1.    Introduction

In the second half of the 20^th Century much has changed with respect to availability, use and societal consequences of psycho-active substances such as alcohol, licit and illicit drugs. After reaching adulthood, but increasingly earlier, most people drink alcohol. Other psycho-active substances used for pleasure such as cannabis, although not so widely used as alcohol, began to grow rapidly in popularity in the sixties, and are not very far from general acceptance these days. In the 1960s many young people used to smoke ‘pot’, at least some 50% in the USA and some 20% in some European countries. On the eve of the new millennium, numerous new fashionable drugs began to be available in sub-cultural markets, and more are appearing every year.

With respect to medicinal drugs, things are perhaps even worse. In a small country like the Netherlands with only 15 million inhabitants, more than 3000 different medicines are available, while on an annual basis close to a million prescriptions for hypnotics alone are written. In the post-war research boom many new medicinal drugs emerged onto the market, now prescribed to ambulant patients instead of hospitalised patients. This means that all possible negative side effects of medicinal drugs are ambulant as well now.

Neither society nor legislation kept pace with all these trends. With the increase in the number of vehicles on the road and the increase in the use of both licit and illicit drugs, one should expect a considerable increase in driving while intoxicated. Indeed, there is increasing evidence that drugs and medicines may impair driver functioning and increase accident risk. Little is known, even approximately, about the impact on societal losses. First of all because it is unclear what the exact influence of many substances on performance capabilities is, and then the relationship with accident likelihood is far from transparent. Traffic safety is undoubtedly affected by drug use, the many signs that indicate a strong relationship can not be neglected, but for the time being the authorities are in need of methods to expose the puzzle. However, there is not sufficient knowledge about methods of detection for roadside enforcement to improve safety.

The CERTIFIED EU Research Project (DG TREN Contract No RO-98-RS.3054) specifically aims to contribute to the existing knowledge base, supporting the development of methods for roadside testing applicable to driver impairment from licit and illicit drugs. This project has the following objectives

·        Review impairment and accident risk evidence for drugs and medicines;

·        Review existing methods of impairment testing and propose new methods (including pilot studies of testing efficacy);

·        Formulate verification methodology for testing methods based on user, legal and operational requirements;

·        Identify key issues relevant to policy formulation.

The research work in the project is carried out in three main research Workpackages. These Workpackages respectively dealt with the form of impairment and accident risk associated with drugs and medicines (R1), as well as current methods that may be applied to roadside impairment testing (R2), and the user, legal and operational requirements for future roadside testing protocols (R3). Each of the Workpackages will be elaborated in summary below, after which the main conclusions from the project will be drawn.

2.    WorkPackage 1-Impairment and Accident Risk

Deliverable 1: Prioritisation of Drugs and Medicines for the Development of Roadside Testing

The aim of the first deliverable (DR1 “Prioritisation of Drugs and Medicines for the Development of Roadside Testing”) was to prioritise those drugs and medicines, including alcohol, which represent the greatest risk to road traffic safety.  This prioritisation was achieved by considering

i.                     research evidence of impairment effects;

ii.                   estimates of exposure within the driving population;

iii.                  evidence of association with accident causation.

The prioritisation is framed within the context of risk relative to other known accident factors such as fatigue, time of day, speeding, etc. An overall risk categorisation imposed by drug type as far as relevant to traffic safety arranges the drugs / drug types as follows:

·         High Risk = Alcohol, Benziodiazepines, Cannabis + Alcohol

·         High-Moderate Risk = Cocaine

·         Moderate Risk = Cannabis, Amphetamines

·         Low-Moderate Risk  = Opiates, Methadone, Antihistamines

·         Low Risk = Antidepressants

After the identification of the principal drugs of concern, both licit and illicit, preliminary proposals were made on psychometric methods of impairment testing specific to drug/medicine effects that may be applicable to roadside testing. The estimated risk scores can be used to provide a preliminary prioritisation of the drug groups in terms of relevance to traffic safety. To this end, a standard metric of risk score ranking is proposed. This ranking can then be adjusted where there is justification for rank assignments that are significantly discrepant from expectations.

With the exception of cannabis and anti-depressants, the safety prioritisation of drug groups on the basis of the (developed) standard metric used is generally in accordance with the preliminary conclusions made in the section summarising the absolute risk. The calculated risk score and initial ranking in the high priority group (*) of cannabis may have been exaggerated. The reasons for this are (i) the concurrent use of alcohol with cannabis (although this is likely not unique to this drug); and (ii) the inflated representation of cannabis in accident cases due to the longevity of detected metabolites in the system.  Instead, cannabis may then be, and in fact is considered as belonging to the medium priority group.  Similarly, the initial ranking as medium priority (**) of antidepressants may have been exaggerated by not taking into account the fact that (i) new generation medications have less impairing effects, and (ii) the depressive persons may drive better with these anti-depressants than if not medicated.  Thus, (new generation) anti-depressants should be considered a low priority.

In the case of opiates, the risk priority may be exaggerated by the estimated risk scores from this table because it does not (i) take account of the high tolerance level experienced by chronic users; nor (ii) the potential for addicts to refrain from driving. Moreover, there is suspicion that the prioritisation of amphetamines may be underestimated because of its effect on mood and risk perception that may reduce inhibitions and increase multiple drug use (including alcohol).

The present safety prioritisation of drug groups on the basis of the standard metric is

·        High Priority* = Alcohol, Benzodiazepines

·        Medium Priority** = Amphetamines, Opiates, Cocaine, Cannabis*

·        Low Priority = Methadone, Antihistamines, Antidepressants**

These estimates should NOT be considered as definitive indications of accident risk because of assumptions underlying the reliability, validity and generality of the parameter data (see DR1, Caveats Section).  For example, the impairment estimates (A) are based on different numbers of studies with varying methods and reliability.  Moreover, this method does not consider the magnitude or relevance of the impairment to driving performance and safety. The exposure data (B) is derived from different countries, which may confound case populations and methods of screening.  And the outcome measure (C) may not be a valid or reliable (significant) indicator of the causal involvement of the drug in traffic accidents.  Furthermore, the method of extrapolating missing values may be erroneous.  Indeed, the boundaries between priority categories are arbitrary, and the method of categorisation based on this risk metric may imply a linearity of scaling that is not appropriate.

As such, this exercise is to be considered ONLY reasonable as a first approximation of (relative) accident risk for the purpose of approximating a rank ordering of drugs for the purpose of this project. The primary objective was to select candidate drugs on the basis of safety priority with which to pilot potential impairment testing methods (and target areas for future research). In this case, it would seem sensible to attempt to use candidates from the high (i.e. alcohol) or medium (i.e. MDMA, (±) 3,4-methylenedioxymethamphetamine also known as ecstasy) priority categories for pilot studies of impairment testing methods. Pilot studies of candidate methods of roadside testing with these high and medium group drugs were carried out in WP2 (Roadside Testing Methods). These studies used alcohol and MDMA as exemplary priority drug types in terms of traffic safety. The aim of the pilot study was to better understand some of the main variables important in devising a roadside test. The limited time available and size of the programme precluded development of a new purpose-designed test. The candidate tests (OMEDA and Vienna) were therefore selected from those available and recommended through consultation with the CERTIFIED consortium.

3.    WorkPackage 2-Roadside Testing Methods

Deliverable 2a: Roadside impairment testing methods

Deliverable 2b: A Pilot Study of Impairment Testing of MDMA with a Computer-Based Task

The reports 2a and 2b together form the major deliverable DR2 for Workpackage R2.  The part 2a contains a summary of the literature review for roadside impairment testing methods together with the background to the pilot study. Some tests are proposed below. Part 2b, also summarised below contains the results of trials, together with data interpretation and a discussion of the way forward for impairment testing.

Tests of driving ability tend to fall into three types: laboratory tests, driving simulators, and on-the-road tests, each with advantages and disadvantages.

Laboratory tests are generally reliable, controllable, sensitive, safe, cheap, and convenient, and with sufficient imagination, almost any laboratory test can somehow be related to some aspect of driving. However, the predictive validity of laboratory tests is poor, and very few tests have any adequate theoretical or empirical justification.

Simulators range from marginally dressed-up laboratory tests to sophisticated devices capable of reproducing most features of driving in a realistic traffic environment. Although no demarcation is available, the former type of simulators should actually not be categorised as such, while the latter are only worth the label if they fulfil yet unspecified but high demands of validity. In that case they require considerable investments in time and money, the best are several times more expensive than a real, instrumented vehicle. Their advantages are that they give a high level of reproducibility and they are safe. However, the latter may actually be a disadvantage as well, if concern for safety is considered a motivator for driver behaviour research. Their disadvantages are that no matter how expensive they are, they cannot faithfully reproduce every aspect of driving, and most of them may be deficient in some vital aspects, e.g. kinaesthetic cues, peripheral vision. Only very few simulators in the world fulfil these requirements. Most simulators may be suitable for some forms of basic training, but there are definite limits to the amount of skill transfer.

Real vehicles are generally considered to be more valid than laboratory tests or simulators, but they do have problems of their own.  Set-piece tests with real vehicles permit some degree of experimental control, but are artificial since, while real driving may involve such manoeuvres, it does not do so under tightly controlled experimental conditions.  Using real vehicles on the highway has been proposed as the ultimate answer, but has two main disadvantages.  First, other traffic, and interactions with it are largely uncontrollable; second, the tests are artificial since the subject is aware that he/she is being studied, either by instrumentation or by human observers. Safety is of course also an issue.

For impairment testing by the roadside, neither real vehicles nor simulators are appropriate, and the choice really falls on laboratory-type tests. There are of two main types.

The first type comprises tests of drug effects on psychological performance such as attention, vigilance, cognitive function such as spatial and temporal information processing, and psychomotor function such as tracking and reaction time. Examples, described in the DR2a report include

·         the Schuhfried Vienna Test System;

·         the Leeds University Object Movement Estimation under Divided Attention (OMEDA) test;

·         the Advisory Group on Aerospace Research and Development (AGARD) Standardisation Tests for Research into Environmental Stress (STRES) Battery;

·         the Cognitive Drug Research (CDR) test battery;

·         the CeNeS Cambridge Neuropsychological Test Automated Battery (CANTAB).

The second type comprises tests of drug effects on physiological and psychological function such as blood pressure, body temperature, pulse rate, pupil size, eye movements, and balance and co-ordination. Examples described in the report include

·        the United States Standardised Field Sobriety Test (SFST)

·        tests from the US Drug Evaluation and Classification (DEC) programme

Some of these tests have been submitted to roadside trials by UK Police Forces.

From the Inception Phase of the project, the EC requested that a small-scale pilot study should be carried out to provide experience with testing methods that may be applicable to roadside testing. Workpackage R1 provided a review of the accident risk with drugs, and recommended alcohol and MDMA as sample drugs representing high and moderate risk categories. The present paragraph summarises the methodology used in the planned pilot studies and presents the analysis of results from this pilot study and other relevant evidence of impairment testing.

The pilot study was completed by two Dutch institutions within the CERTIFIED consortium.  The University of Maastricht performed psychometric and psycho-physiological testing of alcohol and MDMA under double-blind laboratory conditions.  In addition to standard psychometric testing of psychomotor performance (e.g., pursuit tracking), the Maastricht study incorporated a new test method devised by the University of Leeds (the OMEDA test, i.e. Object Movement Estimation under Divided Attention). Based on a top-down theoretical account of intersection accidents, this test has been formulated to test effects of age and dementia on higher level cognitive functions including working memory, time-to-contact estimates, collision judgements and divided attention. The application of this specific method to the test of MDMA effects seemed suitable after the literature survey (2a) and the MDMA literature data base. The University of Groningen carried out a driving simulator test of MDMA effects upon traffic safety, amongst house party visiting (young) people, in conjunction with the Maastricht study, to validate the laboratory data (De Waard et al., 2000).

The performance measures showed dissociate effects of MDMA. There was simultaneous improvement and impairment of performance on different tasks. Improvement of performance relative to placebo was clearly seen on the psychomotor task measuring compensatory tracking performance. In addition, the divided attention version of this task, when it is combined with peripheral signal detection showed improvement under the influence of MDMA, while alcohol effects on errors in this task tended to be negative. Impaired performance under the influence of MDMA was seen on the OMEDA task. The essence of this task was also divided attention, but its unique component was the perception of object movement and the subsequent estimation of object movement without vision, i.e. time perception. In particular, the performance on a Time-To-Contact Estimation subtask was impaired under the influence of MDMA. In depth analysis of the effects of MDMA on this task showed that the MDMA influence was especially pertinent when movement of the object was occluded. This is perhaps indicative of the subjects’ impairment under MDMA to adequately make a mental representation of the events in time. MDMA improves psychomotor function, but impairs time perception.

The driving simulator test largely confirmed these findings, showing that psychomotor functioning, i.e. steering and manoeuvring in traffic, was not impaired after MDMA, but judgements in gap-acceptance and car-following tasks were impaired to a dangerous level.

4.    WorkPackage 3-Roadside Testing Requirements

Deliverable 3: Roadside Testing Requirements

Before starting testing procedures, in some countries in Europe the police need an ”initial suspicion” to take measures from a driver suspected of DUI (driving under the influence of alcohol or drugs). Given such a suspicion, the policemen at first will offer a breath alcohol pre-test to the driver. If the pre-test shows an alcohol concentration above the legal limits, no further investigations concerning illicit or licit drugs are done in most cases. If, however, the pre-test shows an alcohol concentration that does not relate to the driving or behavioural decrements observed (especially in the case that the breath test shows 0 %) a suspicion of the influence of drugs other than alcohol may emerge. As a result of missing test procedures concerning the influence of illicit or licit drugs, only those few drivers that show clearly recognisable conspicuous behaviour are processed for drug testing beyond breath testing.

A cost-effective way of combating drug-impaired driving might be found by combining roadside impairment testing and drug recognition by well-trained police officers, optional screening of body fluids, and blood analysis for evidentiary purposes.

Roadside test methods to detect impaired drivers are of different kinds. They may be categorised as behavioural observation, either related or not related to vehicle handling, performance tests, either with special devices or without, physiological tests and toxicological tests. The tests should fulfil a series of quality criteria concerning practicability, suitability and analytical soundness. The methodology should possess reliability and validity, including content, construct and criterion validity, as well as face validity so that the test appears to be a fair test of driving ability to subjects and administrators. An adequate normative database is necessary as well.

With respect to the incidence of drugs, other than alcohol, several studies have been carried out among the EU driving population. Of the drivers involved in a German study, 1.4% tested positive for illicit drugs, predominantly cannabis and opiates. Another 4.1% tested positive for licit drugs, mostly benzodiazepines. In the Netherlands, 4% of the drivers randomly stopped at Friday and Saturday night tested positive for cannabis and 1.4% for other drugs. Only 1% tested positive for prescription drugs, mainly benzodiazepines. Among young male drivers (18-24 years old), 15.3% tested positive for illicit drugs. For the same group the relative risk of alcohol intoxication turned out to be 6.1, as opposed to 3.9 for the total.

In a survey among European experts it was estimated that 1-2% of the EU general driving population are positive for illicit drugs, whereas approximately 10% are positive for impairing prescription drugs. As far as incidence of drugs in road accident fatalities is concerned, recently strong increases are reported, from 3% in 1985-1987 to 17.4% in 1996-1999. The incidence of illegal BACs, on the other hand, decreased from 25% to 20%, whereas the incidence of prescription drugs remained stable: 5.5% in the period 1995-1997 versus 5.8% in the period 1996-1999. However, the results of these European epidemiological studies do not permit an assessment of the accident (or injury or fatality) risk of drug-driving, since all of the studies focused on either the general driving population or accident-involved drivers/riders. Neither case-control studies nor culpability studies have been conducted, to date. Based on studies outside Europe, it can be concluded that users of most psycho-active drugs other than alcohol have an enhanced fatality risk, but that the fatality risk of alcohol intoxication is approximately three times as high. The effects of cannabis alone on road safety are still unclear, although recent culpability studies seem to indicate that THC-concentrations exceeding 2 ng/l have an adverse effect.

Based on all information gathered, it is possible to make some rough estimates on fatality risks:

·        82% of all drivers are drug- and alcohol-free, having a relative fatality risk of 1;

·        12% are drug-positive, having a relative fatality risk of 2;

·        5% are alcohol-positive (BAC > 0.5 g/l), having a relative fatality risk of 6;

·        1% are alcohol- and-drug-positive, having a relative fatality risk of 9.

The total economic cost of all types of road accidents (fatalities, injuries, and material damage only) is estimated to be 100 billion Euro (including the estimated cost of unreported accidents). This equals an economic cost of 2.22 million per reported fatality. The total socio-economic cost, including value of human life, is calculated to be 162 billion, equalling 3.6 million per reported fatality (Commission of the European Community, 2000)

Drug-driving on EU roads would then cause 8.2% of all road fatalities, equalling an economic cost of 8.2 billion, and a socio-economic cost of 13.2 billion. Drink-driving would cause 17.2% of road fatalities, equalling an economic cost of 17.2 billion, and a socio-economic cost of 27.9 billion. And, finally, combined drink/drug-driving would cause 5.5% of road fatalities, equalling an economic cost of 5.5 billion, and a socio-economic cost of 8.9 billion (Commission of the European Communities, 2000).

5.    Conclusions

The measurement of drug impairment is complicated by a large number of variables. The relevance of a test procedure to driving impairment depends on the psychometric properties of the test: sensitivity, reliability and validity. There are also definite constraints with respect to practicality and robustness that have to be considered when selecting adequate tests. In addition, the sheer number and diversity of tests is such that distilling them down into one valid test for roadside testing of the effects of drugs on driving is very difficult.

However, a first step is definitely needed to get things started. The easiest roadside test to implement in the short term would be one capable of detecting gross impairments, as individual differences in performance are likely to mask more subtle effects. If roadside testing is to act as a deterrent to driving under the influence of drugs it will be necessary to establish in the laboratory safe levels of at least the most commonly used drugs. This together with a test to detect drug levels accurately would deter motorists in the same way as the breathalyser, which it would also augment. Alcohol sets an example of setting defined limits and validation, providing an example of how the effects of a drug on driving can be measured and characterised in the laboratory and in driving situations (including simulators).

Psychologists differ on which test of impairment is best because there is no absolute standard to the behaviour that it tries to measure.  There are also practical constraints in devising a suitable test for roadside use. To date, there has been little effort to develop suitable methods of roadside testing of impairment, especially methods based on objective instrumentation. This project has attempted to undertake a preliminary investigation of some candidate technologies that could potentially be applied to future methods of roadside impairment testing.

No readily available test was found, however. The DRT/FIT test, currently available “off the shelf” has a number of disadvantages, including time and effort involved, user training requirements, subjectivity in recording, paperwork, and lack of quantification. The DRT/FIT test does however appear to work in preliminary trials and has shown good overall correlation with positive drug detection (being based on a well-established US methodology). Its feature of a composite set of measurements has the advantage of detecting (if not fully distinguishing) the often subtle differences in effects produced by drugs on different individuals. Recent trials with cannabis show a correlation between the DRT/FIT test and degraded performance on a driving simulator, indicating a risk to driving and the potential importance of being able to test for impairment. It is about to be recommended for adoption in the UK, which will make it the first impairment test to be introduced in the EU. The DRT/FIT test may set a baseline for evaluation of new techniques such as the relatively new OMEDA test. The implication of the OMEDA test finding that MDMA improves psychomotor function, but impairs time perception, confirmed and validated by the driving simulator study, is that under certain conditions OMEDA seems a suitable method to test effects of MDMA in the field. As such, it constitutes a candidate test for the roadside assessment of drug-impaired driving ability. The extent to which other drug effects can be reliably and consistently measured and distinguished with this test at the roadside remains to be established and validated.

In terms of future development, cannabis is a common moderate risk drug, less traffic safety affecting than the other high prevalence types of drugs, alcohol, and benzodiazepines. It is however a high priority drug in terms of its increasing use by drivers and the increasingly common habit of using it together with alcohol. If a test were devised starting with cannabis, it should then be possible to extend our understanding of the effects of other drugs. Similar reasoning can be applied to moderate priority drugs such as amphetamines (e.g. ecstasy). The impairing effects of alcohol alone, as well as drugs and alcohol in combination need to be detected.

The relative importance of drugs as a road safety problem depends on the relative incidence of drugs as a causative factor compared with other possible causative factors, such as inappropriate speed. Current methods of detection in Europe are for detecting either the consumption or impairment due to drugs at the roadside. In the absence of widespread police training in recognising driver impairment due to drugs, and the absence of reliable and convenient roadside screening devices for drugs, it is not surprising that drugs as a contributory factor is reported to be only 0.3%. This is almost certainly a substantial underestimate. In the absence of techniques to improve quantification of drug driving the true figure can only be speculated upon, but a reasonable lower and upper estimate would be 1% and 3% respectively. The extreme end of this range places drugs as a priority relative to other major accident factors.

Due to the fact that, to date, detecting and proving drug-driving is a lot more expensive and time-consuming than detecting and proving drink-driving, it is estimated that the cost of increased drug-driving enforcement equals at least twice the cost of increased drink-driving enforcement. The cost per avoided drug-related fatality can be estimated on the basis of the Commission’s estimated cost of increased drink-driving enforcement. Assuming that the number of alcohol-related fatalities would drop by 1,000 (= 10% of all alcohol-related fatalities), the cost per avoided alcohol-related fatality is estimated to be 100,000-1,000,000 Euro. Correspondingly, the cost of increased drug-driving enforcement and campaigns can be estimated at 333,000-3,333,000 Euro per avoided drug-related fatality (1,000/600 x twice the cost per avoided alcohol-related fatality). This means that increased drug-driving enforcement and campaigns probably will not comply with the “1 Million Test”, which is maintained as the cost-effectiveness criterion for measures promoting EU road safety (Commission of the European Communities, 2000). Large-scale drug-driving enforcement, regardless of the degree of impairment, would require a considerable amount of police and court capacity, which is difficult at short notice in most countries. If no extra capacity would be made available for this purpose, drug-driving enforcement would probably be at the expense of other traffic law enforcement activities, and particularly of drink-driving enforcement. As a result of the lower risk of apprehension for drink-driving, the total number of alcohol-related accidents might increase instead of decrease. The economic cost of even small increases of drink-driving and of alcohol-related fatalities would outweigh the economic benefit of a 10% reduction of drug-driving and of drug-related fatalities as in the example. In fact, a 6% increase of drink-driving would undo the economic benefit of the 10% reduction of drug-driving. Probably, it will be more feasible to meet the “1 Million Test” criterion by specifically aiming drug-driving enforcement activities at clearly impaired drivers. Limiting enforcement activities to specific groups, especially at times and places where enhanced illicit drug use can be expected (week-end nights and mornings, near well-known places where illicit drugs are consumed on a large scale) would be particularly beneficial. A reduction of drug-related accidents can be expected of a measure which is not specifically directed at reducing drug-driving but at reducing drink-driving by subgroups of drivers who tend to combine alcohol and drug use. An example of such a measure is a lowering of the legal BAC-limit for young (or novice) drivers to 0.2 g/l.

The results (and the associated caveats) do highlight the need for additional research to provide the types of evidence necessary to derive a more definitive categorisation of accident risk from drugs and medicines in Europe. The urgent need for research follow up from the ROSITA and CERTIFIED projects is clear, together with police training and commercial or institutional development of suitable roadside screening devices for drugs.

References

Clarke, A., Riedel, W.J. (2000).  Roadside impairment testing methods.  Project Deliverable DR2a, CERTIFIED EU Research Project (Contract No RO-98-RS.3054), School of Psychology, University of Leeds.

Commission of the European Communities (2000). Priorities in EU Road Safety. Progress Report and Ranking of Actions. Communication from the Commission to the Council, the European Parliament, the Economic and Social Committee and the Committee of the Regions. COM (2000) final, Brussels.

De Waard, D., Brookhuis, K.A., Pernot, L.M.C., Lamers, C.T.J., Booij, L., Sikkema, K.L., Munttjewerff, N.D., Vuurman, E.F.P.M., Riedel, W.J. (2000). Een onderzoek naar de effecten van MDMA (Ecstasy) op cognitieve- en psychomotorische functies, rijgedrag in de rijsimulator, en consequenties voor  de verkeersveiligheid. Rapport COV 00-06. Groningen, Centrum voor Omgevings- en Verkeerspsychologie, Rijksuniversiteit Groningen.

Franzén, S., Berghaus, G., Clark, A., Mathijssen, M.P.M., Tunbridge, R. (2000). Roadside Testing Requirements. Project Deliverable DR3, CERTIFIED EU Research Project (Contract No RO-98-3054), School of Psychology, University of Leeds.

Riedel, W.J., Lamers C.T.J. (2000). A Pilot Study of Impairment Testing of MDMA with a Computer-Based Task.  Project Deliverable DR2b, CERTIFIED EU Research Project (Contract No RO-98-RS.3054), School of Psychology, University of Leeds.

Tunbridge, R., Clarke, A., Ward, N., Dye, L., Berghaus, G.  (2000). Prioritising drugs and medicines for development of roadside impairment testing.  Project Deliverable DR1, CERTIFIED EU Research Project (Contract No RO-98-3054), School of Psychology, University of Leeds.

Wetherell, A. (1999). A Review of the Characteristics and Principles of Tests
for the Roadside Testing of Drug Impaired Drivers. Deliverable R2 Task 2.1 CERTIFIED EU Research Project (Contract No RO-98-RS.3054), School of Psychology, University of Leeds.