Plaintiffs William D. Tart and Marion Tart appeal from a judgment for defendants entered in the United States District Court for the Southern District of New York after a jury trial before Judge Edmund L. Palmieri on plaintiffs' medical malpractice claims. On appeal, plaintiffs contend that a supplemental jury instruction by the trial judge misled the jury on a critical issue, and that the judge erroneously excluded evidence admissible under Rule 803(18) of the Federal Rules of Evidence, the "learned treatises" exception to the hearsay rule. For reasons set forth below, we reverse the judgment of the district court, and remand for further proceedings in accordance with this opinion.

Prior to the events at issue in this case, plaintiff William D. Tart was employed by the Celanese Corporation as an Assistant Chief Pilot.1 His employer required him to take an annual physical exam, which included a stress test. During the course of a stress test of the type at issue here, a patient walks or jogs at varying speeds and varying inclines on a treadmill while the physician monitors various lead systems attached to the patient's chest. Since 1973, plaintiff had been taking stress tests at the Life Extension Institute, a partnership of defendants Dr. John P. McGann and Dr. Ronald E. Costin. In September 1979, plaintiff took another such test there, administered by defendant Dr. Robert Mooney. Fifteen to twenty minutes after completing the test, while still at the Life Extension Institute, plaintiff suffered a heart attack. He was subsequently hospitalized for over a week. As a result of the heart attack, plaintiff lost his pilot license, and could therefore no longer continue to serve as a pilot for his employer.

According to plaintiffs, defendants failed to monitor and administer the stress test properly, in violation of good medical practice. Among other things, plaintiffs allege that Dr. Mooney should have stopped the test during the fourth stage, when Tart complained of "heavy fatigue." Instead, Dr. Mooney continued the test for several minutes, until the fourth stage of the test was completed. The jury found for defendants. Whether it did so on flawed instructions is the first issue before us.

On the fourth day of the trial, Judge Palmieri charged the jury in the morning. The jury deliberated the rest of the day and in the evening, and had certain testimony re-read. Finally, at 9:30 PM, the jury asked Judge Palmieri the following question:

Your Honor, can Mr. Tart's expression of fatigue be legally regarded as his request to stop the test?

The answer to this question may have been critical to the decision of this case. Although there was some dispute between the parties as to when a doctor should stop a stress test in response to physical indications of the patient's fatigue, the expert witnesses for both parties agreed that a stress test should be stopped if the patient himself requests that it be stopped.

Judge Palmieri answered the jury's question as follows:

Mr. Tart's expression of fatigue cannot be equated with a request to stop the test. It is a factual matter and not one that is subject to legal construction.

The judge apparently understood the jury's question to be whether Tart's expression of fatigue was the legal equivalent of a direction to stop, i.e., whether Tart's words had to be so construed. On this theory, the judge's answer was correct. But it is equally likely that the jury was asking not whether it had to so regard what Tart said, but whether it could do so. On that theory, the first sentence of the judge's response may have seriously misled the jury into thinking that it could not find Tart's expression of fatigue to be the factual equivalent of a request to stop the test. The judge's further explanation--that the issue was a factual matter not subject to legal construction--was correct. But after being told that Tart's statement "cannot be equated with a request to stop the test," the jury may not have understood that it could find just the opposite as "a factual matter." A better response to the jury's question would have been something like the following:

The expression of fatigue may be regarded as a request to stop, or it may not be; it is a question of fact for you to determine. In other words, whether a doctor in the position of Dr. Mooney should reasonably have regarded Mr. Tart's expression of fatigue as a request to stop is a question for you, the jury, to determine in light of all the circumstances, including Mr. Tart's condition, the duration of the test, the stage of the test, his previous medical history, and Dr. Mooney's acquaintance with that condition and history.

We recognize that trial judges in the heat of a trial are forced to respond quickly to jury requests of the sort at issue here, and that it is certainly far easier for an appellate court to fashion an appropriate response some months later, after much reflection. Nonetheless, the district court "must exercise special care to see that inaccuracy or imbalance in supplemental instructions do not poison an otherwise healthy trial." United States v. Carter, 491 F.2d 625, 633 (5th Cir.1974) (finding supplemental instructions in criminal case prejudicial to defendant). See also Arroyo v. Jones, 685 F.2d 35, 39 (2d Cir.), cert. denied, --- U.S. ----, 103 S. Ct. 468, 74 L.Ed.2d ---- (1982). This is especially true since the judge's "last word is apt to be the decisive word." Bollenbach v. United States, 326 U.S. 607, 612, 66 S. Ct. 402, 405, 90 L. Ed. 350 (1946) (reversing criminal conviction because of erroneous supplemental instruction). In this case, one of the issues before the jury was whether the stress test should have been stopped in response to plaintiff's expression of fatigue. After deliberating most of the afternoon and evening, the jury asked Judge Palmieri for additional guidance on this issue; having received it, the jury announced the verdict only a few minutes later. Under those circumstances, an unambiguous answer was critical. Since the instruction in this case was "on a vital issue and misleading," id., although unintentionally so, we believe that a new trial is called for. Cf. Bollenbach v. United States, supra, 326 U.S. at 612-13, 66 S. Ct. at 405; Norfleet v. Isthmian Lines, 355 F.2d 359, 362-64 (2d Cir.1966) (reversing on ground of erroneous supplemental charge).

Plaintiffs' second argument on appeal concerns the admissibility of so-called learned treatise evidence. During the course of the trial, plaintiffs attempted to bolster the testimony of their expert witness by introducing into evidence an article entitled "Maximal Exercise Testing." The article was written by Dr. Robert A. Bruce, who devised the Bruce-Protocol, which was used as the basis for the stress test taken by plaintiff William Tart. A stress test protocol is a set of directions that describes how a treadmill stress test should be conducted. It indicates, for example, the proper speed and incline of the treadmill at various stages of the test. The district court refused to allow plaintiffs to introduce the Bruce article into evidence as an exhibit or to quote from it on the direct examination of their expert.

The district court also severely restricted plaintiffs' use of learned treatise evidence on the cross-examination of defendants' expert. Plaintiffs sought to cross-examine defendants' expert by asking him whether he agreed with statements and figures contained in a booklet on exercise testing published by the American Heart Association. In particular, plaintiff wished to quote from a table on recommended target heart rates. Judge Palmieri refused to admit the table into evidence, and also refused to allow plaintiffs to quote from it. Judge Palmieri also instructed plaintiffs not to quote from the Bruce article, except to the extent it contradicted defendants' expert.

Judge Palmieri's evidentiary rulings on this issue evidently were based at least in part on a misunderstanding of the status of learned treatise evidence under the current Federal Rules. Judge Palmieri stated in explanation of his decision to prohibit use of the Bruce article as substantive evidence his belief that medical literature can only be used "as a cross-examination tool." But Rule 803(18) of the Federal Rules of Evidence provides as follows:

The following are not excluded by the hearsay rule, even though the declarant is available as a witness:

(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.

Thus, the Rule explicitly permits the admission of medical literature as substantive evidence "to the extent called to the attention of an expert witness upon cross-examination or relied upon by him in direct examination ...," as long as it is established that such literature is authoritative. The admissibility of such evidence is amply confirmed by the relevant case law, see, e.g., United States v. Mangan, 575 F.2d 32, 48 (2d Cir.) (admission of charts on handwriting characteristics), cert. denied, 439 U.S. 931, 99 S. Ct. 320, 58 L. Ed. 2d 324 (1978); Johnson v. Ellis & Sons Iron Works, Inc., 609 F.2d 820, 822-23 (5th Cir.1980) (admission of safety codes and standards); Maggipinto v. Reichman, 607 F.2d 621, 622 n. 4 (3d Cir.1979) (admission of statements from medical texts), and by the commentators, see, e.g., 11 J. Moore, Moore's Federal Practice Sec. 803(18); 4 J. Weinstein & M. Berger, Weinstein's Evidence p 803(18).

Prior to the enactment of Rule 803(18), learned treatises were generally usable only on cross-examination, and then only for impeachment purposes. See Weinstein, supra, p 803(18). Most commentators found the hearsay objections to learned treatise evidence unconvincing, and recommended that treatises be admitted as substantive evidence. Some commentators went so far as to suggest that treatises be admitted independently of an expert's testimony. Id. p 803(18). The Advisory Committee rejected this position, noting that a treatise might be "misunderstood and misapplied without expert assistance and supervision." Fed.R.Evid. 803(18) advisory committee notes. Accordingly, the Rule permits the admission of learned treatises as substantive evidence, but only when "an expert is on the stand and available to explain and assist in the application of the treatise ...." Id.

Nonetheless, it may not have been improper for Judge Palmieri to exclude the evidence at issue in this case. First, there was some dispute over the relevancy of the challenged evidence, particularly the chart on recommended target heart rates. Second, it does not appear that the evidence was properly proffered under Rule 803(18); the only rule explicitly referred to by plaintiffs was Rule 703, which deals with the bases of opinion testimony by experts. Cf. Johnson v. Ellis & Sons Iron Works, Inc., supra, 609 F.2d at 823 (refusal to admit safety code evidence not supportable on the basis that it was not properly proffered, because plaintiff did specifically refer to Rule 803(18)). Finally, it appears that the substance of much of the disputed evidence was made available to the jury. Thus, the exclusion of the evidence at issue may have been harmless error. But see id., 609 F.2d at 823 (direct quotation may be "more dramatic" and "more persuasive" than otherwise equivalent testimony). We need not decide whether exclusion of the evidence at issue was reversible error. We hold only that if on remand the disputed medical literature is shown to be relevant and authoritative, it may be read into evidence if properly proffered under Rule 803(18).2

The judgment below is reversed and the case is remanded to the district court for further proceedings in accordance with this opinion.