Traffic Injury Prevention
ISSN: 1538-9588 (Print) 1538-957X (Online) Journal homepage: http://www.tandfonline.com/loi/gcpi20
Mature Driver Laws and State Predictors of Motor Vehicle Crash Fatality Rates Among the Elderly: A Cross-sectional Ecological Study Teresa M. Bell, Nan Qiao & Ben L. Zarzaur To cite this article: Teresa M. Bell, Nan Qiao & Ben L. Zarzaur (2015) Mature Driver Laws and State Predictors of Motor Vehicle Crash Fatality Rates Among the Elderly: A Cross-sectional Ecological Study, Traffic Injury Prevention, 16:7, 669-676, DOI: 10.1080/15389588.2014.999858 To link to this article: http://dx.doi.org/10.1080/15389588.2014.999858
Accepted author version posted online: 08 Jan 2015. Published online: 08 Jan 2015. Submit your article to this journal
Article views: 85
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=gcpi20 Download by: [University of Nebraska, Lincoln]
Date: 05 November 2015, At: 17:48
Traffic Injury Prevention (2015) 16, 669–676 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2014.999858
Mature Driver Laws and State Predictors of Motor Vehicle Crash Fatality Rates Among the Elderly: A Cross-sectional Ecological Study TERESA M. BELL1, NAN QIAO2, and BEN L. ZARZAUR1 1
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
2
Department of Surgery, Indiana University School of Medicine, Indianapolis, Indiana Department of Economics, Indiana University, Indianapolis, Indiana
Received 1 October 2014, Accepted 15 December 2014
Objective: State-level data have indicated that motor vehicle crash (MVC) fatality rates among the elderly vary widely across states. To date, the majority of states have implemented mature driver laws, which often require more frequent license renewals, in-person renewal, and vision testing for drivers above a certain age. We sought to evaluate the impact of mature driver laws on states’ MVC fatality rates among the elderly while examining other state-level determinants of MVC-related deaths. Methods: We performed a cross-sectional ecological study and modeled state MVC fatality rates for the population over age 65 as a function of state transportation policies and demographic, health system, population health, travel, and climate characteristics using a general linear model. Principal component analysis was used to reduce dimensionality of the data and overcome multicollinearity of state predictor variables. Results: Higher average temperature, higher gas prices, and a greater number of emergency medicine physicians to population size were significantly associated with lower MVC fatality rates. Positive predictors of MVC fatality rates were percentage of population overweight or obese and percentage with college degree over the age of 65. Having any restriction on elderly drivers was associated with a higher MVC fatality rate and no individual component of mature driver laws (shortened renewal cycle, in-person renewal, and vision testing) was significantly associated with lower fatality MVC rates for adults over 65. Conclusions: Mature driver laws are not associated with lower state MVC fatality rates among the elderly. Keywords: elderly, fatality, licensing, old drivers, older occupant, regulations
Introduction In the context of the aging babyboom generation, public concern for the safety of older drivers has continued to grow. As of 2012, there were nearly 36 million licensed drivers over the age of 65 in the United States (Federal Highway Administration, Department of Transportation 2014). Compared to younger drivers, older drivers have more accidents per mile driven and are more likely to be fatally injured due to increased fragility (Li et al. 2003; U.S. Department of Transportation 1997). One study has projected that as the population of older drivers in the United States grows, there will be a 178% increase in motor vehicle crashes and a 155% increase in fatal accident involvement by 2030 (Lyman et al. 2002). Despite these reports, recent studies have indicated that older drivers are less likely to be involved in accidents than younger drivers and
Associate Editor Jessica Jermakian oversaw the review of this article. Address correspondence to Teresa Maria Bell, PhD, Assistant Professor of Surgery, 702 Rotary Circle, Room 022A, Indianapolis, IN 46202. E-mail: [email protected]
that the rate of crash fatalities in older drivers is decreasing nationally (Cheung and McCartt 2011; Cicchino and McCartt 2014). Although national trends have demonstrated a recession in fatal crash involvement for older drivers, state-level data indicate that motor vehicle crash (MVC) fatality rates among the elderly vary widely across states, from approximately 1.5 deaths per 100,000 people over the age of 65 in Massachusetts to 20.7 per 100,000 in Mississippi (Centers for Disease Control and Prevention 2013). Currently, 33 states have implemented mature driver laws, which require vision testing and more frequent license renewals for drivers above a certain age (Governor’s Highway Safety Association 2013). Research has indicated that drivers with visual and cognitive impairments associated with aging are at greater risk for motor vehicle crashes; therefore, mandatory in-person license renewal and vision screenings may reduce the number of licensed older drivers with vision and cognition problems (Anstey et al. 2005; Rubin et al. 2007). However, previous studies have had mixed findings in regards to whether or not state licensure laws for the elderly have impacted fatality rates for older drivers (Grabowski et al. 2004; Langford et al. 2004, 2008; Levy et al. 1995; Morrisey and Grabowski 2005).
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
670 To date, there have been no previous studies that simultaneously investigated the effect of mature driver policies and state characteristics on MVC fatality rates among the elderly. It is unknown whether the large gap in fatality rates is explained by variation in state sociodemographics, health status, or health care systems in addition to licensing policies. Baker (1975) introduced a framework for determinants of injury that attempts to explain how individual characteristics such as age result in differential injury death rates for population subgroups. She proposed that age, socioeconomic, cultural, physical, and other individual characteristics modify death rates for groups of people by altering 4 categories of injury determinants including (1) the quantity and quality of exposure to environments conducive to injury; (2) the ability to perceive and respond to hazards; (3) injury thresholds (i.e., ability to withstand mechanical insults to the body); and (4) probability of recovery (e.g., access to life-saving emergency care). It follows that health-related factors such as states’ chronic disease burden, quality of health care, and development of emergency response systems could potentially account for differences in MVC fatality rates by modifying the latter 2 categories of determinants, whereas driving policies typically target the first 2 categories. Additionally, contextual factors such as the economic costs of driving, population density, and climate have been established in the literature as contributors to motor vehicle accidents (Chi et al. 2012; Theofilatos and Yannis 2014; Zwerling et al. 2005). These factors are likely associated with driving exposure, crash risk, and ability to respond to hazards, which can also be assigned to Baker’s (1975) first 2 categories of injury determinants. Examining additional state characteristics that influence MVC fatality rates through mechanisms that are both similar to and different from licensing policies will help parse out the effects of each factor so that future research on the design of policy interventions is more informed. The objective of this cross-sectional ecological study is to examine the relationship between state mature driver laws and MVC fatality rates among the elderly. It also aims to investigate the effect of mature driver laws relative to other state-level factors that potentially influence MVC fatality rates among the elderly such as demographic, health system, population health, travel, and climate characteristics.
Bell et al. number of driver’s licenses per 1,000 residents came from Our Nation’s Highways (Federal Highway Administration 2009). Data on the number of person-miles driven for elderly drivers was obtained from the 2009 National Household Travel Survey (U.S. Department of Transportation 2009). Climate data were provided by the National Oceanic and Atmospheric Administration’s National Climatic Data, which is based on state averages from 1971 to 2000 (National Climatic Data Center 2013). Population health status and provider data were obtained from Kaiser Family Foundation State Health Facts (Kaiser Family Foundation 2010). State trauma system data were obtained from the American Trauma Society (2009).
Study Variables State elderly MVC fatality rate was the dependent variable used in this study. For the purposes of this article, we define elderly as individuals over the age of 65. The death rate from MVCs was age-adjusted to account for states that have a greater proportion of older senior citizens. Stable rate estimates for Alaska, Hawaii, and the District of Columbia could not be produced and these states were excluded from the final analysis. The independent variable of interest in the study was the presence of state laws that imposed additional requirements on older drivers in order to obtain driver’s licenses. Laws were categorized as shortened renewal cycle, in-person renewal requirement, and vision test requirement. We also incorporated state-level control variables into our model. Because many population demographic and health variables are collinear, we used principal component analysis to reduce dimensionality of the data. We selected a variety of state-level covariates that accounted for amount of driving done by residents over 65, cost of driving, population health, access to care, quality of care, and environmental factors. Principal component analysis was used to reduce dimensionality of the data and overcome multicollinearity of state predictor variables. We identified 11 distinct components from 35 potential state covariates. Covariates with eigenvalues over 1 were included in the regression analysis. The extracted covariates accounted for 90.85% of variance.
Regression Analysis
Methods Data Source State motor vehicle fatality data were obtained from the Centers for Disease Control and Prevention’s Web-based Injury Statistics Query and Reporting System database, which compiles vital statistics data (Centers for Disease Control and Prevention 2013). Because death due to motor vehicle crash is a rare outcome among the elderly, we aggregated MVC fatality data from 2006 to 2010 to ensure stability of rate estimates. Information on state licensure laws was obtained from the Governor’s Highway Safety Association (2013). The American Community Survey (U.S. Census 2011) provided state demographic data used in the analysis (U.S. Census 2011). The
We used a general linear model to determine the impact of elderly licensure laws on MVC fatality rates in the over 65 population. Models assessing the effect of the presence of any restriction on older drivers as well as individual components of laws were generated. Model 1 examined the effect of all mature driver laws jointly on fatality rates. Models 2, 3, and 4 examined the effect of each law component separately (shortened renewal cycle, in-person renewal requirement, and vision test requirement, respectively). Each of the final models included the following covariates: average annual person-miles driven (among persons 65 and older), population density, average temperature, average precipitation, percentage of the population over 65 with college degrees, percentage of population overweight or obese, percentage of Medicare
Mature Driver Laws and Fatality Rates
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
beneficiaries with prescription drug coverage, percentage of population with access to a level I or II trauma center within 60 minutes, emergency medicine physicians per 10,000 members of the population, average gas price per gallon, and driver license fee amount. We also controlled for the number of years between required license renewals for states. We used principal component analysis to determine model covariates rather than automated variable selection procedures such as forward, backward, or stepwise regressions in order to ensure that the most relevant variables representing each category of Baker’s (1975) injury determinants were included. IBM SPSS Statistics 22 (IBM, Armonk, NY) was used for all analyses.
Results We found that annual MVC fatality rates among people over the age of 65 varied greatly across states. Fatality rates ranged from approximately 1.5 deaths per 100,000 people over the age of 65 in Massachusetts to 20.7 per 100,000 people over 65 in Mississippi. The Northeastern states tended to be in the quintile with the lowest fatality rates, whereas the Midwestern states typically fell into quintiles with the highest rates (Figure 1). Thirty-two out of 50 states had at least one restriction on elderly drivers. Of these 32, 19 states shortened renewal cycles for older drivers, 14 required in-person renewal, and 11 required a vision test. The age at which these laws went into effect varied considerably between states. Shortened renewal cycle laws went into effect for drivers between the ages of 60 to 85, with 65 and 75 being the most common age requirements (n = 4 for each). The age range for in-person renewal requirements ranged from 64 to 80, with 70 being the most common age at which the law went into effect (n = 5). Lastly, vision test requirements tended to go into effect at even younger ages than the 2 other requirements and ranged from 40 to 80 years old. Mature driver laws were most common in Western states. Six of the 10 states with the lowest MVC fatality rates among the elderly had no restriction on elderly drivers, whereas 7 of the 10 states with the highest MVC fatality rates among the elderly did have mature driver laws in place (Figure 2). State characteristics also varied across quintiles of MVC fatality rates among the elderly. In general, states in the highest quintile had the lowest population density, lowest average temperature, highest prescription drug coverage rates, poorest access to level I and II trauma centers, lowest driver’s license fees, and longest license renewal cycles. States in the highest fatality rate quintile also had the greatest number of states with mature driver laws in place (Table 1). Regression results found that average temperature, gas price, and ratio of emergency medicine physicians to population size were significant negative predictors of MVC fatality rates. In 3 of the 4 models we found that higher annual temperatures were associated with a reduction in fatality rates (P = .0499). In model 1, gas price was a significant negative predictor, with higher gas prices being associated with a decrease in fatality rates (P = .022). A greater number of emergency
671 medicine physicians per 10,000 members of the population was also associated with a significant reduction in fatality rates in all 4 models (P < .001; Table 2). We found several positive predictors of MVC fatality rates among the elderly. Positive predictors included the percentage of population overweight or obese and the percentage with college degrees over the age of 65. The percentage of the population overweight or obese was strongly associated with a positive increase in MVC fatality rates among the elderly in all 4 models (P = .014, .020, .049, .033, respectively). We also found that a higher percentage of older adults with college degrees predicted significantly higher fatality rates in all 4 models (P = .006, .004, .024, .012). Lastly, model 1 indicated that having any restriction on elderly drivers was associated with a higher MVC fatality rate (P = .036). Model 2 also found that shortened renewal cycles predicted higher MVC fatality rates (P = .042). Neither inperson renewal nor vision test requirements had significant negative or positive effects on MVC fatality rates among the elderly (P = .862, .636).
Discussion Overall, our study found that mature driver laws were not associated with lower state MVC fatality rates among the elderly. Having any additional requirements for older drivers and a shortened renewal cycle were associated with higher fatality rates; however, in-person renewal and vision test requirements were not significant predictors of fatality rates. Our results differ from previous studies that reported a beneficial effect of elderly driver laws on MVC fatality rates (Nelson et al. 1992; Shipp 1998). Because our study is ecological, we can only determine that mature driver laws are associated with higher MVC fatality rates. It is possible that laws are endogenous to state characteristics and that higher MVC fatality rates among the elderly prompt state legislators to pass mature driver laws. These results could also be due to policies having unintended effects. For example, older drivers may feel that such policies are discriminatory and potentially unsafe drivers may be more motivated to renew their licenses as a result. It is also possible that these policies remove safe older drivers from the road, who then rely on drivers in age groups at higher crash risk for transportation. Future studies should assess individuals who did not renew driver’s licenses due to mature driver laws and determine whether or not they were fit to drive. This result is in agreement with other studies that have reported that mature driver laws did not significantly reduce MVC fatalities in the elderly (Grabowski et al. 2004; Morrisey and Grabowski 2005; Rock 1998). There are several potential explanations for these findings. One reason may be that we examined MVC fatality rates for the entire population over 65 and there may be a heterogeneous response to the laws in younger old and older old drivers. Studies that have reported significant effects of mature driving laws found that the laws were only significantly associated with reduced fatality rates in drivers over 80 years old (Grabowski et al. 2004). Considering that the rate of MVC fatalities in people between the ages
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
672
Bell et al.
Fig. 1. MVC fatality rates among people age 65 and older. State MVC fatality rates divided into quintiles are depicted. Rates were estimated based on data from 2006 to 2010 and were age-adjusted to account for state populations with a greater proportion of older elderly. Stable rate estimates of Alaska, Hawaii, and the District of Columbia could not be produced.
of 60 and 70 is lower than that of younger age groups, states with mature driver laws that impose restrictions at younger ages may not be targeting the most at-risk age group (Federal Highway Administration, Department of Transportation 2014). The challenge of implementing additional requirements and restrictions on older drivers may also be reflected in our results. For example, studies examining the role of visual
impairment in motor vehicle crashes have reported that glare sensitivity and visual field loss are significant predictors of crash involvement but static visual acuity is not; however, vision tests for drivers’ licensure are based primarily on static visual acuity (Keeffe et al. 2002; Rubin et al. 2007). Several studies have also reported that crash involvement can be predicted with various performance-based cognitive tests, yet implementation of these tests in real-world settings has been problematic
Fig. 2. Mature driver laws across states. The presence of a law is indicated by grey shading. Data were obtained from the Governors Highway Association in December 2012. States that passed laws after 2010 were not considered to have mature driver laws in our analysis, which examined data from 2006 to 2010.
Mature Driver Laws and Fatality Rates
673
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
Table 1. Variation in state characteristics by quintiles (Q) of MVC fatality rates among the elderly
MVC fatality rate 65+ Average annual person-miles driven, persons 65 and up Population density Average temperature (◦ F) Precipitation (inches) % College degree 65 and up Per capita income (US$) % Overweight or obese % Medicare beneficiaries with known prescription drug coverage % Population with access to level I or II trauma center within 60 minutes Emergency medicine physicians per 10,000 population Average gas price (US$/gallon) Driver’s license fees (US$) License renewal cycle (years) Any restriction on elderly drivers Shortened renewal cycle for elderly In-person renewal requirement Vision test requirement
Q1: Lowest
Q2: Low
Q3: Medium
Q4: High
Q5: Highest
2.5 (0.7) 265.8 (243.5)
4.5 (0.5) 354.4 (483.1)
6.7 (0.6) 444.0 (360.7)
9.4 (1.5) 367.8 (348.8)
15.6 (2.6) 300.5 (168.6)
510.3 (443.5) 53.9 (6.5) 42.8 (12.9) 21.8 (4.3) 28,326.56 (5,281.42) 63.2 (2.8) 40.5 (7.1)
204.8 (145.5) 52.1 (6.1) 37.4 (11.6) 22.8 (5.0) 27,499.70 (3,405.42) 62.9 (3.7) 40.2 (8.4)
80.8 (62.3) 53.8 (9.0) 34.9 (17.2) 19.5 (4.5) 23,993.30 (1,820.46) 63.8 (2.7) 43.4 (10.2)
189.5 (168.0) 51.5 (9.4) 43.0 (8.8) 23.3 (4.7) 27,614.00 (3,303.16) 63.3 (2.1) 41.2 (6.3)
30.8 (19.9) 49.0 (7.0) 28.5 (14.6) 20.6 (3.3) 23,412.56 (1,887.85) 64.2 (2.4) 49.3 (8.3)
84.8 (24.9)
89.2 (8.2)
65.7 (25.3)
85.0 (8.6)
58.56 (20.0)
1.7 (0.5)
1.5 (0.4)
1.1 (0.3)
1.3 (0.1)
1.1 (0.4)
2.76 (0.05) 9.78 (1.56) 5.1 (2.0) 5 (55.6%) 2 (22.2%) 4 (44.4%) 1 (11.1%)
2.80 (0.05) 9.50 (1.18) 6.3 (2.5) 6 (60.0%) 3 (30.0%) 2 (20.0%) 3 (30.0%)
2.78 (0.07) 8.50 (0.53) 5.7 (2.0) 6 (60.0%) 4 (40.0%) 3 (30.0%) 3 (30.0%)
2.77 (0.08) 9.30 (1.25) 5.8 (1.6) 5 (50.0%) 3 (30.0%) 2 (20.0%) 2 (20.0%)
2.76 (0.07) 8.44 (0.53) 6.6 (1.4) 7 (77.8%) 6 (66.7%) 1 (11.1%) 1 (11.1%)
Table 2. Coefficient estimates of mature driver laws and state characteristics Dependent variable: Age-adjusted rate of death due to MVC in population 65 and older
Intercept Average annual person-miles driven, persons 65 and up Population density Average temperature (◦ F) Precipitation (inches) % College degree 65 and up Per capita income % Overweight or obese % Medicare beneficiaries with known prescription drug coverage % Population with access to level I or II trauma center within 60 minutes Emergency medicine physicians per 10,000 population Average gas price (US$/gallon) Driver’s license fees (US$) License renewal cycle (years) Any restriction on elderly drivers Shortened renewal cycle for elderly
(1)
(2)
(3)
(4)
−7.956 (24.633) 0.000 (0.001) 0.000 (0.003) −0.307 (0.123)∗ −0.014 (0.054) 0.626 (0.219)∗∗ 0.000 (0.001) 91.403 (35.126)∗∗ 0.000 (0.002) 0.012 (0.017) −8.186 (1.638)∗∗∗ −0.019 (0.008)∗ 1.559 (1.444) −0.261 (0.328) 2.491 (1.154)∗
−11.647 (24.571) 0.000 (0.001) −0.001 (0.003) −0.294 (0.123)∗ −0.003 (0.055) 0.667 (0.221)∗∗ 0.000 (0.001) 85.990 (35.256)∗ 0.000 (0.002) 0.014 (0.017) −7.761 (1.624)∗∗∗ −0.014 (0.007) 1.472 (1.447) −0.313 (0.343)
−14.616 (26.259) 0.001 (0.001) 0.000 (0.004) −0.274 (0.142) −0.023 (0.058) 0.607 (0.245)∗ 0.000 (0.001) 87.933 (39.513)∗ 0.000 (0.002) 0.014 (0.018) −7.596 (1.754)∗∗∗ −0.014 (0.008) 1.346 (1.547) 0.069 (0.313)
−16.120 (26.342) 0.001 (0.001) 0.000 (0.004) −0.271 (0.131)∗ −0.021 (0.058) 0.626 (0.236)∗∗ 0.000 (0.001) 87.688 (37.587)∗ 0.000 (0.002) 0.016 (0.018) −7.530 (1.739)∗∗∗ −0.014 (0.008) 1.259 (1.555) 0.112 (0.328)
2.494 (1.191)∗
In-person renewal requirement Vision test requirement
Unstandardized beta coefficients for independent variables are reported. Standard error values are in parentheses. ∗ Significant at 0.05 level. ∗∗ Significant at 0.01 level. ∗∗∗ Significant at
ISSN: 1538-9588 (Print) 1538-957X (Online) Journal homepage: http://www.tandfonline.com/loi/gcpi20
Mature Driver Laws and State Predictors of Motor Vehicle Crash Fatality Rates Among the Elderly: A Cross-sectional Ecological Study Teresa M. Bell, Nan Qiao & Ben L. Zarzaur To cite this article: Teresa M. Bell, Nan Qiao & Ben L. Zarzaur (2015) Mature Driver Laws and State Predictors of Motor Vehicle Crash Fatality Rates Among the Elderly: A Cross-sectional Ecological Study, Traffic Injury Prevention, 16:7, 669-676, DOI: 10.1080/15389588.2014.999858 To link to this article: http://dx.doi.org/10.1080/15389588.2014.999858
Accepted author version posted online: 08 Jan 2015. Published online: 08 Jan 2015. Submit your article to this journal
Article views: 85
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=gcpi20 Download by: [University of Nebraska, Lincoln]
Date: 05 November 2015, At: 17:48
Traffic Injury Prevention (2015) 16, 669–676 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2014.999858
Mature Driver Laws and State Predictors of Motor Vehicle Crash Fatality Rates Among the Elderly: A Cross-sectional Ecological Study TERESA M. BELL1, NAN QIAO2, and BEN L. ZARZAUR1 1
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
2
Department of Surgery, Indiana University School of Medicine, Indianapolis, Indiana Department of Economics, Indiana University, Indianapolis, Indiana
Received 1 October 2014, Accepted 15 December 2014
Objective: State-level data have indicated that motor vehicle crash (MVC) fatality rates among the elderly vary widely across states. To date, the majority of states have implemented mature driver laws, which often require more frequent license renewals, in-person renewal, and vision testing for drivers above a certain age. We sought to evaluate the impact of mature driver laws on states’ MVC fatality rates among the elderly while examining other state-level determinants of MVC-related deaths. Methods: We performed a cross-sectional ecological study and modeled state MVC fatality rates for the population over age 65 as a function of state transportation policies and demographic, health system, population health, travel, and climate characteristics using a general linear model. Principal component analysis was used to reduce dimensionality of the data and overcome multicollinearity of state predictor variables. Results: Higher average temperature, higher gas prices, and a greater number of emergency medicine physicians to population size were significantly associated with lower MVC fatality rates. Positive predictors of MVC fatality rates were percentage of population overweight or obese and percentage with college degree over the age of 65. Having any restriction on elderly drivers was associated with a higher MVC fatality rate and no individual component of mature driver laws (shortened renewal cycle, in-person renewal, and vision testing) was significantly associated with lower fatality MVC rates for adults over 65. Conclusions: Mature driver laws are not associated with lower state MVC fatality rates among the elderly. Keywords: elderly, fatality, licensing, old drivers, older occupant, regulations
Introduction In the context of the aging babyboom generation, public concern for the safety of older drivers has continued to grow. As of 2012, there were nearly 36 million licensed drivers over the age of 65 in the United States (Federal Highway Administration, Department of Transportation 2014). Compared to younger drivers, older drivers have more accidents per mile driven and are more likely to be fatally injured due to increased fragility (Li et al. 2003; U.S. Department of Transportation 1997). One study has projected that as the population of older drivers in the United States grows, there will be a 178% increase in motor vehicle crashes and a 155% increase in fatal accident involvement by 2030 (Lyman et al. 2002). Despite these reports, recent studies have indicated that older drivers are less likely to be involved in accidents than younger drivers and
Associate Editor Jessica Jermakian oversaw the review of this article. Address correspondence to Teresa Maria Bell, PhD, Assistant Professor of Surgery, 702 Rotary Circle, Room 022A, Indianapolis, IN 46202. E-mail: [email protected]
that the rate of crash fatalities in older drivers is decreasing nationally (Cheung and McCartt 2011; Cicchino and McCartt 2014). Although national trends have demonstrated a recession in fatal crash involvement for older drivers, state-level data indicate that motor vehicle crash (MVC) fatality rates among the elderly vary widely across states, from approximately 1.5 deaths per 100,000 people over the age of 65 in Massachusetts to 20.7 per 100,000 in Mississippi (Centers for Disease Control and Prevention 2013). Currently, 33 states have implemented mature driver laws, which require vision testing and more frequent license renewals for drivers above a certain age (Governor’s Highway Safety Association 2013). Research has indicated that drivers with visual and cognitive impairments associated with aging are at greater risk for motor vehicle crashes; therefore, mandatory in-person license renewal and vision screenings may reduce the number of licensed older drivers with vision and cognition problems (Anstey et al. 2005; Rubin et al. 2007). However, previous studies have had mixed findings in regards to whether or not state licensure laws for the elderly have impacted fatality rates for older drivers (Grabowski et al. 2004; Langford et al. 2004, 2008; Levy et al. 1995; Morrisey and Grabowski 2005).
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
670 To date, there have been no previous studies that simultaneously investigated the effect of mature driver policies and state characteristics on MVC fatality rates among the elderly. It is unknown whether the large gap in fatality rates is explained by variation in state sociodemographics, health status, or health care systems in addition to licensing policies. Baker (1975) introduced a framework for determinants of injury that attempts to explain how individual characteristics such as age result in differential injury death rates for population subgroups. She proposed that age, socioeconomic, cultural, physical, and other individual characteristics modify death rates for groups of people by altering 4 categories of injury determinants including (1) the quantity and quality of exposure to environments conducive to injury; (2) the ability to perceive and respond to hazards; (3) injury thresholds (i.e., ability to withstand mechanical insults to the body); and (4) probability of recovery (e.g., access to life-saving emergency care). It follows that health-related factors such as states’ chronic disease burden, quality of health care, and development of emergency response systems could potentially account for differences in MVC fatality rates by modifying the latter 2 categories of determinants, whereas driving policies typically target the first 2 categories. Additionally, contextual factors such as the economic costs of driving, population density, and climate have been established in the literature as contributors to motor vehicle accidents (Chi et al. 2012; Theofilatos and Yannis 2014; Zwerling et al. 2005). These factors are likely associated with driving exposure, crash risk, and ability to respond to hazards, which can also be assigned to Baker’s (1975) first 2 categories of injury determinants. Examining additional state characteristics that influence MVC fatality rates through mechanisms that are both similar to and different from licensing policies will help parse out the effects of each factor so that future research on the design of policy interventions is more informed. The objective of this cross-sectional ecological study is to examine the relationship between state mature driver laws and MVC fatality rates among the elderly. It also aims to investigate the effect of mature driver laws relative to other state-level factors that potentially influence MVC fatality rates among the elderly such as demographic, health system, population health, travel, and climate characteristics.
Bell et al. number of driver’s licenses per 1,000 residents came from Our Nation’s Highways (Federal Highway Administration 2009). Data on the number of person-miles driven for elderly drivers was obtained from the 2009 National Household Travel Survey (U.S. Department of Transportation 2009). Climate data were provided by the National Oceanic and Atmospheric Administration’s National Climatic Data, which is based on state averages from 1971 to 2000 (National Climatic Data Center 2013). Population health status and provider data were obtained from Kaiser Family Foundation State Health Facts (Kaiser Family Foundation 2010). State trauma system data were obtained from the American Trauma Society (2009).
Study Variables State elderly MVC fatality rate was the dependent variable used in this study. For the purposes of this article, we define elderly as individuals over the age of 65. The death rate from MVCs was age-adjusted to account for states that have a greater proportion of older senior citizens. Stable rate estimates for Alaska, Hawaii, and the District of Columbia could not be produced and these states were excluded from the final analysis. The independent variable of interest in the study was the presence of state laws that imposed additional requirements on older drivers in order to obtain driver’s licenses. Laws were categorized as shortened renewal cycle, in-person renewal requirement, and vision test requirement. We also incorporated state-level control variables into our model. Because many population demographic and health variables are collinear, we used principal component analysis to reduce dimensionality of the data. We selected a variety of state-level covariates that accounted for amount of driving done by residents over 65, cost of driving, population health, access to care, quality of care, and environmental factors. Principal component analysis was used to reduce dimensionality of the data and overcome multicollinearity of state predictor variables. We identified 11 distinct components from 35 potential state covariates. Covariates with eigenvalues over 1 were included in the regression analysis. The extracted covariates accounted for 90.85% of variance.
Regression Analysis
Methods Data Source State motor vehicle fatality data were obtained from the Centers for Disease Control and Prevention’s Web-based Injury Statistics Query and Reporting System database, which compiles vital statistics data (Centers for Disease Control and Prevention 2013). Because death due to motor vehicle crash is a rare outcome among the elderly, we aggregated MVC fatality data from 2006 to 2010 to ensure stability of rate estimates. Information on state licensure laws was obtained from the Governor’s Highway Safety Association (2013). The American Community Survey (U.S. Census 2011) provided state demographic data used in the analysis (U.S. Census 2011). The
We used a general linear model to determine the impact of elderly licensure laws on MVC fatality rates in the over 65 population. Models assessing the effect of the presence of any restriction on older drivers as well as individual components of laws were generated. Model 1 examined the effect of all mature driver laws jointly on fatality rates. Models 2, 3, and 4 examined the effect of each law component separately (shortened renewal cycle, in-person renewal requirement, and vision test requirement, respectively). Each of the final models included the following covariates: average annual person-miles driven (among persons 65 and older), population density, average temperature, average precipitation, percentage of the population over 65 with college degrees, percentage of population overweight or obese, percentage of Medicare
Mature Driver Laws and Fatality Rates
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
beneficiaries with prescription drug coverage, percentage of population with access to a level I or II trauma center within 60 minutes, emergency medicine physicians per 10,000 members of the population, average gas price per gallon, and driver license fee amount. We also controlled for the number of years between required license renewals for states. We used principal component analysis to determine model covariates rather than automated variable selection procedures such as forward, backward, or stepwise regressions in order to ensure that the most relevant variables representing each category of Baker’s (1975) injury determinants were included. IBM SPSS Statistics 22 (IBM, Armonk, NY) was used for all analyses.
Results We found that annual MVC fatality rates among people over the age of 65 varied greatly across states. Fatality rates ranged from approximately 1.5 deaths per 100,000 people over the age of 65 in Massachusetts to 20.7 per 100,000 people over 65 in Mississippi. The Northeastern states tended to be in the quintile with the lowest fatality rates, whereas the Midwestern states typically fell into quintiles with the highest rates (Figure 1). Thirty-two out of 50 states had at least one restriction on elderly drivers. Of these 32, 19 states shortened renewal cycles for older drivers, 14 required in-person renewal, and 11 required a vision test. The age at which these laws went into effect varied considerably between states. Shortened renewal cycle laws went into effect for drivers between the ages of 60 to 85, with 65 and 75 being the most common age requirements (n = 4 for each). The age range for in-person renewal requirements ranged from 64 to 80, with 70 being the most common age at which the law went into effect (n = 5). Lastly, vision test requirements tended to go into effect at even younger ages than the 2 other requirements and ranged from 40 to 80 years old. Mature driver laws were most common in Western states. Six of the 10 states with the lowest MVC fatality rates among the elderly had no restriction on elderly drivers, whereas 7 of the 10 states with the highest MVC fatality rates among the elderly did have mature driver laws in place (Figure 2). State characteristics also varied across quintiles of MVC fatality rates among the elderly. In general, states in the highest quintile had the lowest population density, lowest average temperature, highest prescription drug coverage rates, poorest access to level I and II trauma centers, lowest driver’s license fees, and longest license renewal cycles. States in the highest fatality rate quintile also had the greatest number of states with mature driver laws in place (Table 1). Regression results found that average temperature, gas price, and ratio of emergency medicine physicians to population size were significant negative predictors of MVC fatality rates. In 3 of the 4 models we found that higher annual temperatures were associated with a reduction in fatality rates (P = .0499). In model 1, gas price was a significant negative predictor, with higher gas prices being associated with a decrease in fatality rates (P = .022). A greater number of emergency
671 medicine physicians per 10,000 members of the population was also associated with a significant reduction in fatality rates in all 4 models (P < .001; Table 2). We found several positive predictors of MVC fatality rates among the elderly. Positive predictors included the percentage of population overweight or obese and the percentage with college degrees over the age of 65. The percentage of the population overweight or obese was strongly associated with a positive increase in MVC fatality rates among the elderly in all 4 models (P = .014, .020, .049, .033, respectively). We also found that a higher percentage of older adults with college degrees predicted significantly higher fatality rates in all 4 models (P = .006, .004, .024, .012). Lastly, model 1 indicated that having any restriction on elderly drivers was associated with a higher MVC fatality rate (P = .036). Model 2 also found that shortened renewal cycles predicted higher MVC fatality rates (P = .042). Neither inperson renewal nor vision test requirements had significant negative or positive effects on MVC fatality rates among the elderly (P = .862, .636).
Discussion Overall, our study found that mature driver laws were not associated with lower state MVC fatality rates among the elderly. Having any additional requirements for older drivers and a shortened renewal cycle were associated with higher fatality rates; however, in-person renewal and vision test requirements were not significant predictors of fatality rates. Our results differ from previous studies that reported a beneficial effect of elderly driver laws on MVC fatality rates (Nelson et al. 1992; Shipp 1998). Because our study is ecological, we can only determine that mature driver laws are associated with higher MVC fatality rates. It is possible that laws are endogenous to state characteristics and that higher MVC fatality rates among the elderly prompt state legislators to pass mature driver laws. These results could also be due to policies having unintended effects. For example, older drivers may feel that such policies are discriminatory and potentially unsafe drivers may be more motivated to renew their licenses as a result. It is also possible that these policies remove safe older drivers from the road, who then rely on drivers in age groups at higher crash risk for transportation. Future studies should assess individuals who did not renew driver’s licenses due to mature driver laws and determine whether or not they were fit to drive. This result is in agreement with other studies that have reported that mature driver laws did not significantly reduce MVC fatalities in the elderly (Grabowski et al. 2004; Morrisey and Grabowski 2005; Rock 1998). There are several potential explanations for these findings. One reason may be that we examined MVC fatality rates for the entire population over 65 and there may be a heterogeneous response to the laws in younger old and older old drivers. Studies that have reported significant effects of mature driving laws found that the laws were only significantly associated with reduced fatality rates in drivers over 80 years old (Grabowski et al. 2004). Considering that the rate of MVC fatalities in people between the ages
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
672
Bell et al.
Fig. 1. MVC fatality rates among people age 65 and older. State MVC fatality rates divided into quintiles are depicted. Rates were estimated based on data from 2006 to 2010 and were age-adjusted to account for state populations with a greater proportion of older elderly. Stable rate estimates of Alaska, Hawaii, and the District of Columbia could not be produced.
of 60 and 70 is lower than that of younger age groups, states with mature driver laws that impose restrictions at younger ages may not be targeting the most at-risk age group (Federal Highway Administration, Department of Transportation 2014). The challenge of implementing additional requirements and restrictions on older drivers may also be reflected in our results. For example, studies examining the role of visual
impairment in motor vehicle crashes have reported that glare sensitivity and visual field loss are significant predictors of crash involvement but static visual acuity is not; however, vision tests for drivers’ licensure are based primarily on static visual acuity (Keeffe et al. 2002; Rubin et al. 2007). Several studies have also reported that crash involvement can be predicted with various performance-based cognitive tests, yet implementation of these tests in real-world settings has been problematic
Fig. 2. Mature driver laws across states. The presence of a law is indicated by grey shading. Data were obtained from the Governors Highway Association in December 2012. States that passed laws after 2010 were not considered to have mature driver laws in our analysis, which examined data from 2006 to 2010.
Mature Driver Laws and Fatality Rates
673
Downloaded by [University of Nebraska, Lincoln] at 17:48 05 November 2015
Table 1. Variation in state characteristics by quintiles (Q) of MVC fatality rates among the elderly
MVC fatality rate 65+ Average annual person-miles driven, persons 65 and up Population density Average temperature (◦ F) Precipitation (inches) % College degree 65 and up Per capita income (US$) % Overweight or obese % Medicare beneficiaries with known prescription drug coverage % Population with access to level I or II trauma center within 60 minutes Emergency medicine physicians per 10,000 population Average gas price (US$/gallon) Driver’s license fees (US$) License renewal cycle (years) Any restriction on elderly drivers Shortened renewal cycle for elderly In-person renewal requirement Vision test requirement
Q1: Lowest
Q2: Low
Q3: Medium
Q4: High
Q5: Highest
2.5 (0.7) 265.8 (243.5)
4.5 (0.5) 354.4 (483.1)
6.7 (0.6) 444.0 (360.7)
9.4 (1.5) 367.8 (348.8)
15.6 (2.6) 300.5 (168.6)
510.3 (443.5) 53.9 (6.5) 42.8 (12.9) 21.8 (4.3) 28,326.56 (5,281.42) 63.2 (2.8) 40.5 (7.1)
204.8 (145.5) 52.1 (6.1) 37.4 (11.6) 22.8 (5.0) 27,499.70 (3,405.42) 62.9 (3.7) 40.2 (8.4)
80.8 (62.3) 53.8 (9.0) 34.9 (17.2) 19.5 (4.5) 23,993.30 (1,820.46) 63.8 (2.7) 43.4 (10.2)
189.5 (168.0) 51.5 (9.4) 43.0 (8.8) 23.3 (4.7) 27,614.00 (3,303.16) 63.3 (2.1) 41.2 (6.3)
30.8 (19.9) 49.0 (7.0) 28.5 (14.6) 20.6 (3.3) 23,412.56 (1,887.85) 64.2 (2.4) 49.3 (8.3)
84.8 (24.9)
89.2 (8.2)
65.7 (25.3)
85.0 (8.6)
58.56 (20.0)
1.7 (0.5)
1.5 (0.4)
1.1 (0.3)
1.3 (0.1)
1.1 (0.4)
2.76 (0.05) 9.78 (1.56) 5.1 (2.0) 5 (55.6%) 2 (22.2%) 4 (44.4%) 1 (11.1%)
2.80 (0.05) 9.50 (1.18) 6.3 (2.5) 6 (60.0%) 3 (30.0%) 2 (20.0%) 3 (30.0%)
2.78 (0.07) 8.50 (0.53) 5.7 (2.0) 6 (60.0%) 4 (40.0%) 3 (30.0%) 3 (30.0%)
2.77 (0.08) 9.30 (1.25) 5.8 (1.6) 5 (50.0%) 3 (30.0%) 2 (20.0%) 2 (20.0%)
2.76 (0.07) 8.44 (0.53) 6.6 (1.4) 7 (77.8%) 6 (66.7%) 1 (11.1%) 1 (11.1%)
Table 2. Coefficient estimates of mature driver laws and state characteristics Dependent variable: Age-adjusted rate of death due to MVC in population 65 and older
Intercept Average annual person-miles driven, persons 65 and up Population density Average temperature (◦ F) Precipitation (inches) % College degree 65 and up Per capita income % Overweight or obese % Medicare beneficiaries with known prescription drug coverage % Population with access to level I or II trauma center within 60 minutes Emergency medicine physicians per 10,000 population Average gas price (US$/gallon) Driver’s license fees (US$) License renewal cycle (years) Any restriction on elderly drivers Shortened renewal cycle for elderly
(1)
(2)
(3)
(4)
−7.956 (24.633) 0.000 (0.001) 0.000 (0.003) −0.307 (0.123)∗ −0.014 (0.054) 0.626 (0.219)∗∗ 0.000 (0.001) 91.403 (35.126)∗∗ 0.000 (0.002) 0.012 (0.017) −8.186 (1.638)∗∗∗ −0.019 (0.008)∗ 1.559 (1.444) −0.261 (0.328) 2.491 (1.154)∗
−11.647 (24.571) 0.000 (0.001) −0.001 (0.003) −0.294 (0.123)∗ −0.003 (0.055) 0.667 (0.221)∗∗ 0.000 (0.001) 85.990 (35.256)∗ 0.000 (0.002) 0.014 (0.017) −7.761 (1.624)∗∗∗ −0.014 (0.007) 1.472 (1.447) −0.313 (0.343)
−14.616 (26.259) 0.001 (0.001) 0.000 (0.004) −0.274 (0.142) −0.023 (0.058) 0.607 (0.245)∗ 0.000 (0.001) 87.933 (39.513)∗ 0.000 (0.002) 0.014 (0.018) −7.596 (1.754)∗∗∗ −0.014 (0.008) 1.346 (1.547) 0.069 (0.313)
−16.120 (26.342) 0.001 (0.001) 0.000 (0.004) −0.271 (0.131)∗ −0.021 (0.058) 0.626 (0.236)∗∗ 0.000 (0.001) 87.688 (37.587)∗ 0.000 (0.002) 0.016 (0.018) −7.530 (1.739)∗∗∗ −0.014 (0.008) 1.259 (1.555) 0.112 (0.328)
2.494 (1.191)∗
In-person renewal requirement Vision test requirement
Unstandardized beta coefficients for independent variables are reported. Standard error values are in parentheses. ∗ Significant at 0.05 level. ∗∗ Significant at 0.01 level. ∗∗∗ Significant at
