Post by CCADP on Aug 28, 2005 7:28:33 GMT -5
Human error calls validity of DNA testing into question
Discipline is as reliable as ever, but workers prone to mistakes in collecting, analyzing samples
- Tom Jackman, Washington Post
Sunday, August 28, 2005
Washington -- Washington hairdresser Andre Chreky gladly agreed to a DNA test when a former employee hit him with a paternity suit.
The claim was absurd, Chreky said he remembers thinking. He had stopped dating the woman years before she gave birth to the boy, now a teenager. This would all be over soon. DNA doesn't lie.
The results were back in a month, on a two-page report from Laboratory Corp. of America, or LabCorp, one of the largest paternity testers in the country and the state of Virginia's exclusive contractor: "The probability of paternity is 99.99 percent."
"It's crazy," Chreky, 50, who lives with his wife and two children in Great Falls, Va., recalled saying. "We need to take this to battle."
The fight lasted two years. When it ended in May, Fairfax County Circuit Court Judge David Stitt not only ruled in Chreky's favor but also raised serious questions about the reliability of DNA testing during a time when it is relied on to prove paternity, guilt, innocence and more.
"I thought LabCorp's performance was shoddy," Stitt said at a hearing in May after ruling that the state did not prove Chreky was the father. "I think something unfair happened in this case, where a citizen was put to the greatest extent to defend himself against what really has turned out to be a moving target as far as where LabCorp is concerned. ... I'm concerned about what level of oversight is being exercised by the commonwealth of LabCorp's work."
The state is not appealing Stitt's ruling.
LabCorp handles more than 100,000 DNA paternity tests for many public and private clients every year. But evidence at Chreky's trial showed that the company has only five people reviewing the data and making paternity determinations -- with one supervisor testifying that he issues an average of 1 paternity report every 4 minutes during a 10-hour shift.
DNA experts say Chreky's case underscores a growing problem in the burgeoning field of DNA testing: People make mistakes, and people collect the DNA samples and perform the analysis. So, they say, although DNA is as reliable as ever as a definitive science, the people reading and analyzing that science are imperfect. The volume of DNA testing also keeps rising.
The ruling in Chreky's case came as Virginia Gov. Mark Warner ordered a review of DNA testing at the state's criminal forensic lab after an audit detected human error in an analysis of a death row inmate's case.
Laurence Mueller, an evolutionary biology professor at UC Irvine who has been tracking lab errors in DNA cases for years, said DNA labs "use techniques that have been automated, like Hostess Twinkies on an assembly line. Most of the time, the Twinkies are fine. But once in a while, you see a bad one."
The bad ones, some biologists say, are coming more frequently.
This month, Illinois fired its DNA lab, Fairfax County-based Bode Technology, for failing to detect semen in 11 out of 51 rape cases. State police said the errors had not wrongly freed or convicted anyone, but they said they would have to reanalyze evidence in 1,200 rape cases.
At a July murder trial in Michigan, prosecutors acknowledged that a DNA test on evidence from 1969 matched someone who would have been 4 years old at the time of the slaying and couldn't possibly have been involved. Additional tests led to a second man, who was convicted.
In Las Vegas in 2001, a man spent a year in jail after being wrongly accused of committing two sexual assaults in the 1990s. Investigators later found that his DNA sample had been switched with another inmate's.
Human error "has always existed in all of the forensic sciences," said William Shields, a professor at the State University of New York in Syracuse who has testified in numerous DNA cases. "It exists in all the sciences."
Brad Smith, a LabCorp spokesman, said criticism from the judge in Chreky's case appeared to be the result of "some good lawyering on the challenge side."
"We are confident that we reported the correct results and that we followed appropriate procedures and good science," he said.
Smith added that he had worked in the identity and paternity testing field since 1982 and that "we've never had a result like this and/or a (judge's) statement like this."
Nathaniel Young, director of the Virginia Division of Child Support Enforcement, which pursued Chreky's paternity, said in a statement that he could not comment on the case, but he said procedures are under review.
LabCorp has performed Virginia's paternity testing since 2001 and charges the state $39.50 per test, or about $120 per case. State statistics show LabCorp was paid $797,000 last year and did almost 20,000 tests.
Stitt found LabCorp's "99.99 percent" report "not statistically valid" and ruled that the state had "failed to prove by clear and convincing evidence" the case against Chreky.
How could a judge discard a seemingly definitive DNA report? Experts said it was virtually unheard of.
But Mueller pointed to a number of incidents of lab error in recent years, including allegations of problems with crime labs in Houston and Richmond, Va. Crime labs in Philadelphia and Minnesota were later discovered to have sent out "false matches."
"It's a terribly important issue," Mueller said. "People involved in doing these techniques make mistakes that are not involved with technology. ... Until you get humans out of the system, these things can happen."
Chreky is no scientist. He said he just knew that this was something he needed to fight. Most people don't have the means to contest a "99.99 percent" finding. His wife, Serena, said the couple spent more than $200,000 to fight the case.
Chreky said he spent much of the past three years overwhelmed with anxiety about the case. "I've been getting up at 3:30, sleeping a couple of hours a night," he said. "I tried to keep busy. You don't want to think about it."
Page A - 3
URL: sfgate.com/cgi-bin/article.cgi?file=/c/a/2005/08/28/MNGS5ECGCU1.DTL
Discipline is as reliable as ever, but workers prone to mistakes in collecting, analyzing samples
- Tom Jackman, Washington Post
Sunday, August 28, 2005
Washington -- Washington hairdresser Andre Chreky gladly agreed to a DNA test when a former employee hit him with a paternity suit.
The claim was absurd, Chreky said he remembers thinking. He had stopped dating the woman years before she gave birth to the boy, now a teenager. This would all be over soon. DNA doesn't lie.
The results were back in a month, on a two-page report from Laboratory Corp. of America, or LabCorp, one of the largest paternity testers in the country and the state of Virginia's exclusive contractor: "The probability of paternity is 99.99 percent."
"It's crazy," Chreky, 50, who lives with his wife and two children in Great Falls, Va., recalled saying. "We need to take this to battle."
The fight lasted two years. When it ended in May, Fairfax County Circuit Court Judge David Stitt not only ruled in Chreky's favor but also raised serious questions about the reliability of DNA testing during a time when it is relied on to prove paternity, guilt, innocence and more.
"I thought LabCorp's performance was shoddy," Stitt said at a hearing in May after ruling that the state did not prove Chreky was the father. "I think something unfair happened in this case, where a citizen was put to the greatest extent to defend himself against what really has turned out to be a moving target as far as where LabCorp is concerned. ... I'm concerned about what level of oversight is being exercised by the commonwealth of LabCorp's work."
The state is not appealing Stitt's ruling.
LabCorp handles more than 100,000 DNA paternity tests for many public and private clients every year. But evidence at Chreky's trial showed that the company has only five people reviewing the data and making paternity determinations -- with one supervisor testifying that he issues an average of 1 paternity report every 4 minutes during a 10-hour shift.
DNA experts say Chreky's case underscores a growing problem in the burgeoning field of DNA testing: People make mistakes, and people collect the DNA samples and perform the analysis. So, they say, although DNA is as reliable as ever as a definitive science, the people reading and analyzing that science are imperfect. The volume of DNA testing also keeps rising.
The ruling in Chreky's case came as Virginia Gov. Mark Warner ordered a review of DNA testing at the state's criminal forensic lab after an audit detected human error in an analysis of a death row inmate's case.
Laurence Mueller, an evolutionary biology professor at UC Irvine who has been tracking lab errors in DNA cases for years, said DNA labs "use techniques that have been automated, like Hostess Twinkies on an assembly line. Most of the time, the Twinkies are fine. But once in a while, you see a bad one."
The bad ones, some biologists say, are coming more frequently.
This month, Illinois fired its DNA lab, Fairfax County-based Bode Technology, for failing to detect semen in 11 out of 51 rape cases. State police said the errors had not wrongly freed or convicted anyone, but they said they would have to reanalyze evidence in 1,200 rape cases.
At a July murder trial in Michigan, prosecutors acknowledged that a DNA test on evidence from 1969 matched someone who would have been 4 years old at the time of the slaying and couldn't possibly have been involved. Additional tests led to a second man, who was convicted.
In Las Vegas in 2001, a man spent a year in jail after being wrongly accused of committing two sexual assaults in the 1990s. Investigators later found that his DNA sample had been switched with another inmate's.
Human error "has always existed in all of the forensic sciences," said William Shields, a professor at the State University of New York in Syracuse who has testified in numerous DNA cases. "It exists in all the sciences."
Brad Smith, a LabCorp spokesman, said criticism from the judge in Chreky's case appeared to be the result of "some good lawyering on the challenge side."
"We are confident that we reported the correct results and that we followed appropriate procedures and good science," he said.
Smith added that he had worked in the identity and paternity testing field since 1982 and that "we've never had a result like this and/or a (judge's) statement like this."
Nathaniel Young, director of the Virginia Division of Child Support Enforcement, which pursued Chreky's paternity, said in a statement that he could not comment on the case, but he said procedures are under review.
LabCorp has performed Virginia's paternity testing since 2001 and charges the state $39.50 per test, or about $120 per case. State statistics show LabCorp was paid $797,000 last year and did almost 20,000 tests.
Stitt found LabCorp's "99.99 percent" report "not statistically valid" and ruled that the state had "failed to prove by clear and convincing evidence" the case against Chreky.
How could a judge discard a seemingly definitive DNA report? Experts said it was virtually unheard of.
But Mueller pointed to a number of incidents of lab error in recent years, including allegations of problems with crime labs in Houston and Richmond, Va. Crime labs in Philadelphia and Minnesota were later discovered to have sent out "false matches."
"It's a terribly important issue," Mueller said. "People involved in doing these techniques make mistakes that are not involved with technology. ... Until you get humans out of the system, these things can happen."
Chreky is no scientist. He said he just knew that this was something he needed to fight. Most people don't have the means to contest a "99.99 percent" finding. His wife, Serena, said the couple spent more than $200,000 to fight the case.
Chreky said he spent much of the past three years overwhelmed with anxiety about the case. "I've been getting up at 3:30, sleeping a couple of hours a night," he said. "I tried to keep busy. You don't want to think about it."
Page A - 3
URL: sfgate.com/cgi-bin/article.cgi?file=/c/a/2005/08/28/MNGS5ECGCU1.DTL