According to the Center for Disease Control and Prevention (CDC), road traffic collisions are a leading cause of death in the United States for people aged 1–54. In 2019, there were 33,244 fatal motor vehicle crashes involving 36,096 deaths in the country, which translates into a crash rate of 11 deaths per 100,000 people and 1.11 deaths per 100 million miles traveled. By State, the fatality rate per 100,000 people ranged from 3.3 in the District of Columbia to 25.4 in Wyoming and the death rate per 100 million miles traveled ranged from 0.51 in Massachusetts to 1.73 in South Carolina. [1]

Vision Zero is a road safety project that aims to completely eliminate all traffic deaths and severe injuries, promoting safe, healthy, and equitable mobility for all. Its underlying principle is that “it can never be ethically acceptable that people are killed or seriously injured when moving within the road transport system.” [2] The strategy has proved its worth across Europe and is gaining momentum in major American cities.

The Vision Zero community recommends that the cities implementing this strategy develop High Injury Networks (HINs) and pay more attention to the corridors spotlighted in the networks. Creating High Injury Networks is an important Vision Zero exercise that implies mapping of stretches of roadways where high severity collisions concentrate with an emphasis on pedestrians and bicyclists. This exercise helps identify corridors that carry a higher risk of injury within a transportation network. Developing a High Injury Network can prove helpful for a variety of reasons, including:

  • Determining geographic areas where crashes are concentrated, and the causes of these crashes, so that efforts can be focused on the most challenging areas and crash factors.
  • Strengthening collaboration to focus street improvements and education campaigns (e.g., Go Human) along the HIN.
  • Prioritizing investments within these areas to reduce collisions. [3]

Example of a High Injury Network – San Francisco, CA [7]

Cities across the United States have developed different approaches to come up with their High Injury Networks. The accuracy of a HIN is highly dependent on the data source used to create the network. Most cities, for example Los Angeles and Atlanta, use police crash reports as a primary dataset for identification of crash locations. In most states, crash reports are completed by officers at the scene of the collision and are stored electronically in a statewide database. However, there are some challenges associated with using this data source:

  • It may lack detail about the collision that could inform future analyses,
  • It could include inaccurate location information,
  • It may not be available in a timely manner. [4]

Recent studies suggest that many traffic collisions also go unreported in these datasets. For example, a study by the National Highway Traffic Safety Administration indicated that approximately 10% of injury-crashes were not reported to police. To account for the unreported cases, the San Francisco Department of Public Health (SFDPH) updated its High Injury Network in 2017 by mapping both hospital and police injury data. Using data from Zuckerberg San Francisco General Hospital and Trauma Center, SFDPH and San Francisco Municipal Transportation Agency (SFMTA) found that:

  • 7% of the pedestrian severe injuries were not included in its crash database,
  • 1% of the bicyclist severe injuries were not included, and
  • 1% of the vehicle severe injuries were not included. [4]
Community Health and Equity Index Map - Los Angeles, CA [6]

Health and Equity Index Map – Los Angeles, CA. [6]

There are a variety of methodologies that cities use to process the available crash data.

For example, the City of Los Angeles, CA identified corridors based on intersection scores, which were calculated as a sum of Fatality multiplied by 1.5, Severe Injury, Child or Senior Dummy, and Target Community Dummy. Their emphasis was on fatal and serious injury (FSI) crashes involving people walking and bicycling. The city used Health and Equity Index as identified in the City Health Atlas to identify burdened communities. [5,6]

An Illustration of the quarter-mile overlapping “corridorized” sections - San Francisco, CA [7]

Graphic Illustrating the Quarter Mile Overlapping “Corridorized” Sections in San Francisco’s Methodology. [7]

The City of San Francisco, CA, as mentioned earlier utilized both hospital and police crash data. Their methodology converted each street segment block into quarter mile overlapping “corridorized” sections. It selected the highest scoring (i.e., most injuries per mile) sections using an injury per mile threshold to capture high crash locations. [7]

The City of Atlanta, GA used GIS software to spatially sum up number of fatalities and injuries within 25 feet of each roadway segment. Their methodology used equitable target areas (ETAs) reflecting concentrations of poverty, minority, and/or senior populations. If a roadway segment fell within an equitable target area, a score of one through three was assigned. Otherwise, a score of zero was used. The purpose of using ETAs was to provide a slight increase in priority to traditionally under-invested communities. Final score for each feature was a sum of Number of Fatalities multiplied by 5, Number of Injuries, and Equitable Target Area Score divided by 3. [8]

The City of Jersey City, NJ mapped all police-reported crashes where people were killed or seriously injured and analyzed them to identify the most dangerous streets in the city. Their final HIN comprised of two versions: one that included all roads in the City regardless of jurisdiction and another one with City-maintained roads only. [9]

The findings from different cities are different but suggest a similar pattern. Most severe and fatal traffic collisions occur on a limited number of streets representing a relatively small share of the roadway network. Thus, in Los Angeles over 65% of all severe and fatal traffic collisions involving people walking occur on just 6% of the City streets. [6] In San Francisco, about 77% pedestrian, 71% bicyclist, and 74% vehicle FSIs occur on 13% of the city streets. [7] Approximately 71% of the fatalities and 42% of the injuries in the City of Atlanta occurred on 5.5% of its roadway network. [8] The City of Jersey City’s HIN captured 80% of fatal crashes on 16% of the total roadway network. [9]

The City of Atlanta’s High-Injury Network and Percent Minority Population by Block Group Map. [4]

The findings also suggest a strong correlation between HINs and race and income level distribution in a region. Overall, neighborhoods with High Injury Network streets had lower median incomes, a larger share of African American residents, higher rates of walking and taking transit to work, and lower rates of vehicle ownership. [10]

Atlanta’s recently adopted transportation plan shows that neighborhoods with a higher concentration of high injury corridors have some of the lowest sidewalk coverage in the city. Additionally, less than one mile of the High Injury Network is located in neighborhoods with median income in the top 20% in the city; two-thirds of the entire network is located in areas whose median income is in the bottom 40%. [10]

In San Francisco, the new methodology integrating hospital reports with police data shows that communities where both children and seniors concentrate see more traffic deaths and injuries, as do majority African American neighborhoods. [11]

The findings of a High-Injury Network help in identifying locations for future investment to reduce traffic fatalities and serious injuries. They show where concentrations of severe injuries and fatalities occur so that city leaders, city staff, mobility and disability advocates, law enforcement, public health officials, and all residents can make informed decisions, prioritizing limited resources and safety improvements along high-frequency and high-risk corridors.















Cover Image: Screenshot of Different Rankings of Roadway Corridors in New Brunswick, NJ. (Data Source: NJTPA 2019 Network Screening)