Saturday, November 7, 2009

Drug Nutts

Some years ago, I attended a celebration dinner at a well-known British university. A head of department had just been nominated to the position of vice-chancellor of another equally prestigious seat of learning, and we were gathered to celebrate his achievement. The food was more than acceptable and the wine both drinkable and in copious supply - so that when, over coffee, the speeches began, we were all in good humour. First in line to speak was the guest of honour himself and, as usual, he gave a brilliant and witty oration. After that, however, matters went rapidly downhill as a succession of mediocre wits - all heads of department - rose to their feet. Listening to one especially dull contribution, the Dean of Studies, next to whom I was seated, whispered "...there are professors and professors (pause) unfortunately".
Many years later the spat (November 2009) between the UK's Home Secretary Alan Johnson and Professor David Nutt - recently sacked as head of the Drugs Misuse Advisory Council - brought that Dean of Studies' pithy aside back to mind. The good Professor Nutt had claimed that the drug Ecstasy was no more harmful - perhaps less so - than horse-riding. Since he appeared to have most of the press and a handsome proportion of the pundits on his side, I decided to throw in a protest - choosing for my target a gruffly strident anti-Johnson polemic in Open Democracy.
Please understand, I'm no defender of any of the major UK political parties - but still less am I ready to be bludgeoned into intellectual submission by professors like Nutt who prefer to be believed (and even obeyed) because they are professors rather than because they are learned.
Here are my two submissions; the second in response to a suggestion by another contributor that I might not have understood the professor's use of statistics.

Submission One

The chattering classes are having a ball with this one - with everyone who thinks they are on the progressive wing of political correctness lambasting Alan Johnson for his sacking of Professor Nutt. Although I am cautiously in favour of legalization - and therefore probably on the professor's side with respect to policy, I find the widespread belief that a professor's advice should be taken as gospel to be no more credible than the fantasy of papal infallibility. When the professor demonstrates evidentially that alcohol and tobacco cause more harm (to health?) than ecstasy and cannabis, I assume he knows his stuff; but when he moves from there to speculating about the different effects of government policy, he seems to me to be laying claim to a level of authority and wisdom that exceed his professional qualifications. I even wonder if he has truly evaluated - scientifically - the evidence for his statement that ecstasy (also crack? Heroin? skunk? LSD?) and horse-riding are about equal in the degree of harm they cause. If so, I would be interested in seeing that evidence and would be mightily impressed - and astonished - if it proved to be watertight. Statements of this kind are a form of playing to the gallery: unnecessary if the audience consists of fellow scientists, but otherwise merely provocative.

Nor are they illustrative of anything. Lots of things can be said to cause harm. It might just as easily be shown, for example, that walking at night, or jogging are as dangerous as horse-riding. The argument rests on the fallacy of assuming that a comparison of two dissimilar elements sheds light on either of them. And if the professor truly thinks that banning horse-riding would be easy, one wonders what he's been smoking.The statement itself could not be more revealing of professorial naivety.

History shows that scientists frequently get things wrong. But even if Professor Nutt is entirely correct in his analysis, this doesn't mean his advice is equally correct. Politicians have to consider a great deal more than their scientific advisers: international treaties and understandings on the issue in question (the drugs trade), public opinion, the tabloids, the welfare of vulnerable members of society, and not least the opinions of other advisers (or are we to assume that professors always agree with each other?) etc.

Advisers merely give advice. They should not expect their advice always to be taken, still less expect ministers in effect to obey them. Nor should they conflate the right to freedom of speech with throwing a tantrum if their advice is rejected. If they want to influence policy, they should stand for office. They might then learn something about how difficult these issues are to deal with politically, however straightforward they may seem in the white-coated confines of the campus laboratory.

Submission Two

Okay. Let's have some fun.

The use of the word ecstasy in this context presumably means the practice of ecstasy consumption. The data obtained on its harmfulness are derived from the population of ecstasy consumers (NOT the proportion of the total population that consumes the drug).

Similarly the data on harm caused by horse-riding are derived from the population of horse riders.

Statistical data are, of course, taken from population samples.

In order for a comparison to be valid, the two samples have to be derived from the same (or at least very similar) populations. A simple random sample of the UK population will not work because it could not be guaranteed that it would contain any ecstasy consumers or horse riders. So in order to conduct a comparative analysis of "harm", we need a sample of ecstasy consumers and a sample of horse riders.

The population of horse riders in the UK includes:

1. Adults who ride horses as part of their profession: jockeys,
police officers, cavalry etc.

2. Adults who ride purely for recreation.

3. Hunters (or cross-country animal chasers)

4. Sports men and women, some of whom participate in national and international competitions.

5. Children

6. Circus and other performers.

The population of ecstasy consumers is...well no doubt Professor Nutt can tell us who consumes ecstasy and under what circumstances.

It's a fair bet that Professor Nutt's ecstasy population sample is different from that of horse-riders (if he ever used one). Children are unlikely to be represented among the drug-takers. Nor sports riders for that matter. Nor professional consumers - i.e. people who get paid for taking ecstasy.

You may think that if you remove professional horse-riders and children from your "rider" population and, say, restrict it to adults who ride for recreation, you can select ecstasy consumers of the same age range and thereby get comparable samples. The problem here is that the two samples would be selected differently and not randomly, which would invalidate the comparison.

Now let us look at how you define each activity. By ecstasy consumers do we mean anyone who consumes the drug once, or regular consumers. If the latter, what constitutes a regular consumer? How many acts of consumption qualify an individual for inclusion in the data sample?

We ask the same kind of questions of horse-riding. How many person-hours of riding per unit of time (say per year) qualify someone as a rider?

And once we have defined our sample populations, we still have to work out how to make one activity equivalent to the other (how many person-units of ecstasy consumption equal a person-unit of horse-riding or vice versa). This is necessary in order to be sure of a roughly equal chance a priori (i.e. before conducting the analysis) of finding "harmful effect" in each sample.

We then have to identify degrees of harm and its victims (the self and other parties). Do we choose, say, admissions to hospital, as our benchmark? Should we also include benefits (i.e. negative harm) as well as positive harm to health, happiness and long-term success or failure? In the case of horse-riding we can probably limit the definition of harm to injury suffered by the rider and the horse (okay there will be some third party injuries too).
Harm from drug-taking, however, is much less straightforward since the relationship of cause and effect may be less easy to establish and might only reveal itself in the long term. Harm to others may be significant - although we would have to be sure that ecstasy was involved (say a drug-induced driving accident), and this, of course, may be a matter of opinion unless we restrict ourselves to the decisions of a court of law following a trial.

All statistics derived from population samples are based on assumptions, but the assumptions themselves have to be reasonable, credible, defensible. The point I am making is that you can't simply take national statistics on aggregate horse-riding injuries, put them against figures (derived from where?) of harm caused by ecstasy consumption and say that one is more or less harmful than the other. To give an extreme example, breathing oxygen - something we all do - is 100% fatal. But it would be meaningless to say that breathing is more dangerous than warfare.

The chances of Professor Nutt having conducted a valid comparative statistical analysis between ecstasy and horse riding are - in my view - vanishingly small; and by making such a comparison he was, therefore, grandstanding. In other words, given that he was a government adviser, he was not making a scientific point but a political one.

JF

PS The above assumes a direct relationship between 'use' and 'harm' (i.e. the more you do the more you are likely to suffer or cause harm). For ecstasy, this is probably a safe assumption. In the case of horse-riding, however, the relationship could be inverse, i.e. harm may be more common among neophytes than experienced riders. If this is so, then Professor Nutt's comparison could be more accurately described as being between ecstasy and inexperience - which would render it even more nonsensical.