There were some questions generated from my last post on this, so I wanted to provide some clarification about why I am skeptical regarding biomechanical studies (dummies, simulation, and modeling) to determine the “cause and effect” relations of triad symptoms.
To examine more closely the deficiencies with biomechanical studies, modeling, and simulation, let’s look at a particular example. Let’s take the example of one of the “triad” symptoms, retinal hemorrhage. What is solely responsible for retinal hemorrhage from head impact is the acceleration experienced by the retina and fluids in contact with the retina, not the skull. Skull impact acceleration forces will be transmitted (and modified) through various tissues and fluids to the retina, and what the retina experiences will be different than what the skull experiences. However, as far as I know, we have no way to directly measure acceleration of the retina, so we have to settle for measuring acceleration of a simulation or model of the skull, and infer how that gets transmitted to the retina. Therein lies part of the problem, and a major reason why modeling or simulating the situation is so difficult. And in fact, measuring skull acceleration is no easy task in itself, because it will depend upon the “flexibility” or “compliance” of any particular skull. I would guess that an infant skull with the fontinel still open would be more compliant than one with the fontinel closed, and certainly more compliant than a fully formed adult skull. (Side note. Some might say, “Use monkeys or cadavers for these studies.” However, the physiology of animal brains and skulls is different, and cadavers have no blood pressure, muscle tone, or reflex reactions.)
For a long time, it was believed, and still is by some, that “short falls” could not cause triad symptoms, because the skull impact acceleration could not exceed the “threshold for head injury” that had been determined by studies. However, none of the “studies” ever stated that once the acceleration threshold had been reached, the probability for triad symptoms would be “x”%. Is the probability 100% at threshold? I doubt it, but nobody knows.
An example of what really needs to be known is represented by this question. For a central frontal forehead impact to an infant, what is the range of accelerations (measured in ft/sec2) of the skull at the point of impact that will cause retinal hemorrhage with the following probabilities? To answer this, we would have to fill in the following chart:
What’s most notable here is that essentially all the entries are question marks. Now, I can go ahead and fill in the 0’s in the first column, because I can safely estimate that zero acceleration will produce retinal hemorrhage zero percent of the time. But there will still be an upper limit of acceleration that can be experienced by the skull that will also produce hemorrhage zero percent of the time. What is that upper limit? I defy anyone to tell us. The same goes for all the question marks in the chart.
The very same principles of physics and statistics also apply to the causation of subdural hematoma and diffuse edema.
Biomechanical studies certainly produce some kind of result that can inferentially guide thinking, but they fail the same question that every forensic discipline (with the exception of DNA) fails – “Show me the data from which I can compute the probability of occurrence.” This question is also the root of the principle cited in Daubert vs. Merrell Dow stating “no known error rate”.
While biomechanical studies certainly produce “a result”, that result cannot be statistically validated, and my fear is that a study result will be used to decide someone’s fate in court, and nobody will actually know if the “result” is correct. We’re still left with somebody’s “best guess” – Oh … or should I say “expert opinion”.
Phil Locke
Wrongful Convictions Blog 