Fatigue, alcohol, and serious road crashes in France: factorial study of national data

BMJ 2001; 322 doi: (Published 07 April 2001) Cite this as: BMJ 2001;322:829
  1. P Philip, consultant (pierrephilip{at},
  2. F Vervialle, statisticianb,
  3. P Le Breton, statisticianb,
  4. J Taillard, research assistanta,
  5. J A Horne, professorc
  1. a Clinique du Sommeil, Centre Hospitalier Universitaire, Bordeaux 33076, France
  2. b Service d'Études Techniques des Routes et Autoroutes, Bagneux, France
  3. c Sleep Research Laboratory, Loughborough University, Loughborough LE11 3TU
  1. Correspondence to: P Philip
  • Accepted 21 December 2000

Editorial by Feyer

France has a high rate of road traffic crashes.1 Although driver fatigue may be an important factor,24 it has not been investigated in France, and no comparisons have been made with alcohol related crashes.5 We investigated the role of fatigue in serious road crashes using the French national database.1

Methods and results

We obtained data from the French Ministry of Transport on all road crashes during 1994–8 (640 670) in which at least one person was severely injured (confirmed by paramedics) or died. Crashes were attended by police officers, who completed a standard ministry questionnaire that covered time of incident, location, road and weather conditions, vehicles involved, mechanical defects, health of driver, and alcohol consumption as well as giving summaries of interviews and probable causes.

As crashes related to fatigue can be difficult to identify, we applied the strict criteria of Horne and Reyner to eliminate many of the confounding factors.2 We assessed only single vehicle crashes that occurred during good weather and road conditions on roads unrestricted by junctions. This excluded most urban road crashes (comprising most crashes), crashes involving pedestrians, and those in which the driver reported taking medication or was suspected to have used illicit drugs. This left 67 671 crashes for analysis.

We identified four categories of crash:

Alcohol related—Blood alcohol concentration >100 mg ethanol/1 l blood (breathalyser or blood analysis).

Fatigue related—Driver could have avoided crash but no avoidance taken (no braking or swerving), with blood alcohol concentration <100 mg/l.

Alcohol and fatigue related—Fatigue related crash with driver's blood alcohol concentration >100 mg/l.

No alcohol or fatigue—No fatigue; blood alcohol concentration<100 mg/l.

About 10% (6770) of the crashes were related to fatigue and 23% (15 670) to alcohol (table 1). These were subdivided into three periods: day (0700–1959), evening (2000–2359), and early morning (0000–0659). Alcohol related crashes were more likely to be fatal during the evening and early morning compared with the daytime (Wald's χ2=4.88, P=0.02 for evening, χ2=18.04, P< 0.01 for early morning), whereas fatigue related crashes were more likely to be fatal during the day than the early morning (χ2= 5.37, P=0.02).

View this table:

For the whole 24 hours, and compared with all other non-alcohol related crashes, the relative risk of death in crashes related to fatigue was 1.65 (95% confidence interval 1.49 to 1.82, χ2=97.09, P=0.001). The risk of severe injuries was 1.5 (1.4 to 1.6, χ2= 226.15, P=0.001).

For alcohol related crashes, the relative risk was 4.2 (3.9 to 4.4, χ2= 2517, P=0.001) for death and 1.9 (1.8 to 2.0, χ2=1057, P=0.001) for severe injuries. For alcohol and fatigue combined, the risk of death was 6.8 (5.7 to 8.0, χ2 =678, P=0.001) and risk of severe injuries 2.6 (2.2 to 3.0, χ2 =141, P=0.001).

We then ran a multivariate analysis on daytime crashes with death as the dependent variable and fatigue, physical handicap, distraction (driver alert but attending elsewhere), and weekend (versus weekday) as independent variables. For non-alcohol related crashes resulting in death, the significant factors were fatigue (odds ratio=1.57, 95% confidence interval 1.42 to 1.74, P<0.001), distraction (0.70, 0.61 to 0.82, P<0.001), and weekends (1.14, 1.05 to 1.23, P<0.001). For alcohol related crashes resulting in death, only fatigue was significant (1.41, 1.15 to 1.73, P<0.001).


We found that fatigue, especially when combined with alcohol, presents a particularly high risk of road crashes resulting in death or serious injury. This has been largely unrecognised in France and elsewhere. There was also a strong relation between time of day and cause of crash, with many alcohol related crashes occurring at night. However, it is likely that police officers will attribute such crashes only to alcohol, even when fatigue is present.


We thank Annie Canel and Michel Labrousse, Service d'Études Techniques des Routes et Autoroutes, for help in organising and Bernard Bioulac, department of neurophysiology, Bordeaux University Hospital, for his support.

The paper was written jointly by PP, FV, JT, and JAH. PLB is the guarantor.


  • Funding This study was sponsored by the Service d'Études Techniques des Routes et Autoroutes.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract