On December 2, 1971, Harrikah I. Stanton, then eight months old, suffered an adverse reaction to Xylocaine, a local anesthetic, and experienced cardiac and respiratory arrest, resulting in severe and irreversible brain damage. Ruby Brooks, Harrikah's mother, brought this diversity negligence and products liability action on behalf of herself and her daughter (the "plaintiffs") against Xylocaine's manufacturer, Astra Pharmaceutical Products, Inc. The suit alleged, inter alia, that Astra had been negligent in failing to file with the Food and Drug Administration (the "FDA") certain reports concerning Xylocaine, as required by federal statutes and regulations, and that the failure to file the reports also rendered Xylocaine a defective product within the meaning of Restatement (Second) of Torts Sec. 402A. Mrs. Brooks further asserted that the failure to file and the defective nature of the drug were substantial factors in causing her daughter's injuries. After an eight-week trial, the jury found Astra liable to plaintiffs on the above bases and awarded compensatory but not punitive damages. When the Pennsylvania Supreme Court subsequently changed the law regarding calculation of damages, however, the district court granted a new trial limited to computation of compensatory damages. Plaintiffs obtained a jury verdict of $2,367,032, on which judgment was entered.

Astra appeals from that judgment, claiming (a) that the federal statutes and regulations asserted as the basis for liability did not apply to Xylocaine and that, at all events, the evidence was insufficient to support the verdict on liability; (b) that the district court erred in resubmitting "special questions" on liability to the jury after the jury had returned the questionnaire containing answers inconsistent with its award of damages; (c) that the court abused its discretion in granting a new trial limited only to damages; and (d) that the court erred in denying Astra's motion for a mistrial based on prejudicial comments by plaintiffs' counsel in his opening statement during the retrial on damages. On cross-appeal, plaintiffs contend that several evidentiary rulings by the district court allegedly prejudiced their ability to prevail on their claim for punitive damages.

For the reasons that follow, we conclude that Astra did have an obligation to comply with the relevant statutes and regulations and that, although the question of causation is extremely close, plaintiffs introduced sufficient evidence to support the verdict in their favor. We also conclude that the district court acted within its discretion in resubmitting the "special questions" to the jury. However, because we hold that the court abused its discretion in granting a new trial limited only to damages, we will vacate the judgments as to liability and damages and remand for a new trial on both issues. Because several of the procedural and evidentiary questions raised in these appeals are likely to recur on retrial, we address those matters as well.

Harrikah Stanton was an eight-month-old infant on December 2, 1971, when she entered Harrisburg Hospital and submitted to a bone-marrow test to determine the cause of the hemolytic anemia from which she had suffered since birth. In performing the test, the hematologist, Dr. Herbert S. Bowman, injected a two-percent solution of Xylocaine into Harrikah's right posterior iliac crest to anesthetize the area from which he would aspirate bone marrow. The Xylocaine, known generically as lidocaine hydrochloride, was a local anesthetic manufactured by Astra.

Shortly after the procedure, Harrikah began convulsing and experienced cardiac and respiratory arrest. Dr. Marita Fabian, a senior resident at the hospital, and other hospital employees attempted to resuscitate her, but to little avail. The cardiac arrest resulted in severe and irreversible brain damage. Harrikah cannot walk, talk, or stand; her development has not progressed beyond that of a three- to four-month old child. She will require constant care for the rest of her life.

On October 24, 1973, plaintiffs commenced this negligence and product-liability action against Astra, Harrisburg Hospital, and Drs. Bowman and Fabian.1 Plaintiffs pressed a host of claims against Astra, charging that the company had been negligent in (a) failing adequately to warn Dr. Bowman and the physician-employees at Harrisburg Hospital of a possible toxic reaction to Xylocaine, even in a proper pharmacological dose, when administered to a child of Harrikah's age; (b) failing adequately to warn Dr. Bowman and the Hospital's physician-employees of the need for immediate availability of resuscitative equipment and short-acting barbiturates when Xylocaine is administered; (c) "overpromoting" Dr. Bowman and the physician-employees so that they no longer heeded the warnings actually given; (d) failing adequately to warn Dr. Bowman and the physician-employees that Xylocaine should not be administered in a two-percent solution to an eight-month-old child; (e) failing to file annual reports and unexpected-adverse-reaction reports with the FDA, as required by 21 C.F.R. Sec. 130.35(e) and (f) (1972);2 (f) marketing a product that was defective because the manufacturer's failure to file the adverse-reaction reports pursuant to section 130.35(f) deprived the FDA of the information it needed to make an informed judgment concerning the conditions under which Xylocaine safely could be marketed; (g) failing to provide Dr. Bowman with adequate directions for computing a safe dosage of Xylocaine for an eight-month old infant; (h) failing to conduct studies adequate to determine what in fact would be a safe dosage for a child of Harrikah's age; and (i) failing to exercise reasonable care to monitor and investigate fully the adverse reactions associated with Xylocaine in order to determine the risks involved in the use of that drug.

After an eight-week trial, the case was submitted to the jury on "special questions." On January 10, 1980, the jury returned a verdict against Astra alone.3 In response to the "special questions," see infra Part III, the jury found that Harrikah had suffered an adverse reaction to Xylocaine; that Astra had acted negligently in failing to file the annual and adverse-reaction reports required by subsections 130.35(e) and (f); that the failure to file these reports rendered Xylocaine a defective product; and that Astra's negligence and the product's defective nature were substantial factors in causing Harrikah's injury. The jury also found that Astra negligently had failed both to conduct adequate dosage studies and to monitor and investigate fully the clinical adverse-reaction experiences, but it did not deem this negligence to have been a substantial factor in causing the child's injuries. Finally, the jury rejected plaintiffs' claims that Astra had failed to issue adequate warnings to users of Xylocaine; that Astra had "overpromoted" those users; and that Astra had failed to provide adequate directions to the users. The jury awarded Mrs. Brooks $60,000 in her own right and $255,000 on behalf of her daughter but refused to award punitive damages, finding that Astra had not acted "with a bad motive or with reckless indifference to the interests of persons who might be administered the drug." The district court entered judgment on the verdict on January 11, 1980.

Before the district court had ruled on post-trial motions, the Pennsylvania Supreme Court decided the case of Kaczkowski v. Bolubasz, 491 Pa. 561, 421 A.2d 1027 (1980), which held that a jury may consider lost future productivity in awarding damages for future losses and that such losses should not be discounted to present value. In the wake of this change in Pennsylvania law, the district court granted plaintiffs' motion for a new trial on compensatory damages but denied it in all other respects.4 Stanton v. Astra Pharmaceutical Products, Inc., Civ. No. 73-610 (M.D.Pa. Oct. 19, 1981) ("Stanton I "). The court denied Astra's motion in toto. Stanton v. Astra Pharmaceutical Products, Inc., Civ. No. 73-610 (M.D.Pa. Oct. 19, 1981) ("Stanton II ").5

The new trial on compensatory damages resulted in a jury verdict awarding Mrs. Brooks $256,000 in her own right and $1,652,371 as Harrikah's parent and guardian. The court awarded damages for delay and molded the award to $2,367,032, pursuant to Pa.R.Civ.P. 238.6 On April 14, 1982, Astra moved for a new trial, claiming that the district court had erred (a) in refusing to dismiss the panel of jurors after the court made allegedly improper remarks in its introductory statement to the panel, and (b) in refusing to declare a mistrial following allegedly improper and prejudicial statements by plaintiffs' counsel in his opening and closing arguments.7 The court denied Astra's motion, Stanton v. Astra Pharmaceutical Products, Inc., Civ. No. 73-610 (M.D.Pa. June 28, 1982) ("Stanton III "), and Astra appealed. Plaintiffs filed a cross-appeal on August 6, 1982, challenging the court's denial of their motion for a new trial on punitive damages.

We now turn to the merits of the various claims raised on appeal, focusing primarily on those relating to the sufficiency of the evidence to support the verdict on liability, the court's resubmission to the jury of the "special questions" used during the first full trial, and the propriety of the granting of a new trial on damages alone. In a final section, we will consider more summarily the propriety of the remarks made by plaintiffs' counsel as well as plaintiffs' claims of error relating to the denial of punitive damages.8

The verdict in plaintiffs' favor was predicated upon the jury's conclusion that Astra had acted improperly when it failed to file with the FDA the annual and adverse-reaction reports required by 21 C.F.R. Sec. 130.35(e) and (f) (1972). Astra initially argues that section 130.35 does not apply to Xylocaine and that the company therefore had no obligation to file any of the specified reports. But even assuming that the regulation does apply, Astra contends, the company did not act negligently when it decided not to file the requisite reports, and the failure to file therefore did not render Xylocaine defective or unreasonably dangerous. Moreover, Astra argues, even if a jury could find that the company either had acted negligently in failing to file the reports or had marketed a defective product, neither the negligence nor the section-402A violation could have been a substantial factor in causing Harrikah's injuries.

We first will consider the question of the applicability of section 130.35; in the course of our discussion, we will outline the rather complex statutory and regulatory schema governing marketed drugs. Because we conclude that Astra had a duty to file the reports required by section 130.35, we then will proceed to the consequences of noncompliance. The jury determined that noncompliance with the FDA's regulations engendered liability based on negligence as well as strict liability under Restatement (Second) of Torts Sec. 402A (1965). We will discuss separately each basis of liability. We conclude that the jury had sufficient evidence to find that Astra's conduct was negligent per se and that such conduct proximately caused Harrikah's injuries; we further conclude that the jury could find that Xylocaine was a defective product within the meaning of section 402A and that the defective product also was a proximate cause of the harm.9

In 1938, Congress enacted the Federal Food, Drug and Cosmetic Act, which prohibits the introduction into commerce of any "new drug" for which a new-drug application has not been filed with and approved by the FDA.10 See Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 612-15, 93 S. Ct. 2469, 2474-76, 37 L. Ed. 2d 207 (1973). Pursuant to this Act, Astra submitted to the FDA in 1948 an application for approval of its new anesthetic, Xylocaine. The FDA approved Astra's application on November 19, 1948; marketing of Xylocaine in 0.5%, 1%, and 2% solutions commenced in 1949.

The Thalidomide tragedies of the early 1960s triggered the next wave of drug-related legislation: the Drug Act Amendments of 1962, 76 Stat. 780 (1962). It is these amendments with which we are principally concerned. Central to the amendments was section 103(a), which provides in relevant part:

In the case of any drug for which an approval of an application filed pursuant to this section is in effect, the applicant shall establish and maintain such records, and make such reports to the Secretary, of data relating to clinical experience and other data or information, received or otherwise obtained by such applicant with respect to such drug, as the Secretary may by general regulation, or by order with respect to such application, prescribe on the basis of a finding that such records and reports are necessary in order to enable the Secretary to determine, or facilitate a determination, whether there is or may be ground for invoking subsection (e) of this section ....11

21 U.S.C. Sec. 355(j)(1) (1976).12 Pursuant to this authorization, the Secretary promulgated regulations on May 28, 1964. These regulations, as we noted above, require companies marketing registered drugs to submit to the FDA annual reports, 21 C.F.R. Sec. 130.35(e) (1971),13 as well as immediate reports of all unexpected adverse reactions to those drugs, id. Sec. 130.35(f).14

Between May 28, 1964, and the end of 1970, Astra received 202 reports of adverse reactions allegedly related to Xylocaine.15 These reactions ranged from minor, temporary effects to death. Yet Astra forwarded none of these reports to the FDA, relying upon the advice of its counsel, Alan H. Kaplan, that Xylocaine was not a "new drug"16 and therefore was exempt from the reporting requirements embodied in 21 C.F.R. Sec. 130.35(e) and (f).17 However, the company prepared the reports and made them a part of the Establishment Inspection Reports seen by non-medical FDA inspectors who came to inspect Astra's facilities between 1964 and 1971.

In deciding not to file the reports ostensibly required by section 130.35, Astra also relied on a letter, dated October 2, 1963, from John F. Palmer, M.D., the Chief of the Department of Health, Education and Welfare's ("HEW's") New Drug Status Branch, in which Dr. Palmer advised Astra that two-percent Xylocaine was "not now regarded" by the Division of Drugs as a "new drug" within the meaning of 21 U.S.C. Sec. 321(p), see supra note 16. App. at A559. On April 25, 1968, however, Arthur M. West, M.D., Acting Director of HEW's Division of Surgical Dental Drugs and Adjuncts, wrote to Astra and requested that the company file all reports prescribed in regulations promulgated pursuant to 21 U.S.C. Sec. 355(j).18

Upon receipt of this letter, Astra wrote back and reminded the FDA of both the 1963 letter from Dr. Palmer and the FDA's announcement on September 12, 1964, postponing the effective date of Sec. 130.35(b)'s reporting requirements, see 29 Fed.Reg. 12,872 (1964). Dr. West responded in July 1969 by referring Astra to a statement published in the Federal Register on May 20, 1968, in which the FDA formally had revoked "all opinions previously given by the [FDA] to the effect that an article is 'not a new drug' or is 'no longer a new drug.' "19 Dr. West further informed Astra that, "since these drugs are presently under review by the National Academy of Sciences-National Research Council ["NAS-NRC"], we are deferring any final decision on the new drug status until the findings of the review are published."20 Letter from Arthur M. West, M.D., to Astra (July 30, 1969), App. at A572. Thus, Astra knew, or should have known, by May 20, 1968, three and one-half years before Harrikah Stanton's adverse reaction, that the FDA deemed Xylocaine to be a "new drug" for purposes of compliance with the reporting requirements of 21 U.S.C. Sec. 355(j) and 21 C.F.R. Sec. 130.35.

Despite the above exchanges between Astra and the FDA, Astra contends that 21 U.S.C. Sec. 355(j) and 21 C.F.R. Sec. 130.35(f) apply only to "new drugs," see supra note 16 (defining "new drug"), and not to "old drugs," or drugs that are generally recognized by experts as safe and effective; Astra further claims that Xylocaine had achieved such recognition by 1964. The district court disagreed and, in a pretrial order, granted plaintiffs' partial motion for summary judgment "to the extent that the court concludes that the regulation 21 C.F.R. Sec. 130.35 ... was applicable to Astra ... as sponsor of the drug Xylocaine." Stanton v. Astra Pharmaceutical Products, Inc., Civ. No. 73-610 (M.D.Pa. Oct. 10, 1979) (Order).

We agree with the district court's ruling, for the very language of section 130.35 does not admit of the exclusion urged by Astra. Section 130.35 applies to "[e]ach applicant for whom a new-drug application or supplement ... became effective or was approved at any time prior to June 20, 1963 ...." 21 C.F.R. Sec. 130.35(a), (b) (emphasis added). This language appears to encompass Xylocaine, which was approved by the FDA in 1948. Moreover, application of section 130.35 to a drug approved in 1948 would seem essential in light of the FDA's statutory obligation continually to monitor drugs already on the market in order to determine whether to withdraw FDA approval pursuant to 21 U.S.C. Sec. 355(e).21

Astra relies both on the 1963 letter from the FDA, see supra, in which the agency advised Astra that it did not consider Xylocaine to be a new drug, as well as on the FDA's 1964 suspension of the effectiveness of certain regulations pending the resolution of a court challenge, see 29 Fed.Reg. 12,872 (1964). But neither the 1963 letter nor the 1964 suspension can avail Astra. While we concede that Astra initially was entitled to rely on the 1963 letter, the FDA effectively nullified the assertions contained in that letter when it announced in 1968, well before Harrikah Stanton's bone-marrow test in 1971, that "all opinions previously given by the [FDA] to the effect that an article is 'not a new drug' or is 'no longer a new drug' are hereby revoked," 33 Fed.Reg. 7758 (1968), see supra note 19. And as for the 1964 suspension: that declaration stayed the enforcement of only 21 C.F.R. Sec. 130.35(b), without any mention of subsections 130.35(e) and (f). Moreover, Astra does not appear to contend that it filed with the FDA the information required in order to qualify for the postponement of section-130.35(b) obligations.

Astra also argues that it had no obligation to file the adverse-reaction reports with the FDA because section 130.35(f) applies only to unexpected adverse reactions, and the reactions reported to Astra were not unexpected. The regulations define "unexpected" as

conditions or developments not previously submitted as part of the new-drug application or not encountered during clinical trials of the drug, or conditions or developments occurring at a rate higher than shown by information previously submitted as part of the new-drug application, or than encountered during such clinical trials.

21 C.F.R. Sec. 130.13(b)(2)(i). Thus, Astra would argue that the FDA already had been apprised of all relevant information and that it was not necessary to add to that data base.

This argument must fail as well. Astra itself contended before the district court that the character of those reactions was a question of fact for the jury. Despite the grant of summary judgment in favor of plaintiffs as to the applicability of section 130.35(f), the district court clearly submitted the question of expectedness to the jury, Transcript, vol. 30, at 114-15, and the jury resolved it adversely to Astra in finding that Astra should have filed the reports. In examining the district court's denial of Astra's motion for judgment n.o.v., we must view the record in this case in the light most favorable to plaintiffs and affirm the judgment unless the record lacks even that minimum quantum of evidence on which a jury reasonably might have based its verdict. Black v. Stephens, 662 F.2d 181, 187-88, 190 (3d Cir.1981), cert. denied, 455 U.S. 1008, 102 S. Ct. 1646, 71 L. Ed. 2d 876 (1982). We certainly cannot say that no rational jury could have believed the reports received by Astra to have related to unexpected adverse reactions. Moreover, it does not appear from the record that Astra previously had informed the FDA of the full import of the adverse reactions reported to Astra during the 1960s.

Accordingly, we conclude that the district court did not err in ruling that subsections 130.35(e) and (f) applied to Astra; to the extent that the jury decided, in reaching its finding of liability, that the unreported adverse reactions were "unexpected," its verdict is supported. We now turn to the bases on which the jury found Astra liable.

1. Was Astra Negligent?

Under Pennsylvania law, the violation of a governmental safety regulation constitutes negligence per se if the regulation "was, in part, intended to protect the interest of another as an individual [and] the interest of the plaintiff which was invaded ... was one which the act intended to protect."22 Majors v. Brodhead Hotel, 416 Pa. 265, 268, 205 A.2d 873, 875 (1965); accord Frederick L. v. Thomas, 578 F.2d 513, 517 (3d Cir.1978); Hunziker v. Scheidemantle, 543 F.2d 489, 497 (3d Cir.1976); Millard v. Municipal Sewer Authority, 442 F.2d 539, 541 (3d Cir.1971); D'Ambrosio v. City of Philadelphia, 354 Pa. 403, 405-07, 47 A.2d 256, 258-59 (1946). Astra cannot seriously dispute that section 130.35 was promulgated to protect individuals such as Harrikah Stanton from precisely the type of harm that here occurred--an unexpected adverse reaction to Xylocaine. It thus would appear that Astra's failure to file the reports constituted negligence per se.23

Astra points out, however, that noncompliance with safety regulations does not result inexorably in a finding of negligence because Pennsylvania law appears to follow the Restatement (Second) of Torts in recognizing a class of "excused violations."24 In particular, Astra argues that its failure to comply with the applicable statutes and regulations is excused by three circumstances: (1) Astra neither knew nor should have known of the need to comply; (2) Astra's counsel advised the company that there was no need to file the reports with the FDA; and (3) the reports actually had been prepared and were available to FDA plant inspectors. We agree that Pennsylvania law allows a defendant to offer excuses for a statutory or regulatory violation; however, the record permitted the jury to reject each of the excuses proffered by Astra.

First, even assuming that Astra at one time had reason to believe that 21 C.F.R. Sec. 130.35 did not apply to Xylocaine, the FDA formally revoked in 1968 all previous opinions as to new-drug status and expressly informed Astra in 1969 that the FDA expected compliance with section 130.35. Thus, at least as of 1968, Astra knew, or should have known, that the FDA expected Astra to file the reports prescribed by section 130.35.

Second, the mere fact that Astra's attorneys interpreted section 130.35 incorrectly does not negate Astra's negligence in failing to comply with the regulations. Astra took a chance, and it is liable for the consequences of its acts. And even assuming that advice of counsel does constitute a legally cognizable excuse, that excuse could insulate Astra from liability only between the time that the FDA promulgated the regulations in 1964 and the time that the agency revoked all private letter-rulings as to new-drug status in 1968. Astra still had three and one-half years before the December 1971 tragedy to file the reports.

Nor can we accept the availability of the reports to plant inspectors as an excuse for Astra's conduct. Plant inspectors are not physicians and do not possess the expertise necessary to facilitate a full evaluation of the reports. Indeed, the jury heard testimony from Dr. Alan K. Done, a Professor of Pediatrics and Pharmacology who had served as Special Assistant to the Director of the FDA's Bureau of Drugs, that "it's not likely that [FDA field investigators] would go and look at those reports." Transcript, vol. 6, at 78. Moreover, the statute requires that manufacturers file reports with the FDA, 21 U.S.C. Sec. 355(j)(1), and make them available to inspectors, id. Sec. 355(j)(2), see supra note 12. It is no answer to say that compliance with section 355(j)(2) excuses a breach of section 355(j)(1).

Under Pennsylvania law, therefore, this case falls within section 288B(1) of the Second Restatement: "The unexcused violation of a legislative enactment or an administrative regulation which is adopted by the court as defining the standard of conduct of a reasonable man, is negligence in itself." The jury therefore had sufficient evidence upon which to find that Astra had acted negligently in failing to file the reports required by section 130.35.25

2. Did the Negligence Proximately Cause the Harm?

That Astra was negligent in failing to file the reports is not in itself sufficient to sustain the finding that Astra was liable. The negligence must also have been a proximate cause of Harrikah Stanton's injury.

It is well established in Pennsylvania that in order to find that defendant proximately caused an injury it must be found that his allegedly wrongful conduct was a substantial factor in bringing about plaintiff's injury even though it need not be the only factor. [Citations omitted.] It is equally well established that defendant's negligent conduct is not a substantial factor in bringing about plaintiff's injury if it would have been sustained even if the actor had not been negligent.

Majors v. Brodhead Hotel, supra, 416 Pa. at 271-72, 205 A.2d at 877; accord Hunziker v. Scheidemantle, supra, 543 F.2d at 496, 498; Greiner v. Volkswagenwerk Aktiengesellschaft, 429 F. Supp. 495, 496-97 (E.D.Pa.1977); Kaplan v. Philadelphia Transportation Co., 404 Pa. 147, 149-50, 171 A.2d 166, 167-68 (1961); Restatement (Second) of Torts, supra, Secs. 431-32. The next question, then, is whether plaintiffs introduced sufficient evidence to support the jury's determination that Astra's failure to file the annual and adverse-reaction reports had been a substantial factor in causing Harrikah's injury.26

In order to establish the requisite causal link between Astra's tortious conduct and Harrikah's injury, plaintiffs relied heavily on the testimony of Drs. Alan K. Done, Henry L. Price, Marvin Seife, and John Adriani.

Dr. Done, at the time of the trial, was a Professor of Pediatrics and Pharmacology and Adjunct Professor of Pharmacy at Wayne State University, as well as Director of the Division of Clinical Pharmacology and Toxicology at Children's Hospital of Michigan. Author of 300 articles, Dr. Done was certified by the American Board of Pediatrics and the American Board of Medical Toxicology. From 1972 to 1975, he served as Special Assistant to the Director of the FDA's Bureau of Drugs. Dr. Done testified that he had reviewed the adverse-reaction reports received by Astra between 1964 and 1970 and had found descriptions of twenty-two fatalities, thirteen near-fatalities,27 twenty-five life-threatening incidents, and forty-five "unknown or nil life threat" instances among the medical cases. Transcript, vol. 4, at 169-70. The dental cases revealed four fatalities, four near-fatalities, thirty-five life-threatening cases, and 152 "unknown or nil threat to life" cases. Id. at 170. Dr. Done observed that the more severe reactions and problems had not necessarily arisen during the more complicated medical or dental procedures; in fact, "[t]he bulk of cases [involved] actually trivial procedures, relatively trivial procedures," id. at 170-71.

Plaintiffs' counsel questioned Dr. Done about the issue of causation:

Q. Now Doctor, based on your review of xylocaine's adverse reaction experience as reported in the yellow books, do you have an opinion as to whether or not the information Defendant, Astra received if not reported to the FDA would bias an evaluation of the safety of xylocaine?

....

A. For clarification, do you mean by received, those that they received as physician or pharmacist complaints?

Q. Yes, sir.

....

A. With that clarification my opinion is that this would seriously hamper the ability of a regulatory agency to carry out its responsibility.

Id. at 145-46. Later, Dr. Done testified that, "as part of the issue of reacting adequately to their own adverse reactions, action was called for earlier than it was taken. Action either by the company or by FDA. The FDA was denied that opportunity by virtue of being denied that information." Id., vol. 5, at 158. Dr. Done reiterated this point still one more time:

Q. Now, if the adverse reaction reports that have been introduced into evidence and to which you had testified, if those reports had been filed by Astra prior to December 2nd, 1971 would there have been an increased chance of some investigation by the FDA as opposed to the fact that one field investigator may have seen the reports at the Astra plant?

....

[A.] The question [sic] is absolutely yes.

Dr. Done also explained two ways in which knowledge of Xylocaine's propensities could have been disseminated to the medical community, had Astra filed the requisite reports. First, Dr. Done discussed the "package insert," a sheet of instructions and warnings that manufacturers package together with the drug. Dr. Done testified that the FDA may order changes in the package inserts; these changes would inform physicians of new or previously unreported developments relating to Xylocaine. Id., vol. 5, at 134-35. The FDA must also approve the package inserts. Id. at 135.

The second means of dissemination would have been via the Physicians' Desk Reference ("PDR"), a manual containing information about thousands of marketed drugs and a basic reference tool for physicians. The PDR is supplemented annually and contains the "minimum" information needed for safe and effective administration of a drug. Id. at 161-63; see also id., vol. 9, at 73 (PDR is one of the best ways for manufacturers to communicate with doctors) (Testimony of Dr. Henry L. Price). According to Dr. Done, the PDR cannot contain any information that differs significantly from that approved by the FDA for inclusion in the package insert. Id., vol. 5, at 160. Moreover, Dr. Done noted, "[a]nything that was in the PDR was in the insert after 1968." Id. at 161. Thus, the jury heard evidence that, had Astra provided the FDA with the information about Xylocaine, the agency could have ordered revision of the package insert, and the updated information also would have appeared in the PDR or one of its annual supplements.

Dr. Henry L. Price, a board-certified Professor of Anesthesiology at Hahnemann Medical College in Philadelphia and a senior examiner for the American Board of Anesthesiology, also testified regarding the impact of Astra's overall course of conduct:

I think that Astra's failure to admit that their drug could cause reactions, which, in fact they did fail to do was responsible in this case for the outcome, including the brain damage because these people who gave the drug were not permitted to know what the experience of Astra was with respect to their own drug because they didn't communicate to the medical community that there could be reactions with this low a dosage. Therefore the people who gave the drug were not aware of what could happen and therefore didn't take precautions which in this case would have obviated the difficulty.

Id., vol. 9, at 62. While Dr. Price did not, in terms, speak specifically of Astra's failure to file reports with the FDA, the jury could have so interpreted his testimony and further could have concluded that, had the FDA been properly informed, it would have required notice via the package insert or PDR of the possibility of reactions from low dosage.

Counsel for plaintiffs also read into the record excerpts from the deposition of Dr. Marvin Seife, a long-time FDA employee who had held such positions as Executive Director of Medical Activities in the Office of Scientific Evaluation, Acting Associate Director for Marketed Drugs, and Deputy Director of the Division of Supplement Review in the Office of Drug Surveillance. During the course of the deposition, Dr. Seife was asked:

"Q. [I]f you had incomplete adverse reaction experience [from a company and its affiliates], would you have been in a position to make an informed judgment with respect to the safety of the drug?

....

What I am trying to get at is, were they [the reports] an important factor in making a determination whether a drug was safe or not?

A. Yes.

Q. Or whether labeling revisions were required?

A. Yes."

Id. at 289-90.

Finally, plaintiffs called Dr. John Adriani, Professor of Pharmacology and Anesthesiology at Louisiana State University, Emeritus Professor at Tulane University, Emeritus Director and former chairman of the Department of Anesthesiology at Charity Hospital in New Orleans, former chairman of the American Medical Association's Council on Drugs, and author of some 600 articles. Dr. Adriani also served as chairman of the FDA's Respiratory and Anesthetic Drugs Advisory Committee and as a member of and then consultant to the FDA's Advisory Committee. The direct examination of Dr. Adriani concluded as follows:

Q Doctor, do you have an opinion to a reasonable medical certainty with respect to whether the absence of [adverse-reaction] reports would bias an evaluation of safety of any drug?

....

A ... You mean if they failed to submit these drug reports; is that what you mean?

Q Yes, sir.

A Would that bias [an] evaluation of a drug?

Q Yes.

A Yes.

Transcript, vol. 26, at 156.

In short, plaintiffs attempted to establish causation by introducing evidence tending to show that the information withheld from the FDA was of great importance and that the agency could not properly perform its regulatory and supervisory roles without access to the unreported data, and that the FDA would have taken action had it been aware of Xylocaine's propensity to cause adverse reactions despite low dosage. Moreover, it was clear that Xylocaine had caused Harrikah's injuries: the jury found that Harrikah Stanton had experienced an adverse reaction, and Astra has never seriously argued that Xylocaine was not a "substantial factor" in causing that reaction.

Astra contends, however, as it did before the district court, see Stanton II, supra, slip op. at 8-11, that the evidence adduced by plaintiffs cannot support the jury's conclusion that the failure to file the reports was a substantial factor in causing Harrikah's injury because no witness stated expressly that, had those reports been filed, the FDA would have taken some action that would have saved this particular child from harm. It is true that the record does not contain such a statement, and that plaintiffs' evidence of causation is not very strong. On this point, the case thus is an extremely close one.29 However, in evaluating the sufficiency of the evidence and the district court's denial of Astra's motion for judgment n.o.v., we are mindful that

our role ... is necessarily a limited one. As we stated in [Huddell v. Levin, 537 F.2d 726 (3d Cir.1976) ], " 'The Seventh Amendment bars appellate review of facts found by a jury in actions at common law....' " 537 F.2d at 736 (quoting 9 C. Wright & A. Miller, Federal Practice and Procedure Sec. 2571, at 681 (1971)). Thus, we are admonished to review the record in this case in the light most favorable to the non-moving party ... and to affirm the judgment of the district court denying the motions unless the record "is critically deficient of that minimum quantum of evidence from which a jury might reasonably afford relief." Denneny v. Siegel, 407 F.2d 433, 439 (3d Cir.1969); accord, Huddell, 537 F.2d at 737.

Dawson v. Chrysler Corp., 630 F.2d 950, 959 (3d Cir.1980), cert. denied, 450 U.S. 959, 101 S. Ct. 1418, 67 L. Ed. 2d 383 (1981); accord Chuy v. Philadelphia Eagles Football Club, 595 F.2d 1265, 1273 (3d Cir.1979) (in banc).

The jury heard the testimony of four well qualified expert witnesses. They heard from these witnesses an analysis of the numerous adverse-reaction reports received by Astra, some of which described incidents of cardiac and respiratory arrest, see, e.g., Transcript, vol. 4, at 193-94. They heard express testimony from these witnesses that Astra's conduct deprived the FDA and the medical community of important information, the lack of which would bias an evaluation of the safety of Xylocaine, and they heard testimony (e.g., from Dr. Price) from which they could infer, keeping in mind the FDA's statutory duty continually to monitor marketed drugs for safety and effectiveness: (1) that had the FDA had these reports, it would have required notice to the medical community (through the package insert or PDR) of the critical information contained in the more than 200 adverse-reaction reports of the incidence of cardiac and respiratory arrest, notwithstanding low dosage of Xylocaine; and (2) that physicians receiving this information would have considered it in deciding how--and whether--to administer Xylocaine to their patients.30 Resolving every inference in plaintiffs' favor, we cannot conclude that the record was devoid of the requisite minimum quantum of evidence supporting the verdict.

Under Restatement (Second) of Torts Sec. 402A, strict liability attaches when a "defective product," sold in an "unreasonably dangerous condition," causes an injury.31 As we noted above, the jury found that Astra's failure to file the required reports rendered Xylocaine a defective product within the meaning of section 402A and that the defect proximately caused Harrikah's injury.32 We first consider whether the finding that whether the finding that Xylocaine was a defective product can be sustained as a matter of law; we then proceed to the question of proximate cause.

1. Was Xylocaine a Defective Product?

Plaintiffs' theory of strict liability is that the federal regulatory scheme creates an expectation on the part of the ordinary user of Xylocaine that the FDA would allow the continued marketing of the drug only after evaluating the information contained in reports required by regulations such as 21 C.F.R. Sec. 130.35. By failing to file those reports, plaintiffs assert, Astra deprived the FDA and Harrikah Stanton's doctors of important information that people such as plaintiffs would have expected the agency and the physicians to have received and considered. Accordingly, the argument goes, the drug therefore was defective because it left "the seller's hands ... in a condition not contemplated by the ultimate consumer, which will be unreasonably dangerous to him," Sec. 402A comment g; moreover, Xylocaine was "unreasonably dangerous" because it caused adverse reactions of which the FDA never learned and therefore was "dangerous to an extent beyond that which would be contemplated by the ordinary consumer who purchases it, with the ordinary knowledge common to the community as to its characteristics," id. comment i.33

We have been unable to uncover any statement of Pennsylvania law that directly resolves plaintiffs' contentions; we therefore must predict what a Pennsylvania court would do if confronted with the facts before us, see Becker v. Interstate Properties, 569 F.2d 1203, 1205-06 (3d Cir.1977), cert. denied, 436 U.S. 906, 98 S. Ct. 2237, 56 L. Ed. 2d 404 (1978). The closest indication of Pennsylvania law derives from the fact that an apposite California case, Toole v. Richardson-Merrell Inc., 251 Cal. App. 2d 689, 60 Cal. Rptr. 398 (1967), was cited approvingly by the Pennsylvania Superior Court in Berkebile v. Brantly Helicopter Corp., 225 Pa.Super. 349, 355, 311 A.2d 140, 143 (1973), aff'd, 462 Pa. 83, 337 A.2d 893 (1975). In Toole, the California Court of Appeals confronted a section-402A claim similar to the one now before us and upheld a jury verdict holding a drug company strictly liable for injuries sustained by a plaintiff where the company had violated reporting provisions of the Food, Drug and Cosmetic Act. Plaintiff had used the drug Triparanol (designed to treat arteriosclerosis), sold under the trade name "MER/29," and had developed cataracts in both eyes; Richardson Merrell's new-drug application had not disclosed the full extent of the company's knowledge regarding similar effects. Finding that the statutory violations rendered Triparanol a defective product because improperly prepared and marketed, the court declared:

In light of the non-disclosure of significant facts which, if known to the FDA, would have enabled its scientists to make a more critical analysis of [the drug], it can hardly be said that the FDA's permission to market the drug was an informed judgment, based on all the known facts, or that it was uninfluenced by [defendant's] non-disclosure.

60 Cal.Rptr. at 413.

We believe, given recent trends in Pennsylvania's law of strict liability, see, e.g., Azzarello v. Black Brothers Co., supra note 31, that a Pennsylvania court would agree with this formulation as an ingredient of section-402A liability, and that it would impose section 402-A liability where the fact finder concluded that the FDA would have, if the missing data had been disclosed, required dissemination to the medical community of the now disclosed facts. We note in this regard that what, in the negligence context, is proximate cause analysis, is collapsed into the elements of liability for defective product. See note 35, infra.

We conclude that a Pennsylvania court applying Pennsylvania law would allow a jury to determine on the facts of this case that Xylocaine was a defective product within the meaning of section 402A. First, plaintiffs adduced evidence that, by failing to comply with 21 C.F.R. Sec. 130.35, Astra deprived the FDA and, derivatively, the medical community at large of important information relating to the dangerousness of Xylocaine and that, without such information, the FDA could not make the professional and medical judgments that reasonable consumers rightfully expected it to make as a precondition of permitting the continued marketing of Xylocaine. Second, although plaintiffs' evidence was not particularly strong, for the reasons stated in Part II.B.2, supra, we believe that plaintiffs presented sufficient evidence that the FDA would have acted had it received the unreported information. The district court therefore did not err in denying Astra's motion for judgment n.o.v. on this issue.34

2. Was the Defective Product a Proximate Cause?

Having established that a jury could have found Xylocaine a defective product under section 402A, we need say little about the issue of proximate cause.35 Section 402A provides that a seller will be held strictly liable "when a defective product he has sold is the proximate cause of a user's injuries." Sochanski v. Sears, Roebuck & Co., supra, note 31, 621 F.2d at 69. As we noted above, see supra Part II.B.2, the jury found that Harrikah Stanton had experienced an adverse reaction to Xylocaine, see Special Question 1, and no one disputes the proposition that the adverse reaction caused the cardiac arrest and ensuing brain damage. The jury therefore could have found that the defective product, Xylocaine, was a "substantial factor" in causing Harrikah's injuries, and the district court properly denied judgment n.o.v. on the question of proximate cause.

III. Resubmission of Special Questions (Fed.R.Civ.P. 49)

At the conclusion of the 1979-80 trial, the jury retired to deliberate the answers to twenty-seven "special questions" prepared by the district court and counsel. Although it returned with an award in favor of plaintiffs, the award was inconsistent with the jury's answers to several of the "special questions."

The jury awarded the $315,000 in response to Special Question 26: "If you determine that any of the defendants are liable in damages to plaintiff, what compensatory damages do you award?" (Emphasis added.) A reasonable reading of that question suggests that any award necessarily would depend on a predicate finding of liability; as we already have discussed, see supra Part II, a finding of liability must subsume a finding of proximate cause. But in determining that Astra had been negligent in failing to file the reports required by section 130.35, see Special Questions 9(a) and 10(a), supra note 26, and that the failure to file the adverse-reaction reports rendered Xylocaine a defective product, see Special Question 11(a), supra note 32, the jury also found that neither the negligence nor the defective nature of the drug had been a substantial factor in causing Harrikah's injury, see Special Questions 9(b), 10(b), and 11(b), supra notes 26, 32. Thus, as the district judge observed, the jury apparently awarded damages "against no one." Transcript, vol. 31, at 26.

The jury compounded the confusion by answering "no" to Special Question 23:

If you determine that the negligence of, or sale of a defective product by, Astra and the negligence of Dr. Bowman were substantial factors in causing plaintiff's injuries, should Astra be relieved of liability because Dr. Bowman's negligence was so extraordinary as not to have been reasonably foreseeable by Astra?

A negative answer--or, for that matter, any answer at all36--assumes a predicate finding that both Astra and Dr. Bowman were liable; the "no" merely means that Dr. Bowman's conduct was not of the kind that should relieve Astra of its own liability. Thus, the jury's decision to answer this question, as well as the answer itself, were inconsistent with the answers to Special Questions 9(b), 10(b), and 11(b), but consistent with the award of damages in Special Question 26.37

Faced with this serious confusion, the trial judge resubmitted the questions to the jurors after explaining to them his reasons for so doing. After further deliberations, the jury changed its answers to several key questions:

1. The answer to 9(b) was changed from "no" to "yes," making Astra's failure to file the annual reports required by section 130.35(e) a substantial factor in causing Harrikah's injuries;

2. The answer to 10(b) was changed from "no" to "yes," making Astra's failure to file the adverse-reaction reports required by section 130.35(f) a substantial factor;

3. The answer to 11(b) was changed from "no" to "yes," making the defective nature of the drug a substantial factor; and

4. The answer to 23 was changed from "no" to "no answer," reflecting the fact that only Astra, and not Dr. Bowman, was liable for the injury.

These revised answers thus supplied a finding of liability to accompany the award of damages.

Astra argues, as it did before the district court, Stanton II, supra, slip op. at 3-8, that the district court erred in resubmitting the questions to the jury. Fed.R.Civ.P. 49,38 Astra points out, provides for submission to a jury of two types of questions: questions submitted under Rule 49(a), which ask the jury only to find facts, rather than to apply the law to those facts and render a general verdict, and questions submitted under Rule 49(b), which, by contrast, merely accompany and clarify the general verdict rendered by the jury. There is no doubt that Rule 49(b) expressly authorizes resubmission of inconsistent answers to special interrogatories. But Astra asserts that the "special questions" used in this case were submitted to the jury under Rule 49(a), not 49(b), and that Rule 49(a) does not contemplate resubmission. Therefore, Astra contends, the court should not have resubmitted the questions but, rather, should have ignored the damages award and molded the verdict for Astra based on the initial finding that Astra was not liable for Harrikah's condition. Had the court so ruled, the case would have ended, and the decision in Kaczkowski v. Bolubasz, supra, would not have required a new trial either on damages or on anything else. Two questions thus arise: (1) Were the questions submitted pursuant to Rule 49(a) or 49(b)? (2) If the former, did the district court abuse its discretion in ordering resubmission?

A. Were the Questions Submitted Under Rule 49(a) or 49(b)?

In deciding whether the district court intended to submit the "special questions" under Rule 49(a) or 49(b), it behooves us to consider the court's own analysis of its actions:

In all candor, in framing the Special Questions, the Court did not consciously intend a Rule 49(a) submission as distinguished from Rule 49(b). It was an honest effort to insure that all issues would be addressed and resolved, including a determination of final liability, in the same manner as if a general verdict was implicated.

Stanton II, supra, slip op. at 3. After analyzing the content of the questions submitted, the court concluded that "[t]he format utilized here performed the function contemplated under 49(b)." Id. at 4. We agree.

As the district court accurately pointed out, Special Question 26 expressly asked the jury to make a finding of ultimate liability: "If you determine that any of the defendants are liable in damages to plaintiff, what compensatory damages do you award?" Special Questions 23, 24, and 25 also called upon the jury to render a general verdict:

If you determine that the negligence of, or sale of a defective product by, Astra and the negligence of Dr. Bowman [Question 23]/Dr. Fabian [Question 24]/ Harrisburg Hospital [Question 25] were substantial factors in causing plaintiff's injuries, should Astra be relieved of liability because Dr. Bowman's/Dr. Fabian's/Harrisburg Hospital's negligence was so extraordinary as not to have been reasonably foreseeable by Astra?

Moreover, the district court's charge reveals that the jury was being asked to consider legal as well as factual matters:

My charge will ... consist of the initial phase being kind of the general charge, that is, the law as it applies generally in all types of litigation, and then later on in the charge, I will get to the specifics of this particular case; the specifics principally as to the law to be applied because I don't expect to do any review of any significance of the ... evidence in the case ....

....

So when I do make reference, as I will, to a particular factual issue or some contention by the parties or some evidence, it is merely to illustrate the legal principle involved ....

Transcript, vol. 30, at 82. The court continued:

[T]he defendants don't rise or fall together. And you can return a verdict in favor of plaintiff against all of the defendants or against one of the defendants or some of the defendants, or you can find in favor of all the defendants and against the plaintiff.

Id. at 84. The jury thus was on notice, from the very beginning of the charge, of its responsibility to render a general verdict on liability.

Rule 49(b) also requires the trial judge to "give such explanation or instruction as may be necessary to enable the jury both to make answers to the interrogatories and to render a general verdict ...." (Emphasis added.) We believe that the district court's charge complied with the mandate of Rule 49(b).

The court began the substantive portion of its charge by explaining to the jury the principles of negligence and proximate cause and also discussed the effect of violating a statute or regulation. Transcript, vol. 30, at 102-17. The jury was fully informed of the requirements of 21 C.F.R. Sec. 130.35 and told to determine whether it was reasonable for Astra not to file the reports either because of a legal excuse or because the adverse reactions therein described were not unexpected. Id. at 110-16. Next came instructions on strict liability and section 402A, id. at 117-19, 125-38, the consequences of a failure to provide proper and adequate warnings about Xylocaine's characteristics, id. at 119-28, and the principle of superseding cause, id. at 140-42. Finally, the court charged the jury on compensatory damages, id. at 142-48, reduction to present value, id. at 148-49, pain and suffering, id. at 149-51, and punitive damages, id. at 152-53. We are confident that these explanations and instructions provided an ample foundation upon which the jury could render a general verdict and answer the "special questions."

Accordingly, we conclude that the "special questions" submitted to the jury fit within the framework of Rule 49(b), and the district court therefore did not abuse its discretion when it resubmitted the inconsistent answers to the jury for clarification.

Even accepting Astra's assertion that the "special questions" were propounded under Rule 49(a), however, we conclude that the district court retained the discretion to resubmit the questions to the jury and that the court did not abuse its discretion in doing so here. The language of the Rule in no way prohibits resubmission, and we see no reason in logic or policy so to distinguish between Rule-49(a) and Rule-49(b) questions.

Astra has pointed to only one case directly holding that a court may not resubmit to a jury questions propounded under Rule 49(a). See McCollum v. Stahl, 579 F.2d 869 (4th Cir.1978), cert. denied, 440 U.S. 912, 99 S. Ct. 1225, 59 L. Ed. 2d 460 (1979).39 The Court of Appeals for the Fifth Circuit has taken the opposite position and allows resubmission even under Rule 49(a). See Alverez v. J. Ray McDermott & Co., 674 F.2d 1037, 1040-41 (5th Cir.1982); Mercer v. Long Mfg. N.C. Inc., 671 F.2d 946, 947 (5th Cir.1982). We are persuaded that the Fifth Circuit has adopted the better view.40 As that court declared in discussing the related question of the permissibility of submitting supplementary interrogatories in an attempt to harmonize responses to Rule-49(a) questions: "It would be anomalous to hold that, while a court pursuant to Rule 49(a) must search for a view of the case which will make the jury's answers consistent, it may not submit an additional interrogatory to the jury to clarify an ambiguity." Morrison v. Frito-Lay, Inc., 546 F.2d 154, 161 (5th Cir.1977).41 Moreover, in purely pragmatic terms, it seems terribly inefficient not to obtain clarification from a still-empaneled jury of the meaning of its answers and verdict, especially when we consider that unclarified inconsistent answers often necessitate a retrial of the entire case. Accordingly, we hold that the district court did not abuse its discretion in resubmitting the "special questions" to the jury, whether those questions be deemed to have been propounded under Rule 49(a) or 49(b).

When the Pennsylvania Supreme Court's decision in Kaczkowski v. Bolubasz, supra, changed the law of damages that had applied at the time the jury made its award in favor of plaintiffs, the district court granted a new trial limited only to damages. Stanton I, supra, slip op. at 2-5. Astra does not contest the necessity of recalculating damages but urges that the retrial also should have encompassed the question of liability.

It is clear that the Federal Rules of Civil Procedure provide for partial new trials, see Fed.R.Civ.P. 59(a); the only question is whether the district court abused its discretion in ordering one in this case.42 The standard for determining whether it is proper to limit a new trial only to certain issues has long been established: "Where the practice permits a partial new trial, it may not properly be resorted to unless it clearly appears that the issue to be retried is so distinct and separable from the others that a trial of it alone may be had without injustice." Gasoline Products Co. v. Champlin Refining Co., 283 U.S. 494, 500, 51 S. Ct. 513, 515, 75 L. Ed. 1188 (1931). Reflecting on Gasoline Products, this Court has declared:

In the forty-six years since the decision in Gasoline Products, the judicial limitation on the power to grant partial new trials has retained its force. It has been read to prevent limited new trials where a tangled or complex fact situation would make it unfair to one party to determine damages apart from liability, 6A Moore's Federal Practice p 59.06 at 59-89 (1973), or where 'there is reason to think that the verdict may represent a compromise among jurors with different views on whether defendant was liable.' 11 C. Wright & A. Miller, Federal Practice and Procedure Sec. 2814 at 96 (1973) (footnotes omitted).

This circuit has consistently observed the stricture of Gasoline Products. For example, this court has required that partial new trials should be granted " 'only in those cases where it is plain that the error which has crept into one element of the verdict did not in any way affect the determination of any other issue.' " Romer v. Baldwin, 317 F.2d 919, 922-23 (3d Cir.1963), quoting Thompson v. Camp, 167 F.2d 733, 734 (6th Cir.), cert. denied, 335 U.S. 824 [69 S. Ct. 48, 93 L. Ed. 378] ... (1948). Accord, e.g., Darbrow v. McDade, 255 F.2d 610, 611 (3d Cir.1958).

Vizzini v. Ford Motor Co., supra note 42, 569 F.2d at 760; accord Heckman v. Federal Press Co., 587 F.2d 612, 619 (3d Cir.1978); see Leizerowski v. Eastern Freightways, Inc., 514 F.2d 487, 491 (3d Cir.1975).

On the facts of this case, we cannot say with assurance that the questions of liability and damages were so separable that the jury's determination of the one had no bearing on its determination of the other; rather, our review of the record strongly suggests that the first jury reached a verdict representing a compromise among jurors with different views as to whether Astra was liable. We therefore cannot assert that "allowing a second jury to determine the issue of damages in isolation from the whole of the circumstances surrounding the case was not an injustice to [Astra]." Vizzini v. Ford Motor Co., supra note 42, 569 F.2d at 761.

The most compelling indication that the first jury reached a compromise is a comparison between the damages awarded and the damages requested. The jury awarded plaintiffs a total of $315,000--$60,000 for Mrs. Brooks and $255,000 for Harrikah. This award, although described by the district court as "one of the highest ever recorded in this district in a personal injury action," Stanton I, supra, slip op. at 5, is fairly small compared to reported verdicts for similar injuries to similarly situated plaintiffs, of which awards we may take judicial notice, and was far smaller than the amount that plaintiffs' evidence might have supported.43 Mrs. Brooks had adduced evidence of damages of $186,360, the sum obtained by multiplying $23,295 (the annual salary of a nurse) by the eight years that Mrs. Brooks, herself a nurse, had spent in caring for Harrikah since the 1971 injury; the jury awarded less than one-third of that sum. Plaintiffs also presented evidence that Harrikah had sustained damages greatly exceeding the $255,000 actually awarded: they sought $529,625 for fifty years' worth of home care,44 $88,664 for therapy, and either $132,000 or $155,212 in lost earnings, depending on whether Harrikah were to have obtained a high-school or a college diploma. Moreover, Harrikah sought additional damages for pain and suffering. The jury thus awarded plaintiffs far less than the amounts requested.

Inconsistent answers to the other "special questions" further suggest that the verdict was a compromise. The answers were inconsistent among themselves and, in some cases, with the damage award itself. We already have described a number of these conflicts, see supra Part III;45 there are other potential inconsistencies as well. For example, the jury found (after resubmission of the "special questions") that Astra had acted negligently in failing to file the various reports required by 21 C.F.R. Sec. 130.35 and that such negligence was a substantial factor in causing Harrikah's injury. But the jury also found that Astra had adequately warned the doctors of the possibility of toxic reactions to Xylocaine and had provided adequate directions for computing a safe dosage of the drug for an eight-month-old child.46 While these various answers may be reconcilable with a finding of liability, see supra note 46, and thus do not undermine the verdict, they nevertheless suggest--at least in the context of the comparatively small damage award--that the jury might have had some reservations about imposing liability on Astra.

In short, there is a great deal of evidence that the jury was at odds, that it was temporizing, and that its solution was a compromise verdict. Under the circumstances, it was an injustice to Astra to retry the damages issue alone, and we therefore hold that the district court abused its discretion in so ordering.

Astra has requested that a new trial on liability be limited to the issues posed in Special Questions 9 through 11, supra notes 26, 32--whether Astra acted negligently in not filing the reports, whether the failure to file the reports rendered Xylocaine a defective product, and whether either Astra's negligence or the product's defective nature was a substantial factor in causing Harrikah's brain damage. We find such an approach inappropriate. Because we are remanding for a new trial, we believe that fairness to plaintiffs requires the liability question to be retried in toto. Plaintiffs therefore will have an opportunity to reintroduce evidence supporting their contentions, inter alia, that Astra failed adequately to warn the medical community of the dangers of Xylocaine, that Astra "overpromoted" the drug, and that Astra failed to conduct sufficient studies to determine a safe dosage of Xylocaine for a child of Harrikah's age.

We now consider in a more summary manner the remaining claims raised by Astra and plaintiffs. We realize that some of the problems raised by the various claims are obviated by our remanding the case for a full retrial, but, because of the age of the case and the likelihood that a number of these questions will recur, we will rule on the germane issues for the guidance of the court and counsel.

Counsel for plaintiffs began his opening statement during the retrial on damages by labeling the case

probably one of the most important civil cases which has ever come before a jury in the United States of America. I think when you see its implications, that it's a test of our judicial system to see if a child who is at the lower end of our society, a person who is absolutely utterly helpless can come before a jury and receive fair and just compensation for the injuries which have been afforded to her.

Trial Transcript, vol. 1, at 49, Feb. 16, 1982. He then sketched for the jury a portrait of Mrs. Brooks, whose first husband was killed in Vietnam in the 1960's and whose second husband--Harrikah's father--was killed "as a victim of street crime" in January 1971, while Mrs. Brooks was pregnant with Harrikah. Id. at 52. Counsel also discussed Mrs. Brooks' third marriage, to a cement mason in Chattanooga, id. at 53, and described the Brooks home:

This is the type of home where children don't talk back to parents, where there is a certain amount of order. They see a father get up every morning at 7:30 and go off to work to provide for the children. There is a certain amount of order and you see the Brooks children, they care for their younger sister. This is not the type of family that some people like to see when they say the family is disintegrating. It is not disintegrating. It is, in fact, a rock of the type that most of us would like to belong to.

Id. at 54.

Finally, plaintiffs' attorney reached the question of plaintiffs' race:

[W]e were concerned about the effect of having black people come to an area where there are not many black people and expecting to get justice from a jury which is mostly white people. We decided to confront this issue and we asked you the questions this morning, and we were really pleased with the responses that we got and we think that this is an impartial jury and everyone here has sworn that they will try this case not on the basis of passions, or prejudice, or economic basis, but on the basis of the facts and the law.

Id. at 54-55. Astra contends that the above statements, taken together, unfairly prejudiced the jury and that the district court erred in denying Astra's motion for a mistrial.

Because we have held that the case must be remanded for a new trial, we need not resolve the question whether the district court should have declared a mistrial; were we to confront that issue, we would be inclined to hold that these remarks did not require a new trial because we would find it difficult to say that "it [was] more than 'reasonably probable' that the verdict was influenced by the prejudicial statements," Draper v. Airco, Inc., 580 F.2d 91, 97 (3d Cir.1978). However, we are not dealing with that question but, rather, with the propriety of the remarks. We hold that the statements go beyond the ambit of proper opening statement. See Government of Virgin Islands v. Turner, 409 F.2d 102 (3d Cir.1969), where we analyzed the legitimate functions of an opening statement, and stated that: "The purpose of an opening is to give the broad outlines of the case to enable the jury to comprehend it ...." Accordingly, the remarks should not be repeated in the opening statement at the retrial. We add that significant portions of the quoted remarks are, at all events, beyond the realm of appropriate advocacy. As we said in Draper, "there must be limits to pleas of pure passion and there must be restraints against blatant appeals to bias and prejudice." 580 F.2d at 95. Justice must not be based on racial sympathy or animosity.

Plaintiffs contend that the district court unfairly prejudiced their ability to prevail on their claim for punitive damages by (1) allowing Astra to introduce evidence of FDA factory inspections to establish an excuse for the company's failure to file the reports required by 21 C.F.R. Sec. 130.35(e) and (f); (2) allowing Astra to introduce the testimony of its attorney, Alan Kaplan, concerning his interpretation of Astra's reporting obligations, in order to establish the reasonableness of Astra's failure to comply with the statute and regulations; and (3) refusing to admit into evidence, for the purpose of establishing statutory fraud and willful, wanton, and reckless conduct, two supplemental new-drug applications that Astra had submitted to the FDA in 1969.47 While we have serious doubts that the evidence presented by plaintiffs would support an award of punitive damages, that question is not now before us; we will rule on plaintiffs' claims for the guidance of the district court, should plaintiffs pursue their punitive-damages claim on retrial. We agree with the district court's rulings on the first two points, but we hold that the court erred in excluding the 1969 applications.

The district court did not err in admitting evidence of the FDA's factory inspections, at least insofar as that evidence related to plaintiffs' claim for punitive damages. Fed.R.Evid. 401 establishes the standard for "relevant evidence": " 'Relevant evidence' means evidence having any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence." (Emphasis added.) The "fact of consequence" here is whether Astra "engage[d] in outrageous conduct" or "act[ed] with a bad motive or with reckless indifference to the interests of persons who might be administered the drug," Special Question 15.

Under the circumstances, the evidence clearly was relevant and admissible as to the punitive-damages claim. In determining the motive with which Astra acted in failing to file the report, a jury certainly could take note of the fact that Astra apparently made the unreported adverse-reaction information available to FDA inspectors. Such conduct could be interpreted as inconsistent with plaintiffs' allegation that Astra was doing everything within its power to conceal from the world the true nature of Xylocaine. The evidence thus bears on the question whether Astra was acting in bad faith when it failed to file the requisite reports and therefore meets the criterion propounded in Rule 401.48

We also agree that the testimony of Astra's attorney, Alan Kaplan, was admissible insofar as it related to the punitive-damages claim because such testimony bore upon the reasonableness of Astra's decision not to file the reports with the FDA.49 Under Pennsylvania law, punitive damages may be awarded "only after consideration of the act itself, together with all the circumstances, including the motive of the wrong-doer, and the relations between the parties." Pittsburgh Outdoor Advertising Co. v. Virginia Manor Apartments, Inc., 436 Pa. 350, 353, 260 A.2d 801, 803 (1970) (emphasis added). In determining Astra's motive, we would think that a jury might consider why Astra did what it did, including whether the company acted on advice of counsel. Pennsylvania courts do not appear specifically to have addressed the question whether advice of counsel may be raised as a defense (albeit not an absolute one) to a claim for punitive damages, but a number of courts have held that "good faith reliance upon advice of counsel may prevent imposition of punitive damages," Henderson v. United States Fidelity & Guaranty Co., 695 F.2d 109, 113 (5th Cir.1983); accord Fox v. Aced, 49 Cal. 2d 381, 317 P.2d 608, 610-11 (1957); see also 22 Am.Jur.2d Damages Sec. 253 (1965) (citing cases). We are persuaded by these decisions, and we predict that Pennsylvania law would allow a jury considering punitive damages to take into account the fact that Astra's lawyer told the company that there was no need to file with the FDA the reports specified in 21 C.F.R. Sec. 130.35. Counsel's testimony to that effect thus was properly admitted as relevant to the claim for punitive damages.

Finally, we come to plaintiffs' claim that the district court erred in excluding the two supplemental new-drug applications that Astra had submitted to the FDA in 1969. Astra had not included in these applications the reports of Xylocaine's adverse-reaction experience, despite the fact that such information was required by 21 C.F.R. Sec. 130.30 (1969), as amended by 32 Fed.Reg. 8087 (1967). Plaintiffs contend that these applications should have been admitted as evidence of a pattern of deception and willful concealment of vital information and that the exclusion of the applications unfairly prejudiced plaintiffs' ability to prevail on their punitive-damages claim. Astra, however, asserts that the applications were properly excluded because the filings "were submitted in error by the defendant, were withdrawn and were not acted upon by the FDA." Brief at 41.

We believe that the district court erred in excluding the 1969 applications. Contrary to Astra's protestations, the applications were "acted upon" by the FDA: the agency thoroughly reviewed the documents, see App. at A529-32, and responded by sending a five-page letter to Astra discussing labeling changes and safety warnings, Letter from John Jennings, M.D., to Astra (Aug. 18, 1969), App. at A533. It therefore is disingenuous for Astra to imply that the FDA did not take seriously the 1969 applications. Nor can Astra prevail by claiming that the applications were "submitted in error." Astra submitted them, even if mistakenly, in the regular course of business. They were not mailed surreptitiously to the FDA by a dissident employee harboring a grudge against the company, and we see no reason why Astra should not have to answer for the good-faith acts of its employees.

Moreover, we agree with plaintiffs that the applications would have been probative of Astra's "determination to thwart the law," Brief at 45. The jury could have considered the circumstances surrounding the 1969 applications as evidence of a larger, carefully planned scheme to hide unfavorable information. Such an interpretation would have rendered Astra's initial decision not to comply with section 130.35 just one more element in that scheme. Accordingly, we hold that the district court erred in excluding the 1969 applications, at least insofar as they related to plaintiffs' punitive-damages claim.

For the reasons stated above, we conclude that Astra had an obligation to file the reports required by 21 C.F.R. Sec. 130.35; that there was sufficient evidence to support the verdict in favor of plaintiffs; that the district court did not err in resubmitting the "special questions" to the jury; and that the court did not abuse its discretion in allowing Astra to introduce evidence of factory inspection and the testimony of its attorney, Alan Kaplan. We have also held that the court erred in excluding from evidence the 1969 supplemental applications, and that, even though the remarks of counsel during his opening statement might not have justified a mistrial, they were improper and should not be repeated on retrial. Finally, we conclude that the court abused its discretion in granting a retrial limited only to damages.

We therefore will vacate the judgment and remand for a new trial on liability and damages. During this new trial, plaintiffs will be free to pursue any and all theories and claims advanced during previous stages of the litigation.