The strik- ing feature of Figure 2 is that for all four sex-race groups the motor vehicle accident mortality rate is higher in the noninspection compared with the in-.
Do states with compulsory motor vehicle inspection show lower mortality rates from motor vehicle accidents? In endeavoring to answer this question the authors conclude that no data are available concerning the effect of increasing doses of more stringent inspection procedures on morbidity and mortality from motor vehicle accidents. Much more investigation is needed to clarify this problem.
MOTOR VEHICLE INSPECTION AND MOTOR VEHICLE ACCIDENT MORTALITY Theodore Colton, Sc.D., and Robert C. Buxbaum, M.D.
THE toll measured in lives lost, injury sustained, and property damage from motor vehicle accidents needs no emphasis. It is a medical, social, and economic problem of national importance. Attention has recently been focused on the vehicle, its design deficiencies and its propensity toward mechanical failure. Thus, the vehicle is viewed as an etiologic agent that, along with other factors such as the driver, the environment, and others, contributes to serious injury and death. We are concerned in this report with the role of mechanical failure in motor vehicle accident mortality. The underlying rationale of this investigation is as follows: If mechanical failure contributes to mortality, then an attempt at identifying and correcting mechanical failure should decrease mortality. Compulsory motor vehicle inspection is one attempt to identify and correct mechanical failure. Some states have statewide inspection programs, others do not. Hence, the question we pose is, do states with compulsory motor vehicle inspection exhibit lower motor vehicle accident mortality rates? 1090
We have previously reported on this and showed that among males age 45-54 in 1960 the mortality was lower in inspection than in noninspection states. Furthermore, this difference was larger for nonwhites compared to whites and was maintained over a variety of other disturbing type variables which included population density, per capita income, and mortality from other accidents. The purpose of this report is to consider all age groups, and both sexes as well.
Material and Methods Our sources of information are publicly available in vital statistics2 and census3 reports. The year 1960 was chosen because calculation of mortality requires data on motor vehicle accident deaths and population at risk classified by state, race, sex, and age. The detailed population information is available only in census years. The states with compulsory motor vehicle inspection4 in 1960 are shown in Figure 1. In this report we have grouped inspection states regardless of their varyVOL. 58, NO. 6. A.J.P.H.
.
ing inspection procedures. Furthermore, some states inspect only once a year, some twice. Some use state-owned and state-run inspection sites; others use private garages and filling stations franchised by the state. With the data at hand, we felt it was not possible to analyze adequately any subgrouping of inspection states. New York, which inspects only those automobiles four or more years old, was classified as an inspection state. Connecticut, which requires inspection of used cars purchased out of state and instate cars ten or more years old when ownership changes, was classified as a noninspection state. Several states have inspection programs only in isolated municipalities. These states were classified as noninspection states. In 1960 there were 17 inspection (including the District of Columbia) and 34 noninspection states. The analysis utilizes age-specific and
......
...
. ..........r
.......... .
N . Jr _.
MOTOR VEHICLE INSPECTION
age-adjusted mortality rates. The ageadjusted mortality rates have been calculated by the direct method using the total population by age of the United States in 1960.3 The elimination of the disturbing or concomitant variables and the significance testing is achieved by weighted covariance analysis. Results Sex-Race-Age-Specific Mortality The age-specific motor vehicle accident mortality rates in the aggregate of the inspection and noninspection states are shown in Figure 2 separately for white males, nonwhite males, white females, and nonwhite females. The striking feature of Figure 2 is that for all four sex-race groups the motor vehicle accident mortality rate is higher in the noninspection compared with the inspection states at almost every age. Figure 2 also indicates some previ-
A I /
. ..
I
Figure 1-Motor vehicle inspection states, 1960. JUNE, 1968
1091
80
706 0
r
oI
0
;60 50
o
0 0 0
0. . °407 6
0
z
!f (r
Z!, .2 8
2
0
10
20
30
50
40
60
70
80
Age (years)
Age (years)
-50
.50 -40
Om
WHITE FEMALES
40 40
NON-WHITE FEMALES
30 ° C; 30
30 0
20_ !
'
-a 20 I0
Non -Inspection States ,-.-
Non-lnspection States
5.%'.0
cr
_: l-
0
l
Inspection Stotes I 10
O)
i
5.
Itspeclion
8
li
States
7
20
30
40
50
60
70
0
20
30
Age (years)
ously well-known information regarding motor vehicle accident mortality by age, race, and sex. First, the rates at each age are higher in males than in females. For females there is little difference in the age-specific rates of whites and nonwhites. For males, both whites and nonwhites have their highest mortality at age 20-25 with a sharp decrease after age 25. For white males there is a fairly steep rise beginning at about age 50, while for nonwhite males there is a moderate rise beginning at this age. It is at the older ages that pedestrian deaths become increasingly important and the rise in over-all motor vehicle mortality beyond age 50 is mainly this pedestrian effect. In the middle years, age 25 to 65, the nonwhite males have appreciably higher mortality rates than their white counterparts. The largest
40
50
60
70
80
Age (yeors)
Figure 2-Sex-race-age-specific motor vehicle accident mortality noninspection states, 1960 (rate per 100,000 population).
1 092
A *0I %
rates in
inspection and
differential in white and nonwhite male mortality is at age 45-54, the subject of our previous report. The age-adjusted mortality rates by sex and race are shown in Table 1. These age-adjusted rates are for all ages and for ages 15-64, the age group which encompasses almost all of the driving population at risk. For each sex-race category the age-adjusted rates are higher in noninspection than in inspection states. The 15-64 age-adjusted rates exhibit a remarkably consistent pattern of a 50 per cent higher rate in the noninspection states. Table 1 also shows male higher than female mortality, nonwhite male higher than white male mortality, and white female approximately equivalent to nonwhite female mortality. Table 2 gives age-specific rates, regardless of sex and race. Again, at each VOL. 58. NO. 6. A.J.P.H.
MOTOR VEHICLE INSPECTION
Table 1-Age-adjusted* mortality rates from motor vehicle accidents by sex and color in inspection vs. noninspection states, 1960 (rate per 100,000 population) All ages Noninspection Inspection states states
Age 15-64
Ratiot
Noninspection states
Inspection states
Ratiot
Males: White Nonwhite
35.0 41.4
25.4 28.5
1.38 1.45
44.0 54.6
30.0 36.1
1.47 1.51
Females: White Nonwhite
12.4 11.4
8.4 8.4
1.48 1.36
13.5 13.1
9.1 9.0
1.49 1.45
Total population
23.8
16.8
1.42
29.1
20.0
1.45
Adjusted to total US population bv age in 1960. t Rate in noninspection states/rate in inspection states.
*
age there are higher rates in noninspeetion states. On a percentage basis, the largest discrepancy occurs between ages 25-54.
Age-Adjusted Mortality by State For each state the age-adjusted mortality rate among white and nonwhite males age 15-64 appears in Table 3. The rank of the state in its male age-adjusted motor vehicle accident mortality is also indicated (1 means lowest mortality). It is clear from Table 3 that (a) there are large differences in motor vehicle accident mortality among the states, and (b) there is overlap in the mortality rates of inspection and noninspection states. The age-adjusted motor vehicle accident mortality among white males 15-64 is lowest in Rhode Island with a rate of 13.4 per 100,000 and highest in Wyoming with a rate of 96.0 per 100,000, more than a sevenfold range in rates. Among inspection states, Colorado, Vermont, New Mexico, and Texas rank high in mortality and illustrate one aspect of overlap. Among noninspection states, Connecticut, Illinois, Maryland, and Washington rank low in mortality, and illustrate the other aspect of overlap. JUNE. 1968
The age-adjusted motor vehicle accident mortality among nonwhite males 15-64 is ranked only in those 24 states where there is a population base of at least 50,000. Rates range from a low of 20.6 in the District of Columbia to a high of 76.0 in Oklahoma. Among the five inspection states, none has an appreciably high rate. Among the 19 ranked noninspection states Illinois, Indiana, Maryland, Michigan, and Ohio have low rates.
Comparing white and nonwhite rates Table 2-Mortality rate from motor vehicle accidents by age in inspection and noninspection states, 1960 (rate per 100,000 population) Noninspection states
Inspection states
Ratio*
9.3 41.6 27.7 22.2 24.4 28.2 38.1
7.0 31.0 18.2 14.0 16.1 20.1 28.6
1.34 1.34 1.52 1.59 1.52 1.41 1.33
population 23.8
16.8
1.42
Age (years)
Under 15 15-24 25-34 35-44
45-54 55-64 65 and older Total *
Rate in noninspection
states/rate
in
inspection
states.
l1093
within states, some states have appreciably higher white than nonwhite rates (Mississippi and Michigan), some have about the same white and nonwhite rates (Indiana, Texas, and the District of Columbia) and others have appreciably lower white than nonwhite rates (Florida, Oklahoma, New Jersey, North Carolina, Maryland). Table 3 also lists, for each state, the change in the age-adjusted motor vehicle accident rate between 1950 and 1960 and the state's rank in change. In some states the rate has risen, in others it has declined. Furthermore, declining rates are not necessarily associated with inspection. For white males, Idaho, a noninspection state, had the largest decline, while Wyoming experienced the largest increase. For the nonwhite males the largest 1950 to 1960 decrease occurred in Virginia while the largest increase occurred in Oklahoma. Table 4 shows the aggregate changes
in rates from 1950 to 1960 for those states inspecting throughout the decade, those initiating inspection between 1951 and 1959 (New Mexico, New York, Rhode Island, Texas and West Virginia), and those that did not inspect at all during the decade. The results for white males do not provide any evidence that continual inspection or the initiation of inspection are associated with an appreciable change in mortality. For nonwhite males there is some indication of a slight trend. On the average, states maintaining inspection throughout the decade exhibited a decrease in mortality (-8.6 per 100,000), those beginning inspection exhibited a very slight decrease (-0.7 per 100,000), while those not inspecting during the decade show a slight increase ( + 0.4 per 100,000).
Influence of Other Variables In examining the role of inspection in motor vehicle accident mortality it is
Table 3-Age-adjusted* motor vehicle accident mortality rates by state for white and nonwhite males 15-64 years old, 1960, and change in rate from 1950 to 1960 (rate per 100,000 population) White males
Nonwhite males
State
1960 pop'n 15-64 (000)
1960 rate
Rank
Colo. Del. D. C. Maine Mass.
495 114 113 276 1,455
49.9 37.5 20.6 29.6 21.0
31 14 2 8 5
Inspection states -1.3 22 16 -3.9 17 18 -6.6 11 121 3.5 34 2 3.0 33 37
N. H. N. J. N. M. N. Y. Pa. R. I.
174 1,683 253 4,583 3,081 254
37.4 20.9 79.1 23.2 28.1 13.4
13 3 47 6 7 1
3.5 0.4 11.1 1.4 0.9 -3.6
35 28 43 30 29 18
1 152 19 438 245 6
(0.0) 41.5 (177.3) 27.7 26.1
Tex. Utah Vt. Va. W. Va.
2,477 241 110 980 504
47.7 42.4 51.4 35.7 42.1
27 21 35 12 20
-12.9 -5.7 4.9 -13.3 0.1
4 13 36 3 26
321 5 0 228 23
47.5 (196.6) (0.0) 40.9 (83.0)
Total
16,791
30.8
1094
Change from '50 to '60 Rank
-1.8
1960 pop'n 15-64 (000)
1,632
1960 rate
(27.6) (54.4) 20.6 (204.6) (20.1)
Rank
1
8 4 2
(21.5)
37.3
Change from '50 to '60
(-9.9) (-32.0) -14.6 (70.7) (-7.8)
(0.0) -3.8 (32.7) 1.7 -2.7
Rank
2
12 15 13
(-80.6) 11
-7.8
7
(161.5) 7
(0.0) -17.7 (45.9)
1
-4.7
VOL. 58. NO. 6. A.J.P.H.
MOTOR VEHICLE INSPECTION
Table 3-Continued White males
Nonwhite males 1960 pop'n
1960 pop'n
15-64
State
15-64 (000)
1960 rate
Rank
Ala. Ariz. Ark. Calif. Conn. Fla.
677 352 406 4,416 724 1,177
57.5 57.7 56.2 47.7 21.0 40.9
42 43 41 24 4 18
Noninspection -7.1 8 2 -15.5 12.8 45 -4.2 16 -2.0 19 6 -10.6
Ga. Idaho Ill. Ind. Iowa Kans.
854 188 2,692 1,269 764 601
51.1 52.6 31.3 47.7 43.0 44.6
34 36 10 28 22 23
0.3 -20.8 -7.1 1.9 1.6 -4.8
27 1 9 32 31 14
279 3 297 74 8 28
Ky. La. Md. Mich. Minn. Miss. Mo.
815 649 784 2,054 941 376 1,135
53.8 49.0 30.3 40.7 41.2 66.5 47.9
38 30 9 17 19 46 29
6.1 8.9 -6.7 -4.4 7.0 14.2 -11.0
38 42 10 15 39 46 5
59 254 149 209 12 208 104
Mont. Neb. Nev. N. C. N. D. Ohio Okla.
189 392 87 1,042 179 2,570 624
62.4 39.8 95.3 53.0 55.5 38.0 49.9
44 16 48 37 40 15 32
5.7 -1.5 14.5 -0.6 19.1 -2.0 8.1
37 21 47 23 48 20 41
6 10 7 295 3 224 55
(155.6) (44.5) (107.1) 71.7 (0.0) 34.3 76.0
Ore. S. C. S. D. Tenn. Wash. Wisc. Wyo.
504 472 185 880 816
1,103 97
50.3 55.4 62.4 45.0 35.5 47.6 96.0
33 39 45 25 11 26 49
-6.1 -8.3 12.7 -0.3 0.0 8.0 25.7
12 7 44 24 25 40 49
11 199 7 149 32 25 2
(89.3) 69.6 (282.9) 55.1 (84.1) (36.5) (88.4)
Total
30,011
44.6
Change from '50 to '60 Rank
-1.8
(000)
1960 rate
states 236 66.7 34 (108.2) 89 53.8 392 47.8 32 (42.6) 246 73.9
3,738
Rank
19
Change from '50 to '60
13.0
(15.9) 14 12
17.7 -7.4
(-3.4)
Rank
21 22 8
23
6.6
17
59.2 (100.3) 26.3 46.8 (10.7) (24.6)
17
-3.9 (17.1) -13.7 -9.3 (-17.6) (-17.5)
11
68.4 60.4 47.4 31.2
20 18 10 5
20 23 9 4
(74.8) 55.5 52.5
16 13
8.5 17.9 -5.3 -12.7 (27.1) 4.2 -5.2
55.2
3 9
3 6
16 10
(26.5) (-5.8) (-66.9) 22
7.9
(-39.7) 6 24
21 15
-11.7 33.6
(-32.3) 1.3 (95.2) 8.1 (24.4) (-58.0)
18 5 24
14
19
(-73.7) 0.4
NOTE: Figures in parentheses indicate the rate has a population base less than 50,000. * All rates adjusted to total United States population by age In 1960.
important to recognize that the inspection states are a self-selected group. Hence, the comparison of mortality in inspection and noninspection states should account for the possibility that the inspection and noninspection states may be very different in several ways other than inspection and that perhaps JUNE. 1 968
these other variables explain the observed mortality difference. For example, population density is inversely related to motor vehicle accident mortal-
ity; the denser the population, the lower the rate. If inspection states tend to be the more densely populated (which, from Figure 1, seems to be the case), 1 095
Table 4-Change in age-adjusted motor vehicle accident mortality rate from 1950 to 1960 for. white and nonwhite males 15-64 years old by inspection status (rate per 100,000 population)
Inspection status of states Began inspection before 1950
Began inspection between 1951 and 1959
1950-1960 change in rate (per 100,000 population) White Nonwhite males males 15-64 15-64 -0.7
-8.6
state. -2.9
-0.7
Did not inspect during 1950 to 1960 11.8
+0.4
then some or perhaps all of the difference between the inspection and noninspection mortality rates is explainable by differences in population density. The question is how much of this difference is attributable to population density? Does any residual difference not explainable by population density provide evidence for a separate effect of motor vehicle inspection? The same argument applies to other variables such as per capita income and mortality from other accidents. (The argument with the latter is that inspection states are more cautious and safetyconscious, and hence tend to be states with lower mortality rates for other accidents as well. In other words, the difference in motor vehicle accident mortality is not specific but is related to or explainable by mortality from other accidents.) Finally, the argument applies to all three of these variables simultaneously. Removal of the influence of these disturbing variables is handled by a weighted covariance analysis and the results are shown in Table 5. The sig1 096
nificance tests of Table 5 examine whether the difference in mortality between the aggregate of inspection and noninspection states is above and beyond what might be expected by chance considering the state-to-state mortality variation. For this purpose we consider the age-adjusted mortality rate in each state as an observation from a random sample, with the state as the sampling unit. The units are unequal in size (size is the population 15-64) and the subsequent analysis takes this into account,5 weighting according to the size of the The results indicate that the difference in white male age-adjusted motor vehicle accident mortality between inspection and noninspection states is
highly significant (P
