http://www.druglibrary.org/schaffer/misc/driving/s3p3.htm

 

The German Roadside Survey 1992-1994. Saliva Analyses from an Unselected Driver Population: Licit and Illicit Drugs

Hans-Peter Krüger, Erwin Schulz, and Hansjörg Magerl

University of Würzburg, Röntgenring 11, D-97070 Würzburg, Germany

ABSTRACT

During the German Roadside Survey from 1992 to 1994, breath alcohol measurements were collected from more than 21,000 drivers. In addition, 13,122 drivers were asked for a saliva sample, and 12,213 (93.1%) agreed to participate. In 1992, samples (n = 3,027) were obtained for analysis, for marihuana, amphetamines, opiates, cocaine, benzodiazepines, and barbiturates. Due to insufficient saliva amounts for some of the samples, 2,234 samples were actually analyzed, with a total of 10,696 single analyses performed. After the results were adjusted to reflect a representative driving population, the following percentages of positives were found: benzodiazepines, 2.7%; opiates (including codeine), 0.7%; marihuana, 0.6%; barbiturates, 0.6%; amphetamines, 0.08%; cocaine, 0.01%. In addition, the saliva was analyzed for acetone and aliphatic alcohols, which have been discussed as markers for alcoholism.

INTRODUCTION

The basic purpose of the present research was to determine the prevalence of psychotropic drugs among the German general driving population, especially compared with the prevalence among the accident-involved population. Until now, information on psychotropic drug use among the German driving population has been based mainly on epidemiologic data from accident victims. Because these data were obtained from highly selected driving subpopulations (intoxicated, injured, or dead drivers) they cannot be meaningfully extrapolated to the general population. The same limitation holds for the re-analyses of blood samples of drivers under the influence of alcohol which was often done in Germany. Therefore, we currently have neither reasonable estimates of exposure nor, consequently, estimates of the accident risks induced by psychotropic substances other than alcohol. Because prevalence information from traffic accidents is lacking, those risks can only be estimated by experimental work in the laboratory. Many laboratory studies have been conducted to date, especially with licit drugs and with cannabis. But interpretation of these results presupposes an accepted model of safe driving which does not yet exist.

METHOD AND SAMPLE

From 1992 to 1994, the German Roadside Survey (GRSS) was conducted in two adjacent regions of Germany: Unterfranken (in Bavaria) and Thueringen (in an area that was part of the former DDR or East Germany). The study is described more detailed in Krueger, Reiss, Kazenwadel, Vollrath, Hilsenbeck & Krause (1995) in this volume. During the first two waves of the GRSS in (1992 and 1993), 13,122 drivers stopped at checkpoints by the police were asked by research teams of the Universities of Wuerzburg (Unterfranken) and Jena (Thueringen) to participate in the GRSS. Drivers who agreed were asked for a short interview, a breath sample, and a saliva sample by means of a salivette (a small roll of cotton like the ones used by dentists). A total of 12,213 (93.1%) agreed to provide the saliva sample. Due to financial constraints, only the samples obtained in Unterfranken in 1992 have been analyzed at this time. Therefore, in the following report, we are referring to this subsample.

As Table 1 shows, it was more difficult to obtain samples from older drivers. The younger the driver, the more willing he or she was to participate. Compared to the breath test, the response rates for saliva samples were approximately 1% smaller. Reasons for refusing to participate were usually quite plausible, such as "being in a hurry." Compliance was lowest during the commuting hours and highest during nighttime hours.

Table 1
Number of drivers stopped during the first wave of the German Roadside Survey in 1992.

 

N

Responders %

 

Interview

Breath Alcohol

Saliva Sample

Total

3027

94.9

94.6

91.7

Sex:

Male

2104

95.8

95.3

92.7

Female

879

95.7

92.0

92.0

Age:

18-24

930

98.6

98.2

97.5

25-49

1610

95.0

93.7

91.3

50 ++

433

95.6

90.7

89.8

The last three columns show the percentages of drivers who agreed to participate in the interview, the breath alcohol test, and the saliva sampling procedure, respectively.

QUANTITY OF SALIVA

A total of 3,252 saliva analyses were performed during the study [The first wave in 1992 in Unterfranken yielded a response rate of 93.7% from 3027 drivers stopped at roadside (i.e. 2836 saliva samples). For a more accurate overview of drug use among drivers, especially about the combined consumption of alcohol and drugs, this sample was augmented with 416 saliva samples supplied from alcohol-positive drivers stopped in 1993 during Wave 2 in Unterfranken. Thus, a total of 3,252 saliva samples form the base for the following evaluation] . The first problem limiting analysis was the quantity of saliva remaining after centrifugation. Of the samples collected, 32.6% were essentially dry, with volumes less than 0.1 mL. The mean of the wet samples was 0.42 mL, with a standard deviation of 0.3 mL. To analyze for any one drug, approximately 0.1 mL of saliva is needed. Of the remaining 67.4% wet samples, 45.1% had sufficient saliva for 6 analyses, 6.3% for 5, 5.1% for 4, 1.6% for 3, 6.6% for 2, and 1.1% for 1. The quantity of saliva was correlated with selected characteristics of the drivers.

Table 2 gives the means and standard deviations for the following stratification variables: BAC (0, up to 0.03 g/dL or 0.03%, greater than 0.03%), age, and sex. Low concentrations of alcohol stimulate the flow of saliva. Elderly persons yield less saliva than younger ones and females have smaller quantities of saliva than males. The smallest average quantity was found among elderly women, the same group that had the highest number of refusals. Obviously, many of them really had a dry mouth during the police checkpoints.

Table 2
Average Quantity of Saliva in Milliliters, Stratified for BAC, Age, and Sex of the Drivers

 

 

N

Mean

Std Dev.

Alcohol

sober

1763

.42

.27

up to 0.03%

304

.55

.57

0.03%-more

143

.47

.37

Age

18-24

805

.43

.30

25-49

1205

.45

.32

50-more

220

.41

.29

Gender

male

1615

.45

.33

female

609

.40

.27

The differences between BAC classes and between males and females were significant at a level of p < 0.01

RESULTS OF ANALYSES

After eliminating the dry samples, 2,235 samples from Unterfranken could be analyzed for the following substances: cannabis, amphetamines, opiates, cocaine, benzodiazepines, and barbiturates. A total of 10,969 single analyses were performed.

Samples were analyzed by means of the Fluorescence-Immuno-Assay for all substances with the exception of benzodiazepines, for which a Radio-Immuno-Assay was used. The procedure is described in more detail in the paper by Magerl & Schulz (1995, in this same volume). The cut-off points determine whether a sample qualifies as positive. The following values were chosen: barbiturates > 100 ng/mL, cocaine >= 200 ng/mL, amphetamines >= 100 ng/mL, cannabis >= 20 ng/mL, benzodiazepines < 5 ng/mL, and opiates >= 100 ng/mL. In the case of benzodiazepines, the cut-off value of 5 ng/mL is conservative. A limit of 3 ng/mL would also have been justified. Therefore, the results are given for two cut-off points. In the case of opiates, a medication with the licit drug, codeine, would also yield positive results in the saliva analysis. Approximately three quarters of the opiate-positive samples are suspected to have been obtained from drivers who had taken codeine.

The analyses yielded the following percentages of positives (using the cut-off scores above): 1.63% for benzodiazepines (20 of 1,592 sampled) >5 ng/mL and 2.3% (37) >3 ng/mL. For other drugs, 0.17% (3 of 1,781) for barbiturates, 0.24% (5 of 2,044) for cannabis, 0.11% (2 of 1,831) for cocaine, 0.12% (2 of 1,736) for amphetamines, and 0.44% (9 of 2,053) for opiates.

WEIGHTING PROCEDURE

The percentages of positive samples depend heavily on the sampling procedure and cannot be interpreted directly. More specifically, the GRSS was designed primarily to determine the effects of raising the BAC limit from 0.0% to 0.08%, which actually occurred in the eastern part of Germany after reunification of the nation. Therefore, the study plan oversampled the weekends and nights, during which the highest rates for drunken driving would be expected. To obtain a valid indication of the prevalence of psychotropic drugs among the general driving population, the saliva sample data must be referred to a representative survey about driving in Germany: the KONTIV [The KONTIV may be translated as Continous Survey on Mobility in Germany and is based on more than 30,000 people. As a reference we used the number of trips undertaking using automobiles]. Thus, the saliva sample was first weighted for age and sex of the drivers, time of day, and day of week to yield the same distribution of trips as in the KONTIV. Because the alcohol-positive drivers were overrepresented in the saliva sample, a second adjustment was performed based on the prevalence of alcohol found in the total GRSS, which included more than 21,000 drivers. The results of the analyses after the weighting procedure are shown in Table 3.

Table 3
Alcohol and Drug Prevalences

Alcohol/Drug

% positive (weighted)

BAC above 0%

5.50

BAC above 0.03%

2.01

BAC above 0.05%

1.20

BAC above 0.08%

0.56

BAC above 0.11%

0.43

Benzodiazepines 3ng/ml cut-off

3.64

Benzodiazepines 5ng/ml cut-off

2.60

Barbiturates

0.53

Cannabis

0.61

Opiates including Codeine

0.70

Opiates excluding Codeine

0.15

Amphetamines

0.08

Cocain

0.01

The benzodiazepines comprise the most prominent drug group, which had the same prevalence as BACs higher than 0.03 g/dL. Barbiturates were found in 0.53% of all cases (a frequency equivalent to that found for BACs of 0.08 g/dL and higher). Cannabis was the most frequently used illicit drug, with a percentage of 0.61% of all trips. The other substances only occurred as single cases. Table 4 provides the 95% confidence intervals for the weighted percentages.

Table 4
Weighted Percentages of Positives and their 95% Confidence Intervals

Substance

N

p

SD

SD%

from %

p %

to %

Benzodiazepines >= 5ng/ml

1592

.0260

.00399

15.35

1.82

2.60

3.38

Benzodiazepines >= 3ng/ml

1592

.0364

.00469

12.90

2.72

3.64

4.56

Barbiturates

1781

.0053

.00173

32.34

0.20

0.53

0.87

Cannabis

2044

.0061

.00172

28.21

0.27

0.61

0.95

Opiates

2053

.0070

.00184

26.34

0.34

0.70

1.06

without Codeine

2053

.0018

.00094

51.97

0.00

0.18

0.36

Amphetamines

1736

.0008

.00067

86.46

0.00

0.08

0.21

Cocaine

1831

.0001

.00021

261.27

0.00

0.01

0.05

BAC > 0

9128

.0554

.00239

4.31

5.07

5.54

6.01

BAC >= 0.03

9128

.0199

.00146

7.35

1.70

1.99

2.28

BAC >= 0.05

9128

.0118

.00113

9.58

0.96

1.18

1.40

BAC >= 0.08

9128

.0053

.00076

14.34

0.38

0.53

0.68

BAC >= 0.11

9128

.0041

.00067

16.31

0.28

0.41

0.54

Column 1 lists the substances. In cases for which two cut-off points were used, two entries are shown. The alcohol results of the GRSS are also included as a reference. (N, sample size; p, probability of positives; SD, standard deviation; SD%, deviation in percent of p.) The next three columns present the lower limit, the center, and the upper limit of the confidence interval in percent.

More important than the mere presence of a drug in the blood of a driver is the question of whether the drug's concentration is high enough to impair driving. In the case of benzodiazepines, the behavioral effects depend not only on the concentration, but also on the particular variety within this drug family. Some varieties differ drastically in their metabolism during which many further metabolites are produced from the psychoactive mother drug; some of them are themselves psychoactive. Therefore, concentration measures provide only a rough estimate of the total amount of psychoactivity. Of the 37 benzodiazepine positives, 17 were less than 10 ng/mL, 13 less than 20 ng/mL, 3 less than 30 ng/mL, and 3 less than 40 ng/mL. The maximum concentration in one case was found to be equal to 44 ng/mL. Realistically, only the 4 cases with the highest concentrations (11.11%) can be discussed in terms of traffic safety. The 3 positive barbiturate cases had concentrations of 110, 127, and greater than 300 ng/mL. Only the latter concentration would be likely to make safe driving questionable.

The concentration-response relationships for illicit drugs are as difficult to determine as for the prescription drugs. The FPIA test for cannabis is sensitive to the psychoactive mother drug, THC, as well as to its first metabolite, THC-COOH, which is more or less neutral for behavior. Only 1 out of 5 positives for cannabis had a concentration of more than 40 ng/mL. One of the 2 positives for amphetamine had a concentration of more than 370 ng/mL, and 1 of the 2 cocaine positives had a concentration of more than 700 ng/mL. The results for opiates are difficult to interpret because the licit drug, codeine -- often used in cough medicine -- also yields a positive indication. We estimate that about three-quarters of the opiate positives were in fact codeine induced.

For the licit drugs, at most about 10% (5 of 40 positives) might qualify as potentially dangerous for traffic safety. In the case of the illicit drugs, at most about 30% (5 of 18) qualify. Therefore, the percentages of positives in Table 3 are the very upper limits of the prevalence rates relevant to traffic safety.

All saliva samples were also tested for ethanol and other alcohols (like, for example, methanol). Most samples could be analyzed for more than one drug. Only one saliva sample was determined to be positive both for benzodiazepines and for opiates. None of the samples positive for benzodiazepines or barbiturates tested also positive for alcohol. The respective ratios for the illicit drugs and alcohol were as follows: cocaine, 0 alcohol positives out of 2; opiates, 3 of 9; cannabis, 3 of 5; and amphetamines, 1 of 2.

COMPARISON WITH OTHER SOURCES

Krüger (1995a, b) reviewed 69 studies of the exposure rates to licit and illicit drugs. The studies took place during the past 20 years and represent locations all over the world. Only 2 roadside studies (RSS) could be found. Of the remaining 67 studies, 39 dealt with injured or killed drivers, 15 reanalyzed blood samples of alcohol positives, and 13 analyzed blood or urine samples from drivers suspected to be under the influence of drugs. The median exposure rates reported in the studies for licit and illicit drugs and for alcohol are given in Table 5.

Table 5
Exposure Rates for Alcohol and Licit and Illicit Drugs as they were found in Studies Using Different Epidemiological Methods

 

Roadside
% positives

Injured
% positives

Fatalities
% positives

Reanalysis
% positives

Drugs
Suspected

Drugs

1

17

19

10

28

Medicaments

4

13

10

7

72

Alcohol

6

35

52

100

*

* usually not reported by authors

The first column (Roadside) gives the results of the German Roadside Survey. The following columns show the median measurements from a total of 69 studies involving (in this order): injured drivers, fatalities, reanalyses of blood samples from drivers under the influence of alcohol, and analyses of blood or urine samples from drivers suspected to be under the influence of drugs.

Detecting 28% of illicit and 72% of licit drugs, the category "suspected drivers" gives only information about the frequency structure of psychoactive substances in the group of traffic offenders. Nothing can be concluded with regard to exposure rates. The most important differences can be found between the results of RSS and studies using other methods. Illicit drugs were found in 1% of the RSS, but in about 15% of the other studies. For licit drugs, this relation is approximately 4:10, whereas for alcohol the ratio is 6:40. Accepting the RSS results as those that correspond best with real-world conditions, the other methods overestimate the exposure rates for alcohol by a factor of 7, for licit drugs by a factor of 2.5, and for illicit drugs by a factor of 15! (This comparison implies that the median proportions found in the 69 studies of which the review is comprised are representative for the exposure rates in Germany. This assumption was proved to be true. The equivalence between the results of the German studies and the median of the foreign studies is sufficient to justify these conclusions).

Apparently, these differences between the RSS and the other methods have three causes. The first is the confounding of exposure rate and risk in the samples of accident victims. High concentrations of drugs put drivers into the category of population-at-risk and, therefore, they are overrepresented in accident populations. On the other hand, a high affinity exists between the use of alcohol and illicit drugs. The results of our RSS showed that, while none of those who tested positive for licit drugs had consumed alcohol, about a third of those who tested positive for illicit drugs had. Thus, if samples of alcohol positives are taken, the proportion of illicit drugs will be overestimated for the general driving population, a finding that accounts for the second difference between the RSS and other methods. Third, the combination of alcohol and drugs (either licit or illicit) multiplies the accident risk, a finding most recently demonstrated by Terhune et al. (1992). The accident samples are characterized by a high proportion of alcohol positives which themselves are susceptible to the combined use of illicit drugs. Taking these three reasons together, the extreme estimation bias in all sampling methods beside the RSS becomes obvious.

CONCLUSION

The results of the German Roadside Survey showed that saliva sampling is a useful epidemiological method which is highly acceptable to drivers. The method is limited by the fact that about a third of the samples are "dry". Modern immunological analysis gives at least a semi-quantitative picture of drug concentrations. In the case of benzodiazepines and cannabis, the results are sometimes difficult to interpret because the test is also sensitive to metabolites that are not psychoactive.

Within these restrictions the Survey found that approximately 3% of all trips in Germany are undertaken by drivers with detectable concentrations of benzodiazepines, 0.5% with barbiturates, and 1% with illicit drugs, primarily cannabis. Only one tenth of the licit drugs and about one third of the illicit ones were found in concentrations likely to induce safety problems. Thus, at the present time, drugs, either illicit or licit, are not a severe problem for traffic safety in Germany.

The Roadside Survey itself as an epidemiological method showed its capability of yielding reasonable estimates about the size of the drug problem. The comparison of its results to those of other methods showed the extreme estimation biases that occur when inferences are made about the exposure rates for the general driving population from the rates found in populations-at-risk. On the other hand, the limitations of the Survey method also became obvious. Of a total of 3252 saliva samples, only 47 were found to be positive. If the scientific interest lies in the description of the features of driving and drivers under the influence of drugs, the Roadside Survey has an extremely negative cost-benefit ratio. To obtain 200 positive cases, about 20,000 drivers must be controlled and their saliva analyzed, which would cost a total of approximately 1.5 millions DM.

REFERENCES

Krüger, H.-P. (1995a). Drogen im Straßenverkehr. Auftreten und Bedeutung. Rechtsmedizin. Beiträge zu aktuellen Themen, 11 (in press).

Krüger, H.-P. (1995b). Auftreten und Risiken von Cannabis im Straßenverkehr. Eine epidemiologische Studie. Paper at the Symposium "Drogen und Verkehrssicherheit" of the Bundesanstalt für Straßenwesen (will 1995 be published within the Technical Reports of the Bundesanstalt für Straßenwesen).

Terhune, K.W., Ippolito, C.A., Hendricks, D.L., Michalovic, J.G., Bogema, S.C., Santinga, P., Blomberg, R., & Preusser, D.F. (1992). The incidence and role of drugs in fatally injured drivers. National Highway Traffic Safety Administration, October 1992 (DOT HS 808 065).