Whiplash injuries remain a barely understood phenomenon. The economic damage caused by whiplash amounts to some 10 billion Euros a year in Europe 1 and 29 billion US Dollars a year in the USA 2. As whiplash occurs as a result of motor vehicle accidents (MVAs), questions inevitably arise regarding who is liable for these costs.

Biomechanical considerations have been based on the assumption that damage to a given material only occurs when the energy that acts on this material is high enough. Thus, energy doses below a defined threshold have been considered harmless 34. In this context, the parameter delta v (ΔV), which describes the velocity change of a motor vehicle during a collision with another vehicle, has become a widely accepted criterion for the energy that acts on the vehicle during a collision 5.

In numerous sled or car crash-test studies, volunteers were subjected to acceleration forces in order to define a threshold below which a cervical spine injury could be excluded 6789101112131415. The results of these studies are rather inconclusive and sometimes contradictory. Thus the scientific community has not yet reached consensus regarding the threshold value for cervical spine injuries after whiplash. Nonetheless, ΔV threshold values were adopted very early in the history of insurance law as a criterion to accept or deny the claim settlement for whiplash-associated disorders (WADs) 16.

Up until now, all volunteer crash-test studies precisely defined the subject's sitting position. While waiting for the collision, the subjects maintained an upright body and head position, with an optimally adjusted headrest. It is obvious that the real-life sitting position in traffic may significantly differ from this laboratory position in one or several points. Furthermore, an increased risk of injury has been observed for various factors such as the seat and headrest settings 1117181920, the distance between head and headrest 212223, the head rotation, and the collision type 24. The inherent variability of these factors makes it unclear how easily the results from laboratory crash tests can be transferred to real-life accident situations. In order to elucidate these issues, this study analyzes the correlation between ΔV and cervical spine injuries in real-life accidents and questions whether ΔV is a valid predictor for cervical spine injuries following whiplash.

The study included 57 patients after a car collision. The patients were recruited either by an engineer's office for vehicle damage assessment and claims adjustment (n = 46) or by the first consultation of an emergency room (n = 11). We obtained the approval of the local independent ethics board and all patients gave their written informed consent to participate in the study.

We enrolled 57 individuals (25 males and 32 females) in the study; these individuals had been the occupants of 51 cars (Table 2). The median age was 33 (range 3 to 90 years) for the males and 30 (range 18 to 59 years) for the females.

A total of 25 (44%) patients complained about pain in the neck. VAS pain scores of 7 to 96 (median = 71) were reported. Fifteen patients reported an immediate onset of pain, four individuals reported a time to onset of minutes to hours, and four patients reported a time to onset of hours to one day. The Pearson's correlation coefficient of r = 0.55 indicated a moderate correlation between the pain that was subjectively reported and ΔV (Fig. 1).

A total of 25 (44%) patients complained pain related neck disability. NDI scores of 4 to 49 (median = 24) were reported. The Pearson's correlation coefficient of r = 0.46 indicated no relevant correlation between the NDI and ΔV (Fig. 2).

Thirty-two patients (56%) were classified as QTF grade 0. Eight patients (14%) presented with QTF grade I, 6 patients (10%) with QTF grade II, and 11 patients (19%) with QTF grade IV. No QTF grade III injuries were scored. The Spearman's correlation coefficient of r = 0.45 indicated no relevant correlation between ΔV and the QTF grade of cervical spine injury (Fig. 3).

Rear-end collision (n = 21, 36%) was the most frequent collision type, followed by side collisions (n = 19, 33%) and front collisions (n = 13, 23%); there were also three multiple collisions and one rollover. For the rear-end collisions, individuals with and without cervical spine injuries were found in a ΔV range between 9 km/h and 37 km/h. This range was 15 km/h to 28 km/h for frontal collisions and 9 km/h to 36 km/h for side collisions. Within these ranges, the percentage of false-positive and false-negative results varied greatly, depending on the predefined cut-off values (Tables 3, 4 and 5). Therefore, for all collision types it was impossible to define a ΔV value that excluded the occurrence of cervical spine injury with acceptable sensitivity while simultaneously predicting the occurrence of cervical spine injury with acceptable specificity.

This study provides evidence that, in real-life accidents, cervical spine injuries may occur at low ΔV values, while it is possible to escape unscathed from collisions with high ΔV values. In particular, the correlation between ΔV and the occurrence of WADs was very low for any of the collision types. Therefore it is impossible to make meaningful statements about the existence of WAD based solely on assessment of the ΔV value. This finding might be of importance for the surgeon's assessment and patient's safety after a car accident. Diagnostic and therapeutic management should not be based solely on information related to trauma impact.

The results of the present study support the findings of numerous sled and car-crash experiments. In those experiments, neck problems were noted after rear-end collisions with ΔVs as low as 7 km/h 14293031. In four other studies 8323334, neck problems occurred at a ΔV < 10 km/h. The neck problems were defined as QTF grade I and QTF grade II, persisting from hours to several weeks in all studies. In contrast, four studies reported rear-end collisions with ΔV values of 13.1 km/h to 50 km/h where the occupants escaped without any signs of injury 493536. In other crash-test studies, frontal impacts at ΔV less than 12 km/h caused no injuries 34. However, different findings were obtained in our study and in a study that performed a retrospective analysis of 24 real-life frontal collisions 37. In that study, 18 of the 24 subjects were classified as QTF grade II. It is noteworthy that 8 of these had neck problems for more than one year. The ΔVs in these cases ranged from 3 km/h to 23 km/h. The authors also reported that one subject suffered a prolapsed disk at C5/6 at a ΔV of 11–15 km/h. The occupant had not been wearing his seat belt and the airbag had deployed. He also had a frontal laceration as a sign of direct head impact. It was assumed that these factors caused the structural injury of the cervical spine at a low ΔV. The occurrence of structural injuries at ΔV values of less than 20 km/h had been considered improbable in expert discussions. However, we also observed a luxation fracture at C5/6 resulting from a frontal collision at a ΔV of 15 km/h and a facet joint fracture at C4 due to a side collision at a ΔV of 10 km/h (Table 1). Both occupants had been wearing their seat belts, there had been no head contact, and the airbag had not deployed. In both cases, it is unclear which factors, either alone or in combination, were responsible for these structural injuries at considerably low ΔV. In accordance with other studies mentioned below, these results are indicative that multiple factors may influence the risk of injury in each individual case. Due to the additive effects of various protective factors, high-energy impacts may be absorbed without injury, while the additive effects of unfavorable factors could explain injuries sustained in low-energy impacts. Some factors have been described to influence the risk of injury, such as sex 3839, head position 40, sitting position 2441, distance between head and headrest 212223 and seat construction 11171819. The duration of the crash pulse is also thought to significantly contribute to the risk of cervical spine injury. These authors stated that an earlier acceleration peak during deformation of the colliding cars was correlated with a higher probability of cervical spine injury 24. However, it remains unclear to what extent each one of these factors influences the risk of cervical spine injury.

The current data exclude the assumption of a linear correlation between ΔV and the risk of suffering a whiplash injury. It is tempting to speculate that the development of a cervical spine injury after whiplash is more like a complex system such as those described in chaos theory 42. Complex systems cannot be simplified into linear correlations. Even small variations of the initial conditions can affect the end result so that it is no longer predictable, such as in the case of the "butterfly effect": the flapping of a butterfly's wings can ultimately result in a different weather pattern 43. Taken together, it can be concluded that ΔV is an irrelevant predictive value for cervical spine injury after a MVA. Nevertheless further studies will be necessary to evaluate the development of pain chronification in dependence of the ΔV to investigate its possible predictive value as "long-term" parameter.

The ΔV value as measured in the trauma impact does not represent a conclusive predictor for cervical spine injury in real-life motor vehicle accidents. This could be important for surgeons and patients in their medicolegal assessment of WADs.

The authors declare that they have no competing interests.

ME drafted the manuscript and performed the medical examination. MK participated in the study coordination and helped in the medical examination. MHL helped to draft the manuscript. EH participated in the study design and its coordination. CD performed the statistical analysis and helped to draft the manuscript. All authors read and approved the final manuscript.