Revised Estimates Of The U.S. Drowsy Driver Crash Problem Size Based On General Estimates System Case Reviews

NHTSA · 1995 · ROSA P / United States. Joint Program Office for Intelligent Transportation Systems

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This study addresses the underestimation of drowsy driver crashes in U.S. traffic safety data. Previous estimates from the General Estimates System (GES) indicated that drowsiness/fatigue was a factor in only 0.9% of police-reported crashes (56,000 annually). However, these figures were widely regarded as conservative due to inconsistent police reporting, lack of verifiable evidence, and coding limitations. The research was motivated by contrasting findings from New South Wales, Australia, which used inclusive trajectory-based criteria to estimate that fatigue caused 6% of all crashes. The authors sought to determine if similar inclusive methods applied to U.S. data would yield accurate results or if they would overestimate the problem by including crashes caused by awake inattention or other factors. To refine these estimates, the authors analyzed 562 Police Accident Reports (PARs) from the 1993 GES dataset. They selected cases that exhibited a “Drift-Out-Of-Lane” (DOOL) trajectory but were not originally coded as drowsiness-related. These cases were divided into two subgroups: “pure” DOOL crashes (single-vehicle, no alcohol/drugs, specific speed and weather conditions) and “other” DOOL crashes. The researchers compared the statistical profiles of these candidate crashes against known drowsiness-cited crashes and conducted detailed case reviews to classify causal factors. Cases were categorized as “definite,” “probable,” or “possible” drowsiness-related based on narrative evidence and crash circumstances, distinguishing them from crashes caused by awake inattention or physiological issues. The results indicated that inclusive trajectory-based definitions significantly overestimate drowsiness involvement. Statistical comparisons revealed that non-cited DOOL crashes differed markedly from drowsiness-cited crashes in time-of-day distribution and other characteristics, such as higher rates of speeding and curve-related incidents. Case reviews found that only 9.3% to 15.2% of “pure” DOOL cases and 1.3% to 2.9% of “other” DOOL cases were actually drowsiness-related. By applying these percentages to national totals, the authors revised the annual estimate of drowsy driver crashes from 56,000 to a range of 79,000 to 103,000. This represents 1.2% to 1.6% of the 6.3 million annual police-reported crashes, a modest increase over the baseline but far lower than the 6% suggested by inclusive Australian methodologies. The study concludes that while GES baseline estimates are conservative, they are not drastically underestimated. The revised range of 1.2% to 1.6% aligns with other broad-sample studies, such as the National Accident Sampling System, which found a 1.5% incidence. The authors emphasize that simply adding crashes based on trajectory definitions inflates estimates by including non-drowsiness causes. The paper highlights the need for more sophisticated data collection, such as in-vehicle instrumentation and post-crash interviews, to accurately assess driver alertness. It also notes that current estimates likely remain conservative for non-police-reported crashes and fatal crashes, where drowsiness may be underreported due to lack of witness testimony or driver candor.

Key finding

Drowsiness/fatigue is a discernible causal factor in 1.2 to 1.6 percent of police-reported crashes, compared to the baseline GES estimate of 0.9 percent.

Methodology

dataset

Sample size: 562

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• incidence prevalence
• drowsiness
• naturalistic crash near crash
• drowsiness detection algorithms
• causation analyses
• pre crash contributing factors

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: crash risk outcomes, observational prevalence
• Methodological Resource: dataset resource