Applying the Crash Trifecta Approach to SHRP 2 Data

The crash trifecta model does not consider crash genesis as a simple unitary element, but rather as a convergence of three separate, converging elements: (1) unsafe pre-incident behavior or maneuver; (2) transient driver inattention; and (3) an unexpected traffic event. Previous results from Phase I of the Crash Trifecta study showed that the presence of all three crash trifecta elements increased as the severity of a safety-critical event (SCE) increased. Given the limited number of crashes available in Phase I, however, it was not possible to identify trends in the presence of specific crash trifecta elements or to break the data down by incident type or crash severity. The current study built on the methods and results from Phase I by applying the crash trifecta model to the Second Strategic Highway Research Program (SHRP 2) Naturalistic Driving Study (NDS), which greatly increased the number of SCEs available for analysis.

The results of Phase II show that elements well within a driver’s control are at the core of the majority of SCEs. Unsafe driving behavior was the most prevalent crash trifecta element, occurring in 70% of crashes and 52% of near-crashes. Unsafe driving behavior combined with transient inattention contributed to over 25% of crashes and almost 33% of at-fault crashes in the current study, compared to 5% of near-crashes and 8% of at-fault near-crashes, indicating that a crash is much more likely to occur if the unsafe driver is also not paying attention.

The prevalence of the remaining two crash trifecta elements (i.e., transient inattention and unexpected event) varied depending on the severity of the SCE. An unexpected event was more likely to be present in near-crashes (74%) compared to crashes (25%), while the opposite was true for transient inattention near-crashes (28%) and crashes (43%).

The increased number of SCEs in Phase II compared Phase I meant that the data set could be broken down by incident type for a more in-depth assessment of the applicability of the crash trifecta model. Of the 16 different incident types, the most common crashes were animal related, rear end (striking), rear end (struck), and road departure (left or right). The most common near-crashes were animal related, rear end (striking), sideswipe (same direction), and turn into path (same direction). The majority of different types of near-crashes tended to be associated with pedestrians, animals, pedalcyclists, or other vehicles behaving unexpectedly. The presence of transient inattention in a number of incident types resulted in a higher proportion of crashes than near-crashes.

As was the case in Phase I, the results of the current Phase II study suggest that assigning a single, unitary critical reason as the proximal cause of the SCE without considering additional contributing factors is likely to be a limitation that does not address the complexities involved in the genesis of a crash.