Schaffer Online Library of Drug Policy Sign the Resolution
Contents | Feedback | Search
DRCNet Home
| Join DRCNet
DRCNet Library | Schaffer Library | Drugs and Driving 

 


The German Roadside Survey 1992-1994: Aims, Methods, and Results

Hans-Peter Krüger, Jürgen Reiß, Thomas Hilsenbeck, Jörg Kazenwadel, Mark Vollrath, Werner Krause

Center for Traffic Sciences, University of Würzburg, Röntgenring 11, D-97070 Würzburg, Germany

ABSTRACT

The roadside survey was conducted in two adjacent states in Germany (one each from the former FRG and GDR). Data were collected from more than 21,000 drivers. Drivers were interviewed and a breath test and saliva sample obtained. Further, drivers were asked to provide an address where they could be contacted for an extended telephone interview. All alcohol-positive drivers and their matched controls were selected for the telephone interview. The purpose of the roadside interview was threefold. First, to obtain biographical data on the driver (e.g., age, sex, driving experience); second, to characterize the circumstances of the trip (e.g., destination, purpose, and presence of passengers); and third, to discover the driver's expertise concerning legal issues (e.g., knowledge about laws and sanctions). The telephone interview addressed drinking behavior, drinking and driving experience, and related topics. More than 95% of the drivers complied with the interview and the breath test, while the response rate for the saliva sample was slightly lower. Thus, the roadside survey is a valid means of estimating alcohol distribution among German drivers. A representative driving study for Germany (KONTIV 89) was used to correct the temporal sample bias (about half the sample was drawn during weekend nights) and the estimate a representative alcohol distribution for trips in Germany. We found that about 5% of the trips involved a drinking driver (BAC > 0%), and were able to determine that certain subgroups are especially vulnerable to driving while intoxicated. Further results are discussed.

INTENTIONS

Although the appropriate legal limit for driving under the influence of alcohol (DUI) is discussed extensively in Germany, reliable information about the extent of the problem is lacking. The re-unification of Germany provided an opportunity to conduct an extended Roadside Survey in parts of the former East and West Germany. This Roadside Survey was done for three basic purposes:

  1. To get a detailed picture of DUI in Germany;
  2. To examine the effects of raising the legal DUI limit from 0.00% to 0.08%, which took place in the former East Germany in 1993 (see Krüger, 1995, in this volume); and
  3. To obtain the information necessary to estimate the alcohol-related accident risk (in combination with an accident study; see Krüger, Kazenwadel & Vollrath, 1995, in this volume).

In addition to using the typical random sampling of drivers, the German Roadside Survey (GRSS) included two innovative methods: First, all passengers in a selected sample of the stopped cars were questioned and asked to provide a breath alcohol sample. This was done to examine the role of the social context of safe driving (see Reiß & Krüger, 1995, in this volume). Second, saliva samples were provided by the drivers and analyzed for a number of licit and illicit drugs and for alcoholism markers like methanol (see Magerl & Schulz, 1995; Krüger, Schulz & Magerl, 1995; in this volume).

METHOD

The German Roadside Survey (GRSS) was conducted in two adjacent states of Germany; one state from the former West Germany (Unterfranken) and one from the former East Germany (Thueringen). Both states have approximately 3 million inhabitants. Three components were done: The first between October and December of 1992, the second between April and June of 1993; and the third between April and June of 1994. Data were collected simultaneously by two research teams, one in Unterfranken and the other in Thueringen. For each of the components, the time schedule for the checks and the control sites were selected following a random sampling plan. More than half of the checks were done at weekend nights. This was necessary to obtain the large sample of intoxicated drivers needed to increase the statistical power for testing possible effects of raising the legal DUI limit. However, additional checks were scheduled for every day of the week, during any time of day. In cooperation with the local police, survey sites were selected with regard to public safety and sufficient traffic flow. Most of the sites were large parking lots, where up to four drivers could be interviewed simultaneously. At each site, interviews were done for about 60 to 90 minutes.

German legislation made it necessary for police officers to stop the cars for our researchers and to do a short check of the driver first (check the drivers license, assess fitness to drive, and so on). The police officers selected only private vehicles (no buses or trucks). After the check, the drivers were asked to drive to our research teams to be questioned (if interviewers were available).

The interviewers conducted a short interview and asked for a saliva sample (only in the first two components) and a breath alcohol test (DRAEGER Alcotest 7410). In addition, the drivers were asked for their names and phone numbers so that subsequent extended interviews could be scheduled. If the driver left the survey site refusing to have any contact with the survey team, age, gender, type of the car, and the number of passengers were recorded by the police. Moreover, all blood alcohol concentrations (BAC) of intoxicated drivers detected by the police were recorded by the survey team.

SAMPLE SIZE AND RESPONDERS

Table 1 gives the sample sizes, for the two regions and each of the components, and the percentage of responders to the different measures. In both regions, responder rates were very satisfying. However, in Thueringen, the responder rates were substantially larger, which may be explained by the history of the East Germans, who are still more willing to comply if asked by a police officer.

Table 1
Sample Size (n) and the Number and Percentage of Responders (%) for the Different Methods

  East West Overall
  n % n % n %
Sample Size 11571 100.0 9627 100.0 21198 100.0
Interview 11447 98.9 8933 92.8 20380 96.1
Saliva (only 92/93) 6644 96.4 5569 89.6 12213 93.2
Address 7565 87.9 5499 83.7 13064 86.1
Breath Alcohol Test 11190 96.7 9131 94.8 20321 95.9
Note: Saliva samples were obtained only in the first and second components (n = 13,109)

Although the responder rates are very high, a thorough analysis of the non-responders is necessary to exclude the possibility that driver intoxication was the reason for refusals.

First, the responders rates are clearly influenced by the gender and age of the driver. Younger drivers are more willing to participate (see Table 2) as are male drivers. In that male drivers are more likely to be driving under the influence of alcohol, this is an indication that other factors besides possible intoxication are responsible for the refusals.

Table 2
Responder Rates (Percentages) in Three Age Groups and for Female and Male Drivers According to the Different Methods

  18 - 24 25 - 49 50 and above female male
  East West East West East West East West East West
Interview 99.5 97.4 99.0 92.0 99.2 92.0 99.0 93.5 99.1 93.3
Saliva 97.6 95.8 96.0 88.2 96.7 87.6 96.1 89.1 96.8 90.5
Address 86.4 89.9 88.1 81.6 91.3 83.1 90.2 83.8 87.6 83.6
Breath Alcohol Test 98.0 98.5 96.7 94.5 96.1 92.3 96.3 93.7 97.0 95.7

The most important factor influencing responder rates is shown in Table 3. Although we tried to standardize the checks done by the police, the duration of a single check varied for different teams of police officers. We classified the duration of a check as being short (shorter than one minute), medium (up to 2 minutes) and long (longer than two minutes). The longer the police spent with the drivers, the lower the responder rates were. In the West, the mean duration of a check was 2 minutes, whereas in the East it was 1 minute, 20 seconds. Accordingly, the responder rates were lower in the West. Probably the drivers became angry and impatient and were less willing to participate in the survey if the police check took too long.

Table 3
Percentage of Responders Agreeing to the Breath Alcohol Test in Thueringen (East) and Unterfranken (West) According to Short, Medium and Long Police Checks

Area Duration of Police Check (Minutes)
- 1 Min. - 2 Min. > 2 Min.
East 97.1 96.7 95.2
West 96.8 95.8 93.7

Finally, there were different types of non-responders for the breath alcohol test (see Table 4). First, some drivers did not stop to be interviewed but passed the survey team. As the police did not inform the drivers that they would be asked to undergo a breath alcohol test, it is unlikely that intoxication was the reason for refusal. Second, some drivers gave a saliva sample but refused to do the breath alcohol test. As we told them that we would analyze the saliva for alcohol, it is again unlikely that they refused the breath alcohol test because they were intoxicated. Subtracting these two groups of drivers from the total number of breath alcohol test non-responders, we get a 2.9% for non-responders (2.3% in Thueringen and 3.5% in Unterfranken).

Table 4
Different types of non-responders refusing the breath alcohol test for Thueringen (East) and Unterfranken (West).

Area Breath Alcohol Tested (%) Total Non-Responders (%) Types of Nonresponders (%)
Passing Saliva Only No Samples
East 96.7 3.3 0.1 0.9 2.3
West 94.8 5.2 0.6 1.0 3.5
The percentage of drivers refusing the breath alcohol test is divided into those passing the interviewer team, those giving a saliva sample and those refusing both saliva and the breath alcohol tests

These analyses showed that some factors increase refusal rates that are probably not related to the driver being intoxicated. Thus, we expect that approximately the same percenctage of non-responders was sober as within the responder group. In light of the large responder rate, we conclude that the GRSS gives a very precise picture of the prevalence of DUI in German drivers.

RESULTS

As the German Roadside Survey focused on weekend nights, it does not supply a representative picture of DUI in Germany with regard to time of day and day of week. However, from a representative study on driving in Germany (KONTIV 89, Emnid 1991) we obtained information about how many drivers are present in traffic depending on time of day, day of week, and age and sex of the driver. By weighting the drivers from the GRSS, we were able to adjust our sample accordingly. These adjusted data give a representative distribution of DUI in Germany.

For the presentation of the results, we took the following BAC classes: BAC equal to 0.00%, BAC less than 0.03%, BAC less than 0.05%, BAC less than 0.08%, BAC less than 0.10%, BAC less than 0.15%, and BAC equal to 0.15% and above. In Tables 5-6, the percentages of drivers in each of these classes is given and the total number of drivers in each subgroup is described. Table 5 gives the results for the drivers from Unterfranken (West) and Table 6 the results for the drivers from Thueringen (East).

Table 5
Percentages of Drivers in the BAC Classes, Number of Drivers in each Sub-Group and Percentage of All Trips (Kontiv %): Unterfranken (West)

  sober below 0.03 below 0.05 below 0.08 below 0.10 below 0.15 above 0.15 n % Kontiv
overall 94.50 3.49 0.81 0.64 0.09 0.27 0.20 9041 100.00
male 93.05 4.27 1.02 0.85 0.13 0.39 0.29 6099 67.46
female 97.51 1.87 0.38 0.22 0.00 0.02 0.00 2941 32.54
18 - 24 96.25 2.44 0.64 0.51 0.09 0.03 0.03 1307 14.46
25 - 49 94.52 3.25 0.82 0.71 0.07 0.33 0.29 5146 56.92
50 ++ 93.56 4.50 0.88 0.59 0.12 0.26 0.10 2587 28.62
workday 95.18 2.94 0.75 0.63 0.06 0.27 0.16 6887 76.18
weekend 92.31 5.25 1.01 0.69 0.17 0.26 0.32 2154 23.82
day 95.57 2.82 0.63 0.53 0.05 0.23 0.18 8166 90.32
night 84.48 9.80 2.53 1.69 0.44 0.65 0.40 875 9.68
workday day 96.01 2.38 0.61 0.57 0.04 0.24 0.15 6423 71.05
workday night 83.68 10.84 2.64 1.49 0.31 0.66 0.36 464 5.13
weekend day 93.94 4.45 0.68 0.40 0.07 0.18 0.29 1742 19.27
weekend night 85.39 8.62 2.42 1.92 0.58 0.63 0.44 411 4.55
male, 18 - 24:
workday day 97.96 1.02 0.58 0.43 0.00 0.00 0.00 468 5.18
workday night 88.34 7.11 1.99 2.16 0.00 0.00 0.40 75 0.83
weekend day 92.18 5.04 1.48 0.61 0.70 0.00 0.00 175 1.94
weekend night 89.46 7.80 1.18 0.98 0.00 0.41 0.17 88 0.97
male, 25 - 49:
workday day 94.78 2.87 0.59 0.92 0.00 0.42 0.42 2274 25.15
workday night 79.60 12.31 3.59 1.58 0.71 1.52 0.69 203 2.25
weekend day 92.34 5.29 0.84 0.62 0.00 0.44 0.46 698 7.72
weekend night 81.94 9.52 3.71 2.28 1.29 0.73 0.54 150 1.66
male, 50 and above:
workday day 93.70 3.80 1.16 0.77 0.19 0.38 0.00 1481 16.37
workday night 81.30 15.65 3.06 0.00 0.00 0.00 0.00 59 0.65
weekend day 93.03 5.72 0.39 0.39 0.00 0.00 0.47 377 4.17
weekend night 76.90 12.95 3.72 3.07 0.66 1.31 1.38 51 0.56
Note: The sum of the subgroups does not always add up to 9041 (all drivers). This is due to the weighting procedure. The three-dimensional distribution at the end of the table is given for male drivers, only. For further explanation, see text.

Table 6
Percentages of drivers in the BAC classes, number of drivers in each sub-group and percentage of all trips (Kontiv %): Thueringen (East)

  sober below 0.03 below 0.05 below 0.08 below 0.10 below 0.15 above 0.15 n % Kontiv
overall 95.88 2.56 0.71 0.38 0.20 0.18 0.09 11069 100.00
male 94.67 3.19 0.99 0.52 0.30 0.20 0.13 7468 67.46
female 98.38 1.25 0.14 0.10 0.00 0.13 0.00 3602 32.54
18 - 24 95.00 3.47 0.64 0.36 0.09 0.42 0.02 1600 14.46
25 - 49 95.83 2.47 0.76 0.45 0.16 0.19 0.14 6301 56.92
50 ++ 96.42 2.27 0.65 0.25 0.35 0.03 0.03 3168 28.62
workday 96.28 2.34 0.63 0.37 0.13 0.16 0.09 8432 76.18
weekend 94.59 3.24 0.97 0.44 0.42 0.24 0.10 2637 23.82
day 96.91 1.99 0.50 0.27 0.18 0.08 0.07 9998 90.32
night 86.23 7.83 2.72 1.41 0.44 1.07 0.30 1071 9.68
workday day 97.16 1.88 0.43 0.28 0.11 0.07 0.07 7865 71.05
workday night 84.15 8.75 3.46 1.51 0.39 1.41 0.33 568 5.13
weekend day 96.02 2.40 0.76 0.23 0.40 0.13 0.07 2133 19.27
weekend night 88.57 6.80 1.88 1.30 0.50 0.69 0.26 504 4.55
male, 18 - 24:
workday day 96.41 2.56 0.36 0.34 0.00 0.34 0.00 574 5.18
workday night 80.49 14.52 1.93 0.98 0.44 1.63 0.00 92 0.83
weekend day 95.73 3.27 0.50 0.50 0.00 0.00 0.00 214 1.94
weekend night 84.70 8.89 2.55 1.57 0.99 1.03 0.28 107 0.97
male, 25 - 49:
workday day 95.79 2.62 0.72 0.48 0.07 0.13 0.19 2784 25.15
workday night 82.72 9.08 4.43 1.74 0.47 1.13 0.43 248 2.25
weekend day 94.39 3.01 1.20 0.45 0.64 0.15 0.16 855 7.72
weekend night 85.78 7.72 2.30 2.03 0.65 0.97 0.55 183 1.66
male, 50 and above:
workday day 96.51 2.08 0.63 0.40 0.39 0.00 0.00 1814 16.37
workday night 86.61 5.01 4.52 0.90 0.90 0.90 1.16 72 0.65
weekend day 95.57 2.71 1.04 0.00 0.68 0.00 0.00 462 4.17
weekend night 89.89 6.97 1.66 0.38 0.43 0.68 0.00 63 0.56
Note: The sum of the subgroups does not always add up to 9041 (all drivers). This is due to the weighting procedure. The three-dimensional distribution at the end of the table is given for male drivers, only. For further explanation, see text.

First the overall percentages are given, followed by the one-dimensional BAC distributions for subgroups of the factors:

  1. Gender,
  2. Age (18 - 24, 25 - 49, 50 or older),
  3. Time of day (day: 4 a.m. to 7.59 p.m.; night: 8.00 p.m. to 3.59 a.m.)
  4. Day of the week (weekend: friday night 8 p.m. to monday morning 3.59 a.m.; workday: monday morning 4 a.m. to friday evening 7.59 p.m.).

This is followed by the two-dimensional distribution of BACs for time of day and day of week and the three-dimensional distribution of time of day, day of week and age for male drivers, only (for females, the percentage driving with any alcohol is too small for this purpose). In the last column, we give number of trips per year which each subgroup contributes to the total number of trips per year in West Germany (taken from the KONTIV 89). In all, a total of 18.640.448.170 trips (= 100%) are undertaken in West Germany each year. Thus, for every number given in the Tables, one may compute the number of trips per year in West Germany to which this would be equivalent. For example, 93.05% of all male drivers in Unterfranken are sober. Male drivers account for 67.46% of all trips. Thus, male sober drivers contribute 67.46 * 0.9305 = 62.77% of all trips or 18.640.448.170 * 0.6260 = 11.700.609.320 trips per year in West Germany.

We would like to encourage groups from other countries to compare these results to their findings and discuss the outcome with us.

REFERENCES

Magerl, H.J. & Schulz, E. Methods of saliva analysis and the relationship between saliva and blood concentration. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (in this volume)

Reiß, J.A. & Krüger, H.P. Accident risk modified by passengers. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (in this volume).

Krüger, H.P., Kazenwadel, J. & Vollrath, M. Grand Rapids effects revisited: Accidents, alcohol and risk. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (in this volume).

Krüger, H.P., Schulz, E. & Magerl, H.J. The German roadside survey 1992-1994. The results of saliva analyses from an unselected driver population: alcohols and licit and illicit drugs. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (in this volume).

Krüger, H.P. Effects of raising a BAC limit in Germany. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (in this volume).