Magazine section: Focus

Unusual suspects

New Scientist vol 158 issue 2135 - 23 May 1998, page 18

Will justice suffer if the police rely too heavily on DNA testing?

A MAN breaks into your flat one afternoon. No one sees or hears him, but when he leaves he cuts his hand on the broken window. The police arrive, collect a few drops of blood and send a sample off to the lab for analysis. Overnight, the blood is matched to a sample on the national DNA database, which contains genetic material from every person in the country. The next day, the police arrest the burglar.

It sounds too good to be true. But it could happen if Peter Gammon, president of Britain's Police Superintendents' Association, gets his way. Earlier this month Gammon suggested that Britain's DNA database should be extended nationwide. At the moment it covers only people who have been charged with or are suspected of committing a crime, and any samples from people who are not convicted must be deleted. Gammon argues that many more criminals would be caught if samples from the crime scene could be matched against the whole population.

The proposal highlights the crucial role DNA profiling now plays in police investigations—it is an unrivalled forensic tool and helps to solve thousands of crimes every year. But it also raises serious questions about privacy and civil liberties that will become more urgent as the technology improves and is more widely used.

Barry Scheck, a well-known defence lawyer and professor of law at Cardozo Law School in New York, warned at a conference in Edinburgh on DNA profiling last month that any DNA database could end up being used for purposes other than those for which it was intended. In the future, he said, "people are going to look for genes that determine sex offenders, addicts and violence", and that they will want to use DNA data banks for this research.

But before any studies were carried out using such databases, he argued, there should be a proper debate about their scientific and ethical implications. "It should not be done by the back door. The samples were put together for ID purposes and should be used just for that."

Other prospective advances in DNA forensics raise equally serious questions. Police are starting to appreciate that a DNA sample could help them build up a profile of what a suspect looks like, rather than just provide corroborating evidence after they are caught. This involves techniques that are radically different, and far less reliable, than simply matching profiles. Nevertheless, the Forensic Science Service (FSS) in Birmingham is already testing markers for skin, hair and eye colour developed by Tony Thody and colleagues at the University of Newcastle. Thody has found a sequence of DNA that can confirm with 85 per cent accuracy that a person has red hair.

The FSS has commissioned David Hopkinson at University College London to carry out a study of the genes that determine facial features. The service is also looking for genetic markers for race. It has been using DNA samples from convicted criminals supplied by the West Midlands police force to test how accurately it can identify the race of a subject. "We think we're looking at five or six wrong diagnoses out of two hundred," says Andy Urquhart, a geneticist at the FSS.

"It's not infallible," says Dave Werrett, director of research and DNA services at the FSS. "It's only a guide." Indeed, it may never be possible for police to generate a complete Photofit of a culprit from a bloodstain on an armchair. And however reliable visual profiling becomes, its use as evidence would be highly controversial. "I do see a problem with using it in the courtroom—specifically if the police direct their search as a result of the evidence," says William Thompson, a lawyer and social psychologist from the University of California at Irvine. This would greatly increase the odds, he says, of the wrong person being accused simply because he had "the very characteristics that are taken as additional evidence of his guilt".

Such scenarios stretch way beyond the present capabilities of DNA profiling, which is increasingly reliable. Since Britain's DNA database was launched in April 1995, 20 000 matches have been made. "We expected a tenth of that," admits Werrett. And each match costs just £42.

The database is run by the FSS on behalf of the Home Office and contains about 300 000 genetic profiles from suspects and convicted criminals and over 31 000 from crime scenes. The profiles are derived from the DNA found in every cell of our bodies—in most cases, the cells come from a mouth swab. Forensic scientists build profiles from stretches of DNA known as short tandem repeats (STRs) that vary greatly from person to person but provide few clues about our physical make-up—the equivalent of number plates on cars.

Britain's police look at STRs at six different locations on the genome to build a DNA profile. This gives highly reliable results—some profiles are so rare that the chances of them matching more than one person are as low as 1 in 50 million. Britain may move to using 9 or 10 locations, which would make the match even more convincing, and the US is talking about 13.

But the better profiling gets, the more courts may rely on it as the main source of evidence. This could be dangerous. "What DNA evidence says is that you have similar DNA to the person who committed the crime—not that you committed the crime," says Paul Debenham, head of life sciences at LGC, a private laboratory in Teddington. Werrett points out that people should never be convicted on DNA evidence alone. Other circumstantial factors must always be considered. "It's not for the scientists to usurp the role of the jury," he says.

Already, however, cases are coming to trial where DNA is more or less the only evidence, says Mark Webster, a former Home Office forensic scientist who now works freelance. This opens the way for corruption, he says. If a police officer knows that a certain individual is already on the database, it is easy enough to plant incriminating evidence. "I wonder what steps are being taken to prevent this."

Moreover, in several cases juries have wrongly convicted suspects because they relied on what appeared to be impressive DNA probabilities and ignored other strong evidence (This Week, 13 December 1997, p 18). David Schiff, who studies law at the London School of Economics, says that DNA profiling is "hugely problematic as evidence in court" and even, in some cases, for focusing an investigation. "It's based on probabilities that [the police] don't understand. They stop looking for other leads."

No one is denying the effectiveness of DNA profiling or its success in helping to solve crimes. But as civil liberties campaigners and some lawyers privately point out, its very success could blind people to its limits. Technology, they stress, should not be allowed to undermine the rights of people—whether they are suspects in a criminal investigation or not.

Alison Motluk