This appeal arises out of the denial of damages to plaintiff James DePass for alleged increased risk of cardiovascular disease and loss of life expectancy. The district court, in a bench trial, found that DePass had not proved the injury by a preponderance of the evidence. This appeal requires us to determine whether the district court was "clearly erroneous" in its finding. Fed.R.Civ.P. 52(a). For the reasons set out below, we affirm the district court.

* On December 9, 1978, DePass was struck by a car driven by an employee of the defendant United States. DePass sustained severe injuries, including a traumatic amputation of his left leg below the knee. DePass brought suit under the Federal Tort Claims Act, 28 U.S.C. Secs. 2671-2680. The United States admitted liability, and the case was tried as to damages only.

At trial, DePass introduced evidence as to the nature and extent of his injuries, and as to past and future pain and suffering. DePass's witnesses included Dr. Jerome D. Cohen, a medical doctor.1 Dr. Cohen testified that he had examined DePass and found that DePass had had a traumatic amputation of the left leg just below the knee. Dr. Cohen then testified that he had read an article, "Traumatic Limb Amputation and the Subsequent Mortality from Cardiovascular Disease and Other Causes," published in volume 33 of the Journal of Chronic Diseases, by Zdenek Hrubec and Richard Ryder ("the Hrubec and Ryder study") (Pl.Ex. 10).2 This article involved a study of 3,890 Americans who had suffered traumatic limb amputations during World War II. The study established a statistical connection between traumatic limb amputations and future cardiovascular problems and decreased life expectancy.

Dr. Cohen testified that DePass fits within the class of persons who had sustained a traumatic amputation of a limb. He testified that, based on his own experience, his examination of DePass, and his analysis of the Hrubec and Ryder study, because DePass is a traumatic amputee he has a greater risk of cardiovascular problems and decreased life expectancy. At the conclusion of Dr. Cohen's testimony, the trial court questioned him as follows:

The Court: Is it your testimony that based on your examination of this plaintiff, Mr. DePass, plus your knowledge in the field of cardiovascular disease and including your analysis and a reading of the article which is Exhibit 10, is your opinion that he, because he is the amputee that he is on the left leg below the knee, that he has a greater risk for developing cardiovascular disease in the future and also a greater risk for shorter mortality, shorter longevity? Is that what you're saying, that he is a risk for that?

[Dr. Cohen]: Yes, that's correct.

The Court: And you're not saying that he will develop it or that he will live shorter, but he is a greater risk for both of those?

[Dr. Cohen]: That's correct.

(Tr. at 289-90).

On cross-examination, Dr. Cohen testified to the existence of several other studies on the relationship between traumatic limb amputation and decreased life expectancy. In particular, he testified as to a 1954 English study of 27,000 amputees that showed no statistical connection between amputation and decreased life expectancy. Several studies conducted on a smaller scale also failed to show a connection. One study, done in Finland, found a statistical connection. He testified that he had not personally reviewed any of the other studies. Dr. Cohen testified that no one knows if in fact the Hrubec and Ryder study is correct. Dr. Cohen then read from the study, "The reasons for the statistically significant relationship demonstrated between limb amputation and cardiovascular disorders are not obvious. Stringent deadlines set by Congress for the completion of this work made impossible detailed studies of individuals." (Tr. at 285).3

The district court, in its order, awarded DePass $800,000 for the nature and extent of his injuries and for his past and future pain and suffering. In its findings of fact, the court found no evidence demonstrating by a preponderance that DePass had suffered a loss of life expectancy. The court found that Dr. Cohen's testimony dealt with "possibilities and speculation."4

On appeal, DePass argues that the district court was clearly erroneous in its findings of fact. He argues that his evidence was clear, convincing, and uncontradicted. He also argues that the United States is bound by the Hrubec and Ryder study as an admission by the government. The United States argues that the district court could reject Dr. Cohen's testimony and the Hrubec and Ryder study. The United States also argues that the evidence at best establishes only that DePass is, on the average, at an increased risk of injury, and that this increased risk is not compensable under Illinois law.

An appellate court will only set aside the findings of a district court when those findings are clearly erroneous. The standard of review is clear. The Supreme Court has stated that "[a] finding is 'clearly erroneous' when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 541, 92 L. Ed. 746 (1948). See also Lee v. National Can Co., 699 F.2d 932, 936 (7th Cir.) cert. denied, --- U.S. ----, 104 S. Ct. 148, 78 L. Ed. 2d 138 (1983). The appellate court cannot reweigh the evidence. West v. Schwarz, 182 F.2d 721, 722 (7th Cir.1950). The inquiry is whether the record contains substantial evidence to support the findings. Id.

Based on the record before us, we cannot say that the district court was clearly erroneous. The findings of fact demonstrate that the district court considered the Hrubec and Ryder study and Dr. Cohen's testimony. The court, however, declined to accept the study and the testimony as establishing the fact of injury by a preponderance. There is substantial evidence in the record to support this. Dr. Cohen testified that he based certain of his conclusions on the Hrubec and Ryder study. The study itself states, and Dr. Cohen testified, that other studies exist that reach a different conclusion. Dr. Cohen stated that no one knows if the Hrubec and Ryder study is correct. The study does not establish a reason for the statistical relationship between amputation and decreased life expectancy.

There was conflicting evidence in the record as to the conclusiveness of the Hrubec and Ryder study. It was for the district court to weigh the evidence; the court did not find the evidence strong enough to be a preponderance.

Moreover, even if there were no contradictory evidence on this point, the district court could still reject the conclusions of the Hrubec and Ryder study. The plaintiff had the burden of proving injury. The court could reject the evidence even when the defendant introduces no contradictory evidence. Nicholl v. Scaletta, 104 Ill.App.3d 642, 60 Ill. Dec. 368, 432 N.E.2d 1267 (1982).

Because we find that the district court could reject the Hrubec and Ryder study as inconclusive, we do not reach the issue of whether merely placing plaintiff in a class of persons subject, on the average, to an increased risk of injury is sufficient to establish a compensable injury under Illinois law. We note, however, that Illinois law is not settled as to whether increased risk of future injury is compensable. Compare Lindsay v. Appleby, 91 Ill.App.3d 705, 46 Ill. Dec. 832, 414 N.E.2d 885 (1980) with Morrissy v. Eli Lilly and Co., 76 Ill.App.3d 753, 32 Ill. Dec. 30, 394 N.E.2d 1369 (1979).

We have considered appellant's other arguments and determine them to be without merit.

Affirmed.

POSNER, Circuit Judge, dissenting.

Although this may seem like a routine personal-injury case, it raises important questions relating to the use of scientific evidence in federal trials. The plaintiff, a 37-year-old man named DePass, was hit by a car owned and operated by the government. He was seriously injured--one of his legs had to be amputated just below the knee, the other was crippled, and one eye was badly injured. He brought suit under the Federal Tort Claims Act and the district court held the government liable and awarded DePass $800,000 in damages. The entire award is for "pain and suffering," since DePass incurred no medical expenses (he received all medical treatment free of charge from the Veterans Administration) and proved no loss of earnings from the accident. This is a generous award, maybe too generous, even though "pain and suffering" does not mean just physical pain and suffering but includes the unhappiness caused by disfiguring and crippling injuries. See, e.g., Riddle v. Memorial Hospital, 43 App.Div.2d 750, 349 N.Y.S.2d 855 (1973). In any event, the judge's failure to explain the basis of the award seems inconsistent with the requirements of Rule 52(a) of the Federal Rules of Civil Procedure. See O'Shea v. Riverway Towing Co., 677 F.2d 1194, 1201 (7th Cir.1982).

But the government has not appealed, so for purposes of this appeal we must accept that $800,000 is a reasonable estimate of DePass's damages, assuming as the judge found that DePass's life was not shortened by the accident. This appeal challenges that finding. Although the additional loss inflicted by shortening what is now likely to be a rather miserable life may be slight in pecuniary terms, especially after being discounted to present value, if DePass proved that the accident probably shortened his life he was entitled to some additional damages and it should be no concern of ours whether the addition would be small or large or whether as an original matter we might think $800,000 adequate or even excessive to compensate him for all of his losses.

If the finding that the accident will not shorten DePass's life had been based in whole or part on the judge's opinion of the believability of a witness I would agree that we should affirm. But there are no "credibility" issues in this case. The government acknowledged at oral argument that DePass's medical expert witness, Dr. Cohen, a cardiovascular epidemiologist, was a candid and truthful witness. I do not understand the district judge to have disbelieved Cohen. I do not even think the judge rejected the finding in the National Institutes of Health study on which Cohen relied (Hrubec & Ryder, Traumatic Limb Amputations and Subsequent Mortality from Cardiovascular Disease and Other Causes, 33 J. Chronic Diseases 239 (1980)) that a person who has had a major amputation--knee or above, or elbow or above--is more likely to develop heart disease and therefore likely to die younger than a person who has not had a major amputation. The judge did remark that a British study apparently had found no association between amputation and longevity. (I say "apparently" because no one connected with this case has ever read the study, which was not published in a medical journal and may not even be available in the United States. All remarks about it are based on a couple of sentences in the NIH study.) Dr. Cohen explained why he thought the British study was probably less pertinent than the NIH study to estimating DePass's life expectancy. The British study was older (1954 versus 1979); and it was of a foreign population in a much earlier period (British veterans of World War I versus American veterans of World War II in the NIH study), a population moreover that has different health characteristics from ours (heart disease is less common in Britain than in the United States). Cohen's testimony was not shaken by cross-examination and the government did not introduce any medical testimony, live or documentary, of its own. Compare Orchard View Farms, Inc. v. Martin Marietta Aluminum, Inc., 500 F. Supp. 984, 995-96 (D.Ore.1980); Mahoney v. United States, 220 F. Supp. 823, 840-41 (E.D.Tenn.1963). It did not even place the British study in evidence. The district judge could have had no basis for rejecting Dr. Cohen's assessment of the relative weight of the two studies.

But as I have said, I do not think the judge rejected DePass's medical evidence because he thought it false (how could he have?) or outweighed by other evidence (there was no contrary evidence). He appears to have rejected the entire class of evidence--statistical evidence--illustrated by the medical evidence in this case. This is suggested by his finding that "the doctor never established that this individual plaintiff would suffer a loss of life expectancy, or even that it was more likely than not that he would suffer the loss. At best, his testimony dealt with possibilities and speculation and failed to establish the existence of such a loss in plaintiff." Another clue is found at the close of Dr. Cohen's testimony when the judge asked him: "Is it your testimony ... that because [DePass] is the amputee that he is ... he has a greater risk for developing cardiovascular disease in the future and also a greater risk for shorter mortality, shorter longevity? Is that what you're saying, that he is a risk for that?" When Dr. Cohen answered, "That's correct," the judge said: "And you're not saying that he will develop [cardiovascular disease] or that he will live shorter, but he is a greater risk for both of those?" When Dr. Cohen again said, "That's correct," the judge remarked: "I won't get into any statistics. I'll just drop it at that."

Maybe the judge just misunderstood the term "life expectancy." When he said that Dr. Cohen had "never established that this individual plaintiff would suffer a loss of life expectancy," he was stating a truism rather than, as he thought, throwing doubt on the cogency of the plaintiff's medical evidence. A man's life expectancy is not his actual life span, which cannot be known till he dies; it is a prediction founded on the average experience of many people having the same characteristics as he. When we say that a heavy smoker has a shorter life expectancy than a nonsmoker we do not mean that Jones who smokes heavily will in fact predecease Smith who does not smoke or that Jones will die sooner than he would have if he did not smoke; we just mean that, on the average, nonsmokers outlive smokers.

But it seems more likely from his remarks that the district judge thought that all probabilities are too uncertain to provide a basis for awarding damages. Yet most knowledge, and almost all legal evidence, is probabilistic. Even the proposition that DePass will die some day is merely empirical. It is of course highly probable that he will die but it is not certain in the way it is certain that 10³ is 1000 or that I am my wife's husband--propositions that are true as a matter of definition rather than of observation. If Dr. Cohen had testified that DePass had heart disease and was therefore likely to die younger than most men in his age group, he would have been making a probabilistic statement; and the probabilities that are derived from statistical studies are no less reliable in general than the probabilities that are derived from direct observation, from intuition, or from case studies of a single person or event--all familiar sources of legal evidence.

All this has long been recognized in personal-injury cases, as it is throughout the law. If a tort victim is seriously injured and will require medical attention for the rest of his life, the court in deciding how much to award him for future medical expenses will have to estimate how long he can be expected to live and it will make this estimate by consulting a mortality table, which is to say by looking at a statistical summation of the experience of thousands or millions of people none of whom is a party or a witness in the case, rather than by studying the lifelines on the victim's palms. And if a study has been made of the mortality of people with the same kind of injury as the plaintiff, the court will consult that study in addition to or instead of standard mortality tables. That is what DePass asked the district judge to do here.

It makes no logical or even legal difference whether the question is how long a permanently injured person will require medical attention or how soon a delayed consequence of the accident (death in this case) will occur. The factual question is the same: when will he die? The Illinois courts (Illinois law governs the substantive questions in this Federal Tort Claims Act suit) recognize that the risk of a future consequence of an accident, even though not certain to materialize, is compensable. See Lindsay v. Appleby, 91 Ill.App.3d 705, 712, 46 Ill. Dec. 832, 837, 414 N.E.2d 885, 890 (1980) ("evidence tending to show an increased risk of further injury caused by defendant's conduct ... was admissible as tending to show plaintiff's future damages"); York v. Grand Trunk W.R.R., 71 Ill.App.3d 800, 806-07, 28 Ill. Dec. 134, 139-40, 390 N.E.2d 116, 121-22 (1979). Although Morrissy v. Eli Lilly & Co., 76 Ill.App.3d 753, 761, 32 Ill. Dec. 30, 37, 394 N.E.2d 1369, 1376 (1979), cited by the majority opinion in the present case, recites the formula, "possible future damages in a personal injury action are not compensable unless reasonably certain to occur," Morrissy is distinguishable from this case. The plaintiff had not yet suffered any ascertainable injury as a result of her mother's having taken DES while pregnant; the complaint was that she might some day develop an illness as a delayed consequence of the DES. DePass was seriously injured; the only issue is the extent of his injury, an issue on which courts traditionally do not impose a heavy burden of proof on plaintiffs. See Story Parchment Co. v. Paterson Parchment Paper Co., 282 U.S. 555, 562-65, 51 S. Ct. 248, 250-51, 75 L. Ed. 544 (1931); Hamil v. Bashline, 481 Pa. 256, 271-73 and n. 10, 392 A.2d 1280, 1288-89 and n. 10 (1978); Hicks v. United States, 368 F.2d 626, 632-33 (4th Cir.1966). Thus in Harp v. Illinois Central Gulf R.R., 55 Ill.App.3d 822, 827, 12 Ill. Dec. 915, 918-19, 370 N.E.2d 826, 829-30 (1977), a case where, as in this case but not in Morrissy, the issue was extent of injury, the "reasonable certainty" formula was invoked but then liberally interpreted to allow an accident victim to recover damages for the possible future rupture of a disc injured in the accident.

Although few reported cases (none from Illinois) deal with the specific question whether a reduction in life expectancy is compensable, the trend is toward allowing recovery in such cases. See, e.g., Downie v. United States Lines Co., 359 F.2d 344, 347-48 (3d Cir.1966); McNeill v. United States, 519 F. Supp. 283, 289 (D.S.C.1981); James v. United States, 483 F. Supp. 581, 586-87 (N.D.Cal.1980); Schultheis & Rheingold, Making Up for Lost Time: Recovering for Shortened Life Expectancy, 19 Trial 44 (Feb. 1983). As it should be. A tortfeasor should not get off scot-free because instead of killing his victim outright he inflicts an injury that is likely though not certain to shorten the victim's life.

In any event, the government conceded at oral argument that a reduction in life expectancy is compensable under Illinois tort law. So the only question is the reliability of the study on which Dr. Cohen relied. The study is of the 30-year medical history of 12,000 Americans injured in combat in World War II of whom almost 4,000 had major amputations as a result of their injuries. (The British study was larger, but we do not know how many of the persons studied were major amputees, or even whether the study distinguished between major and minor--though no one seriously believes that losing your pinky will significantly increase the risk of dying prematurely from heart disease.) Although the NIH study does not establish the exact causal linkage between major amputation and early death from heart disease, it suggests a linkage. Emotional stress and (more important) lack of exercise are believed to be among the causes of heart disease, and a major amputation is a source of stress (reflected for example in higher suicide and alcoholism rates among amputees) and makes exercise more difficult. Consistently with this hypothesis, the study found that the increased incidence of heart disease was greater for men who had lost two legs than for men who had lost one (functionally, DePass is closer to the former than to the latter category), and greater for those who had lost one leg than for those who had lost just an arm.

The NIH study does not pretend to be conclusive and its results are presented with appropriate scientific caution. Subsequent studies may overthrow them. But as I have already noted a tort plaintiff's burden of proving the extent of his injury is not a heavy one. Doubts are resolved against the tortfeasor. All of the medical evidence introduced in this case (the government, it will be recalled, presented none) points in one direction--that DePass will probably die younger because of the accident. On the basis of the study Dr. Cohen testified that DePass had a 44-58 percent higher than normal risk of dying from heart disease, resulting in a 30 percent reduction in DePass's life expectancy--from 37 to 26 years, a difference of 11 years. This reduction in life expectancy is just a convenient summation of the probabilistic results of the NIH study applied to a man of DePass's age. To illustrate very simply what is involved, imagine that before the accident DePass had a 10 percent chance of dying at age 60, a 50 percent chance of dying at age 65, and a 40 percent chance of dying at age 70. Multiply each age by the corresponding percentage, sum up the results, and you have his expected age at death: 66.5 years. Assume the accident changed these percentages as follows: he now has a 40 percent chance of dying at age 60 and a 60 percent chance of dying at age 65. His expected age of death is now 63, so the accident reduced his life expectancy by 3 1/2 years. From the way I have set up this highly stylized example we know for a fact that DePass will not die at the age of 63. All we know is that he is likelier, by given percentages, to die at either 60 or 65 than he was before the accident. But if we want to summarize all of our information in a single figure, we can do so by means of the concept of life expectancy. It provides a perfectly objective basis for awarding damages.

If any corroboration of the applicability of the NIH study to DePass's personal situation was needed, it was supplied by Dr. Cohen's finding, in an examination of DePass, that DePass's cardiovascular function had fallen below normal for men of his age. This is not evidence of heart disease but it shows that DePass is, as one would expect given the nature of his injuries, perforce leading an extremely sedentary life--which according to the NIH study is probably the main cause of the much greater incidence of heart disease among major amputees.

It is of course unlikely that DePass will die exactly 11 years earlier because of the accident. He may never get heart disease. Or he may be cut down by it long before he is within 11 years of the expected end of his life. It was unlikely that the accident victim in Abernathy v. Superior Hardwoods, Inc., 704 F.2d 963, 972-74 (7th Cir.1983), to pick a recent personal-injury case almost at random, would die 40 years after the accident--a number plucked from a mortality table, reflecting as the NIH study reflects statistical experience. And yet Abernathy was awarded damages as if he would die in 40 years. That is how damages are calculated in personal-injury cases; and a judge is not free to say, in my court we do not allow statistical inference. Knowledge increasingly is statistical, and judges must not let themselves lag too far behind the progress of knowledge. As a matter of fact they have not lagged. The kind of evidence that the district judge rejected in this case, evidence of probability of survival, invariably based on studies of a group of people rather than of just the individual plaintiff, is an increasingly common basis for awarding damages. See, e.g., Herskovits v. Group Health Coop., 99 Wash.2d 609, 664 P.2d 474 (1983); Bowman v. Twin Falls Construction Co., 99 Idaho 312, 316-19, 581 P.2d 770, 774-77 (1978); Daniels v. Hadley Memorial Hospital, 566 F.2d 749, 758 (D.C.Cir.1977); McCallister v. Workmen's Compensation Appeals Bd., 69 Cal. 2d 408, 71 Cal. Rptr. 697, 445 P.2d 313 (1968); Wolfstone & Wolfstone, Recovery of Damages for the Loss of a Chance, 1978 Personal Injury Annual 744.

The district judge's rejection of such evidence, if widely followed, would lead to systematically undercompensating the victims of serious accidents and thus to systematically underdeterring such accidents. Accidents that require the amputation of a limb, particularly a leg, are apparently even more catastrophic than one had thought. They do not just cause a lifetime of disfigurement and reduced mobility; they create a high risk of premature death from heart disease. The goal of awarding damages in tort law is to put the tort victim as nearly as possible in the position he would have occupied if the tort had not been committed. This goal cannot be attained or even approached if judges shut their eyes to consequences that scientists have found are likely to follow from particular types of accident, merely because the scientists' evidence is statistical. But unless I have mistaken the true grounds of the district judge's decision in this case that is what he did.

The finding that DePass failed to prove a reduction in life expectancy as a result of the accident should be vacated as clearly erroneous and the case should be remanded to the district court for a determination of the amount of damages necessary to compensate DePass for an 11-year reduction in his life expectancy.