Chances of having HIV and positive (1/20)*(99/100) =99/2000
=====
Chances the test says positive 118/2000
Chances a positive result is wrong 19/118, about 16%
Chances of not having HIV and negative (19/20)*(99/100)=1881/2000
Chances of having HIV and negative (1/20)*(1/100) = 1/2000
=======
Chances that the test says negative 1882/2000
Chances that a negative result is wrong: 1/1882, about 0.05%
The test is highly unlikely to miss a correct HIV diagnosis. The total error rate is around 16%, almost all of it false positives. If a million people take the test:
Positive Test Negative Test
950000 don't have HIV 9500 940500
50000 have HIV 49500 500
==== =====
59000 941000
Good luck!
You are describing (more or less) the Positive and Negative Predictive Value of a test. In my previous post to this thread, I discussed test Sensitivity and Specificity. Sensitivity and Specificity are purely characteristics of a test. Positive and Negative Predictive Value combine Sensitivity and Specificity with the prevalance of a disease in a given population to (as you so cogently described) give the chances of a true negative or positive in that population.
Positive and Negative Predictive Value tell us how useful it is to apply a given test in a particular population. They're a kind of epidemiologic analysis.
Let's say you took your same hypothetical HIV test and applied it to a population of IV Drug Abusers that is 90% HIV positive:
Chances of not having HIV and positive (10/100)*(1/100)
= 1/1000
Chances of having HIV and positive (90/100)*(99/100)
= 891/1000
=====
Chances the test says positive 892/1000
Chances a positive result is wrong 10/892, about 1%
Chances of not having HIV and negative (10/100)*(99/100)
= 99/1000
Chances of having HIV and negative (90/100)*(1/100)
= 9/1000
=======
Chances that the test says negative 118/1000
Chances that a negative result is wrong: 9/188, about 8%
Bottom line: If you want to know how accurate your result on a test is, you need to use the test's Sensitivity and Specificity.
... I think. From a practical point of view, if you KNOW that you practice risky behaviors that increase your chance of HIV, the chances go up that the test will give you a false negative (the worst situation). If you don't know what behaviors are risky, donate blood to the ARC and steal a copy of the questionaire you have to fill out.
I think you shoult look at one or more of the alternatives below if you practice these behaviors.
- stop behaviors. Find other ways to express your sexuality. Don't use needle drugs.
- modify behaviors to make them less risky. cut risky behavior incidents in half. Use barrier protection. Buy yourself a bottle of clorax & use it.
- regular multiple independent testing (vs regular testing. e.g. do two tests each time you would do one test). Use (clinically) different tests if possible.
The reason why you get a high percentage (16%) is that the HIV negative population which is the population susceptible to a false positive result is so much larger (19 times) than the HIV positive population in your calculation. On an individual bases the test is as advertise. The chance of false positive test is 1 in 100. If the test is truly independent than the chance doing the test twice and getting a positive result both times is 1 in 10,000 Doing the test 3 times and getting a positive result all 3 times is 1 in a million and so on and so on. No medical professional would base a diagnosis on a single test.
Your calculations on the chances of a false positive repeating the same test multiple times is not quite correct.
False positives may be caused by many factors. For instance, some False Positives are caused by random variations in the way the test was manufactured. If you got a False Positive for this reason and you repeated the test with the same manufacturing batch, you should get the same False Positive result.
Other False Positive can be caused by random variations in the way the test was performed. If this was the cause of your False Positive, repeating the test will probably yield a different result but you can't tell which result was the true one!
In practice, HIV testing is done by first performing a high Sensitivity/low Specificity (few False Negatives, relatively many False Positives) Enzyme Linked ImmunoAssay (ELISA). If the result is positive a high Specificity (few False Positives) Western Blot test is done to weed out the ELISA's False Positives.
By using two different tests with different characteristics, you get both high Sensitivity and high Specificity.