Whiplash / Neck Pain

Crash Severity and Injury Risk: A New Study Provides the Definitive Answer

Arthur Croft, DC, MS, MPH, FACO

Can one gauge the risk for injury on the basis of property damage in a motor vehicle collision (MVC)? This question lies at the very heart of some of the most hotly debated legal jousting. Fairly or otherwise, it often forms the crux of a defense strategy for denying responsibility. It has gradually become the perfect foil; a conclusion that plays easily and seductively on the intuitive logic of triers of fact. But the tactic has a more tenebrous side.

In the mid-1990s, the insurance industry developed what is now called the "minor impact, soft tissue (MIST) segmentation" policy (a term arguably intended to marginalize the common whiplash injury). Many insurers have gradually adopted this policy. It is based on the assumed relationship between property damage and injury risk, with a cutoff point set at $1,000. Its application is simple: A person claiming to be injured in a collision in which the property damage of his or her vehicle falls below this threshold is deemed to be misrepresenting the facts, and the case is "segmented" to another department. In plainer English, it is sent to the company's special investigation unit (SIU). This is the unit that investigates fraud. Thus, in addition to the fact that personal injury lawsuits often turn on the simple question of property damage, so too do fraud investigations. Some might argue that, unless it could be demonstrated that this practice was based on some tangible and valid evidence, this practice might constitute a form of bad faith. Can a fraud accusation and denial of services or compensation be reasonably, responsibly or ethically based on a purely administrative decision and arbitrary cutoff point?

This field of personal injury litigation actually relies on a relatively well-established series of arguments, and I have explored many of them in this column over the years. This editorial will focus on a single question: Is there any valid way to predict injury risk, injury severity, or the likelihood of long-term symptoms by looking at the amount of property damage? And thus, is the MIST segmentation policy itself valid? This is a critical question to answer definitively, because when the chips are down and the defense has played its best hand, they can just about always rely on the tried and true "no crash - no cash" defense. This tactic typically is accompanied by the display of a poster-sized photo of the vehicle showing little or no damage.

The scope and magnitude of this tactic are not subtle. With an annual comprehensive cost of $43 billion in the United States alone, whiplash injury has become one of the most expensive health care problems Americans face. Of this amount, insurers pay out an estimated $19 billion. From the epidemiological community emerges the fact that zero-property-damage crashes may account for up to 40 percent or more of whiplash claims. Thus, if an insurer can reliably convince (say 70 percent of the time) the triers of fact that injury is simply not possible when there is little or no property damage, it can potentially write off a substantial proportion of its share of this money - no mean feat in the world of big business.

And so, the central question: Is there or is there not a correlation? At the risk of sounding like a lawyer, the answer is yes and no. Consider a range of crashes from 0.5 mph to 150 mph. In the lowest velocity range, it is unlikely that anyone would be injured. And a low-velocity crash is likewise unlikely to result in any structural damage to the vehicle. At the other end of the extreme, property damage will be very severe and the likelihood of survival low. Thus, considering this entire range, there is clearly a relationship or correlation between property damage and occupant risk. But now let's consider a narrower range of crash velocities, say from 2 mph to 25 mph. This is, in fact, the range in which more than 85 percent of all litigation occurs.

Here, the relationship between property damage and injury risk is nonlinear. In the lower velocity range, vehicles are relatively stiff, the collisions more elastic. In engineering parlance, there is a relatively higher coefficient of restitution in lower velocity crashes. Then, as crash velocity increases, so too does the amount of residual crush of the vehicle. Interestingly, though, the actual risk to the occupants decreases as this occurs. This has been demonstrated in several studies. So, over this range of crash velocities, there is no good correlation between injury risk, injury severity, or long-term outcome and property damage. This is, of course, an important distinction to understand.

For many years, clinical and, more importantly, epidemiological evidence have been accumulating that rather consistently undermine this crash/risk correlation fallacy, but, other than for the small community of my loyal readers, students, and fellow crash research geeks, this evidence has remained fairly obscure (if not occult) for most. Moreover, the nagging questions remained: Has the question been sufficiently studied? Epidemiological evidence is, after all, one thing, but the best way to answer this question would be to conduct a formal study in which patient outcome is the dependent variable and crash damage is held as the independent variable.

I decided to conduct a meta-analysis of this very question in hopes of formalizing our current level of confidence in either assuming or rejecting a relationship between crash damage and risk.1 This is an important analysis because it is the first of its kind and, therefore, will stand as the definitive work, at least until other outcome studies become available. I encourage readers to download the free paper from the Web site and distribute it to other interested parties. The study's conclusions were as follows:

In our comprehensive literature review, we found only four papers that compared property damage resulting from low-velocity motor vehicle crashes to any of three injury categories (injury risk, injury severity, or duration of symptoms) which were conducted using acceptable scientific rigor and statistical assessment of the results. Two of the papers reviewed in this analysis followed the same group of 32 subjects. Another study followed only 26 subjects. In the largest dataset (n = 5,083 claims), the authors did not interview or examine the subjects. They reported injury claims and, in cases in which the records allowed them to determine it, symptoms exceeding six months. These claims apparently were not all verified and no information was available regarding injury severity or long-term symptoms outside of retrospective claims review of insurance files by nonphysician claims persons.

Damage assessment was made on the basis of repair costs and did show a positive correlation between increasing costs and property damage. However, it was not clear whether a continuous relationship existed across all crash severity categories, since differences between the two lowest ranges were marginal and since one comparison group was omitted from analysis without explanation. Selection bias in the other two studies tended to exclude persons who were not injured. As a result, none of the studies can be used to develop risk tables regarding property damage and injury risk, injury severity, or injury duration or persistence. Our best evidence synthesis demonstrates that while there appears to be some relationship between property damage and injury risk or severity, this may only be true when considering a wider property damage range (e.g., minor vs. severe or moderate vs. severe), but this metric neither holds for males nor correlates significantly with long-term symptoms for persons of either sex.

A substantial number of injuries are reported in crashes of severities that are unlikely to result in significant property damage. Thus, property damage is not a valid predictor of acute injury risk or of symptom duration. Other factors, such as head restraint geometry, awareness of the impending crash, sex, and prior injury are likely to impose competitive or stronger outcome effects, particularly regarding long-term outcome. Based upon our best-evidence synthesis, the level of vehicle property damage appears to be an invalid construct for injury presence, severity or duration. The MIST protocol for prediction of injury does not appear to be valid.

In a nutshell, the available research does not answer the question in the most satisfying way, but existing research tells us that crash damage alone is not a reliable index of injury risk, severity or outcome in low-velocity crashes. Human factors, which are quite variable, collectively can have a profound effect on outcome and often can be more important than crash metrics in determining risk. Crash stiffness also varies considerably across the low-velocity crash continuum. With so many factors potentially affecting outcome, it is unlikely that any single factor will have sufficient predictive ability by itself. Instead, physicians should consider all potential risk factors and weigh them according to their relative importance.

When displaying posters of photos of vehicles with minimal or no property damage, lawyers often ask the jury to "use your common sense." But what common sense do non-engineers have about low-velocity crashes? Here is a piece of common sense most jurors might actually have. It is not at all uncommon to hear of a very severe crash in which one occupant was killed while others walked away relatively unscathed. Common sense would thus argue that crash metrics alone did not allow the prediction of the outcome, since it was the same for all occupants.

Reference

  1. Croft, AC and MD Freeman, Correlating crash severity with injury risk, injury severity, and long-term symptoms in low velocity motor vehicle collisions. Medical Science Monitor, October 2005;11(10):RA316-321. The full-text article can be downloaded for free on the Medical Science Monitor's Web site: www.medscimonit.com/pub/vol_11/no_10/8008.pdf.
March 2006
print pdf