A New Tool to Assess the Costs and Effectiveness of Traffic Crash Interventions


Apr 4, 2016

Emergency responders helping at a traffic accident

Photo by inhauscreative/iStock

This commentary originally appeared in ITE Journal on April 1, 2016.

Motor vehicle crashes are a leading cause of accidental death in the United States. In 2014, more than 32,600 Americans were killed and more than 2.3 million were injured in crashes.[1] The direct and indirect costs of crashes are substantial. One study found that crash-related costs—which include a variety of costs, such as medical care, productivity, and travel delay—reached at least $242 billion in 2010.[2] And this estimate does not fully capture the price of all consequences, including long-term harm to family and community quality of life.

A wide range of evidence-based policy and program interventions can help prevent vehicle crash injuries and deaths. However, with limited resources, state policymakers must choose the interventions that will provide the greatest reductions for the money spent. To do this, they need to understand how much interventions will cost and how effective they will be in their state.

While information on effectiveness has been available for specific interventions, the information is generally specific to the particular location where the intervention was implemented. In addition, information on the costs of implementing these interventions is relatively limited.

To address this problem, the RAND Corporation, with funding from the National Center for Injury Prevention and Control at the Centers for Disease Control and the Robert Wood Johnson Foundation, developed an online tool called Motor Vehicle Prioritizing Interventions and Cost Calculator for States (MV PICCS) that generates state-specific cost-effectiveness estimates. The tool is available for free at www.cdc.gov/motorvehiclesafety/calculator.

MV PICCS allows users to compare 14 interventions on cost and effectiveness. These 14 interventions were selected based on the following four criteria. The interventions:

  1. Are meant to change driver or passenger behavior (as opposed to changes to roadway or vehicle engineering);
  2. Can be implemented (or influenced) by states;
  3. Are demonstrated to be effective, based on past research; and
  4. Are not already in widespread use.

Table 1 describes the 14 interventions, along with their estimated effectiveness in reducing deaths within specific types of crashes. For example, red-light cameras are expected to reduce deaths that occur at signalized intersections by 17 percent.

Using MV PICCS to Identify Cost-Effective Interventions by State

MV PICCS determines the cost-effectiveness of each intervention in each state. Costs are defined as the cost paid by the state to implement the intervention. For some costs, we were able to scale them to each state based on state characteristics (e.g., wage rates, population, roadway miles). For example, the average cost of police time varies by state. For others, we developed a “most common cost” and used it across all states (for example, an average per-camera cost based on actual per-camera costs for speed cameras in different cities). The costs of upfront items, such as passive alcohol sensors, were spread over five years. We also estimated the fines and fees that would be paid to the state by offenders, since these could offset some of the costs.

Table 1. List of 14 Interventions in MV PICCS with Estimates of Reductions in Deaths.[3]

Intervention Description Type of Crash Reduction in Deaths
Red-light cameras Cameras used to capture images of vehicles whose drivers fail to stop for red lights, primarily at intersections. Tickets are sent to offenders by mail. Occurred at intersection with light 17%
Speed cameras Cameras are used to capture images of vehicles whose drivers are exceeding posted speed limits. Mobile speed cameras are often used to cover multiple road segments. Speed-related 12%
Alcohol interlocks These devices prevent a vehicle from starting until the driver has blown into a tube to prove sobriety. Previous DWI 24%
Sobriety checkpoints At a specific location, teams of police officers stop cars to check whether drivers are intoxicated. Alcohol-related 8%
Saturation patrols Police patrol selected locations, looking for suspicious driving behavior in an attempt to identify alcohol-impaired drivers. Alcohol-related 18%
Bicycle helmet laws This law mandates that children who ride bicycles wear helmets. Bike 15%
Motorcycle helmet laws This law requires all motorcyclists, regardless of age or experience level, to wear helmets that meet U.S. Department of Transportation safety standards. Motorcycle 29%
Primary enforcement of seat belt laws States with seat belt laws vary in their enforcement; a primary law allows police to ticket offenders exclusively for not wearing seat belts. Passenger-vehicle occupants 7%
Seat belt enforcement campaign This policy combines the intense enforcement of seat belts over a fixed period with a publicity campaign. Passenger-vehicle occupants 5%
License plate impoundment This intervention requires a driver who has been convicted of DWI to surrender the vehicle's license plate, which is either impounded or destroyed. Previous DWI conviction 27%
Limits on diversion and plea agreements These rules prevent DWI arrestees from diverting cases out of the judicial system or pleading out of charges. Previous DWI conviction 11%
Vehicle impoundment This intervention requires that a DWI offender's vehicle be confiscated for a period of time, after which the offender either reclaims or surrenders the vehicle. Previous DWI conviction 30%
In-person license renewal This intervention requires any driver over age 70 to renew his or her driver's license in person at a department of motor vehicles, instead of using mail-in or online renewal. Drivers over 70 9%
Higher seat belt fines This intervention adds $75 to a state's existing fine, which represents a significant increase over existing seat belt fines in most states. Passenger-vehicle occupants 7%

NOTE: DWI = driving while intoxicated. All estimates are the same for injuries, except for sobriety checkpoints, with an estimated 20% reduction in injuries.

Effectiveness is defined in two ways. First, we used estimates from the literature of the percent of injuries and deaths that could be avoided with each intervention. These are specific to crash types, as shown in Table 1. We chose the estimates from the best available study or meta-analysis of each intervention and translated that into state-specific effects based on the number of crash deaths in that state for that particular crash type. Even in two states with similar populations, one may have a greater number of lives saved due to motorcycle helmet laws, based on the number of past motorcycle deaths.

Second, we translated effectiveness into dollars based on the estimated value of saving a life or preventing an injury. We used existing literature to estimate these values for individual states, based on nine categories, such as medical costs, property damage, lost productivity, and insurance.[4]

Based on these dollar values, the cost-effectiveness score is the ratio of the effectiveness to the cost. The higher the score, the more cost-effective the intervention. Scores of less than one are not cost-effective, meaning that the cost is greater than the benefit.

These assumptions are all programmed into MV PICCS. Users can then customize their analysis by specifying the five inputs identified in Table 2.

Table 2. User Inputs to MV PICCS

Variable Comment
State The tool can analyze only one state at a time.
List of interventions to analyze The user can investigate any combination of the 14 interventions.
Analysis type The user must select one of two analysis types. The cost-effectiveness analysis looks at each policy separately, while the portfolio analysis takes related policies into account.
Run type The user can choose to include the fines and fees that offenders pay to the state.
Budget The user must tell the tool how much money is available to implement the selected interventions.

With this combination of inputs, users can find the most cost-effective way to spend limited dollars on traffic safety, and develop state-specific annual estimates for costs and effectiveness of policies and programs. The tool can analyze one policy or all 14. MV PICCS can also run in “portfolio analysis” mode, meaning that it adjusts for related policies. For example, if we simply add the lives saved from the three seat-belt related interventions, the estimate will be too high, because we expect some overlap between the policies.

Using MV PICCS Data to Compare National Spending on Traffic Safety

The tool analyzes one state at a time. However, based on the underlying data that we compiled for MV PICCS, we can look across states and identify the most cost-effective ways to spend money on traffic safety at a national level. We considered three policy questions from a national perspective:

  • Should interventions be selected nationally or state-by-state?
  • What is the most cost-effective way to allocate an increase in funding for interventions?
  • What is the most cost-effective way to reduce drunk driving?

To do this, we developed a cost-effectiveness estimate for each intervention in each state—714 in all (14 interventions in 50 states, plus the District of Columbia; for convenience we will refer to 51 states). We then took some of these out of consideration, for two reasons. First, we eliminated three interventions—two (saturation patrols and high-visibility enforcement) because we did not find reliable information on where they are already implemented, and one (increased seat belt fines) because the implementation cost is zero. So this analysis covers 11 interventions, not the original 14. Second, we removed those interventions that are already implemented in individual states. This leaves 298 potential interventions across the 51 states. If we implement all 298, we would save 3,939 lives for a cost of $2.1 billion (this and other figures for lives saved are on an annual basis. Comparisons of this baseline to the three policy questions are shown in Table 3.

Should Interventions Be Selected Nationally or State-by-State?

To address this question, we generated the cost-effectiveness of each intervention at a national level by summing the costs and effectiveness of the intervention in all states where it is not currently implemented. The cost-effectiveness of these interventions ranges from 130 for alcohol interlocks to 0.8 for limits on diversion and plea agreements. Under the national approach, if all states implemented the three most cost-effective national interventions (alcohol interlocks, universal motorcycle helmet laws, and license plate impoundment), 1,219 fatalities would be prevented for a total cost of approximately $55 million (See Table 3). Universal motorcycle helmet laws alone would prevent 745 fatalities and cost $41 million to implement. Under a state-specific approach, where each state implements the most cost-effective intervention based on the state-level estimates, 928 fatalities would be prevented and it would cost about $60 million.

The national approach is more cost-effective, in that more fatalities are prevented at a lower cost, but it does not spread the reduction in fatalities across all states. The main reason that the state-based approach is less cost-effective overall is that it leads some states, where the most cost-effective interventions are already in place, to implement interventions with low cost-effectiveness ratios. Maryland is an extreme example of this because it has already implemented 9 of the 11 interventions considered in this analysis. As a result, even the most cost-effective prospective additional intervention (among those we considered) in Maryland has costs that exceed the expected benefits.

Table 3. Comparison of Responses to Policy Questions.[5]

If we: It saves this many lives and prevents this many injuries in this many states for a total benefit of at a cost of for a cost-effectiveness ratio of
Implement all 11 interventions in states that do not already have them 3,939 430,100 51 $14.1 billion $2.1 billion 6.8
Should interventions be selected nationally or state-by-state?
Implement the top-ranked intervention in each state 928 176,000 51 $4.9 billion $60 million 81.8
Implement the 3 most cost-effective interventions in all states that do not already have them 1,219 214,000 45 $6 billion $55 million 109.9
Implement universal motorcycle helmet laws in all states that do not currently have them (subset of the response above) 745 197,000 30 $5 billion $41 million 122.4
What is the most cost-effective way to allocate an increase in funding for interventions?
Increase each state's traffic safety funding by 10% 660 46,600 47 $1.9 billion $28 million 66.0
Implement the most cost-effective interventions, regardless of state, with the same amount of money 1,320 225,800 44 $6.4 billion $58 million 110.7
What is the most cost-effective way to reduce drunk driving?
Implement all DWI interventions in states that do not currently have them 1,182 52,700 49 $2.7 billion $764 million 3.5
Implement the 10 most cost-effective DWI interventions 170 6,100 10 $344 million $2 million 167.0

What Is the Most Cost-effective Way to Allocate an Increase in Funding for Interventions?

For this question, we considered the effects of a 10-percent increase in federal funding to states (approximately $57.9 million divided among the 51 states) to implement additional interventions and compared two ways to allocate this increase. The first is to increase each state's individual allocation by 10 percent and implement those interventions that are most cost-effective within that state. This approach spends only $28.4 million, because many states cannot use the full allotment as the cost of many interventions exceeds the 10-percent increase. Allocating funds in this way would save 660 lives in 47 states.

The second approach is to take the same $57.9 million and spend it on the most cost-effective interventions, regardless of state. We ranked interventions by their state-specific cost-effectiveness ratio and selected the most cost-effective until the cumulative cost reached $57.9 million. This approach would save 1,302 lives using the majority of the available funding ($56.9 million) in 44 states. This approach is far more cost-effective than giving each state a 10-percent increase in funding. One drawback, however, is that fewer states benefit (44 vs. 47).

What Is the Most Cost-effective Way to Reduce Drunk Driving?

Using five interventions that target driving while intoxicated (DWI) specifically, we look at the cost-effectiveness ratios across all states. There are 255 possible intervention–state combinations (51 states times five interventions), and excluding those that are already in use leaves 119. If we implemented all 119 interventions in the states where they are not in place, 1,182 lives are saved at a cost of $764 million. Ten of these have cost-effectiveness ratios exceeding 100; if just these ten were implemented, they would save 170 lives in ten states at a cost of $2.1 million. This is a striking difference in cost-effectiveness; we can save 14 percent of the fatalities with less than 0.5 percent of the cost.

Limits and Assumptions

This analysis has three important limitations:

First, it is limited to 14 interventions. The tool does not include every possible intervention because of the criteria we used to select them. For example, the tool omits bans on cell phone use while driving for which the evidence is still conflicting. It also excludes engineering interventions such as improved vehicle safety or road quality.

Second, many assumptions were needed to generate these estimates. Four of the most important are summarized here. First, the cost-effectiveness estimates reflect assumptions about the level and characteristics of implementation (for example, the number of cameras needed for red-light and automated speed-camera enforcement). Second, the effectiveness estimates from the literature are based on conditions in a specific jurisdiction, which might not reflect the conditions in others. Third, effectiveness estimates for injuries were not available for most interventions, so, without more-specific information, we assumed that the reduction was the same as for fatality reductions. Finally, for some components of the state-specific estimates we used data from national databases, which may or may not be accurate for any particular state. Full details about the assumptions are provided in a documentation report.[6]

Third, the tool is not continually updated. We used best available data at the time of the research. However, we have not updated two key sets of data: the number of fatalities by type per state and the implementation status of each intervention in each state. If, for example, deaths in vehicle crashes have declined considerably in some states since 2010, the tool's estimates of lives saved will be too large.

Despite these limitations, we believe that the analyses can be of great use to state policymakers. They provide policymakers a sense of the relative costs and effects of the different interventions under consideration, and a variety of ways to conduct state-specific analysis.


  • [1] National Center for Statistics and Analysis, “Traffic Safety Facts 2013: A Compilation of Motor Vehicle Crash Data from the Fatality Analysis Reporting System and the General Estimates System,” National Highway Traffic Safety Administration, 2014.
  • [2] 2. Blincoe, L., T. R. Miller, E. Zaloshnja, and B. A. Lawrence, “The Economic and Societal Impact Of Motor Vehicle Crashes, 2010 (Revised),” National Highway Traffic Safety Administration, Washington, DC, May 2015.
  • [3] Ringel, J., J. Zmud, K. Connor, D. Powell, B. Chow, L. Ecola, et al., “Reducing Motor Vehicle Deaths and Injuries Through Behavioral Interventions: Project Report and Online Tool Documentation,” RAND Corporation, Santa Monica, CA, 2015.
  • [4] Blincoe, et al., 2015.
  • [5] Ecola, L., B. Batorsky, and J. Ringel, “Using Cost-effectiveness Analysis to Prioritize Spending on Traffic Safety,” RAND Corporation, Santa Monica, CA, 2015.
  • [6] Ringel, et al., 2015.

Liisa Ecola is a senior project associate and transportation researcher at the RAND Corporation. Jeanne S. Ringel is a senior economist at RAND and director of the Population Health Program.