OLR Research Report

January 11, 2006




By: James J. Fazzalaro, Principal Analyst

You asked for an explanation of the so-called “Fatal Fifteen” list released to the press at the end of November 2005 by the National Safety Council and an interest group that calls itself “End Needless Death on Our Roadways” (END). The list purports to show the 15 “deadliest” states in the country for alcohol impaired driving based on a ranking of each states' percentage of alcohol related motor vehicle fatalities. You wanted background on this list and an assessment of its meaning and significance with respect to Connecticut's inclusion on the list.


The “Fatal Fifteen” list released jointly by the National Safety Council and END on November 28, 2005 claims to identify the 15 “deadliest” states in the country with respect to alcohol impaired driving based on the fact that the jurisdictions that made their list all had alcohol-related fatalities in 2004 that comprised 41% or more of their total traffic fatalities as reported through the federal Fatality Analysis Reporting System. They identified Connecticut as one of the jurisdictions that made their list in both 2003 and 2004.

The National Highway Traffic Safety Administration (NHTSA) maintains the database from which these statistics are drawn and it is generally recognized as the primary source for traffic fatality data. NHTSA defines a fatality as “alcohol-related” if either a driver or non- motorist involved in the accident had any measurable or estimated blood-alcohol concentration (BAC) of .01% or more. When dealing with “alcohol-related” statistics using this definition, it is important, in order to avoid misinterpreting their meaning, to recognize several of the characteristics of the definition including:

1. it counts equally both those with alcohol levels that constitute intoxication under the law (.08%) and those with alcohol levels typically produced by consuming only one drink;

2. it does not account for causation, i.e., whether the alcohol present in the accident participant actually contributed to or caused the accident is not determined;

3. not all fatalities classified as alcohol related are based on a known presence of alcohol—for a significant portion of fatalities NHTSA classifies it as either alcohol-related or not through a statistical manipulation that attributes a specific BAC level to the participant based on other characteristics of the accident, such as the day and time, type of vehicle and its role in the accident, age and sex of participant, restraint use, and previous traffic violations of the participant.

The two organizations publishing the “Fatal Fifteen” list assert that states with higher percentages of alcohol-related fatalities are inherently more dangerous than those with lower percentages of alcohol-related fatalities. However, NHTSA warns in its literature that “great caution” must be used in comparing levels of alcohol involvement among states.

The characterization of states as either more or less dangerous with respect to alcohol-impaired driving based entirely on the relative percentage of all their traffic fatalities that are alcohol-related according to NHTSA's definition is inherently flawed in that it relies entirely on a statistic the magnitude of which is completely dependent on how large a number of people die in traffic accidents who are not found to have consumed any alcohol. In 2004, Connecticut and Iowa experienced approximately the same number of alcohol-related fatalities (127 and 110 respectively), however Iowa had a total of 390 traffic fatalities compared to Connecticut's 291, even though Iowa has a smaller population and fewer licensed drivers than Connecticut. Thus Iowa's proportion of alcohol-related fatalities was 28.2% which made it the second “safest” state according to the Fatal Fifteen analysis while Connecticut's proportion of 43.6% made it the 13th “deadliest.” However, on a per capita basis, Connecticut's overall traffic fatality rate is less than 60% of Iowa's. Had Connecticut experienced traffic fatalities in 2004 at the same rate as occurred in Iowa, its percentage of alcohol-related fatalities would have been 32.6% instead of 43.6% and it would have been the ninth “safest” state instead of the 13th “deadliest.”

If state-to-state comparisons must be made, the appropriate way to make the comparison would be to compare the rate at which fatalities occur based on a uniform measure such as population, number of licensed drivers, number of registered vehicles, or, most appropriately, the number of vehicle miles of travel occurring in each state. Using any of these measures, Connecticut consistently ranks among the best overall fatality rates in the country. In 2003, its fatality rates of 11.05 per 100,000 licensed drivers and 9.71 per 100,000 registered vehicles were the 2nd lowest in the United States. Its fatality rates of 8.44 per 100,000 of population and 0.94 per 100 million vehicle miles traveled were the 3rd lowest in the United States. Its alcohol-related fatality rate of 0.42 fatalities per 100 million vehicle miles of travel was the 8th best (lowest) in the United States. Over the last five years, the lowest Connecticut's alcohol-related fatality rate has ranked has been 15th best among the 50 states, Puerto Rico, and the District of Columbia.


On November 28, 2005, the National Safety Council and END jointly released a list of what they claimed were the fifteen “deadliest” states in the country for alcohol impaired driving. They called this listing the “Fatal Fifteen” although it actually contains 17 U.S. jurisdictions not 15. The organizations claim that these 17 jurisdictions are the deadliest in the country for alcohol impaired driving because in each of them the percentage of all traffic fatalities that were “alcohol-related” in 2004 was 41% or more. The organizations do not explain why the 41% figure was selected as the threshold for their list as opposed to a higher or lower percentage of alcohol-related fatalities. Data for their list is taken from the National Highway Traffic Safety Administration's Fatality Analysis Reporting System database—the nationally recognized clearinghouse for traffic fatality data.

END is an organization headquartered in Chicago, Illinois that describes itself on its Internet website as “an organization of physicians and other medical professionals, dedicated to developing and promoting new and innovative strategies for addressing impaired and other dangerous driving behavior.” The National Safety Council describes itself as a “nonprofit, nongovernmental, international public service organization dedicated to protecting life and promoting health.” Besides appearing on the joint press release on the Fatal Fifteen listing, the National Safety Council also issued its own separate press release which made the same assertions and presents the same information as the joint press release with END.

The two organizations' “Fatal Fifteen” press releases state that last year (2004), nearly 17,000 motorists were killed nationwide in alcohol-related traffic crashes and that nearly 6,000 of those fatalities occurred in the “Fatal Fifteen.” Of note, however, is that had the list included only the 15 places with the highest percentage of alcohol-related traffic fatalities (14 states plus Puerto Rico) rather than 17 jurisdictions (D.C. and Pennsylvania in addition to the other jurisdictions), the total number of alcohol-related fatalities represented by the “Fatal Fifteen” would have been 4,490 rather than 6,000. Including the extra two jurisdictions allowed Pennsylvania's total of 1,490 alcohol-related fatalities to be included in the total specified in the press releases.


Although ranking one state against others may be a valid comparative exercise, it is important that the basis of the ranking be an appropriate method for comparison. The relative percentage of alcohol-related fatalities to total traffic fatalities used as the basis of the “Fatal Fifteen” comparison of the “deadliest” states for alcohol impaired driving is both an inadequate and somewhat misleading way to compare states for several reasons, as explained below.

Vagueness of the Term “Alcohol-Related”

As used by NHTSA, and consequently by most others, a fatality is considered to be “alcohol-related” or “alcohol-involved” if either a driver or a nonmotorist (usually a pedestrian or bicyclist) involved in the accident had any measurable or estimated blood alcohol concentration of .01% or more (emphasis added). (Although estimation methods vary, an average 170-pound male who consumes one 12-ounce beer in one hour typically would have a BAC of approximately .01%.) What the term means, in effect, is that some measurable amount of alcohol was found in the blood of someone associated with the accident. But the alcohol that causes the accident to be labeled alcohol-related does not have to be in the person who may have been considered responsible for the accident. For example, a driver who has had no alcohol may run a stop sign and hit another driver who may have a blood-alcohol content of .03%. Even though the driver who caused the accident had not consumed alcohol, the accident would still be considered “alcohol-related” because the driver who was hit had some alcohol in his blood. The same would be true if a driver who had consumed no alcohol struck an intoxicated pedestrian who walked in front of his vehicle. NHTSA estimates that 36% of all fatalities of a pedestrian over age 15 involve an intoxicated pedestrian, that is, the pedestrian has a BAC of .08% or more.

However, the NHTSA data also includes fatalities where NHTSA has attributed the fatality to alcohol-involvement even though there is no evidence or information that it was actually present in someone involved in the accident. NHTSA estimates that on average, in more than 50% of fatal accidents, alcohol involvement as determined by actual blood testing is not known. Whenever blood alcohol test results are not available or not known for all drivers and nonoccupants involved in a fatal accident, NHTSA decides whether or not to consider it an alcohol-involved accident by applying a statistical formula that assigns an alcohol status to a driver or nonoccupant of “sober,” “had some alcohol,” or “intoxicated” based on the characteristics of the accident and not on actual knowledge of the presence of alcohol in someone.

It determines this based on the time of day, day of the week, crash type and location; the type of vehicle and its role in the accident; the age, sex, restraint use, and previous driving violations of those involved in the accident; and whether or not the state has a 21-year-old minimum drinking age law. Thus, for example, a single vehicle accident involving a 25-year old male driver with a prior driving violation who runs off the road on a weekend night and who was not wearing his seat belt is highly likely to be assigned as an alcohol-involved accident while an accident involving a 45-year old female driver without prior violations who was wearing a seat belt and ran off the road at 5:00 PM on a weekday afternoon probably would not be assigned as an alcohol-involved accident. NHTSA does this statistical manipulation of the data because it believes alcohol-involvement would be under reported if the data relied solely on known alcohol test results.

Alcohol-involved means neither alcohol impaired nor intoxicated (as defined by most state criminal drunk driving laws), although it is being increasingly used somewhat carelessly and erroneously in some contexts to imply those conditions. NHTSA clearly states in many of its alcohol-related statistical compilations that the term “alcohol-related,” “does not indicate that a crash or fatality was caused by the presence of alcohol.” However, many organizations that make use of the NHTSA data do not make this distinction as clearly, if at all. NHTSA also states that, “Great caution should be exercised in comparing the levels of alcohol involvement among States. Differences in alcohol involvement can be

due to any of a number of factors not necessarily directly related to a State's alcohol traffic safety program.” (See “Traffic Safety Facts 2004 Data—State Alcohol Estimates”, DOT HS 809 916, p.1).

Some of the factors NHTSA identifies that affect alcohol involvement in fatal accidents include:

1. Population demographics and the economic environment

● Older drivers and female drivers have lower levels of alcohol involvement

● Drivers of older vehicles show higher levels of alcohol involvement

● Pedestrians killed in traffic crashes show high levels of alcohol involvement

2. Degree of urbanization

● Alcohol involvement in single and multiple vehicle crashes tends to be greater in urban fatal crashes

● Alcohol involvement in nonoccupant fatal crashes is higher in rural areas

3. Type of vehicle

● Motorcycle operators show high levels of alcohol involvement, followed by drivers of light trucks and vans

● Drivers of medium and heavy trucks show the lowest levels of alcohol involvement

One of the most significant differences between states in this area is the difference in the degree of blood alcohol testing for driver and nonoccupant BACs. These differences in testing affect the accuracy and reliability of the NHTSA alcohol involvement estimates. In 2004, the difference in testing ranged from a low of 7% known BACs in Alabama to a high of 82% known BACs in Nebraska. States with higher rates of known BACs yield more complete and accurate alcohol involvement estimates.

Connecticut is a relatively high BAC testing percentage state. The Connecticut Department of Transportation reports that in 2003, Connecticut recorded BAC test results for 86% of fatally injured drivers and 30% of surviving drivers involved in fatal accidents. Both of these testing rates were higher than the respective national averages of 66% and 24%. Also in 2003, Connecticut data show that 78% of fatally injured pedestrians and bicyclists over age 15 had known BACs compared to 61% nationally (The Connecticut Highway Safety Strategic Plan FY 2006, Connecticut Department of Transportation, p. 21). Connecticut's percentage of BAC-tested surviving drivers is likely to increase in future years since legislation adopted in 2003 makes it mandatory that law enforcement officers obtain a blood or breath sample from any surviving driver involved in an accident resulting in a death or serious physical injury of another person, whenever the officer has probable cause to believe that the surviving operator may have been operating his vehicle while under the influence of alcohol, drugs, or both.

Ineffectiveness of Alcohol-Related Fatality Percentage as a Comparative Measure

As discussed above, the term “alcohol-related” fatality is a vague and overly general term that captures both known and assumed BAC levels in an accident participant regardless of whether the presence of alcohol in such a participant has a causal link to the accident result. In many cases it may, but in some others it may not, but the accident is considered “alcohol-related” regardless. However, since it is used so exclusively in the statistical databases, and has more or less general acceptance among users of this data, it maintains its existence as the primary way to categorize accidents where alcohol is present and where it is not.

Consequently, in terms of this analysis, although a more precise way of determining whether the presence of alcohol had a causal impact on the fatality data would be preferable, use of the term “alcohol-related” as it exists in the statistical databases is more or less unavoidable. But it should be used with a clear understanding of its limitations as a means of conducting state-by-state comparisons such as the “Fatal Fifteen.” The question then becomes what the most valid way to use the data in such a comparison would be.

Using the percentage of one state's traffic fatalities that are “alcohol-related” to compare it to another state's percentage and thereby determine which state is “deadlier” is, we believe, an inadequate basis for comparison. NHTSA has recognized this as noted above. The reason this is not an appropriate way to make such comparisons is simple. Since a traffic fatality must be either alcohol-related or not alcohol-related under this scheme, a particular state's percentage of all of its traffic fatalities that are alcohol-related depends entirely on the number of fatalities out of the total that are not “alcohol-related.” Thus, the more fatalities a state experiences that do not include an alcohol-involved participant, the better (or lower) its percentage of alcohol-related fatalities will appear to be, even if the rate at which these are occurring is higher.

To illustrate, State A might experience 100 traffic fatalities that are determined to be alcohol-related and 150 that are determined not to be alcohol-related. In that case, its percentage of alcohol-related traffic fatalities would be 40%. However, State B might also experience 100 alcohol-related traffic fatalities but have 200 that are not alcohol-related. State B's percentage of alcohol-related fatalities would be 33% rather than 40%. Applying the logic used by the National Safety Council and END in their Fatal Fifteen listing, the likely conclusion would be that State A is “deadlier” than State B in terms alcohol impaired driving, even though the actual number of alcohol-related fatalities in each state is identical.

The effect that the number of non alcohol-related traffic fatalities has on a state's placement on a list based on alcohol-related fatality percentage becomes even clearer when a concrete comparison is made between Connecticut and another state on the list. Connecticut's alcohol related fatality percentage in 2004 was 43.6% based on a total 291 traffic fatalities of which 127 were considered to be alcohol-related. One of the states that was closest to Connecticut's number of alcohol-related fatalities in 2004 was Iowa. It had 110 alcohol-related traffic fatalities, but had a fatality total of 390, almost 100 more than Connecticut. Consequently, Iowa's percentage of alcohol-related traffic fatalities was only 28.2%, which according to the rationale used in the Fatal Fifteen list made it the second best state in the country behind only Utah. However, had as many non-alcohol-related fatalities occurred in Connecticut as occurred in Iowa in 2004, Connecticut's alcohol-related fatality percentage would have been 32.6% instead of 43.6%, which would have made it the ninth “safest” state instead of the 13th “deadliest” state with respect to impaired driving. Thus it becomes clear that any state-by-state comparison that considers only states' percentage of alcohol-related fatalities and not the rate of such fatalities introduces a bias under which the more non-alcohol related fatalities experienced, the better a state is likely to look with respect to its alcohol-related fatality percentage, even if this is not the case.

Evaluating Connecticut on the Basis of Fatality Rates

Although the percentage of alcohol-related fatalities is not an appropriate way to compare states against one another, such comparisons have more validity when fatality rates, rather than absolute numbers of fatalities or percentages, are the basis for the comparison. By calculating each state's alcohol-related fatality rate, its relative number of alcohol-related fatalities can be expressed against a common measure—whether that is population, number of licensed drivers, number of registered vehicles, or vehicle miles traveled (VMT). While any of these can be a useful measure, VMT is generally the most frequently used because actual travel is one of the best measures of accident exposure.

Based on any of these measures, Connecticut has one of the best fatality rates in the country. In 2003, the most recent year for which NHTSA has calculated state fatality rates, Connecticut fatality rates were as follows:

● Fatalities per 100,000 Population—8.44 (3rd lowest in U.S.)

● Fatalities per 100,000 Licensed Drivers—11.05 (2nd lowest in U.S.)

● Fatalities per 100,000 Registered Vehicles—9.71 (2nd lowest in U.S.)

● Fatalities per 100 Million VMT—0.94 (3rd lowest in U.S.)

As shown in Table 1 below, Connecticut's ranking among the best fatality rates in the United States is consistent over time. Since 1998, the lowest Connecticut has ever ranked in any of the four rate measures has been seventh best in the United States.

Table 1: Connecticut Overall Traffic Fatality Rate and Rank Among U.S. Jurisdictions—1998 Through 2003


Fatalities per 100,000 Population


Fatalities per 100,000 Licensed Drivers


Fatalities per 100,000 Registered Vehicles


Fatalities per 100 Million VMT



8.44 (3rd)

11.05 (2nd)

9.71 (2nd)

0.94 (3rd)


9.31 (6th)

12.05 (3rd)

10.82 (4th)

1.03 (4th)


9.11 (5th)

11.77 (2nd)

10.51 (3rd)

1.01 (2nd)


10.41 (7th)

12.89 (3rd)

11.76 (5th)

1.1 (4th)


9.17 (6th)

12.68 (3rd)

10.67 (2nd)

1.0 (2nd)


10.05 (7th)

14.01 (4th)

11.96 (4th)

1.1 (3rd)

Connecticut also appears to have one of the better records among the states with respect to its rate of alcohol-related traffic fatalities based on vehicle miles of travel. In 2003, Connecticut's alcohol-related fatality rate was 0.42 alcohol-related fatalities per 100 million VMT—the eighth best fatality rate in the country behind only Utah, Vermont, Indiana, Massachusetts, New Hampshire, New Jersey, and New York.

Connecticut's relatively low alcohol-related fatality rate is also fairly consistent over time. As shown in Table 2 below, the worst alcohol-related fatality rate Connecticut has experienced since 1999 was the 15th best rate in the United States.

Table 2: Connecticut Alcohol-Related Fatality Rate per 100 Million Vehicle Miles Traveled 1999-2003


Connecticut Alcohol-Related Fatality


U.S. Alcohol-Related

Fatality Rate


0.42 (8th)



0.46 (10th)



0.52 (12th)



0.52 (15th)



0.45 (9th)


The significant difference between Connecticut's actual performance with respect to alcohol-related traffic fatalities and the inaccurate portrayal evidenced in a compilation such as the National Safety Council/END Fatal Fifteen list that is based on alcohol-related fatality percentages and not rates can best be seen by comparing Connecticut's alcohol related fatality rate against the rates for the 15 states that would be considered the “best” states according to their alcohol-related fatality percentage. Only four of the 15 states with the lowest percentage of alcohol-related fatalities had an alcohol-related fatality rate lower than Connecticut.