An interesting quote on several levels….
For those of you who don’t follow The Drudge Report, there was an interesting story of a mistaken HIV diagnosis there today.
Defending the liberal tradition in history, science, and philosophy.
Support Dean’s World
{ 10 comments }
“This Machine Does Not Lie”…a very, very dangerous point of view, in medicine or any other field.
About 10 years ago, a Washington Metro train slammed into the back of the train in front of it, killing the driver. Conditions were icy, and the driver had requested permission to switch from automatic to manual control, which was denied. According to the NTSB report, requests to switch to manual were routinely denied, with comments like “let the train do what it’s supposed to do.”
The report explains why this was unwise, to put it mildly, given the actual behavior of the train control system.
Automated systems can be very useful, but should never be regarded as oracles.
There are 2 huge problems with the HIV test:
1. It detects antibodies, not the virus itself.
This is like seeing a fire-engine, and then concluding there is a fire. There could be a fire, but there could also be a cat stuck in a tree, or a group of firemen going to a car wash. It is an indirect measurement.
2. It was approved by the FDA as a screening test.
When it was patented in 1984, it was sold as a method to prevent contamination of the blood supply. In essence, the test would be given to all blood donations — if the test “reacted”, the blood donation was rejected; if the test did not “react”; the blood donation was accepted.
Somehow, this “screening test,” morphed into a “medical diagnosis” test.
Analogy: Airport screening — we screen out metal objects, because we don’t want weapons made of metal on the plane. Perfectly kosher practice.
However, now if the machine goes beep, we not only declare that you have a metal object, but we presume that you are a terrorist, intent on causing harm.
In sum, the test is very non-specific and generates numerous false positives, all documented in the scientific literature.
The machine doth lie.
–HB
Surely, its always scientifical honesty to hop on an authoritative New York Post article (cough) and claim it as a jumping off point to sell ones authentic agenda (cough).
McK,
I will agree that article is short on science.
But I don’t see why that should limit it as a jumping off point for a discussion that is based on science.
“Scientifical honesty?”
What makes the New York Post less “authoritative” than, say, the New York Times?
It was a “human interest story.” The people were real and the story was painful to read and quite complex. Somebody sent it to me minutes after it came out and I sent it to many others. It felt like a play, like a Pinter play or something, this intact family, devoted, solid, and then the devil comes in. Tiny crack. Nobody thinks twice about it…Until…the fire has burned down what was sacred in the family. Trust.
Imagine the possibilities. One quick swab.
Do you know how many of these are going to keep coming up as time progresses? Tip of the iceberg.
McK, let me tell you something. I have written for the New York Post on a few occasions. They had the very best, most rigorous fact-checking department I’ve ever worked with, just about. They checked EVERY word, with the source.
NYT, I think, has no fact checking dept. If I’m not mistaken.
Not to cast aspertions on your Post colleagues, Celia, but this story as is doesn’t pass even the most basic tests of credibility.
The description of what was supposed to have happened at the hospital is scanty, but bears little relationship to actual diagnostic procedures.
For example, confirmatory HIV testing, which is mandatory before a positive diagnosis can be given, is not done by machine.
Checking a story with your source does nothing to make it credible if the source is not credible in the first place.
Perhaps there was a failure of communication or understanding here, or perhaps there was malice involved. A court might sort this out, but something tells me this “case” isn’t going to get anywhere near being heard in an actual court.
Hank, from what you have written you seem to be under the impression that people are diagnosed as having HIV solely on the basis of a reactive screening test, and one that uses circa 1984-5 technology at that. This isn’t the case, any more than anyone is determined to be a terrorist solely because they set off a metal detector.
And it appears that you have decided there is something fundamentally flawed with the entire scientific discipline of diagnostic serology. There’s nothing unusual about using antibodies to identify various types of infections. Where did you get that idea?
Hi Fraser:
Hank, from what you have written you seem to be under the impression that people are diagnosed as having HIV solely on the basis of a reactive screening test, and one that uses circa 1984-5 technology at that.
No, I’m fully aware of the “algorithm” of HIV testing, involving repeated ELISA tests and Western Blots (for antibodies only)
And it appears that you have decided there is something fundamentally flawed with the entire scientific discipline of diagnostic serology.
Not at all — but I think there is something fundamentally flawed with ANY test that yields so many false positives.
More so, I think there is something fundamentally flawed with unvalidated diagnostic serology. There is no large study (that I am aware of) that has validated the proposition that detection of antibodies equates with detection of virus.
The lack of validated HIV tests explains all the false positives I linked to above.
The numerous false positives likely explains the lawsuit about which Jaymaster has written above.
This type of lawsuit has happened before in the Audrey Serrano Case , although that was more doctor error, than test error.
I do appreciate your civility and vast knowledge on the topic — well done.
–HB
The New York Post circa 2009:
A city hospital nearly destroyed a New Jersey woman’s life and wrecked her marriage after misdiagnosing her with terminal HIV, hepatitis and herpes, according to a bombshell lawsuit
The scientific literature circa 1992:
“Multiple False-positive Serologic Tests for HIV, HTLV-1 and Hepatitis C Following Influenza Vaccination.
“Our study data revealed that influenza vaccination was associated with the occurrence of multiple false-positive viral ELISA’s among blood donors. The proposed mechanism is believed to be due to an early nonspecific IgM response..” Mackenzie, page 1017.)
–HB
Hank, I find it strange that you cite the Audrey Serrano case in a post claiming that diagnostic serology is inaccurate. In fact, diagnostic serology provided the crucial evidence that Serrano did not have HIV, and helped her win her case.
Her doctor was found to be negligent for treating her for HIV without ordering diagnostic serology. Instead, according to Dr Lai, “Serrano convinced her she had HIV when she took her personal history”. *raises eyebrow*
“… I think there is something fundamentally flawed with ANY test that yields so many false positives.”
Cross reactions can of course occur with any individual serological test, which is why an HIV diagnosis is only made after an algorithm of tests which provide multiple lines of serological evidence, and in light of the clinical data.
The reliability of an HIV diagnosis rests on the postive predictive value of the diagnostic algorithm as a whole, not just on the specificity of a single component of it.
You claim there are “so many” false positive diagnoses of HIV based on serological testing algorithms. I don’t deny that these can occasionally occur but they are extraordinarily rare, and usually result from using an algorithm of insufficient positive predictive value for the clinical situation (for example interpreting a weakly reactive EIA and a two-band WB as “positive” rather than “indeterminate” in a person whose clinically assesed pretest probability of HIV infection is extremely low).
There is actually some skill involved in making an HIV diagnosis, which is why it is generally left up to people trained and qualified in diagnostics – i.e. physicians. Like any professionals they can sometimes make mistakes – and Dr Lai’s looks like a doozy – but the diagnostic tools provided by HIV serology are about as good if not better than any other medical test. While the process can look like a black box to people without that training, no HIV diagnosis is made by a “machine”.
It’s telling that the case you cite in your argument that serological testing algorithms are unreliable in fact demonstrates the opposite. Dr Lai was found negligent for making a diagnosis without making use of serological evidence. Once the testing was eventually carried out, it proved her error.
“There is no large study (that I am aware of) that has validated the proposition that detection of antibodies equates with detection of virus.”
This is a difficult assertion to refute, because the goalposts aren’t fixed. How large is “large”? In fact, there are numerous studies from the late 80s on that demonstrate that an HIV diagnosis based on clinical and serological evidence equates with present HIV infection. Dissidents who claim otherwise almost always cite studies from around 1984-6 when viral detection methods for HIV were in their infancy, and often very insensitive.
[Edit] Hank, yes, no-one is denying that there are a number of well-known factors that can cause an EIA to react falsely. With the fourth generation EIAs you can expect that between 1 and 2 truly negative samples per 1000 will yield a false positive reaction, usually only slightly above cutoff. Which is why no positive HIV diagnosis is made on an EIA (ELISA) test result alone.
No experienced physician is going to get excited about an EIA reacting at 1.2 or 1.5 in someone where there isn’t a strong clinical suspicion of recent seroconversion (the cutoff for “reactive” being 1.0). It simply triggers the next part of the testing algorithm, which in the US is usually Western Blot testing. It’s a nuisance in many cases, especially in people at low risk for HIV infection, but you need to do it to get a reliable and definitive answer one way or another.
Of course, if this isn’t handled well, it can cause immense anxiety to the person being tested. Which is another reason why diagnosis, including HIV testing and its interpretation should be left to clinicians who know what they are doing.
Even if you believe the “consensus” wisdom on HIV, there’s nothing inherently fishy about the story; hospitals and doctors do bungle you know.
The one thing that screams out here is, why didn’t her husband get tested? Never mind if you think the tests are junk, it would have probably helped them both if he tested negative.
By the way, this would be far from the first time HIV testing confusion like this has nearly–or actually–ruined lives.
Comments on this entry are closed.