Predicting road crashes from a mathematical model of alertness regulation—The Sleep/Wake Predictor

Sleepiness is related to factors such as the time of day, the time since awakening and the duration of prior sleep. The present study investigated whether actual road crashes could be predicted from a mathematical model based on these three factors (the Sleep/Wake Predictor—SWP). Data were derived from a population-based case-control study of serious injury crashes. Data on accident time (or control sampling time) and start and end of prior sleep were entered into the model (blind). The predicted sleepiness values were used in logistic regressions. The results showed a highly significant odds ratio (OR) = 1.72 (confidence interval = 1.41–2.09) for each incremental step of sleepiness on the output sleepiness scale (nine steps). There was also a significant interaction with blood alcohol level, showing high OR values for high sleepiness levels and alcohol levels above 50 mg% (0.05 g/dl). It was concluded that the model is a good predictor of road crashes beyond that of alcohol level, and that interaction between the two carries a very high risk.