A Virus as a Witness: The Schmidt Case & Novel Science

Every so often a case does more than decide one defendant’s fate. It sets the terms for a whole category of evidence. The prosecution of Lafayette, Louisiana physician Richard J. Schmidt was one of those cases. To prove that Schmidt had tried to kill his former girlfriend by injecting her with HIV-infected blood, prosecutors turned to a technique no American criminal jury had ever been allowed to consider: phylogenetic analysis, the science of reconstructing a genetic family tree. It worked. Schmidt was convicted, and the courts upheld the use of that evidence all the way up. For anyone trying to understand how a brand-new forensic method makes the leap from the laboratory to the jury box, Schmidt is the case to study.

What the State Had to Prove

The accusation was chilling in its specifics. Schmidt, a gastroenterologist, had been giving his former girlfriend (an intensive-care nurse) what she believed were routine vitamin injections. Prosecutors alleged that in the summer of 1994, after she moved to end their long affair, he gave her a shot that was not a vitamin at all, but blood drawn from one of his own HIV-positive patients. Months later she tested positive for HIV. Schmidt pleaded not guilty and said he had been home with his wife that night. The hard part for the prosecution was proof of source: how do you show a jury that the virus in the victim came from that particular patient and not from anywhere else in the world?

How You Put a Virus on the Stand

The answer came from the biology of HIV itself. The virus mutates quickly, so over time separate infections drift into genetically distinguishable variants. That means scientists can compare viral sequences from different people and measure how closely related they are which means, in effect, building a family tree, or phylogeny, of the virus. Investigators sequenced HIV from the victim, from Schmidt’s HIV-positive patient, from a set of other HIV-positive people in the Lafayette area, and against sequences in a national database. The result was striking: the victim’s virus was most closely related to, and in the language of the analysis “nested within,” the lineage of the patient’s virus. That pattern was consistent with the virus having traveled from the patient to the victim. One detail from the State’s own published account is worth keeping, because it is the kind of thing a cross-examination is built on: the two genes analyzed did not fully agree. The reverse transcriptase gene showed the nesting pattern the State relied on. The envelope gene showed the opposite pattern, weakly, with statistical support near fifty percent. It is worth being precise about what this evidence was and was not. It did not identify a person the way a standard DNA profile does. It reconstructed the genealogy of a virus and offered it as circumstantial proof of a route of transmission.

Two Texas Scientists at the Center

The science that anchored the State’s case came largely out of Texas. Dr. Michael Metzker, then at Baylor College of Medicine in Houston, carried out the sequencing and analysis and testified to it. Dr. David Hillis, a leading phylogenetics expert and director of the School of Biological Sciences at the University of Texas at Austin, testified as well and vouched for both the method and the result. When the defense raised the possibility that the samples had been contaminated, the State answered with more science rather than argument: it commissioned a second, independent sequencing by an outside expert, and Hillis confirmed that the new results matched the first and showed no evidence of contamination. The team’s peer-reviewed account of the work was later published in the Proceedings of the National Academy of Sciences.

The Admissibility Fight and Why the Evidence Got In

Schmidt’s lawyers fought hard to keep the phylogenetic evidence away from the jury. They argued it was far more uncertain than ordinary DNA fingerprinting, which courts had long accepted, and that the controls and lab work in this case were flawed. They lost, and the reasons they lost are the lesson. Louisiana, like the federal courts, screens novel scientific evidence under the Daubert standard, which asks whether a technique has been empirically tested, whether it has been through peer review and publication, whether its rate of error can be assessed, and whether it is generally accepted in the relevant scientific community. Phylogenetic analysis cleared every one of those touchstones: it is a mainstream method in evolutionary biology, heavily published, testable, and honest about its uncertainties rather than hiding them. It also helped that the State narrowed what it was offering. It told the court it would use the testimony only to establish that the two samples were closely related, not to prove that the virus had actually passed from one to the other. The trial court admitted the evidence after a pretrial hearing at which five experts testified, three for the State and two for the defense. The Louisiana Third Circuit Court of Appeal affirmed the conviction and the evidence in 2000, the Louisiana Supreme Court let that stand, and in 2002 the United States Supreme Court declined to take the case thus ending the appeals and cementing Schmidt as the first American criminal prosecution to use phylogenetic evidence.

Novel Is Not the Same as Junk

It is tempting to sort every unfamiliar forensic technique into one of two bins, proven or junk. Schmidt shows why that is too simple. Phylogenetic analysis was novel to the courtroom, but it was neither new nor shaky as science. It rested on decades of peer-reviewed work and well-understood methods. That is precisely why it came in. The gatekeeping question is not “have we seen this in a courtroom before.” It is whether the method is genuinely reliable and whether it was properly applied in this case. Watch how both halves surfaced in Schmidt. The reliability of phylogenetics as a discipline answered the first half. And when the defense attacked the second half, the lab work and the risk of contamination, the State did not simply argue the point; it ran an independent re-analysis. What the court did with that fight is worth knowing, though, because it complicates the tidy version of this story. It held that whether the protocols had been properly followed was a question for the jury rather than a gatekeeping question for the judge, and much of the defense’s evidence on the point was kept out of the hearing on that basis. The re-analysis happened. The gate did not test it. That is the discipline the gate is supposed to impose.

What Came After Schmidt and Its Limits

Schmidt opened the door, but American courts have walked through it carefully. Phylogenetic analysis has been used as criminal evidence in the United States only a handful of times since including a 2004 prosecution in Washington State and the 2009 prosecution of Philippe Padieu in Collin County, both involving allegations that a single source infected multiple partners. Padieu was convicted on six counts of aggravated assault with a deadly weapon and sentenced to forty-five years on five of them. In each case the strain identified as the source traced back to the accused. Tellingly, though, the convictions did not rest on the viral family tree alone. In the Texas case, prosecutors backed the phylogenetics with old-fashioned investigation, including contact tracing and HIV testing of most of the complainants’ former partners. The genetic evidence was a piece of the case, not the whole of it, which is exactly how this kind of proof is meant to be used.

As the technique spread, the scientific community sharpened its view of what phylogenetics can and cannot honestly claim, and the refinement matters enormously for the defense. The touchstone paper, published in 2007 by an international group of phylogenetics and HIV experts, set out cautions that still govern the field. Its central conclusion is blunt: phylogenetic analysis cannot prove that HIV passed directly from one specific person to another. What it can do is show that two strains are closely related or exclude a defendant altogether by showing the strains are unrelated. That asymmetry is the heart of it because the method is far more powerful as a tool of exoneration than of proof.

The institutions have since caught up to that paper. UNAIDS took the position in 2013 that phylogenetic evidence alone is not sufficient to establish, to the criminal standard, that one person infected another. In 2018 twenty scientists led by Françoise Barré-Sinoussi, who shared the Nobel for discovering HIV, published a consensus statement endorsed by the International AIDS Society and by UNAIDS reaching the same conclusion: the analysis alone cannot prove beyond reasonable doubt that one person infected another, although it can exonerate. That is not a fringe view of this technique. It is the mainstream one.

Two limits do most of the work. First, relatedness is not direction. Even when two people’s strains cluster together, the virus may have reached both of them from an unsampled third person who was never tested. A tree that looks like “A infected B” can be equally consistent with “A and B were both infected by C.” When the scientists from Schmidt’s case later claimed, in a journal article analyzing the newer prosecutions, that their analysis could reliably establish the direction of transmission, other experts who had served as court advisors called that conclusion unwarranted, and more recent work has put numbers on it. Tested against 112 transmission pairs where the true direction was independently known, the accuracy of direction inference ran from thirty-two percent to ninety-three percent depending on the shape of the tree. Thirty-two percent is worse than a coin flip. The authors concluded that these approaches are unsuitable for individual-level analysis such as forensic investigations. Second, the controls decide the outcome. A phylogenetic result is only as trustworthy as the comparison sequences it is measured against, and controls that are not epidemiologically and temporally appropriate can make an ordinary degree of relatedness look strikingly unique. That is not hypothetical. It is the precise vulnerability Schmidt’s own defense tried to exploit, and it is still the first place a competent challenge should look.

None of this means the evidence should never reach a jury. It means the honest version is narrower than jurors might assume from the words “DNA evidence,” and the gap between what the science can support and what a prosecutor might argue is where the defense works. A phylogenetic tree that is consistent with transmission is not the same as proof of transmission and a good cross-examination makes a jury feel that difference.

The Same Gate, in Texas

For a Texas audience the two frames look alike, and they part company at exactly the place this case turned. Louisiana and the federal courts apply Daubert. In Texas criminal court the test is Kelly, which asks three things: is the underlying scientific theory valid, is the technique valid, and was the technique properly applied on this occasion.

Read that third prong against what the Louisiana court actually did. When Schmidt’s lawyers tried to show that the protocols had not been followed, the court of appeal held that whether those protocols were properly applied was a question for the trier of fact and not a gatekeeping function of the trial court. Kelly points the other way. All three prongs have to be proven to the judge, outside the presence of the jury, by clear and convincing evidence. The application fight that Louisiana sent to the jury is a fight a Texas judge has to hear first.

Which makes the next part worth saying plainly. No Texas appellate court has ever decided the question. A Texas trial court admitted this evidence in Collin County in 2009. Padieu’s appeal raised ineffective assistance of counsel and was resolved on a silent record, so it never reached the science. The evidence has been in front of a Texas jury, and the standard that would govern the next one has never been applied to it. The defense’s leverage is greatest at that threshold hearing, not after a jury has been told what “the science” shows.

Why It Still Matters

Schmidt was decided a generation ago, but it reads like a preview of the present. The court that admitted a viral family tree in the 1990s opened a door that molecular and sequence-based forensics have been walking through ever since. The newest arrival at that same gate is DNA age estimation from methylation (the “epigenetic clock” we cover in a companion post) which, like phylogenetics before it, is powerful, genuinely scientific, and not yet fully settled as courtroom proof. As one HIV-strain expert remarked when Schmidt was decided, “the horse is out of the barn.” New forensic tools will keep arriving. The question is never whether they are impressive. It is whether they are reliable, whether they were properly applied, and whether they were tested at the gate before a jury heard a word.

Facing an Unfamiliar Kind of Forensic Evidence?

From viral genetics to DNA age estimation to breath and blood testing, new forensic methods reach the courtroom faster than the law that governs them. The place to challenge a novel technique is at the reliability gate before a jury hears a conclusion. Deandra Grant Law brings genuine forensic training to that fight, testing the science on its own terms.

Call (214) 225-7117 • texasdwisite.com

Sources & Further Reading

• Science (AAAS), reporting on the Schmidt prosecution and the first courtroom use of phylogenetic analysis (1998). science.org
• Metzker, Mindell, Liu, Ptak, Gibbs & Hillis, “Molecular Evidence of HIV-1 Transmission in a Criminal Case,” Proceedings of the National Academy of Sciences 99(22):14292 (2002). ncbi.nlm.nih.gov
• State v. Schmidt, 699 So. 2d 448 (La. App. 3 Cir. 1997) (pretrial writ application; this is where the Daubert analysis lives, and the 2000 opinion defers to it), and 771 So. 2d 131 (La. App. 3 Cir. 2000) (affirming the conviction). caselaw.findlaw.com
• Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993); in Texas criminal cases, Kelly v. State, 824 S.W.2d 568 (Tex. Crim. App. 1992).
• Bernard et al., “HIV forensics: pitfalls and acceptable standards in the use of phylogenetic analysis as evidence in criminal investigations of HIV transmission,” HIV Medicine (2007). Phylogenetics cannot prove direct transmission, can exonerate, and depends heavily on appropriate controls. pubmed.ncbi.nlm.nih.gov
• aidsmap, “Claims that phylogenetic analysis can prove direction of transmission are unfounded, say experts” (2010). Experts respond to the later Metzker/Hillis analysis of the Washington (2004) and Texas (2009) cases. aidsmap.com
• “Using phylogenetics to infer HIV-1 transmission direction” (PNAS 2022). Tested against 112 pairs with known direction; concludes direction inference is unsuitable for individual-level (forensic) analysis. ncbi.nlm.nih.gov
• Related, from our blog: DNA methylation “age clocks” and the Kelly standard, the newest novel-forensics fight, and how Kelly differs from Daubert. texasdwisite.com