Just off Barbara’s kitchen is the room that her husband, a gastroenterologist named Richard Schmidt, used as his study before he went to prison. For her, this is where the story begins. On a hot July day in 1995, she came home to find several plainclothes police officers rummaging through Richard’s papers. She asked what they were looking for. “Evidence of B12,” she remembers them saying, with no further explanation. She called her husband at work to find out what was going on. The investigators had showed up there too. “He seemed clueless,” she recalls.
It didn’t take long for all the sordid details to emerge. The police were investigating a complaint by Janice Trahan, a nurse who had worked at Schmidt’s practice and had a decade-long affair with the doctor. As Trahan later testified in court, Schmidt had entered her house on the night of August 4, 1994, soon after she’d ended their relationship. She and her 3-year-old son were asleep in bed. Schmidt had woken her up, saying he wanted to give her a vitamin B12 shot, something he’d occasionally done during their relationship when she wasn’t feeling well. Trahan protested. Schmidt insisted. He pushed the needle into the muscle of her upper arm, depressed the plunger, and quickly left the house. Several months later, Trahan started feeling ill and was eventually diagnosed with HIV. She believed that Schmidt, in a fit of jealousy, had injected her with tainted blood from one of his patients.
Barbara has a different recollection of August 4, 1994, one she described to both the police and later the jury at her husband’s trial. The family had recently returned from a trip to Florida, she says, and Richard had injured his back lifting heavy luggage off the carousel. A record in her checkbook reminded her that she’d gone out to buy new knobs for the kitchen cabinets that day, which she had to show him as he lay in bed. “He wasn’t crazy about them,” she says. After that, Barbara maintains, she showered and they both went to sleep. (The Louisiana Department of Corrections declined to make Richard Schmidt available for an in-person interview.)
When the police searched Schmidt’s office, however, they found something strange. Buried beneath a mound of old files was the notebook in which his nurses had recorded the previous summer’s blood draws and other lab work. Curiously, the notebook was only half full. The final entry showed that an HIV-positive patient had come in for an appointment on August 4. He had been scheduled for a blood draw, but the notebook didn’t confirm whether it actually took place. The police theorized that Schmidt had taken his patient’s blood, with or without the help of a conspiratorial nurse, and kept the vial hidden so he could use it on his former lover. Still, the evidence was circumstantial. One of Schmidt’s nurses, in her testimony, described the notebook as “just a jot sheet,” not an official record. The prosecuting attorney, Keith Stutes, needed harder proof.
At the time, the US justice system was in the midst of a forensics revolution. Just over a decade earlier, in 1983, scientists had invented a technique known as polymerase chain reaction, or PCR, which made it possible to copy DNA from tiny samples of blood or tissue in large enough quantities to study. Within a few years, investigators in England had used DNA profiling to prove that a teenager who’d confessed to two rape-murders had not committed the crimes; they also used it to identify the man who did. The first US case came a year later. Soon enough, DNA testing was everywhere you looked, according to attorney Barry Scheck, who in 1992 cofounded the Innocence Project, which exonerates wrongfully convicted people using DNA. One judge called it “the single greatest advance in the search for truth since the advent of cross-examination.”
For Stutes, then, it made sense to turn to genetics. “We were conscious that DNA could make comparisons of body fluids,” he says. “The question was, is there a science that will be able to compare the viruses?” The criminologists his office usually worked with didn’t have an answer, but they recommended that he contact Michael Metzker, a graduate student at Baylor College of Medicine, who was developing technology for studying the transmission of HIV as part of his thesis work. As Metzker recalls the conversation, Stutes asked him “whether two strains from different individuals could be matched” using some sort of test. Not that he knew of, Metzker told Stutes. But there was a technique that could parse the relationship between two different viruses. It was called phylogenetics.
Just as each living person occupies a branch on a family tree extending generations back in time, so does each living organism. These trees, called phylogenies, tell us that humans once shared an ancestor with mushrooms, giraffes with beetles, elephants with elephant shrews. Modern phylogenetics got its start soon after Darwin published On the Origin of Species, but its potential wasn’t fulfilled until the 1990s, when biologists gained the ability to sequence DNA cheaply and run the long strands of genetic code through complex computer algorithms. With these tools, they could re-create the evolutionary history of any organism on earth. And they weren’t limited to the broad strokes of change that occur over eons or millennia. They could also track the person-to-person movements of modern pathogens, tracing an outbreak of swine flu to the town where it originated, uncovering the trade routes that brought hepatitis C to the United States, and probing new hot spots of HIV infection.
Indeed, HIV is a good fit for the technique. Its genetic material is made from RNA, which mutates much faster than DNA, producing new strains at a furious pace. Each of the 30 million or so persons currently living with HIV harbors a slightly different variant. Even the tiny virus particles circulating in an individual’s bloodstream differ, if ever so slightly, one to the next. But within a given geographic area, each risk group—intravenous drug users, men who have sex with men—tends to carry a similar strain. In the early 1990s, a few years before Stutes approached Metzker, a phylogenetic analysis led by the Centers for Disease Control and Prevention suggested that a dentist in Florida had infected six of his patients with HIV. They constituted their own miniature risk group—in phylogenetic terms, their own small branch of the giant tree.
Metzker planned to mimic the CDC’s approach in the Florida case. If he found that Trahan’s virus closely resembled that of the HIV-positive patient in Schmidt’s notebook, her accusation would be supported. If, on the other hand, the connection was distant, then her account would seem weaker. The police sent Metzker blood samples from Trahan, Schmidt’s patient, and about 30 other HIV-positive people from the local area; this last group would serve as a scientific control, allowing Metzker to gauge the relative similarity of all the strains.
HIV hides out in its host’s own genome, switching from RNA to DNA after it successfully infects a cell. Metzker isolated some of these cells from each sample, extracted their DNA, and used PCR to create huge quantities of copies. Then he sequenced the DNA and created a tree. The first two strains, Metzker concluded, were more similar to each other than to any of those in the control group. Some of the sequences were identical, with portions of the patient’s HIV DNA showing up in Trahan—a phenomenon called nesting. It was as though her virus had lifted a verse from his genetic song.
The results gave Stutes the evidence he’d been looking for. On July 23, 1996, a year after the police searched Schmidt’s home, the doctor was arrested on charges of attempted murder. The case immediately became what one local headline referred to as a “media magnet.” Schmidt’s eldest daughter was in the family’s backyard pool when she heard the news; her brother’s girlfriend called to say she’d seen their father on TV. (Now a 39-year-old lawyer in Lafayette, the daughter still hesitates to tell people her maiden name.) Producers from 20/20, Dateline, and The Montel Williams Show jockeyed for interviews. “The newspaper articles all tainted Richard in a negative way,” says Schmidt’s Bronx-born defense attorney, Mike Fawer. And Schmidt, with his helmet-like black toupee, thick brows, and heavy-lidded eyes, fit the part. “They made him look so sinister,” Barbara says.
Schmidt’s case marked the first time that a US criminal court would consider phylogenetic evidence. At his pretrial hearing, which began on December 6, the defense and the prosecution argued over whether Metzker’s analysis was admissible—in part, whether it met the guidelines established by the US Supreme Court in a 1993 case called Daubert v. Merrell Dow Pharmaceuticals. In Daubert, the court advised that a forensic test should be presented to a jury only if the methods on which it was based had been published in a peer-reviewed journal and gained “widespread acceptance within a relevant scientific community.” In addition, the court said, the test should have a known error rate.
After several days of testimony, the judge, Durwood Conque, ruled that Metzker’s analysis met the Daubert criteria. He stipulated, however, that the phylogenetic evidence could be used only to show that two HIV sequences were related. It could never establish the “likelihood or suggestion of direct transmission,” he said.
To the defense team, even this qualified acceptance seemed improper. “I thought we were going to win on the Daubert hearing,” Fawer says. “But we got the back of the hand.” Phylogenetics was far from a proven forensics tool. Its only prior use in a criminal court had been to investigate a rape conviction in Sweden. And while a description of the Florida dentist study had been published in Science, a peer-reviewed journal, in 1992, that case involved civil charges, not criminal ones, and never went to trial. That left little basis for the scientific community to accept the method or discern its susceptibility to error. In fact, Metzker believes that one of the reasons Stutes hired him, an inexperienced graduate student, was that more seasoned scientists did not want to be involved with what they considered a controversial practice.
Schmidt’s trial began in mid-October 1998. It was held in the courthouse in downtown Lafayette, a huge block of concrete rising from a manicured lawn. The 12 jurors heard Janice Trahan, now known as Janice Allen, describe her tumultuous relationship with Schmidt—his “obsessed, jealous, controlling, volatile” behavior, the “extremely painful” shot he’d forced upon her. They heard Jim Craft, the lead detective investigating Trahan’s accusation, discuss the memory recall techniques he’d given her to help her remember the date of the alleged injection. They heard Barbara Schmidt’s conflicting account of that date. The doctor himself never took the stand, never discussed the odd notebook. “It surely was not clearly explained,” Fawer, the defense attorney, acknowledges.
Finally, it was time to present the forensic evidence. Five expert witnesses eventually took the stand. The transcripts of their testimonies read like a molecular biology textbook: There are complex discussions of statistical significance, “bootstrap values,” “branch lengths,” and “gene loci.” Even the more accessible arguments are delivered in all but impenetrable language. A defense witness named Bette Korber, one of the country’s top phylogeneticists, gave a lengthy and complicated testimony that included cases from Thailand, Amsterdam, and a large US database. Eventually she offered her succinct take on the HIV sequences from Trahan and Schmidt’s patient: “I would say that they are probably unrelated.” But, in the transcript, this insight comes toward the end of 50-plus pages of technical science.
One academic stood out from the rest. David Hillis, an evolutionary biologist at the University of Texas at Austin, had advised the CDC on phylogenetics and edited Systematic Biology, one of the field’s most prestigious journals. “I was very impressed with him,” says Stutes, who calls him “a powerful witness.” At the prosecution’s request, Hillis had performed his own analysis of the HIV samples, attempting to replicate Metzker’s findings. When he took the stand, Stutes asked him what he had discovered.
Hillis spoke plainly. The HIV sequences in Trahan and Schmidt’s patient were, he told the jury, as closely related as two sequences isolated from different individuals could be. He was unflappable in the face of Fawer’s grilling. Toward the end of the cross-examination, Hillis testified that “there’s less than a one in a million chance you’d see the association that you did at random.” The prosecutor liked this line so much that he used it in his closing arguments. Twenty years later, Barbara is still haunted by it. She believes that those words—“one in a million”—more than any others, stayed with the jury, possibly because it was the only phrase they could easily understand. Trahan remembers it too.
On October 23, 1998, the jury found Schmidt guilty of attempted second-degree murder. (Ten of the 12 jurors agreed; until last November, a Jim Crow–era law in Louisiana allowed for convictions that were not unanimous.) Four months later, he was sentenced to 50 years in prison.
The process of moving science from the cloistered realm of the laboratory to the messy environment of the courtroom is inevitably fraught. Although criminal law readily embraced DNA forensics at first, debate arose quickly. In 1989, Barry Scheck questioned the legitimacy of a test linking blood from a murder victim to blood on his client’s watch. That first challenge marked the start of a battle. “People talked about it being the DNA Wars,” says Simon Ford, a research scientist at Forensic Bioinformatics, a firm that seeks to “cast light into the black box of forensic analyses.” By 1992, the same year the Innocence Project began, the tension had spurred the National Research Council to draw up formalized guidelines. Among other things, those guidelines advised experts to stop saying that DNA testing was infallible.
Forensic phylogenetics never got its own NRC report. In part that may be because, unlike DNA profiling, it is rarely invoked in court. The technique is used only in cases of criminalized HIV or hepatitis C transmission, and not all attorneys seek it. There are no designated crime labs, no special training protocols for witnesses. The only existing recommendations are in piecemeal publications by small groups of academics. Amid this sparse landscape, Schmidt’s case stands out. Anne-Mieke Vandamme, a Belgian biochemist and leading expert in forensic phylogenetics, calls it a “textbook case”—proof that the technique works. But she and others also see it as a warning sign, an example of what can happen when complex, unfamiliar science is brought before juries.
At the time of Schmidt’s trial, defense witnesses raised several possible issues with Metzker’s study. For one thing, his lab notes referred to Trahan and Schmidt’s HIV-positive patient by their first names, which meant he knew which sequence belonged to whom, potentially skewing the results. What’s more, the defense witnesses argued, Metzker didn’t take the proper precautions with his PCR analysis. DNA fragments have a way of flying around the laboratory once the PCR test tubes are opened. They “can fall on pipettes, gloves, bench tops, into other tubes,” says microbiologist James Mullins, who worked on the Florida dentist case and testified at Schmidt’s pretrial hearing. Once that happens, tracking the fragments back to their sources becomes about as easy as figuring out which log the ashes in your firepit came from. DNA from different sequences can easily become mixed. That’s why “you shouldn’t ever work on both subjects in the same laboratory,” Mullins adds. Metzker didn’t follow this protocol either.
Today, he acknowledges both oversights. “We should have blinded it,” Metzker says, using the technical term for when scientists mask information about their test subjects to avoid bias. Yet he notes that Hillis’ follow-up study, which was blinded and used newly drawn blood samples, came to the same conclusions. Hillis observed the right PCR protocols too. “The truth of the matter is we’ve done our due diligence,” says biologist David Mindell, who assisted Hillis with the work.
But not all the questions hovering around the results were addressed by the second study. The defense also voiced doubts about a process called contact tracing, which seeks to uncover all possible transmission routes for an infectious disease. Trahan had volunteered the names of several men she’d slept with, all of whom the police confirmed did not have HIV. But that inquiry, the defense argued, wasn’t equivalent to a rigorous probe by an expert epidemiologist. Trahan’s work as a nurse in an intensive care unit, for instance, could have exposed her to the virus. Although Vandamme cautions that she isn’t familiar with all the specifics of Schmidt’s case, she says she’d likely have zeroed in on this weakness if she’d been asked to testify.
At the pretrial hearing, Mullins also contended that there were problems with the control group, the HIV-positive patients from around Lafayette. The ideal forensic phylogenetic study, he says, draws its controls from the same risk group as the accuser—in this case, Trahan—and from people infected around the same time. But in Schmidt’s trial, nearly two-thirds of the samples came from gay men, some of whom had tested positive up to 10 years earlier. Had they come from a wider pool, Mullins says, Metzker might have found strains even more closely related to Trahan’s HIV than the one circulating in Schmidt’s patient, who was a gay man. That wouldn’t necessarily have absolved Schmidt, but it might have broadened the jury’s perspective.
Hillis balks at this. For him, the fact that Trahan’s sequences were embedded, or nested, within the patient’s sequences is indisputable evidence of their relatedness. “This could only have happened if the physician’s patient was the source of the infection in the sample from the victim,” he says. An expanded control group, he adds, would not have made a difference.
John Huelsenbeck, a phylogeneticist at UC Berkeley, is unwavering in his belief that the jury in Schmidt’s case “made the right decision.” Like Hillis, he sees the nesting as decisive. “That’s all that matters,” he wrote in an email. Vandamme holds a similar view. In general, she says, phylogenetics can only uncover whether there is a close relationship between two strains, not which one came first—just as Judge Conque advised the jury during the pretrial hearing. But the nesting phenomenon transcends this restriction, allowing for the conclusion that Schmidt’s patient’s virus was older than Trahan’s.
Yet even that seemingly pivotal piece of evidence goes only so far. Vandamme emphasizes that the nesting doesn’t eliminate the possibility of an intermediary between Schmidt’s patient and Trahan. “You can never prove that one infected the other,” she says. In the early days of forensic phylogenetics, she adds, the few juries that encountered it sometimes got the mistaken impression that an analysis showing a connection was proof of guilt. But the science can’t ever show that. “You can do the best possible phylogenetic analysis,” she says. “It’s still not going to prove transmission.”
It’s not uncommon for questions to linger after a criminal conviction, particularly when the charges are as serious as attempted murder. You can believe in Schmidt’s guilt and still see problems with how the scientific evidence in his case was handled. But beyond the complex debates about PCR methodology, contact tracing, and all the rest, Vandamme says her biggest concern about forensic phylogenetics may be that academics don’t always “have sufficient appreciation for how what they say is interpreted by the jury.” Mullins, the defense witness, maintains that Hillis crossed a boundary with his “one in a million” comment, using persuasive language that may have caused the jury to lend phylogenetics more weight than Judge Conque intended—or than the science warranted. “He really stretched credulity,” Mullins says. “It was an overreach.” One juror, who asked to remain anonymous, says, “I think the DNA was probably the major reason, and the other evidence just supported that.”
Schmidt appealed his case multiple times, mostly on the basis of allegedly flawed phylogenetic evidence. In 2000 the Louisiana Court of Appeals denied his request, finding that the circumstantial evidence was enough to eliminate reasonable doubt even without the science. Two years later, the US Supreme Court declined to hear his case, offering no comment for its decision. The Louisiana Supreme Court denied another appeal in 2005. Schmidt and his attorneys continue to fight the verdict, though they know they are unlikely to succeed. He will be eligible for early release in 2023.
Among the clippings in the box under Barbara Schmidt’s bed is a single photograph of her and her husband. They look like they’re headed for a night out, he in a shirt and tie with his arm around his equally well-dressed wife. But in fact they’re leaving the courthouse after a day of pretrial testimony—the fateful first chapter in a textbook case.
John Rowland/Lafayette Daily Advertiser via USA Today Network (Schmidt); Lafayette Daily Advertiser File via USA Today Network (newspaper); Getty Images (red texture)
This article appears in the April issue. Subscribe now.
Listen to this story, and other WIRED features, on the Audm app.
Let us know what you think about this article. Submit a letter to the editor at email@example.com.
More Great WIRED Stories
- A more humane livestock industry, thanks to Crispr
- Coders’ primal urge to kill inefficiency—everywhere
- For gig workers, client interactions can get … weird
- For avalanche safety, data is as important as proper gear
- How hackers pulled off a $20 million Mexican bank heist
- 👀 Looking for the latest gadgets? Check out our latest buying guides and best deals all year round
- 📩 Get even more of our inside scoops with our weekly Backchannel newsletter