"You should go ahead and assume that you have HIV," the doctor said.
It was just before 7 on a frigid January evening. I was alone in my apartment in Washington, D.C.; I hadn't been expecting a call from the doctor. Several weeks earlier, I had come down with a fever and spent a good week laid up in bed. When I finally called my primary care physician, whom I'd chosen only a month earlier due to his proximity to my apartment, his receptionist signed me up for the next available appointment, which was a week away. Desperate, I went to a hospital, then an urgent care center, where a doctor suspected some kind of infection and prescribed a hefty dose of antibiotics. Within a few days, my fever had faded, but I decided to keep my appointment regardless. My new doctor, whom I'll call Dr. Smith, gave me a checkup, drew blood, and promised to run a full panel of tests. I hadn't realized that HIV would be one of them.
I realized it now. "Your ELISA test went up to positive," Dr. Smith told me, "so it triggered a Western Blot, and that came back indeterminate. Let me tell you what that means. You're probably early on in your infection, so it's not showing completely yet. But these results, in my experience, always indicate HIV positivity. Why don't you come in tomorrow morning and we can get an RNA test and a confirmation?"
"A confirmation," I asked, "that I have HIV?"
"Yes," the doctor said. "We should definitely assume right now that you have HIV."
This assumption would be proven false after a week—a week of anxiety, anger, and, most of all, confusion. My HIV test showed indications of early infection, signs portentous enough to prompt a doctor to diagnose me on the spot. Yet as I received more tests and more information, it became increasingly clear that Dr. Smith's verdict was a misdiagnosis—that I did not have HIV.
How can a test so common, so regulated, so accurate fail? How could I, a healthy adult with virtually no risk of infection, be misdiagnosed with HIV? And if it could happen to me, who else could it happen to?
Most HIV tests don't test for HIV. Rather, the primary diagnostic tool, an immunoassay, tests for antibodies that react to HIV. (Earlier generations of immunoassays were called ELISAs, pronounced "Eliza"; most immunoassays today are a newer generation called EIAs, but many doctors still call them ELISAs as shorthand.) The immune system floods the body with these antibodies soon after HIV infection as it attempts, in vain, to fight off the virus.
If an immunoassay test turns up HIV-related antibodies, it's considered "reactive." But a reactive immunoassay is not considered proof of an HIV infection. It must be followed up with a confirmatory test, usually a Western Blot. These tests also search for antibodies, but rather than turning "reactive" or "nonreactive," they detect about 10 different "viral bands." These bands are specific antibodies that bind to different components of the virus. If a Western Blot turns up two or more reactive bands, the patient is considered HIV-positive. If it turns up one band or several "equivocal" bands, the test is considered "indeterminate" and should be run again in about a month.
That monthlong wait speaks to perhaps the most insidious aspect of the AIDS virus: It hides. After infection, the virus conceals itself completely for about 10 days—the so-called "eclipse period." (A person in the eclipse period is almost certainly not infectious.) Once that phase ends, the virus begins attacking the immune system, provoking the antibodies detectable by immunoassays and Western Blot tests. Yet the process of producing antibodies takes time, and so there is an additional "window period" between antibody production and immunoassay reactivity. For today's fourth-generation tests, the time between infection and detectability is about 15 days; for third-generation tests, some of which are still used, it's 25; second- and first-generation tests need 50 to 60 days of infection to detect any antibodies. Western Blots require about two months of infection before turning positive.
This period of undetectability also happens to be the most highly infectious phase of the disease. Between 30 and 50 percent of all HIV transmissions occur during the window period. Thus, it presents perhaps the greatest challenge to combatting HIV: How can a highly infectious person be stopped from spreading the virus if he can't know he has it?
I received a third-generation immunoassay, which measured an antibody level of 1.23, just faintly reactive. (A level below 1.00 is considered nonreactive.) Following protocol, the lab then ran a Western Blot, which turned up "indeterminate." Dr. Smith told this to me when I went into the office the following morning. I asked for further explanation.
"Well," he said, looking at my lab report, "it's actually kind of a funny result. You have two nonviral bands but no viral ones. I've actually never seen a case like this."
"Could that mean I don't have HIV?" I asked.
"Probably not," he told me. "It probably means you're just seroconverting."
Seroconversion occurs during the window period of infection, when HIV attacks the immune system and it responds with antibodies. It is sometimes—though not always—accompanied by flu-like symptoms, such as high fever, respiratory distress, bodily aches, and night sweats. Seroconversion, in other words, is a hypochondriac's worst nightmare, masking a deadly virus with fairly routine symptoms. One distinction between seroconversion and a standard flu is swollen lymph nodes, a frequent symptom of acute HIV infection. But these can also be caused by a variety of other maladies.
"You said you were sick recently, right?" Dr. Smith continued. "Had a fever? That may well have been seroconversion. If so, you'll test a true positive soon."
I had, in fact, learned about seroconversion the previous week, when a different doctor had floated the idea—just possibly because I am gay. When my fever refused to budge after a week, my boyfriend took me to the hospital, where I was hooked up to a saline drip and given a battery of (inconclusive) tests. After I introduced the doctor to my boyfriend, a strange look came over her face and she excused herself. About an hour later, she came back, pulled the curtain closed, and sat down next to my bed.
"Before you go, one last possibility to consider is HIV," she said. "You do have a fever, which can be a symptom of acute HIV infection. It's called seroconversion."
Somewhat alarmed, I asked if any of my other symptoms matched up.
"No," she conceded. "Your lymph nodes are normal, and you don't have any respiratory issues. But still," she continued, lowering her voice and glancing toward my boyfriend, "it's something to consider."
Floating such a drastic diagnosis may sound like an overreaction, but in fact, the doctor was merely following generally accepted guidelines for HIV diagnosis. The CDC considers homosexuality high-risk in and of itself. Being monogamous and practicing safe sex doesn't matter much when you're a "man who has sex with men." What matters is your demographic. And my demographic makes me a likely vector for HIV. Nine days after my visit to the hospital, my HIV test came back indeterminate.
Indeterminate Western Blot tests are a rare but intractable problem. Because they're used only as confirmatory tests, Western Blots have been the subject of much less technological advancement than immunoassays; it's more important to detect likely HIV than to confirm it beyond doubt. Some doctors and scientists see no place for Western Blots in the future of HIV testing; in fact, a few labs have begun to phase them out in favor for RNA tests. These search the blood for the virus itself, and have the immeasurable benefit of detecting HIV within only 15 days of infection. It is precisely because of this sensitivity and complexity, however, that RNA tests are rarely used as the first confirmation test: They remain prohibitively expensive for most labs and patients.
There are a number of reasons why a Western Blot could be indeterminate. Most indeterminate results arise from medical conditions that affect a patient's blood. Pregnancy, for instance, can cause a Western Blot to find traces of one viral band, called P-24. Autoimmune disorders can cause other viral bands to show up, as can the flu vaccine. Most of the time, these bands disappear in follow-up tests. If they don't but no further bands show up, the indeterminate result should be considered negative.
But there's another kind of indeterminate result, one so rare that Dr. Smith had never seen or even heard about it. This result indicates the detection of nonviral bands. These signal an error in the test itself. To run a Western Blot test, lab technicians grow HIV in a cell culture, isolate parts of the virus called antigens, and then expose them to a patient's blood serum. If the patient is HIV-positive, antibodies in his blood serum react to certain antigens, producing viral bands. But very rarely, the test can malfunction: Lab technicians may insufficiently purify HIV, allowing debris from the cell culture as well as antigens to be exposed to the blood serum. In turn, this debris—rather than the HIV antigens—may react with antibodies. The result: nonviral bands.
These bands do not indicate the presence of HIV. They don't even hint at it. In 1999, the National Confirmatory Testing Laboratory of the American Red Cross suggested that Western Blots with nonviral bands be considered immediate negatives, since "no individual exhibiting non-viral banding has been associated with either seroconversion detection of different HIV subtypes or other disease agents." The same year, the Association of Public Health laboratories stated that "non-viral bands should not be required to be reported [because] since 1991 no individual exhibiting non-viral banding has been associated with either early seroconversion, detection of different HIV-I subtypes, or other disease agents." And in 2000, the Blood Products Advisory Committee of the FDA proposed that nonviral band indeterminate Western Blots be considered completely HIV-negative, and thus safe to use in blood donations.
All of these groups recognized the same fact: Nonviral bands show up as a result of a lab error, not a blood infection. Yet none of their recommendations were ever taken. Many doctors are unaware of the nonviral bands and think them to be a sign of early seroconversion. Because the problem is so rare, it has attracted little research or attention from the press. When a patient receives this anomalous result, he is usually thrust into the formal process of an HIV diagnosis.
There are several ways to tell a patient he has HIV. Some doctors refuse to tell patients over the phone, choosing instead to make an appointment and tell them in person so they can quell the patient's panic and ask crucial follow-up questions. Others, however, believe that calling a patient and simply asking him to come in to discuss an HIV test is an obvious giveaway of an unwanted result, and that the news should simply be broken immediately, over the phone. This method has the benefit of ensuring that the patient knows his status whether or not he ever follows up. Rapid test immunoassays are quickly solving this conundrum: Clinicians can tell a patient his status in only 20 minutes. Rapid tests are popular in public health clinics but are not yet widely used by most primary care doctors, and so for now, the quandary remains.
Dr. Smith gave me his diagnosis over the phone and scheduled an appointment for the next morning. I spent that evening frozen in terror and confusion: There was simply no way, so far as I could see, that I could have been exposed to HIV. When I told my boyfriend, he was equally baffled: We each dug up the paperwork from our last two HIV tests, both of which were unambiguously negative. Nothing had occurred between those tests and this new one that could have exposed either of us to HIV. But this knowledge did little to allay the panic. We did not sleep that night.
The next morning at my appointment, Dr. Smith reiterated his strong suspicion that I was early into HIV infection and drew blood for a confirmatory RNA test. But he also suggested a rapid test immunoassay, which, by that point, should have turned positive. I sat for 20 minutes waiting for a second line to appear: Rapid tests contain colloidal gold, which reacts to HIV antibodies by producing a red line on a small screen.
As I waited for the test result, I began to think about what my life with HIV would look like. The virus, of course, is no longer a death sentence; those who have it can still live long, happy, mostly healthy lives. Moreover, the stigma once attached to HIV has largely dissipated, as HIV-positive people have rejected victim-blaming and empowered themselves both in their medical treatment and in the respect and decency they demand from society. The Internet provides myriad resources for those living with the infection, and virtual support groups foster a sense of community. No one has to face HIV alone, and no one should feel ashamed about being HIV-positive. If I did have HIV, I decided, I would not let it destroy my life. But I would much rather not have HIV. After 20 interminable minutes, no second red line showed up on my test—a negative result. My confusion grew deeper, though my spirits started to lift. I had now received one positive, one indeterminate, and one negative HIV test. I still didn't like my odds. But the situation seemed slightly brighter than it had the night before. Six days later, I received mailed lab results from the urgent care clinic. Unbeknownst to me, they had run a highly sensitive immunoassay—and it was completely negative. My blood was not testing positive for HIV. The immediate sensation was not one of comfort, or even relief. It was bewilderment. Nothing that had happened to me seemed to make sense. And when it abruptly stopped happening, that made little sense, either.
Finally, more than two weeks later, my RNA test came back negative. Dr. Smith didn't explain the delay—the test takes less than a week—nor did he attempt to explain his original misdiagnosis.
"I have no idea why the Western Blot was indeterminate," he said. "But we should run one more immunoassay."
He ran one more immunoassay. It came back negative.
In all, I received six HIV tests in the span of less than a month. Without insurance, the process would have cost more than $2,000; an RNA test alone is about $500. Had the recommendations of the Confirmatory Testing Laboratory, the Association of Public Health laboratories, and the Blood Products Advisory Committee been adopted more than a decade ago, I could have avoided the entire experience. My indeterminate test was the result of a lab error; nonviral Western Blots always are. For the foreseeable future, patients who receive nonviral Western Blots will be misdiagnosed. And like me, they will live, however briefly, with the belief that they have HIV.
I asked Bernard Branson, the associate director for HIV laboratory diagnostics at the CDC, whether he thought indeterminate results would ever disappear completely.
"No test is perfect," he told me, "not even RNA tests." The CDC is currently working to develop immunoassays that dramatically cut the rate of false positives, false negatives, and indeterminate results, but there may always be a few cases like mine, no matter how advanced the technology. It's hard to know how many people get a false positive result. A 2005 study suggested a false positive rate from an immunoassay plus Western Blot of 1 in 250,000, but a lot of false positives are unreported, and the problem has not been the subject of much research.
"Why," he asked me in return, "are you interested?"
I explained to him my own experiences, my week of anxiety, and my hope that others in my situation might not face the same difficulties I did. As I spoke, I felt some of the confusion, the fear, the trepidation of that awful week lurch back into my mind. I had received my final confirmatory test more than a month earlier, but the pall of HIV still hung over my life in an ineffable way. Even now, several months later, I wake up some mornings with the same sudden panic I felt the moment I got the call.
"How can something like this still happen?" I asked Branson. That, at heart, was the question that had been haunting me. I wanted an answer beyond biology, beyond virology, an explanation larger than lab mistakes and hasty diagnoses. I wanted that week of my life back, and if I couldn't get it, I wanted to know why I lost it in the first place—why a widely used, highly sensitive test could malfunction, be misinterpreted, and bring me to the brink of despair.
Branson considered my question for a moment.
"How can it happen?" he said. "Bad luck."
When I finished the interview, I scoured my room for all of the results I had received and tucked away during my month of HIV tests. Then I tore them up and threw them away.