Objective: Falls from height cause significant mortality in the urban environment, but reliable prognostic factors have not been identified. Even the intuitive relation between the distance fallen and mortality rate has been questioned. Our objective was to determine factors predictive of increased mortality rate in victims of falls from height.
Design: Clinical observational study, retrospective for January 1998 to May 1999 and prospective from June 1999 to September 2000.
Setting: The study population was drawn from Seine-Saint-Denis, an urban region near Paris with 1.3 million inhabitants treated by a French out-of-hospital medical emergencies unit.
Patients: Patients were victims of falls from height >3 m, age >12 yrs. Study entry was performed on the scene by an emergency physician from the medical emergencies unit.
Interventions: None.
Measurements and main results: Studied data included age, gender, circumstances of fall, height of fall, nature of the impact surface (soft or hard), transient impact preceding final impact, and part of the body touching the ground first. The primary end point was mortality. The study included 287 patients, 116 (40%) during the retrospective phase and 171 (60%) during the prospective phase. Ninety-seven patients (34%) ultimately died. In multivariate analysis, age (mean, 41.6 +/- 16.6 yrs in patients who died vs. 34.9 +/- 14.9 in survivors; odds ratio, 1.05; p < .0005); height of fall (median, 5.0; 3.8-8.0 vs. 2.0; 1.2-3.0 floors; odds ratio, 1.24; p < .0001); nature of the impact surface (hard in 39% vs. soft in 22%; odds ratio, 2.7; p < .05); and head, anterior, and lateral body surfaces touching the ground first (with respectively mortality rates of 44%, odds ratio, 16.7, p = .0001; 57%, odds ratio, 10.6, p < 0.005; 32%, odds ratio, 11.1, p < .001) were independently correlated with the final mortality rate.
Conclusions: Patient age, height of fall, impact surface nature, and body part first touching the ground are independent prognostic factors in victims of falls from height.