SPATIAL AND TEMPORAL PATTERNS OF FATIGUE RELATED CRASHES IN HAWAII By Karl E. Kim Eric Y. Yamashita Hawaii CODES Project Traffic Records Forum July 29 - August 2, 2001 New Orleans, Louisiana
Overview Background Data & Methods Findings Countermeasures Conclusion
Background Fatigue-related crashes difficult to identify; Underreporting & misclassification; Occur most often in the early morning hours with a smaller peak in the mid-afternoon; Nationwide, younger drivers (<30 of age) accounted for 2/3 of drowsy driving crashes; Majority of fatigued drivers are males.
Typically occur on long straight stretches of roadway; 1 percent of all the crashes nationwide are cited as being fatigued-related In Hawaii, 3 percent of all the crashes that occurred from 1986 to 1995 were fatiguerelated crashes According to NHTSA, alcohol was consumed in nearly 20% of all fatiguerelated crashes.
Fatigue and Alcohol On 4 hours of sleep, 1 can of beer can have the same impact as a six-pack; 18 hours of sustained wakefulness produces performance impairment equivalent to.05 BAC; 24 hours without sleep =.10% BAC Source: National Sleep Foundation, 2000
Data & Methods Crash Outcome Data Evaluation System (CODES) Project Funded by the U.S. D.O.T., NHTSA. Data includes linking Crash, EMS, Hospital, and Claims Data Linked with Automatch Hawaii s Traffic Safety GIS Statistical Analysis + Spatial Analysis = Problem Identification
Findings Driver Characteristics Gender Age Fatigue Related Non-Fatigue Related Z-test Frequency Percent Frequency Percent Male 4966 76.2 134186 68.0 13.94 Female 1550 23.8 63015 32.0-13.94 Total 6516 100.0 197201 100.0 15-19 1274 19.7 26944 14.5 11.68 20-24 1647 25.5 31979 17.2 17.26 25-29 1099 17.0 27371 14.7 5.10 30-34 738 11.4 23413 12.6-2.79 35-39 513 7.9 18900 10.2-5.83 40-44 347 5.4 14633 7.9-7.37 45-49 222 3.4 10362 5.6-7.40 50-54 158 2.4 7298 3.9-6.06 55-59 140 2.2 6074 3.3-4.91 60 & over 330 5.1 19189 10.3-13.64 Total 6468 100.0 186163 100.0 Mean Age 30.0 34.9 Std. Deviation 13.2 15.9 t-test Prob.= 0.0001 The are 3.2 times more males than females involved in fatigue related crashes. 4966 v. 1550. Sixty-two percent of the fatigued drivers were under 30 years old.
Findings Vehicle Characteristics Fatigue Related Non-Fatigue Related Z-test Frequency Percent Frequency Percent Crash Type Single Vehicle 4252 65.1 41606 20.2 86.86 Pedestrian 9 0.1 5316 2.6-12.44 Bike/Moped 12 0.2 4636 2.3-11.24 Vehicle-to-Vehicle 2255 34.5 154347 75.0-73.04 Total 6528 100.0 205905 100.0 Vehicle-to-Vehicle Crashes Head On 246 10.9 4398 2.8 22.40 Rear End 1093 48.5 67726 43.9 4.36 Sideswipe Same Dir. 283 12.5 16112 10.4 3.25 Sideswipe Opp. Dir. 236 10.5 5915 3.8 16.10 Angle Same Dir. 99 4.4 11669 7.6-5.67 Angle Opp. Dir. 159 7.1 15366 10.0-4.58 Broadside 138 6.1 32999 21.4-17.61 Other 1 0.0 162 0.1-0.89 Total 2255 100.0 154347 100.0 Vehicle Manuever Prior to Crash Straight Ahead 5811 89.2 110726 54.6 55.38 Changing Lanes 107 1.6 12228 6.0-14.79 Merging 19 0.3 2965 1.5-7.84 Overtaking 23 0.4 2794 1.4-7.06 Slow/Stopping 53 0.8 10471 5.2-15.81 Right Turn 73 1.1 8519 4.2-12.33 Left Turn 106 1.6 29427 14.5-29.39 Other 321 4.9 25713 12.7-18.65 Total 6513 100.0 202843 100.0 65 percent of the fatigue related crashes are single vehicle crashes. 89 percent of the vehicles involved in crashes were going straight at the time of the crash.
Findings Spatial Factors Fatigue Related Non-Fatigue Related Z-test Frequency Percent Frequency Percent Freeway/Highway Yes 3098 47.4 76135 36.9 17.28 No 3433 52.6 129995 63.1-17.28 Total 6531 100.0 206130 100.0 Urban/Rural Urban 4404 67.4 170527 82.7-31.86 Rural 2127 32.6 35599 17.3 31.86 Total 6531 100.0 206126 100.0 Intersection Yes 1195 18.3 81320 39.5-34.54 No 5336 81.7 124810 60.5 34.54 Total 6531 100.0 206130 100.0 Vertical Road Alignment Level 4154 64.5 149525 73.6-16.35 Unlevel 2291 35.5 53631 26.4 16.35 Total 6445 100.0 203156 100.0 Horizontal Road Alignment Straight 5050 77.8 180669 88.5-26.22 Curved 1444 22.2 23577 11.5 26.22 Total 6494 100.0 204246 100.0 67 percent of fatigue related crashes occur on urban roads. 78 percent occur on straight roads 64 percent occur on level roads
Temporal Factors 700 18000 600 16000 Freq. (Fatigue) 500 400 300 200 100 Late Night or Early Morning Late Afternoon 14000 12000 10000 8000 6000 4000 2000 Freq. (No Fatigue) Fatigue Related Non- Fatigue Related 0 0 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Hour of the Day Two peak periods exist - Late night/early morning and Evening. 58 percent of the fatigue-related crashes occur during the hours of 12 AM to 6 AM.
Temporal Factors 1500 38000 Freq. (Fatigue) 1300 1100 900 700 500 300 36000 34000 32000 30000 28000 26000 24000 22000 Freq. (No Fatigue) Fatigue Related Non- Fatigue Related 100 20000 Monday Tuesday Wednesday Thursday Friday Saturday Sunday Year 42 percent of fatigue-related crashes occur on the weekend. The number of non-fatigue crashes decline on the weekends. The weekend fatigue related crash involves more alcohol, and risk-taking behavior.
Findings Logistic Regression Model of Fatigued Drivers Involved in All Vehicle Crashes Variable Parameter s.e. Prob. X- Square Odds Ratio Intercept -4.1602 0.0534 <.0001. Early Morning (12:00 AM to 6 AM) 2.1853 0.0296 <.0001 8.893 Driver Age (15 to 29) 0.4150 0.0297 <.0001 1.514 Rural Area 0.7616 0.0327 <.0001 2.142 Highway 0.2425 0.0295 <.0001 1.274 Male 0.2276 0.0333 <.0001 1.256 Serious or Fatal Injury 0.5391 0.0778 <.0001 1.714 Weekends 0.3995 0.0298 <.0001 1.491 Level Grade -0.2254 0.0318 <.0001 0.798 Straight Alignment -0.2802 0.0387 <.0001 0.756 Speeding -0.8316 0.0505 <.0001 0.435 Likelihood Ratio X 2 8086.4880 (with 10 Df, p=0.0001) 8.8 times more likely to occur in the hours 12AM to 6AM. 1.4 times more likely to occur on weekends. 2.1 times more likely to occur on rural roads. 1.2 times more likely to occur on highways.
Injury Outcomes 14 Time of Day Effects for Fatigued-Related Crashes Required EMS Transport Vs. Non-EMS Transport Frequency 12 10 8 6 4 2 0 Require EMS No EMS -2 1AM 3AM 5AM 7AM 9AM 11AM 1PM 3PM Time of Day 5PM 7PM 9PM 11PM Again two peak periods exist for fatigue-related crashes
Fatigue Related Non-Fatigue Related Z-test Frequency Percent Frequency Percent Driver Injury (EMS Score) DOA 3 1.7 13 0.8 1.30 Extremely Critical 0 0.0 19 1.1-1.41 Critical 4 2.3 24 1.4 0.91 Severe 40 22.7 428 25.1-0.68 Minor 126 71.6 1184 69.3 0.62 None 3 1.7 40 2.3-0.54 Total 176 100.0 1668 100.0 Required EMS Transport Yes 176 23.8 1708 8.5 14.33 No 562 76.2 18465 91.5-14.33 Total 738 100.0 20173 100.0 23.8 percent of the fatigue-related crashes required EMS Transport, as oppose to 8.5 percent for nonfatigue related crashes.
Single Vehicle Fatigue-Related Crashes by Crash Type Required EMS Transport EMS as a Crash Type Frequency % Frequency % of Crashes Guardrail 43 13.40 8 18.60 Culvert 4 1.25 3 75.00 Bridge / Overpass 1 0.31 1 100.00 Underpass/Bridge Support 1 0.31 0 0.00 Building 0 0.00 0 0.00 Island/Median/Curb 40 12.46 7 17.50 Embankment/Retaining Wall 60 18.69 11 18.33 Fence 11 3.43 1 9.09 Utility Pole 88 27.41 26 29.55 Traffic Signal/Stop Sign 26 8.10 3 11.54 Tree 35 10.90 6 17.14 Hydrant 12 3.74 0 0.00 Animal 0 0.00 0 0.00 Total 321 100.00 66 27 Percent of the single vehicle fatigue-related crashes involved collisions with utility poles, and 30 percent of those required EMS Transport.
Hospital and Claims Analysis of At Fault Drivers and Fatigue-Related Crashes At Fault Driver Hospital Stay Days Hospital Billing ($) Total Claim ($) Mean SD Mean SD Mean SD Fatigue-Related 2.5 24.6 1153.22 11611.30 1326.85 11699.50 Non-Fatigue-Related 1.4 17.5 623.12 5821.40 783.11 6026.40 The mean hospital stay days for at fault fatigue-related drivers was 2.5 days as opposed to non-fatigue which was 1.4 hospital stay days. The mean dollar in hospital billing for at fault fatigue-related drivers was 1.85 times greater than for non-fatigue-related drivers. The mean dollar in insurance claims for at fault fatigue-related drivers was 1.69 times greater than for non-fatigue-related drivers.
Fatigue Related CBD Fatigue-related crashes have greater spatial dispersion than non-fatigue crashes.
Single Vehicle Fatigue-Related Crashes 3D mapping provides another means of visualizing & locating areas to implement corrective measures Location of Greatest Problem
Single Vehicle Fatigue Related Crashes on Highways (12:00 AM to 6:00 AM) Use statistical model to identify significant areas of concern. Use GIS to map locations of concern. Focus measures at these locations to improve conditions that cause fatigue related crashes.
Education: Countermeasures Raise public awareness about the risks of driving while fatigued. Focus education on the young male population. Educate shift workers on the problems associated with fatigue related crashes. Educate those with active lifestyles that restricts sleep. Engineering: Installation of rumble strips in problem areas. Increase lighting in areas of high fatigue related crashes. In the vehicle alerting devices.
Conclusion Understanding both the spatial and temporal factors associated with fatigue-related crashes can aid in the prevention of these events. Using Crash data and Statistical / Spatial Analysis Techniques provides a better understanding of the fatigue-related crash.