The DNAinfo archives brought to you by WNYC.
Read the press release here.

Medical Examiner Uses Shoddy Science for DNA Matching, Legal Aid Alleges

By Janon Fisher | June 17, 2013 9:07am
 An analytical tool developed by Dr. Theresa Caragine (left) and Adele Mitchell for the Office of Chief Medical Examiner is under fire.
An analytical tool developed by Dr. Theresa Caragine (left) and Adele Mitchell for the Office of Chief Medical Examiner is under fire.
View Full Caption
Nicole Bengiveno/The New York Times/Redux

NEW YORK CITY — The city medical examiner's new method of singling out a suspect's DNA from the abundance of unrelated genetic material found at a crime scene is junk science and should be discontinued, the Legal Aid Society has charged.

The group has mounted a challenge to three criminal cases based on the Office of the Chief Medical Examiner's Forensic Statistical Tool, or FST — an algorithm developed in 2010 to allow technicians to determine the probability that a suspect's DNA is contained in a mixture of multiple people's DNA.

The Legal Aid Society is fighting to block prosecutors from using the technique, saying that it has never been fully vetted by the scientific community and has yielded at least one false positive DNA match, in a Brooklyn sexual assault case.

"I think the validition studies are incomplete, not thoroughly peer reviewed and have holes in them," said Alan Gardner, the supervising attorney in Legal Aid's DNA Unit. "It shouldn't be used."

The technique was developed by Dr. Adele Mitchell and former deputy lab director Dr. Theresa Caragine, who left the office recently after a scathing audit of her department found that a lack of organizational and leadership skills promoted conflict in the lab.

The FST algorithm compares both the suspect's DNA and DNA from a control group of 61 donors to the tangle of DNA found at crime scenes. When the suspect's DNA proves a better match for the crime scene DNA than the control group's DNA, prosecutors can use that match as evidence in court.

But Legal Aid lawyers claim that the control group does not reflect the diversity of New York's population. They found that 54 percent of the donors in the control group were white, while the city's overall population was just 44 percent white in 2010, according to the U.S. Census Bureau.

Legal Aid lawyers added that many of the crime scene samples were old and degraded, making comparisons more difficult and increasing the possibility of bad test results.

The OCME declined to comment on Legal Aid's challenge.

One of the three cases Legal Aid is challenging is the case of Andrew Peaks, 28, of Brooklyn, who was charged with two separate sexual assaults after the ME's office claimed they found his DNA on two articles of clothing from two attacks in 2010. They ran the tests after he was identified by two eyewitnesses.

After getting a cheek swab from the defendant, the lab determined that there was a "moderate" chance that his DNA was on a piece of clothing found at the scene.

But the first "match" to Peaks that the lab found using the FST method turned out to be faulty, prosecutors subsequently admitted in court papers.

Legal Aid lawyers Jessica Goldthwaite and Clint Hughes had contested the DNA match after discovering that this new method of calculating the probability of a DNA match had never been fully vetted in the scientific community.

“Dr. Caragine did not say that the OCME’s FST had been subjected to validation, or peer reviewed, by any other laboratory or scientific body,” Goldthwaite wrote in court papers.

During hearings questioning the science behind FST, Mitchell said under cross-examination that she performed validation studies testing her method, but she never documented her process or recorded data to back up her hypothesis.

The Medical Examiner's office also hasn't allowed outside forensic experts to examine their computer code.

“Dr. Caragine has asserted that the FST software program had been tested extensively within OCME, but acknowledged that it is proprietary software and has neither been tested nor validated by any other lab or scientific body," Goldthwaite stated in her motion.

Caragine did not respond to calls for comment.

Prior to the new FST technique, the ME's office only attempted to make a DNA match to a suspect when the evidence found at the crime scene clearly contained a single person's DNA. The new system was designed to broaden the ME's ability to make matches when the evidence was less clear-cut.

John Jay College of Criminal Justice Professor Lawrence Kobilinsky said that the ME's method resembles something called the "likelihood ratio" which assigns a number reflecting the probability of a DNA match. The technique is used in Great Britain, Italy and Spain, but he said the method favors prosecutors.

"It's a pro-prosecution statistic," he said. "Regardless of the number, the jury interprets it the same way. Jurors don't understand statistics, they don't really understand DNA, but when you say there's a match, they understand that."

FST has been used in more than 500 cases in New York City since 2010, but only made it to trial 25 times, a law enforcement source said.

Even if Legal Aid succeeds with the challenge, "it's not a get-out-of-jail-free card," said the source. "Without FST, we would go forward with cases but the jury would have less information to make a decision with."

Still, the challenge comes at a low point for the Medical Examiner's office, which has recently been plagued by scandal and high-level departures.

Allegedly shoddy work by lab tech Serrita Mitchell prompted a review of more than 800 rape cases in January. The director of quality assurance was fired over the incident and lab director Dr. Mechthild Prinz was forced to resign.

Problems in the forensic department have prompted two City Council bills to increase transparency and prevent similar future problems.

The first piece of legislation would require the OCME to give the score of all proficiency reports taken by the lab staff.

The council is also proposing a law that would require the ME to report any "significant event," including any "failure, lapse, error" that undermines the mission of the office.

"Clearly, the OCME is going through one heck of a hard time," Kobilinsky said.