I was idly pondering today the suggestion by Mr Justice Ryder’s Family justice modernisation group that some agreed research on some key topics be prepared, in order to have some basic key principles agreed with some research that the profession can have confidence in.
Here’s a suggestion for one – are hair-strand tests for alcohol and drug misuse (as they presently exist) sufficiently accurate for the court to rely on, or is the desire to have something that proves for definite whether a clandestine and unwanted activity is taking place racing ahead of what the science can reliably achieve at this time?
I don’t think I’m madly atypicical for a Local Authority care lawyer, nor is my authority madly atypical. I comfortably do 15 hair-strand tests per year (all of these figures are going to be under-estimates, so the final figure I come up with is even more of an under-estimate). I have colleagues who do about the same amount – I’ll under-estimate that as SIX, although it is more.
Let’s say that my authority does 75 hair strand tests per year. Now, although the costs of that are divided between the Local Authority and the publicly funded parties, it is all ultimately taxpayers money. The cost of a hair-strand test varies on how many months, and how many substances you want to check for, but let’s say £1,000 is about average (again, under-estimating)
£75,000 per year, for one Local Authority.
My quick unscientific count of counties in England is 37, each of whom has at least one LA who deals with care proceedings. Let’s assume (again, underestimating) that we add another 13 LA’s (London, Manchester and those metropolitan boroughs around the outskirts of Birmingham easily give you more than 13).
So, fifty Local Authorities running care proceedings, and each year they spend (together with the LSC) £75,000 on hair-strand testing.
That’s £3,750,000 per year.
I remind you, that this is an under-estimate, but on that under-estimate, the taxpayer is spending THREE and THREE-QUARTERS OF A MILLION POUNDS on hair-strand testing. And the only times that the Court have really grappled with the issue of how accurate this science is, it hasn’t been all that compelling. * THREE and THREE-QUARTERS OF A MILLION POUNDS each year.
[*RICHMOND LONDON BOROUGH COUNCIL v (1) B (2) W (3) B (4&5) CB & CB (BY THEIR CHILDREN’S GUARDIAN) (2010)  EWHC 2903 (Fam)
There was a need for considerable caution when hair tests were being interpreted and relied upon, both generally and particularly in isolation. When used, hair tests should be used only as part of the evidential picture, although findings of very high levels might form a significant part of the picture. Because of the respective strengths and weaknesses of the two tests, both tests should be used if hair tests were to be undertaken. The tests could produce conflicting results. The results should be used only for the purposes of determining whether they were consistent with excessive alcohol consumption by use of the cut-off levels. If the concentration found was below those levels, the results would be consistent with abstinence or social drinking; if it was above, the results would be consistent with excessive alcohol consumption. Further, at the cut-off levels, evidence suggested that 10 per cent of results would be false positives. The tests could not establish whether a person had been abstinent: non-detection on either test did not mean that the subject had not consumed alcohol; detection on either test below the cut-off levels did not mean that they had; and neither test was designed to establish abstinence or social drinking. The cut-off levels for both tests was for the proximal 3cm segment of hair. No levels had been established for 1cm segments, nor was there sufficient published data on testing such segments to enable the validity of such tests to be established. Accordingly, any evidence based on the testing of 1cm segments was unlikely to be sufficient to support conclusions as to the level of alcohol consumption. Further, in the absence of any peer reviewed and agreed cut-off between abstinence and social drinking, a court would need specific justification before accepting any such evidence]
and I’m also aware of another High Court case involving drugs which has not been reported, which was not quite as bleak as Richmond, but wasn’t a glowing endorsement either.
I am not a scientist – though in my younger days, I had aspirations to become one; so one of the things I understand is the concept of “false positives” and “false negatives”. And it is absolutely key to determing how reliable these tests are, to know exactly what the rate of both is.
For those who aren’t science geeks, a false positive is where the test pops up an answer that says you used drugs/alcohol when you didn’t, and a false negative is where the test pops up an answer that says you are clean, when you had in fact used drugs/alcohol.
In terms of alcohol, the High Court established that the rate of false positives and false negatives with hair strand testing were each around 10%. [I have personal experience of asking the question of hair-strand testers what their rates of false positives and false negatives are, and haven’t had a satisfactory answer to date. The companies understandably don’t want, given that they are in competition, to give away commercially sensitive information like that]. I believe those figures were taken from studies where a proper blind control was used, of people known to be teetotal, submitting hair-strand tests and the false positive rate was around 10% there – I think they were teetotal Scandanavian students.
Okay, that’s fine – if there’s only a 10% risk that the test shows a false positive, that means they are 90% accurate, which for the purposes of family proceedings means that for any invidual test that shows the presence of drugs or alcohol, it is massively MORE LIKELY THAN NOT to be accurate. But remember the numbers we’re using – an underestimated quick calculation shows 3,750 per year. Any single case in that amount that produces a false positive has a hugely detrimental impact on the poor parent who was statistically unlikely to be telling the truth when they say the test was wrong but WERE in this individual case, because SOME of those tests will produce a false positive.
[I’m not going to fall into the trap of suggesting that it would be 10% of 3,750 – statistics and probability work in a more complicated way than that, and it would be easy to get caught out by approaching it in a common-sense way – cough, Professor Mea- cough. If you ever want to feel like you no longer have any grasp of common sense, working out some problems on statistics and probabilities is a really good way to make you feel that nothing is real or as it seems – for example, that if you put 23 people in a room together, it is more likely than not that two of them will have the same birthday, even though there are 365 days in a year, and you’d think that you’d need 180 people to get those odds]
Whilst the plural of anecdote is not data, I am fairly sure that any family lawyer reading this can think of several examples from their caseload fairly immediately, of parents who admitted drug use before the results came back, but they came back negative, or of parents who got negative tests but later admitted significant use. Those are the false negatives, and when deciding whether a parent has conquered their drug problem are important. The Court relying on a false negative can be just as bad as relying on a false positive. (Of course, the false negatives are the ones that stick in our mind, because we have confirmation from the admissions that the test is wrong, whereas a denial that the results are true doesn’t confirm to us whether the test is wrong, or the subject is lying)
My point is, the taxpayer is spending a significant amount of money each year on hair-strand tests – the outcome of these tests can be pivotal to the decisions the Court makes about the future of children. It would probably be worth spending a few thousand on research to establish, categorically whether the money is being well-spent.
To be provocative – here’s another scientific method for establishing whether someone has taken drugs, and it is based on science, and the accuracy rate is said to be higher than 90% *, and would be substantially cheaper. You hook the person up to a polygraph (or lie-detector, to use the common name) and ask them. I suspect that you are now recoiling from the idea of lie-detector evidence being used in family courts and would hate to see it used.
And I’m being provocative of course, because I feel exactly the same, but it is worth contemplating why it provokes that unease, disquiet, even revulsion. Is it fundamentally because, despite all scientific evidence to the contrary, we believe that human beings are good at spotting when someone is lying and that the best judge of whether someone is lying is… well, a Judge? Is it because a polygraph machine just summons up old-fashioned science-as-magic imagery, like mesmerism or seances, or is my own personal inner-confidence that I could beat a lie-detector test subconsciously shared by many of us so that we inherently doubt its accuracy, or do we just plain not trust looking at a bunch of lines on a piece of graph paper and calling a man a liar as a result of it?
*caveat – the accuracy of lie-detector tests is probably one of the most controversial debates in the american justice system, and the accuracy is said to be very high by people who support it, and very poor by people who are against it… And bizarrely, the one thing both groups seem to agree on is that the lie-detector is better at spotting when a guilty person is lying than when an innocent person is telling the truth. So, is our in-built prejudice based on the “fact” that they’re just not very good?
I’d like to see some research done on methods of testing for drug/alcohol misuse, with some proper blind studies (where there are samples taken from people who are unequivocally known to be users of substances, or categorically abstinent, and the people doing the testing have no idea which – and ideally neither do the people sending the samples) and I think a provocative, but interesting measure would be whether the science of hair-strand testing proves to be more accurate than lie-detectors… or indeed someone just looking at a person closely while they plead that they’re not a drug-user…
By the way, you probably didn’t know that there has already been a pilot study about polygraph use in the UK, and legislation permitting its use – on assessing the risk of sex offenders in the West Midlands. (The Polygraph Rules 2009 if you’re sceptical) There’s also a very weird bit of research showing that when sex offenders were hooked up to an entirely bogus lie-detector, they were more honest about their offending as a result simply by the effect of believing that the machine WOULD reveal that they were lying, they became more honest. http://www.springerlink.com/content/dm51518228jk8208/
I strongly suspect that proper scientific research would show that hair-strand testing for drugs IS reliable enough for family proceedings, but might assist us with how accurate it is, and the circumstances in which a second opinion or second test should be carried out (not that there would be time in 26 week model for anyone to challenge the outcome) and the Court should KNOW what the false positive and false negative rates are. I am far less confident about alcohol testing – everything I’ve read and heard suggests that the method will eventually be refined and reliable but we’re not quite there yet. But my overall point is that it shouldn’t be for me, or the Judge in a case to weigh up how confident one can be in the testing, we should have the research (done by scientists who have no financial stake in the outcome) to tell us how reliable these tests are, and thus whether these tests are great value for money, or a waste of time, or somewhere in the middle.