The United States appeals a judgment holding it liable to Leonard and Jean Harbeson and two of their children for the failure of three doctors at an Army medical center to respond to the Harbesons’ inquiries by disclosing the risks of an anticonvulsant medication on unborn children. The government contends the judgment against it is barred by collateral estoppel and that the district court erred in its findings of fact and conclusions of law. We affirm.

BACKGROUND

In December 1970, during Leonard Harbeson’s tour of' duty at an air force base, his wife, Jean, was diagnosed as having epilepsy. She received a prescription for the anticonvulsant drug, “Dilantin Kapseals” (Dilantin), from doctors at the air force base. Mrs. Harbeson was then pregnant with the couple’s first child, Michael, who was born in 1971 and is a healthy and normal child.

In the ensuing year, several other anticonvulsants were prescribed for Mrs. Harbeson, only to be discontinued after she suffered adverse reactions. Following her husband’s transfer to McChord Air Force Base in the State of Washington, Mrs. Harbeson was referred in May 1972 to Madigan Army Medical Center (Madigan) for evaluation and treatment. Dr. Green, a neurologist at Madigan, prescribed Dilantin and Phenobarbitol to control Mrs. Harbeson’s seizures.

Around this time the Harbesons contemplated having more children. Because of suspicions about the drug, the couple in November 1972 consulted Dr. Green about the possible risks of taking Dilantin during pregnancy. Later that month the Harbesons made the same inquiry of Dr. Weis, an intern at Madigan’s gynecology clinic. In July 1973 Mrs. Harbeson presented the same question to Dr. Larson, a resident in Madigan’s gynecology clinic, when she saw him because of a difficulty in conceiving. Each of the three Madigan doctors informed the Harbesons that taking Dilantin during pregnancy could cause cleft palate, which could be surgically repaired, and hirsutism, a temporary condition of excess hair. In reliance on this advice, the Harbesons decided to have additional children. Jean gave birth to Elizabeth in April 1974 and to Christine in May 1975, while taking Dilantin.

In responding to the Harbesons’ inquiries, none of the doctors conducted a literature search or consulted other sources for specific information concerning the effect of Dilantin on an unborn child, with the possible exception of Dr. Green’s consultation of the “Physicians’ Desk Reference” (PDR). The district court found that a literature search would have revealed several articles regarding the correlation of Dilantin and birth defects, including a “hallmark” article, Maternal Epilepsy & Abnormalities of the Fetus and Newborn, authored by B.A. Speidel and S.R. Meadow and published in October 1972 in the British journal. The Lancet.

After being alerted to the drug’s possible side effects on unborn children, the Harbesons took their daughters to a Seattle hospital for examination. Dr. Sterling Clarren of the University of Washington Dysmorphology Clinic diagnosed the children as having “Fetal Hydantoin Syndrome” (FHS). The first article describing this actual syndrome was published in the Journal of Pediatrics in August 1975. FHS may involve birth defects such as cleft palate, hirsutism, growth deficiencies, cardiac defects, skeletal anomalies, developmental defects, and mild to moderate retardation. Elizabeth and Christine Harbeson suffer from mild to moderate growth deficiencies, mild to moderate developmental retardation, and other physical, mental, and developmental defects.

The Harbesons brought an action against both Parke Davis, Inc., the manufacturer of the drug, and the United States. The action against the United States was pursuant to the Federal Tort Claims Act (FTCA), 28 U.S.C. § 2674 et seq. Both cases were tried simultaneously in November 1981.

A jury tried the Harbeson case against Parke Davis, and on December 7, 1981 returned a special verdict finding that Dilantin did not cause the children’s injuries. On that same day the district court entered judgment in favor of Parke Davis. Pursuant to 28 U.S.C. § 2402 , the court tried the Harbesons’ case against the United States. By agreement of the parties, the court took into consideration the pleadings and evidence submitted in the Parke Davis case. Testimony of additional expert witnesses was presented in the trial against the United States that was not presented in the jury trial.

On December 29, 1981 the court issued findings of fact and conclusions of law in favor of the Harbesons in their action against the government. On that same day the court on its own motion certified questions of law to the Washington Supreme Court. The court’s question was whether Elizabeth and Christine Harbeson could maintain a “wrongful birth” action. See Harbeson v. Parke Davis, Inc., 98 Wash.2d 460 , 656 P.2d 483 (1983). The Washington Supreme Court answered affirmatively. Id. On February 15, 1983 the district court entered judgment in favor of the Harbesons. At no time did the government move the court that the jury verdict on causation should collaterally estop the Harbesons’ action against the government.

I. Collateral Estoppel

A. Standard of Review

A question concerning the waiver of an affirmative defense involves the interpretation of Rule 8(c) of the Federal Rules of Civil Procedure and, as such, is a question of law reviewed de novo. Cf. Southeast Alaska Conservation Council, Inc. v. Watson, 697 F.2d 1305, 1309 (9th Cir.1983) (statutory interpretation). The applicability of the doctrine of collateral estoppel is primarily a question of law that may necessitate resolution of factual questions. See United States v. Geophysical Corp. of Alaska, 732 F.2d 693, 697 (9th Cir.1984).

B. Analysis

The government argues that the special verdict of the jury in the Parke Davis action that found Dilantin did not cause the birth defects should have precluded the district court from finding for the Harbesons in their action against the government. We do not reach the merits of this argument because we hold that the failure of the government to raise this issue below constituted a waiver of this affirmative defense.

Collateral estoppel is an affirmative defense that ordinarily is waived if not specially pleaded. See Fed.R.Civ.P. 8(c); Blonder-Tongue Laboratories, Inc. v. University of Illinois Foundation, 402 U.S. 313, 350 , 91 S.Ct. 1434, 1453 , 28 L.Ed.2d 788, 812 (1971). Because the action against Parke Davis was tried simultaneously with the action against the government, this case involves a situation where there was no opportunity to raise the defense in the pleadings. This court has stated that a preclusion defense that arises after the pleadings have been filed can be raised only by a motion for leave to file a supplemental answer under Rule 15(d). Weekes v. Atlantic National Ins. Co., 370 F.2d 264, 271 (9th Cir.1966). In so holding, however, we liberally treated an attempt to raise the defense as if it were a motion for leave to file a supplemental answer. See id.

The fact that collateral estoppel can be raised for the first time after the pleadings have been filed does not foreclose the possibility that the defense can be waived. The purpose behind requiring the pleading of this defense “is to give the opposing party notice of the plea of estoppel and a chance to argue, if he can, why the imposition of an estoppel would be inappropriate.” Blonder-Tongue, 402 U.S. at 350 , 91 S.Ct. at 1453 . This in turn allows each party to supplement the record to show why the defense should or should not be imposed. See id. The district court can then make whatever factual findings may be necessary, see Geophysical Corp., 732 F.2d at 697 , and determine as a matter of law whether the defense should be applied. This policy rationale is effectuated with equal force by requiring a party in the situation in which the government finds itself here to have raised the defense before the district court if it had a reasonable opportunity. Cf Blonder-Tongue, 402 U.S. at 350 , 91 S.Ct. at 1453 (case remanded to allow party to amend pleadings to assert estoppel for first time because party had no previous opportunity to plead estoppel); accord, Exxon Corp. v. Texas Motor Exchange of Houston, Inc., 628 F.2d 500 , 507 n. 3 (5th Cir.1980) (collateral estoppel will not be considered for the first time on appeal). To hold otherwise would be inconsistent with the spirit of Rule 8(c) of the Federal Rules of Civil Procedure, and would undermine judicial economy.

Based on the foregoing, we hold that the United States waived its right to assert the doctrine of collateral estoppel by failing to raise it before the district court. The government had more than a reasonable opportunity to raise the defense below. The judgment in favor of Parke Davis was entered on December 7, 1981 and judgment in this case was not entered until February 15, 1983. In the time since judgment in favor of Parke Davis, the government finished the trial of the Harbesons’ FTCA claim, made oral argument on motions to dismiss on other grounds, submitted written closing arguments, and participated in certification of questions of law to the Supreme Court of Washington. It is difficult to conceive of a situation where waiver would be more evident.

2. Negligence.

A. Standard of Review

We review the district court’s findings of fact under the clearly erroneous standard. Fed.R.Civ.P. 52(a). We must accept the district court’s factual findings unless we are left with the definite and firm conviction that a mistake has. been committed. United States v. United States Gypsum Co., 333 U.S. 364, 395 , 68 S.Ct. 525, 541 , 92 L.Ed. 746 (1948). The district court’s conclusions of law are reviewed de novo. United States v. McConney, 728 F.2d 1195, 1201 (9th Cir.1984) (en banc).

In an FTCA action, we must apply the law of the state in which the alleged act occurred. See 28 U.S.C. § 2674 . We “must ascertain from all available data” the applicable Washington law. Mardirosian v. Lincoln National Life Insurance Co., 739 F.2d 474 at 476 (9th Cir.1984). This Circuit reviews de novo the district court’s interpretation of the law of the state in which it sits. Matter of McLinn, 739 F.2d 1395 at 1403 (9th Cir.1984).

B. Analysis

Under Washington law, there are basically three theories under which a medical malpractice action may be brought — basic negligence, negligence as a matter of law, and informed consent. See Harbeson v. Parke Davis, Inc., 98 Wash.2d 460 , 656 P.2d 483, 489-91 (1983) (and statutes cited therein). The carefully worded findings of fact and conclusions of law of the district court reveal that its decision is consistent with any or all of these tort principles. The crucial findings of the court are that the doctors did not conduct a literature search and that a literature search would have revealed several articles, including the Speidel and Meadow article, regarding the correlation between Dilantin and birth defects. Because we hold the district court did not clearly err under the informed consent doctrine, we need not consider the applicability of the other theories to this case.

In 1972, the Washington Supreme Court adopted the doctrine of informed consent in ZeBarth v. Swedish Hospital Medical Center, 81 Wash.2d 12 , 499 P.2d 1, 8-9 (1972). See Harbeson, 656 P.2d at 490 , see also Watkins v. Parpala, 2 Wash.App. 484 , 469 P.2d 974 (1970) (informed consent theory first recognized in Washington). This cause of action may arise even though the doctor’s actions have not been negligent in any other way. See Holt v. Nelson, 11 Wash.App. 230 , 523 P.2d 211, 216-17 (1974). The Washington Supreme Court expressly found this doctrine applicable to the disclosure of “material information as to the likelihood of future children[] being born defective.” Harbeson, 656 P.2d at 491 .

The doctrine is premised on the fundamental principle that a competent individual has a right to determine what shall be done with her own body. Smith v. Shannon, 100 Wash.2d 26 , 666 P.2d 351, 354 (1983). To allow this determination the health care provider must provide the individual with sufficient information to make an “intelligent ” decision. Smith, 666 P.2d at 354 (emphasis in original); Canterbury v. Spence, 464 F.2d 772, 786-87 (D.C. Cir.), cert. denied, 409 U.S. 1064 , 93 S.Ct. 560 , 34 L.Ed.2d 518 (1972).

The elements of the doctrine of informed consent are: (1) the existence of a material risk unknown to the patient; (2) the failure to disclose the risk; (3) that had the risk been disclosed the patient would have chosen a different course; and (4) resulting injury. Harbeson, 656 P.2d at 490 ; see R.C.W. 7.70.050. If the plaintiff establishes those elements, the burden falls on the defendant to prove a defense to justify the failure to impart the material information. Miller v. Kennedy, 11 Wash. App. 272 , 522 P.2d 852, 861 (1974), aff’d 85 Wash.2d 151 , 530 P.2d 334 (1975); see Holt, 11 Wash.App. 230 , 523 P.2d 211, 218-19 (1974).

This doctrine does not require a physician to disclose “every risk which could be disclosed, if only because of the time required to disclose every remote risk.” Shannon, 100 Wash.2d 26 , 666 P.2d 351, 354 (1983) (quoting Waltz and Scheuneman, Informed Consent "to Therapy, 64 Nw.U.L.Rev. 628, 635 (1970)) (emphasis in original). A physician is required to disclose “only those [risks] of a serious nature.” Id., 666 P.2d at 355 . The guide for disclosure is materiality. Id.

We now reach the crucial issue confronting us — whether the risks Dilantin posed to unborn children were material. This concerns the first element of the informed consent doctrine as stated in Harbeson, 656 P.2d at 490 . The government does not seem to question seriously whether the other elements have been met.

“The test of materiality is an objective one incorporating the underlying concept of patient sovereignty.” Smith, 666 P.2d at 355 .

Id. (quoting Miller, 522 P.2d at 863 ). The determination of materiality is a two-step process. Shannon, 666 P.2d at 356 . The first step “is to define the existence and nature of the risk and the likelihood of its occurrence.” Adams v. Richland Clinic, Inc., P.S., 37 Wash.App. 650, 681 P.2d 1305, 1310 (1984); see Shannon, 666 P.2d at 356 . “Some” expert testimony is necessary to establish this aspect of materiality because only a physician or other qualified expert “is capable of judging what risks exist and the likelihood of occurrence.” Id. The second prong of the materiality test is for the trier of fact to decide whether the “probability of that type of harm is a risk which a reasonable patient would consider in deciding on treatment.” Id. This determination of materiality does not require expert testimony. Id. The focus is on whether a “reasonable person in the patient’s position probably would attach significance to the specific risk.” Miller, 522 P.2d at 863 . The crucial indicator is the “patient’s need.” Canterbury, 464 F.2d at 786 .

Against this backdrop, we now analyze the district court’s findings that the risks posed by Dilantin were material. The record contains expert testimony concerning what risks exist and their likelihood of occurrence. Dr. Robert G. Scherz testified that literature available to the medical profession in 1972 and 1973, such as the Speidel and Meadow article, “establishes to a reasonable probability that Dilantin at least causes cleft palate and growth deficiencies when taken during pregnancy and possibly skeletal anomalies, cardiac defects, developmental delays and multiple defects.” Furthermore, the PDR, a standard guide in the medical profession, contained during the relevant period the following warning concerning Dilantin: “Although evidence of a teratogenic [that which causes fetal malformations] effect in the human has not been established, the use of this drug in pregnancy requires that its potential benefits be weighed against possible hazards to the fetus.” Dr. Davis A. Stumpf, a witness for the government, testified that this warning alone would put a doctor on notice of developmental problems. He stated that this warning implies that there is evidence of teratogenic effects, even if such effect has yet to be proven. He considered developmental delay to be one of the common teratogenic effects of a drug.

Moreover, Dr. Morton Stenchever testified that the Speidel and Meadow article revealed in its results that major congenital malformations occurred with twice the expected frequency in children born to mothers taking anticonvulsants, of which Dilantin is one. The article expressly stated the “risk appears to be 2-3 times greater than normal.” Speidel and Meadow, Maternal Epilepsy and Abnormalities of the Fetus and Newborn, The Lancet, Oct. 21, 1972, p. 842. Dr. Stenchever also indicated the Speidel and Meadow study revealed that of the 192 pregnancies of the women taking Dilantin, nine had babies born with malformations. As stated by Dr. Stenchever, that is approximately five percent. This percentage apparently includes the risks of cleft palate and hirsutism, defects of which the Harbesons’ were warned. Dr. Stenchever further testified that risks from Dilantin were small. Nonetheless, he admitted that he used the Speidel and Meadow article in 1972 and 1973 when he advised his patients.

In the article the authors cautioned there is a “probability” that they “under-reported the incidence of anomalies in the epileptic group ... [because] the same method of ascertainment has yielded fewer anomalies in the control group than national figures lead one to expect.” Speidel and Meadow, supra at 842. Furthermore, “certain anomalies that are asymptomatic at birth may not have been recorded.” Id.

In light of the above evidence, we are satisfied that the first prong of the materiality test laid out. in Shannon has been met. See Shannon, 666 P.2d at 356 . Expert evidence disclosed the nature of the risks and their likelihood of occurrence. See Adams, 681 P.2d at 1310 (interpreting Shannon). In light of that evidence, these risks were “reasonably foreseeable.” See Holt, 523 P.2d at 218 . We further are satisfied that the second prong of the materiality test has been met. Based on the types of potential risks and their likelihood of occurrence, the district court did not err in finding as a fact that a reasonable patient would have considered these risks in deciding on treatment. See Shannon, 666 • P.2d at 356. Combining the findings on the two prongs of the test, we cannot say that the district court clearly erred in finding that the risks posed by Dilantin were material.

The government’s argument that the risks are so small as to not require disclosure is not persuasive. We do not believe the risks posed by this drug are but “imperceptible risk[s]” that need not be disclosed, Mason v. Ellsworth, 3 Wash.App. 298 , 474 P.2d 909, 920 (1970). The risk in Mason found to be so small that it need not be disclosed occurred “at most” .75 percent of the time. The likelihood of the risk here exceeds that likelihood in a significant manner. The fact that the risks enunciated by Speidel and Meadow included cleft palate and hirsutism, of which the Harbesons were warned, does not persuade us otherwise. Those were not even the most common risks. The most common anomaly discovered in that study was congenital heart disease, Speidel and Meadow, supra at 842, of which the Harbesons were not warned. Even the Physicians Desk Reference should have given the doctors notice of the teratogenic effects of the drug. The potential teratogenic effects of Dilantin posed “grave risks” that should have been disclosed. Harbeson, 656 P.2d at 494 (citing ZeBarth, 81 Wash.2d 12 , 499 P.2d 1 (1972)).

The government raises several other arguments in its attempt to convince us that the district court clearly erred. We are unconvinced. We have no reason to question the high value of Dilantin in the treatment of epileptics. The Speidel and Meadow article itself recognizes that the “value of anticonvulsants is great.” Speidel and Meadow, supra at 843. Our decision is consistent with any findings that the medication should not have been withdrawn from the market. The doctrine of informed consent does not exist to tell health care providers whether or not to offer certain treatment. The doctrine is intended to coexist with medically acceptable treatment forms. It seeks to allow a competent patient to weigh the value of the treatment against the risks posed. See generally Canterbury, 464 F.2d at 780-81 . The goal is to make the patient an active participant in the decisionmaking process. See Katz, Informed Consent, A Fairy Tale? Law’s Vision, 39 U.Pitt.L.Rev. 137, 160 (1977).

The government claims that Mrs. Harbeson's medical condition required the use of Dilantin and that if informed of the risk she may have stopped taking the medication. We construe this argument as invoking the exception to the doctrine based on the disclosure being detrimental to the patient’s best interest. See Holt, 523 P.2d at 218 . This argument has no merit. The Harbesons consulted the doctors to decide if they should have additional children, not to decide whether Mrs. Harbeson should terminate her treatment.

Further, we reject the government’s contention that the Speidel and Meadow article “did not contain credible evidence of a material risk” to warrant disclosure. First, we note that the district court found that “several articles” existed concerning the correlation of Dilantin and birth defects. For example, there is expert testimony that the warning in the PDR would put a physician on notice of the potential teratogenic effects of Dilantin. The district court’s decision, therefore, did not rest solely on the Speidel and Meadow article. Second, we believe a risk must be disclosed even if it is but a potential risk rather than a conclusively determined risk. A leading article on informed consent noted that

Waltz and Scheuneman, supra at 640 (1970) (emphasis added). Although the use of Dilantin may not be readily viewed as an innovative therapy, the policy expressed in this statement is applicable here. The Speidel and Meadow article pointed out what are at least potentially material risks. It may be that those risks had not yet been documented or accepted as a fact in the medical profession. Nonetheless, under the doctrine of informed consent, those risks should have been disclosed. Medical knowledge should not be limited to what is generally accepted as a fact by the profession. To hold otherwise would defeat the purpose of the doctrine, give little weight to exploratory medical research, and invite impossible line drawing.

Finally, we address the government’s concern expressed at oral argument as to “how a doctor ought to know that he doesn’t know” whether there is information that need be disclosed. To justify ignorance of this type of risk would insulate the medical profession beyond what is legally acceptable. Here, there is expert testimony of Dr. Scherz that it would be “just good basic medicine” to conduct a literature search or contact specialists in response to a direct question to a physician such as the one posed here. With the demands of their profession, no one can expect doctors to have all material information stored in their minds. We do not decide the extent to which a literature search must reach. Some limits are appropriate. This may best be defined by reference to what is material and reasonably available. As we have stated, a risk is not material unless expert testimony can establish its existence, nature, and likelihood of occurrence. See Adams, 681 P.2d at 1310 . Presumably, absent literature documenting research on these risks, there can be no expert testimony to establish these factors. A literature search will thus put a physician on notice of these risks. In this case there is evidence that the warning in the PDR was sufficient to put physicians on notice. In addition, the district court found several other articles would also have provided notice. It was not unreasonable to expect the doctors in this case to discover the risk.

CONCLUSION

The government’s failure to raise the collateral estoppel argument below has resulted in its waiver of that affirmative defense. The district court properly applied the law and its findings of fact are not clearly erroneous.

AFFIRMED.

. Because we rest our decision on waiver, we do not decide whether a jury verdict may have preclusive effect over an issue in an FTCA action. By statute, an FTCA action must be tried to the court. 28 U.S.C. § 1346 (b). The applicability of the collateral estoppel doctrine in this situation was not briefed by the parties and we find it unnecessary to reach it. . Compare Hayes v. Hulswit, 73 Wash.2d 796 , 440 P.2d 849 (1968) (basic negligence — breach of standard of the care determined by degree of care and skill expected of the average practitioner in the class to which the defendant belongs) with Helling v. Carey, 83 Wash.2d 514 , 519 P.2d 981 (1974) and Keogan v. Holy Family Hospital, 95 Wash.2d 306 , 622 P.2d 1246 (1980) (negligence as a matter of law — reasonable prudence requires certain actions irrespective of the standards of the profession). . The government contends in a footnote to its brief that the Washington Supreme Court "held" in Harbeson v. Parke Davis, Inc., 98 Wash.2d 460 , 656 P.2d 483, 494 (1983) that the applicable standard under which this case should be judged is that enunciated in ZeBarth v. Swedish Hospital Medical Center, 81 Wash.2d 12 , 499 P.2d 1 (1972). Although we would not label it as a holding, the Washington Supreme Court in Harbeson noted that at the time in question here the Washington doctrine of informed consent “was governed by” ZeBarth. Harbeson, 656 P.2d at 494 . Although recognizing some change or clarification in the doctrine of informed consent between the ZeBarth standard and the present standard, we are not convinced that we must apply the ZeBarth test instead of giving retroactive effect to the present test. See generally Bradley v. School Board of City of Richmond, Virginia, 416 U.S. 696, 711 , 94 S.Ct. 2006, 2016 , 40 L.Ed.2d 476 (1974) (court is to apply law in effect at the time it renders its decision, unless doing so would result in "manifest injustice”); City of Great Falls v. U.S. Dept. of Labor, 673 F.2d 1065, 1068-69 (9th Cir. 1982) (manifest injustice analysis). The parties’ briefs do not discuss this issue. We need not decide this issue because we would affirm the district court un der either the present or former standard for informed consent. We agree with the Washington Supreme Court that the teratogenic effects of Dilantin constituted "a grave risk” of which the Harbesons should have been informed. Harbeson, 656 P.2d at 494 (interpreting ZeBarth ). There is sufficient evidence to convince us that "a reasonably prudent medical specialist” in 1972 and 1973 would have informed the Harbesons of these “serious risks,” ZeBarth, 499 P.2d at 1 . See, e.g. Affidavit of Dr. Scherz, Testimony of Dr. Stumpf. The factual findings of the district court, ZeBarth, 499 P.2d at 10 , are not clearly erroneous. See also Mason v. Ellsworth, 3 Wash.App. 298 , 474 P.2d 909, 919-20 (1970) (discussed in n. 4, infra ). . While we conclude that the risks of Dilantin are material under the above analysis, we note that we could rest our decision on an analysis that avoids comparison of percentages. Because the Harbesons made a specific inquiry concerning the risks of Dilantin on pregnancy, this case presents a different angle on the materiality issue than did Mason and the other Washington cases. Although noting no specific request was made there, the court in Mason held, among other things, that if a patient inquired “as to any risks,” she must be informed of them. Mason v. Ellsworth, 3 Wash.App. 298 , 474 P.2d 909, 919 (1970). The court stated: If, at the time the physician discloses reasonably foreseeable risk[s], the patient inquires concerning any or all risks, then complete disclosure on the part of the physician is required. Correlatively, if the patient wishes to be informed of any risks beyond those reasonably foreseeable he has the duty of inquiry. Id. The omission of the doctors here would be negligent under this analysis as well.