The scope of deadly hazards such as texting and drug use by drivers may be underestimated and not adequately addressed because police aren’t collecting enough information at crash scenes, according to a new report.

The report, released today by the National Safety Council, also found that no state systematically records whether crashes involve vehicles with self-driving features, such as collision-avoidance systems.

The group said more attention must be focused on the problem with a shift from an “accident-report mentality” to crash investigation. It is important to know not just what happened, but why it happened, said Deborah Hersman, chief executive of the safety council, a nonprofit group.

“Better data enables us to make better decisions when it comes to our priorities, our investment and our technology,” she told FairWarning.

The report follows the safety council’s recent preliminary estimate that traffic deaths in 2016 were 14 percent higher than in 2014 – totaling 40,200. If the figures are borne out, 2016 would be the first time since 2007 that deaths have exceeded 40,000.

That comes despite vehicles being more crashworthy and the council’s estimate that the miles traveled during 2016 increased only three percent. Federal officials have yet to release their 2016 fatality report.

Gaps in the forms

In compiling their report the safety council’s researchers reviewed accident-report forms used by highway and law enforcement agencies in all 50 states. They checked whether those forms have a specific space or code to note factors such as texting while driving. Gaps in the forms make collecting and sharing information with researchers and officials difficult, if not impossible, according to the safety council.

Among its findings:

• Although 46 states ban texting behind the wheel, 26 states lack a space or code to record whether a driver was texting.

• Of the eight states that have legalized marijuana for recreational use, only four have a specific space or code to report its use. They are Alaska, California, Oregon and Washington.

• No state has a specific code to note a vehicle with self-driving features such as collision avoidance.

• Only 25 states have a space or code to note whether a driver has a learner’s permit.

Hersman said a major concern is the failure to record information about crashes involving new models with self-driving features. “As we move towards automated vehicles we really need to understand when active technologies are engaged,” she said.

“We have to get better addressing these technologies and their efficacy, how they work, when they fail. If there is a failure we have to understand so we can get on top of it early.”

Several thousand self-driving vehicles will be sold in 2020, increasing to about 4.5 million in 2035, according to an analysis last year by IHS Markit, a market research firm.

Safety researchers already conduct crash tests and computer simulations trying to determine how well a vehicle will protect its occupants. But detailed information from a crash is important to understand what happens in the real world, said Charles Farmer, the vice president for research at the Insurance Institute for Highway Safety, a research group funded by the insurance industry.

Researchers like Farmer have yearned for that information for at least three decades.

Since 1998, the National Highway Traffic Safety Administration, the federal agency that regulates traffic safety, has been working with the Governors Highway Safety Association trying to get police to collect more detailed and standardized information. Their recommended Model Minimum Uniform Crash Criteria has 110 items.

Every state uses that form “to some degree and most states use most of the data elements,” said Barbara Harsha, former GHSA executive director.

Despite NHTSA and GHSA working on the issue for almost two decades, the safety council report concluded that “no state is adequately capturing the crash data we need to understand why crashes are rising, and form an effective path forward.”

Kara Macek, a GHSA spokeswoman said in an email that “GHSA agrees that there is always room for improvement.”

But, she said, it also requires the cooperation of state agencies “and balancing the data needs against the limited amount of time and resources law enforcement have to complete the crash investigation.”

NHTSA officials didn’t respond to an interview request.

Probable undercount

However, the agency has expressed concern about needing more information in areas such as distracted driving. It estimated there were about 400 deaths in 2014 linked to the use of a cell phone. But the agency concedes that was probably an undercount, due to the difficulty of determining if the driver was using a wireless device.

The need for better crash information has also been noted by the National Transportation Safety Board, which investigates significant transportation accidents. Its concerns include “improving and refining the Model Minimum Uniform Crash Criteria,” Christopher O’Neil, an NTSB spokesman, told FairWarning in an email.

Last August safety consultant Carl Nash, a former NHTSA official and now chief scientist with the Center for Auto Safety, a nonprofit advocacy group, told NTSB officials that police collection of crash information is “hopelessly out-of-date.” He said NHTSA has struggled for decades “to get comprehensive, accurate, representative highway crash data.”

Nash said police should be relying more on “modern computer technologies” to gather and share information. That includes getting information from event data recorders — the “black box” — found on most vehicles. In a crash it can record information including speed, braking and seat-belt use.

The safety council’s Hersman said she hopes concern over an increase in deaths will provide a “clarion call” for local, state and federal officials to focus more sharply on gaps in information.

Forms and coding aside, there is also the issue of manpower and the police officer at the scene, said Kenneth Spears, the executive director of the American Association of State Troopers.

“I guess there is a balance there for the agency trying to determine what data actually needs to be collected versus having the police officers’ time best used for other activities,” he said.

About FairWarning

This story was reported by FairWarning (www.fairwarning.org), a nonprofit news organization based in Pasadena, Calif., that focuses on public health, safety and environmental issues.