Fatal falls from a height: the use of mathematical models to estimate the height of fall from the injuries sustained.

The authors undertook a review of fatal falls from a height, that occurred in 1991-92 in Singapore, with the objective of constructing mathematical models relating the height of fall to the injuries sustained. The 603 cases studied showed a mean age of 41.4 years with a male to female ratio of approximately 2:1. A sub-sample of 416 (69%) of these subjects had fallen from known heights (mean, 26.9 m; range, 3-69.6 m) and were studied in further detail. Bivariate analysis of this group showed that their injury severity score (ISS) was significantly correlated with the height of fall (H) and age (P < 0.01; r = 0.412 and 0.187, respectively). As the ISS is not strictly a continuous variable and varied markedly with H, it was categorised into bands (ISSB) before being subjected to further analysis. Regression modelling to adjust for mutual confounding showed that both height of fall and age were significant independent determinants of the ISSB (P < 0.0001). A model with H as the dependent variable was then constructed to relate the height of fall to ISSB and other statistically significant indicators of the extent and the severity of the injuries sustained. A second model with bands of height (HB) as the dependent variable was similarly constructed to assess the effect of banding both height and ISS. Our findings suggested that the height of fall was significantly associated with age, ISS and the extent of injury (mostly AIS > or = 3), and confirmed the usefulness of these models for investigative purposes. Statistical models could be designed and used to assess any apparent discrepancy between injury severity as determined at autopsy and the suspected/alleged height of fall.