Concept
Where But-For Strains
But for and substantial causes
Where But-For Strains
Ordinary causation doctrine asks two things of a plaintiff. Under Restatement (Second) of Torts § 432(1), negligent conduct is not a substantial factor in bringing about harm that would have been sustained even if the actor had not been negligent.1 Under § 328A(c), the plaintiff must prove that the defendant’s failure was a legal cause of the harm suffered.2 Together the two sections require evidence that the harm more probably than not would not have occurred with proper care.
Two situations make that showing impossible even though a defendant plainly did something wrong. In the first, several defendants were careless in the same way, one of them certainly injured the plaintiff, and no one can say which. Each defendant can then argue that the plaintiff has not proved that this defendant’s negligence was necessary to the harm, and the plaintiff recovers from no one. In the second, a negligent delay cost the plaintiff a chance of a better outcome, but the chance was already below even, so the plaintiff cannot show that proper care would more probably than not have changed the result.
Courts have answered both problems with departures that are narrow, divided, or both. Each departure rests on a particular feature of the problem, and each leaves the but-for baseline in place for the ordinary case. The first concerns the plaintiff who cannot say which negligent defendant caused the harm.
Unidentified Defendants
Summers v. Tice shows how the but-for burden can defeat a plaintiff who was plainly wronged. Two hunters, both negligent, fired their shotguns, and a pellet from one gun or the other injured the plaintiff’s eye. The odds were fifty-fifty that either defendant’s shot had hit the eye, so the plaintiff could not show that it was more probable than not that either pellet was the one.3 Under strict but-for analysis he would lose against both.4
The California Supreme Court would not let him lose. Both defendants had done the same thing, both were negligent, and one of them had injured the plaintiff, so it seemed unfair to excuse both merely because the plaintiff could not pinpoint the shooter. The court shifted the burden to each defendant to prove that he had not caused the harm. Neither hunter could do that any better than the plaintiff could, and both were held liable.5 The liability was joint and several unless one of them could prove that the shot came from the other’s gun.6
The court gave two reasons. The defendants were better placed than the plaintiff to establish whose act had caused the harm, and as between two negligent defendants and an innocent victim, the defendants should pay even without proof of which one fired the pellet.7 The first reason did little work on the facts, because the hunters knew no more than the plaintiff did.8 The second carries the rule: without the shift, each careless hunter would be shielded by the other’s carelessness.
The tomato case has the structure of the hunters’ case. Both farmers were careless in the same way, one tomato surely did the harm, and the diner cannot show whose it was. The difficulty is one of proof, and the burden-shifting solution answers it: each farmer must prove that the tomato was not his, neither can, and each is jointly and severally liable. Presumably one farmer pays and seeks contribution from the other, so the two end up splitting the damages.9
The solution has a cost. Because the diner ate only one tomato, only one farmer caused her any harm, yet neither can prove that the other’s tomato was the one, so the farmer who caused no harm is liable too. Traditional doctrine says that a defendant’s negligence must cause injury before liability attaches, and the rule does not honor that. Courts that adopt it judge the departure better than sending the plaintiff away empty-handed.10
The tomato case also differs from the overdetermined harm in Anderson. There, two causes were each sufficient to produce the harm, so each defendant could argue that its negligence was not a but-for cause because the other’s would have done the same damage. Each cause was clearly sufficient, though, and the difficulty lay in the but-for test itself. Here only one cause operated, and the difficulty is identifying which defendant supplied it.11
Each condition answers a different risk. Proof that every defendant was negligent means that only proven wrongdoers can be charged with a loss they may not have caused. Substantially similar acts mean that the uncertainty is truly even, with no evidence pointing toward one defendant. And the requirement that one of the acts caused the harm confines the rule to the either/or problem, in which the only gap in the proof is identity.
An exam stem that names several defendants and leaves the cause unproved invites a citation to Summers, and the invitation should be declined unless the conditions are met. The case does not hold that the burden shifts whenever a plaintiff sues more than one defendant. It addresses a narrow anomaly in which but-for analysis fails, and it is not a general abdication of the plaintiff’s burden on causation.13
The pole case falls outside Summers. The defendants’ acts differ, so the chances are not equal, and the jury can weigh the evidence in the ordinary way. The pedestrian keeps the usual burden against each defendant, including the burden of proving that each was negligent.14 The doctrine also does not fit two successive negligent acts that probably each contributed part of one injury. Where a plaintiff is hurt in two collisions, as in Osteen’s case, the first collision may have caused some of the injury and the second some more. In Summers the pellet was either one defendant’s or the other’s, so one defendant had caused no harm and the other had caused it all.15
Courts do not agree about Summers itself. A number have followed it, while others have rejected it entirely and left the plaintiff to the traditional but-for test.16 Market-share liability is a related variation. The plaintiff must join defendants representing a substantial share of the relevant market at the relevant time, and each is liable for a fractional share of the damages measured by its market share. Jurisdictions differ on whether to recognize it.17
Lost Chance of Survival
The same two sections that govern the ordinary case produce an awkward result when a physician’s negligent delay reduces a chance of survival that was already below even. In Herskovits v. Group Health Cooperative of Puget Sound, a physician’s missed diagnosis of lung cancer cut the plaintiff’s chance of survival from 39 percent to 25 percent.19 The statistics suggest that the outcome might have been different without the negligence but probably would not have been. Of every hundred patients like him, 61 were doomed either way.20 A plaintiff who must show that proper care would more probably than not have saved the decedent cannot do so, although the late diagnosis deprived him of a 14 percent chance to survive.21
The difficulty lies in what the evidence proves. The plaintiff has lost something real, a significant chance that her husband would still be with her, and the same statistics that prove the loss also show that the delay probably did not change the outcome.22 The all-or-nothing treatment of that chance produces an anomaly. Under the rule typified by Cooper v. Sisters of Charity, Inc., a plaintiff who proves that, but for the negligence, the decedent had a 51 percent chance of survival recovers for the whole death, even though there was a 49 percent chance it would have occurred anyway. A plaintiff who proves a 49 percent chance recovers nothing.23 The physician’s carelessness is identical in the two cases, and only the starting odds differ. The Herskovits court objected that denying recovery would release doctors and hospitals from liability whenever the chance of survival was under 50 percent, regardless of how flagrant the negligence.24
Courts have taken three positions. The first keeps the traditional cause-in-fact requirement and denies all recovery unless the victim’s chances were initially over 50 percent.25 Its premise is that liability requires proof that the wrong probably changed the result, and a patient with a 40 percent chance probably would have died anyway.
The second relaxes the requirement while keeping the vocabulary of causation. Under this “relaxed causation” rule, of which Herskovits is the usual example, the jury may award damages if it finds the negligence a “substantial factor” in hastening or precipitating the adverse result, and the plaintiff need not prove that the initial chance exceeded 50 percent. A plaintiff who wins recovers 100 percent of the damages, even though the lost chance may have been below 50 percent before the negligence.26 This use of “substantial factor” is not the multiple-sufficient-causes use taught earlier, since no second sufficient force is present, and the instruction simply loosens the plaintiff’s burden. Jurors have no principled way to apply it. They cannot place Herskovits in the 14 percent group that would have lived with earlier diagnosis rather than among the 61 percent who were doomed either way, so the approach licenses a finding for the plaintiff on intuitive grounds, despite the likelihood that the negligence changed nothing.27
The third position changes what counts as the injury. On this view a plaintiff who cannot prove by a preponderance that proper care would have cured the disease can still prove that the delay deprived him of a chance of cure, say 30 percent, and the probability of survival is reflected in the amount of damages.28 An increasing number of courts now allow recovery for the lost chance itself under this value-of-the-chance rule. Damages equal the percentage difference attributable to the defendant’s negligence multiplied by the plaintiff’s total damages.29 The defendant pays for what the delay actually took, so the payment tracks the size of the wrong. Liability for the lost chance has so far been confined mainly to medical negligence cases, and it seems proper to limit it to these or similar cases where the defendant’s duty was to maximize the plaintiff’s chances.30
Sven’s case shows how far the strict and proportional rules diverge. Sven first saw Dr. Kildare, and the medical testimony puts his chance of recovery at 40 percent had he been diagnosed then and at 15 percent by the time he reached the emergency room. His survivors sue Kildare for wrongful death.31 Assume that Kildare’s failure to diagnose was negligent. Under strict but-for causation the survivors lose, because his chance with proper care was below 50 percent. Under the relaxed approach the jury may find the delay a substantial factor, and a verdict for the survivors carries full damages. Under the value-of-the-chance rule they recover 25 percent of the wrongful-death damages, the difference between 40 percent and 15 percent.32
The first family is paid for a death the delay did not cause, and the second receives only a quarter of a loss the delay wholly caused. That is the standard objection to the percentage solution: it overcompensates in every case in which the decedent would have died anyway and undercompensates in every case in which the delay caused the death.33 Courts that favor loss of a chance answer that it rests on the same justifications as increased-risk recovery: deterring negligent physician conduct and compensating real harms to patients that fall below the 50 percent threshold of traditional doctrine.34
Applied to the Herskovits numbers, the three positions give three answers. The strict rule awards nothing, because a 39 percent chance cannot be turned into a more-probable-than-not showing. The relaxed approach sends the case to the jury, and a plaintiff’s verdict carries the full wrongful-death damages. The value-of-the-chance rule awards 14 percent of those damages, the share of the chance the delay removed.35
Students most often skip the second step. A plaintiff whose chance with timely care was above 50 percent needs no lost-chance doctrine at all, because she can show that proper care would probably have saved the patient and may recover for the entire death.36 Applying a percentage discount to that plaintiff shortchanges a claim that already satisfies ordinary causation.