A perspicacious psychologist was once asked his opinion on a matter of public interest: Would recently introduced tough penalties for drivers affected by alcohol encourage people to drive more carefully? Yes, he replied,but not if they'e drunk. It was a telling point. Legislation intended to deter people from driving after they have been drinking will always be most effective in deterring the least drunk drivers and least effective in deterring the most drunk drivers.

In spite of this obvious problem there has been for many years a world-wide recognition of the need to protect the public from people who are in no condition to be in charge of a motor vehicle.

Driving under the influence was for many years around the world the attempted solution to the problem¥a legal attempt to put away the drunk and incapable driver and remove a menace from the road. Any person who drives or attempts to drive a motor vehicle while under the influence of alcohol, drugs or alcohol and drugs to such an extent as to be incapable of having proper control of the vehicle commits an offence.. That or a similar statement described the offence. But the difficulty in proving that someone was so drunk they should not be permitted to take charge of a motor vehicle caused never ending problems in prosecution.

In principle, it should be simple to devise a test which will assess a person' driving ability and determine whether this is so affected by intoxication as to make the driver incapable of having proper control. In practice, such a test could not be implemented for a very simple reason: as will be shown, many legally drunk drivers are safer than sober drivers. (Any test which removed sober drivers and set scot-free drunk drivers would be unacceptable. It is better to be sober but incapable than intoxicated and competent.)

Blood-alcohol legislation is one way of circumventing this difficulty but one which introduces a curious philosophico-legal oddity. A person who drives or attempts to drive a motor vehicle while the percentage of alcohol in his blood equals or exceeds 0.08 per centum commits an offence.... The offence of bad driving need no longer be proved, or indeed, have occurred. The imbibing driver might be the best in the world but if his blood does not come up to scratch then he is guilty of an offence which may put him in gaol.

Random breath testing is the way to enforce such a law. If drunk drivers cannot be detected visually by their driving inability then they will be detected biochemically by their blood constitution.

It might be argued that such legislation can do no harm. (After all, few of us fit the dangerous drunk description even if most of us have an occasional drink.) This is not the case for a number of reasons.

First, there is a subset of the motoring population who routinely have illicit levels of alcohol in their blood as a consequence of their metabolism and not as a result of drinking alcohol¥which indeed they may have never consumed. Yet they may well be very safe drivers. A law which has a side-effect of rendering such a minority either undetected criminals or excludes them from the roads is surely unjust.

Second, people who fear that their illicit blood biochemistry might be discovered at the spur-of-the-moment whim of a police officer may act in ways which will endanger both themselves and others. There are three main ways in which this can happen. 1. Good Samaritans may be less likely to stop should they witness an accident if they risk a breath test. At best they may slow down and assure themselves that there appears to be no serious injury, at worst they will pass by rapidly. This is understandable given the penalties they might face. 2. More serious is the possibility that they will not help someone they have themselves injured; with their employment if not their freedom at stake, there will be a strong temptation to abandon the drunken pedestrian who falls in front of their car. 3. Others may ask their sober but less competent spouse or companion to drive them home after a party. The sober but inexperienced friend may well feel they are doing the right thing until the hazards of unfamiliar night-driving lead to tragic consequences.

By their very nature it is difficult to determine how frequent these side-effects are (except perhaps hit-and-run offences) but there is already anecdotal evidence to show that they certainly have occurred. It would be surprising if they had not.

But one of the oddest ill-effects of random breath-testing is at first blush paradoxical: it will necessarily and selectively remove from the roads our safest drivers.

When L.F. Borkenstein carried out his classic drink-driving study thirty years ago, (the Grand Rapids Study) the Borkenstein Breathalyzer was used as an aid to determining the role of blood alcohol in road accidents. He also examined a number of other variables which might be of importance. He found that high blood alcohol levels were indeed associated with road accidents, particularly the most severe and those which were single vehicle collisions. On the other hand he concluded that the safest drivers were white, well-educated, upper-class, married, middle-aged men with more than average driving experience. Though some aspects of this best-driver profile were controversial, these paled in comparison with his two remaining conclusions: the best drivers were also daily drinkers with a blood-alcohol of 0.03%! The immediate interpretation, that a few drinks improve driving ability, caused an academic uproar. In fact, this interpretation is wrong, as further analysis of his data showed. The Figure (Borkenstein) (taken from P.M. Hurst in the Journal of Safety Research, 1973) shows the relative probability of involvement in an accident as a function of blood alcohol. One of the characteristics of the motor vehicle drivers of Grand Rapids which had interested Borkenstein was how often they consumed alcohol. As shown in the figure, he rated them in five categories from daily drinkers to yearly or less. Two things should be noted.

First, all drivers have their driving ability impaired as a result of drinking even quite small amounts. Second, the daily drinkers were safer than the general sober population even when they had substantial blood alcohol levels. At the other extreme, yearly or less drinkers are worse than the average driver even when sober but particularly so at even moderate blood alcohol levels. The three intermediate drinking frequency groups fell between these two extremes.

Because of Borkensteins scrupulous attention to detail and because of the scale of the study (5,000 drivers involved in accidents were compared with 7,000 who were not involved) his work has been both acclaimed and widely accepted. (It is presumably partly for this reason that his accurate and reliable Breathalyzer was for so long widely incorporated into traffic legislation around the world.) What has been almost completely unappreciated is that if the Breathalyzer is used to randomly test drivers and where applicable remove them from the road then it will selectively remove the safest drivers. If a sample of daily drinkers is tested on some particular day it is likely that many of them will have positive blood alcohol levels, some exceeding the legal limit.

In a similar sample of yearly-drinkers it would be most unlikely to catch them on that one day of the year when they drink. So random breath testing must selectively remove daily-drinkers¥even though they are safer than the general population. Random breath testing therefore not merely fails to improve the average driving ability, it has the opposite effect. The number of accidents per driver-kilometre will increase.

Are there no benefits from random breath testing? Yes, it will occasionally catch people who are so drunk that they are a menace to themselves and others, but with what inefficiency and at what cost? Only a tiny fraction of those tested are found to be over the limit. Given the financial constraints of an imperfect world, time devoted by the police to RBT is necessarily time that is unavailable for other, perhaps more important, duties. Random testing or examination is generally neither practised nor indeed legal in the detection and prevention of other offences. RBT might also deter some drivers from drinking and driving. But to reiterate, those who are in most need of deterrence are the least likely to be dissuaded.

How then should the problem of alcohol-related traffic accidents be addressed? BorkensteinOs conclusion that high blood alcohol levels are strongly correlated with frequent, severe and otherwise avoidable accidents is uncontroversial and generally accepted and supported by everyday experience. Something ought to be done. But it is the dangerous driving, whether due to a loose wheel, drunkenness or plain incompetence which should be stopped.

If we wish to concentrate on drivers whose erratic or dangerous behaviour on the road is due to drunkenness, then the old driving under the influenceO law can be substantially strengthened by the use of blood-alcohol measurements as confirmation of their drunken state. The person who weaves across lanes before ending up on the kerb and who is found to have a positive blood alcohol reading will have committed an offence. But the offence would consist in his actions and his blood alcohol would be corroborative evidence as to the cause of his actions. Injustice would still occur but the absurdity of penalising drivers solely for their bad blood would cease. Who are we really trying to stop¥bad drivers or drinkers?

J.R.Johnstone is a former Senior Research Fellow with the NH&MRC.He is the author of Health Scare: The Misuse of Science in Public Health Policy, published by AIPP and available from the Institute of Public Affairs. Modified 17 Feb 2001