Aa
Sensitivity of tests and testing question
I've been reading this forum and have learned a lot about HIV transmission and testing from you both. Thanks for providing this service. I'm male and I had an encounter about 8 weeks ago with a female escort. Some oral sex (with condom) followed by vaginal sex (with the same condom, actually). Afterwards I checked the condom and it didn't break.
6 weeks later, for whatever reason, I started to worry about STDs, particularly HIV. I bought a test at the drugstore (Oraquick, using the oral swab) and it was negative at 41 days. From other responses you've written recently, I see you think the oral test has sensitivity (i.e., probability of the test being positive if infected) of 90% at 6 weeks. Is this correct?
Given I wore a condom, even if she had HIV, my belief is that this is very strong evidence that I did not catch HIV from this encounter. Is this correct?
Here's why I believe this (FYI I have a degree in statistics). The probability of transmission (assuming she was positive, which is unlikely) is 1/2000 multiplied by the probability of condom not working (unlikely), let's say 1/20 (i.e., 95% effectiveness) = 1/40,000. Assume the test is 90% sensitive and, based on the FDA, is 99.98% specific (prob. of testing negative if not infected). This gives a negative predictive value (prob. of not being infected when testing negative) of 99.9997%. In other words, there's about a 1 in 400,000 chance of being infected based on these (conservative) assumptions. Does this seem right?
I'm wondering if this sufficient proof or whether I should test again. It is 8 weeks. I think I shouldn't because if the oral test is now 95% sensitive, the negative predictive value becomes 99.9999% or 1 in 800,000, which isn't a big difference. Do you agree?
Note: I fleshed out the calculation because this might help others in similar situations. It helped me put things in perspective. Hearing a test is "only 90% accurate" seems scary, but not when you do the math.
6 weeks later, for whatever reason, I started to worry about STDs, particularly HIV. I bought a test at the drugstore (Oraquick, using the oral swab) and it was negative at 41 days. From other responses you've written recently, I see you think the oral test has sensitivity (i.e., probability of the test being positive if infected) of 90% at 6 weeks. Is this correct?
Given I wore a condom, even if she had HIV, my belief is that this is very strong evidence that I did not catch HIV from this encounter. Is this correct?
Here's why I believe this (FYI I have a degree in statistics). The probability of transmission (assuming she was positive, which is unlikely) is 1/2000 multiplied by the probability of condom not working (unlikely), let's say 1/20 (i.e., 95% effectiveness) = 1/40,000. Assume the test is 90% sensitive and, based on the FDA, is 99.98% specific (prob. of testing negative if not infected). This gives a negative predictive value (prob. of not being infected when testing negative) of 99.9997%. In other words, there's about a 1 in 400,000 chance of being infected based on these (conservative) assumptions. Does this seem right?
I'm wondering if this sufficient proof or whether I should test again. It is 8 weeks. I think I shouldn't because if the oral test is now 95% sensitive, the negative predictive value becomes 99.9999% or 1 in 800,000, which isn't a big difference. Do you agree?
Note: I fleshed out the calculation because this might help others in similar situations. It helped me put things in perspective. Hearing a test is "only 90% accurate" seems scary, but not when you do the math.
MEDICAL PROFESSIONAL
Thanks for the thanks. I'm glad to have helped.
Hi Dr. Handsfield. Just wanted to thank you for your reply and for confirming the approach I took to make these calculations. Hopefully this thread is helpful to others, too.
MEDICAL PROFESSIONAL
Welcome to the forum. Thanks for carefully reviewing other questions similar to yours. You have correctly analyzed your risks and my replies.
The main thing I would add, which you probably saw elsewhere but don't mention, is that it's generally believed that most escorts (expensive female CSWs by appointment) are at low risk for HIV and other STDs, owing to consistent condom use, awareness of risks, frequent testing, and often careful selection of their clients. And that often their main clients are themselves at low risk, often monogamous -- i.e. men like you.
The official data for the oral fluids HIV test show similar performance of other antibody tests, approaching 100% reliability by 6-8 weeks. However, a couple of research studies suggest that's optimistic and that the figures you quote may be more reflective of reality. But it's an open question -- so to err on the side of caution, we recommend that only blood tests be used for testing earlier than 3 months after an exposure of concern. Anyway, your 6 week result probably is indeed at least 90% sensitive, perhaps higher.
I would modify your statistical calculation. I would start with 1 in 1,000 chance your partner had HIV (or 1 in 100, if you want to be conservative) -- so that even without a condom, it would calculate as 1 chance in 2 million (or maybe 1 in 200,000) you were infected. Second, I would assume 99-100% condom effectiveness. The concept of "use effectiveness" versus biological effectiveness of condoms enters in. The first takes into account that condoms sometimes break or are not properly used. In other words, over time, someone who relies exclusively on condoms to prevent HIV (or pregnancy, other STDs, etc) probably reduces his chance by 95%. However, if a condom is properly used and does not rupture, then biological effectiveness -- close to 100%, since HIV cannot pass through intact latex or polyurethane -- is the proper value to use.
Based on the preceding calculation, I agree with your analysis of negative predictive value. But as just discussed, it's probably even better than the figures you cite.
Finally, I agree exactly with your concluding statement -- which is why Dr. Hook and I often go through this sort of statistical analysis for questioners who haven't done it on their own, as you did. For both you and other readers, here's another thread that discusses the importance of interpreting test results not strictly on test peformance, but in light of the prior probability of catching HIV:
http://www.medhelp.org/posts/show/1596259
I hope these comments have further boosted your confidence that you were at little risk of HIV and that you were not infected. Best wishes--
HHH, MD
The main thing I would add, which you probably saw elsewhere but don't mention, is that it's generally believed that most escorts (expensive female CSWs by appointment) are at low risk for HIV and other STDs, owing to consistent condom use, awareness of risks, frequent testing, and often careful selection of their clients. And that often their main clients are themselves at low risk, often monogamous -- i.e. men like you.
The official data for the oral fluids HIV test show similar performance of other antibody tests, approaching 100% reliability by 6-8 weeks. However, a couple of research studies suggest that's optimistic and that the figures you quote may be more reflective of reality. But it's an open question -- so to err on the side of caution, we recommend that only blood tests be used for testing earlier than 3 months after an exposure of concern. Anyway, your 6 week result probably is indeed at least 90% sensitive, perhaps higher.
I would modify your statistical calculation. I would start with 1 in 1,000 chance your partner had HIV (or 1 in 100, if you want to be conservative) -- so that even without a condom, it would calculate as 1 chance in 2 million (or maybe 1 in 200,000) you were infected. Second, I would assume 99-100% condom effectiveness. The concept of "use effectiveness" versus biological effectiveness of condoms enters in. The first takes into account that condoms sometimes break or are not properly used. In other words, over time, someone who relies exclusively on condoms to prevent HIV (or pregnancy, other STDs, etc) probably reduces his chance by 95%. However, if a condom is properly used and does not rupture, then biological effectiveness -- close to 100%, since HIV cannot pass through intact latex or polyurethane -- is the proper value to use.
Based on the preceding calculation, I agree with your analysis of negative predictive value. But as just discussed, it's probably even better than the figures you cite.
Finally, I agree exactly with your concluding statement -- which is why Dr. Hook and I often go through this sort of statistical analysis for questioners who haven't done it on their own, as you did. For both you and other readers, here's another thread that discusses the importance of interpreting test results not strictly on test peformance, but in light of the prior probability of catching HIV:
http://www.medhelp.org/posts/show/1596259
I hope these comments have further boosted your confidence that you were at little risk of HIV and that you were not infected. Best wishes--
HHH, MD
You are reading content posted in the HIV - Prevention Forum
Condoms are the most effective way to prevent HIV and STDs.
PrEP is used by people with high risk to prevent HIV infection.
Can I get HIV from surfaces, like toilet seats?
Can you get HIV from casual contact, like hugging?
Frequency of HIV testing depends on your risk.
Post-exposure prophylaxis (PEP) may help prevent HIV infection.
