METHODS

Data sources

Multiple databases were used to study the risk per passenger vehicle driver by driver age for four categories of road users: drivers, their passengers, occupants of other passenger vehicles colliding with them in two-vehicle crashes, and non-occupants in one-automobile crashes, including pedestrians, bicyclists, and motorcyclists.

Drivers

Active drivers by driver age, with the exception of ages 30–34, were estimated using the 1995 Nationwide Personal Transportation Survey (NPTS).¹⁸ NPTS data from telephone interviews of a national sample of United States residents (97 881 persons in 1995) were used because standard federal licensure data include drivers dying before scheduled license renewals and license holders who no longer drive.²⁴ Thus, standard data overstate the number of older persons who drive, which results in underestimates of crash rates. For ages 30–34, standard federal data were used because the NPTS estimate for this age group was 11.5% higher—about the same as the census estimate. Compared with federal licensure statistics, NPTS estimated about 29% fewer drivers aged 80 and older, which likely is more accurate than standard data. For other age groups, the two databases yielded similar estimates of drivers.

To estimate active drivers during 1993–97, NPTS 1995 driver estimates were multiplied by five. This procedure appeared reasonable because licensed drivers among the different age groups increased linearly during 1993–97; also, the licensed driver counts for 1993–97 resembled the 1995 NPTS counts multiplied by five, except for ages 80+ and 30–34.²⁴

Deaths

The Fatality Analysis Reporting System (FARS), the information source for motor vehicle related deaths during 1993–97, is a census of fatal crashes in the United States occurring on public roads in which a death takes place within 30 days of the crash.²⁵

Non-fatal injuries

The General Estimates System (GES), the information source for non-fatal motor vehicle injuries during 1993–97, is a national sample of police reported crashes.²⁶ The 55 000 annual crashes in the GES are weighted according to their probability of selection so as to provide national crash estimates. The GES translates police judgments of injury severity to the KABCO scale (K = killed, A = incapacitating injury, B = visible but non-incapacitating injury, C = complaint of pain, O = no injury). This study analyzed both A and B injuries.

Insurance claims

Insurance databases provide a comprehensive picture of the most common types of crashes because police are not notified of many crashes reported to insurers, and police reported crashes are more likely to involve severe injury. Also, data on driver fault are more directly addressed than for involvements in police reported crashes. Another advantage of using insurance data was that the claims experience for ages 75–79, 80–84, and 85 and older could be computed separately rather than combining them in a single age category because, unlike FARS, GES, and NPTS, there were sufficient numbers of insured drivers in these groups to permit stable claim rate estimates.

Insurance claims by driver age—primarily non-fatal crashes—were supplied by nine insurers for vehicles three years old or newer: model years 1999–2001 during calendar years 1998–2001. The actual ages of drivers in crashes resulting in insurance claims were unknown; rather, the ages of drivers listed for those vehicles in insurance records—referred to as rated drivers—were analyzed. The rated driver usually is the main driver for that vehicle, but some insurers assign the highest risk driver in a household to its most valuable vehicle. This may have led to some misclassification of crashes by driver age, but likely had little effect on older drivers, who usually share a household only with other older people.

For injury claims, this study included only those made for medical expenses under bodily injury liability policies. Bodily injury liability claims are made by persons injured by a driver deemed at fault and typically are occupants of other passenger vehicles or non-occupants struck by the insured vehicle. Bodily injury liability claims, which are the strongest indicator of responsibility for other road users’ injuries, were studied in all states and the District of Columbia.

Property damage liability claims, which cover property damage to other vehicles caused by insured drivers, also were analyzed in all 50 states and the District of Columbia. Property damage liability claims, which are filed more often than injury claims, are not necessarily associated with injuries, but are another indicator of the crash patterns of drivers in different age groups.

Analytic methods

The magnitude of the public health problem of motor vehicle injuries by driver age is shown by absolute numbers of vehicle related deaths and non-fatal injuries. Risks per driver or per insured vehicle are another measure of impact on the population exposed to vehicle travel.

Rate ratios (RR) and their 95% confidence intervals (CI) were calculated to quantify differences among driver age groups with regard to deaths per driver, persons with non-fatal injuries (incapacitating and visible non-incapacitating) per driver, and claims per insured vehicle year. Ages 30–59 were used as a reference group to facilitate comparisons among younger, older, and middle range drivers. Analyses were done using SAS, Microsoft Excel, and SUDAAN.^27–29

RESULTS

Deaths

Young drivers

Table 1 presents deaths during 1993–97 and NPTS estimates of active drivers by driver age. The greatest numbers of deaths arose from collisions involving teenage drivers and drivers aged 20–29. Per driver younger than 30—especially those under 25—raised RRs of death were observed among all road user types compared with drivers aged 30–59. Deaths among non-occupants and other vehicle occupants frequently occurred in collisions with younger drivers. A striking finding was a RR of 5.79 (95% CI 5.50 to 6.09) for passenger deaths in vehicles of teenage drivers. Passenger deaths per driver also were significantly raised among drivers in their 20s.

Older drivers

Among drivers 75 and older, significantly raised RRs were observed only for driver deaths (RR 3.02; 95% CI 2.86 to 3.19) and passenger deaths (RR 2.52; 95% CI 2.39 to 2.66) relative to ages 30–59 (table 1; fig 1). Smaller elevations were observed among drivers aged 70–74 (RR for driver deaths 1.35; RR for passenger deaths 1.31). The absolute number of passenger deaths associated with the oldest drivers was far lower than passenger deaths for young drivers.

Most passengers of older drivers were elderly. In both police reported towaway and fatal crashes during 1993–97, about 75% of passengers transported by drivers aged 75 and older were 70 and older.^25,30

Among drivers aged 60–74, risks of death among occupants of other passenger vehicles and non-occupants were significantly lower compared with ages 30–59. When drivers reached age 75, there was an upturn in fatality risk among these types of road users, but their RRs relative to 30–59 year old drivers remained below 1.0.

Two thirds of the deaths in crashes involving drivers 75 and older were the drivers themselves. Drivers younger than 60 comprised fewer than 50% of motor vehicle deaths in their collisions.

Non-fatal injuries

General Estimates System

Persons with non-fatal injuries described by police as incapacitating or visible but non-incapacitating were studied. RRs for the two injury severities were similar so they were combined (table 2; fig 2).

Young drivers

Younger drivers had much higher involvement rates in crashes resulting in non-fatal injuries to themselves than 30–59 year old drivers. As was true for fatalities, teenage drivers posed a greater risk of non-fatal injury to their passengers, non-occupants, and occupants in other vehicles than drivers in any other age group. The fatal and non-fatal injury rates per driver of passengers of teenage drivers both were more than five times as high as those of 30–59 year old drivers. For both fatal and non-fatal crashes, drivers in their 20s had lower risks than teenage drivers but their risks still were raised.

Older drivers

The pattern for non-fatal injuries was for drivers aged 60–74 to have RRs for injuries well below 1.0 compared with ages 30–59, but for drivers 75 and older to have modest upturns in risk that were not as high as those for teenage drivers.

With some exceptions, results for non-fatal injuries among older drivers had patterns resembling those for deaths. One difference was that RRs for non-fatal injuries among older drivers and their passengers (1.16 and 0.86, respectively) were far lower than those for deaths (3.0 and 2.5, respectively).

Another difference is that although older drivers had lower rates of collisions resulting in occupant deaths in other passenger vehicles, drivers 75 and older had a 10% increase (statistically non-significant) compared with ages 30–59, in two-vehicle crash involvements resulting in non-fatal injuries to other vehicle occupants (RR 1.10; 95% CI 0.98 to 1.24). Compared with ages 60–74, the increase was statistically significant.

Insurance claims

Table 3 shows distributions of bodily injury liability claims (primarily non-fatal injuries), property damage liability claims, and insured vehicle-years by driver age, together with RRs and confidence intervals. Figure 3 depicts claims per 1000 insured vehicle-years.

Young drivers

Teenage drivers had the highest bodily injury liability frequencies per insured vehicle-year (RR 3.32; 95% CI 3.26 to 3.38). Drivers aged 20–24 had twice the bodily injury liability claim rate as drivers aged 30–59 (RR 2.0; 95% CI 1.97 to 2.03) and drivers 25–29 had a smaller elevation in bodily injury liability claim rates. Property damage liability claim rates per insured vehicle-year also were raised among drivers younger than 30, with risks more than doubling among drivers aged 16–19 (RR 2.13; 95% CI 2.11 to 2.15).

Older drivers

As with crashes leading to fatal and non-fatal injuries, older drivers in their 60s had significantly lower bodily injury liability claim frequencies per insured vehicle-year compared with ages 30–59, with RRs of about 0.80. At age 70, bodily injury liability claim rates and RRs started to increase, with higher rates for each successive five year age group. The bodily injury liability claim rate was 1.8 times as high for drivers 85 and older as for drivers 30–59 (95% CI 1.71 to 1.89), albeit lower than that for teenage drivers. Property damage liability claims per insured vehicle-year among the oldest drivers (aged 85+) were as high as among teenage drivers (RR 2.09 for drivers 85+; 95% CI 2.05 to 2.13).

DISCUSSION

Public concern exists about potential hazards from the aging driver population. Older people suffered the most serious consequences of crashes involving older drivers, consistent with other research.^21,23 Significant increases in fatalities per driver did not begin until ages 70–74, and the most likely victims were the older drivers followed by their passengers, who also were older and thus more fragile. Throughout the older ages, there were low rates of crashes fatal to other road users, including other vehicle occupants, after declining greatly from high levels per driver among teenagers and 20–29 year olds.

Analyses of non-fatal injuries and insurance claims showed that other road users did experience some of the injury burden and property losses associated with older drivers. This study differed from past studies, which observed increased risk to other road users per mile, but generally not increased risks per driver.^20–22 Although older drivers did not pose an excess risk of non-fatal injury to passengers of their vehicles or to non-occupants, risks to occupants of other passenger vehicles (and to the drivers themselves) did increase when drivers reached age 75. The effect of older drivers’ crashes on other road users was most evident for insurance liability claims, suggesting increased at-fault crashes. Relative to ages 30–59, bodily injury liability claims nearly doubled at ages 85 and older, and, after a low point at ages 60–69, property damage liability claims increased with age so that the claim rate against drivers 85 and older doubled.

To put this into perspective, it is important to recognize that younger drivers, both teenagers and people in their 20s, represent a substantially greater public health problem than older drivers for both fatal and non-fatal motor vehicle injuries. With the exception of driver death rates among drivers 75 and older, teenage drivers posed the highest risks of death and non-fatal injury to themselves, their passengers, occupants of other passenger vehicles, and non-occupants. Risks declined among drivers in their 20s, but were significantly greater compared with 30–59 year olds.

Property damage liability claims per insured vehicle year were similar among the youngest and oldest drivers, but teenage drivers had much higher bodily injury liability claim rates. This suggests that the oldest drivers have a comparable risk of involvement in at-fault crashes, but are less likely than younger drivers to injure others.

Why older drivers were associated with an increased risk of bodily injury liability claims, property damage liability claims, and non-fatal injuries, but not fatal injuries, to occupants of other passenger vehicles is unknown. The travel patterns of older drivers, who travel less at night and more on roads with speed limits below 55 mph, may reduce fatality risk.³¹

One important point when interpreting the current study is that injuries described as incapacitating or visible non-incapacitating by police typically are minor.³² A limitation of insurance databases is that they are confined to newer vehicles (1–3 years old), which might have higher annual mileage and healthier older drivers with more resources than is typical. Another limitation is that confounding factors such as travel speeds and alcohol impairment could not be studied.

Screening to identify older drivers at very high risk of crashing might protect both older drivers and other road users, but consensus has not been reached on feasible and valid methods to detect unsafe drivers and avoid falsely characterizing safe drivers as hazardous.¹⁷ Research on driver screening should continue, because screening might provide useful information to drivers and physicians when considering whether modifications in driving might be appropriate for individuals. Efforts to enhance the protection afforded to older vehicle occupants, such as improving restraint systems and vehicles, should be vigorously pursued because older people are more susceptible to injury and will continue to travel in vehicles whether or not they are driving.^19,33