Trimodal temporal distribution of fatal trauma—Fact or fiction?

Injury, Int. J. Care Injured (2008) 39, 960—966

www.elsevier.com/locate/injury

LETTER TO THE EDITOR Trimodal temporal distribution of fatal trauma–— Fact or fiction? Dear Sir, We would like to comment on the released reports in INJURY on fatal trauma epidemiology, temporal death distribution, and modern trends concerning polytrauma, as highlighted in the editorial by Civil.5 We have recently addressed several of these issues as well,15 and support some of the thoughts expressed, while coming to somewhat different conclusions in other regards. First, the epidemiology of fatal and non-fatal trauma appears to be changing, in particular with respect to a definitively growing elderly population. Aldrian et al.1 found that the age increased by about 0.75 years per year over the decade investigated. Similarly, we found that the mean age of the trauma patients who died during a 9-year period increased significantly by almost a decade, from a mean of 42  24 years to about 51  25 years ( p = 0.04).15 While age is increasing, the number of trauma patients who die from multi-organ failure (MOF) is decreasing.9 MOF is reported in 20% numbers from a quarter of a century ago.2 Also, we noted that age significantly increased (r = 0.43) and ISS decreased (r = 0.52) with longer time from injury to death ( p < 0.001). Those dying of MOF were significantly older, and less severely injured, compared to deaths from head-injuries and exsanguination.15 Second, a trimodal temporal death distribution has been questioned by several authors6,11,17 since the seminal report by Trunkey.16 However, it has been recognised that the temporal distribution is dependant on the circumstances under which it is investigated.3,10,15 This includes not only the place, time and location, but also registry-criteria,4 and definitions of the time-sequences chosen for analysis. While we concur with Pang et al.11 that trimodality may no longer be a valid concept per se in the 21st century, we have shown that the temporal death

distribution is in fact model-dependent and unimodal, bimodal, or trimodal patterns may be seen.15 As such, the ‘‘trimodal temporal death distribution’’ may be a fiction. Clearly, when investigating trauma deaths they obviously occur in a strictly inverse exponential relation from the time of injury until death actually ensues. This can be seen in a standard Kaplan—Meier curve in a timeto-event relationship.15 However, demonstrating this on a reasonable scaled timeline in any publication or presentation is difficult as the deaths occur in the time range of seconds, minutes, hours, weeks and months after primary injury impact. Thus, for the sake of clarity, time sequences (i.e.