The unformatted htm text below is from

Copyright © 2004 by Leonard Evans

Information about

How to purchase

Table of Contents

Text

Reviews

The text below is © Leonard Evans.  Portions totaling not more than 400 words my be freely used with appropriate citation.
For more extensive use, permission in writing must be first obtained by emailing the author 

7 Older drivers

This  html version contains only the text (no figures, tables equations, or summary and conclusions) To check printed book appearance see pdf version of Chapter 1 or pdf version of Chapter 16.

Introduction
Older drivers have been the focus of an extensive and expanding literature, reflecting concerns that projected increases in the number of older drivers will increase societal harm from traffic crashes. , A 1998 review focusing on just one aspect of older-driver research lists 428 references. Much additional literature has appeared since then documenting relationships between advancing driver age and medical conditions relevant to driving, , including vision and cognition. Specific driver performance skills have been shown to decline with increasing age.
In the United States, influential organizations, including the National Highway Traffic Administration, the American Medical Association, the American Association for Retired Persons, and the American Automobile Association, have ongoing activities and policies relating to older drivers. Other motorized countries have similarly high interest in older drivers.
This chapter provides a broad epidemiological overview of how various risk measures change as drivers age. The approach is to display graphically many different rates as a function of gender and age. As discussed in Chapter 1, measures in traffic safety are nearly always rates. Different rates provide answers to different questions. The question, "Does risk in traffic increase as drivers grow older?" is insufficiently specific to have an answer. If the risk is specified as say, the risk that a driver will be killed in same distance of driving, then the question does have a quantitative answer, which will be provided below.
Although the main emphasis here is on older drivers, age is treated as a variable. While all quantitative results are presented in terms of this variable, the terms young and old are used in the text for expository convenience, just as one would use cold and hot if the variable were temperature. In neither case would quantitative definitions of these terms be necessary or helpful. Risks to older drivers are measured relative to risks to drivers of other ages, particularly risks to younger drivers. Hence risks to older and younger drivers are intrinsically components of the same broad technical question of how risk depends on age.
Two distinct types of risks
Two conceptually distinct types of risks change as drivers become older:
1. Risks to the drivers themselves.
2. Threats the drivers pose to other road users.
These risks are of a different nature. There is near universal agreement that there should be more, and stronger, laws to prevent people from harming others than to prevent them from harming themselves. Indeed, the view that there should be no laws aimed at preventing people from harming only themselves always finds some support. Public safety makes a stronger claim on public resources than does personal safety, some aspects of which are generally supported using personal resources. Differences between the risks we assume ourselves and threats we pose to others have important impacts on legislation, licensing policy, and police enforcement.
Changing risks drivers face as they age
The risks that drivers face as they age are examined in terms of the numbers of them killed per year while driving (Fig. 7-1) as recorded in FARS data. Drivers includes drivers of any motorized vehicle. The values plotted are the sums of values for the three years 2000, 2001 and 2002 divided by three. The resulting average number of deaths per year provides a more reliable estimate than data for a single year, especially at ages for which there are few fatalities in any one year. The peak numbers occur at age 19 years (plotted at 19.5) for both genders. The general pattern for drivers is similar to that for all fatalities shown previously in Fig. 3-9, p. 46. This is to be expected, as the majority of fatalities in US traffic are drivers. However, note how similar the patterns also are for non-drivers (Fig. 3-10, p. 46).

Figure 7-1. Number of drivers of any type of motorized vehicle killed per year versus driver gender and age. Average values for FARS 2000-2002.9
One of the reasons why the number of driver deaths declines with increasing age is that there are fewer people of older age in the population. It is therefore appropriate to examine the population rate, the number of deaths per million people (Fig. 7-2). This was derived using Bureau of the Census estimates for 2001, so this and other plots may be interpreted to apply to the central year, 2001. The population rate shows a U-shaped pattern between ages 20 and 80 that recurs for many rates. The rate begins to increase with age in the early 60s for men and women, peaking in the mid 80s at values close to the maximum values for younger drivers.

Figure 7-2. The population rate, the number of driver fatalities per million population versus gender and age. FARS 2000-20029 and US Bureau of the Census data for 2001.10
Drivers and licensed drivers
The population rate is based on dividing the number of drivers killed by the number of people in the population without regard to whether or not they are licensed drivers. Many people in the US do not have driver licenses, as is illustrated by the Federal Highway Administration data shown in Fig. 7-3. For ages prior to the mid 20s, males and females have remarkably similar probabilities of being licensed to drive. With increasing age women have increasingly lower rates of licensure than men. This reflects earlier social norms. A woman aged 85 in 2001 was born in 1916 and entered the years of license eligibility in the 1930s, a period when it was not so common for women to drive.


Figure 7-3. The number of driver licenses per capita, which is approximately the same as the probability that a person has a driving license. (A commercial truck driver, for example, may have multiple licenses). FHWA data for 2001.11

Some of the drivers included in Figs 7-1 and 7-2 were killed while driving without licenses. There are many reasons for not having a license, including being too young to be eligible, not applying, failing license tests, not renewing through forgetfulness or fear of failing tests, or having a license suspended or revoked for traffic law violation. The mix of reasons will be different for younger and older drivers. Figure 7-4 shows that the probability that a fatally injured driver did not have a valid driver license at the time of his or her fatal crash depends strongly on gender and age. Unlicensed drivers killed are overwhelmingly male, with FARS data for 2002 recording 3,524 males and 595 females - a male-to-female ratio of 5.9 to 1. The increases in Fig. 7-4 at older age may reflect that some older drivers whose licenses are revoked on medical or competency grounds still continue to drive, sometimes with fatal consequences. The increases at younger ages relate more to driving before applying for or obtaining a license.
The plot in Fig. 7-4 does not include data for 336 boy and 75 girl drivers aged 13 and younger, for a male-to-female ratio of 4.5 to 1 (more on this topic on p. 227-228). As all these drivers are below the age of licensure in all US states, their data would correspond to points at 100 percent. The first point plotted is for 14-year-old fatally injured drivers, where the percents without valid licenses are 93% for boys and 88% for girls. Seven US states issue learner-driver permits to drivers under age 15 (Table 8-1, p. 200), although the most common US practice is to issue learner permits at age 15.
The data plotted in Fig. 7-5, unlike in the earlier figures, include only those fatally-injured drivers who had valid driver licenses at the time of their fatal crashes. The license rate, the number of licensed drivers killed per million licensed drivers, shows a much clearer increasing trend with increasing age, the effect being similarly prominent for men and women. The downward trend in the population rate at the oldest ages (Fig. 7-2) was largely reflecting a lower rate of licensure.
Distance of travel
The 2001 National Household Travel Survey (NHTS) was a survey of travel for all members of about 66,000 households contacted by telephone between April 2001 and May 2002. This provided the estimates of the average distance of travel per driver in 2001 shown in Fig. 7-6.
The number of licensed drivers killed per billion kilometers of licensed driving, the distance rate, is estimated by multiplying the estimates in Fig. 7-6 by the numbers of licensed drivers. The distance rate (Fig. 7-7) varies so much more than the population or license rates that it is plotted on a logarithmic scale. Relationships similar to Fig. 7-7 are found for cars and light trucks separately using estimated travel distance derived from odometer readings of vehicles in NASS data. (p 68-74) Drivers older than 80 suffer more driver deaths for the same distance of travel than the highest values for drivers in the late teens and early twenties.

The reason why the distance rate increases at young and old ages so much more than the other rates is because of the variation in travel as a function
of driver age. One of the largest changes to occur as drivers age is that
they drive less. A major contribution to this is reduced commuting trips to places of employment. Such trips account for about half of personal travel. Another contribution comes from decisions to drive less because of concerns about driving competency in general, and more particularly at night, on freeways, or in inclement weather. In general, aging is accompanied by reductions in activities requiring travel. The data in Fig. 7-7 are for licensed drivers. As people become older it is increasingly likely that they will not have a license and cease driving entirely (Fig. 7-3). While safety is one component of what has been referred to as the older driver problem, another major factor is reduced mobility. ,
Fragility - increased risk of death in given crash as drivers age
Increases in the population, license, and distance rates with increasing age should not be interpreted to reflect increases in the driver's risk of crashing with aging. Such an interpretation misses the crucial point that the number of drivers of given gender and age killed is the product of two factors:
1. The number of involvements in severe crashes.
2. The probability that involvement proves fatal.
The first factor reflects influences due to use and behavioral factors, such as amount of driving, environmental conditions, roadway classification, intoxication, driver capabilities, and, most importantly, driver risk taking. Chapter 6 showed that when a crash occurs, the risk of death depends on the gender and age of the occupant. Alcohol consumption was also shown to affect survival as a factor separate from its much larger influence
on the risk of crashing. One behavioral factor that has a large influence on survival is the use of a safety belt. This does not depend too strongly on gender and age, the variables of interest in this chapter. We therefore
assume that when a crash occurs the dependence on survival is represented by the previously derived Eqn 6-8, p. 134, for male drivers and Eqn 6-9, p. 137, for female drivers. We use these relationships to estimate involvement rates in crashes of similar severity by considering crashes in a severity range greater than sufficient to likely kill 80-year-old male drivers, for which case Eqn 6-8 gives Rmale(80) = 4.54.
Consider the mix of crashes in which N fatalities occur to 80-year-old males. If these crashes were repeated keeping all factors except the drivers the same, then we would expect N/(4.54) fatalities for 20-year-old male drivers and,
by using Eqn 6-9, 1.311´N/(4.54) = N/(3.46) fatalities for 20-year-old female drivers. In order to obtain the same number of fatalities, 4.54 times as many crashes by 20-year-old male drivers, and 3.46 times as many crashes by 20-year-old female drivers are required. In this way we can use the observed numbers of fatalities to infer involvement rates in crashes in a severity range sufficient to likely kill 80-year-old male drivers.
Figure 7-8 shows the number of involvements in similar severity crashes per licensed driver versus gender and age. This is obtained from the licensed driver deaths in Fig. 7-5 by the calculation described above. The point for 80-year-old males has the same value in Fig. 7-8 as in Fig. 7-5, but the value of the point for 20-year-old males in Fig. 7.8 is 4.54 times that in Fig. 7-5. In contrast to the earlier figures, there is only a modest increase in involvements in severe crashes per licensed driver with increasing age, showing that a major component of increasing fatality risk with increasing age is greater risk of being killed in the same crash. Also note that the number of severe crashes per licensed driver for older drivers never approaches even close to the high values for younger drivers.
Severe crash involvements for the same distance of travel (Fig. 7-9) increase with increasing driver age for ages above about 60. However, the increase is modest compared to that for the distance fatality rate (Fig. 7-7). Even for the oldest age category (85 and older) severe crash involvements for the same distance of travel are less than that for the youngest age category (16-20 years).

Threat to other road users
The threat drivers impose on other road users is estimated by examining the number of pedestrians killed as a function of the gender and age of the involved driver. Only single-vehicle crashes are included, so a typical case will be a vehicle striking and killing a pedestrian but without injuring the driver. No assumption is made regarding legal responsibility in pedestrian fatality crashes. FARS data show about one third of fatally injured pedestrians have blood alcohol concentrations that would be illegal for drivers (there are no laws relating to blood alcohol concentration for pedestrians or passengers, only laws against behavior in public resulting from intoxication). I suggest in Chapter 16 that regardless of the behavior of the pedestrian, there should be a default legal presumption of driver responsibility.
The number of pedestrians killed versus the gender and age of the involved driver shows that the main threat to other road users is overwhelmingly from young drivers (Fig. 7-10). Even after adjusting for the fewer numbers of older people in the population, it is still the young who pose the greatest threat to other road users (Fig. 7-11). Another way of interpreting Fig. 7-11 is that it represents the threat to other road users of the average individual of given gender and age, and that the greatest threat comes from 20-year-old males.
The number of pedestrians killed per licensed driver (Fig. 7-12) shows no more than a modest increase with increasing age at the oldest age interval
(85 years and older) for women drivers. In general, a license holder of almost any younger age poses a greater threat to other road users than even the oldest license holder for whom data are available.
For the same distance of travel, drivers pose increasing risks to other road users as they age past about 60 (Fig. 7-13). However, even at ages past 80 they still pose lower risks than do 20-year-old drivers.
Pedestrian involvements in fatal and severe crashes
Above we noted that as people become older they drive less. Some driving may convert into more walking. In this section we examine how pedestrian risks depend on pedestrian age.
Figure 7-14 shows the distribution of pedestrian fatalities by pedestrian gender and age. The same data normalized by population are shown in Fig. 7?15. After age 60 the risk of pedestrian death per person increases steeply to a peak, and then declines (likely reflecting reduced walking).

Part of the large increase in pedestrian fatalities per capita at older ages in Fig. 7-15 is due to the greater likelihood that the older person is killed by an impact that would not kill a younger person. In order to estimate the risk that a pedestrian is struck by a vehicle, as distinct from the consequences of being struck, we again use the relationships between risk of death from the same impact and gender given in Eqns 6-8 and 6-9 (p. 134 and 137). Figure 7-16 shows the number of pedestrian involvements in crashes in the severity range equal to or greater than would likely kill an 80-year-old male pedestrian. Like the driver fatality data, the pedestrian fatality data show peaks at the late teens or early 20s. The increasing involvement in severe pedestrian crashes with increasing age at ages above about 65 probably reflects decreasing perceptual skills and agility, and also perhaps increased pedestrian exposure related to driving less. The declines after age 80 likely reflect reduced activity.
Cross sectional compared to longitudinal analyses
All the above analyses, in keeping with nearly all analyses of age effects in traffic, have been cross sectional. That is, rates for people of different ages are compared in a specific year. While results have been discussed in terms of drivers aging, strictly speaking cross sectional analysis cannot address what happens as individuals age. Apparent aging effects would be generated even if every driver were assigned a risk at birth that remained unchanged throughout his or her life, but drivers born a long time ago were given higher rates than those born recently. The overall declines in fatality rates over time presented in Chapter 3 indeed suggest that today's drivers of a given age have lower risks than drivers of the same age in earlier decades.
Studies that trace the characteristics of a given group, or cohort, of people as they age are referred to as longitudinal. There is a convention in epidemiology to connect data points only for longitudinal data, but not for cross sectional data, as we have done in all the figures. The non-adherence to this convention in
the interests of clarity is unlikely to generate ambiguity. In order to explore how the risks of a group of 20-year-old drivers changed as they aged into 85-year-old drivers, data from 1935 through 2001 would be required, but the earliest
FARS data are for 1975. Even if the data were available, many other factors that influence safety have changed dramatically since 1935, so that there does not seem any way, even in principle, to estimate how a driver's risk changes with aging that is free from substantial uncertainty. It is cross sectional
data, such as presented here, that provide information relevant to policy decisions for today's population. Decisions rarely require information about how safe today's 80-year-olds were 60 years ago, or how safe today's 20-year-olds will be in 60 years. It is today's drivers who are licensed, or have their licenses revoked.
While not so central to policy decisions, longitudinal analyses add to understanding of how aging impacts driving, and can add insights into long-term future trends. One longitudinal study tracing a group of drivers over 16 years of FARS data (1975-1990) found that risks increased later and less steeply than in cross sectional comparisons. Another longitudinal study examined how crash types varied for drivers over age 60 in 1987-1995 Finnish data, and found that the proportion of older-driver crashes that were at intersections depended on cohort and age. A longitudinal analyses using the much longer series of FARS data now available would provide useful new information.
Traffic deaths relative to all deaths
2,403,351 people died from all causes in 2000 in the US. Of these, 41,945, or 1.75%, died in traffic crashes. The vast majority of deaths are from diseases, for which risk generally increases steeply with increasing age, whereas traffic deaths peak at young ages. The age dependencies for disease and traffic combine to make the probability that a given death is a traffic fatality depend very steeply on age, as shown in the logarithmic plot in Fig. 7-17. An exponential decline (straight line in the graph) fits the data well (r2 = 0.99) for both genders between ages 20 and 80, leading to,
7-1
and
7-2
where a is age in years. Thus the probability that a death is a traffic fatality declines at just under 8% per year for both genders during most of their lives.
Given the much higher numbers of young males than young females killed in traffic, the absence of a clear difference between the genders in Fig. 7-17 might seem surprising. The explanation is that most deaths of younger people are from injuries, not diseases. The gender-dependence for non-traffic injuries, in the US mainly firearm deaths, follows a pattern similar to the dependence for traffic crashes, so the ratio of traffic deaths to all deaths ends up relatively gender independent.
Table 7-1 shows illustrative ages selected from the plotted data. Given that a death occurs to a 20-year-old, the probability that it is a traffic fatality is over 30%. Given that a 17-year-old girl dies, it is about as likely to be due to a traffic crash as from all other (injury plus disease) causes combined. As people age, the risks from other causes of death increase much more rapidly than any increase of risk in traffic. Given that a 65-year-old dies, the probability that death is due to a traffic crash is less than one percent. For an 80-year-old it is less than half a percent, and less than a quarter of a percent for ages above 90.

Table 7-1. Given that a death occurs, the percent probability that it is a traffic fatality of any kind. Based on data from FARS 2000 and NCHS(CDC) 2000.18

Types of crashes
Not only do the numbers of crashes vary by gender and age, but the types of fatal crashes in which older drivers are involved differ from the mix for the overall population.
Rollover and driver age
Fig. 7-18 shows that as drivers age, when a driver is killed, the probability that it occurs in a crash in which rollover is the most harmful event declines, reaching very low values at the oldest ages. For age 90 and older, FARS 2000-2002 documents 299 male driver fatalities - eight of them in rollover crashes. For female drivers, 118 fatalities, three in rollovers. For both genders, given a driver death at age 90 or older, there is a less than 3% chance that it is a rollover, compared to a 25% chance for drivers in their mid-twenties. The higher rates for males and for younger drivers supports that rollover is related more to driver risk-taking than vehicle characteristics. Ninety percent of rollover crashes are single vehicle.

Figure 7-18. Probability that, when a driver is killed, rollover is the most harmful event. FARS 2000-2002.
Two-vehicle crashes and side impact
Given that a driver is killed the probability that it is in a rollover crash, or more generally a single-vehicle crash, decreases steeply with age. As a consequence, the probability that the death results from a two-vehicle crash must increase with age (Fig. 7-19). The type of two-vehicle crash also depends on driver gender and age. Figure 7-20 shows the percent of all drivers killed in two-vehicle crashes who were killed in vehicles struck on the side.
The three figures (Figs 7-18 to 7-20) show patterns consistently interpretable in terms of risk taking. Consider a hypothetical extremely risky male driver who crashes into the first available object. This is unlikely to be a vehicle - there are far more trees than vehicles. His crashes will be overwhelmingly single vehicle, likely involving rollover. Although a small percent of his crashes will be two-vehicle crashes, he will still be involved in more two-vehicle crashes than an average driver. If he is involved in a two-vehicle crash, his vehicle is more likely to have a frontal than a side impact. Now consider a hypothetical very safe female driver. She is going to avoid single-vehicle, and therefore also rollover crashes, or crashing head-on into other vehicles. Her main risk in

traffic is being side impacted by other vehicles, so that a high percent of her low crash risk is from side impact crashes.
The high proportion of older drivers who are killed in side impacts, and the high risks faced by older pedestrians (Fig. 7-15) suggests that the most productive way to protect older road users is to address the behavior of younger road users.
All the above relationships are based on relative risks, so the comments in the section Interpreting risk ratios (p. 76-77) should be kept in mind. Fig. 7-20 does not imply that an elderly woman driver is more likely to be killed on her next trip by side impact than is a young male driver. It implies that if she is killed, the probability that it is from a side impact is higher than the probability that the young male driver, if he is killed, will be killed by a side impact.
Risk comparisons using specific examples
The material above shows many measures, each addressing how some specific risk changes as drivers age. Below, specific examples are presented in tabular form to summarize and simplify the main features of the different measures. The values quoted are in many cases obtained by interpolating between values plotted, and in a few cases are based on relatively small sample sizes.
Threats to other road users
Table 7-2 compares the risks imposed on other road users by drivers aged 20, 45, and 70 years, as indicated by the number of single-vehicle pedestrian fatality crashes in which the divers were of those ages. For fatalities to others per licensed driver (the first row), all the relative risks shown are substantially less than one. This means renewing the license of a 70-year-old for another year imposes far less risk on other road users than renewing the license of a 45- or 20-year-old. The effects are so large and clear as to leave little doubt that, from a public safety perspective, there should be less concern over renewing the license of an average 70-year-old applicant than of applicants of any younger age. For example renewing the license of a 70-year-old female imposes 41% less risk than renewing for a 45-year-old female, and 77% less risk than for a 20-year-old female. If the rate is for the same distance of travel rather than per year, then 70-year-olds impose more risk than 45-year-olds, but about half the risk of 20-year-olds.
Table 7-3 shows parallel information for 80-year-old drivers. Renewing the license of an 80-year-old for another year imposes substantially less risk on other road users than renewing the licenses of younger drivers, but by lesser amounts than noted above for 70-year-old drivers.
While data become sparse at ages beyond 80, the indications are clear that renewing for another year the license of drivers at any age up to the maximum for which data are available (near 90) poses less risk to others than renewing licenses of substantially younger drivers (Fig. 7-12).
Risks older drivers themselves face
As drivers age, some measures indicate increases in risk (Tables 7-4 and 7-5). A major contributor to this is that the same severity crash is more likely to lead to the death of an older person. In terms of the measures which best reflect the behavioral aspects of driving, namely, driver involvements in severe crashes for the same distance of travel (Table 7-4), and crashes in which pedestrians are killed for the same distance of travel (Table 7-2), the values for 70-year-old male drivers are within 19% of those for 45-year-old male drivers. (Many factors could contribute to the lack of a larger difference, such as the older drivers confining driving to safer periods, less alcohol use, etc.). The involvement rates for 80-year-old drivers are 80% above those for 45-year-old drivers (Tables 7-3 and 7-5).

Examples using hypothetical families
The illustrations above can be simplified further by comparing risks to members of hypothetical families. Let us consider a family with the following three members:
son age 20
dad age 45
granddad age 70
We focus on male members because males comprise 70% of driver fatalities. Assume all have driving licenses, and crash rates equal to the averages reported above. Based on the results in Figs 5, 7, 8, 9, 12, and 13, the risk associated with each of the family members can be ranked from the highest risk (rank 1 in Table 7-6) to the safest (rank 3). The first row indicates that the son is most likely to die in the forthcoming year as a driver, the granddad least likely. The second row indicates that in a trip of fixed distance, say to the drugstore, the son is most likely to die, the dad the least likely to die.

Table 7-6. Comparison of risks associated with 3 family members. The first row indicates that the son is most likely (rank 1) to die in the forthcoming year as a driver, the granddad least likely (rank 3).

For all the measures listed, the son has the highest risk, in all cases by large amounts. In a few of the comparisons, indicated by asterisks, the risks for the dad and granddad were quite close, but the ranks listed reflect the nominal values read from the plots.
For the case when the granddad is 80, the rankings are all unambiguous (Table 7-7). In a trip to the drugstore, the 80-year-old granddad is more likely to be killed than the 20-year-old son. For all the other cases listed, the son has the highest risk.
The older driver problem - how is it changing?
The findings here are based on data centered in 2001. Analyses with features in common with the present have been applied previously to data for the mid-1990s and the early 1980s. , The more recent the study, the more data were available at older ages. Indeed, the studies using early 1980s data selected 65 as the main older driver age category because of the paucity of data at ages older than this.20,21 The plots presented here have many data at ages above 90.
The decision of those setting up the FARS system to allow only two digits for age, although well justified by early 1970s experience, turns out to be unfortunate. The oldest specific age that can be coded is 96 (not 99, because fields are needed for such categories as age unknown). FARS 2002 codes 12 people aged 96. The next age category, which includes everyone 97 and older, has 17. It would be valuable to know the specific ages of those in this open-ended category, especially as the numbers at very old ages are expected to increase in the future. Also, accumulating small values over many FARS years often provides useful information.
The studies from the three periods indicate fairly robust relationships, with no more than minor differences in detail. The stability of the rates means that even as the number of older drivers increases, the present rates can be applied to larger numbers of drivers. The threat that older drivers pose to other road users was similarly small in all the studies.
The above discussion has focused on how various measures depend on average age. Not only do various measures of driver performance decline with age, but variability among individuals also increases, underlying the importance of not judging an individual's fitness to drive on the basis of chronological age. Licensing decisions should be based on tests that apply to all without regard to age. The admonitions about the uses and pitfalls of average values at the end of Chapter 3 (p. 60) apply here also.
The younger driver problem
The graphs and tables presented here consistently show that young male drivers have the highest fatality and crash rates, and pose the greatest threats to other road users. The same finding emerges from the prior studies discussed, and from many additional sources. While some measures for older drivers indicate above average risk, the amount that they are above average appears to be trending downwards in time. Such a tendency will elevate the already dominant contribution of young, especially young male, drivers to even greater prominence. One of the grand themes at the center of traffic safety in every country in the world is that traffic crashes are overwhelmingly a problem of young male drivers.

Summary and conclusions (see printed text)

References for Chapter 7 - Numbers in [ ] refer to superscript references in book that do not correctly show in this html version.  To see how they appear in book see the pdf version of Chapter 1.

[1]   Lyman S, Ferguson SA, Braver ER, Williams AF.  Older driver involvements in police reported crashes and fatal crashes: Trends and projections.  Inj Prev. 2002; 8: 116-120.

[2]   Bedard M, Stones MJ, Guyatt GH, Hirdes JP. Traffic-related fatalities among older drivers and passengers: Past and future trends.  Gerontologist. 2001; 41: 751-756.

[3]   Eby DW, Trombley DA, Moinar LJ, Shope TJ.  The assessment of older drivers’ capabilities: A review of the literature.  Report UMTRI-98-24, University of Michigan Transportation Research Institute, Ann Arbor, Michigan; August 1998

[4]   Owsley C, McGwin G Jr, Sloane M, Wells J, Stalvey BT, Gauthreaux S.  Impact of cataract surgery on motor vehicle crash involvement by older adults.  J Am Medical Assoc. 2002; 288: 841-849.

[5]   Lyman JM, McGwin GJ, Simms RV.  Factors related to driving difficulty and habits in older drivers.  Accid Anal Prev. 2001; 33: 413-421.

[6]   McGwin G Jr, Chapman V, Owsley C. Visual risk factors for driving difficulty in older drivers.  Accid Anal Prev. 2000; 32: 735-744.

[7]   Lundberg C, Hakamies-Blomqvist L.  Driving tests with older patients: Effect of unfamiliar versus familiar vehicle.  Trans Res, Part F. 2003; 6: 163-173.

[8]   Warshawsky-Livne L, Shinar D.  Effects of uncertainty, transmission type, driver age and gender on brake reaction and movement time.  J Safety Res. 2002; 33: 117-128.

[9]   Fatality Analysis Reporting System (FARS) Web-Based Encyclopedia.  Data files and procedures to analyze them.    http://www-fars.nhtsa.dot.gov

[10] Population Projections Program, Population Division, US Census Bureau, Department of Commerce Washington, DC (http://www.census.gov).  Specific data used was “(NP-D1-A) Projections of the Resident Population by Age, Sex, Race, and Hispanic Origin: 1999 to 2100.”

http://www.census.gov/population/projections/nation/detail/d2001_10.pdf

[11] Federal Highway Administration, Office of Highway Policy Information.  Highway Statistics 2001.    http://www.fhwa.dot.gov/ohim/hs01/xls/dl20.xls

[12] State-by-state driving rules for teenage drivers.    http://golocalnet.com/drivingage

[13] 2001 National Household Travel Survey (NHTS).                              

http://nhts.ornl.gov/2001/html_files/introduction.shtml

[14] Kahane CJ.  Vehicle weight, fatality risk and crash compatibility of model year 1991-99 passenger cars and light trucks.  Report DOT HS 809 662 2.  Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration; October 2003.

[15] Stalvey BT, Owsley C.  The development and efficacy of a theory-based educational curriculum to promote self-regulation among high risk older drivers.  Health Promot Pract. 2003; 4: 109-119.

[16] Keeffe JE, Jin CF, Weih LM, McCarty CA, Taylor HR.  Vision impairment and older drivers: Who’s driving?  Br J Ophthalmol. 2002; 86: 1118-1121.

[17] Hakamies-Blomqvist L, Henriksson P.  Cohort effects in older drivers’ accident type distribution: Are older drivers as old as they used to be?  Trans Res Part F. 1999; 2: 131-138.

[18] National Center for Health Statistics, Center for Disease Control.  Table 310. Deaths by single years of age, race, and sex.   http://www.cdc.gov/nchs/data/statab/wktbl310.pdf

[19] Evans L.  Risks older drivers face themselves and threats they pose to other road users.  Int J Epidemiology. 2000; 29: 315-322.

[20] Evans L.  Older driver involvement in fatal and severe traffic crashes.  J Gerontology: Soc Sciences. 1988; 43: S186-S193.

[21] Evans L.  Traffic Safety and the Driver.  New York, NY: Van Nostrand Reinhold; 1991. Chapter 2.  Effects of sex and age, p. 19-43.