Abstract
This study compares traffic volume and motor vehicle crash injuries before, during, and after state-of-emergency and stay-at-home orders in the state of Ohio from January to July 2020 vs the same period in 2019.
To minimize transmission of coronavirus disease 2019 (COVID-19), most US states in spring 2020 passed policies promoting social distancing through stay-at-home orders prohibiting nonessential travel.1 While vehicle miles traveled in the US decreased by 41% in April 2020 compared with 2019,2 the effect of this mobility decrease on motor vehicle crashes (MVCs) is poorly understood. We estimated associations between COVID-19–related social distancing policies, traffic volume, and MVC-related outcomes in Ohio.
Methods
Our observational study compared MVCs and traffic volume data from two 7-month periods: January 1, 2020, through July 31, 2020, and January 1, 2019, through August 1, 2019 (accounting for the leap-year day in 2020). Motor vehicle crash data were obtained from the Ohio Department of Public Safety’s Electronic Crash Submission database.3 Traffic volume data were obtained from the Ohio Department of Transportation through permanent count stations positioned on interstate, state, and US routes.4
Three state-level policies demarked 4 study periods in 2020: period 1, January 1 through March 8; period 2, March 9 (state-of-emergency declaration) through March 22; period 3, March 23 (stay-at-home order) through May 11; and period 4, May 12 (retail reopening) through July 31. Mean daily counts were calculated and compared across periods for 3 types of crash-related outcomes: (1) number of people (motor vehicle drivers and passengers, pedestrians, motorcyclists, and bicyclists) involved in MVCs (MVC involvements), (2) number of people having any injuries in an MVC (MVC injuries), and (3) number of people having a severe or fatal injury in an MVC (MVC severe or fatal injuries), along with (4) traffic volume.
Daily interrupted time-series analyses with ordinary least-squares linear regression and Newey-West standard errors were used to estimate slope changes. All outcome variables were log transformed. Crash month, weekday or weekend occurrence, gasoline price, and unemployment rate were included in the analysis to control for seasonality and confounding. Statistical significance was defined as a 95% CI that excluded 0. As this study used publicly available, deidentified secondary data reported on an aggregated level, it did not undergo institutional review board review per institutional guidelines.
Results
From January 1 through July 31, 2020, MVCs were experienced by 284 128 individuals, with 27 809 having some level of injury and 3719 having severe injuries; there were 621 fatalities. These numbers were compared with MVCs during the 2019 study period, in which 382 098 individuals were involved in MVCs, 33 365 had some level of injury, 4243 had severe injuries, and there were 619 fatalities. When separated by period during 2020, all outcomes substantially declined during period 2 and reached their lowest levels directly following the stay-at-home order before gradually increasing through periods 3 and 4 (Figure).
Figure. Daily Counts of Motor Vehicle Crash–Related Outcomes and Traffic Volume.
The vertical dotted lines demark the study periods (P1-P4) in 2020. Period 1 was January 1 through March 8; period 2, March 9 (state-of-emergency declaration) through March 22; period 3, March 23 (stay-at-home order) through May 11; and period 4, May 12 (retail reopening) through July 31. The x-axis is in reference to 2020 dates, accounting for the leap-year day. MVC indicates motor vehicle crash.
Comparing slopes across periods, period 2 saw significantly larger daily changes than any other period of 2020 across all outcomes: for MVC involvements, −7.08% (95% CI, −8.31% to −5.82%); for MVC-related injuries, −5.08% (95% CI, −6.48% to −3.65%); for MVC-related severe or fatal injuries, −5.61% (95% CI, −8.19% to −2.95%); and for traffic volume, −4.07% (95% CI, −5.14% to −2.99%) (Table).
Table. Mean Daily Counts and Changes in Motor Vehicle Crash Involvements, Injuries, Severe or Fatal Injuries, and Traffic Volume.
| Outcome and perioda | Mean daily count (95% CI)b | Difference, % (95% CI)c | 2020 Daily % change (95% CI)d | ||
|---|---|---|---|---|---|
| 2019 | 2020 | Period in 2020 vs same period in 2019c | Period in 2020 vs previous period in 2020c | ||
| Crash involvement | |||||
| Period 1 | 1759 (1634-1884) | 1697 (1585-1810) | −3 (−13 to 6) | Not applicable | 0.11 (−0.60 to 0.82) |
| Period 2 | 1555 (1421-1689) | 1116 (907-1325) | −28 (−45 to −11) | −34 (−49 to −19) | −7.08 (−8.31 to −5.82) |
| Period 3 | 1767 (1670-1865) | 788 (736-839) | −55 (−62 to −49) | −29 (−50 to −9) | 0.97 (0.63 to 1.33) |
| Period 4 | 1881 (1817-1945) | 1404 (1359-1449) | −25 (−30 to −21) | 78 (69 to 87) | 0.13 (−0.17 to 0.42) |
| Injuries | |||||
| Period 1 | 138 (128-148) | 132 (125-140) | −4 (−13 to 5) | Not applicable | −0.23 (−0.89 to 0.44) |
| Period 2 | 127 (117-137) | 100 (84-116) | −21 (−37 to −6) | −24 (−39 to −10) | −5.08 (−6.48 to −3.65) |
| Period 3 | 155 (146-163) | 82 (76-88) | −47 (−54 to −40) | −18 (−36 to 0) | 0.49 (−0.32 to 1.30) |
| Period 4 | 178 (173-184) | 164 (158-170) | −8 (−12 to −3) | 100 (90 to 110) | 0.20 (−0.18 to 0.58) |
| Severe or fatal injuries | |||||
| Period 1 | 19 (17-20) | 16 (15-17) | −14 (−25 to −3) | Not applicable | −0.41 (−1.21 to 0.40) |
| Period 2 | 18 (16-21) | 16 (13-18) | −15 (−36 to 6) | −3 (−22 to 16) | −5.61 (−8.19 to −2.95) |
| Period 3 | 22 (20-24) | 15 (13-16) | −34 (−47 to −21) | −7 (−28 to 14) | −0.27 (−1.43 to 0.90) |
| Period 4 | 27 (26-29) | 28 (26-30) | 4 (−6 to 13) | 94 (75 to 113) | 0.16 (−0.48 to 0.80) |
| Traffic volumee | |||||
| Period 1 | 686 (653-720) | 758 (728-788) | 10 (4 to 17) | Not applicable | 0.27 (−0.05 to 0.59) |
| Period 2 | 789 (731-848) | 667 (572-763) | −15 (−30 to 0) | −12 (−26 to 2) | −4.07 (−5.14 to −2.99) |
| Period 3 | 821 (791-851) | 463 (438-488) | −44 (−48 to −39) | −31 (−47 to −15) | 0.50 (0.28 to 0.72) |
| Period 4 | 833 (810-857) | 673 (652-695) | −19 (−23 to −15) | 45 (38 to 52) | 0.27 (0.09 to 0.45) |
Between-year comparisons are adjusted to accommodate for the leap-year day in 2020. Period 1: January 1, 2020, through March 8, 2020, vs January 1, 2019, through March 9, 2019 (adjusting for the leap-year day in 2020). Period 2: March 9, 2020 (state-of-emergency declaration) through March 22, 2020, vs March 10, 2019, through March 23, 2019. Period 3: March 23, 2020 (stay-at-home order) through May 11, 2020, vs March 24, 2019, through May 12, 2019. Period 4: May 12, 2020 (retail reopening) through July 31, 2020, vs May 13, 2019, through August 1, 2019.
Mean number of outcomes per day were calculated from raw totals.
The 2-sample t test with unequal variance was used to calculate the difference between periods. Differences were scaled by the reference period’s point estimate to determine percentage difference.
Daily percent change was derived from interrupted time-series analysis using the slope of each period. Slope changes were estimated using Newey-West standard errors and daily interrupted time-series analyses with ordinary least-squares linear regression. All outcome variables were log transformed with a 7-day lag period. Models control for month, weekend vs weekday, weekly gasoline price, and monthly unemployment rate.
Indicates daily traffic volume divided by 10 000.
Relative to the same 2019 period, period 3 showed the largest difference: a −55% (95% CI, −62% to −49%) change in MVC involvements, a −47% (95% CI, −54% to −40%) change in injuries, a −34% (95% CI, −47% to −21%) change in severe or fatal injuries, and a −44% (95% CI, −48% to −39%) change in traffic volume. In period 4, mean daily counts of MVC-related injuries and severe or fatal injuries approached 2019 levels.
Discussion
The period beginning with Ohio’s state-of-emergency declaration was associated with the greatest daily percentage decrease in MVC involvements, injuries, and traffic volume compared with other state-level policies implemented during early stages of the pandemic. These findings coincided with behavior change likely associated with gubernatorial state-of-emergency declarations: schools suspended in-person classes, sporting events restricted spectators, and large gatherings were banned. A return to 2019 levels in the number of MVC injuries and severe or fatal injuries was observed in period 4, perhaps due to increased alcohol and cannabinoid use, speeding, harsh acceleration and braking events, and mobile phone use observed among drivers following easing of COVID-19 lockdowns.5,6
This study has limitations. As injury severity in Ohio crash reports was identified by police officers rather than medical professionals, nondifferential misclassification may exist. Additionally, the public’s response to the pandemic may have been influenced by factors outside of policy (eg, media coverage). Also, generalizability beyond Ohio may be limited. Results were presented by various periods to facilitate cross-state comparisons.
As the pandemic continues, policy makers should consider the effects of lockdown and reopening policies on factors beyond COVID-19 infection, including MVC-related injuries and deaths.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.Mervosh S, Lu D, Swales V. See which states and cities have told residents to stay at home. New York Times. Updated April 20, 2020. Accessed November 20, 2020. https://www.nytimes.com/interactive/2020/us/coronavirus-stay-at-home-order.html
- 2.Office of Highway Policy Information, US Federal Highway Administration . Traffic volume trends: April 2020. Accessed November 20, 2020. https://www.fhwa.dot.gov/policyinformation/travel_monitoring/20aprtvt/20aprtvt.pdf
- 3.Ohio Department of Public Safety . Crash Statistics System. Accessed September 1, 2020. https://ohtrafficdata.dps.ohio.gov/crashstatistics/home
- 4.Ohio Department of Transportation . Traffic Monitoring Management System. Accessed September 1, 2020. https://odot.ms2soft.com/tcds/tsearch.asp?loc=odot
- 5.Thomas FD, Berning A, Darrah J, et al. Drug and Alcohol Prevalence in Seriously and Fatally Injured Road Users Before and During the COVID-19 Public Health Emergency. National Highway Traffic Safety Administration Report DOT HS 813 018. Published October 2020. Accessed November 20, 2020. https://rosap.ntl.bts.gov/view/dot/50941
- 6.Katrakazas C, Michelaraki E, Sekadakis M, Yannis G.. A descriptive analysis of the effect of the COVID-19 pandemic on driving behavior and road safety. Transportation Research Interdisciplinary Perspectives. 2020;7. doi: 10.1016/j.trip.2020.100186 [DOI] [PMC free article] [PubMed] [Google Scholar]