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


Drink/Driving, Daily Self-reported Drinking, and Analysis of Alcohols in Saliva Sampled Daily

M. W. Perrine, L. S. Lester, J. S. Searles, J. C. Mundt, J. L. Ferguson

Vermont Alcohol Research Center, 2000 Mountain View Drive, Colchester, VT 05446, USA

Supported by US PHS grant #PO1-AA07203 from NIAAA to M.W. Perrine


Saliva testing for ethanol and methanol content was included in a larger study designed to assess the validity of daily self-reported alcohol consumption. A newly developed "interactive voice response telephone system" (IVR) allowed each respondent to report daily, using the touch-tone telephone keypad, to pre-programmed questions concerning: quantity of beer, wine, liquor, and cigarettes consumed; whether respondent drove after drinking and , if so, at what self-rated level of intoxication; plus questions on stress, mood, and general health. Every night during the 4-week study, breath and saliva samples were collected from the 30 respondents of their homes. Usable saliva volumes (>=.2ml) were obtained in almost all cases (97.3%). Saliva ethanol values ranged from 0 to 295.4 mg/dl, with subjects showing elevated (>=50 mg/dl) readings on 1.4 study days, on average. Methanol values ranged from 0 to 1.3 mg/dl, with subjects showing elevated (>=0.3 mg/dl) readings on 11.8 study days, on average. Ethanol values from saliva testing were highly correlated (r = .95) with ethanol readings from breath samples and with the IVR relf-reports of drinking (r = .72). Methanol values were moderately correlated with breath (.40) and saliva (.43) ethanol values. Subjects in the higher ethanol group (>=50) showed significantly higher self-reported number of drinks consumed, intoxication in general, intoxication while driving, and problem severity than those in the lower ethanol group (0-49.99).


A new technology for obtaining daily self-reports of drinking and drink/driving has recently been developed, implemented, and validated. It uses the touch-tone telephone to enter data into an automated interactive voice response (IVR) system. Three telephone-reporting studies of alcohol consumption have been conducted at the Vermont Alcohol Research Center: a 21-day non-IVR pilot study (Perrine et al., unpublished report), a 112-day IVR study (Mundt et al., in press; Mundt et al., in press; Searles et al., in press), and a 28-day IVR validity study (Perrine et al., in press[b]). This report is based on a portion of the latter study.

The three major sources of data regarding driving after drinking and the role of alcohol in traffic safety are: alcohol involvement in highway crashes, alcohol presence at roadside surveys (random testing), and responses to relevant questions in retrospective surveys of drinking drivers. It is noteworthy (and understandable) that all three approaches gather cross-sectional rather than longitudinal data. Extremely important insights into such problems as the role of alcohol in crashes have been obtained with the first two approaches; however, they are too sporadic and too coarse for a detailed study of drinking drivers over time. Drink/driving behavior tends to persist consistently over time (Perrine et al., in press[a]), and, therefore, longitudinal studies of daily drinking and of drink/driving would provide new insights into the problem, possible relationships with associated factors, and leads for developing new prevention measures. Now, a highly specialized modification of survey methodology is available to collect detailed longitudinal data on a daily basis: the interactive voice response (IVR) system.

This paper: (1) relates daily self-reported drinking to self-reported driving after drinking and to self-ratings of intoxication while driving, as well as to drinking locations and driver records; (2) examines saliva analysis' utility for determining presence of ethanol and methanol; and (3) investigates interrelations among breath ethanol, saliva ethanol, and saliva methanol.



Newspaper ads and posters were used to recruit 30 males (aged 23-60 years, mean age = 37.3) from the local population. Eligibility criteria were: at least 21 years old, current alcohol consumers, living in a "stable" (at least 6 months) relationship with a female partner (i.e., the collateral reporter in the main study), and available at home every evening of the 28-day study (July 10 - August 6, 1994) to provide saliva and breath samples.

Interactive Voice Response (IVR) System

Subjects attended an orientation session for a review of the IVR manual and all data collection procedures, a detailed description and demonstration of the IVR system, and an explanation of the payment system whereby they could earn $500 for providing complete data (i.e., daily IVR self-reports, breath and saliva samples) throughout the 28-day study. To call the IVR system, subjects used a dedicated 800 (toll-free) number connected to a computer-automated, subject/menu-driven system that collected daily reports on the questions in the IVR script (Figure 1). Subjects responded to the questions (e.g. "How many beers did you drink yesterday?") by pushing the appropriate numbers on their touch- tone telephone keypad. Subject-entered data were collected automatically and associated with an individual account for storage, analysis, and evaluation. A research assistant reviewed incoming data daily and reminded subjects to call in if they had not done so by the evening meeting at the subject's residence.

Figure 1
Script used by recorded interviewer for daily monitoring with interactive voice response (IVR) system

1. BEER: How many beers did you drink yesterday?

2. LIQUOR: How many drinks containing liquor did you have yesterday?



4. INTOXICATION: Rate your highest level of intoxication yesterday on a scale of 1 to 10, with 1 being perfectly sober and 10 being as drunk as you've ever been.

5. DRIVING: Rate your highest level of intoxication while driving yesterday on a scale of 1 to 10, with 1 being perfectly sober and 10 being as drunk as you've ever been. Press 0 if you did not drive yesterday after having 1 or more drinks.

6. PROBLEMS: Rate the seriousness of any problems you had because of your drinking yesterday on a scale of 1 to 10, with 1 being no problems and 10 being serious problems.

7. LOCATION: For the following locations, press 1 if you had at least 1 drink yesterday and 2 if you didn't: Home; Friend's; Bar; Restaurant; Other.


4. REASONS FOR NOT DRINKING YESTERDAY: For each of the following statements, press 1 if true or 2 if false:

a. AVOIDANCE: I purposefully avoided drinking.

b. OPPORTUNITY: I had no opportunity to drink.

c. THOUGHT: It didn't occur to me to drink.

d. MONEY: I couldn't afford to drink.

e. DESIRE: I had no desire to drink.

f. ILLNESS: I didn't feel well.

g. HANGOVER: I was too hungover.

8. CIGARETTES: How many cigarettes did you smoke yesterday?

9. MOOD: How would you rate your mood yesterday on a scale of 1 to 10, with 1 being very sad and 10 being very happy?

10: HEALTH: How did you feel physically yesterday on a scale of 1 to 10, with 1 being very bad and 10 being very good?

11: STRESS: How stressful was your day yesterday, on a scale of 1 to 10, with 1 being low stress and 10 being high stress?

Breath Ethanol Concentration (BEC)

Breath samples were obtained from each subject each evening at his residence. Just prior to sample collection, the subject was instructed to rinse his mouth with a sip of water and spit it out. He was then asked to take a second sip, rinse his mouth, and swallow the water. Immediately thereafter, a breath sample was obtained using a hand-held device (Alco-Sensor III). The BEC readings were never shared with subjects.

Saliva Alcohol (Ethanol and Methanol) Concentration

After BEC determination, data collectors obtained saliva samples using the Salivette device. Subjects were previously instructed to refrain from placing anything in their mouths for 20 minutes prior to sample collection (e.g., no food, drink, cigarettes, chewing tobacco, gum). The subject was instructed to remove the cotton wad from the Salivette and place it in his mouth, keeping it there for approximately 3 minutes, while rolling it around like a hard candy rather than chewing on it. The samples were kept in coolers containing ice packs during evening data collection. Refrigerated samples were shipped to the Forensic Science Institute of Ohio, Franklin County Coroner's Office, Columbus, Ohio, where they were extracted from the Salivettes by centrifuging and analyzed for ethanol and methanol content by Perkin-Elmer headspace gas chromatography, using procedures and standards developed at the University of Würzburg, Institute of Forensic Medicine.


Alcohol Consumption

The 30-subject sample adequately represents the drinking patterns of "social drinkers" in the general population, based on responses to standard questions in a brief interview following the orientation session (Table 1). During the 28-day study, drinking was reported on 55.2% of all possible drinking occasions ("occasions" = number of subjects x number of days in the study = 840). Further, subjects consumed only one drink on approximately one third (32%) of the occasions on which they reported drinking.

Table 1
Selected Alcohol Consumption Data from 30 Subjects

Preferred Beverage Usual Quantity of Drinking Usual Frequency of Drinking Light Drinking Days (Average Week) Heavy Drinking Days (Average Week)
Beer 73% 1 Drink 30% 1-3/month 7% Mean Days/wk 1.9 Mean Days/wk 0.8
Wine 13% 2-3 Drinks 50% 1-6/week 87% Mean Drinks 0.87 Mean Drinks 3.6
Liquor 13% 4 or more 20% Daily 7% <= 1 Drink 80% >= 4 Drinks 43%

Saliva Analysis

The saliva collection procedure was highly successful in terms of obtaining both subject cooperation and sufficient sample volumes (>= .2 ml) to analyze ethanol and methanol content (97%). Saliva ethanol values ranged from 0 to 295 mg/dl, with subjects showing elevated (>=50 mg/dl) readings an average of 1.4 study days. Methanol values ranged from 0 to 1.33 mg/dl. Taking methanol values as a function of the number of self-reported drinks on a given day, the data fell into four, distinctly different categories. The mean methanol values associated with the categories 0, 1, 2-4, and 5 or more drinks were, respectively: .21, .25, .32, and .42 mg/dl. A threshold methanol value of .3 mg/dl was set to define onset of levels possibly indicative of heavy drinking (Krüger et al., 1993). Our subjects showed elevated (>= .3 mg/dl) readings on an average of ll.8 study days. Looking at drinking and associated problems as a function of these ethanol and methanol cutoffs, subjects with elevated (vs. non-elevated) ethanol values showed significantly higher self-reported number of drinks consumed, intoxication rating in general, intoxication rating while driving, and problem severity rating. A very similar pattern was noted when viewing the data in terms of elevated and non-elevated methanol levels.

Across all reporting occasions and intake levels, saliva ethanol values were highly correlated (r=.95) both with breath ethanol readings taken on the same day and with the IVR self- reports of drinking (r=.72) for that day. Methanol values were moderately correlated with breath (.40) and saliva (.43) ethanol values and with total self-reported drinks (.44) from the same reporting day. Since methanol levels may remain elevated for some time after a drinking episode, we also determined the correlation between methanol values on a given reporting day with breath (.15) and saliva (.20) ethanol values and total drinks (.19) reported for the previous day. Selecting only those subjects who reported consuming five or more drinks on a given day, correlations between the (non-zero) methanol values on the heavy drinking day and the ethanol measures for that day were .61 (breath) and .64 (saliva). The correlation between methanol values and total number of drinks reported for the heavy drinking day was .46. Considering a possible delayed heavy drinking effect on methanol levels, we calculated the correlations between the (non-zero) methanol levels on the day after the heavy drinking day with the ethanol levels (.22, breath; .36, saliva) and the total number of drinks from the heavy drinking day (.46).


With reference to all drinking occasions (i.e., 55% of all possible occasions), driving after drinking was reported by subjects on 43% of drinking occasions, with 33% of reports rated as having been "perfectly sober" at the time and 9% as feeling some level of intoxication while driving. Further examination of "perfectly sober" drivers versus "intoxicated" drivers revealed significant differences between their (mean [SD]) methanol (0.27 mg/dl [0.13] vs. 0.34 mg/dl [0.11]), saliva (8.77 mg/dl [22.95] vs. 21.06 mg/dl [30.9]), and breath [8.59 mg/dl [20.61] vs. 22.4 mg/dl [35.3]) ethanol levels.

The drinking location data are also relevant, assuming that those who drink in more than a single location are likely to be at risk for operating a motor vehicle under the influence of alcohol. On drinking days, drinking at a single location was reported 84.7% of the time. Drinking only at home was reported 61.9% of the time, whereas drinking at two, three, or four different locations (e.g., home, bars, restaurants, friends' houses) on the same day was reported 13.3%, 1.3%, and .4% of the time, respectively. Dividing subject reports based on one versus two or more drinking locations revealed that, relative to single locations, those associated with multiple locations showed significantly (t-tests, p<.01) higher breath and saliva ethanol readings, more self-reported drinks consumed, higher levels of intoxication while drinking, and more problems due to drinking. No differences between these two groups were detected with regard to methanol levels or self-reported mood and stress level. Excluding data from subjects who reported that they never drove after drinking and those who always reported being perfectly sober while driving, reports from those who drank at multiple locations indicated significantly higher intoxication ratings while driving than did those associated with a single location.

Driver Records

The subjects' driver records were also reviewed. Of the 30 subjects, one had multiple DUI convictions, one had one DUI, and one had an offense in which alcohol was probable. One of these individuals reported an intoxication rating while driving on four occasions during the present study. Neither of the two heaviest drinkers in the study had any past record of driving offenses; one of them reported that he never drove after drinking.


The main conclusions to be drawn from the results are outlined below.

  1. In the parent study, it was concluded that "IVR technology provides an innovative user- friendly methodology for obtaining valid measures of daily alcohol consumption" (Perrine et al., in press[b]). In addition, the data analyzed for the present paper indicate a strong systematic relationship between self-reported driving after drinking and elevated intoxication ratings and measured ethanol levels on 9% of the drinking occasions.
  2. Further, those who reported drinking at 2 or more locations (vs. only 1) showed significantly higher ethanol levels, more drinks consumed, higher intoxication ratings in general and while driving, and more drinking-related problems. Although small in number, 10% of the sample had alcohol-involved offenses on official driving records, approximately the same percentage found in our larger samples of the general driving population. Further, one of these subjects reported driving on 4 occasions during the study with elevated intoxication ratings.
  3. This pilot work with the saliva procedures was extremely successful, both logistically (collection, handling, and shipping) and analytically, especially given the loss of only 3% of the samples due to insufficient volume. Thus, saliva analysis' utility for determining the presence of ethanol and methanol was clearly demonstrated, by a user-friendly method. Further, we found very high correspondence between the saliva and breath ethanol measures.
  4. Saliva procedures are valuable in determining methanol presence. Despite the limited amount of heavy drinking among these subjects, elevated methanol levels were nevertheless found among the heavier drinkers, consistent with the findings of Krüger et al. (1993).
  5. We conclude that the IVR method, especially with concurrent measures of saliva and/or breath alcohols, has enormous potential for future research on drink/driving.


Krüger, H.-P., Schulz, E., Magerl, H., Hein, P.M., Hilsenbeck, T., & Vollrath, M. (1993). Incidence and meaning of psychoactive substances in highway traffic [Ger]. Würzburg: Zentrum für Verkehrswissenschaften der Universität Würzburg.

Mundt, J.C., Perrine, M.W., Searles, J.S., & Walter, D. (in press). An application of interactive voice response technology to longitudinal studies of daily behavior. Behavior Research Methods, Instruments, and Computers.

Mundt, J.C., Searles, J.S., Perrine, M.W., & Helzer, J.E. (in press). Cycles of alcohol dependence: Frequency-domain analyses of daily drinking logs for matched alcohol-dependent and nonclinical subjects. Journal of Studies on Alcohol.

Perrine, M.W., Meyers, A.M., & Yu, J. (in press[a]). Roadside BAC, alcohol use, and driver record (in this volume).

Perrine, M.W., Mundt, J.C., Searles, J.S., Lester, L.S. (in press[b]). Validation of daily self- reported alcohol consumption using interactive voice response (IVR) technology. Journal of Studies on Alcohol.

Searles, J.S., Perrine, M.W., & Mundt, J.C. (in press). Self-report of drinking by touch-tone telephone: Extending the limits of reliable daily contact. Journal of Studies on Alcohol.