Introduction

S

ingle episodes of drinking, persistent alcohol abuse, and alcohol dependence can result in a variety of adverse social consequences that can affect a host of people, ranging from an individual drinker to society as a whole. For example, alcohol can play a significant role in various accidents, including motor vehicle crashes and falls. Such consequences exact significant human suffering and economic costs (see Chapter 11, Economic Issues in Alcohol Use and Abuse).

Study findings reflect the seriousness of alcohol's role in accidental injuries and deaths. Popham et al. (1984) noted that the estimated relative risk of accidental death was 2.5 to 8 times greater among males defined as heavy drinkers or alcohol dependent than among the general population. Estimates suggest that alcoholics are nearly 5 times more likely than others to die in motor vehicle crashes, 16 times more likely to die in falls, and 10 times more likely to become fire or bum victims (Eckardt et al. 1981).

Anda et al. (1988) used data from a sample of 13,251 noninstitutionalized U.S. adults included in the National Health and Nutrition Examination Survey Epidemiologic Follow-up Study to estimate relative risk of fatal injury. This study is noteworthy because baseline data on alcohol use were collected approximately 10 years before followup data on fatal injuries. After adjusting for gender, age, race, and education, the investigators found that the relative risk of death from injury was significantly greater among

Single episodes of drinking, persistent alcohol abuse, and alcohol dependence can result in a variety of adverse social consequences that can affect a host of people, ranging from an individual drinker to society as a whole.

those reporting five or more drinks per occasion than among those who reported drinking fewer than five drinks per occasion (Anda et al. 1988). Among individuals reporting nine or more drinks per occasion, the relative risk of fatal injury was more than three times greater than among those reporting fewer than five drinks per occasion. Self-reported quantity of consumption per drinking occasion was found to be a stronger predictor of fatal injury than frequency of consumption.

This chapter highlights findings from studies that address the relationship between alcohol use and a variety of adverse social outcomes. Included are brief discussions of the role of alcohol in motor vehicle, aviation, and water safety. Alcohol involvement in personal and criminal violence and high-risk sexual behavior is also considered.

Alcohol and Trauma:
An Overview

Alcohol use and the impairment that results from it have been empirically linked to an array of

serious and fatal injuries, including spinal cord injuries (Branche et al. 1991), traumatic brain injuries (Jones 1989), and injuries incurred in general aviation crashes (Gibbons 1988), drownings (Wintemute et al. 1990), bicycle crashes (Olkkonen and Honkanen 1990), spousal violence (Miller et al. 1989), suicides (Gomberg 1989), and homicides (Welte and Abel 1989). Recent studies based on medical examiner reports and records of emergency room patients present a panoramic perspective on alcohol-related injury.

Various factors have complicated efforts to obtain precise assessments of acute alcohol involvement in trauma events. Until recently, for example, the World Health Organization's International Classification of Diseases (ICD) included diagnostic codes appropriate only for identifying chronic alcohol-related conditions. The inclusion of alcohol involvement codes (code Y90, “evidence of alcohol involvement determined by blood alcohol content," and code Y91, "evidence of alcohol involvement determined by level of intoxication") in the 10th revision of the ICD (ICD-10) promises to facilitate measurements of alcohol's role in accident-related injuries and casualties (World Health Organization 1990).

Studies of Medical Examiner/Coroner Reports

Studies based on coroner reports allow direct comparisons of alcohol involvement in different fatal injury events. These studies also permit comparisons by gender, age, and ethnicity. Some caution should be exercised when interpreting results from these studies because not all victims are tested for blood alcohol, and testing rates may vary by injury type and by the idiosyncrasies of the events. Events leading to fatal injury are not necessarily representative of less severe injury events.

Goodman et al. (1991) examined death certificates for all intentional (n=5,776) and unintentional (n=10,624) injury deaths in Oklahoma between 1978 and 1984. Unintentional injury deaths include all injuries (such as fatal traffic crashes) except suicide and homicide. Blood alcohol testing was performed on 83 percent of homicide victims and 56 percent of unintentional injury victims. Evidence of alcohol was found in approximately half of the homicide and unintentional injury victims tested (figure 1). Among those tested, nearly 24 percent of suicide victims, 34 percent of homicide victims, and 38 percent

of unintentional injury victims had blood alcohol concentrations (BACS) of .10 percent (the legal level of intoxication in many jurisdictions) or greater. Only 20 percent of individuals who died of natural causes were alcohol positive. Goodman et al. (1991) note that because BAC testing for victims of suicide and unintentional injury deaths was selective, varying by the victim's age, by the time of day of the injury, and by the county in which the injury occurred, biases are present in the data. Accordingly, the researchers suggest that the results can be generalized to all injury deaths only with caution.

Smith et al. (1989) reported similar results in an analysis of more than 100,000 deaths investigated by North Carolina medical examiners from 1973 to 1983. BACs were determined for approximately 64 percent of the victims. Testing varied by victims' ages, gender, and manner of death. Homicide victims were tested most frequently (86 percent), followed by victims of suicide (78 percent) and unintentional injury death (68 percent). Positive BACS were detected in nearly 50 percent of unintentional injury deaths and approximately 35 percent of suicide deaths. Higher rates of alcohol involvement (63 percent) were observed among homicide victims. BACS exceeding the legal limit of .10 percent were reported for approximately 40 percent of unintentional injury victims, 27 percent of suicide victims, and 52 percent of homicide victims. Only 8 percent of those who died of natural causes had BACS of .10 percent or greater. The researchers noted that because alcohol and drug testing is performed routinely in North Carolina, regardless of the manner of death, this practice increases availability of BAC results and enables definition of the relationship between alcohol and all fatal injuries.

Although rates of alcohol-related traumatic injury deaths vary significantly by day and time, the pattern is similar to day-and-time variations reported for alcohol-related motor vehicle fatalities (National Highway Traffic Safety Administration (NHTSA) 1991). Day-to-day variability in the proportion of alcohol-related homicide and unintentional injury deaths is particularly striking. On weekends, a significant proportion of traumatic injury deaths involve alcohol. Indeed, Smith et al. (1989) observed that more than 60 percent of the homicides occurring on Saturdays were alcohol related, compared with approximately 47 percent of homicides occurring on Wednesdays. A similar pattern was observed for deaths from unintentional injuries.

When assessing variability by time of day, several researchers reported that rates of fatal injury events involving alcohol are lowest in the morning, slightly higher in the afternoon, and significantly higher during evening and late-night hours (Goodman et al. 1991; Smith et al. 1989). In particular, traumatic injury deaths between 9 p.m. and 3 a.m. are likely to involve alcohol. Within 24-hour time periods, alcohol-related unintentional injury deaths and homicides exhibit the most pronounced time variations. Researchers found comparatively small day or time variations in alcohol involvement among those who died of natural causes (Goodman et al. 1991; Smith et al. 1989).

Evidence also indicates that alcohol involvement in fatal injury events is associated with victims' gender and ethnicity. Males are disproportionately represented in most serious and fatal injury events. Moreover, the probability of alcohol involvement in fatal injury events is considerably higher for male victims (Goodman et al. 1991; Smith et al. 1989).

Alcohol appears to be a prominent factor in traumatic injury deaths among Native Americans. In this population, however, alcohol involve

ment in traumatic injury deaths varies little between men and women, in contrast to the sharp gender differences reported for other ethnic groups. Goodman et al. (1991) found that approximately 80 percent of unintentional injury deaths among Native Americans were alcohol related, regardless of gender. These investigators also reported that a high proportion of homicides and suicides among Native Americans involved alcohol. Yet it is important to note that rates of alcohol-involved deaths vary among tribes. Christian et al. (1989) observed a considerable disparity in alcohol-related deaths among 11 Oklahoma tribal groups.

Hispanic males are more likely to experience drinking problems than African-American or white males (Caetano 1986) (see Chapter 1, Epidemiology of Alcohol Use and AlcoholRelated Consequences). In addition, Hispanic males seem especially susceptible to alcoholrelated traumatic injury death. For example, Goodman et al. (1991) reported that alcohol was involved in a greater proportion of homicide deaths among Hispanic men than among white or African-American men. Nearly 90 percent of male Hispanic homicide victims tested alcohol positive, and approximately 70 percent had BACS of .10 percent or greater (Goodman et al. 1991). Alcohol was detected in more than half

breath alcohol concentrations to be supplemented with self-report information on drinking behavior before and after the precipitating event. The reported prevalence of alcohol involvement among emergency room patients usually ranges from 15 percent to 25 percent (Cherpitel 1989a, 1989b; Holt et al. 1980; Wechsler et al. 1972; Yates et al. 1987a, 1987b). These differences in prevalence estimates may stem from the differences in populations served by emergency rooms in different locations or from variations in sampling designs and subject selection criteria among studies.

The reported prevalence of alcohol involvement among emergency room patients usually ranges from 15 percent to 25 percent.

Teplin et al. (1989) evaluated alcohol involvement patterns in a sample of 640 patients presenting for treatment at the emergency room of Northwestern Memorial Hospital in Chicago. During a 9-month period, data were collected 7 days a week from 4 p.m. to 12 p.m.-hours when alcohol use tends to be high. Approximately 24 percent of the patients tested alcohol positive, and 16 percent had BACs of .10 percent or greater. The rate of alcohol involvement found in this study was somewhat higher than in other emergency room studies and may have

resulted from the hours during which data were collected. Findings from this study indicated that several variables had a statistically significant effect on BAC at the time of emergency room presentation. Specifically, unemployment, male sex, age range 31 to 54 years, lower education, and injury significantly contributed to higher BAC levels. With regard to race, African-American males showed significantly higher BACS than white males. No differences in BACs were found between African-American and white females.

In another emergency room study, Cherpitel (1989a) analyzed BACS and measures of selfreported alcohol consumption collected from patients admitted to the San Francisco General Hospital emergency room. Patients in the study were sampled systematically on a 24-hour basis for 3 months. Of the 2,516 subjects selected, 1,896 provided breath samples and interviews. No significant demographic differences were found between those interviewed and those not interviewed. Cherpitel (1989a) noted that BAC measurements taken at the time of emergency room presentation could seriously misrepresent BACS at the time of the event, because only 36 percent of study patients arrived at the emergency room within 6 hours of the precipitating event. To circumvent this problem, she separately analyzed BACS obtained within 6 hours of emergency room arrival (n=1,747) and BACs obtained within 6 hours of the event for those who reported no drinking after the event (n=427). These analyses yielded similar results. Measures of self-reported alcohol consumption were used to supplement breath alcohol measurements.

Consistent with other emergency room studies (Teplin et al. 1989; Wechsler et al. 1969), Cherpitel (1989a) found that patients who presented as a result of injury rather than illness were more likely to test alcohol positive (figure 2). Males were more likely to present in emergency rooms than females, and the relationship between alcohol involvement and injury was stronger among males than among females. Injured males were nearly twice as likely as noninjured males to have a BAC of .10 percent or greater (figure 2). Analysis of breath alcohol samples taken within 6 hours of the event suggested that alcohol involvement rates did not differ between injured and noninjured females.

In addition to measuring breath alcohol levels, Cherpitel (1989a) conducted short interviews with emergency room patients. Approximately 23 percent of patients reported drinking in the 6 hours before being injured or, for the nonin

jured, in the 6 hours before realizing they had a medical problem. Regardless of injury, males were more likely than females to report drinking. Those seen because of injury were nearly twice as likely to report drinking in the 6 hours before the event (figure 3). The magnitude of the relationship between injury and self-reported alcohol consumption was similar for males and females.

Cherpitel (1989a) also conducted a detailed analysis of alcohol involvement in injuries of different types and causes in this study. This analysis revealed that although injury type was not significantly associated with 6-hour breath analyzer readings or self-reported alcohol consumption, cause of injury was significantly associated with alcohol consumption. Figure 4 presents rates of self-reported alcohol involvement and breath analysis readings for seven categories of injury cause examined by Cherpitel (1989a). As in other studies (Cherpitel 1989b; Yates et al. 1987 a), alcohol involvement was found to be especially prevalent among those injured in fights and assaults. Comparatively high rates of alcohol involvement for injuries caused by motor vehicle collisions and falls were also found (figure 4).

Findings from a study by Goodman et al. (1991) indicated that alcohol involvement varied by type of unintentional injury death. More than half the victims of fatal automobile crashes who were tested had positive BACs. Approximately 35 percent of fatalities due to falls and 43 percent of those due to burns were alcohol related. In addition, alcohol involvement in homicides and suicides varied depending on the instrument causing death. More than 60 percent of homicide victims killed by cutting instruments tested alcohol positive, and alcohol was detected in 52 percent of those killed by firearms. Regardless of the method, more than 40 percent of homicide victims had positive BACS. High rates of alcohol involvement were also found among suicide victims who used firearms. However, these findings may be confounded by the fact that BAC testing may vary according to the sex and age of victims. Smith et al. (1989) reported that for all manners of death, men were tested more frequently than women (68.6 percent compared with 61.9 percent). In addition, individuals 15 through 59 years of age were tested more frequently than those 60 years or older.

Studies based on emergency room patients and medical examiner reports document high rates of alcohol involvement in a range of fatal and nonfatal traumatic injuries. These studies

also demonstrate that rates of alcohol involvement are higher for injury than noninjury victims, and that alcohol involvement rates vary by type of injury. These studies strongly suggest that alcohol may increase injury risk in a broad range of human activities. But demonstrating high rates of alcohol involvement in traumatic injury allows only weak causal inferences, because the number of injuries that would have occurred without alcohol involvement is usually unknown. With few exceptions, research outside the traffic safety arena has not used methodologies allowing direct estimates of increased risk.