Full text of DOI:10.1093/gerona/57.3.M186

Journal of Gerontology: MEDICAL SCIENCES
Copyright 2002 by The Gerontological Society of America
2002, Vol. 57A, No. 3, M186–M191
Risk Factors for Motor Vehicle Crashes in
Older Women
Karen L. Margolis,1 Roxanne Pieper Kerani,2 Paul McGovern,3 Thomas Songer,4 Jane A. Cauley,4
and Kristine E. Ensrud3,5 for the Study of Osteoporotic Fractures Research Group
1Division of Clinical Epidemiology, Hennepin County Medical Center, Minneapolis, Minnesota.
2Division of Allergy and Infectious Diseases, School of Medicine, University of Washington, Seattle.
3Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis.
4Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pennsylvania.
5Center for Chronic Disease Outcomes Research, Minneapolis VA Medical Center, Minnesota.
Background. Motor vehicle crash and fatality rates are higher per mile driven for elderly drivers, with an exponen-
tial increase above age 75. Identifying elderly drivers who are at risk for automobile crashes may help direct interven-
tions to reduce their high rate of injuries and deaths.
Methods. Subjects were 1416 women aged 65 to 84 enrolled in the Portland, Ore. site of the Study of Osteoporotic
Fractures. Motor vehicle crash information for the years 1986–1995 for each participant was obtained from the Oregon
State Department of Transportation. Items from questionnaires, interviews, and physical examinations were tested pro-
spectively for associations with the occurrence of motor vehicle crashes.
Results. About one third of participants (415 of 1416) had a motor vehicle crash during a mean follow-up time of 5.7
years. After adjustment for age and weekly driving mileage, risk factors significantly associated with motor vehicle
crashes were a fall in the previous year [hazard ratio (HR) 1.53, 95% confidence interval (CI) 1.26–1.86], a greater
orthostatic systolic blood pressure drop (HR 1.11 per 12.5 mm Hg, 95% CI 1.01–1.22), and increased foot reaction time
(HR 1.10 per 0.06 second, 95% CI 1.00–1.22). Other neuromuscular tests, functional status, medical diagnoses, vision
tests, and cognitive tests did not predict motor vehicle crashes in this study population.
Conclusions. This prospective study with extended follow-up of a large cohort of elderly women has identified crash
risk factors that can be measured in the clinical setting. Further study is needed to determine if interventions aimed at
these risk factors can decrease the risk of motor vehicle crashes.
Y the year 2020, it is estimated that there will be 50
million elderly persons eligible to drive in the United
tions, and functional impairments and motor vehicle crashes
are also controversial (6,10,12–14).
B
States (1). Motor vehicle crash rates adjusted for miles
driven are higher for elderly drivers, with an exponential
increase above age 75 (1). A similar pattern is observed
with the driver fatality rate (1). Identifying elderly drivers
who are at high risk for automobile crashes may help di-
rect interventions to reduce the high rate of injuries and
deaths.
The health status of older drivers has been the central fo-
cus of previous studies on motor vehicle crashes among the
elderly. Age-related changes in processing of sensory input,
cognitive capacity, and neuromuscular function may all af-
fect an elderly driver’s ability to operate a motor vehicle
safely. Reduced visual acuity, poor contrast sensitivity,
glaucoma, cataracts, and hearing impairment have been re-
lated inconsistently to motor vehicle crashes (2–8). The
Useful Field of View, a test of visual processing speed, has
a stronger and more consistent relationship with motor vehi-
cle crashes (6,8,9). Some studies have shown associations
between cognitive impairment (6,10) or depression (10,11),
and the risk of automobile crashes, but results have been in-
consistent (6,8–11). Associations between the use of medi-
cations with central nervous system effects, medical condi-
Most previous studies have been limited by cross-sec-
tional or retrospective study designs, small sample sizes,
and limited information on potential risk factors. The Study
of Osteoporotic Fractures (SOF) overcomes these limita-
tions with its prospective design, detailed clinical examina-
tions, extended follow-up, and large sample of elderly com-
munity-dwelling women. The purpose of this study was to
describe the motor vehicle crash risk over a 10-year period
and to identify risk factors for motor vehicle crashes in el-
derly women.
METHODS
Study Population
The Study of Osteoporotic Fractures is a prospective
study designed to identify risk factors for fractures in el-
derly women (15). Nonblack women aged 65 and older
were recruited (N ꢀ 9704) for the baseline examination us-
ing population-based lists from 1986–1988 in Baltimore
County, Md.; Portland, Ore.; Minneapolis, Minn.; and the
Monongahela Valley near Pittsburgh. Seven subsequent ex-
ams have been conducted at approximately 2-year intervals.
M186

CAR CRASHES AND OLDER WOMEN
M187
Motor vehicle accident data were only available from the
state of Oregon; therefore, only participants from the Port-
land site are included in the present study (n ꢀ 2421). We
further excluded 152 women who did not attend Visit 2, 250
women with drivers’ licenses issued by states other than Or-
egon, 554 women who reported at Visit 2 that they had not
driven in the past 12 months, 43 women who declined to
provide their drivers’ license numbers, and 6 women who
were aged 85 years or older at baseline. Thus, our analyses
were conducted on 1416 women.
Contrast sensitivity was measured using the Vistech Vision
Contrast Test System 6500 (24) at 10 feet for spatial fre-
quencies ranging from 1.5 to 18 cycles per degree. The
scores for low and high spatial frequencies were averaged
separately (25). Near depth perception was measured with
the Randot Stereotest Contoured Circles test (26). Distance
depth perception was measured using the Howard Dolman
apparatus and was scored as the standard deviation of four
trials (27). The Mini-Mental Status Examination (28) was
administered at Visit 1, and the Trail Making Part B (29)
and Digit Symbol Substitution (30) tests were administered
at Visit 2 to assess cognitive function.
Data Collection
All participants were asked to provide their drivers’ li-
cense numbers during Visit 2. Motor vehicle crash informa-
tion for the years 1986–1995 for each participant was ob-
tained from the Driver and Motor Vehicle Services, division
of the Oregon State Department of Transportation. The date
of any crash that resulted in a police report was recorded.
Participants who had any motor vehicle accident from en-
rollment in SOF until the end of July 1995 were considered
to have the outcome of interest.
Data from the baseline exam (1986–1988), Visit 2 (1988–
1989), and Visit 4 (1992–1994) are included in this analysis.
Unless noted otherwise, all data used in this analysis were
collected at the baseline visit. Driving exposure was mea-
sured at Visit 2 and Visit 4 by asking participants if they had
driven a car in the past 12 months, and if so, on average,
how many miles per week they had driven.
At baseline, the participants were asked the number of
falls they had experienced in the previous year. Physical ac-
tivity was measured using a modified Paffenbarger ques-
tionnaire, whereby an estimate of total kilocalories ex-
pended per week in recreational activities was calculated
(16,17). Participants were also asked, “Do you walk for ex-
ercise?” and “At least once per week, do you engage in any
regular activity long enough to work up a sweat?”
Functional status was assessed by self-report of any diffi-
culty performing each of six activities of daily living: walk-
ing two to three blocks on level ground, climbing up 10
steps without stopping, walking down 10 steps, preparing
meals, doing heavy housework, and shopping for groceries
or clothes (18,19). At Visit 2, participants completed the
Geriatric Depression Scale (20). The presence of depression
was defined as a positive response to six or more symptoms
(21). Participants were asked whether they had ever been di-
agnosed by a physician with diabetes, stroke, osteoarthritis,
or cataracts. Use of medications to aid sleep in the past 12
months was assessed, and the names of the medications
were recorded. Participants were asked to report the average
number of alcoholic drinks they consumed per week during
the past year.
Data Analysis
Bivariable relationships were examined with chi-square
and t tests. Alcohol consumption was highly skewed, with a
large proportion of nondrinkers. Therefore, drinks per week
were categorized into five levels, with the first level repre-
senting those who reported never drinking, the second level
representing former drinkers, and the next three levels rep-
resenting the remaining participants divided into tertiles
based on reported drinks per week. Cox proportional haz-
ards regression was utilized for multivariable analyses. The
hazard ratio was determined before and after adjustment
both for age in 5-year increments and for weekly miles
driven in 50-mile increments. For all other continuous vari-
ables, it was expressed per one standard deviation. Indepen-
dent variables that had a relationship at a significance level
of p ꢁ .10 with the outcome variable after adjustment for
age and miles driven per week were included in a multivari-
able model. We tested for interactions between variables
that entered the final multivariable model by adding an in-
teraction term to the model.
Blood pressure was measured both in the supine position
after a 5-minute rest period and after standing for 1 minute.
Neuromuscular function tests included ability to rise from a
chair five times without using the arms, number of step-ups
completed in 10 seconds, average of right and left grip
strength using a hand-held dynamometer (Preston Grip Dy-
namometer, Takei Kikikogyo, Tokyo, Japan), walking
speed over a 6-meter course, time to complete a walk in a
tandem position, and the number of seconds to complete 10
taps with the dominant foot. At Visit 2, dominant hand and
foot reaction times were measured by depressing a pad in
response to a light cue. The results of 10 trials were aver-
aged.
There were 772 participants who reported that they had
continued driving through Visit 4 and had no record of a
motor vehicle accident before July 31, 1995 (the latest date
for which motor vehicle information was available). These
individuals were considered to be at risk from their date of
enrollment until July 31, 1995. Persons who had a motor ve-
hicle accident (n ꢀ 415) were considered to be at risk from
their time of enrollment until the date of their first accident.
Participants who reported that they had stopped driving be-
tween visits 2 and 4 (n ꢀ 89) were censored halfway be-
tween their Visit 2 and Visit 4 dates, as the specific date on
which they stopped driving was not available. Nonattenders
at Visit 4 (n ꢀ 51) were censored halfway between their
Visit 2 dates and the midway point of Visit 4 (July 31,
1993), as information regarding whether they were still
driving was not available. Participants who died at any
time after Visit 2 (n ꢀ 89) were censored on their date of
death.
Corrected visual acuity was measured by the letter charts
using the Bailey and Lovie method (22), and acuity was ex-
amined as both a dichotomous “impaired acuity” variable
(visual acuity of 20/40 or worse) and as a continuous vari-
able [log minimum angle of resolution (logMAR) (23)].

M188
MARGOLIS ET AL.
RESULTS
interaction between walking for exercise and falling in the
The demographic and driving characteristics of the study
past 12 months. An increased risk of motor vehicle crashes
among women who fell was primarily restricted to the
women who did not walk for exercise, with an adjusted risk
ratio of approximately 2. The association between minimal
alcohol consumption and motor vehicle crashes also was
present in the multivariable model. Both systolic blood
pressure drop and foot reaction time were associated with
motor vehicle crashes in the multivariate model.
population are shown in Table 1. At baseline, the average
age of participants was 71 years. The mean number of re-
ported weekly miles driven was 55, or just under 3000 an-
nual miles driven. About one third of participants (415 of
1416) had at least one motor vehicle crash during an aver-
age follow-up time of 5.7 years (Table 1). The rate of
crashes was 0.07 per driver per year.
Self-reported risk factors and their associations with
crashes are shown in Table 2, before and after adjustment
for age and miles driven per week. Each 5 years of age in-
creased the relative risk of a motor vehicle crash by 16%.
Each additional 50 miles driven per week increased the risk
of a motor vehicle crash by 14%. A history of one or more
falls increased the risk of a motor vehicle crash by approxi-
mately 50%. Women who reported they walked for exercise
had a borderline lower risk. Compared with women who did
not drink alcohol, we found an increased risk of motor vehi-
cle crashes in women who drank but consumed one drink or
less per week, but not in women who drank somewhat more
heavily.
Only three physical examination findings showed an as-
sociation with motor vehicle crashes (Table 3). For each
standard deviation increase in the orthostatic systolic blood
pressure drop and foot reaction time, there was an approxi-
mately 10% higher risk of motor vehicle crashes. There was
also a small increase in the motor vehicle crash risk with
worsening visual acuity modeled continuously as logMAR.
The results of the multivariable model are shown in Table
4. Age and miles driven per week remained associated with
motor vehicle crashes. We found a statistically significant
DISCUSSION
This prospective study provides important new informa-
tion on motor vehicle crash risk in elderly women drivers.
One in three women in our study had a motor vehicle crash
serious enough to generate a police report during approxi-
mately 6 years of follow-up. Motor vehicle crashes could be
predicted by several self-reported variables: increased age,
increased number of miles driven weekly, and a history of
falling in those who did not walk for exercise. Compared to
nondrinkers, very light drinkers had an increased risk of
motor vehicle crashes, whereas moderate drinkers did not.
Only two physical examination findings were weakly asso-
ciated with the risk of motor vehicle crashes: a drop in sys-
tolic blood pressure upon standing and foot reaction time.
None of the SOF measurements of functional status, medi-
cal diagnoses, vision, cognitive function, depression, or use
of sleep medications were predictors of motor vehicle
crashes in this study population.
Three studies have examined the association between a
history of falls and motor vehicle crashes, but results were
inconsistent (6,11,31). In the case-control study by Sims and
colleagues (6), older drivers with at least one at-fault crash
in the previous 6 years and a history of a fall in the previous
2 years had more than twice the odds of a motor vehicle
crash. In a later prospective study, the same authors (11)
also found a doubled motor vehicle crash risk in people who
reported frequent falling or tripping. Marottoli and cowork-
ers (32) found that people who reported not walking had a
two-fold increased risk of motor vehicle crashes, but did not
report on whether any association with falls was found. Our
data support the limited evidence that falls are associated
with motor vehicle crashes and further suggest that the mo-
tor vehicle crash risk in women who fall is confined to those
who are sedentary.
Table 1. Characteristics of the Study Population*
Characteristic
Value
Age, y
Mean (SD)
65–69, %
70–74, %
75–84, %
Education
71.3 (4.2)
39.9
38.7
21.4
Less than high school, n (%)
236 (16.7)
High school or greater, n (%)
1180 (83.3)
Miles driven per week
The mechanism of the association of falls and motor ve-
hicle crashes may be mediated by the association of falls
with other risk factors. In the SOF study, falls are associated
with increasing age, poorer health status, the presence of
one or more medical conditions, not walking for exercise,
functional impairment, trouble with dizziness, depression,
weight loss, slow gait speed, inability to stand from a chair,
poor visual acuity, benzodiazepine use, anticonvulsant use,
and antidepressant use (33). Although most of these vari-
ables were not individually associated with motor vehicle
crashes in our study sample, we speculate that falling may
be a marker for elderly women in whom these risk factors
combine to produce impairments that also impede driving
ability.
Mean (SD)
0–25, %
26–50, %
51–100, %
55.5 (57.2)
38.1
31.0
20.3
10.5
ꢂ100, %
Number of motor vehicle crashes, 5.7 years avg. follow-up (%)
0
1
2
3
4
5
6
7
1001 (70.7)
311 (22.0)
84 (5.9)
14 (1.0)
3 (0.2)
2 (0.1)
0 (0.0)
1 (0.1)
*Age and education at baseline visit, miles driven per week at Visit 2, and
motor vehicle crashes from the Oregon State Department of Transportation
(1986–1995).
Our study also found a weak association between ortho-
static systolic blood pressure change and motor vehicle

CAR CRASHES AND OLDER WOMEN
M189
Table 2. Association of Self-Reported Risk Factors With Crashes
Risk Factor
Mean (SD)/ Prevalence (%)
Crude HR (95% CI)
p Value
Adjusted HR (95% CI)*
p Value
Age, 5 y
1.16 (1.04–1.30)
1.14 (1.07–1.23)
1.53 (1.26–1.86)
0.93 (0.85–1.03)
0.83 (0.68–1.01)
1.01 (0.91–1.12)
1.03 (0.95–1.12)
1.41 (0.86–2.29)
.009
.0001
.0001
.17
1.18 (1.06–1.32)
1.16 (1.08–1.24)
1.50 (1.23–1.83)
0.93 (0.84–1.03)
0.84 (0.69–1.02)
1.00 (0.90–1.11)
1.04 (0.96–1.13)
1.43 (0.88–2.32)
.003
.0001
.0001
.17
Miles driven per week, 50 mi
Any fall in the past 12 months
Physical activity, kcals/week**
Walks for exercise
Activity to induce sweating
Any functional impairment
Geriatric Depression Scale score ꢃ6
Medical Diagnosis
Diabetes
31.4
2482 (2135)
60.5
44.2
31.9
.06
.08
ꢂ.20
ꢂ.20
.17
ꢂ.20
ꢂ.20
.15
3.4
5.7
3.5
59.4
26.1
1.06 (0.84–1.33)
1.10 (0.64–1.87)
0.96 (0.79–1.17)
1.18 (0.95–1.47)
ꢂ.20
ꢂ.20
ꢂ.20
.12
1.08 (0.86–1.37)
1.11 (0.65–1.90)
0.93 (0.76–1.14)
1.10 (0.88–1.38)
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
Stroke
Arthritis
Cataracts
Habits/Medication
Sleep medication
Long-acting benzodiazepines
Alcoholic drinks per week
Abstainers
18.2
9.9
1.02 (0.80–1.31)
1.01 (0.73–1.40)
ꢂ.20
ꢂ.20
1.01 (0.79–1.30)
1.00 (0.70–1.39)
ꢂ.20
ꢂ.20
16.9
27.0
22.5
14.8
18.7
REF
REF
Former Drinkers
0.1–1
1.1–3
1.03 (0.76–1.40)
1.36 (1.01–1.85)
1.06 (0.75–1.51)
0.91 (0.65–1.28)
ꢂ.20
.05
ꢂ.20
ꢂ.20
1.05 (0.77–1.44)
1.39 (1.02–1.88)
1.09 (0.76–1.54)
0.93 (0.66–1.31)
ꢂ.20
.04
ꢂ.20
ꢂ.20
ꢂ3
Notes: HR ꢀ hazard ratio; CI ꢀ confidence interval; REF ꢀ reference value for the hazard ratio, which equals 1.0.
*Adjustment was for age and miles driven per week.
**Hazard ratio for this variable is reported per one standard deviation change.
crashes. No association was reported in two other studies, in
which blood pressure drop upon standing was modeled as a
categorical variable (6,11). We did not find a significant as-
sociation between systolic blood pressure drop of 20 mm
Hg or more and motor vehicle crashes, a characteristic
found in 22% of the participants, nor was self-reported diz-
ziness upon standing associated with motor vehicle crashes
(data not shown). This study is the first to examine the asso-
ciation of reaction time with motor vehicle crashes. Slower
foot reaction time could be predicted to interfere with re-
moving the foot from the accelerator or applying it to the
brake pedal, but these results need confirmation.
We did not find an association between other neuromus-
cular tests or functional status and motor vehicle crashes.
Table 3. Association of Physical Examination Findings With Crashes
Physical Exam Finding
Mean (SD)/ Prevalence (%)
Crude HR (95% CI)
p Value
Adjusted HR (95% CI)*
p Value
Orthostatic systolic blood pressure drop, mm Hg**
Orthostatic diastolic blood pressure drop, mm Hg**
Neuromuscular function
10.6 (12.5)
1.1 (7.8)
1.11 (1.01–1.22)
1.04 (0.96–1.13)
.03
ꢂ.20
1.10 (1.00–1.21)
1.04 (0.96–1.13)
.06
ꢂ.20
Unable to rise from chair 5 times, %
Number of step-ups in 10 seconds**
Grip strength, kg**
2.4%
9.7 (2.4)
21.5 (4.2)
1.05 (0.23)
11.1 (5.0)
4.7 (1.9)
1.22 (0.93–1.60)
1.00 (0.91–1.10)
0.99 (0.90–1.09)
1.01 (0.92–1.11)
1.01 (0.92–1.11)
0.97 (0.87–1.09)
1.03 (0.94–1.14)
1.10 (1.00–1.22)
.14
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
.06
1.22 (0.93–1.60)
1.01 (0.92–1.11)
1.01 (0.92–1.12)
1.02 (0.92–1.12)
1.00 (0.91–1.10)
0.96 (0.86–1.07)
1.03 (0.93–1.14)
1.09 (0.98–1.20)
.15
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
ꢂ.20
.10
Walking speed, m/s**
Time to complete tandem walk, s**
Time to complete 10 foot taps, s**
Hand reaction time, s**
0.36 (0.06)
0.36 (0.06)
Foot reaction time, s**
Vision
Visual acuity 20/40 or worse, %
Visual acuity, logMAR**
Contrast sensitivity, low spatial frequencies**
Contrast sensitivity, high spatial frequencies**
Distance depth perception**
4.7%
1.20 (0.77–1.89)
1.10 (1.00–1.21)
0.96 (0.87–1.06)
0.92 (0.83–1.02)
1.04 (0.95–1.14)
ꢂ.20
.05
ꢂ.20
.10
1.14 (0.73–1.80)
1.08 (0.97–1.19)
0.99 (0.89–1.10)
0.94 (0.85–1.04)
1.01 (0.92–1.11)
ꢂ.20
.16
ꢂ.20
ꢂ.20
ꢂ.20
0.08 (0.12)
73.7 (33.3)
33.5 (27.0)
2.02 (2.44)
ꢂ.20
Cognition
Mini-Mental State Examination Score 25 or 26, %
Trail Making Part B, completed in 3 min, %
Digit Symbol Substitution, number correct**
71.2%
75.0%
42.4 (11.0)
0.99 (0.93–1.07)
1.08 (0.83–1.40)
1.02 (0.92–1.14)
ꢂ.20
ꢂ.20
ꢂ.20
0.99 (0.93–1.07)
1.07 (0.82–1.40)
1.04 (0.93–1.15)
ꢂ.20
ꢂ.20
ꢂ.20
Notes: HR ꢀ hazard ratio; CI ꢀ confidence interval.
*Adjustment was for age and miles driven per week.
**Hazard ratio for this variable is reported per standard deviation change.

M190
MARGOLIS ET AL.
Table 4. Multivariable Model (n ꢀ 1226)
tion with motor vehicle crashes. If the association between
very light drinking and motor vehicle crashes is due to
physically frail individuals giving up drinking, former
drinkers should have an elevated motor vehicle crash risk
compared to life-long abstainers, which we did not find. It is
possible that the finding is due to chance, given the number
of associations examined.
Variable
Adjusted HR
95% CI
p Value
Age, 5 y
1.14
1.16
REF
(1.01–1.29)
(1.07–1.25)
.03
.0002
Miles driven per week, 50 mi
No fall in past 12 months,
walked for exercise
No fall in past 12 months,
did not walk for exercise
Any fall in past 12 months,
walked for exercise
Any fall in past 12 months,
did not walk for exercise
Alcoholic drinks per week
Abstainers
1.01
1.20
2.07
(0.77–1.33) ꢂ.20
(0.90–1.60) ꢂ.20
A major methodological issue in research on motor vehi-
cle crashes in the elderly is the measurement of driving ex-
posure. This is particularly important given that older driv-
ers voluntarily curtail their driving as they develop
progressive sensory, cognitive, and physical impairment
(12). Studies that do not adequately account for driving ex-
posure will thus underestimate the impact that these impair-
ments have on motor vehicle crash risk. In our study,
weekly driving mileage was self-reported using a single
question, and we lacked specific information about when
women stopped driving. Data from the National Personal
Transportation Survey shows that women over age 65 drove
3308 miles annually in 1983 and 4750 miles annually in
1990 (38). This suggests that there may have been a modest
degree of underreporting of miles driven by the women in
our sample. Furthermore, driving exposure was not mea-
sured until 2 years after the start of motor vehicle crash as-
certainment. Changes in driving behavior between Visit 1
and Visit 2, including a reduction in weekly mileage driven
in response to a motor vehicle crash, are not captured by our
data. This may have biased some associations.
Several of our potential motor vehicle crash risk factors
(depression, hand and foot reaction times, Trail Making Part
B, and Digit Symbol Substitution) were also measured at
Visit 2 in 1988–1989. Because motor vehicle crashes from
1986–1995 were included, about one third preceded the
measurement of these five variables. However, limiting fol-
low-up from Visit 2 onward would have unacceptably low-
ered our study power. The lack of true prospective follow-
up with respect to these variables may have obscured or
weakened any association with motor vehicle crashes.
Our results should be interpreted with caution for a num-
ber of other reasons. The study sample consisted of predom-
inantly white women from Oregon. Our outcome data were
limited to motor vehicle accidents that resulted in a police
report; therefore, near misses, moving violations, and less
serious crashes were not ascertained, nor do we know which
of the motor vehicle crashes resulted in injuries. We may
have underestimated the number of crashes because we did
not have the ability to ascertain crashes that occurred out-
side of Oregon. No information was available on other dis-
eases, in particular cardiovascular disease.
(1.53–2.79)
.0001
REF
0.99
1.46
1.13
0.86
1.10
1.10
Former Drinkers
0.1–1.0
1.1–3.0
ꢂ3
(0.71–1.38) ꢂ.20
(1.05–2.02)
(0.78–1.65) ꢂ.20
(0.59–1.24) ꢂ.20
(1.00–1.23)
(0.99–1.22)
.02
Systolic blood pressure drop, mm Hg*
Foot reaction time, s*
.05
.08
Notes: HR ꢀ hazard ratio; CI ꢀ confidence interval; REF ꢀ reference value
for the hazard ratio, which equals 1.0.
*Hazard ratio per standard deviation change (12.5 mm Hg for systolic blood
pressure and 0.06 sec for foot reaction time).
This is in general agreement with previous studies (6,
11,14,32). Most previous studies have not found an associa-
tion of cognitive impairment with motor vehicle crashes
when standardized screening instruments have been used
(9,11,32,34). However, the women in SOF were fairly
healthy and cognitively intact when they entered the study;
therefore, the lack of association of these variables with mo-
tor vehicle crashes is not unexpected.
Our data are consistent with the majority of other studies
that have found no association of diabetes (6,11,31,34) and
osteoarthritis (6,11,31,34,35) with motor vehicle crashes.
The evidence for stroke as a risk factor for motor vehicle
crashes is less consistent, with two positive studies (11,34)
and two negative studies (6,31). Women in SOF were not
asked if they had heart disease until Visit 4, so we were un-
able to investigate heart disease as a risk factor for motor
vehicle crashes. However, one cross-sectional study (35)
and two case-control studies (31,34) have found an age-
independent association between heart disease and motor ve-
hicle accidents. Our findings are in accord with other stud-
ies that have found no association of visual acuity, contrast
sensitivity, or ocular disease with motor vehicle crashes
(2,7,9,11,32).
Previous studies have had inconsistent findings with re-
gard to depression as a motor vehicle crash risk factor, with
two studies reporting an association (11,14) and three stud-
ies reporting no association (6,32,34). Only 3.5% of the
women in SOF scored high on the Geriatric Depression
Scale; therefore, our study’s power to find an association of
depression with motor vehicle crashes is limited.
Although three other prospective studies found that the
use of benzodiazepines was not associated with motor vehi-
cle crashes (6,11,34), these findings are contradicted by sev-
eral case-control studies with much greater power due to the
large number of subjects studied (36,37). Two other studies
of older drivers found no association of alcohol consump-
In conclusion, this prospective study identified motor ve-
hicle crash risk factors that can be measured in the clinical
setting: increased age, increased weekly mileage driven, a
history of falls in sedentary women, increased systolic
blood pressure drop, and foot reaction time. Other neuro-
muscular, vision, or cognitive tests and knowledge of func-
tional status and medical comorbidity appeared to contrib-
ute little to predicting motor vehicle crash risk in this
relatively healthy, unimpaired group of women. Future re-
search is needed to examine whether change in physical and
cognitive functions predicts motor vehicle crashes and

CAR CRASHES AND OLDER WOMEN
M191
terrelationship and reliability of activity monitoring, activity surveys,
and caloric intake. J Gerontol. 1983;38:394–397.
18. Lawton MP, Brody EM. Assessment of older people: self-maintaining
and instrumental activities of daily living. Gerontologist. 1969;9:179–
186.
whether the risk of motor vehicle crashes can be lowered by
interventions aimed at ameliorating modifiable risk factors,
such as a history of falls or slow foot reaction time.
Acknowledgments
19. Fitti JE, Kovar MG. The supplement on aging to the 1984 National
Health Interview Survey. Vital Health Stat 1. 1987;21:1–115.
20. Yesavage J. Geriatric Depression Scale. Psychopharmacol Bull. 1988;
24(4):709–711.
21. Gerety M, Williams J, Mulrow C, et al. Performance of case-finding
tools for depression in the nursing home: influence of clinical and
functional characteristics and selection of optimal threshold scores. J
Am Geriatr Soc. 1994;42(10):1103–1109.
22. Bailey IL, Lovie JE. New design principles for visual acuity letter
charts. Am J Optom Physiol Optics. 1976;53(11):740–744.
23. Westheimer G. Scaling of visual acuity measurements. Arch Ophthal-
mol. 1979;97:327–330.
24. Scialfa CT, Adams EM, Giovanetto M. Reliability of the Vistech con-
trast test system in a life-span adult sample. Optom Vis Sci. 1991;68:
270–274.
25. Ginsburg AP. A new contrast sensitivity vision test chart. Am J Optom
Physiol Optics. 1984;61(6):403–407.
26. Simons K. A comparison of the Frisby, Random-Dot E, TNO, and
Randot Circles stereotests in screening and office use. Arch Ophthal-
mol. 1981;99:446–452.
27. Gibson JJ. The Perception of the Visual World. Westport, CT: Green-
wood Press; 1950.
28. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: a practi-
cal method for grading the cognitive state of patients for the clinician.
J Psychiatr Res. 1975;12:189–198.
29. Reitan R. Validity of the Trail Making Test as an indicator of organic
brain damage. Percept Mot Skills. 1958;8:271–276.
30. Wechsler D. Wechsler Adult Intelligence Scale–Revised. New York:
Psychological Corporation; 1988.
31. Koepsell TD, Wolf ME, McCloskey L, et al. Medical conditions and
motor vehicle collision injuries in older adults. J Am Geriatr Soc.
1994;42:695–700.
32. Marottoli RA, Cooney LM, Wagner DR, McPhil JD, Tinetti ME. Pre-
dictors of automobile crashes and moving violations among elderly
drivers. Ann Intern Med. 1994;121(11):842–846.
33. Ensrud K, Blackwell T, Hanlon J, et al. Central nervous system active
medications and risk of falls and fractures in older women. J Bone
Miner Res. 1999;114(suppl):526S.
34. McGwin G, Sims R, Pulley L, Roseman J. Relations among chronic
medical conditions, medications, and automobile crashes in the el-
derly: a population-based case-control study. Am J Epidemiol. 2000;
152(5):424–431.
35. Gallo JJ, Rebok GW, Lesikar SE. The driving habits of adults aged 60
years and older. J Am Geriatr Soc. 1999;47:335–341.
36. Ray WA, Fought RL, Decker MD. Psychoactive drugs and the risk of
injurious motor vehicle crashes in elderly drivers. Am J Epidemiol.
1992;136(7):873–883.
This research was supported by Public Health Service grants AG05407,
AR35582, AG05394, AR35584, and AR35583. Dr. Margolis was sup-
ported by K23 award HL03996.
Address correspondence to Karen Margolis, MD, MPH, Division of
Clinical Epidemiology, Hennepin County Medical Center (865B), 701
Park Avenue South, Minneapolis, MN 55415. E-mail: margo006@tc.
umn.edu
References
1. Retchin SM, Anapolle J. An overview of the older driver. Clin Geriatr
Med. 1993;9(2):279–296.
2. McCloskey LW, Koepsell TD, Wolf ME, Buchner DM. Motor vehicle
collision injuries and sensory impairments of older drivers. Age Age-
ing. 1994;23:267–273.
3. Szlyk JP, Seiple W, Viana M. Relative effects of age and compro-
mised vision on driving performance. Hum Factors. 1995;37(2):430–
436.
4. Marottoli RA, Richardson ED, O’Connor TZ, Peduzzi P, Cooney LM,
Tinetti ME. Prospective predictors of crashes among older drivers. J
Am Geriatr Soc. 1998;46(9):S23.
5. Shipp MD. Potential human and economic cost-savings attributable to
vision testing policies for driver license renewal, 1989–1991. Optom
Vis Sci. 1998;75(2):103–118.
6. Sims RV, Owsley C, Allman RM, Ball K, Smoot TM. A preliminary
assessment of the medical and functional factors associated with vehi-
cle crashes by older adults. J Am Geriatr Soc. 1998;46:556–561.
7. Ivers RQ, Mitchell P, Cumming RG. Sensory impairment and driving:
the Blue Mountains Eye Study. Am J Public Health. 1999;89(1):85–87.
8. Voelker R. Crash risk among older adults studied. JAMA. 1999;
282(17):1610–1611.
9. Owsley C, Ball K, McGwin G, et al. Visual processing impairment
and risk of motor crash among older adults. JAMA. 1998;279:1083–
1088.
10. Hu PS, Trumble DA, Foley DJ, Eberhard JW, Wallace RB. Crash risks
of older drivers: a panel data analysis. Accid Anal Prev. 1998;30(5):
569–581.
11. Sims RV, McGwin G, Allman RM, Ball K, Owsley C. Exploratory
study of incident vehicle crashes among older drivers. J Gerontol Med
Sci. 2000;55A:M22–M27.
12. Guibert R, Duarte-Franco E, Ciampi A, Potvin L, Loiselle J, Philibert
L. Medical conditions and the risk of motor vehicle crashes in men.
Arch Fam Med. 1998;7(6):554–558.
13. McGwin G, Sims RV, Pulley L, Roseman JM. Diabetes and automo-
bile crashes in the elderly: a population-based case-control study.
Diabetes Care. 1999;22:220–227.
14. Foley DJ, Wallace RB, Eberhard J. Risk factors for motor vehicle
crashes among older drivers in a rural community. J Am Geriatr Soc.
1995;43:776–781.
15. Cummings SR, Black DM, Nevitt MC, et al. Appendicular bone den-
sity and age predict hip fracture in women. JAMA. 1990;263:665–668.
16. Paffenbarger RS, Wing AL, Hyde RT. Physical activity as an index of
heart attack risk in college alumni. Am J Epidemiol. 1978;108(3):161–
175.
37. Thomas RE. Benzodiazepine use and motor vehicle accidents. Can
Fam Physician. 1998;44(April):799–808.
38. Federal Highway Administration. 1990 National Personal Transpor-
tation: Summary of Travel Trends. Washington, DC: United States
Department of Transportation; 1992.
17. LaPorte RE, Black-Sandler R, Cauley JA, Link M, Bayles C, Marks B.
The assessment of physical activity in older women: analysis of the in-
Received June 18, 2001
Accepted September 25, 2001