ERVIN, Circuit Judge.

Jefferey Dale Ellis, administrator of the estate of his wife, Margaret Ann Ellis, appeals from the district court's denial of a motion for a new trial. Margaret Ellis, a user of Playtex tampons, died unexpectedly in May of 1981. Convinced his wife had died from toxic shock syndrome (TSS), and that the TSS had been induced by a Playtex Super Deodorant tampon, Jefferey Ellis brought a wrongful death action against Playtex International, Inc. After a full trial the jury found for Playtex. Ellis immediately moved for a new trial on the grounds that the verdict was not supported by the weight of the evidence and that certain critical exhibits had been improperly excluded. His motion was denied.

On appeal, Ellis argues that the district court committed a variety of reversible errors during the course of the trial. Specifically, Ellis contends that (1) the court applied the wrong standard in determining whether the jury's verdict was supported by the clear weight of the evidence; (2) the court erred in refusing to admit Toxic Shock Syndrome Studies carried out by the Center for Disease Control (CDC) and various state health departments; (3) the court improperly excluded a list of complaints received by Playtex from users of Playtex tampons; (4) the court erred in failing to permit the plaintiff's experts to read from a medical treatise; and (5) the court tendered an incorrect charge to the jury on the defendant's duty to warn.

We conclude that the CDC and state TSS studies were improperly excluded from evidence. Because we believe that the studies may have played a critical role in the outcome of the trial, we reverse the judgment of the lower court and remand for a new trial.

I.

On Friday, May 22, 1981, Margaret Ellis informed her husband that her menstrual period had begun and asked him to purchase a box of tampons for her at a local store. Jefferey Ellis complied with her request and promptly returned home with a package of Playtex Super Deodorant tampons.

Three days later, on Monday, May 25, 1981, Mrs. Ellis announced that she was not feeling well and took to her bed. The following day she remained in bed, complaining of a sore throat, aches, and a fever. Later that day Mrs. Ellis asked her son to run an errand for her at a local store. When he returned he discovered that his mother had collapsed on the bathroom floor. A rescue squad was summoned and Mrs. Ellis was immediately hospitalized. Early Wednesday morning Mrs. Ellis died.

At trial, Ellis advanced two theories of liability. The first alleged that Mrs. Ellis' death was proximately caused by the negligence of Playtex. The second claimed that Playtex breached an implied warranty of merchantability by manufacturing and marketing a dangerous product. In order to prevail under the implied warranty theory, Ellis had to prove that (1) Mrs. Ellis had TSS; (2) TSS is causally related to tampon use; and (3) the Playtex tampon used in this case was responsible for the toxic infection. To prevail on the negligence theory, the plaintiff had to prove, in addition to these first three elements, that Playtex knew of the dangers associated with tampon use and failed to provide an adequate warning to consumers.

The first factual issue — the presence of TSS — was hotly contested. Expert medical witnesses on both sides offered conflicting theories to explain Mrs. Ellis' sudden death. All five of the doctors who examined and treated Mrs. Ellis testified that she died from a heart attack that had been induced by toxic shock syndrome. They noted that high levels of toxins had been found in several of Mrs. Ellis' major organ systems, and that the damage to her heart resembled that found in other toxic shock cases. They also noted that there was no evidence of pre-existing heart disease. Each of the five doctors concluded that the most likely source of the toxins was the Playtex tampon that Mrs. Ellis had been using before she became ill.

Playtex, on the other hand, offered medical testimony by three physicians who concluded that it was Mrs. Ellis' unique susceptibility to heart trouble, and not TSS, that caused her death. These doctors pointed out that Mrs. Ellis did not exhibit several symptoms commonly associated with toxic shock syndrome and that staphylococcus aureus — the bacteria many experts believe has been responsible for toxin production in TSS-related cases — was not found in her vagina. Moreover, these experts noted that Mrs. Ellis had suffered a series of fainting spells in the past. In their view, therefore, it was more likely that the cardiac arrest was related to a fainting spell than to TSS.

The second issue — causation — was also strenuously contested at trial. To support its position the plaintiffs sought to introduce two epidemiological studies authored by the Federal Center for Disease Control in Atlanta (CDC) and an epidemiological study supervised by the Wisconsin Minnesota, and Iowa state health departments (Tri-State Commission). These studies indicated that a statistically significant link existed between tampon use and TSS. The CDC data was particularly helpful to the plaintiff because it revealed that the more absorbent tampons, like the Playtex deodorant tampon used by Mrs. Ellis, had a higher statistical correlation with TSS than the less absorbent tampons. Furthermore, the CDC data revealed that most TSS cases began on the second day of the victim's menstrual cycle — the same day Mrs. Ellis' illness began.

The court, however, refused to admit the studies and data into evidence on the ground that they constituted hearsay. The court reasoned that there was insufficient foundation testimony about the methodology used in the studies to "secure for the reported information some essence of reliability or trustworthiness." J.A. at 6-7.

Like the causation issue, the failure to warn issue was also heavily disputed at trial. Playtex maintained that it had made every effort to warn consumers through newspaper, magazine, and television advertisements as soon as it learned of the potential TSS risk in 1980. Playtex also argued that an explanatory warning had been placed on all boxes of Playtex Super Deodorant Tampons produced after November, 1980, and on store shelves. Ellis testified, however, that he had not seen a warning when he purchased the tampons, and he offered into evidence a Playtex tampon box, purchased in January 1982, that did not contain a warning or explanation of any kind. Ellis also sought to admit records of consumer complaints received by Playtex from Playtex tampon users. The records purported to reveal the types of injuries that users were experiencing. Ellis offered the records to indicate the types of risks Playtex either was aware of or should have foreseen and included in a warning.

As with the CDC and Tri-State reports, the court refused to admit the Playtex records into evidence. The court reasoned that because the "truth or falsity" of the consumer complaints was not in issue, the records need not be actually admitted into evidence as long as the jury was informed that complaints had been received.

At the end of a full trial the jury returned a verdict for Playtex. Shortly thereafter Ellis' motion for a new trial was denied.

II.

We address first Ellis' contention that the district court applied the wrong standard in reviewing the verdict. It is well established in this circuit that a trial judge must employ two different legal standards when considering a motion for a directed verdict and a motion for a new trial. A motion for a directed verdict does not require the court to weigh the plaintiff's evidence against that offered by the defendant. Rather, the court's sole duty is to examine the sufficiency of the evidence tendered by the party opposing the motion. If the evidence is sufficient, the judge should not direct the verdict. The court, however, need not examine the evidence tendered by the party seeking a directed verdict.

A motion for a new trial, on the other hand, involves a "comparison of opposing proofs." William v. Nichols, 266 F.2d 389, 393 (4th Cir. 1959). The district court may set aside a verdict supported by substantial evidence if "it is contrary to the clear weight of the evidence, or is based upon evidence which is false," or is "necessary to prevent a miscarriage of justice." Aetna Casualty v. Yeatts, 122 F.2d 350, 354 (4th Cir. 1941).

The distinction between this standard and that of a directed verdict was succinctly summarized in Garrison v. United States, 62 F.2d 41 (4th Cir. 1932), in which we stated:

'Where there is substantial evidence in support of plaintiff's case, the judge may not direct a verdict against him, even though he may not believe his evidence or may think that the weight of the evidence is on the other side; for, under the constitutional guaranty of trial by jury, it is for the jury to weigh the evidence and pass upon its credibility. He may, however, set aside a verdict supported by substantial evidence where in his opinion it is contrary to the clear weight of the evidence, or is based upon evidence which is false; for, even though the evidence be sufficient to preclude the direction of a verdict, it is still his duty to exercise his power over the proceedings before him to prevent a miscarriage of justice.'

62 F.2d at 42, quoted in Aetna Casualty v. Yeatts, 122 F.2d 350, 354 (4th Cir. 1941).

Ellis contends that the lower court failed to discern the difference between these two standards and mistakenly applied a directed verdict standard to a motion for a new trial. In the past we have not hesitated to reverse a judgment when a lower court has abused its discretion and applied the wrong standard in reviewing a motion for a new trial. Williams v. Nichols, 266 F.2d 389 (4th Cir. 1959); Virginia Railway Co. v. Armentrout, 166 F.2d 400, 408 (4th Cir. 1948). In this case, however, we are simply not convinced that the lower court applied the wrong standard. The lower court opinion expressly cites Aetna and paraphrases the proper standard. Although there is relatively little discussion of the respective merits of the evidence tendered by each side, we are not prepared to say that the court failed to "exercise its independent judgment after weighing all of the evidence." A fuller discussion might have been preferable, but brevity is not grounds for finding reversible error.

Furthermore, it is not clear that the jury's verdict was contrary to the clear weight of the evidence or that it represented a "miscarriage of justice." Although it would be reasonable to attach greater weight to the testimony of the five treating physicians than that of the three nontreating physicians, credible expert testimony was presented by both sides. The factual issues, particularly the "failure to warn" question, were reasonably close. We do not believe, therefore, that in applying the Aetna standard the lower court reached a result so unsupported by the evidence as to constitute an abuse of discretion.

III. A.

We now turn to Ellis' first evidentiary objection that the CDC and Tri-State studies were improperly excluded from evidence. But before addressing this claim it is important to review briefly the nature and form of the scientific data that was offered.

i

Toxic Shock Syndrome was first identified in 1978, but it was not until 1980 that the symptoms of the disease were linked to menstruation. Faced with a rapid and severe outbreak of a novel disease, the federal government and a group of state health agencies undertook a series of studies designed to isolate the cause of the disease. The CDC and Tri-State reports at issue here were among the first of these epidemilogical studies. Focusing on a pool of TSS cases that had been reported to state health officials, the CDC and the state health departments of Wisconsin, Minnesota, and Iowa administered a questionnaire to a selected group of the TSS victims and to a control group. The studies revealed that the vast majority of TSS victims had recently used some type of tampon. Playtex tampons were among those that TSS victims reported having used.

The CDC studies were conducted in the spring and summer of 1980. The first results of the CDC inquiry were published in the May 23, 1980, issue of Morbidity and Mortality Weekly Report (MMWR). Subsequent findings, published in the June 27, September 19, October 3, and October 17 issues of MMWR confirmed the correlation between menstruation and TSS. The findings as published in the MMWR, were in the form of "raw data," accompanied by an editorial note. It was this MMWR data that constituted the first of the two studies the plaintiffs sought to admit at trial. The data consisted of tables, graphs, and charts showing the factors used to evaluate cases and controls and the results obtained. The data was organized into two self contained studies entitled "Epidemiologic Data, Toxic-Shock Syndrome, CDC Case-Control Study # 1, June 1980," and "Epidemiologic Data, Toxic-Shock Syndrome, CDC Case-Control Study # 2, July/August 1980."

The Tri-State Study was conducted during the summer and fall of 1980. Some of the data used in the study, like the CDC data, was published in the MMWR before the study was finished. The Tri-State findings were announced in January 1981 and promptly forwarded to the Food and Drug Administration (FDA). The FDA immediately published a summary of the results in the federal register. 46 Fed.Reg. 23766-68 (Apr. 28, 1981) (codified at 21 C.F.R. § 801.430 (1984)). At trial the plaintiff offered the Tri-State data in the same form in which it had been originally presented to the FDA: a five page letter summarizing the methodology, findings, and conclusions.

At trial Ellis contended that the studies fell within an exception to the hearsay rule covering investigative public records and reports. Fed.R.Evid. 803(8)(C). On appeal, Playtex insists that the district court ruled correctly in refusing to apply this exception because (1) the data lacks traditional indicia of reliability and trustworthiness and (2) the data was offered without proper foundation testimony to explain the methodology used by the CDC and the State Health Departments. Playtex argues in the alternative that (1) even if the exception was applicable the evidence was more prejudicial than probative and (2) if exclusion was erroneous, it was harmless error. We do not find Playtex's arguments persuasive.

ii

Fed.R.Evid. 803(8)(C) states that the following are not excluded by the hearsay rule:

[r]ecords, reports, statements, or data compilations, in any form, of public offices or agencies, setting forth . . . factual findings resulting from an investigation made pursuant to authority granted by law, unless the sources of information or other circumstances indicate lack of trustworthiness.

The rule is premised on "the assumption that a public official will perform his duty properly and the unlikelihood that he will remember details independently of the record." Fed.R.Evid. 803(8)(C) advisory committee note, citing Wong Wing Foo v. McGrath, 196 F.2d 120 (9th Cir. 1952). "Admissibility in the first instance" is assumed because of the reliability of the public agencies usually conducting the investigation, and "their lack of any motive for conducting the studies other than to inform the public fairly and adequately." Kehm v. Proctor Gamble, 724 F.2d 613, at 618, 619 (8th Cir. 1983). This is not to say that a report containing the findings of a public agency, made pursuant to an investigation authorized by law, is always admissible. To the contrary, if "sufficient negative factors are present" to indicate the report is not trustworthy, it should not be admitted. Fed.R.Evid. 803(8)(C), advisory committee note. The factors that may be used to determine admissibility include: (1) the timeliness of the investigation; (2) the special skill or experience of the official; and (3) possible motivation problems. Id. But the burden is on the party opposing admission to demonstrate that the report is not reliable. See Kehm, at 618; Baker v. Elcona Homes Corp., 588 F.2d 551, 558 (6th Cir. 1978), cert. denied, 441 U.S. 933, 99 S.Ct. 2054, 60 L.Ed.2d 661 (1979).

Placing the burden on the opposing party makes considerable practical sense. Most government sponsored investigations employ well accepted methodological means of gathering and analyzing data. It is unfair to put the party seeking admission to the test of "re-inventing the wheel" each time a report is offered. Rather than requiring the moving party to spend considerable time and money to ensure that the experts who conducted the study are available at trial, it is far more equitable to place that burden on the party seeking to demonstrate why a time tested and carefully considered presumption is not appropriate.

We do not believe scientific reports should be treated any differently from other public findings of fact under a Fed.R.Evid. 803(8)(C) analysis. Indeed, in this instance the rationale behind the rule argues strongly for admission of the contested studies. First, both studies were carried out by public offices "pursuant to authority granted by law." The CDC is a branch of the United States Department of Health and Human Services, and the Tri-State study was conducted by members of the Wisconsin, Minnesota, and Iowa Departments of Health. Both studies consist of "data compilations . . . setting forth factual findings . . ." Fed.R.Evid. 803(8)(C). Although there has been some disagreement over whether "evaluative" public records or reports were intended by Congress to be considered "factual findings" for the purposes of Fed.R.Evid. 803(8)(C), it is well established that the phrase should be interpreted broadly. See Kehm, at 618; Melville v. American Home Assur. Co. 584 F.2d 1306 (3rd Cir. 1978); Baker v. Elcona Homes Corp., 588 F.2d 551 (6th Cir. 1978), cert. denied, 441 U.S. 933, 99 S.Ct. 2054, 60 L.Ed.2d 661 (1979); Wetherill v. University of Chicago, 518 F. Supp. 1387 (N.D.Ill. 1981); Zenith Radio Corp. v. Matsushita Electrical Industrial Co., Ltd., 505 F. Supp. 1125 (E.D.Pa. 1980); 4 Weinstein's Evidence ¶ 802(8)[03], at 803-205-08. Thus, the fact that the CDC and Tri-State studies contained tentative conclusions as well as statistical findings does not affect the applicability of the rule. See United States v. American Telephone Telegraph Company, 498 F. Supp. 353, 360 (D.D.C. 1980).

Second, the studies possess ample indicia of trustworthiness. The CDC and the state health departments of Minnesota, Iowa, and Wisconsin are highly skilled in the study of epidemiology. Both the CDC and the state health departments use uniform procedures and methods in their epidemiology studies that are widely accepted by their peers. Data is not reported in the MMWR unless it is statistically significant. We find nothing in the record that would lead us to doubt the reliability of those procedures as they were applied in this particular instance. Furthermore, there is no conceivable motive for carrying out the studies in any other manner than to inform the public fairly and accurately. At the time the studies were undertaken, TSS had become an epidemic of significant proportion. There is no reason to believe that the CDC and Tri-State group approached this epidemic with anything but the professional impartiality they regularly apply to similar public crises. See Reyes v. Wyeth Laboratories, 498 F.2d 1264 n. 1 (5th Cir.), cert. denied, 419 U.S. 1096, 95 S.Ct. 687, 42 L.Ed.2d 698 (1974) (CDC studies admitted to show number of reported incidences of polio). Finally, the studies were clearly timely.

With the foregoing principles and considerations firmly in mind, we find Playtex's specific challenges to the trustworthiness and reliability of the studies unpersuasive. As an initial matter, Playtex argues that the trustworthiness of the reports is suspect because the studies constituted unpublished "preliminary epidemiologic data" and not "final findings of fact." Playtex notes that Ellis did not offer into evidence a December 18, 1980, article published in the New England Journal of Medicine that discussed and analyzed the CDC data.

We find no merit in this argument. All of the CDC data and much of the Tri-State data offered at trial were published in the MMWR. The findings of the Tri-State study were reported in published FDA regulations. See 21 C.F.R. Part 801. Under Fed.R.Evid. 803(8)(C), data need not be published to be admitted. The rule refers to findings "in any form." But the fact that these data were published adds to their reliability. The MMWR is the official publication of the CDC and the chief means by which that organization reports all major epidemiological findings. Doctors and scientists across the country regularly rely on the MMWR to keep track of important epidemiological developments. As a governmental publication and the voice of a prestigious research center, the MMWR carries with it an imprimatur of impartiality and reliability. No scientific facts are ever "final," for they are forever subject to further discoveries. Publication and survival of peer review, however, is one further indication that the reliability of the contested data far exceeds that normally accorded "preliminary data" and "speculation."

Playtex also argues that the studies are not reliable because no experts testified at trial about the methodology used to obtain the data. Playtex points out that the investigators conducting the studies did not have firsthand knowledge of the interviewee's symptoms or disease; the interviews occurred in some cases months after the alleged occurrence of TSS; the victims could not have been disinterested given the prospect of products liability litigation; and the studies did not indicate that there was a formal procedure or stated methodology for gathering data.

The burden rested with Playtex, however, not Ellis, to demonstrate that the methodology was flawed. The hearsay exception assumes regularity of procedure absent a showing to the contrary. See generally 4 Weinstein Evidence ¶¶ 803(8)[01] and [03]. Thus, whether the declarant or the investigator was available to testify about the methodology, or whether the report stated its methodology, are not by themselves, reasons to exclude the studies. As the Eighth Circuit has noted:

[A] party which would exclude evidence falling within the apparent scope of the hearsay exceptions must make an affirmative showing of untrustworthiness, beyond the obvious fact that the declarant is not in court to testify.

Kehm, at 618. See also Robbins v. Whelan, 653 F.2d 47 (1st Cir.), cert. denied, 454 U.S. 1123, 102 S.Ct. 972, 71 L.Ed.2d 110 (1981).

We do not find sufficient substance in the allegations of methodological flaws to conclude that Playtex has met its burden. First, practical considerations make it virtually impossible to interview all disease victims immediately after they have taken ill. In most cases researchers have no reason to undertake a study until well after an outbreak has occurred and there is cause for concern. There is little reason why a hiatus of several months should make a statistical study less reliable. Indeed, were we to accept Playtex's argument, very few scientific studies such as those at issue here would ever be held admissible.

Similarly, the sheer number of TSS victims interviewed makes it extremely difficult for investigators to have firsthand knowledge of each interviewee's symptoms. A patient afflicted with a serious disease, however, has a strong incentive to speak candidly with his doctor. Indeed, it is this practical consideration that forms the basis of the hearsay exception for statements made to a nurse or doctor concerning medical history, treatment, and diagnosis. See Fed.R.Evid. 803(4). There is little reason to believe, therefore, that statistical findings are less trustworthy where interviewers rely on doctors' reports rather than on patient interviews.

Finally, allegations of bias are purely speculative. All epidemiological studies that might implicate a manufactured product are conducted with the possibility of litigation on the horizon. It is highly unlikely, however, that a TSS victim would distort the nature of her injury when the possibility of a cure or preventive advice depends on accurate answers. The odd interviewee who might speak against interest in this way would not diminish the reliability of a study based upon a large statistical pool. Furthermore, the cure for bias — requiring cross-examination of each interviewee — would again have the practical result of precluding all public reports from being used at trial.

In our view, Playtex's concern about the methodology of the studies should have been addressed to the relative weight accorded the evidence and not its admissibility. This position, we believe, is consistent with the rationale behind the 803(8)(C) exception. Although the rule is designed to assume the admissibility of a report in the absence of affirmative indicia of untrustworthiness, there is no indication that Congress intended for the reports to escape searching examination. Allowing the jury to evaluate the reports after careful cross-examination and the presentation of expert testimony, therefore, serves both of these functions well; it permits admission without sacrificing scrutiny. See Migliorini v. United States, 521 F. Supp. 1210, 1218 (M.D.Fla. 1981) (CDC reports on swine flu vaccinations admitted along with testimony criticizing CDC study as methodologically flawed and biased).

Moreover, this approach — emphasizing the jury's role — avoids many of the problems that have plagued courts which have adhered to the traditional rule that the judge sit in judgment on scientific data. Until recently, most courts followed the "Frye" rule, which required the trial judge to determine the admissibility of scientific data based on whether the techniques used were "generally accepted by the relevant scientific community." In recent years, however, the Frye rule has come under increasing attack because of the importance it places on the judge's subjective ability to "count heads" among experts in the scientific community. Critics and courts that have rejected the Frye rule have argued that the acceptability of scientific data should be debated by experts before the jury. In an era when scientific data is playing an increasingly important role at trial, they argue, there is no reason not to let the jury see and evaluate the same data that the experts — whether they be doctors or other specialists — are relying on to reach their conclusions.

Evidence that falls within the 803(8)(C) exception, however, has already met the Frye standard. Because the evidence has been gathered and presented by a public agency it may be presumed to reflect methodologies accepted by the scientific community. Application of the federal rule, therefore, avoids the need to "count heads" and encourages increased examination of evidence by the jury in a manner not inconsistent with the rationale behind Frye. In short, by applying 803(8)(C) and thus requiring Playtex to challenge the credibility of the reports before the jury and not the judge, our decision today creates a result that permits the trial court and the litigants to skirt the difficulties posed by Frye without sacrificing the continued survival of the rule.

iii

Playtex argues that even if admission of the studies would have been proper under Fed.R.Evid. 803(8)(C), the court was justified in excluding the evidence because the inevitable debate over research techniques would have forced consideration of "an array of tangential issues" that would have unnecessarily confused the jury, prolonged the trial, and prejudiced Playtex. We disagree.

Under Fed.R.Evid. 403 the trial court is given discretion to balance the probative value of a piece of evidence against the likelihood that it will confuse and unduly prejudice the jury. The data that Ellis offered were highly probative of an issue of causation that was critical to the outcome of the case. As reported in the MMWR, the statistics were presented in a straightforward, objective manner. We believe that the data were understandable to a jury of average intelligence and experience. Indeed, the Tri-State data were contained in a letter from the author of the study that summarized the data and explained the methodology used.

The weight that a juror might ascribe to the data, of course, would turn on the credibility and persuasiveness of the experts that each side offered to explain the data. If one side failed to explain the data adequately, the jury would be entitled to give it little weight during its deliberations. The absence of such expert testimony, however, does not affect the admissibility of the data.

Given the importance of the causation issue to the outcome of the suit, an extended debate on the validity of the data was not only permissible; it was to be encouraged. That Playtex foresaw it would take considerable time and effort to explain why the data were not relevant or reliable is not a reason to exclude the data. To the contrary, if Playtex had a case to make, there is all the more reason to insist that the jury hear a full debate from the experts on the reliability of the data.

iv

As a final argument in opposition to the admission of the data, Playtex insists that even if the court's decision to exclude was erroneous, the error was harmless. Playtex points out that the data was introduced only to demonstrate that Playtex had notice of the TSS risk and numerous witnesses testified that the CDC data had been received by Playtex. Moreover, Playtex notes that even if the data were introduced for another purpose, the jury did not need to see the CDC data because six witnesses testified about the results of the CDC studies.

We conclude from the record that the data were introduced for the sole purpose of proving causation. We do not believe that the initial error of exclusion was rectified by reference made to the data in subsequent testimony. As discussed above, this data was highly relevant and probative on the issue of causation. The discovery of the link between TSS and menstruation was so recent that the jury may have viewed the plaintiff's theory with skepticism. In this situation there was no substitute for unedited data in the hands of the jury.

Because the jury verdict did not specify the basis for the jury's decision, we simply cannot be certain that the absence of the data did not prejudice the outcome. If, for example, the jury concluded Playtex had negligently failed to warn, the jury may have denied recovery because it believed Ellis had not established causation.

In sum, then, taking into account the rationale behind the public records hearsay exception and relevant case precedent, we conclude that the lower court committed reversible error in excluding the CDC and Tri-State data.

B.

Ellis claims that the district court erred in excluding consumer complaints received by Playtex because the complaints were probative of Playtex's knowledge about problems with their tampons at the time the tampons allegedly responsible for Margaret Ellis' death were manufactured. Ellis correctly points out that Playtex could not be found negligent for failing to warn unless it could be shown that the company was aware of the dangers its tampons posed.

The truth or falsity of the content of the complaints, however, was not in issue. The complaints were offered solely to establish notice, and this purpose was adequately served by the court's statement to the jury that Playtex had received complaints ranging from minor grievances to reports of TSS. Had the complaints been admitted, it was likely that the trial would have been unduly prolonged while Playtex challenged the substance of each complaint and sought to identify the precise type of tampon that each complainant used. These questions were clearly tangential to the main questions facing the jury. We do not believe, therefore, that the lower court abused its discretion in choosing to exclude the complaints under Fed.R.Evid. 403.

C.

At trial, an expert medical witness testifying on Ellis' behalf attempted to read from an article published in the medical journal Obstetrics and Gynecology. The article discussed the results of a study designed to determine whether tampon use caused alterations in the vaginal wall. The methodology employed by the study required the use of a colposcope to make a microscopic examination of vaginal mucosa. Alterations were identified by the presence of "microulcerations."

The lower court refused to permit the expert witness to read from the article. On appeal Ellis contends that the court committed reversible error because the testimony that the witness sought to give fell within the learned treatise exception to the rule against hearsay, and the testimony was probative of the health risks associated with Playtex Super Deodorant tampon use.

We simply are not convinced that the lower court abused its discretion in concluding that the potential prejudicial effect of the testimony outweighed its probative value. Apparently, Mrs. Ellis' vagina was not colposcopically examined, and there was no evidence that she suffered vaginal ulcerations. Moreover, the study did not indicate the type of tampon used by the subjects and their physical condition at the time of the examination. We do not believe, therefore, that it was unreasonable for the district court to conclude that under these circumstances the article was likely to cause confusion or invite unwarranted speculation by the jury about the facts of this case.

IV.

The appellant's final claim focuses on the charge that the court submitted to the jury on the failure to warn issue. The court instructed the jury that

[T]he plaintiff must prove by a preponderance of the evidence each of the following:

One, that the tampon product used by Margaret Ann Ellis was manufactured and distributed by the defendant.

Two, that the defendant knew or should have known that the tampon was dangerous for the use for which it was offered.

And three, that the defendant had no reason to believe that Margaret Ann Ellis, for whose use the chattel was offered, would realize its dangerous condition, and that such failure to warn was a proximate cause of the death of Margaret Ann Ellis.

* * * * * *

For a warning to be sufficient, to shift the risk of harm from the manufacturer to the user, the warning of a danger must be of a character reasonably calculated to bring home to a reasonably prudent person the nature and extent of the danger the manufacturer should reasonably foresee.

To be of such character, the warning must embody two characteristics.

First, it must be in such form that it could reasonably be expected to catch the attention of a reasonably prudent woman in the circumstances of its use.

Secondly, the content of the warning must be of such nature as to be understandable to the average users and to convey a fair indication of the nature and extent of the danger to the mind of a reasonably prudent user.

Ellis argues that this charge did not accurately reflect Virginia law. He insists that the court should have instructed the jury that:

A mere general warning of danger may be insufficient, and an insufficient warning is in legal effect no warning. Where a manufacturer is obligated to give an adequate warning of danger, the giving of an inadequate warning is as complete a violation of its duty as would be the failure to give any warning.

We have repeatedly held that a district court has "broad discretion in framing his charge to the jury." Chavis v. Finnlines Ltd., O/Y, 576 F.2d 1072, 1084 (4th Cir. 1978); see Smith v. University of North Carolina, 632 F.2d 316 (4th Cir. 1980). Taken as a whole, the court's charge was entirely consistent with Virginia law. See Featherall v. Firestone, 219 Va. 949, 252 S.E.2d 358 (1979). The critical elements of the appellant's proposed charge was effectively conveyed in the instructions the jury received. We do not believe Virginia law required the court to state expressly that an insufficient general warning is legally no warning.

V.

Because we believe that the exclusion of the CDC and Tri-State reports prejudiced the outcome of the trial, the matter is remanded for a new trial consistent with this opinion.

REVERSED AND REMANDED.

. . . . . .

(18) Learned treatises. To the extent called to the attention of an expert witness upon cross-examination or relied upon by him in direct examination, statements contained in published treatises, periodicals, or pamphlets on a subject of history, medicine, or other science or art, established as a reliable authority by the testimony or admission of the witness or by other expert testimony or by judicial notice. If admitted, the statements may be read into evidence but may not be received as exhibits.