Twelve-Month Prevalence and Changes in Driving After Drinking

United States, 1991–1992 and 2001–2002

S. Patricia Chou, Ph.D., Bridget F. Grant, Ph.D., Ph.D., Deborah A. Dawson, Ph.D., Frederick S. Stinson, Ph.D., Tulshi Saha, Roger P. Pickering, M.S.

Laboratory of Epidemiology and Biometry, Division of Intramural Clinical and Biological Research, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Department of Health and Human Services, Bethesda, MD.

Reprinted from Drug and Alcohol Dependence 80(2). S. Patricia Chou, Bridget F. Grant, Deborah A. Dawson, Frederick S. Stinson, Tulshi Saha, and Roger P. Pickering, “Twelve-Month Prevalence and Changes in Driving After Drinking,” pp. 223–230, 2005, with permission from Elsevier.

Background: Drinking and driving has been identified as one of the most important contributors of motor vehicle fatalities. This paper addressed the existing gap in our public health knowledge regarding the current prevalence of driving after drinking and how this has changed over the past decade. Methods: Prevalence rates of drinking and driving in 2001–2002, and changes in those prevalence rates between 1991–1992 and 2001–2002, were examined in two large nationally representative surveys of the U.S. population. Results: Overall, the prevalence of driving after drinking was 2.9 percent in 2001–2002, representing approximately six million U.S. adults. This rate was about three-quarters of the rate observed in 1991–1992 (3.7 percent), reflecting a 22-percent reduction. Generally, the male–female differentials in the rate of driving after drinking decreased over the past decade. However, the sex ratios increased substantially for underaged youth over the past decade, reflecting the sharp decrease in prevalence of driving after drinking among 18- to 20-year-old women. Constant and emerging subgroups at high risk for drinking and driving included Whites, Native Americans, males, underaged young adults, and 21- to 25-year-olds. Conclusions: The results of this study highlighted the need to continue to monitor prevalence and changes in driving after drinking. Results are discussed in the context of strengthening existing prevention and intervention efforts and developing new programs with the sociodemographic differentials observed in this study in mind.

Key words: Alcohol-impaired driving; Drinking and driving; Sociodemographic characteristics; High risk subgroups; Changes in drinking and driving

INTRODUCTION

Road traffic injuries and fatalities are a global problem affecting all sectors of society. The World Health Organization (WHO) predicted that 1.39 million people would die in road traffic crashes in 2000 and that 2.34 million people will die in 2020 (Murray and Lopez, 1996). Among them, more than 50 percent of traffic fatalities occurred among young to middle-aged adults ages 15–44. In 2000, road traffic crashes ranked as the ninth leading cause of mortality and morbidity, accounting for 2.8 percent of all global deaths and disability. WHO’s projections suggest that by 2020, road traffic injuries could rank third among all causes of death and disability worldwide (WHO, 2003).

The practice of drinking and driving has been consistently identified as one of the strongest predictors of motor vehicle fatalities worldwide (WHO, 2003). In the United States, approximately 40 percent of all motor vehicle-related deaths are attributed to alcohol each year (National Highway Traffic Safety Administration [NHTSA], 1995). In the year 2001 alone there were over 16,000 alcohol-related fatalities and 275,000 alcohol-related injuries as a result of motor vehicle crashes (NHTSA, 2002). Accordingly, the consequences of drinking and driving have received wide media coverage and intense attention from public policy makers, and the practice of drinking–driving has been studied using a variety of data sources.

The National Highway Traffic Safety Administration (NHTSA) maintains the Fatal Accident Reporting System (FARS), which collects detailed information for every fatal traffic crash occurring in the United States. Yearly data are released for public use by the following year and have been used to identify demographic and drinking characteristics associated with a high risk of fatal injuries (e.g., Kennedy et al., 1996). FARS data have also been studied in conjunction with blood alcohol concentration (BAC) data collected in a series of national roadside surveys. Using such an approach, Lund and Wolfe (1988, 1991) and Zador (1991) reported that each 0.02-percent increase in the driver’s non-zero BAC nearly doubled the risk of being involved in a fatal crash. In addition, the 1973, 1986, and 1996 national roadside surveys indicated the prevalence of drinking drivers (BAC > 0.05 percent) fell sharply from 1973 (13.7 percent) to 1986 (8.4 percent) but decreased slightly between 1986 and 1996 (7.7 percent). Voas and colleagues (1998) further reported that there was no reduction in the prevalence of drivers at the highest BACs between 1973 and 1996 despite the significant decline in the overall prevalence of drinking drivers on American roadways.

The Behavioral Risk Factor Surveillance System (BRFSS) is a nationwide annual telephone survey of U.S. adults that has also collected data on drinking–driving in addition to other health-related behaviors. Smith and Remington (1989), using data from the 1986 BRFSS, estimated that there were 150 million drinking–driving events in the United States in 1986. Based on data from the 1993 BRFSS, Liu and colleagues (1997) estimated there were 123 million episodes of alcohol-impaired driving among U.S. adult drivers, a decrease of about 20 percent compared with the 1986 estimate. This rate of reduction was comparable with the 30-percent decline in alcohol-related traffic fatalities during the period of 1982 and 1992 (Centers for Disease Control and Prevention, 1993).

As evidence indicates that automobile fatalities are the leading cause of death among American youth 15–20 years old (Centers for Disease Control and Prevention, 1996), numerous studies have focused specifically on young drivers. Using the FARS data, a special report on young drivers compiled by NHTSA (2002) reported that the number of young drivers (15–20 years old) who were intoxicated at the time of the fatal crash dropped 24 percent between 1991 and 2001. In addition, prior studies have analyzed data collected in national surveys to investigate drinking-and-driving behaviors among youth. For example, data collected in the 1991 Youth Risk Behavior Survey (Escobedo et al., 1995) indicated that the prevalence of drinking and driving was positively associated with frequency of alcohol use and binge drinking and years elapsed since the initiation of alcohol use, and data from the 1999 National Household Survey on Drug Abuse (NHSDA) indicated that college attendance increased the likelihood of drinking–driving regardless of any pre-existing conditions (Paschall, 2002). Likewise, data from the National Institute on Alcohol Abuse and Alcoholism’s (NIAAA) 1991–1992 National Longitudinal Alcohol Epidemiologic Survey (NLAES) and the 1999 Harvard College Alcohol Survey showed that drinking–driving was inversely related to age at first drink and at first intoxication (Hingson et al., 2002, 2003). None of these studies based on national survey data examined changes over time in drinking and driving among youth. However, based on drinking driver fatalities reported in the FARS data, NIAAA surveillance reports (NIAAA, 1993, 2003) showed a 15-percent decline in alcohol-related fatal traffic crashes among youth ages 16–24 between the years 1991 and 2001 (31.4 percent versus 26.9 percent).

From a review of previous research, it became apparent that few studies provided prevalence information on alcohol-impaired driving at the national level or within important subgroups of the population. Nor did many of the studies track changes in the prevalence of alcohol-impaired driving over time within major subgroups of the population. From a public health policy standpoint, it is essential to be able to assess the prevalence of driving after drinking too much, along with changes of the prevalence over time both at the national level and within major subgroups of the population. In light of the bold national goal established in 1995 to reduce alcohol-related automobile crash fatalities to 11,000 by 2005 (NHTSA, 1997), the progress in achieving this goal has been slow, and the change that occurred in the past few years has been in the wrong direction (Sweedler et al., 2004). Accurate and timely information will help formulate effective and empirically derived prevention and intervention programs. This is of critical importance for those highly vulnerable subgroups of the population, namely college students and young adult males. However, available data sources to address these concerns are surprisingly sparse.

The purpose of this paper, therefore, was to address the existing gap in the public health knowledge regarding the 12-month (current) prevalence and changes in drinking and driving over the past decade among important subgroups of the U.S. population. To address changes in drinking-and-driving practices, we compared data from two large nationally representative samples of the U.S. population: the 1991– 1992 National Longitudinal Alcohol Epidemiologic Survey (NLAES, n = 42,862) and the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC, n = 43,093).

METHODS

Samples

Both the 1991–1992 NLAES and the 2001–2002 NESARC are nationally representative samples of the general adult population (age 18 and older) of the United States conducted by the National Institute on Alcohol Abuse and Alcoholism. Details of these two surveys have been described in detail elsewhere (Grant et al., 1992, 2003). Face-to-face interviews were conducted with each respondent drawn from the civilian, noninstitutionalized population of the United States for both NLAES and NESARC. NESARC additionally included a subsample of respondents residing in group quarters. Overall survey response rates were 90 percent for NLAES and 81 percent for NESARC.

To ensure adequate numbers of respondents for analytic purposes, oversampling of Blacks in NLAES and of Blacks and Hispanics in NESARC was implemented at the design phase. Within each household unit, a person age 18 or older was randomly selected. Young adults (ages 18–29 in NLAES and ages 18–24 in NESARC) were oversampled at a rate of 2.25:1.00 at this stage of sample selection to secure a greater representation of this heavier drinking subgroup of the population. Both samples were weighted to adjust for oversampling young adults and nonresponse at the household and person levels. The data were then adjusted to be representative of the U.S. population for a variety of socioeconomic variables including region, sex, race, and ethnicity using the 2000 Decennial Census.

Measure of Driving After Drinking

On NESARC, driving after drinking was explored by asking “In your entire life, did you EVER more than once drive a car, motorcycle, truck, boat, or other vehicle after having too much to drink?” A followup question, “Did that happen in the past 12 months?” was asked if the answer to the preceding question was “yes.” There was a slight variation in the question on the NLAES survey, in which the question read, “In the last 12 months, did you drive a car, motorcycle, truck, boat, or other vehicle after having too much to drink?” A followup question—“About how many times did this happen in the last 12 months?”—was asked if the answer to the preceding question was “yes.” To equate these measures, only NLAES respondents reporting driving after having too much to drink more than once were included in the analyses presented here.

Analysis Procedures

To account for the complex sampling design of both NLAES and NESARC, a specialized software program, SUDAAN (Research Triangle Institute, 2004), was used to estimate the standard errors of all current prevalence estimates. This software adopts Taylor series linearization to take into account the design effects of complex sample surveys like NLAES and NESARC. Differences in prevalence rates of driving after drinking for the total sample and among important sociodemographic subgroups of the population were conducted using t-tests designed for independent samples.

RESULTS

Twelve-Month Prevalence of Driving After Drinking: 2001–2002

The 12-month prevalence rates of driving after drinking by age, sex, and race are presented in Table 1. In the total sample, the 12-month prevalence of driving after drinking was 2.9 percent, representing approximately six million American adults. The prevalence of driving after drinking among males was 4.4 percent, whereas the corresponding rate among females was 1.5 percent, representing 4.4 million and 1.6 million U.S. adults, respectively.

Table 1. Twelve-Month Prevalence of Driving After Drinking by Age, Sex, and Race: United States, 2001–2002

Age (years)		Male			Female			Total
		%	S.E.	Population Estimatea	%	S.E.	Population Estimatea	%	S.E.	Population Estimatea
Total
	Total	4.4	0.22	4388	1.5**	0.11	1585	2.9	0.13	5973
	18–29	7.8	0.57	1759	2.9**	0.30	659	5.3	0.33	2418
	30–44	5.2	0.36	1653	1.8**	0.20	598	3.5	0.23	2251
	45–64	3.0	0.31	929	0.9**	0.16	301	1.9	0.18	1231
	> 65	0.3	0.10	47	—	—	26	0.2	0.07	73
White
	Total	5.0	0.25	3516	1.7**	0.13	1296	3.3	0.15	4812
	18–29	9.9	0.76	1364	3.5**	0.38	492	6.7	0.42	1857
	30–44	6.1	0.45	1301	2.5**	0.28	542	4.2	0.29	1842
	45–64	3.4	0.38	810	1.0**	0.19	251	2.2	0.22	1061
	> 65	0.3	0.12	41	—	—	11	0.2	0.06	52
Black
	Total	2.6	0.35	258	0.7**	0.17	90	1.5	0.19	348
	18–29	3.1	0.74	84	1.2*	0.45	40	2.1	0.42	124
	30–44	3.5	0.74	118	0.4**	0.15	18	1.8	0.33	136
	45–64	1.8	0.43	54	—	—	32	1.3	0.30	85
	> 65	—b	—	2	0.0	0.00	0	0.1	0.05	2
Native American
	Total	5.9	1.61	124	2.4	0.92	56	4.1	0.84	180
	18–29	—	—	32	—	—	18	5.8	2.24	50
	30–44	9.9	3.25	69	—	—	11	5.5	1.70	80
	45–64	—	—	19	—	—	12	2.0	0.50	31
	> 65	—	—	4	—	—	15	—	—	19
Asian
	Total	2.0	0.61	89	0.8	0.34	35	1.4	0.33	124
	18–29	4.3	1.63	57	2.6	1.29	32	3.5	1.00	89
	30–44	—	—	24	—	—	4	—	—	27
	45–64	—	—	8	0.0	0.00	0	—	—	8
	> 65	0.0	0.00	0	0.0	0.00	0	0.0	0.00	0
Hispanic/Latino
	Total	3.3	0.48	402	0.9**	0.18	107	2.1	0.27	509
	18–29	5.0	1.00	222	2.0**	0.50	76	3.6	0.60	298
	30–44	3.1	0.59	141	0.6**	0.16	24	1.9	0.32	165
	45–64	1.5	0.42	39	—	—	7	0.9	0.22	45
	> 65	0.0	0.00	0	0.0	0.00	0	0.0	0.00	0

^a Population estimates are in thousands.
^b Estimate does not meet precision standard.
* p < 0.05 for gender comparison.
** p < 0.01 for gender comparison.

Age. Young adults ages 18–29 (5.3 percent) were significantly more likely to drive after drinking in the last 12 months, compared with 30- to 44-year-olds (3.5 percent) and 45- to 64-year-olds (1.9 percent), with age differentials of 1.5 and 2.8, respectively. Respondents age 65 or older had the lowest prevalence of operating a motor vehicle after drinking (0.2 percent, p < 0.001). When examined within the race–ethnicity and sex groups, the prevalence of driving after drinking was significantly (p < 0.001) greater among young adults (ages 18–29) compared with the rates of all other age groups, but only among White males.

Race–Ethnicity. Native Americans exhibited the highest rate of driving after drinking (4.1 percent), which was significantly (p < 0.05) greater than those of Blacks (1.5 percent), Asians (1.4 percent), and Hispanics (2.1 percent). Notably, the rates of the latter three ethnic groups did not significantly differ. When prevalence was examined with respect to gender, significantly greater rates were observed among Native Americans relative to Whites, Blacks, Asians, and Hispanics, and this difference was statistically significant among males and females.

Gender. Overall, in the last 12 months, males were significantly more likely to operate a motor vehicle after drinking too much than females (4.4 percent versus 1.5 percent), a ratio of about 2:93. Among race–ethnic groups, however, we observed no gender difference among Native Americans and Asians. In contrast, males showed significantly greater rates than females among the other ethnic groups (i.e., Whites, Blacks, and Hispanics). The sex ratios varied from 2:9 to 3:7 for these three ethnic groups.

Gender differences were observed within age–ethnicity subgroups of the population. In general, males were significantly more likely to drive after drinking compared with females for most age groups among Whites, Blacks, and Hispanics. Due to several prevalence rates not meeting precision standards, we were not able to assess the gender differences among the age–ethnic subgroups of Native Americans and Asians.

18- to 29-Year-Olds. To address the elevated prevalence of driving after drinking among the young adult group (18- to 29-year-olds) we disaggregated this age category into single ages and present the prevalences and associated population estimates by sex in Table 2. This detailed table showed that among male and female adults, the prevalence of driving after drinking peaked at ages 22 and 23. With respect to gender differentials in the prevalence of 12-month driving after drinking, male rates were significantly greater than female rates in all age groups except 21- and 23-year-olds.

Table 2. Twelve-Month Prevalence of Driving After Drinking by Sex and Age: United States, 2001–2002

Age (years)	Male			Female			Total
	%	S.E.	Population Estimatea	%	S.E.	Population Estimatea	%	S.E.	Population Estimatea
Total
18	5.5	1.43	122	1.0**	0.58	20	3.4	0.80	142
19	6.4	1.50	140	2.8*	0.94	50	4.8	0.91	191
20	9.3	2.34	175	1.8**	0.72	39	5.3	1.17	214
21	7.2	1.60	135	5.7	1.67	119	6.4	1.18	254
22	11.5	2.12	199	4.2**	1.31	77	7.8	1.26	276
23	10.4	2.16	183	7.1	1.48	124	8.7	1.29	307
24	6.9	1.49	127	2.8*	0.84	54	4.8	0.83	182
25	9.4	2.62	179	2.3*	0.84	44	5.9	1.40	223
26	6.9	1.96	119	1.4**	0.47	26	4.1	1.01	145
27	7.5	1.92	138	2.4*	0.86	43	5.0	1.08	181
28	7.1	1.69	121	2.0**	0.69	38	4.4	0.86	159
29	6.3	1.68	121	1.3**	0.52	24	3.9	0.92	145
30–44	5.2	0.36	1653	1.8**	0.20	598	3.5	0.23	2251
45–64	3.0	0.31	929	0.9**	0.16	302	1.9	0.18	1231
> 65	0.3	0.10	47	—b	—	26	0.2	0.07	728

^a Population estimates are in thousands.
^b Estimate does not meet precision standard.
* p < 0.05 for gender comparison.
** p < 0.01 for gender comparison.

Changes in Prevalence of Driving After Drinking Between 1991– 1992 and 2001–2002

In the total sample, the prevalence of driving after drinking significantly declined 22 percent over the past decade, from 3.7 percent to 2.9 percent (Table 3). Significant declines in driving after drinking were also observed among males (5.8 percent versus 4.4 percent) and Whites (4.1 percent versus 3.3 percent). There were no significant changes in the rate of driving after drinking observed among females. With regard to age groups in the total sample, the only significant decreases in the rates of driving after drinking were among 18- to 29-year-olds overall and among 18- to 29-year-old Whites.

Table 3. Changes in the 12-Month Prevalence of Driving After Drinking by Age, Sex, and Race: 1991–1992 and 2001–2002

Race–ethnicity/age (years)		Male				Female				Total
		NLAES (1991–1992)		NESARC (2001–2002)		NLAES (1991–1992)		NESARC (2001–2002)		NLAES (1991–1992)		NESARC (2001–2002)
		%	S.E.	%	S.E.	%	S.E.	%	S.E.	%	S.E.	%	S.E.
Total
	Total	5.8	0.22	4.4**	0.22	1.7	0.11	1.5	0.11	3.7	0.13	2.9**	0.13
	18–29	11.6	0.53	7.8**	0.57	4.2	0.34	2.9*	0.30	7.9	0.33	5.3**	0.33
	30–44	5.8	0.34	5.2	0.36	1.8	0.18	1.8	0.20	3.8	0.20	3.5	0.23
	45–64	3.0	0.33	3.0	0.31	0.4	0.08	0.9**	0.16	1.6	0.16	1.9	0.18
	> 65	0.3	0.11	0.3	0.10	—a	—	—	—	—	—	0.2	0.07
White
	Total	6.4	0.26	5.0**	0.25	2.0	0.12	1.7	0.13	4.1	0.15	3.3**	0.15
	18–29	13.9	0.67	9.9**	0.76	5.5	0.42	3.5**	0.38	9.8	0.42	6.7**	0.42
	30–44	6.7	0.41	6.1	0.45	2.0	0.21	2.5	0.28	4.3	0.24	4.2	0.29
	45–64	3.1	0.36	3.4	0.38	0.4	0.10	1.0**	0.19	1.7	0.18	2.2	0.22
	> 65	0.3	0.13	0.3	0.12	—	—	—	—	—	—	0.2	0.06
Black
	Total	2.9	0.46	2.6	0.35	0.6	0.16	0.7	0.17	1.7	0.23	1.5	0.19
	18–29	4.7	1.07	3.1	0.74	0.4	0.20	1.2	0.45	2.4	0.54	2.1	0.42
	30–44	2.8	0.69	3.5	0.74	1.5	0.44	0.4*	0.15	2.1	0.40	1.8	0.33
	45–64	1.9	0.73	1.8	0.43	—	—	—	—	0.9	0.33	1.3	0.30
	> 65	—	—	—	—	0.0	0.00	0.0	0.00	—	—	—	—
Native American
	Total	4.2	2.05	5.9	1.61	5.8	2.40	2.4	0.92	5.1	1.49	4.1	0.84
	18–29	—	—	—	—	13.2	5.46	—	—	9.2	3.15	5.8	2.24
	30–44	—	—	9.9	3.25	—	—	—	—	—	—	5.5	1.70
	45–64	—	—	—	—	0.0	0.00	—	—	—	—	2.1	0.50
	> 65	0.0	0.00	—	—	0.0	0.00	—	—	0.0	0.00	—	—
Asian
	Total	1.1	0.51	2.0	0.61	1.2	0.77	0.7	0.34	1.2	0.46	1.4	0.33
	18–29	2.3	1.39	4.3	1.63	—	—	2.6	1.29	2.8	1.33	3.5	1.00
	30–44	—	—	—	—	—	—	—	—	—	—	—	—
	45–64	0.0	0.00	—	—	0.0	0.00	0.0	0.00	0.0	0.00	—	—
	> 65	0.0	0.00	0.0	0.00	0.0	0.00	0.0	0.00	0.0	0.00	0.0	0.00
Hispanic
	Total	5.4	0.79	3.3*	0.48	0.6	0.19	0.9	0.18	3.0	0.41	2.1	0.27
	18–29	8.8	1.49	5.0*	1.00	0.6	0.25	2.0*	0.50	4.7	0.77	3.6	0.60
	30–44	3.6	0.99	3.1	0.59	1.2	0.48	0.6	0.16	2.4	0.57	1.9	0.32
	45–64	—	—	1.5	0.42	—	—	—	—	2.1	1.00	0.9	0.22
	> 65	0.0	0.00	0.0	0.00	0.0	0.00	0.0	0.00	0.0	0.00	0.0	0.00

^a Estimate does not meet precision standard.
* p < 0.05.
** p < 0.01.

Among males, significant decreases in the rates of driving after drinking were observed among Whites overall (6.4 percent versus 5.0 percent), Hispanics overall (5.4 percent versus 3.3 percent), and among 18- to 29-year-old males (11.6 percent versus 7.8 percent), 18- to 29-year-old Whites (13.9 percent versus 9.9 percent), and 18- to 29-year-old Hispanics (8.8 percent versus 5.0 percent).

The picture was different for females. Significant reductions in the rates of driving after drinking were noted for 18- to 29-year-old females overall (4.2 percent versus 2.9 percent) and 30- to 44-year-old Black females (1.5 percent versus 0.4 percent) over the last decade. In contrast, small but significant increases in the corresponding rates were observed among females ages 45–64 (0.4 percent versus 0.9 percent) and among 18- to 29-year-old Hispanic females (0.6 percent versus 2.0 percent).

To further examine changes in driving after drinking among 18- to 29-year-olds, we disaggregated age into 1-year intervals for that age group (Table 4). In the total sample, significant declines in the prevalence of driving after drinking were only observed among 18-, 26-, 27-, 28-, and 29-year-olds. Among males, the rate of driving after drinking significantly decreased only among 21-, 26-, 27-, and 29-year-olds.

Table 4. Changes in the 12-Month Prevalence of Driving After Drinking by Sex and Age: 1991–1992 and 2001–2002

Age (years)	Male				Female				Total
	NLAES (1991–1992)		NESARC (2001–2002)		NLAES (1991–1992)		NESARC (2001–2002)		NLAES (1991–1992)		NESARC (2001–2002)
	%	S.E.	%	S.E.	%	S.E.	%	S.E.	%	S.E.	%	S.E.
Total	5.8	0.22	4.4**	0.22	1.7	0.11	1.5	0.11	3.7	0.13	2.9**	0.13
18	8.3	2.00	5.5	1.43	4.1	1.30	1.0*	0.58	6.2	1.17	3.4*	0.80
19	7.5	1.75	6.4	1.50	5.8	1.90	2.8	0.94	6.7	1.28	4.8	0.91
20	10.3	1.80	9.3	2.34	5.0	1.24	1.8*	0.72	7.9	1.10	5.3	1.17
21	14.0	2.48	7.2*	1.60	3.1	0.87	5.7	1.67	8.0	1.24	6.4	1.18
22	15.6	2.25	11.5	2.12	6.0	1.35	4.2	1.31	10.8	1.38	7.8	1.26
23	12.6	2.05	10.4	2.16	5.4	1.20	7.1	1.48	9.2	1.25	8.7	1.29 <