Greaney, J.
In this products liability case, the plaintiff Florence Vassallo claimed that the defendants, Baxter Healthcare Corporation and Baxter International, Inc., were liable to her for damages because silicone breast implants, manufactured by a predecessor company to the defendants (Heyer-Schulte Corporation), that had been implanted in her were negligently designed, accompanied by negligent product warnings, and breached the implied warranty of merchantability, with the consequence that she was injured. The plaintiff Vincent Vassallo claimed a loss of consortium. The plaintiffs also asserted a claim for violation of G. L. c. 93A, §§ 2 (a) and 9.
A jury in the Superior Court heard the negligence and breach of warranty claims, returned verdicts (in response to special questions) on those claims in favor of the plaintiffs, and assessed damages. The judge entered a separate memorandum of decision on the G. L. c. 93A claim in which she found the defendants liable for a violation of that statute. The judge concluded that the defendants’ conduct was not knowing or wilful, determined that an assessment of additional compensatory damages would duplicate the damages determined by the jury, and awarded the plaintiffs their reasonable attorney’s fees and costs. The defendants appealed from the judgment, and we allowed their application for direct appellate review.
We conclude that the determinations of liability as to the negligence and G. L. c. 93A claims are correct, and the judgment can be affirmed on this basis. As a result, we need not consider the defendants’ arguments concerning the warranty findings.
We conclude, however, that we should change our products liability law to conform to the clear majority rule regarding what has to be shown to recover in a breach of warranty claim [4]*4for failure to warn of risks associated with a product, and we do so in Part 3 of this opinion.
The jury could have based their verdicts on the following evidence. In February, 1977, at the age of forty-eight, Mrs. Vas-sallo underwent breast implantation surgery. The silicone gel breast implants that Mrs. Vassallo received were manufactured by Heyer-Schulte Corporation in October, 1976. Through a series of corporate transactions, the defendants assumed responsibility for breast implant products manufactured by Heyer-Schulte.
In 1992, Mrs. Vassallo underwent a mammogram after complaining of chest pains that extended up under her left armpit. The mammogram revealed that her breast implants possibly had ruptured. The silicone gel implants were subsequently removed in April, 1993, and were replaced with saline implants. During the course of the explant surgery, the surgeon noted severe, permanent scarring of Mrs. Vassallo’s pectoral muscles which she attributed to the silicone gel. The implants themselves were encapsulated in scar tissue with multiple nodules of silicone granulomas. Dissection of the scar tissue capsules revealed that the left implant had ruptured, releasing free silicone gel, while the right implant was intact, but had several pinholes through which silicone gel could escape.
The plaintiff’s pathology expert, Dr. Douglas Shanklin, indicated that, based on the cellular responses shown in the pathology slides of Mrs. Vassallo’s breast tissue taken at the time of explant, the rupture had been longstanding, perhaps for several years. According to Dr. Shanklin, Mrs. Vassallo’s pathology slides showed silicone granulomas, giant cells, lymphocytes, and macrophages, all of which indicated a chronic immunological and inflammatory reaction to the silicone implants. Dr. Shan-klin also identified deposits of silica and lymphocytic vasculitis, which, he testified, were evidence that Mrs. Vassallo suffered from an autoimmune disease caused by the silicone gel.
Doctor Christopher Batich, professor of materials science and engineering at the University of Florida, testified for the plaintiffs on the effects of silicone in the body. He discussed animal studies that demonstrated migration of silicone to various organs both from ruptured gel implants and after intramuscular injection of “radio-labeled” liquid silicone. Doctor Batich also discussed the mechanism by which silicone can degrade to [5]*5low molecular weight materials in the body,3 and explained that the use of silicone gel in breast implants was unreasonably dangerous “[bjecause the material can get out of the implant, it can break up into small particles, it can travel through the body, and it can undergo chemical transformation into things that have biochemical effects.”
Doctor Bruce Freundlich, chief of rheumatology at Graduate Hospital in Philadelphia and an associate professor of medicine at the University of Pennsylvania, indicated that silicone gel breast implants can cause atypical connective tissue disease with a variety of symptoms that can include joint pain, dry eyes and mouth, difficulty sleeping leading to chronic fatigue, breast pain, fever, reduced sensation in the hands and feet, hair loss, itching, problems swallowing, and heartburn. Doctor Freundlich also offered his opinion that Mrs. Vassallo was suffering from atypical autoimmune disease, based on a review of her medical records and a physical examination that revealed the following symptoms: “tobacco pouch mouth,” or a tightening of the face around the mouth, which has been associated with scleroderma or mixed connective tissue disease; puffy fingers; an ulceration on one finger; thickening of the skin on her face and neck; te-langiectasia, or small red blood vessels, of the nose; hyperre-flexia; nocturnal myophonic jerking; dry eyes; elevated levels of antinuclear antibodies and IGA immunoglobin antibodies4; numbness and tingling in her hands; chronic fatigue; hair loss; difficulty swallowing; and problems with memory loss. According to Dr. Freundlich, Mrs. Vassallo’s problems were related to her exposure to silicone gel, and her future was “guarded.”
Doctor Eric Gershwin, chief of the division of rheumatology, immunology, and allergy at the University of California at Davis School of Medicine, discussed his own clinical research, and internal Dow Coming studies, to support his conclusion that silicone gel acts as an adjuvant to stimulate the body’s immune [6]*6system.5 Based on his research and treatment of more than 700 women with silicone gel breast implants, Dr. Gershwin stated that there is a unique constellation of symptoms seen in approximately five per cent of women with silicone breast implants, and that these symptoms, taken together, constitute an atypical autoimmune disease. Doctor Gershwin also reviewed Mrs. Vassalio’s medical records and concluded that her symptoms were consistent with this atypical autoimmune disease and were caused by her ruptured silicone gel breast implant.
There was also extensive testimony as to knowledge, attributable to the defendants, of the risks of silicone gel breast implants up to the time of Mrs. Vassalio’s implant surgery in 1977. According to Heyer-Schulte’s own internal correspondence, the company was aware of a “Talk Paper,” issued by the United States Food and Drug Administration in 1976, that documented migration to the brain, lungs, and heart, and death following injections of liquid silicone into the human body. In 1976, Heyer-Schulte received a report of an animal study, partially funded by Heyer-Schulte and conducted using miniature silicone gel implants supplied by Heyer-Schulte, that documented migration of gel from ruptured implants to the surrounding connective tissues and local inflammatory responses with fibroblastic activity and giant cell formation. The authors of the study stated: “The present tendency by manufacturers of breast implants towards ever thinner envelopes and a filler that is getting further away from gel and closer to silicone liquid must be looked at in the light of these experimental findings, and the question must be asked whether the possible advantages of these changes outweigh the disadvantages.” Heyer-Schulte was also aware that some of their implants were rupturing, having received 129 complaints of ruptured gel implants in 1976. In fact, the president of Heyer-Schulte had written in 1975 that “[presently, mammary implants have been designed to be increasingly fragile in response to plastic surgeons’ demand for softness, realistic feel and mobility.” As a result, Heyer-Schulte knew that its implants were “not consistent as far as durability or destructibility is concerned.” The encapsulation of the implant, and the viscous nature of the silicone gel, made it difficult to detect that a rupture had occurred, allowing the silicone to leak [7]*7into the body for long periods before explantation. By 1975, Heyer-Schulte also knew that, even without a rupture of the implant shell, the silicone gel could leak (known as “gel bleed”) through to the exterior surface of the implant and possibly produce “detrimental effect[s]” in the body.
Despite this knowledge of the possible adverse long-term consequences of leaking silicone in the body, Heyer-Schulte conducted few animal, and no clinical, studies to document the safety and efficacy of its silicone gel implants. When Heyer-Schulte began using silicone gel manufactured by Dow Coming in 1976, it relied primarily on the animal testing conducted by Dow Corning, despite the observations of a Heyer-Schulte scientist that “the data . . . [did] not answer questions concerning migration,” and “was lacking in quality and left many questions unanswered.” Heyer-Schulte did conduct toxicity testing on the Dow Coming gel; the gel passed the seven-day and thirty-day toxicity tests, but failed the ninety-day toxicity test based on the microscopic tissue evaluation that showed considerably greater fibrous tissue reaction and inflammation to the silicone gel than to the control material. There is no indication in the record that Heyer-Schulte ever repeated this ninety-day toxicity test, and the company continued to use the Dow Corning gel in the manufacture of its silicone gel breast implants.
Heyer-Schulte did furnish warnings to physicians concerning their silicone gel implants in a product insert data sheet (BIDS). The 1976 version of the BIDS that accompanied Mrs. Vassallo’s implants included warnings that the implant shell could be easily cut or ruptured by excessive stresses, and that Heyer-Schulte could not guarantee gel containment in the case of a rupture. The warnings did not address the issue of gel bleed, the fact that a rapture could result from normal stresses and could persist undetected for a significant time period, or the consequences of gel migration in the body. The BIDS also contained a list of potential complications associated with breast implants, but this list did not address the risks of chronic inflammation, permanent tissue scarring, or possible effects on the immune system. Broposed revisions to the BIDS, which would have included “a warning to the effect that uncontained silicone gel may have untoward consequences,” and complications of “migration of the silicone, with mild to severe consequences, including reduction of breast size and absorption of the silicone by the blood [8]*8and lymph systems, resulting in damage to the liver and kidneys,” were rejected by Heyer-Schulte’s president in March, 1976. The president did issue a letter to doctors dated August 23, 1976, which stated that “[i]f a shell is tom[ ] with time and normal stresses the gel will migrate,” and that “mild inflammation and polynuclear giant cell response characterized as mild foreign body reaction” had been associated with the silicone gel implants. Once again, this letter did not completely address the potential effects of silicone migration on the body’s immune system. Mrs. Vassallo stated that, if she had known that the implants could cause permanent scarring, chronic inflammation, and problems with her immune system, she would not have gone ahead with the implantation procedure. We now turn to the issues appropriate for discussion.
1. The defendants filed three motions in limine to exclude the testimony of the plaintiffs’ experts, Drs. Gershwin, Freundlich, and Batich. As to Dr. Gershwin’s anticipated testimony, the defendants’ motion sought to exclude his opinions that “silicone breast implants cause disease and that plaintiff[’s] breast implants causefd] her alleged illness.” As to Dr. Freundlich’s expert testimony, the defendants’ motion sought to exclude his opinions that “(1) silicone can cause an atypical connective tissue disease in some women; (2) plaintiff[’s] illness may be classified as an undifferentiated connective tissue disease; and (3)
. . . [the plaintiff’s] disease is related to her silicone gel breast implants.” As to Dr. Batich’s anticipated testimony, the defendants’ motion sought to exclude opinions “about what silicone does in the human body.” Each motion put forth a number of grounds as to why the expected opinions by the plaintiffs’ experts should be excluded.6 The plaintiffs filed a detailed opposition to the defendants’ motions. Together, the defendants’ motions, and the plaintiffs’ opposition, placed before the judge considerable documentary and other materials. The record prepared on the motions included references to medical and scientific publications, affidavits expressing the views of experts, and the transcript of a four-day evidentiary hearing, [9]*9held in the United States District Court for the District of Oregon,7 pursuant to the Daubert decisions,8 on defense motions in limine in several consolidated products liability actions to exclude expert testimony offered by the plaintiffs on causation issues relating to alleged injuries claimed from silicone breast implants.
The judge held a nonevidentiary pretrial hearing on the defense motions in limine. Although the defendants’ counsel initially argued that an evidentiary hearing should be held on the three motions, similar to the hearing conducted in the Federal District Court in Oregon, they abandoned that position during the hearing and maintained, as to the testimony of Drs. Gershwin and Freundlich, that their causation opinions could be admitted at trial only if “classical epidemiological studies” existed to support their causation conclusions.9 On this point, counsel for the parties indicated that their respective positions could be resolved on the conflicting materials submitted by each side concerning the relatively narrow issue whether classical epidemiological studies were a necessary predicate to the opinions expected to be offered by Drs. Gershwin and Freund-lich. As to the anticipated testimony of Dr. Batich (and to a part of Dr. Gershwin’s testimony), the defendants’ trial counsel argued, primarily, that the testimony was unreliable because experts supplying data for Dr. Batich’s conclusions had relied on certain studies involving the breakdown of silicone which had used, in the defendants’ opinion, a scientifically questionable technique called nuclear magnetic resonance spectography (NMRS).10 Once again, as had been done before, the parties left [10]*10the judge to decide the issue on the paper record without a separate evidentiary hearing.
The judge did not rule on the motions in limine at the conclusion of the pretrial hearing. The judge’s final rulings were made at the trial, when each of the three plaintiffs’ experts offered their opinions. At trial, the defendants’ counsel objected to opinions elicited from Drs. Freundlich and Gershwin as to whether “silicone gel breast implants can cause an atypical connective tissue disease in someone.” At this critical point, in response to a direct question asked by the judge, the defendants’ counsel specified that their objection was based solely on the ground previously advanced at the pretrial hearing on the motions in limine that the opinions to be put before the jury by Drs. Gershwin and Freundlich were inadmissible “without epi-demiologic studies.” The judge ruled on the precise objection then made as follows: “Given that that’s the ground, given the record that we have here on the motion[s] in limine, I’ll deny the motion[s].”
With respect to Dr. Batich’s testimony at trial, the defendants’ counsel made only the argument, referred to above, that his testimony was inadmissible because he had relied on studies using the NMRS technique. The judge overruled the objection to Dr. Batich’s testimony, stating that, on the basis of the record developed at the Federal District Court hearing in Oregon, including the testimony of a court-appointed independent expert in chemistry, “there is a sufficient basis to accept Dr. Batich’s testimony, including any reliance on the [disputed NMRS] experiments and articles to satisfy a standard ... of reliability.” The judge left the defendants’ criticism of Dr. Batich’s testimony for cross-examination.
The defendants make a wide assortment of arguments on appeal that the three challenged expert opinions, to use the caption headings in their appellate brief, did not meet “the Frye Test of General Acceptance in the Scientific Community,”11 that the “Medical Community Does Not Accept Plaintiffs’ Experts’ Conclusions,” and that “Other of the Lanigan/Daubert Factors Do Not Support the Trial Court’s Denial of [the Defendants’] [11]*11Lanigan Motions.”12 We need not consider any of the defendants’ arguments for the reasons now discussed.
The defendants did not press for an evidentiary hearing on their motions in limine (and they do not argue now that they should have had one). The defense was content, as the trial approached, to have the judge consider and decide their contentions on the record placed before her (which was primarily a paper record), and the defense limited its objection to a single point in the case of each expert. The particular point was reiterated at the crucial moments in the trial when the judge was called on to make a final ruling on the motions in limine.
On a record where objections are properly preserved, we may examine, on a de novo basis, a trial judge’s decision made pursuant to the standards set forth in Commonwealth v. Lani-gan, 419 Mass. 15 (1994) (Lanigan), to admit or exclude certain scientific expert evidence. Commonwealth v. Vao Sok, 425 Mass. 787, 797 (1997). Where, however, the party who has lost the evidentiary ruling preserves only a limited basis for review, we ordinarily examine only that basis, consistent with the established principle governing appellate review in civil cases that issues not properly raised in the trial court will not be considered on appeal. This principle was stated by Justice Quirico, writing for the Appeals Court in Commonwealth v. Olson, 24 Mass. App. Ct. 539, 544 (1987), as follows:
“A lawyer cannot try a case on one theory and then, having lost on that theory, argue before an appellate court about alleged issues which might have been, but were not, raised at the trial. That is true whether or not the trial lawyer and the appellate lawyer are the same person, and whether or not they are in any way associated. The appeal must be based on what took place at the trial, not on anything which is presented for the first time before an appellate court.”
See Huber v. Huber, 408 Mass. 495, 497 (1990); Fishman v. Brooks, 396 Mass. 643, 649 (1986); Kagan v. Levenson, 334 [12]*12Mass. 100, 107 (1956). See also Commonwealth v. Shea, 401 Mass. 731, 740 (1988) (defendant failed to renew objection when evidence sought to be suppressed was introduced at trial); Commonwealth v. Gabbidon, 398 Mass. 1, 7 (1986) (filing of motion in limine to prevent the introduction of evidence does not, by itself, preserve appellate rights); Freyermuth v. Lutfy, 376 Mass. 612, 616 (1978) (“The consequence of the failure to object is to waive the objection to the testimony”).
The judge properly concluded that the causation opinions of Drs. Gershwin and Freundlich could be admitted in the absence of classical epidemiological studies. Both witnesses possessed the knowledge, skill, and experience to qualify as experts in their fields, having conducted research, published in peer reviewed journals on silicone related topics, and treated hundreds of women with silicone gel breast implants in their clinical practices. At the Daubert hearing held by the Federal District Court in Oregon, the transcript of which was included in the record before the judge, Dr. Gershwin described the bases for his opinion that silicone gel breast implants cause disease in women as including animal studies and controlled, blinded, clinical studies showing stimulation of the immune system by silicone; animal studies showing migration of the silicone; clinical studies showing local complications of silicone, including chronic inflammation; and the differential diagnosis method of identifying a patient’s symptoms as being among those associated with silicone gel breast implants and ehminating any other causes for these symptoms. Although there was conflicting testimony at the Oregon hearing as to the necessity of epidemiological data to establish causation of a disease, the judge appears to have accepted the testimony of an expert epidemiologist that, in the absence of epidemiology, it is “sound science ... to rely on case reports, clinical studies, in vivo tests, [and] animal tests.”13 The judge may also have relied on the affidavit of the plaintiffs’ epidemiological expert, Dr. David S. Egilman, who identified several examples in which disease causation has been established based on animal and clinical case studies alone to demonstrate that “doctors utilize epidemiological data as one tool among many.”
[13]*13The judge’s decision, that the plaintiffs’ evidence of general causation (specifically the testimony of Drs. Gershwin and Freundlich) was scientifically valid despite the absence of supporting epidemiological data, is consistent with decisions by other courts. See Benedi v. McNeil-P.P.C., Inc., 66 F.3d 1378, 1384 (4th Cir. 1995) (holding that epidemiological studies were not necessary to provide causal link between acetaminophen and liver damage in a person who regularly consumed alcohol, because the methodology underlying the expert’s conclusion was sound); Hopkins v. Dow Corning Corp., 33 F.3d 1116, 1124 (9th Cir. 1994), cert, denied, 513 U.S. 1082 (1995) (upholding the District Court’s decision to admit plaintiff’s expert testimony regarding causal link between plaintiff’s silicone gel breast implants and her mixed connective tissue disease, because the “experts based their opinions on the types of scientific data and . . . techniques relied upon by medical experts in making determinations regarding toxic causation where there is no solid body of epidemiological data to review”); Jennings v. Baxter Healthcare Corp., 152 Or. App. 421, 437 (1998) (concluding that the trial court did not err in allowing testimony as to silicone-related disorders based on animal research and clinical experience using the differential diagnosis technique). Contrast In re Breast Implant Litig., 11 F. Supp. 2d 1217 (D. Colo. 1998) (excluding expert testimony regarding causal link between silicone gel implants and autoimmune or systemic disease); Hall v. Baxter Healthcare Corp., 947 F. Supp. 1387 (D. Or. 1996) (excluding all testimony concerning a causal link between atypical connective tissue diseases and silicone gel implants as not based on accepted scientific testimony and not admissible under Daubert); Kelley v. American Heyer-Schulte Corp., 957 F. Supp. 873 (W.D. Tex. 1997) (excluding expert testimony by plaintiff’s epidemiologist and rheumatologist regarding a causal link between plaintiff’s implants and her symptoms of Sjogren’s Syndrome). Based on our review of the trial record, we cannot say that the judge erred in allowing the opinions of Drs. Gershwin and Freundlich as to causation in the absence of supporting epidemiological data. We emphasize that the relevant issue was the scientific validity of the methods relied on by the experts to form their conclusions. See Commonwealth v. Lanigan, supra at 25. Challenges to the conclusions themselves go to the weight, not the admissibility, of the evidence, see Commonwealth v. Gomes, [14]*14403 Mass. 258, 273 (1988), and the defendants were given ample opportunity to challenge these conclusions through cross-examination and the direct testimony of their own expert witnesses.
The judge also had a basis for concluding that the opinion of Dr. Batich on the degradation of silicone in the human body should be admitted. Doctor Batich’s testimony was based on his own in vitro and in vivo research on the effects of silicone in the body, internal studies conducted by Dow Coming, and the NMRS studies conducted by Dr. Leoncio Garrido, see note 10, supra. Doctor Batich’s opinion was also supported by the testimony of Dr. Shanklin, who observed silica, a by-product of silicone degradation, in the tissue samples removed from Mrs. Vassallo at the time of explant. Doctor Shanklin’s testimony preceded the judge’s ruling on the defendants’ motion in limine with respect to Dr. Batich, and was received without objection. The only basis for the defense motion in limine preserved for appeal was the scientific validity of Dr. Garrido’s NMRS study.14 This issue had been discussed in detail at the Daubert hearing, held by the Federal District Court in Oregon, by expert witnesses who both opposed and supported the scientific reliability of Dr. Garrido’s research. In addition to the transcript of that hearing, the judge had before her the draft report of the independent chemistry expert appointed to review the testimony in the Oregon hearing, Dr. Ronald McClard, who concluded that, although Drs. Garrido and Macdonald (who conducted similar research) reached different results in their studies, “[Dr.] Garrido’s claimed observations of [silicone] in the NMR is supported by scientifically valid reasoning and methodology . . . and are clearly relevant to the matter at hand.” The judge stated that she relied on both the transcript of the Oregon hearing and the report of Dr. McClard in ruling that Dr. Batich’s testimony was sufficiently reliable to satisfy the applicable standard for [15]*15admissibility. Given the conclusion of Dr. McClard, we cannot say that the judge erred in denying the defense motion in limine with respect to Dr. Batich.
In view of the narrow issue before us, we need not decide whether, as argued by the plaintiffs, the opinions of their experts fall within those cases which accept expert testimony based on personal observations, clinical experience, or generally accepted scientific techniques without application of the possibly more rigorous analysis set out in Lanigan, supra,15 or whether, as argued by the defendants, the opinions require detailed consideration under Lanigan for determination of their inherent scientific reliability. We note, however, that the Federal courts (and other State courts) have taken the latter view and have conducted inquiries under the Daubert decisions (or the particular State counterpart). See, e.g., Hall v. Baxter Healthcare Corp., supra; In re Breast Implant Cases, 942 F. Supp. 958 (E. & S.D.N.Y. 1996); Jennings v. Baxter Healthcare Corp., supra. We also point out as well that a national panel of experts has been appointed pursuant to Fed. R. Evid. 706 to try to arrive at a consensus concerning the admissibility of expert testimony of the type involved in this case. See Hall v. Baxter Healthcare Corp., supra at 1394. The report of this rule 706 panel may resolve the controversy surrounding claims of atypical connective tissue disease related to silicone breast implants.
2. We next discuss the other issues argued as a basis for a new trial by the defendants.
(a) The judge properly prevented the defendants’ experts (as well as the plaintiffs’ experts) from testifying on direct examination to the out-of-court opinions of other scientists in the [16]*16absence of some specific exception to the hearsay rule (none was shown). The judge followed our established rule. Grant v. Lewis/Boyle, Inc., 408 Mass. 269, 273 (1990). Her ruling conformed with our decision in Department of Youth Servs. v. A Juvenile, 398 Mass. 516, 531 (1986), in which we declined to adopt Proposed Mass. R. Evid. 703 (the equivalent of Fed. R. Evid. 703), and instead took “a modest step by permitting an expert to base an opinion on facts or data not in evidence if the facts or data are independently admissible and are a permissible basis for an expert to consider in formulating an opinion.” See Commonwealth v. Waite, 422 Mass. 792, 803 (1996) (“When an expert provides the jury with an opinion regarding the facts of the case, that opinion must rest on a proper basis, else inadmissible evidence might enter in the guise of expert opinion”). There is nothing in Commonwealth v. Lanigan, supra, that casts doubt on, or suggests the need to reconsider, this rule. The judge correctly ruled that the defendants could use scientific studies and literature in cross-examination of the plaintiffs’ experts.16
(b) There was no error in the judge’s rulings on certain items of evidence pertaining to the issue of notice to Heyer-Schulte of the safety of the gel implants. The defendants claim that they were prejudiced by the judge’s rulings allowing the plaintiffs to introduce product complaints and Dow Corning studies that postdated Mrs. Vassallo’s implants, while preventing the defendants from introducing documents allegedly reporting on implant safety assessments post-1977. The record does not support the defendants’ contentions.
First, the judge properly prohibited the defendants from asking their experts about the specifics of post-1977 published literature, including the results of a 1982 consensus development conference on silicone gel breast implants, because this testimony was hearsay with no independent basis for admission. See Grant v. Lewis/Boyle, Inc., supra. The judge did not abuse her discretion in rating that the post-1977 silicone gel published literature was irrelevant on the issue of notice to Heyer-Schulte of problems related to Mrs. Vassallo’s implants. This rating was equally enforced against both the plaintiffs and the defendants.
[17]*17Second, the internal Dow Coming studies were admissible on grounds other than notice, that is, as business records providing evidence concerning the characteristics of the silicone gel used in Mrs. Vassallo’s implants. Post-1977 Dow Coming studies remained admissible on these grounds because the gel composition did not change during these later years. The defendants took advantage of the ruling allowing post-1977 Dow Coming studies by introducing in evidence a 1986 Dow Coming study that favored their case. The defendants’ reliance on the “curative admissibility doctrine” to justify this action is misplaced; this doctrine is only applicable if the original evidence being rebutted was improperly admitted, which is not the case here. See Commonwealth v. Ruffen, 399 Mass. 811, 813-814 (1987).
Third, the defendants never objected to any testimony concerning post-1977 complaints, and again, took advantage of this evidence to ask their materials expert, Dr. Donald Uhl-mann, about the number of mpture complaints received from physicians from 1974 to 1990 for implants manufactured in 1976. The defendants did object to sending post-1977 complaints to the jury during a discussion of the exhibits after the close of testimony. At that point, the judge properly mled that the post-1977 complaints were admissible to show notice to Heyer-Schulte of specific problems experienced by purchasers and users of their silicone gel implants and were relevant to the issue of Heyer-Schulte’s continuing duty to warn of risks discovered after tire product at issue had been manufactured. See doCanto v. Ametek, Inc., 367 Mass. 776, 784-785 (1975).
(c) The judge properly allowed the plaintiffs to introduce evidence of complaints made to Heyer-Schulte on a year-by-year basis with an appropriate cautionary instruction to the jury that the complaints were “not admitted for the truth of what the complaints say,” but for the jury’s “consideration of what Heyer-Schulte was being told and was learning about its product.” The complaints of substantially similar defects and consequences were admissible as evidence of notice to Heyer-Schulte of defects in the integrity of its product about which it had failed to warn. See Burnham v. Mark IV Homes, Inc., 387 Mass. 575, 585 (1982); H.P. Hood & Sons v. Ford Motor Co., 370 Mass. 69, 75 (1976).
Moreover, the defendants’ contention that the plaintiffs were allowed to present “undifferentiated boxes of complaints without any foundational showing that the products or griev-[18]*18anees at issue were related to Mrs. Vassallo’s” is not supported by the record. Prior to submitting the complaints to the jury, those concerning the single lumen gel implants were identified separately from the double lumen implants, which contained both gel and saline, and all complaints were categorized by year and type of problem (i.e., rupture, migration, and tissue reaction). Complaints concerning the double lumen implants were relevant with respect to rupture, because the shells of these implants were made of the same silicone material as the gel implants. The judge gave a limiting instruction to the jury that complaints concerning the types of implants other than the gel implant received by Mrs. Vassallo were only relevant insofar as notice of problems with the implant shell. This limiting instruction was sufficient to overcome any prejudice complained of by the defendants.
(d) There was no error in the admission of internal Dow Coming research documents concerning silicone. As previously discussed, these studies were admissible as business records, see G. L. c. 233, § 78, providing evidence concerning the characteristics of the silicone gel used in Mrs. Vassallo’s implants, and both parties took advantage of the judge’s ruling. At trial, the defendants concurred that the studies were Dow Coming business records, but maintained that they were still inadmissible hearsay. The defendants now offer the specific argument that the Dow Corning studies were inadmissible as business records because they largely contained opinions. See Julian v. Randazzo, 380 Mass. 391, 393 (1980) (police report, comprising investigating officer’s opinion and recommendation, not admissible); Burke v. Memorial Hosp., 29 Mass. App. Ct. 948, 949 (1990) (“Opinions contained in business records are not admissible unless they fall within some other exception to the hearsay rule”). However, the Dow Coming studies can be distinguished from the cases offered by the defendants. The police report in Julian, supra at 394, was “second-level” hearsay, consisting of information gained by interviewing other police officers and bystander witnesses. The records offered in Burke, supra at 950, were employee performance evaluations that “by their very nature consisted almost entirely of judgmental evaluations and opinions.” By contrast, the Dow Coming documents were scientific studies containing primarily factual data. While these studies may also have contained opinions regarding the authors’ interpretations of the data, the [19]*19defendants’ objections were to the studies as a whole. No request was made to strike or exclude those portions containing inadmissible opinions. The defendants’ motion was, therefore, overbroad as including competent evidence relevant to the composition and physiological effects of the silicone gel used in Mrs. Vassal-lo’s implants. See Griffin v. General Motors Corp., 380 Mass. 362, 365 (1980).
(e) The defendants correctly point out that evidence of failure adequately to test a product is relevant to claims of design, manufacturing, or warning defects, but does not furnish a separate, independent basis for liability. See J.R. Nolan & L.J. Sartorio, Tort Law §§ 292-295 (2d ed. 1989 & 1998 Supp.); Restatement (Second) of Torts § 402A (1965). The judge’s instruction that the jury could consider product testing in evaluating the design defect claim was not in error. However, the defendants argue that the judge based her findings of negligence and breach of warranty in connection with the G. L. c. 93A decision in part on inadequate testing when, as contended by the defendants, there was no evidence that additional testing would have disclosed otherwise unknown information affecting the design of the product or the provision of warnings. See Mason v. General Motors Corp., 397 Mass. 183, 192 (1986) (failure to test not relevant absent a showing that useful information would have been derived from test). Regardless of whether additional testing performed by Heyer-Schulte would have revealed additional risks associated with the silicone gel implants, any error committed by the judge related to the issue of testing was nonprejudicial as the judge had independent bases to find negligence and a violation of G. L. c. 93A.
3. The evidence warrants the jury’s findings (and their verdicts) on the negligence claims, and the defendants’ arguments seeking judgment in their favor or a new trial on the negligence claims have been considered and rejected. No issue is raised as to the amount of damages awarded.
Because the plaintiffs’ recoveries can be upheld on the jury’s findings of negligence, we need not address the defendants’ claims of error concerning the breach of warranty count. We take this opportunity, however, to consider the defendants’ argument that we should change our products liability law concerning the implied warranty of merchantability from what is stated in Hayes v. Ariens Co., 391 Mass. 407, 413 (1984), and that the law should be reformulated to adopt a “state of the art” standard [20]*20that conditions a manufacturer’s liability on actual or constructive knowledge of the risks.
Our current law, regarding the duty to warn under the implied warranty of merchantability, presumes that a manufacturer was fully informed of all risks associated with the product at issue, regardless of the state of the art at the time of the sale, and amounts to strict liability for failure to warn of these risks. See Simmons v. Monarch Mach. Tool Co., 413 Mass. 205, 207-208 n.3 (1992); Hayes, supra. This rule has been justified by the public policy that a defective product, “unreasonably dangerous due to lack of adequate waming[s], [is] not fit for the ordinary purposes for which [it is] used regardless of the absence of fault on [a defendant’s] part.” Simmons, supra.
At trial, the defendants requested a jury instruction that a manufacturer need only warn of risks “known or reasonably knowable in light of the generally accepted scientific knowledge available at the time of the manufacture and distribution of the device.” The judge declined this request, and instead gave an instruction using language taken almost verbatim from that in Hayes, supra. While the judge’s instruction was a correct statement of our law, we recognize that we are among a distinct minority of States that applies a hindsight analysis to the duty to warn.17
The majority of States, either by case law or by statute, follow the principle expressed in Restatement (Second) of Torts § 402A comment j (1965), which states that “the seller is required to give warning against [a danger], if he has knowledge, or by the application of reasonable, developed human skill and foresight should have knowledge, of the . . . danger.” See, e.g., M.F. Daller, Product Liability Desk Reference — a Fifty-State Compendium (1996 ed.); Annot., Strict Products Liability: Liability for Failure to Warn as Dependent on Defendant’s Knowledge of Danger, 33 A.L.R.4th 368, 377 (1984 & 1997 Supp.); Restatement (Third) of Torts: Products Liability, Reporters’ Note to comment m, at 104 (1998) (“An overwhelming [21]*21majority of jurisdictions supports the proposition that a manufacturer has a duty to warn only of risks that were known or should have been known to a reasonable person”). At least three jurisdictions that previously applied strict liability to the duty to warn in a products liability claim have reversed themselves, either by statute or by decision, and now require knowledge, or reasonable knowability as a component of such a claim. See Fibreboard Corp. v. Fenton, 845 P.2d 1168, 1172-1173 (Colo. 1993); Feldman v. Lederle Labs., 97 NJ. 429, 455 (1984); La. Rev. Stat. Ann. § 9:2800.59(B) (West 1997). The change in the law of New Jersey is particularly relevant, because we relied in part on New Jersey law in formulating the strict liability standard expressed in the Hayes decision. See Hayes, supra at 413, citing Beshada v. Johns-Manville Prods. Corp., 90 NJ. 191, 202-207 (1982).
The thin judicial support for a hindsight approach to the duty to warn is easily explained. The goal of the law is to induce conduct that is capable of being performed. This goal is not advanced by imposing liability for failure to warn of risks that were not capable of being known. See Henderson, Doctrinal Collapse in Products Liability: The Empty Shell of Failure to Warn, 65 N.Y.U. L. Rev. 265, 274 & n.32 (1990).
The Restatement (Third) of Torts: Products Liability § 2(c) (1998), recently approved by the American Law Institute, reaffirms the principle expressed in Restatement (Second) of Torts, supra at § 402A comment j, by stating that a product “is defective because of inadequate instructions or warnings when the foreseeable risks of harm posed by the product could have been reduced or avoided by the provision of reasonable instructions or warnings . . . and the omission of the instructions or warnings renders the product not reasonably safe.” The rationale behind the principle is explained by stating that “[ujnforeseeable risks arising from foreseeable product use ... by definition cannot specifically be warned against.” Restatement (Third) of Torts: Products Liability, supra at § 2 comment m, at 34. However, comment m also clarifies the manufacturer’s duty “to perform reasonable testing prior to marketing a product and to discover risks and risk-avoidance measures that such testing [22]*22would reveal. A seller is charged with knowledge of what reasonable testing would reveal.” Id.18
We have stated that liability under the implied warranty of merchantability in Massachusetts is “congruent in nearly all respects with the principles expressed in Restatement (Second) of Torts § 402A.” Commonwealth v. Johnson Insulation, 425 Mass. 650, 653-654 (1997), quoting Back v. Wickes Corp., 375 Mass. 633, 640 (1978). The main difference has been our application of a hindsight approach to the duty to warn of (and to provide adequate instructions regarding) risks associated with a product. We recognize that this approach has received substantial criticism in the literature, see note 18, supra, and, in fact, has not been uniformly applied by Massachusetts State and Federal courts. See, e.g., Boston v. United States Gypsum Co., 37 Mass. App. Ct. 253, 261-262 (1994) (noting, but not addressing, jury instruction that for a breach of warranty claim, the “duty to warn extends only to such dangers or defects about which the manufacturer either actually knew . . . [or] which it reasonably should have known”); Welch v. Keene Corp., 31 Mass. App. Ct. 157, 163 & n.2 (1991) (noting that, for a breach of warranty claim, “a manufacturer has a duty to warn only as to those dangers about which the manufacturer actually knew or about which it reasonably should have known”); Kotler v. American Tobacco Co., 926 F.2d 1217, 1231-1232 (1st Cir. 1990), vacated on other grounds, 505 U.S. 1215 (1992) (same); Anderson v. Owens-Illinois, Inc., 799 F.2d 1, 4 (1st Cir. 1986) (same); Wasylow v. Glock, Inc., 975 F. Supp. 370, 378 (D. Mass. 1996) (same).
In recognition of the clear judicial trend regarding the duty to warn in products liability cases, and the principles stated in Restatement (Third) of Torts: Products Liability, supra at § 2 [23]*23(c) and comment m, we hereby revise our law to state that a defendant will not be held liable under an implied warranty of merchantability for failure to warn or provide instructions about risks that were not reasonably foreseeable at the time of sale or could not have been discovered by way of reasonable testing prior to marketing the product. A manufacturer will be held to the standard of knowledge of an expert in the appropriate field, and will remain subject to a continuing duty to warn (at least purchasers) of risks discovered following the sale of the product at issue. In accordance with the usual rule governing retroactivity in this type of action, the standard just expressed will apply to all claims on which a final judgment has not been entered, or as to which an appeal is pending or the appeal period has not expired, and to all claims on which an action is commenced after the release of this opinion. See McCarthy v. Litton Indus., Inc., 410 Mass. 15, 25-26 (1991); Payton v. Abbott Labs, 386 Mass. 540, 568-570 (1982).19
4. The judge made detailed findings of fact to support her rulings that G. L. c. 93A had been violated. The judge’s findings are not clearly erroneous, her legal conclusions are sound, and, having found no error in the trial, we reject the defendants’ argument that the G. L. c. 93A decision cannot stand. The judge could properly decline to assess additional compensatory damages on the G. L. c. 93A claim because such an assessment would duplicate the jury’s award of damages. See Calimlim v. Foreign Car Ctr., Inc., 392 Mass. 228, 235 (1984). The fact that Mr. Vassallo was not included in the demand letter sent under G. L. c. 93A, § 9, caused the defendants no prejudice because the award of attorney’s fees and costs would, in any event, be attributable in its entirety to" Mrs. Vassallo’s G. L. c. 93A claim.
5. The judgments are affirmed. The plaintiffs may apply to a single justice of this court for an award of appellate attorney’s fees and costs pursuant to G. L. c. 93A, § 9 (4), and Yorke Mgt. v. Castro, 406 Mass. 17, 19-20 (1989).
So ordered.