Charlotte Chaney appeals from a judgment entered after a jury verdict for Smithkline Beckman Corporation on a products liability claim arising from the death of her husband. She claimed that her husband's fatal stomach cancer was caused by his ingestion of the drug Tagamet, an ulcer medication manufactured by Smithkline. She challenges a partial summary judgment and numerous discretionary rulings. We affirm.

John Chaney was prescribed Tagamet for stomach ulcers and took the drug from July 18, 1979 to December 8, 1980. He died from stomach cancer on October 26, 1981. Charlotte Chaney brought this action, alleging that Smithkline failed to warn physicians that Tagamet caused cancer or that the drug masked cancer symptoms, precluding timely diagnosis. Chaney claimed that these failures to warn proximately caused her husband to consume Tagamet and die. On the basis of Smithkline's motion in limine, which was treated as a motion for partial summary judgment, and following a pretrial hearing, the district court1 refused to let the jury consider whether Tagamet caused Chaney's cancer.2 The case was then tried to the jury on the masking issue only. A verdict was directed for defendant Skaggs Companies, Inc., and the jury found in favor of the other defendants.

The critical issue in this appeal is the summary judgment on the cancer causation issue. The district court based its ruling on depositions of Chaney's experts. The first, Dr. Martins, was asked whether he was prepared to testify under oath that Tagamet caused Mr. Chaney's cancer. He responded: "No. I'm not because there has been no proof that carcinoma is caused by Tagamet." Martins Deposition at 28. The second, Dr. Proctor, testified that there was a "strong possibility" that Tagamet caused the stomach cancer. When he was asked to quantify the possibility, he answered that the "probability that [Tagamet] was responsible for Mr. Chaney's disease is probably greater than 20 percent and probably less than 80 percent."3 Proctor Dep. at 6, 26. On the basis of these depositions, the district court observed:

What we have is Dr. Proctor saying that I can't say whether it is more probable than not. I estimate the probabilities. In other words, it's a double set of probabilities. He is now estimating the probabilities as being somewhere in the range of 20 to 80 percent. With that type of a showing, plaintiff would like to say to the jury, "Although the expert opinion in this field has not jelled, that nobody is yet willing to say that it is more likely so than not, or that it meets the medical probability test of the Arkansas law, that nevertheless you may so find."

And that is particularly inappropriate where the lay jury--the whole technical information is outside the knowledge of the lay jury.

T. 39-40.

In reviewing a summary judgment, the question is whether there is a genuine issue of material fact and the moving party is entitled to judgment as a matter of law. Fed.R.Civ.P. 56(c). The primary issue here is whether there was a sufficient showing that Smithkline's failure to warn proximately caused Chaney's death, or more specifically, whether Tagamet caused the stomach cancer. We treat the summary judgment as equivalent to a directed verdict. See generally 10 C. Wright, A. Miller & M. Kane, Federal Practice and Procedure Sec. 2713.1 (1983) (comparison of the summary judgment motion with the directed verdict motion). Thus, the question is whether Chaney presented a submissible issue of causation. The judgment must be sustained if all the evidence points one way and is susceptible of no reasonable inferences sustaining Chaney's position. See Brown v. Syntex Laboratories, 755 F.2d 668, 671 (8th Cir.1985) (federal standard); see also Dobson v. Bacon Transport Co., 607 F.2d 805, 807 n. 1 (8th Cir.1979) (equating federal and Arkansas directed-verdict standards). The most favorable characterization of Chaney's proof is that there is a 20-80% probability that Tagamet causes cancer. The jury, however, is not permitted to speculate about the substantive elements of a case. Expert medical testimony to the effect that a causal connection between an occurrence and death or injury was merely "possible" is insufficient to take a case to the jury. "[S]uch testimony leaves the issue in the field of conjecture and permits the jury to speculate or guess as to the cause of death." Chicago Great Western Ry. v. Smith, 228 F.2d 180, 182 (8th Cir.1955) (quoting Bearman v. Prudential Insurance Co., 186 F.2d 662, 665 (10th Cir.1951)); see Norland v. Washington General Hospital, 461 F.2d 694 (8th Cir.1972). The "strong possibility" or 20-80% probability characterization invites such speculation.

Moreover, this is not a case in which the jury could rely on common learning to narrow the range of possibilities. The subject of carcinogens is highly complex. Proctor's unwillingness to narrow the probability indicates the uncertainties that plague even the experts in this area. Notwithstanding the expert's vast learning, he could not diagnose the issue with greater certainty. To ask jurors who lack scientific training to make a more precise diagnosis would be folly. Such a request would invite the jury to simply guess where the causation probability fell in the 20-80% range. Thus, the district court did not err in granting summary judgment on this issue.

Chaney also complains about a number of rulings by the district court, including the denials of motions to compel discovery and for a continuance, a host of evidentiary rulings, and the failure to instruct the jury on the loss of earnings from an investment business and punitive damages. Many of these claims of error are not supported by references to the transcript as required by Federal Rule of Appellate Procedure 28(e). See United States v. Cohen, 738 F.2d 287, 289-90 (8th Cir.1984); Rebuck v. Vogel, 713 F.2d 484, 487 (8th Cir.1983). Insofar as we can determine the bases for these arguments, we cannot conclude that there has been an abuse of discretion. With respect to denial of the instruction on punitive damages, Chaney points to no evidence demonstrating that Smithkline knew or ought to have known that its conduct would naturally or probably result in injury and that it recklessly continued such conduct. These are the findings required by Arkansas substantive law on punitive damages. See Freeman v. Anderson, 279 Ark. 282, 651 S.W.2d 450 (1983). Thus, the court properly refused to give a punitive damages instruction. Similarly, the refusal of certain other instructions has not been shown to be error.

The judgment is affirmed.