CENTRAL ISLIP, N.Y. (AP) — The way prosecutors see it, Oneil Sharpe Jr. was drunk when he raced down a Long Island highway at nearly 100 mph this summer and slammed into a car carrying a family home from a church gathering, causing a fiery wreck that killed a father and his two children.
Sharpe's blood-alcohol reading, taken about four hours later, was 0.06 percent — below the legal threshold of 0.08. But he was still charged with drunken driving and vehicular homicide because a forensic technique estimated that his blood-alcohol level at the time of the crash actually was 0.12.
That technique, known as "retrograde extrapolation," has been used to win convictions nationwide for decades, but has increasingly come under scrutiny by drunken-driving experts as an unreliable measure of a person's intoxication. Some defense attorneys have even labeled it junk science.
"Government lawyers and puppet scientists know retrograde extrapolation is hogwash and will say so when it benefits them, but mostly they pretend otherwise because it is useful in gaining convictions," said D. Timothy Huey, a Columbus, Ohio, attorney specializing in drunken-driving cases.
"Retrograde extrapolation is about as scientifically reliable as astrology," added Jonathan Manley, a former prosecutor who is representing Sharpe in the Long Island crash. "It relies on the assumption that a person's blood-alcohol content peaked prior to the arrest without any basis to prove that."
While there are no national statistics to document the use of retrograde extrapolation, prosecutors in many states, including New York, North Carolina, Michigan, Colorado and Illinois, have offered evidence of estimated intoxication levels at trial. But courts in some other states have severely restricted its use, requiring prosecutors to use only the blood-alcohol readings taken at the time of a person's arrest.
"The general rule in virtually every state is that it is up to a jury to decide its relevance," said attorney Patrick Barone, who has written a textbook titled "Defending Drinking Drivers" and teaches a class at Western Michigan University's law school that includes a two-hour lecture on retrograde extrapolation.
"I challenge it every time it comes up," Barone said. "Juries are too frequently bedazzled by science because science as a whole is not well understood or taught."
In the Long Island case this summer on the Southern State Parkway, the 24-year-old Sharpe wasn't tested immediately, because a friend allegedly whisked him away from the scene. Prosecutors believe Sharpe fled because he knew he was drunk, and they bolstered their case by releasing a cellphone video purportedly showing him tossing a bottle of tequila into the woods after the crash.
Prosecutors who have used retrograde extrapolation swear by it as a proven technique that doesn't reward drunken-driving suspects for fleeing the scene and avoiding immediate blood-alcohol testing.
"It's been standard procedure for many years in many states across the country," said Beadle County, South Dakota, State's Attorney Michael Moore, a past president of the National District Attorney's Association. "It's been proven to be reliable and upheld by courts all across the country."
It was most famously, if not successfully, used after the Exxon Valdez oil spill in Alaska in 1989, when Capt. Joseph Hazelwood was given a blood test 10 1/2 hours after the grounding and recorded a 0.06. A toxicologist estimated his reading at the time of the accident was 0.14.
Hazelwood was ultimately convicted of negligent discharge of oil and fined $50,000, but was cleared of charges of being drunk.
Experts say the intoxicating effects of alcohol are not experienced until it is absorbed into the bloodstream. After a person stops drinking, the blood-alcohol level peaks when the most alcohol has been absorbed and the least amount of alcohol has been eliminated.
Defense attorneys argue alcohol absorption and elimination rates vary widely depending on a person's gender, drinking habits, the type of beverage, what a person ate and how much, and whether a person had experienced trauma, which sometimes slows the rate.
Leonard R. Stamm, a Maryland attorney and dean of the National College for DUI Defense, said prosecutors should avoid retrograde extrapolation "unless they're able to line up every conceivable factor, and I don't know that they're ever able to do that."
Mary Catherine McMurray, a forensic expert who once worked for the Wisconsin state patrol, says she has seen increased use of retrograde extrapolation in the past five years, but cautions "the further away from a person's last drink, the less useful retrograde is, and prosecutors usually have to take the word of the person on when that last drink occurred."
"It's hard for me to call it junk science, because it's got merits," she said, "but I understand why people say it is junk science."
Associated Press researcher Barbara Sambriski in New York contributed to this report.