WELLFORD, Circuit Judge.

There is no dispute that Joseph Novak, a young man of 37 who had previously been in good health, died of a rare disorder known as dermatomyositis on April 11, 1977. The plaintiff, his widow, sued the United States under the Federal Tort Claims Act (FTCA) and the Swine Flu Act claiming that her husband's untimely death was the result of his vaccination for swine flu on November 10, 1976.

The Swine Flu Act, enacted in 1976, was an attempt to inoculate the entire adult population of the United States to avoid a feared epidemic. This act marshalled the forces of public and private health care organizations to give free influenza inoculations to everyone requesting them. During the administration of this program, 45 million adults were vaccinated. Although the feared swine flu epidemic never materialized, several people suffered injuries or died as a claimed consequence of the vaccination. Foreseeing this potentiality, Congress permitted a cause of action in tort against the United States, and allowed recovery for all injuries proved to be the direct result of the vaccination. 42 U.S.C. § 247b(k). For details of the reasons for its enactment, see generally Hasler v. United States, 718 F.2d 202 (6th Cir. 1983), cert. denied, 469 U.S. 817, 105 S.Ct. 84, 83 L.Ed.2d 31 (1984), and Gicas v. United States, 508 F. Supp. 217 (E.D.Wis. 1981).

Dermatomyositis (DM/PM), akin to polymyositis, is a "perplexing disease of unknown origin, thought to affect the body's auto-immune system which causes the immune system to destroy muscle and/or skin tissue." Then, and now, the scientific and medical community does not know the exact cause of DM/PM.

The district court found that Mr. Novak suffered from DM/PM within fifteen days of his inoculation with the swine flu vaccine. His wife testified that on November 25, 1976, around Thanksgiving, he complained of soreness in his leg and shoulder muscles, and was unable to engage in his normal athletic and other activities. She conceded that he became severely ill in January 1977, shortly after seeing a family doctor on December 27, 1976. His condition persisted, resulting in a hospital admission during January, at which time he was diagnosed as suffering from the DM/PM disorder. Mr. Novak unfortunately never recovered, and died within a few months.

A. Proximate Causation.

To support her claim that Novak's death was caused by the swine flu vaccination, Mrs. Novak relied primarily on the testimony of Dr. Joseph Bellanti, an expert in immunology and microbiology and a professor at Georgetown University Medical School. Dr. Bellanti gave an opinion that Mr. Novak's DM/PM "was related to the vaccine that he had received." In support of his conclusion, Dr. Bellanti drew inferences from the fact that Novak was in good health prior to the shot, that the lapse of time between the shot and the illness was relatively short, and that the disorder first manifested itself in the same arm in which Novak had received the shot. He also relied on research studies and his own experience. Dr. Bellanti believed clinical evidence showed that DM/PM may be caused by introducing a virus or virus-like particle into the body. In this respect, Bellanti's testimony was corroborated to some degree by other experts, Doctors Pearson and Lawry.

Based on this testimony, the district court found that the swine flu inoculation probably caused Novak's death, and entered judgment for the plaintiff. Looking to the "totality of circumstances," the judge concluded that causation was shown, although he was aware of and troubled by the problem and difficulty of demonstrating proximate cause. The district judge conceded that "the medical experts agree that no scientific data or epidemic logic data exists that DM/PM is a viral caused disease." The government appealed from the adverse judgment.

Federal Rule of Civil Procedure 52(a) applies in this case, and the district court's causation finding is reviewed under the clearly erroneous standard. Hasler v. United States, 718 F.2d 202 (6th Cir. 1983), cert. denied, 469 U.S. 817, 105 S.Ct. 84, 83 L.Ed.2d 31 (1984). We may not overturn the decision, therefore, unless we are "left with a 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). The plaintiff must make out her case under Ohio tort law, the place of the inoculation and the death of Mr. Novak. 42 U.S.C. § 247b(k)(2). In actions of this type, we construe the state law in favor of the plaintiff, the intended beneficiary. See Unthank v. United States, 732 F.2d 1517, 1521 (10th Cir. 1984). A judgment, however, based upon speculation or supposition cannot be sustained. We have considered the standards applied in Hasler v. United States, 718 F.2d 202 (6th Cir. 1983), cert. denied, 469 U.S. 817, 105 S.Ct. 84, 83 L.Ed.2d 31 (1984). The issue there, as here, was proximate cause, and we found it necessary in Hasler to reverse the lower court's judgment rendered for the plaintiff in a swine flu vaccination case.

Hasler involved the plaintiff's claim that her swine flu inoculation caused her affliction with a type of rheumatoid arthritis known as Still's Disease. 718 F.2d at 204. The district court found for the plaintiff in Hasler because 1) she was in good health shortly before her inoculation; 2) she contracted Still's Disease just ten days after the vaccination; and 3) the experts agreed that a reaction to the vaccine could cause the affliction. Id. at 204-05. We reversed the judgment of the district court, holding:

Given these facts, to conclude that the swine flu vaccination was the cause in fact of Ms. Hasler's injuries would be mere conjecture. * * * While an antibody antigen reaction can cause rheumatoid arthritis, there is no showing that it did cause the plaintiff's disease in this case. Moreover, the plaintiff must show that any antigen reaction she may have had was a reaction to the swine flu shot.

Id. at 205 (emphasis in original).

This ruling parallels the position of our court that an expert's opinion must be based on a theory that is generally accepted in the relevant scientific community. In addition, it must not be based on speculation or surmise. Sterling v. Velsicol Chemical Corp., 855 F.2d 1188 (6th Cir. 1988); United States v. Kozminski, 821 F.2d 1186 (6th Cir.), aff'd, ___ U.S. ___, 108 S.Ct. 2751, 101 L.Ed.2d 788 (1988). This is also the standard in Ohio. State v. Thomas, 66 Ohio St.2d 518, 423 N.E.2d 137 (1981); State v. Whitman, 16 Ohio App.3d 246, 475 N.E.2d 486 (1984). In Sterling, we reversed a district court's verdict for the plaintiffs because there were no scientific studies linking pollution of area waterways to damage of the plaintiffs' immune systems. Despite the experts' opinions that pollution did cause bodily damage, we reversed on the causation question, stating:

Without the requisite clinical tests and a widely accepted medical basis for reaching its conclusions, plaintiffs' experts' opinions are insufficient to sustain plaintiffs' burden of proof that the contaminated water damaged their immune system.

855 F.2d at 1209. This position mirrors the view of the Ohio courts. E.g., Shumaker v. Oliver B. Cannon Sons, Inc., 28 Ohio St.3d 367, 504 N.E.2d 44 (1986).

Experts presented by Mrs. Novak thought that the swine flu vaccine was a probable cause of Novak's death, reasoning that there was a "discovered link" between the vaccine in question and a disease of the nervous system known as Guillain-Barre Syndrome (GBS), which causes the immune system to attack the myelin sheaths covering the nerves. They believed that the swine flu vaccine contained chemicals closely resembling myelin, and that this substance caused the body to attack its own nerves as it tried to attack the virus. The government and its experts conceded that the Center for Disease Control reported an increase in the number of reported cases of GBS for a limited period at or about the time of the swine flu vaccination program. There is, however, no direct relationship between GBS and DM/PM. See Hixenbaugh v. United States, 506 F. Supp. 461 (N.D.Ohio 1980), for a discussion concerning various side effects associated with GBS.

The plaintiff's experts submitted that, like GBS, DM/PM could be caused by a virus. They testified that analysis of those suffering from the DM/PM disorder indicates the presence of virus-like particles in their bodies, and that certain experiments performed on animals show that when injected with such particles, the animals develop symptoms resembling DM/PM. Finally, they pointed to evidence that children who were twice injected with a "killed measles vaccine" developed an irritation in their muscles, suggesting that a "killed virus" vaccine, such as swine flu, could at least theoretically cause a reaction leading to DM/PM.

In summary, based on this reasoning, and an assumption that DM/PM is caused by a virus, the plaintiff's experts testified that the swine flu vaccine could have been the cause of Novak's death. They relied heavily on the relatively short period between Novak's inoculation and development of his sickness and on Novak's previous good health to support their conclusion. The conclusion that the swine flu vaccine was a cause of Mr. Novak's condition was reached even though the experts admitted that, unlike the case of GBS, there was no evidence that the serum contained any muscle cells which could start a reaction like the myelin elements cause in cases of GBS.

The United States insists that the scientific community is not sure whether DM/PM is caused by a virus at all; it contends that there is no medical consensus or any degree of medical certainty that the swine flu virus could cause DM/PM. Indeed, no witness for the plaintiff could say with scientific or medical certainty that the particular vaccine at issue here caused Novak's disorder. It is clear, taking a favorable view of plaintiff's evidence, that no medical evidence states with assurance that any vaccine could cause DM/PM.

Dr. Allan B. Kirschner, a rheumatologist, and one of the physicians treating Mr. Novak, diagnosed Novak's medical condition and observed that it would be "conjecture" to state that the vaccine played a part in Mr. Novak's illness. Dr. Kirschner could not rule out butazolidin, a muscle relaxant prescribed to Mr. Novak, as a possible cause of Novak's condition. The Novak family physician had prescribed this drug after Novak's initial complaints about pain in his arm.

The district court made reference in its decision to testimony of Dr. Raymond Adams, Chief of Neurology at Massachusetts General Hospital in Boston, editor, author, researcher, recipient of a number of honorary degrees, and an expert presented by defendant, indicating that he "knew of cases of children developing DM after viral inoculations and injections." Dr. Adams, however, agreed with the diagnosis of Dr. Kirschner that no causal relationship existed between the swine flu vaccine and the DM/PM disorder. Dr. Adams was emphatic that the cause of DM/PM is unknown. He further testified that the "mere seeing of viral particles in muscle does not prove they are the cause of any abnormality in the muscle." He added that "the status of the viral causation of polymyositis is totally unsettled." (Emphasis added). We find that the district court had no basis to use Dr. Adams' studies, research, or testimony as supportive of any causal relationship on this critical issue in this case.

Since our decision in United States v. Green, 548 F.2d 1261 (6th Cir. 1977), we have required that in order for the testimony of experts to be admissible under Federal Rule of Evidence 702, it must conform to "generally accepted explanatory theory" accepted or recognized by the relevant scientific or medical community. In a proposed article, two of plaintiff's experts, Drs. Pearson and Lawry, theorized that DM/PM could be caused by the swine flu virus. Their article was rejected for publication by three medical journals. Dr. Adams taught Dr. Pearson and he disagreed completely with Dr. Pearson's opinion that the swine flu vaccine could cause DM/PM. Dr. Adams stated that the opinions and findings of the studies relied upon by Dr. Pearson did not support Dr. Pearson's conclusions.

Dr. Lawry acknowledged that there were no biochemical tests to determine the etiology of DM/PM, and indicated that he and Dr. Pearson agreed that there was " no evidence to support myotoxicity from a direct viral infection of muscle." (Emphasis added). Dr. Lawry conceded, moreover, that a theory of causal relationship between the swine flu vaccination and Mr. Novak's DM/PM is "not supported by any epidemiologic studies." Dr. Lawry also admitted that there was "no objective medical study or report which shows a causal association between the Swine Flu vaccine and DM or PM," and that medical science was not certain whether DM and PM are "auto-immune diseases."

Dr. Bellanti's theory regarding the origins of DM/PM was merely "suggested" by an "experimental body of information." He stated that DM/PM "may derive, most likely derives, from an immunologic basis," and he presumed from certain data that DM/PM might result from an infection or inoculation with the swine flu shot. Bellanti conceded throughout that his theory could not be stated or advanced with certainty and that no article or medical writing supported the causal connection he advocated.

We are persuaded by Sterling v. Velsicol Chemical Corp., 855 F.2d 1188 (6th Cir. 1988); United States v. Kozminski, 821 F.2d 1186 (6th Cir.), aff'd, ___ U.S. ___, 108 S.Ct. 2751, 101 L.Ed.2d 788 (1988); and United States v. Green, 548 F.2d 1261 (6th Cir. 1977), decisions of this court, that this expert evidence, advanced by plaintiff and relied upon by the district court, was not admissible in light of the totality of the medical evidence and the overwhelming proof that any specific cause of DM/PM is unknown and conjectural at best. Each of the plaintiff's experts conceded that the scientific and medical community was unsure about, and could not state with any degree of medical certainty, what causes DM/PM. In fact, although the plaintiff's experts suspect that the disorder may be caused by a virus introduced into the body through an injection, they produced no hard evidence linking DM/PM to a virus or to the swine flu shot. They agreed that when the source of a disease cannot be determined microbiologically or etiologically, the scientifically responsible method is to see whether reported cases of the disease have increased significantly following a given incident or episode. As in Hasler, there is no such support for this theory in the instant case.

During the course of trial, and before the presentation of all the evidence, the following colloquy occurred (during the testimony of Dr. Krakauer, an expert medical witness from Cleveland Clinic Foundation presented by the defendant):

THE COURT: There is uncontroverted evidence, Mr. McLaughlin, that on Thanksgiving Day Mr. Novak was declining to participate in any athletics; he declined to do furniture moving — and everybody's conclusion is in January, and I will make that as a finding of fact at this point. The onset was at Thanksgiving.

MR. McLAUGHLIN [defendant's attorney]: The onset of what, Your Honor?

THE COURT: Dermatomyositis.

MR. McLAUGHLIN: Your Honor, the evidence in the record is that —

THE COURT: I don't care what the evidence in the record is. There is uncontroverted testimony of the children and Mrs. Novak that he was complaining of weakness on Thanksgiving. He declined to do what he usually did, rearrange furniture, and he absolutely refused to go out and participate with all the cousins. That is uncontroverted.

* * * * * *

MR. McLAUGHLIN: The evidence I heard, Your Honor, is, I think that it is clear that on Thanksgiving, the testimony is that Mr. Novak had soreness, complained of soreness in his left arm.

THE COURT: Not at the point of injection, but at the lower left arm, if you recall.

MR. McLAUGHLIN: That is what the medical records say. Mrs. Novak said in the upper arm. The testimony is clear he was complaining of soreness of the left arm, that is not being disputed.

THE COURT: How do you explain his change of life style and not participating in athletic games with the children and the cousins, which he normally did on Thanksgiving Day, and the fact he felt unable to rearrange the furniture for dinner?

MR. McLAUGHLIN: That has been explained by testimony of the witness in that his left arm bothered him such he could not participate.

MR. BELLI [plaintiff's attorney]: I didn't hear.

MR. McLAUGHLIN: His left arm bothered him so much he could not participate in those activities. That is factual.

THE COURT: From there it went downhill. I would say that was the first time he complained about something that led to his eventual demise.

* * * * * *

THE COURT: On November 10 he got a swine flu shot, which had the usual sore reaction that you get from any kind of inoculation. By November 25, Thanksgiving Day, he was complaining of soreness in his entire left arm, and he was unable to participate in any athletic activities he normally did.

(Emphasis added).

Based on the entire record in this case, we are left with a definite and firm conviction that the district court erred in finding that the onset of the disease occurred in November 1976. Dr. Adams' study of the Novak medical records indicated that, to a reasonable degree of medical certainty, the onset of the DM/PM disease occurred only a few days before Novak was admitted to the hospital in January 1977, seven or eight weeks after Novak received the flu shot. Dr. Bellanti, plaintiff's principal expert witness, stated in his deposition that based upon the medical records, the onset of DM/PM was "at the end of December, 1976, or early January, 1977." Dr. Lawry, another of plaintiff's experts, testified that, based on the medical records and information Mr. Novak related to the doctors, the onset occurred in December or early January. All of defendant's medical experts testified, and the medical records themselves indicated, that the onset was very late December or early January. No doctor or medical evidence indicated that Novak was afflicted with DM/PM within fifteen days of being inoculated with the swine flu vaccine.

Mrs. Novak testified only that around Thanksgiving her husband experienced "soreness in that left arm around the area of the shot" and complained that it was difficult for him to lift things. She indicated that the next week Mr. Novak said that his arm was not getting better, that it "looked maybe a little swollen," and that "maybe I could have pulled my muscles." Despite these complaints, in early December, there was no "suggestion of going to the doctor." Mr. Novak's mother-in-law said that on Thanksgiving Day Novak complained over the telephone that he was "having trouble with his arm," and that it hurt him. Novak never notified his employer of any problems, and carried on his full business duties, including traveling, through December. Mrs. Novak indicated on the hospital admittance records that her husband became ill on January 5, 1977, and she admitted that Novak continued to play tennis during December. We are satisfied that the district court was in error in indicating this finding about the time of onset during the course of testimony in the face of these facts and the virtually unanimous medical conclusions to the contrary.

The burden is upon the plaintiff in this case to establish proximate cause, and because it involves a medical condition or illness, the plaintiff must show by a reasonable degree of medical certainty that the disease was caused by the negligence of the government and/or by its defective product. See Shumaker v. Oliver B. Cannon Sons, Inc., 28 Ohio St.3d 367, 504 N.E.2d 44 (1986). It is not enough, as the district court found in this case, to base a judgment for the plaintiff carrying such a burden on a determination that "the scientific data does not rule out a viral cause." Rather, the plaintiff must show by the preponderance of the evidence that Novak's DM/PM was indeed caused by a virus and that the causative virus or source of Mr. Novak's fatal illness was injected or introduced into Mr. Novak through the government's swine flu vaccine. We are satisfied that the plaintiff's evidence failed in this regard, despite energetic and creative efforts on the part of plaintiff's counsel to prove the necessary connections. We are left with a "firm conviction" in this respect that the district court committed error in finding the required causal link.

We find neither of the Ohio cases cited by the district court to be supportive on the issue of proximate cause involved in this case. There is no allegation here of a tainted vaccine or of a product "defective" in the sense of heavy machinery as analyzed under Ohio strict liability law.

For the reasons stated, including an erroneous finding of fact about the time of the onset of the fatal disease, we conclude that the district court's finding of proximate cause as a matter of reasonable medical or scientific certainty was clearly erroneous. The medical theories about a direct connection between the flu shot and DM/PM advanced by Drs. Bellanti, Pearson, and Lawry are neither "widely accepted" nor "generally accepted" by the medical community.

For expert testimony to be admissible under Rule 702, a four-part test must be met: (1) a qualified expert; (2) testifying on a proper subject; (3) in conformity to a generally accepted explanatory theory; (4) the probative value of which outweighs any prejudicial effect. United States v. Green, 548 F.2d 1261, 1268 (6th Cir. 1977) (emphasis supplied); United States v. Smith, 736 F.2d 1103, 1105 (6th Cir.), cert. denied, 469 U.S. 868, 105 S.Ct. 213, 83 L.Ed.2d 143 (1984).

United States v. Kozminski, 821 F.2d 1186, 1194 (6th Cir.), aff'd, ___ U.S. ___, 108 S.Ct. 2751, 101 L.Ed.2d 788 (1988); see also Sterling v. Velsicol Chemical Corp., 855 F.2d 1188 (6th Cir. 1988); State v. Thomas, 66 Ohio St.2d 518, 423 N.E.2d 137 (1981); State v. Whitman, 16 Ohio App.3d 246, 475 N.E.2d 486 (1984). Even if the theories of plaintiff's experts were admissible, taking their proof and testimony as a whole, they indicate that no objective scientific tests or adequate studies clearly support the thesis that they advocate. As in Hasler, 718 F.2d at 206, we find "any causal connection between the swine flu inoculation" and Mr. Novak's DM/PM condition to be "merely conjectural."

B. Adequacy of Warning.

There is also a serious controversy in this case concerning the nature of the warning given to those inoculated with the swine flu vaccine. Dr. Bellanti agreed with the government's position that in 1976, when the vaccine was administered to Mr. Novak, there was no medical basis for a requirement to warn those taking the swine flu vaccine of a specific potential risk of myositis, which would include the DM/PM disorder.

The district court, however, concluded: [T]he warning was inadequate as to the decedent because of his prior `auto-immune' viral encephalitis. The failure of the warning was a proximate cause of Mr. Novak's taking the swine flu shot, and constituted negligence under applicable Ohio law. . . .

The district court then found negligence on the part of the United States due to an inadequate warning.

Dr. Bellanti conceded that in light of scientific knowledge at the time of the swine flu vaccination program there was no duty to warn about the potentiality of myositis. He also agreed that the notice or warning reflected in the consent form signed by Mr. Novak was adequate. The district court based its conclusion of inadequacy on the government's failure to alert persons to a danger or special risk if they previously had suffered from viral encephalitis. Here again, there is simply insufficient medical evidence in the record to support the district court's deduction that there was some proximate causal relationship between the fifteen-year-old episode of viral encephalitis suffered by Mr. Novak, the flu shot, and the development, after a matter of several weeks, of the disorder which brought about Mr. Novak's death.

A consent form furnished by the government and signed by Mr. Novak at the time he was inoculated included the following warning, in relevant part, of possible or potential hazard:

Possible Vaccine Side Effects

Most people will have no side effects from the vaccine. However, tenderness at the site of the shot may occur and last for several days. Some people will also have fever, chills, headache, or muscle aches within the first 48 hours.

Special Precaution

As with any vaccine or drug, the possibility of severe or potentially fatal reactions exists. However, flu vaccine has rarely been associated with severe or fatal reactions. In some instances people receiving vaccine have had allergic reactions. You should note very carefully the following precautions:

* * * * * *

People with known allergy to eggs should receive the vaccine only under special medical supervision.

People with fever should delay getting vaccinated until the fever is gone.

People who have received another type of vaccine in the past 14 days should consult a physician before taking the flu vaccine.

If you have any questions about flu or flu vaccine, please ask.

Mr. Novak's signature on this consent form acknowledged his opportunity to inquire and the risk involved.

Although acknowledging that the advisory form signed by Mr. Novak "warned of the `possibility of severe or even fatal reactions'," the district court found it inadequate because it did not refer "to other auto-immune reactions or viral infections such as encephalitis." For the reasons stated in our discussion on proximate cause, we find that the district court erred in predicating a finding of negligence on the defendant's part because of a failure to refer to "auto-immune reactions," "viral infections" and/or "encephalitis." There was no adequate admissible medical proof showing a causal relationship between these conditions and DM/PM. There was, therefore, no reason for the defendant to make the references deemed important or vital by the district court, and there was no duty on the part of the defendant to warn about any of these conditions. Particularly, the district court erred in finding the warning inadequate, negligent, and insufficient because it did not specifically caution those who may have experienced "viral encephalitis." It was not proven that Mr. Novak had actually suffered from viral encephalitis, and no medical opinion identified any connection between "viral encephalitis" and DM/PM.

We conclude that the circumstances of this case do not give rise to any presumption that the failure of the consent form to allude to viral encephalitis or auto-immune reactions directly caused Mr. Novak's illness and death. The only case cited by the plaintiff in support of the district court's holding about the adequacy of the warning, Seley v. G.D. Searle Co., 67 Ohio St.2d 192, 423 N.E.2d 831 (1981) holds, among other things, that

[a] plaintiff asserting strict liability claims based on failure to provide adequate warnings not only must convince the fact finder that the warning provided is unreasonable, hence inadequate, but he also must establish the existence of proximate cause between the drug and the fact of the plaintiff's injury.

423 N.E.2d at 838 (footnote omitted).

It has already been established in our discussion on proximate cause that the inoculation of Mr. Novak with the swine flu vaccine was not proximately and directly related to his becoming ill with DM/PM. The warning was not demonstrated by medical evidence to be inadequate, but even if Mr. Novak had been inadequately advised by the form he signed, the plaintiff has still failed Seley's second prong, causation. See also Basko v. Sterling Drug, Inc., 416 F.2d 417, 426 (2d Cir. 1969) ("no duty to warn of unknown or unforeseeable risks"); Chambers v. G.D. Searle Co., 441 F. Supp. 377, 381 (D.Md. 1975) (duty to warn measured on date drug used), aff'd, 567 F.2d 269 (4th Cir. 1977).

Finally, we conclude that the warning in question was legally sufficient under the circumstances and that its reasonableness, or lack thereof, is not a basis upon which plaintiff may recover.

We accordingly REVERSE the judgment of the district court and direct entry of a judgment for defendant for the reasons indicated.