From: Sandy Steel <>
To: 'Robert H Stevens' <>
CC: 'Neil Foster' <>
Date: 09/04/2009 23:57:54 UTC
Subject: RE: Causation/Loss of chance in medical cases

Thanks for the response.  

So we agree that

(i) Loss of chance represents an epistemic uncertainty regarding some state

of the world.

I also agree that

(ii) ""Negligence in the air" isn't enough to prove a wrong". (There are

however three interrelated concepts under the "no negligence in the air"

idea: that the tortious aspect of the conduct must be a factual cause of

some harm; that the harm is not pre-empted; and that (roughly) the harm is

within the risk: Wright (1985) Cal.L.Rev)

So turning to the proposed explanation of the cases where the law imposes

liability for loss of chance:

"Negligence "in the air" isn't enough to prove a wrong. You need to be able

to show that it was your rights which were violated, which the patient can

do as against the doctor who has assumed responsibility towards her, but

those who live in the vicinity of the chemical factory cannot"

But the no "negligence in the air" point does not map onto the point that

loss-of-chance liability can be explained by instances where there has been

a rights-infringement and loss-of-chance damages are awarded as

consequential losses. (Obviously, the negligence in the air point (in its

first two meanings) is an application of the principle that the right must

have been infringed: but the principle is obviously broader, applying

outside negligence, and outside wrongs which require causation of some harm,

e.g. trespass). I don't see why an "assumption of responsibility" meets the

negligence in the air point, it does not establish (absent a contract) an

infringement of right where the duty imposed by the assumption of

responsibility is breached (it does not establish causation)

If the response to that is: well, that shows how flimsy the distinction is

between bringing the claim in contract and tort for loss of a chance, then

we can't say that liability is "explained" by where there is an infringement

of right since we acknowledge in the response that that is a distinction

(between bringing in contract where we have a nominal breach and

infringement of rights and tort where we don't) without a difference.

And if we then say, more substantively, that "assumption of responsibility"

explains recovery for pure economic loss, and recovery for loss of chance of

avoiding economic loss, and recovery for loss of chance of avoiding physical

injury then we are, I think, pushing it a little. Though as I said, some

subscribe to a view along those lines : Goldberg and Zipursky in the article

I cited; and Perry (1995, ed. Owen, Philosophical Foundations of Tort). But

this line still leads to the problem of why recovery should ever be more

than proportionate in this context. If we say in principle C should recover

in full where he can show D wronged him on the BoP, then do we treat the

loss-of-chance claim as a subsidiary claim? I don't think this is met by

Rob's point, elaborated in his Torts and Rights (2007), that we have to

distinguish with regard to Smith v. Leech Brain between an infringement of

right and the quantification of consequential loss. I think this distinction

can be accommodated by a better concept of causation, like NESS, which

relegates Baker v. Willoughby/Jobling-like problems to valuation of damages.

But I don't see why putting this doctrinally as matter of damages should

hide the fact that damages are reduced to account for the probability that

due to a non-tortious factor it might have happened anyway.

Secondly, Rob wrote:

"The last point [whether consequential damages had to be proved in the US to

be reasonably likely to occur] is an issue of remoteness and is the same in

England. Where you seek recovery for consequential loss you have to satisfy

Hadley v Baxendale."

I would be wise not to disagree with that! But I didn't mean to dispute

anything about remoteness. It is one thing whether damages are too remote

(whether they are not within Hadley limbs (1) and (2)). It is another thing

whether, even if they are not too remote, what standard of proof will be

applied in determining whether the defendant caused them. In the US, I

think, one must show with regard to future loss (and past loss - no weird

distinctions based on third parties' actions as far as I know) that the type

of damage is reasonably foreseeable (remoteness), but also that is more

likely than not that the D caused it. Whereas in England, one can claim for

the loss of chance of obtaining some past/future benefit. That effectively

means that a lower standard of proof is applied to some future loss (one can

obtain damages for a harm less than 50% likely to occur).

Finally, sorry for rambling on, but one last point. I disagree with the

conclusions Rob draws from his analysis of the factory-chemical example:

"If I negligently release a noxious chemical which 100,000 people inhale, if

25 years later there are 150 cases of lung cancer when without the release

there would have been only 100, all of those who suffer from the disease can

show that I have caused them to lose the chance of avoiding the disease but

none can show that I have injured/wronged them as on the balance of

probabilities each would have suffered the disease anyway".

Ex ante, Rob argues, the risk exposure each of the 100,000 suffers is

1/2000. Ex post, only the 150 who suffer actually lose a chance of 1/3:

"Only the 150 people who develop the disease suffer a lost chance (1/3) of

avoiding the cancer they have". This is supposed to show that the increased

risk of avoiding harm in the future is not the same as the loss of chance of

avoiding a harm.

This is true, but it is only true because we have more data when we know

that the 150 people have suffered cancer; we have a better idea as to the

probability that the defendant has caused the individual C's harm. We know

that 100 were doomed anyway and that 50 got cancer through the D's agency.

We do not know who belongs in which class. It is therefore impossible to say

that all those who suffer the cancer "actually" lost a chance (Rob does not

use this word, but it is a contrast implied in his analysis)- some of them

didn't - they were doomed: all we have is just a better approximation for

each of the 150 that the defendant caused their cancer. There would thus be

no inconsistency for the law to award damages for loss of chance before the

harm actually occurs: some people are doomed; some are not. We are in

exactly the same situation when the harm occurs; we just have better


I'll be amazed if anyone makes it to the end of this,