Consider the following undisputed findings about a hypothetical substance and a hypothetical disease published in a wide variety of respected, peer-reviewed scientific journals:

  • There's a very strong correlation between the presence of the substance and the subsequent diagnosis of the disease, so much so that the presence of the substance more than doubles the risk of being diagnosed with the disease.
  • The more of the substance to which you're subjected, the greater the risk you'll be diagnosed with the disease.
  • The findings above have been consistently demonstrated across numerous studies.
  • Some people carry a gene that causes them to actually make the substance and lots of it; they have a significantly increased risk of developing the disease.
  • Several biological models have been developed and published that plausibly demonstrate how and why the substance causes the disease.
  • Animal experiments also demonstrate that the substance and the disease go hand-in-hand, and animals injected with the substance develop similar pathologies.

A strong association, dose-response, consistency, temporality and biological plausibility: What more would you possibly need to make a judgment about so-called "general causation" in toxic tort case? Nothing more, at least in many courts these days, which means that the substance in our hypothetical, which, as you guessed, was not so hypothetical, could be found to be the legal cause of the equally not-so-hypothetical disease.

So what are they? Amyloid-beta and Alzheimer's disease, respectively; recent announcements blowing gaping holes in the decades-old hypothesis that amyloid-beta causes Alzheimer's ought to give courts pause before allowing even the best untested hypothesis to be passed off as "scientific knowledge."

The amyloid hypothesis, or amyloid cascade hypothesis, certainly was a sound hypothesis. Since 1907, when Alois Alzheimer noticed the amyloid plaques in the brains of the victims of what we now call Alzheimer's, the two have been inextricably causally linked in the minds of many, if not most, researchers, and, over the years, more and more observations seemed only to confirm the connection.

Eventually, there was so little doubt that amyloid causes Alzheimer's that hundreds of millions, if not billions, of dollars were spent to develop drugs to combat it. Some of the money was spent elucidating that pathway whereby amyloid is made in the body so that drugs could be designed to stop its production. Some of it was spent learning how amyloid molecules link together and fold up into tangles and pile up into plaques so drugs could be designed to prevent the folding or to untangle the tangles. Still, more money was spent learning how amyloid was cleared, or removed, from the body so that drugs could be made to wash it away. Now the drugs have been tested.

The drugs worked as designed, and the amyloid was reduced in volume or more neatly folded or scrubbed away. Unfortunately, the victims weren't helped, and, in fact, the symptoms of many were made markedly worse.

Why? Because correlation isn't causation even if you do have a good narrative to go along with your coincidental association. Imagine, by way of example, that amyloid plaque deposition is a defense mechanism that guards against or ameliorates the effects of the assault of some unknown agent on neurons, which assault ultimately leads to Alzheimer's. If that were the case, it would be no more surprising to find amyloid near neurons than platelets near wounds, and, thus, you'd no more blame amyloid for the observed neurological damage than you'd blame platelets for road rash (get well soon, Dr. Rock, we're glad you were wearing your helmet!).

And guess what? Though the fortune wagered on the amyloid hypothesis being true surely attests to the degree of belief held by those making the wager one of the great things about skepticism is that not every scientist bought in to the amyloid hypothesis, and they're out there now saying, "See? I told you so," and finally getting people to look at studies show nuns with tons of amyloid plaques and yet no Alzheimer's, mice that crank out the amyloid yet suffer no neurodegeneration and intriguing hints that amyloid may play a role in combating marauding microbes.

There's nothing wrong with how the amyloid hypothesis is playing out. It's how science works. It's finding a way through a vast maze, making your best guess, admitting when you've hit a wall, updating your beliefs, making your way back to the last fork that looks promising and doing it all over again.

Of course, if you bet on an untested hypothesis, and it doesn't pan out, you lose your time and money. If you're a scientist, engineer or investor, that is. On the other hand, in civil courts across the country, vast fortunes are made on the basis of untested hypotheses, and the great thing is that even if the hypothesis is tested and fails, you don't have to give the money back.

Why some courts don't demand that hypotheses be tested before bestowing upon them the certificate of scientific knowledge remains a mystery. Both the Bradford Hill causal criteria and their predecessor Koch's postulates include the not very surprising requirement that before causation is assigned, you ought to see if preventing exposure prevents disease.

Indeed, it goes right to the heart of "but-for," or counterfactual, causation — if the exposure hadn't occurred, the disease wouldn't have occurred. Anyway, that's a mystery for another day. For today, please ponder the following question, and let us know your thoughts if you're so inclined.

In those courts, where something like the amyloid hypothesis would easily have passed the test to be certified as scientific knowledge, how would one go about getting it "decertified," as it were? Would it be a sort of un-Frye test whereby you'd have to line up scientists to testify that the "knowledge" was now generally unaccepted? Or would it be a Daubert affair, and, if it was, how would it work in a court in which a hypothesis need not be tested to be considered scientific knowledge? Think about it.

How could the subsequent failing of a test that wasn't necessary to deem a hypothesis scientific knowledge in the first place logically alter the status of the hypothesis? See our point? Courts that lower the standard for what constitutes scientific knowledge may be opening a box that cannot be reclosed. Maybe that's the point. 

Appeared in the October 12, 2012 edition of Product Liability Law360