The association between a statewide stay-at-home order and motor vehicle injury rates among population sub-groups in West Virginia

Abstract

Objective:

Little is known about the relationship between Stay-At-Home orders issued by state governments due to the COVID-19 pandemic and their impacts on motor vehicle-related injuries. The purpose of this study was to determine whether the presence of a Stay-At-Home order was associated with lower rates of motor vehicle-related injuries requiring emergency medical treatment among population sub-groups in West Virginia (i.e., males, females, 0-17, 18-25, 26-45, 46-65, ≥66 years old).

Methods:

A Stay-At-Home order was in effect in West Virginia from March 23-May 4, 2020. Counts of individuals who incurred motor-vehicle-related injuries that required emergency medical treatment were obtained from the Centers for Disease Control and Prevention’s National Syndromic Surveillance Program from January 1 thru September 6 of 2019 and 2020. Counts were obtained by week-year and by population sub-group in West Virginia. The presence of the Stay-At-Home order was binary coded by week. Negative binomial regression was used to assess the relationship between the presence of a Stay-At-Home and injury rates. 2019 population sub-group estimates were obtained from the United States Census Bureau and used as offsets in the models. Models were also adjusted for year and vehicle miles traveled by week-year.

Results:

There were 23,418 motor-vehicle related injuries during the study period. The presence of the Stay-At-Home order was associated with 44% less injuries overall [Incident Rate Ratio (IRR)=0.56, 95% CI 0.48, 0.64]. Females experienced fewer injuries than males (IRR = 0.49 vs 0.63, respectively) and the number of injuries decreased with age (p-value 0.031) when comparing time periods when the Stay-At-Home was in effect compared to times when it was not.

Conclusions:

West Virginia’s Stay-At-Home order was associated with lower motor-vehicle injury rates requiring medical treatment across all population sub-groups. Most population sub-groups likely altered their travel behaviors which resulted in lower motor-vehicle injury rates. These findings may inform future policies that impose emergency travel restrictions in populations.

Introduction

Stay-at-Home (SAH) orders (also known as Safer-at-Home, Shelter-in-Place, or lockdowns) are public safety policies that are intended to limit population movement by restricting unnecessary travel and person-to-person interaction (Jacobsen and Jacobsen 2020). From March 1 to May 31, 2020, 42 states or territories in the United States (US) implemented these measures to reduce the spread of the SARS-Cov2 virus throughout their populations (Moreland et al. 2020). During the beginning of the SARS-Cov2 pandemic, the specifics of these enacted policies varied between states and sometimes even within local, neighboring jurisdictions. Some SAH orders lasted a few days while others lasted for several weeks. Some SAH orders differed in how ‘unnecessary travel’ was defined, who was permitted to travel, the number of individuals allowed to gather, or what types of business or organizations could be open to the public during the restrictive period (Moreland et al. 2020). Despite these differences, several studies determined that SAH orders were generally effective at not only slowing the spread of the SARS-Cov2 virus (Sen et al. 2020), but also in deterring population movement (Hsiehchen et al. 2020; Jacobsen and Jacobsen 2020; Moreland et al. 2020).

Because the majority of the US population relies heavily on motor vehicles for mobility, it was reasoned that SAH orders would naturally impact motor vehicle-related travel behaviors. For example, a study conducted among drivers 16-18 years of age in Birmingham, Alabama found that 52% of participants changed their travel behaviors due to the SARS-Cov2 pandemic; among those who reported a change, 89% reduced the miles or days they drove after the SAH order was enacted (Stavrinos et al. 2020). Another study conducted in California, which used highway patrol data, revealed that traffic volume decreased 55% on select highways after the SAH order was enacted; this study also found that drivers’ speed increased possibly due to reduced traffic congestion (Shilling 2020). Another study conducted in Florida, which used traffic counting stations to estimate traffic volume, revealed that there was a 48% decrease in traffic counts comparing March 2020 data to March 2019 data (Parr et al. 2020). Studies conducted in one Michigan county, and another conducted in Birmingham, Alabama, which both used traffic count data, reported decreased traffic volumes when SAH orders were enacted (Stavrinos et al. 2020; Stoker et al. 2020). A statewide study conducted in Connecticut, which used a mobile phone tracking database to estimate driving distances, reported that the mean daily vehicle miles traveled among drivers decreased 43% in the post-SAH order vs. pre-SAH order periods in that state (Doucette et al. 2021).

Because travel behaviors were likely altered, SAH orders may have also impacted motor vehicle collisions and resulting injuries as well. Several studies reported that motor vehicle collisions in general decreased during the time period when SAH orders were enacted (Stoker et al. 2020; Sutherland et al. 2020; Doucette et al. 2021). A study that investigated crashes in one Michigan county found that collisions decreased 46% in March 2020 compared to March 2019 (Stoker et al. 2020). Another study conducted in Connecticut found that after accounting for reduced traffic volumes, single vehicle crash rates increased while the average number of daily crashes decreased 55% after the SAH order was enacted compared to before it was enacted (Doucette et al. 2021). In terms of injuries, one California study found that there was a 50% reduction in injurious and fatal collisions and a 50% reduction in crashes involving cyclists and pedestrians during the SAH order (Shilling 2020). One New Hampshire trauma center reported that motor vehicle-related trauma encounters were 81% lower in the SAH order time period in 2020 compared to the same time frame the 3 years prior (Kamine et al. 2020). One Michigan trauma center reported that there was a 67% decrease in vehicle-related trauma encounters comparing March-April 2020 compared to the same time period the three years prior (Stoker et al. 2020). A statewide study using traffic records from Missouri found that during societal lockdown there were no significant changes in serious or fatal crashes, while there was a significant reduction in traffic collisions which resulted in minor or no injuries after the SAH order was enacted in that state (Qureshi et al. 2020). In Connecticut, there was a 56% reduction in non-injury crashes and a 52% decrease in injury crashes after the SAH order was enacted (Doucette et al. 2021). Another study which utilized patient data from trauma centers and medical examiners located in Charlotte, North Carolina, Jacksonville, Florida, Miami, Florida, Baltimore, Maryland, and Worcester, Massachusetts, found that alcohol, marijuana, and opioid use increased among motor vehicle-related trauma patients; 65% of drivers tested positive for at least one drug after March 17, 2020, compared to 51% of drivers that tested positive before March 16, 2020 (Thomas et al. 2020).

Studies conducted in other countries experienced similar trends. An emergency department in Anhui province, China had a 40% decrease in patients during lockdown, which was mainly attributed to less motor vehicle-related injuries. During lockdown periods in Australia, travel by automobile significantly decreased along with collisions, while Spain, Greece, South Africa, and England also experienced decreases in traffic collisions during lockdown. In Japan, speed-related fatal motor vehicle collisions actually increased during lockdown (additional references can be found in Online Appendix).

While the current research suggests that the enactment of SAH orders was generally associated with decreased motor vehicle-related travel and subsequent injuries, there are still several gaps in the extant literature. For example, most of the existing research was conducted in metropolitan areas or within more populated states. Secondly, studies which investigated the association between motor vehicle-related injuries and SAH orders typically involved data from only one trauma center. Additionally, it is unknown whether SAH orders are associated with reduced motor vehicle-related injuries among different population sub-groups. It is entirely possible that SAH orders may not be followed equally nor benefit all population sub-groups (i.e., different age groups or sex). Thus, the purpose of this study was to investigate whether the SAH order enacted in West Virginia, a rural Appalachian state, was associated with reduced motor-vehicle related injuries requiring medical treatment across different population sub-groups using data from multiple emergency departments located throughout the state. It was hypothesized that motor vehicle-related injuries would be reduced after the SAH order was enacted, but differences in population-subgroups would be observed. These findings could inform future policies that impose emergency travel restrictions on populations.

Methods

Data sources

Three data sources were used for this analysis. The primary data source was the Centers for Disease Control and Prevention’s (CDC) National Syndromic Surveillance Program. A network of nearly 6,000 healthcare facilities from 49 states and the District of Columbia contribute patient data incurred from emergency department visits to CDC’s data collection platform daily. In West Virginia specifically, 86% of the state’s emergency departments actively contribute data to this system. This system allows public health officials to track and detect potential disease outbreaks or health events almost in real time. The data system contains information such as the patient’s chief complaint, primary and secondary discharge diagnoses, along with some demographic data. The health records can be searched by diagnostic codes and/or by free text. Additional information about CDC’s system is described in detail elsewhere (Centers for Disease Control and Prevention 2021). The second data source used for this analysis was 2019 West Virginia population data, which was obtained from the United States Census Bureau (2020). Population data from 2019 was the most recent year available. The data were collected for the total population and by sex and age group (i.e., 0–17, 18–25, 26–45, 46–65, ≥66 years of age). The third data source was the estimated vehicle miles traveled by month-year in West Virginia during the time periods of interest, which were obtained from the Federal Highway Administration (2021).

Study population

The study population included any individual who received medical treatment for a motor-vehicle related injury in a West Virginia emergency department either from January 1, 2019 thru September 6, 2019 or from January 1, 2020 thru September 6, 2020. Patients were identified using CDC’s standard case definition for motor vehicle-related injuries which incorporates both diagnostic codes and free text and is applied to patients’ chief complaints and discharge diagnoses. This search string has been validated by CDC to minimize the number of false-positives and false-negatives. Because emergency departments are not mandated to participate in the National Syndromic Surveillance Program, their participation can fluctuate. This analysis excluded individuals who were treated in facilities that did not participate in both study periods. The study population also included all ages as anyone could theoretically be injured by a motor vehicle as a driver, passenger, pedestrian, pedal cyclist, etc.

Variables

The primary dependent variable in this analysis was the number of motor vehicle-related injury encounters requiring emergency medical treatment. These injury encounters were aggregated by week-year overall and by population sub-group (i.e., weeks 1–37 of 2019 and 2020). The primary independent variable was whether or not a SAH order was in effect during the week-year, which was binary coded (i.e., yes, or no). West Virginia’s SAH order was enacted from March 23 to May 4, 2020, which encompassed weeks 13–18 of 2020. Other covariates included year of the injury, patients’ sex (male, female) and patients’ age at time of the injury which was categorized as 0–17, 18–25, 26–45, 46–65, ≥66 years old. The number of vehicle miles traveled in West Virginia by month-year was linearly interpolated to week-year (these data are graphed and shown in Figure A1 in the Online Appendix).

Statistical analyses

In addition to descriptive statistics, negative binomial regression was utilized to determine the incidence rate ratio regarding the association between the presence of the SAH order and motor-vehicle related injuries per capita. This model was chosen as these were count data and there were concerns regarding overdispersion (Hilbe 2011). This model was run for the overall population and then stratified by sex and age group. The total population for each sub-group were used as the offset in the respective models. All models were also adjusted for year and vehicle miles traveled by week-year. Collinearity was assessed between year and vehicle miles traveled using variance inflation factor and tolerance. Additionally, in order to determine whether sub-group differences existed, effect measure modification was assessed using an interaction term between the presence of a SAH order and the population sub-group. Thus, an interaction p-value for sex was estimated by running an additional negative binomial regression model with sex and an interaction term between sex and the SAH order adjusted for year and vehicle miles traveled. The interaction p-value for age group was estimated by running an additional negative binomial regression model with age group and an interaction term between age group and the SAH order adjusted for year and vehicle miles traveled. All data management and statistical analyses were conducted using SAS version 9.4 using two tailed hypothesis tests with α = 0.05. Institutional Review Board approval was obtained from West Virginia University.

Additional references

Additional references can be found in the bibliographic section of the Online Appendix.

Results

Between Jan 1 and Sept 6 of both 2019 and 2020, 23,418 individuals were injured in motor vehicle-related encounters that required immediate medical attention in West Virginia (Table 1). Overall, there were slightly more females treated compared to males (50.4% vs. 49.6%, respectively), and more 26–45-year-olds (33%) injured compared to other age groups. Individuals ≥66 years of age were the least prevalent age group to be injured (8.6%) in a motor vehicle compared to all other age groups.

Table 1.

Demographic characteristics of individuals injured in motor vehicle related incidents that received medical treatment between Jan 1-Sept 6 in 2019 and 2020 in West Virginia (N = 23,418).

	2019		2020		Total
Characteristic	N	%	N	%	N	%
Sex
Males	5,978	49.4	5,824	51.4	11,614	49.6
Females	6,115	50.6	5,499	48.6	11,802	50.4
Missing	2		0		2
Age Groups (in years)
0–17	1,938	16.0	1,746	15.4	3,684	15.7
18–25	2,504	20.7	2,406	21.3	4,910	21.0
26–45	4,004	33.1	3,782	33.4	7,786	33.3
46–65	2,586	21.4	2,441	21.6	5,027	21.5
>66	1,063	8.8	948	8.4	2,011	8.6

After adjusting for year and vehicle miles traveled, injuries were 44% lower during the SAH order compared to when the ban was not in effect (Table 2). There were statistically significant differences among all demographic sub-groups when comparing injury rates before and after the SAH order. While all demographic sub-groups experienced fewer motor vehicle-related injuries during the SAH order, women experienced 51% fewer injuries during the ban, while males experienced 37% fewer injuries during the ban compared to when it was not in effect after adjusting for year and vehicle miles traveled. As age increased, there was a larger association between the presence of the SAH order and less injuries (p = 0.031). Compared to when the ban was not in effect, individuals 0–17 years were 34% less likely to be treated for injuries, while individuals ≥66 years were 59% less likely to be injured when the ban was in effect compared to when it was not after adjusting for year and vehicle miles traveled. Figures comparing injury counts in 2019 and 2020 in relation to the enactment of the SAH order for the overall population are shown in Figure 1; additional figures comparing injury counts in 2019 and 2020 in relation to the enactment of the SAH order by sex and by age group are shown in the Online Appendix (Figures A2 and A3, respectively).

Table 2.

The association between the presence of a Stay-at-Home order and incidence rate ratios of motor-vehicle related injuries requiring medical treatment stratified by demographic sub-groups in West Virginia.^a

	Rate Ratio	95% CI	Interaction p-value
Overall	0.56	0.48, 0.64
Sex
Males	0.63	0.53, 0.74	0.038
Females	0.49	0.41, 0.57
Age Groups (in years)			0.031
0–17	0.66	0.51, 0.85
18–25	0.59	0.48, 0.70
26–45	0.58	0.50, 0.68
46–65	0.48	0.40, 0.59
>66	0.41	0.30, 0.56

Abbreviations: CI = Confidence interval.

^aNegative binomial regression was used to assess incidence rates of injuries when a Stay-at-Home order was/was not in effect. The dependent variable was counts of injuries by week-year, while the independent variables were the presence of a Stay-at-Home order, incident year, and vehicle miles traveled by week-year. The offset was per capita counts. The interaction p-value for sex was estimated by running an additional negative binomial regression model with sex and an interaction term between sex and the Stay-At-Home order adjusted for incident year and vehicle miles traveled by week-year. The interaction p-value for age group was estimated by running an additional negative binomial regression model with age group and an interaction term between age group and the Stay-At-Home order adjusted for incident year and vehicle miles traveled by week-year.

Figure 1.

The number of motor vehicle-related injuries requiring immediate medical treatment for the overall population by week of year for 2019 and 2020 in West Virginia. The black reference line shows when the Stay-At-Home order was in effect during 2020.

Discussion

This study determined that the SAH order issued in West Virginia during the beginning of the SARS-Cov2 pandemic was associated with fewer motor vehicle injuries requiring immediate medical treatment overall and across all demographic sub-groups. While this may suggest that different facets of the population likely altered their travel behaviors, there were statistically significant differences between these groups. In terms of injuries, this may indicate that some population sub-groups, especially women and older individuals, benefited more from the SAH order compared to other groups, which could have public health implications.

The overall findings of this study were similar to other studies conducted in more populated states. While WV saw a 44% reduction in motor-vehicle related injuries overall during the SAH order, four trauma centers and hospitals in Sacramento, California similarly saw a 38% decrease in motor vehicle-related trauma injuries between March-April 6, 2020 (Shilling 2020). Additionally, individual trauma centers in New Hampshire and Michigan saw 81% and 67% reductions, respectively, in motor vehicle-related trauma comparing March-April 2020 to the three years prior (Kamine et al. 2020; Stoker et al. 2020). While the reductions in motor vehicle-related injuries during the SAH order in West Virginia were not as great as those seen in Michigan and New Hampshire, this may be explainable. West Virginia is a rural, mountainous state and many communities are geographically isolated. Many individuals must drive great distances to reach essential services such as healthcare, pharmacies, and grocery stores. The SAH order in West Virginia did not restrict people from these services. Thus, some individuals may have continued to drive even with the SAH order in place to reach essential services. Conversely, it could be that some West Virginia drivers simply did not obey the SAH order, continued to drive, and were more at risk of injury compared individuals in states such as New Hampshire or Michigan. One study suggested that individuals from states with greater proportions of Republicans tended to travel more and not obey SAH orders as much as states with lower proportions of Republicans (Hsiehchen et al. 2020). In general, West Virginia is a conservative, Republican state. Additionally, this study looked at emergency departments across the state and not just individual trauma centers; trauma teams did not have to be activated in all encounters.

While no previous studies have investigated the relationship between SAH orders and motor vehicle-related injuries across different population sub-groups, the findings confirming that women and older age groups tended to experience fewer motor-vehicle related injuries during the SAH order may not be surprising. With the exception of certain distracted driving laws, previous studies have found that women tend for follow traffic-related laws more than males and that older age groups tend to be more compliant with traffic laws than younger age groups (Yagil 1998a, 1998b; Rudisill et al. 2019). Prior to the SARS-Cov2 pandemic, studies investigating compliance with emergency travel restrictions were virtually non-existent in the peer-reviewed literature. Only one study, which was conducted in Beijing, China, found that women tended to be more compliant with travel restrictions than males (Wang et al. 2014). Additionally, it is also well-documented in the traffic safety literature that males tend to drive more than females, and younger age groups (i.e., 25–55 years of age) tend to drive more than the oldest of age groups. Thus, women and older drivers may have been less at risk of injurious collision simply because they were driving less in general, and the ban magnified this effect.

Collectively, these findings are important from a public health perspective. This study found that SAH orders were associated with fewer motor vehicle-related injuries even in rural states such as West Virginia and that certain facets of the demographic experienced fewer injuries than others. Thus, if a SAH needed to be enacted again, the whole population could likely be targeted from a public health messaging/interventional standpoint.

Limitations

While the strength of this study is that it utilized data from 85% of West Virginia’s emergency departments and is thus representative of the state population, it is not without limitation. First, it is entirely possible that some emergency encounters could have been missed due to potential coding errors to how the injury was coded (i.e., the encounter was not coded as a motor vehicle injury when it should have been or vice versa). However, this study utilized CDC’s case definition for motor vehicle-related injuries which are evaluated for sensitivity and specificity. Second, this study involved emergency encounters. Because the data were de-identified, it is entirely possible that a person could be counted more than once if they were involved in more than one collision over the study period. Third, some individuals who were treated may not have been West Virginia residents which could have influenced counts. It is also possible that some individuals were hurt in a motor vehicle collision, but did not seek medical care. Fourth, the SAH order in West Virginia started and ended on a Monday and the weeks were coded Sunday thru Saturday. Fifth, vehicle miles traveled were only available for the entire population; if vehicle miles traveled were available for each of the demographic groups, the findings may have differed. As mentioned previously, it is well-documented in the traffic safety literature that females and older adults tend to drive less than males and younger age groups. Additionally, these findings may not be generalizable to other states, and it is entirely possible that other confounding factors, such as those related to the economy, weather, etc. could have also influenced travel in the population. While all models were adjusted for year and vehicle miles traveled, which should have helped control for these other confounding factors, it does not eliminate the potential for residual confounding. Last, as seen in this study and others, motor vehicle-related injuries were decreasing before the ban was enacted (Sutherland et al. 2020). However, this is likely attributed to the public health messages announcing that the bans were becoming effective; as seen in this study and others, injury counts typically resumed after SAH orders were lifted (Qureshi et al. 2020).

Nevertheless, the SAH order in West Virginia, which was enacted during the beginning of the SARS-Cov2 pandemic, was associated with significantly lower motor vehicle-related injuries requiring immediate medical treatment across various population sub-groups. However, some population sub-groups, such as females or older individuals, may have experienced greater reductions in injuries during the ban compared to other groups. These findings may inform future policies that impose emergency travel restrictions particularly in rural populations.