Merrell Dow Pharmaceuticals, Inc. v. Havner

MERRELL PHARMACEUTICALS, DOW

INC., Petitioner, Marilyn

Ernest HAVNER and Havner on Kelly

Behalf of their minor child

HAVNER, Respondents.

No. 95-1036.

Supreme Court of Texas.

Argued March 1996. July

Decided 1997. Overruling

Order Rehearing Nov. 1997. well, It in other countries as

1983. wаs sold but was called Debendox the British Com- monwealth,, Ireland, Leno- and Australia and Germany. Mari- tan The Bendectin West lyn components: ingested Havner had two succinate, doxylamine is an antihista- mine, pyridoxine hydrochloride, which is and Prior to Bendectin had vitamin B-6. dicylomine hy- component, contained a third drochloride, antieholergenic. which is an thirty Approximately women took million three-ingre- two- or Bendectin either the dient form. twenty years ago, questions

More than possible were raised about Bendectin and its The FDA in- association with birth defects. Hill, Austin, Miller, concerns, vestigated John L. Russell W. but failed to con- Dallas, Bridges, Essig, James E. Kamela clude that Bendectin increased the'risk Houston, Dickson, Robert L. Hall R. Mar- thirty birth defects. More than studies on ston, Monica, CA, George Berry, E. Santa Bendectin and birth defects have been con- Williams, Beaumont, Wiley, Gene M. Rob L. published peer-reviewed scien- ducted and Goode, Austin, for Petitioner. Steven journals questions tific since and medical were first raised. None of these studies Allison, Grillo, Corpus Guy H. Kevin W. women who took concludes that children of Christi, Nace, DC, Barry Washington, J. in- during pregnancy had an Bendectin Christi, Hilliard, Corpus Roberrt C. Rebecca creased risk of limb reduction birth defects. Rockwall, Flood, Hamilton, E. T. Cor- John affirmatively conclude Some of these studies Christi, pus Respondents. that there is no association between Bendec- tin and birth defects and that Bendectin is a OWEN, Justice, delivered the approval FDA of Ben- drug. Although safe PHILLIPS, in which Court Chief revoked, dectin has never been Merrell Dow Justice, GONZALEZ, HECHT, drug withdrew the from the market ABBOTT, Justices, CORNYN, ENOCH year Kelly a little over a after Havner was join. born. there The issue this case is whether any drug evidence that the Bendectin caused The Havners’ suit is based on theories of Kelly Havner to be born with a birth defect. design, negligence, defective and defective legally hold that the evidence offered is

We marketing. brought It is one of thousands causation. Accord- insufficient to establish predecessors against Dow and its Merrell ingly, judgment we reverse the of the court distribution of Bendec- the manufacture and appeals. 907 S.W.2d 535. litigation, virtually tin. all the Bendectin the scientific relia-

the central issue has been I to es- bility expert testimony offered challenged tablish causation. Merrell Dow Kelly Havner was born with a limb reduc- the Havners’ causation evidence several fingers right tion birth defect. The her junctures proceedings. in these It filed Kelly’s hand were not formed. mother had summary judgment, contending motion for prescription drug taken the Bendectin scientifically evi- there is no reliable during pregnancy her to reheve nausea causes limb reduction dence Bendectin morning symptoms associated with Kelly or that it caused Havner’s birth defects formulated sickness. Bendectin was motion, denying the birth defect. Before predecessors its and mar- Merrell Dow and hearing at which the scien- trial court held a in the from 1957 to keted United States reliability summary granted damages. tifie of the Havners’ Id. at 564. We Merrell judgment extensively application evidence was aired. Dow’s for writ of error. trial, challenges legal Dow sufficien- reliability Merrell

Just before the scientific cy causation evidence and of the Havners’ again the Havners’ evidence was raised admissibility sought Dow in motions in of some of that evidence Merrell limine testimony process rights contends that its due to exclude the of certain of the further experts its Havners’ and other causation evi- under the United States Constitution and ’ rights requested dence. One of these motions due course under the Texas Constitu- disposition until tion were denied. Because of our about causation be excluded case, prima we reach the no evidence facie case had been established that this statistically point significant there was a elevated of error.

risk that a child with limb would born reduction birth defects the child’s mother II ingested sought Bendectin. motion Another expert All the witnesses on causation have preclude the Hаvners’ witnesses from re- appeared in in which other cases Bendectin lying on in vitro and in vivo animal studies. limb was claimed to have caused reduction sought entirely Other motions to exclude birth defects. The Sixth Circuit commented testimony of three of the Havners’ causation on a that the Bendectin suits are “variations briefed, fully witnesses. The issues were theme, an orchestra which somewhat like lengthy hearing, and after a the trial court halls, substituting music travels different denied each of the motions. playing musicians from time to time but es- jury bifurcated trial ensued. sentially repertoire.” Turpin the same v. liability phase, the Havners called five ex- Pharms., Inc., Merrell Dow 959 F.2d perts on question. (6th Cir.1992). the causation Merrell 1351 objected some, Dow to the admission of but extensively The federal courts have dealt all, of this evidence. Merrell Dow also date, litigation. plain- with Bendectin To no unsuccessfully moved for a directed verdict prevailed ultimately tiff has in federal court. on the issue of causation at the close of the The evidence those cases has been similar Havners’ As evidence. can be seen from the by to that offered the Havners. The federal record, question reliability of scientific decisions of the have discussed the substance repeatedly. was raised detail, testimony evidence in and often the At liability phase, the conclusion of the Palmer, scrutiny that of Drs. under included jury Newman, Glasser, found in favor Gross, Swan, of the Havners and the Hav- punitive awarded million. In the dam- $3.75 ners’ These are not witnesses. decisions million, ages stage, jury Court, provide awarded but binding they $30 on our but do that amount was reduced the trial court to extensive consideration scientific relia- pursuant million to former Tex. bility Crv. $15 of the causation evidence. §

PRAC. & Rem.Code 41.007. Merrell Dow Some federal courts have concluded appealed. expert legally evidence of causation is panel appeals of the court of insufficient. See Elkins v. Richardson-Mer (6th rell, Inc., Cir.1993); оriginally Turpin, heard case reversed and ren- 8 F.3d 1068 1349; judgment that dered the Havners take noth- 959 F.2d Brock v. Merrell Dow Pharms., (5th Inc., Cir.), ing, holding that the evidence of causation 874 F.2d 307 modi (5th Cir.1989); legally reh’g, insufficient. 907 at 548. 166 S.W.2d fied Richardson-Merrell, Inc., panel concluded that Havners Richardson v. 857 “[t]he (D.C.Cir.1988); bring anything have failed to 823 LeBlanc v. Merrell forward (E.D.La. Pharms., Inc., suspicion F.Supp. on the essential element of Dow 932 782 Inc., Pharms., banc, 1996); rehearing causation.” Id. On en Hull v. Merrell Dow (S.D.Fla.1988); v. disagreed. F.Supp. divided court It affirmed the trial Monahan Labs., 83-3108-WD, No. damages, court’s award of actual but re- Merrell-National (D.Mass. Dec.18, 1987). punitive versed and rendered the award 1987WL 90269 However, recently trial court en- expert have found the a state Other federal courts against jury on a verdict Raynor judgment tered to be inadmissible. See v. Pharms., Inc., finding included a of fraud. Merrell Dow that 104 F.3d 1371 Merrell opinion, highly the court was (D.C.Cir.1997); In a written v. Merrell Dow Daubert Cir.) (on (9th Pharms., Inc., critical of the evidence offered Merrell 43 F.3d 1311 — Dow, ample denied, -, concluding that there was evi- remand), cert. U.S. misrepresenta- (1995); Merrell Dow had made Ealy dence 133 L.Ed.2d 126 S.Ct. FDA, misrepresenta- including Richardson-Merrell, Inc., tions to the 897 F.2d 1159 on Bendectin. (D.C.Cir.1990); tions about its animal studies Lynch v. Merrell-National Pharm., Inc., (1st Cir.1987); No. 1027 Labs., Blum v. Merrell Dow DeLuca 830 F.2d 1190 1996) (Pa.Ct.C.P. (appeal pending). Pharms., Inc., Dec. F.Supp. v. Merrell Dow (3d (D.N.J.1992), aff'd, Cir. 6 F.3d 778 granted has sum- At least one state court Richardson-Merrell, Inc., 1993); Lee v. Dow on the mary disposition for Merrell (W.D.Tenn.1991), F.Supp. aff'd, New- that the of Drs. basis (6th Cir.1992); v. Mer F.2d 1577 Cadarian man, Palmer, inadmissible. and Swan was F.Supp. rell Dow 83-303467-NM, Navarro, DePyper v. No. (E.D.Mich.1989); Ambrosini v. Richardson- 1995) (Mich.Cir.Ct. Nov.27, 1995 WL 788828 Merrell, Inc., 86-278, No. 1989 WL 298429 testimony inad- (holding plaintiffs’ experts’ (D.D.C. 1989), aff'd, 946 F.2d 1563 June Davis/Frye rule and ren- missible under Richardson-Merrell, (D.C.Cir.1991); Will v. Dow). dering judgment for Merrell (S.D.Ga.1986). Inc., F.Supp. 647 544 found, only we have appellate decision initially circuit court found the One federal federal, upheld that has a verdict state or expert testimony and reversed a admissible plaintiff in a Bendectin case is favor of summary judgment for Dow. DeLu Merrell appeals District of from the court of for the Pharms., Inc., ca v. Merrell Dow 911 F.2d Dow Phar in Oxendine v. Merrell Columbia (3d Cir.1990). 941, However, re (D.C.1986) 952-59 maceuticals, Inc., 506 A.2d 1100 again the trial court once found the mand notwithstanding the (reversing judgment and, entering evidence inadmissible after ex remanding reinstatement of verdict and for findings fact conclusions of tensive determination of compensatory damages and law, granted summary judgment for Merrell However, the subse punitive damages). judg Dow. The Third Circuit affirmed that history that case is somewhat ex quent opinion. DeLuca unpublished ment with an court, trial traordinary. Upon remand to the Pharms., Inc., F.Supp. v. Dow 791 Merrell appeals’ following the court of instead of (3d (D.N.J.1992), aff'd, 6 778 Cir. 1042 directive, granted trial court Merrell 1993). trial and vacated Dow’s motion new ensued, appeal and the judgment. Another have A few federal district courts denied that a instructions case was remanded with Dow on the summary judgment for Merrell the verdict. Oxen judgment be entered on question. raised a fact basis that Pharmaceuticals, Inc., v. Merrell Dow dine Pharms., Inc., Longmore Dow 737 v. Merrell (D.C.1989). 330, 331, Judgment 338 563 A.2d (D.Idaho 1990); In re Bendec F.Supp. 1117 taken, appeal was entered. Yet another was F.Supp. 744 Litig., tin Prods. Liab. 732 for lack of appeal was dismissed but (E.D.Mich.1990); Hagen v. Richardson-Mer punitive dam finality question because the (N.D.Ill.1988); rell, Inc., see F.Supp. 697 334 Merrell Dow to be tried. ages remained Pharms., Inc., v. Merrell Dow also Lanzilotti Oxendine, Pharms., v. 593 A.2d Inc. (E.D.Pa. 10, 82-0183, 1986 July No. WL (D.C.1991). remand, judgment Following verdict). 1986) (denying motion for directed entered, sought relief but Merrell Dow light post-trial devel judgment from the in which Merrell Dow obtained Decisions opments including epidemiological v. jury in its favor include Wilson verdict Pharmaceuticals, completed at the time of trial. 893 F.2d that were Merrell Dow (10th appellate Cir.1990), relied on deci Merrell Dow also and In re Bendectin Cir.1988). (6th appel- first heels of the sions decided on the Litigation,

7H light late decision in Oxendine had concluded be considered most favorable to scientifically that there was no party rehable evi- favor the verdict has been whose rendered, dence of causation in the Bendectin eases. every reasonable inference de The trial court indulged declined set aside ducible from the evidence to be judgment. Seale, Merrell Dow Inc. v. party’s v. favor. Harbin Oxendine, (D.C.1994). (Tex.1970). 649 A.2d A no evidence S.W.2d ensued, appeal appellate (a) fourth and the court point will be sustained when there is a the case fact, remanded to the trial court for a complete absence of evidence of a vital (b) determination whether Merrell Dow could the court is barred rules of law or of newly demonstrate “that the evi- discovered weight giving evidence from to the evi probably produce (c) dence “would fact, a different prove dence offered to a vital ” granted.’ verdict if a new trial were Id. at prove evidence offered to a vital fact is no remand, (d) 832. On extensively scintilla, the trial court more than a mere evi evidence, including reviewed the the testimo- conclusively opposite dence establishes the ny Newman, Swan, or affidavits of Drs. Calvert, the vital fact. Robert W. “No Evi Palmer, Gross, Glasser, granted re- dence” and Evidence” Points “Insufficient verdict, lief from the rendering judgment Error, 38 Tex. L.Rev. Merrell Dow. Oxendine Merrell Dow More than a scintilla of evidence exists when Pharms., Inc., 82-1245, No. WL supporting finding, the evidence as a 1996) (D.C.Super.Ct. (appeal pend- Oct. whole, “‘rises to a level that would enable ing). people reasonable and fair-minded to differ ” Burroughs their conclusions.’ Wellcome Thus, we are not the first court to wrestle (Tex.1995) Crye, Co. v. 907 S.W.2d presented with the issues the Bendectin Moriel, (quoting Transportation Ins. Co. v. litigation. (Tex.1994)). 879 S.W.2d Ill experts Several of the Havners’ testi cases, As in most the Bendectin the fied Bendectin can cause limb reduction central issue before us is not whether the that, birth Dr. defects. Palmer testified to a plaintiffs’ possessed adequate witnesses cre- degree certainty, Kelly reasonable of medical dentials, skills, experience testify about Havner’s birth was defeсt caused the causation. The qualifica- witness whose ingested during preg Bendectin her mother Palmer, tions have challenged been is Dr. held, however, nancy. We have that an ex experience whose in identifying the cause of pert’s opinion bare will not suffice. See Bur questioned birth defects is by Merrell Dow. Wellcome, 499-500; roughs 907 S.W.2d at Longoria, United Blood Servs. v. 938 Cf. Ass’n, Employers’ v. Texas Ins. 612 Schaefer 29, (Tex.1997); S.W.2d 30-31 Broders v. (Tex.1980). 199, S.W.2d 202-04 The sub Heise, (Tex.1996). 148, 924 S.W.2d testimony stance of the must be considered. Indeed, witnesses, the Havners’ causation in- Wellcome, 499-500; Burroughs at S.W.2d Palmer, cluding Dr. in a testified case that Schaefer, 612 S.W.2d at 202. Court, reached the Supreme United States case, Schaefer, compensation In a workers’ and that Court deemed their credentials “im- plaintiff the suffered from atypical tuberculo pressive.” Daubert v. Merrell Dow sis, some strains of which were carried 579, 2, 2786, 509 U.S. 583 & n. 113 S.Ct. expert fowl. An ‍‌​‌‌​​‌​‌‌‌​​​​‌‌‌‌​‌‌‌‌​​‌‌‌‌‌‌‌​​‌‌​‌‌​​​​‌‌‌‌‍testified that based on rea n.& 125 L.Ed.2d 469 The probability, plaintiffs sonable medical the us, previously issue before as in most of the employment disease resulted from his as a cases, cited Bendectin is whether the Hav- plumber in which he was to soil ners’ evidence is scientifically rehable and the contaminated with feces birds. Schae support thus some judgment evidence to Nevertheless, fer, 612 at 202. this S.W.2d in their favor. testimony entirety, Court looked at the in its determining noting accept expert’s opinion whether there is no that to probative support “simply evidence of force to some evidence because he used the jury’s finding, all the magic effectively record evidence must words” would remove the Wellcome,

jurisdiction appellate Burroughs courts to deter- see also at 907 S.W.2d sufficiency mine the legal of the evidence 499-500. any expert case requiring testimony. Id. at Gonzalez, in writing Justice for the 202-05. considering After the record in Court, gave examples rather colorful of unre Schaefer, this Court held that there was no in E.I. liable scientific evidence du Pont de despite evidence of causation because Robinson, & Nemours Co. v. S.W.2d used, “magic testimony language” expert (Tex.1995), he said an when that even proba- was not based on reasonable medical expert degree should able to bility possibility, specu- but instead relied on flat, testify that the moon is world lation, and Id. at surmise. 204-05. cheese, green made of the Earth is recognized that Other courts have likewise system. center of the solar If for some simply expert says it is not so because it “an testimony reason such were in a admitted Co., is so.” Viterbo v. Dow Chem. 826 F.2d objection, reviewing without trial would a (5th Cir.1987). 420, expert When obliged accept court be it evi as some little, “br[ings] to court than his creden- In concluding dence? answer is no. subjective opinion,” tials and a this is not testimony that this scientifically unreliable judgment. evidence that would Id. evidence, however, therefore no a court at 421-22. The Fifth Circuit in af- Viterbo necessarily beyond expert what looks summary judgment firmed a and the exclu- Reliability by looking said. is determined unreliable, expert testimony sion including numerous those forth in factors set holding opinion fundamentally an “[i]f testimony Robinson Daubert. unsupported, expert then it offers no assis expert generally opinion testimony. 422; jury.” tance to Id. also see Whether rises to the level evidence is Ciba-Geigy Corp., Rosen v. evidence, rules of in determined under our (7th Cir.) (“[A]n expert nothing supplies who cluding requires Rule courts supplies nothing but a bottom value to line if testimony determine will assist — judicial denied, process.”), cert. U.S. jury in deciding a fact issue.1 While -, (1996); S.Ct. L.Ed.2d 33 admissibility Rule 702 deals with the of evi Pharms., Inc., Turpin v. Merrell Dow dence, guidelines it offers in de substantive (6th Cir.1992) (holding evi termining is some legally dence insufficient in case Bendeetin probative evidence of value. no understandable basis was scientific

stated). expert’s Similarly, say that the testimo ny is some under our evidence standard argued beyond It could the looking simply expert the review because testified testimony reliability to determine sci- technique underlying methodolo incompatible entific is with оur no supporting opinion gy generally his or her is reviewing If evidence standard of review. accepted by community put the scientific light is to court consider the evidence in the ting the cart before the horse. As we said verdict, argument most favorable to the Robinson, an expert’s bald assurance of va runs, beyond should not court look n lidity enough. is not 923 S.W.2d at 559 expert’s testimony to determine if it is reli- (quoting Daubert v. Merrell Dow argument simplistic. able. But an too such (9th Cir.) (on Inc., 43 F.3d re It reduces no evidence of re- standard mand) (holding expert’s of va assertion meaningless looking view to a exercise of lidity enough; objective, is not there must be appear transcript see what words in the methodology), independent expert’s testimony, in fact validation of of the not whether there is — denied, -, an t. U.S. rejected some We have such evidence. cer (1995)). 205; 189, 133 approach. Schaefer, 612 at See S.W.2d S.Ct. L.Ed.2d skill, knowledge, experience, provides: training, or edu- 1. Rule 702 cation, technical, scientific, testify opinion in the form an specialized If or other knowledge will assist the fact to under- otherwise. trier of stand issue, evidence or to determine a fact Civ. Evid. TexR. 702. by qualified expert a witness as an The view that courts not look be- judgment should remains free to direct a ... and yond grant summary judgment. an averment likewise to expert underlying opinion data his or her are the 595, 113 atU.S. S.Ct. at 2798.

type experts of data on which reasonably The Court cited two Bendectin decisions in rely rejected by has likewise been statement, Turpin, of this 959 F.2d underlying courts. The data should be inde- 1349, and Brock v. Merrell Dow Pharmaceu pendently determining (5th evaluated in ticals, Inc., Cir.), 874 F.2d 307 modified See, (5th Cir.1989). opinion e.g., reh’g, itself is reliable. In re on 884 F.2d 166 In Turpin, the Paoli R.R. Yard Sixth Circuit held that the scien Litig., PCB 35 F.3d evidence, (3d tific light viewed most favor- Cir.1994); Richardson v. Rich plaintiffs, able to the was not sufficient to ardson-Merrell, Inc., 857 F.2d jury probable allow a to find that it was more (D.C.Cir.1988); Agent Orange In re Liab. inju than not that the defendant caused the (E.D.N.Y.1985), Litig., F.Supp. Brock, ry. Turpin, 959 F.2d at 1350. In (2d Cir.1987). 'd, 818 F.2d 187 In aff judgment Fifth Circuit rеversed a on entered Supreme the wake of the Court’s decision in jury verdict because the evidence of causa Daubert, the Third prior Circuit overruled its Brock, legally tion was insufficient. 874 F.2d holding in DeLuca v. Merrell Dow Pharma 315; Raynor at see also v. Merrell Pharms. ceuticals, (3d Inc., 911 F.2d Cir. (D.C.Cir.1997) (af 1990), expert’s that an averment that his or firming judgment notwithstanding the ver testimony her type is based on the of data on noting expert testimony dict and that even if experts reasonably rely generally Daubert, were admissible under it was “un enough to survive a Federal Rule of Evi- likely” jury that a reasonably could find it Paoli, inquiry. dence 703 In re 35 F.3d at causation). sufficient to show 747-48. The persuaded Third Circuit was discussed, already As a number of other Judge opinion Weinstein’s Agent in In re “ decisions in the litigation Bendectin have Orange: ‘If underlying data are so lack- held that the causation legally evidence was ing probative reliability force and that no insufficient, setting jury sometimes aside a reasonable opinion could base an summary verdict in other granting cases them, an entirely upon which rests judgment supra or a directed verdict. See ” them must be excluded.’ (quot- Id. at 748 709. The decision in Richardson-Merrell ing Agent Orange, re F.Supp. said in no uncertain terms that the trial court 1245). If expert’s scientific in granting judgment did err notwith- reliable, it is not evidence. The thresh- “[wjhether standing the verdict because old reliability determination of does not run expert’s opinion adequate has an basis” is an afoul of our no evidence standard of review. “falling province issue within the court.” 857 F.2d at 833. Indeed, the Supreme United States Court many There are decisions outside the Ben- agree would that a determination of scientific litigation dectin that have the relia examined reliability appropriate is in reviewing the le- bility of scientific evidence in a review of the gal sufficiency of evidence. While admissibil- See, sufficiency legal e.g., of the evidence. ity sufficiency rather than was the focus of Corp., Conde v. Chem. 24 Velsicol F.3d Supreme Daubert, the Court’s dеcision (6th Cir.1994) (stating that even if evi explained that Court that when “wholesale Daubert, dence is admissible under it can inappropriate is exclusion” the and legally still be insufficient to withstand sum admitted, sufficiency a review of its is not mary judgment); v. Wade-Greaux Whitehall foreclosed: Labs., Inc., 1441, 1485-86 (D.Vi.) F.Supp. the event the [I]n trial court concludes that (granting summary judgment in toxic tort presented support- the scintilla of evidence case when evidence of causation was insuffi ing position is insufficient to allow a verdict), aff'd, jury cient to sustain a 46 F.3d juror (3d posi- reasonable to Cir.1994); conclude that the see also v. Tele Vadala (1st likely true, Indus., Inc., tion dyne than not is the court 44 F.3d Cir. 1995) underly testimony If (noting expert that even the foundational data admitted, unreliable, ex plane ing opinion testimony crash are about cause of were opinion jury pert permitted to will not be to base an permit would not sufficient to Paoli, favor); any opinion from plaintiffs’ In on that data because drawn find re (“[I]f Further, an that data is unreliable. 750 n. 21 the of evidence likewise scintilla unreliable when expert’s even presented to reasonable is insufficient allow a underlying are sound if the data the juror position that the more like conclude from that based on true, draws conclusions data ly free to than not is the court remains methodology. expert’s A flaw in flawed the summary judgment grant ... [or] direct reasoning from the data render reliance judgment.”); In re Eastern & Joint cf. in 1124, on a and render the unreasonable Litig., F.3d Southern Dist. Asbestos (2d Cir.1995) ferences drawn therefrom dubious. Under (finding evidence of circumstance, expert’s tes the scientific case sufficient legally causation in asbestos and, timony legally, no evidence. unreliable notwith reversing judgment and trial court’s verdict); standing Alpha v. Thera the Gruca next some the difficult We consider Cir.1995) (7th peutic Corp., 51 F.3d proof of in a surrounding issues causation respon (holding trial court abdicated its case this. toxic tort such as admissibility sibility by refusing to rule on by instructing and for the defen verdict IV ease; assuming in a admissi dant blood bank limb The not contend that all Havners do evidence, legally bility of it would be the by Ben- birth are reduction defects caused sufficient). Elec. But see Joiner v. General always limb causes dectin or Bendectin (11th Cir.1996) (Birch, Co., even taken at reduction birth defects J., sufficiency (stating concurring) development. limb Ex- the critical time of scope weight beyond are of evidence Dow perts for Havners Merrell — analysis), cert. U.S. granted, of Daubert are agreed that limb reduction defects some -, 137 L.Ed.2d 325 S.Ct. experts agreed also that the genetic. These large percentage cause of a limb reduction Robinson, we set forth some of undis- birth defects is unknown. Given these looking factors that courts should consider facts, puted plaintiff what must establish beyond expert. the bare fact whether Bendectin raise a issue on Those factors include: birth caused an individual’s defect? (1) theory has the extent this one question of cases like causation tested; can be been or engendered debate. Courts has considerable (2) technique to which relies extent al- have the issue have not addressed upon subjective interpretation have ex- ways agreed, and commentators expert; widely divergent quan- views pressed necessary to quality tum and of evidence (3) theory subjected has been whether recovery. sustain a publication; peer review Sometimes, in toxic tort cases causation (4) error; technique’s potential rate specific general discussed terms of (5) theory underlying or tech- whether See, Raynor Merrell e.g., causation. nique generally accepted as has been (D.C.Cir. Inc., 104 F.3d commu- by the relevant scientific valid Sanders, 1997); Joseph From to Evi Science nity; and Testimony on Causation in dence: (6) been non-judicial uses that have Cases, 1, 14 46 Stan. L.Rev. Bendectin theory technique.' or made of the (1993). a sub General causation is whether inju Robinson, capable causing particular a The issue stance is of 923 S.W.2d at 557. See evidence, general population, or condition in the admissibility ry in of Robinson was is a sub specific factors causation whether explained as the same while but we have injury. particular of caused a individual’s may a no review stance applied evidence experi- cases, controlled scientific In some scientific evidence. if a in Demon- Epidemiological ments can be carried out to determine Use Evidence of Cause-In-Fact, causing particular strating substance of 7 Haev. Envtl. L. capable a condition, Gold, objec- (1983); injury 429, or and there will be Steve Causa- Rev. Proof, which it tive criteria can be determined Stan- tion in Toxic Torts: Burdens of certainty Persuasion, particular with reasonable that a Evi- and Statistical dards of (1986). injury by exposure dence, 376, individual’s was caused to Dr. L.J. Yale However, many a given Havners, toxic Glasser, substance. gave a witness for eases, experimentation tort direct cannot be if study designed given a to see a example done, and will be no there rehable study drug though rashes. a causes Even specific of causation. people drug took may that ten who show rash, appeared a while rashes on exhibited direct, scientifically In the absence reli- people drug, take three who did not causation, proof able at- claimants explained Dr. that cannot Glasser tempt exposure to demonstrate that to the got the rash tell us which ten substance at of then- issue increases the risk things drug. because of the We know that particular injury. The finder of fact is asked drug other than the cause rashes. demonstrably to infer the risk that because greater general population in the to ex- due Recognizing epidemiological that studies substance, posure injury to the the claimant’s the actual of an indi- cannot establish cause was than likely that caused condition, injury ques- difficult vidual’s a a theory substance. that sci- Such concedes plaintiff tion for the courts is how a faced particular ence cannot tell us what caused a a fact with this conundrum can raise issue on plaintiffs injury. policy It based on likely causation and meet the “more determination the incidence proof. Generally, not” more re- burden injury sufficiently disease or elevated due willing recognize cent have decisions been substance, exposure to a who someone epidemiological showing an in- studies exposed to that substance and exhibits the may support recovery. Judge creased risk injury disease or can question raise a fact on Weinstein, Agent whose decision generally causation. See Daubert v. Merrell litigation widely Orange has discussed been Pharms., Inc., Dow n. 13 followed, has observed that courts have — (9th Cir.) (on remand), denied, cert. U.S. “strong” been between divided -, 116 S.Ct. 133 L.Ed.2d 126 preponderance “weak” versions of the rule. rely to a Havners extent on considerable Orange” Litig., “Agent re Prod. Liab. epidemiological proof general (E.D.N.Y.1985) F.Supp. (citing Accordingly, causation. we consider the use Rosenberg, The Connection in David Causal of epidemiological studies and the “more like Exposure A “Public Law” Mass Cases: Vi- ly proof. than not” burden System, sion the Tort 97 HaRV. L.Rev. (1984)). “strong” requires version A plaintiff epidemiological to offer both evi-

Epidemiological existing probability studies examine causation ex- dence that populations attempt percent fifty exposed population to determine there in the ceeds proof is an between a “particularistic” association disease or condi- that the substance suspected causing tion and a “weak” factor harmed the individual. The version See, solely e.g., disease or condition. Bert Black & to be on statisti- allows verdicts based Lilienfeld, E. Epidemiologic Rosenberg, supra, David cal 97 Haev. L. evidence. Proof Litigation, Judge Toxic Tort Rev. at 857-58. Weinstein concluded 52 Fordham L.Rev. (1984). However, Orange plaintiffs Agent witnesses were re- quired Havners and uni- offer causation was commentators this area evidence that formly acknowledge percent probable,” “more than given plaintiffs’ F.Supp. studies cannot establish individu- and that the ex- perts required myriad al out contracted a disease or condition due to to “rule were See, exposure particular drug possible afflic- agent. causes the veterans’ Dore, tions,” e.g., at 1263. Commentary Michael id.

716 any the 1.0 mark have courts have likewise found that whether studies over

Other Id.; requirement probability a more than also v. significance. of 50% see Allen United (D.Utah States, 247, epidemiological that must F.Supp. means evidence 588 418-19 1984) injury that show the risk of or condition rejecting greater than (explicitly double population more than 50% standard of causation connection unexposed popula the risk control evidence), grounds, rev’d on statistical other See, Daubert, 1320 e.g., tion. 43 F.3d at (10th Cir.1987); 816 F.2d 1417 Grassis plaintiff's that (requiring to show Bendeetin 446, N.J.Super. Corp., 248 Johns-Manville ingestion than drug mothers’ of the more 671, (App.Div.1991)(holding A.2d 591 674-76 defects); of birth De doubled likelihood in precluding trial court erred that Pharms., Inc., 911 F.2d Luca v. Merrell Dow epidemiological on studies based (3d Cir.1990) 941, (requiring 958 that Ben- 2.0). less showing relative risks of than plaintiffs risk of limb dectin establish relative risk” in toxic “doubling of the issue arising epidemiologi from reduction defects for the provided ground cases has fertile tort 2.0, which to more equates cal data of at least scholarly plow. who that Those advocate risk); doubling than a of the Hall v. Baxter something doubling the risk is short of a 1387, 1403 Corp., F.Supp. Healthcare liability or adequate support who advocate (D.Or.1996) breast-implant plain (requiring liability proportionate type that some exposure to breast tiffs to demonstrate Farber, imposed be include Daniel A. should their implants more than doubled the risk of Causation, 1219, 71 Minn. L.Rev. Toxic which, injuries, alleged epidemiological Gold, (1987); supra, at 96 Yale L.J. 1237-51 terms, requires than a relative risk of more 395-401; L. Hall & K. Sil Khristine Ellen States, 2.0); F.Supp. Manko v. United Reaрpraising Epidemiology: A Re bergeld, (W.D.Mo.1986) (stating that a rel Dore, Harv. sponse to Mr. L.Rev. Envtl. epidemiological risk of in an ative 2.0 (1983); Rosenberg, supra, 97 445-46 likely than not means the disease 859-60; 2 American see also L.Rev. Harv. event), was caused in relevant affd Responsibility Enterprise Inst., Law (8th Cir.1987); Marder v. part, 830 F.2d 831 Injury (1991) (discussing 369-75 Personal Co., F.Supp. Searle & G.D. propor suggesting tort toxic cases and (D.Md.1986) litigation, that in a (stating IUD compensation to all with the disease tionate showing a preponderance causation attribut or disorder should be based terms, evidence, re epidemiological causation); Ap Kaye, D.H. able fractions of 2.0), aff'd, risk of at quires relative least ples Oranges: Confidence Coefficients (4th Cir.1987); F.2d 655 v. United Cook Persuasion, 73 and the Burden Cornell (N.D.Cal.1982) States, F.Supp. L.Rev. case, (stating that in vaccine when relative 2.0, greater greater ‍‌​‌‌​​‌​‌‌‌​​​​‌‌‌‌​‌‌‌‌​​‌‌‌‌‌‌‌​​‌‌​‌‌​​​​‌‌‌‌‍risk is than there is spectrum Mi- On other end injury 50% that the was caused than chance Dore, epidemiological who asserts chael vaccine). by the alone, cannot, cau- standing establish Dore, Commentary Use on the sation. See contrary reached a con Some courts have Evidence, supra, 7 Epidemiological Harv. clusion, holding 434; Michael D. L. see also Envtl. Rev. doubling showing something less Green, Expert Sufficiency Witnesses finding may support jury’s the risk Litigation: Evidence in Toxic Substances causation. In In re Joint Eastern & South Orange Legacy Agent and Bendeetin Litigation, ern District Asbestos (1992) (2d Litigation, 86 Nw. U.L.Rev. Cir.1995), 1124, 1134 Circuit Second in- (concluding in the of other absence no au the district court cited observed that formation, would be doubling of the risk thority for “bold” assertion that stan verdict, plaintiff’s inadequate mortality ratios of are statisti dardized 1.5 might a lower advocating that risk upon but cally insignificant and cannot be relied elimi- risk factors could be sufficient jury. court that it was by a The circuit held nated); Thompson, Causal Moore jury Melissa preferable far to instruct the on statisti Implications Epidemiology: significance jury let decide cal and to Inferenсe B Litigation, Toxic Tort N.C. L.Rev. (1992) (arguing strong that a associ- recognize that Although we there ation risk than or requires greater ratio legal fit science and precise is not a between 8.0, although equal to moderate association of proof, persuaded that burdens we are coupled 3.0 to suffice if 8.0 could with properly designed epidemiolog and executed factors). may part ical be of the evidence studies tort supporting causation a toxic ease and particularly Some commentators have been relating is a for that there rational basis attempts critical of the courts to meld the that more than a “dou requirement there be probability requirement more than 50% bling of the risk” to our no standard evidence epidemiological the relative risks found in likely and to than not review the more determining studies in if the studies were proof. generally DeLuca v. burden of See admissible or were some would evidence that Inc., Dow Merrell support an award for the claimant. But (3d Cir.1990); Lilienfeld, 958-59 Black & su disagreement epidemiological there is on how 767; pra, see 52 Fordham L.Rev. also studies be used. Some commentators should Cook, Daubert, 1321; F.Supp. 43 F.3d at contend that more than probability 50% at 308.

requirement stringent, is too others while argue epidemiological that have no studies naturally that a Assume condition occurs legal requirement relation to the “more 1,000 people they six when out of even are likely Gold, Compare supra, than not.” drug. not to a certain If studies of (advocating L.J. at 395-97 relaxed Yale people drug who did take show that nine Petitti, proof), threshold of with Diana B. disease, 1,000 out contracted it is still on Epidemiology, Guide Juri Reference likely than other than more not that causes (1996) (finding J. 167-68 no metrics drug responsible any given were oc- for support in of epidemiology textbooks or from it currence the disease since occurs in six empirical proposition for the 1,000 anyway. out of individuals Six of thе agent risk attributable exceeds 50% an statistically nine attrib- incidences would be likely is more than to be the cause of the drug, utable to causes other than the disease), plaintiffs Thompson, supra, 71 therefore, it probable is not more N.C. L.Rev. (asserting 264-65 use any drug caused one of disease. incidence satisfy statistical association to a more This would amount to evidence that the likely than not is “misguided”). standard See drug could have caused the disease. Howev- al., also Carl F. Cranor et Bound Judicial er, 1,000 than if more twelve out of who take ary Drawing and the Need Context- disease, drug contract the then it be Sensitive Science Toxic Torts Dau after statistically likely given more that a than not Pharmaceuticals, bert v. Merrell Dow drug. disease was individual’s caused (1996) (arguing Va. Envtl. L.J. 37—40 that epidemiological oversimplification evidence should This is statistical causation, simply relating general excluded because it relative reveals a as we 2.0, below, risk less than unless thinking there is no other but it discuss illustrates evidence); Kaye, supporting supra, 73 doubling requirement. behind the of the risk Cor- vein, (arguing viewpoint L.Rev. at 69 is falla For another in this same see nell cious to “if Bernstein, reason that the data are more P. Charrow E. Robert & David Legal Washington another, probable hypothesis Foundation, under one Scientific Admissibility hypothesis likely then the former is more in the Evidence Courtroom: Significance latter”); be true than the James Robins & After Dau- Statistical Greenland, (1994), Probability Sander Causa who bert advocate there is mathematically relationship tion Under a Model Stochastic Individu demonstrable (1989) Risk, likely al relative risk more between and the Biometrics (concluding proportional liability They than not standard. contend that a slightly schemes based on risk of than 2.0 will cannot be relative alone). ever, legal rarely, satisfy data causation perspective, miological From a requirement standard. mathematical studies and the of probability general changes of thе causation doubling more than a of a the risk strikes as significance changes. the level of statistical legal sys- balance between the needs of our may Id. at 29-31. A relative risk of 2.2 be tem and limits the of science. to probabili-

sufficient show more than a 50% (5 ty 0.05 at the level chances out of 100 that C chance), by but the result occurred not at hold, however, We do not that a relative (10 100). 0.10 level chances of out With risk of more than 2.0 is a litmus test or a attempt calculations that we do out to set single epidemiological legally is test sufficient here, example these commentators offer an evidence of causation. Other factors must be which a relative risk ratio of 2.75 results noted, already epidemiologi- considered. As probability a general of causation of about only cal studies show association. There 0.05, significance with a statistical but 52% of may relationship in fact be no causal if even probability general about a 43% of cau- high. example, risk is For relative stud- significance sation with a statistical of 0.10. have is Id. ies found that there an association at 31-32. implants between silicone breast and reduced recognize, as We does federal Refer- This rates breast cancer. does not neces- Evidence, ence Manual on that a Scientific sarily implants mean breast caused the disease or condition is or is caused either reduced E. rate breast cancer. See David by exposure suspected agent to a and Bernstein, Admissibility data, of Scientific frequency such the incidence of ad- Evidence Daubert Merrell Dow After general population effects in verse Pharmaceuticals, Inc., 15 Cardozo L.Rev. exposed, actual cannot indicate the cause of (1994) al., (citing H. Berkel et given individual’s disease or condition. See Augmentation: Breast Factor Risk al., Bailey A. Linda et on Guide Reference Eng. Cancer?, Breast 326 New J. Med. Epidemiology, in Federal Center, Judioial (1992)). Likewise, particular study if a even REFERENCE MANUALON EVIDENCE SCIENTIFIC risk, reports may in a low relative there fact But the law must balance the relationship. strong be causal consen- to compensate need those who have been among epidemiologists sus is that injured conclusions by wrongful actions of another drawn, all, if about causation not be concept with should at deeply in our imbedded jurisprudence until number criteria have been consid- defendant cannot be widely by set of injury prepon- found liable for an unless the ered. One criteria used supports epidemiologists published by derance the evidence cause Sir Austin scientifically epide- fact. The use of reliable Bradford Hill 1965.2 Another set crite- Temporality. 2.The Hill criteria are summarized as 4. "Which the cart and Bradford is follows: which the horse?" Biological gradient. "Fifthly, 5. if associ- Strength upon my 1. of association. "First biological ation one can reveal a put strength I To list would of association. curve, gradient, dose-response we then very example, comparing take old carefully look for such evi- occupations patients should dence_ most cancer scrotal response occupations patients presenting The clear-dose curve ad- with the diseases, simple explanation obviously mits with other Percival Pott could reach light." puts in a correct the enor- the case clearer conclusion because of chimney Plausibility. helpful 6. would mous increase of scrotal cancer "It be sweeps.” biologically plausible. suspect we causation my Consistency. we 2. "Next on list of features But this is a feature I am convinced cannot specifically place biologically plausible I demand. What de- considered would consistency pends biological knowledge day.” re- of association. Has it been peatedly persons, in dif- 7. inter- observed different Coherence. "The cause-and-effect seriously places, pretation ferent circumstanсes and of our data should not con- times?” generally Specificity. ... 3. "If the association is limit- flict with the known facts of the specific particular history biology sites the disease." ed workers and natural possible types Experiment. "Occasionally no 8. it is of disease and there is association dying, appeal experimental ... evidence.... between the work other modes of clearly strongest support strong argument then that is a in favor Here the for the causation hypothesis be revealed.” of causation.” its, Science, 14,1995, by epidemiologists July studying Similarly, ria used dis- 168. ease the Henle-Koch-Evans Temple, drag Postulates.3 evalua- Robert the director Although epidemiologists do not FDA, consider “[m]y tion has said that basic necessary that all these criteria be met be- if the relative isn’t at least rule is risk three causation, drawing they fore inferences about four, forget point it.” We hasten to Id. part are of ac- methodology generally sound are out that these statements contained in cepted by community. the current scientific popular press, what is more akin to the journals, peer-reviewed scientific and the requires Sound also methodology that altogether of those is not context statements design of epidemiological and execution stud any from of clear. We draw no conclusions be example, ies examined. For can bias foregoing point articles than to other out dramatically reliability affect the scientific why See, there are a number reasons epidemiological study. Bailey an e.g., et al., bright- on a of 2.0 as a Epidemiology, on rebanee relative risk Guide Reference boundary bne would not be accordance Reference Manual on Evi Scientific 138-43; supra, Thompson, supra, dence, at methodology with sound scientific in some 71 N.C. L.Rev. at 259-61. Bias result exploration expbeation can Careful cases. factors, bias, confounding from selection methodology what is rebаble in a scientific Thompson, supra, information bias. 71 N.C. given necessary. context is L.Rev. at 260. We will not an undertake ways extended many discussion D bias cause results of a to be A few courts that have embraced the misleading. We epidemiologi note standard have in- more-than-double-the-risk subject cal studies “are to many biases and instances, epide- dicated in dicta that in some present problems therefore formidable in de miological studies with relative risks of less sign prob greater execution and even might than if there 2.0 suffice were other interpretation.” lems Marcia Angell, See, Daubert, e.g., evidence causation. Interpretation Studies, Epidemiologic Hall, 16; 1321 n. 947 F.Supp. F.3d at Eng. J. New Med. 1398, 1404. We decide in need not this case We also note that some of the literature epidemiological whether evidence with a rela- epidemiologists

indicates that a consider rela 2.0, coupled less tive risk with other tive risk of less than three to a weak indicate evidence, may legally and rebable credible be Thompson, supra, association. See 71 N.C. empha- sufficient to We causation. (citing Wynder, L.Rev. at Ernest L. size, however, of causation evidence Epidemiology Guidelines to the As Weak scientifically from whatever must be source sociations, 16 Preventive Med. hoc, speculative testimony rebable. Post will (1987)). The executive editor of the New not suffice. England Medicine, Journal An- Marcia gelí, physician, treating physician, has A even a general stated a “[a]s rale of thumb, looking we who has a are for a risk of seen skewed data relative sample, three or such accepting paper as one of a few infants who [before defect, publication], particularly biologically position if it has a in a birth is not implausible finding.” or if community it’s brand-new infer causation. The scientific Gary Taubes, Epidemiology accept methodologically Lim- Faces Its would not sound 3. See, Lilienfeld, Analogy. e.g., supra, 9. “In some Blаck circumstances & 52 Fordham judge by analogy. 762-63; Callahan, would be fair to theWith Christopher L.Rev. L. Es effects of thalidomide and rubella us we before Litigation, Causation in Toxic Tort tablishment of surely ready accept slighter would be but Dore, (1991); 23 Ariz LJ. Michael A St. drug similar evidence with another viral disease in or another Evaluating Proposed Epi Standard For Use of pregnancy." demiological Evidence in Toxic Tort and other Bernstein, supra, 15 L.Rev. at 2167-68 Cardozo Cases, Injury Personal 28 How. LJ. Hill, (quoting Austin Bradford The Environment ah, (1985); Bailey see also et Guide on Reference Causation?, and Disease: Association or Proc. Epidemiology, in Reference Manual Scientific (1965)); Soc'y Royal see also Med. supra, at 160-64. Evidence,

Thompson, supra, 71 N.C. L.Rev. at 268-74. doubling expert reporting that that fail indicate more than “study” such an ingestion particular drug by the risk. Similarly, mother caused the defect. birth E expert’s physical assertion that a exami- be

nation confirmed causation should not raise issue and thus To a fact on causation accepted at face value. O’Conner review, sufficiency a claimant legal to survive Co., 13 Commonwealth Edison simply into must do more than introduce (7th Cir.1994), treating physician testified that show a knew cata- he what radiation-induced substantially must elevated risk. claimant clinically they racts like because are looked in the show that he or she similar those and definable and “cannot proof deseribable studies. This include would anything else.” at 1106. person mistaken for Id. injured same sub Nevertheless, exposure stance, his exposure that the or dose levels were plaintiffs radiation caused the cataracts was greater in the comparable to or than those studies, had no found to be inadmissible because it exposure before the occurred the literature on which injury, timing scientific basis. The and the of the the onset of expert support relied did not his assertion injury expe with that onset of was consistent cataracts be di generally rаdiation-induced could study. rienced those in the See at agnosed by visual examination. Id. Thompson, supra, 71 L.Rev. 286-88. N.C. at good Further, evils 1106-07. For a discussion of the plausible are causes if there Bernstein, nature, of this see negat “evidence” could injury of the or condition that be Fur supra, ed, plaintiff 15 Cardozo L.Rev. 2148-49. the must offer evidence exclud ther, VI(A), expert certainty. discuss in Part an ing as we with those causes reasonable choosing study, picking a cannot dissect and & generally E.I. du Pont de Nemours See data, (Tex. “reanalyze” Robinson, a the data to derive Co. v. 923 S.W.2d process 1995) risk not higher relative if this does the the (finding that failure of methodology. comport scientific damage with sound rendered rule out other causes speculation); opinion little than his more promulgated regulations The FDA has Co., Liab. Ins. Employers Parker v. Mut. in requirements for clinical that detail the (Tex.1969) a (holding that S.W.2d vestigations safety of the and effectiveness “probable” only when “in the cause becomes drugs. § These 21 C.F.R. 314.126 explana causal of other reasonable absence case re regulations “[i]solated state that likely more than not that tions becomes experience, lack ports, reports and random result”). injury was a ing permit evalua scientific details sum, emphasize that courts must In we Id. tion not considered.” will be reliability from all make a determination 314.126(e). reject § should likewise Courts party, allow a the evidence. Courts should scientifically not because it is such evidence defendant, present the best plaintiff or out, points physicians reliable. As Bernstein evidence, passes muster assuming it available methodology, not following scientific would Robinson, a then should under patients in un patient or examine a several totality of from the evi- court determine settings whether controlled to determine dence, affecting all considering factors effects, nor particular drug has favorable studies, reliability particular whether reports they rely on case to determine would is sufficient evidence to legally there Bern whether a substance is harmful. See judgment. 2148-49; stein, supra, 15 Cardozo L.Rev. is Finally, cognizant that science are Rosenberg, supra, 97 Harv. L.Rev. we see also theo- constantly reevaluating conclusions and particular (arguing at 870 that anecdotal time, only scientific and that over accomplishes no than ries ized evidence more methodology in a knowledge scientific knowl but appearance false actual direct causal, may have strived particular evolve. We relationship). Expert testi field edge of a in holdings scientifically cannot to make our observations' mony reliable that is current, accepted generally scientific light of up epidemiological to shore used However, 2.67, methodology. person should not is are that a courts ratio odds exposed drug likely is times possibility in to the 2.67 foreclose advances sci- study. develop the disease under require ence reevaluation what is “good science” in future cases. person risk” Similarly, “relative that a drug develop particular who took a will

V study. in disease can be determined a cohort conducting are Id. at 176. The risk is calculat- Certain conventions used relative studies, by comparing ed the incidence of disease scientific and statistics are used to population reliability exposed the incidence evaluate scientific endeavors population. If the disease in the control and to determine what the results tell us. 1.0, exposed indi- relative risk the risk opinion, this we consider of the basic some unexposed viduals the same as concepts currently individuals. used scientific studies 1.0, If greater the relative risk is than analyses and statistical those con how exposed greater risk in cepts legal sufficiency individuals than mesh with our stan exposed. those not If the relative risk is less dard of review. For discussion extended 1.0, risk in methodology individuals epi statistical and its use studies, exposed. less than in those not For demiological see DeLuca Merrell risk, Pharmaceuticals, doubling result to Dow indicate (3d Cir.1990). greater relative must See Turpin 945-48 See risk be 2.0. also Pharms., Inc., id. at Dow 147-48. Merrell (6th Cir.1992); al., 1353 n. Bailey et Refer Perhaps is the the most useful measure Epidemiology, ence on Guide in Reference risk, proportion attributable of which is the supra, at Evidence, on Manual Scientific relationship statistical measure of a factor’s 138-43, attempt 171-78. do not We dis population. to a represents disease in the It cuss the aspects all multifaceted of the scien “proportion among the ex of the disease statistics, tific method and but focus on the posed with the individuals is associated

principles light particular shed on the words, exposure.” Id. at 149. it facts and issues this case. percentage inju reflects the the of disease or ry prevented by eliminating be could

A exposure the For de substance. a more way study populations by One a is retro of use tailed discussion the calculation and of spective case-comparison epi case-control or risk, proportion the of id. at attributable see demiological study. For example, type 149-50; Lilienfeld, this supra, & Black 52 Ford study of identifies individuals with disease Thompson, 760-61. L.Rev. at See also ham group and a suitable people control of with supra, 71 N.C. L.Rev. at 252-56. out disease the and then looks back to exam The numeric value of an odds ratio at postulated ine of the causes disease. See risk, equal least to the relative but odds the al., Bailey Epidemiol et Guide on Reference risk, espe ratio often overstates the relative ogy, in Reference Manual on Scientific cially if the event is not the occurrence of 136-38, supra, at Evidence, 172. Another example rare. For an of the difference be type study cohort tween mathematical calculation study, study, prospec or incidence which is risk, odds ratio and thе see Barbara relative study groups tive that identifies and observes Page, Hazard & Batten Munro Ellis Statis them group over time to see one is more tical for Health Care Methods Research 134-36, likely develop Id. disease. at 173. (2d 1993). example given In the ed. 3.91, by Page,

An “odds ratio” be can calculated Munro and the odds ratio was study. example, at case-control Id. 175. For while the risk was 3.0 based on relative al., Bailey odds ratio could be used to the odds of data. also et show same set See drag ingestion Epidemiology, of a is associated with a on Guide Ref Reference particular Evidence, compares The disease. odds ratio erence Manual on Scientific 149; having exposed supra, supra, Thompson, odds the disease when at 71 N.C. drag exposed. to the versus when not If L.Rev. at n. 22. may expressed alge- important

The relative risk because is the basis of statis- braically tical as: test. Id.

—RR study deciding may A contain error in R Ic hypothesis, reject accept and this error ‍‌​‌‌​​‌​‌‌‌​​​​‌‌‌‌​‌‌‌‌​​‌‌‌‌‌‌‌​​‌‌​‌‌​​​​‌‌‌‌‍risk, RR where is the relative is the inci- Ie Id.; types. one can be of two Moore & exposed popula- in the dence the disease supra, Type A I error 482-87. McCabe, tion, and is the incidence of disease in the Ie hypothesis when the null is true but occurs sample population. control A calculation is Type rejected, has been and a II error oc follows: hypothesis null but has curs when the false (cid:127) exposed of the incidence disease accepted. supra, at been 55. Page, Munro & (Ie) per- per is 30 cases individuals example types given An two error sons, or 0.3 comparison Page is a of two Munro and (cid:127) in the unex- incidence disease taught groups people who have been sta- (Ic) per posed individuals is 10 cases Group A tistics different methods. Id. persons, or 0.1 significantly Group B a higher than scored (cid:127) the relative risk is the incidence in the The knowledge test of their statistics. (0.3) exposed group by the inci- divided hypothesis is that thеre is no difference null (0.1), unexposed group in the dence methods, teaching because between but equals 3.0 difference, was a indicated there hypothesis rejected. Suppose, null was Using hypothetical, this can we conclude however, Group composed people who are are three times ability that in people higher math likely to contract disease than those actuality teaching not matter method did necessarily. who are not? Not The result rejection hypothesis null all. any given study comparison not be Type I Id. error. representative population. of the entire may by The result have occurred chance. making a I probability Type error discipline has means of statistics determined by changing level can be the decreased telling significant how results us the is, that significance, probability that the the may study be. by chance. Id. If the level results occurred in significance had been five one hundred (0.05), only in there is a five one hundred

B by that result chance chance the occurred step understanding signifi- The first in If one in significance alone. level of is the testing is to how research cance understand (0.01), in is a one one one hundred there hy- is often A researcher tests conducted. result occurred chance that the hundred potheses by testing whether the and does so However, significance as the chance alone. particular hypothesis. data (e.g., from 0.05 stringent is level made starting point hypothesis, is the null 0.01), to find a to it be more difficult will that there is no difference or no assumes signifi- significant Altering Id. result. studying If you effect. were the effects also increases the cance level in this manner Bendectin, hypothesis example, the null for error, accepting a Type risk of a II which is be it has no effect. The research- would Type To II hypothesis. null Id. avoid false hypothe against er tries find evidence errors, be low- significance can level George sis. See David S. Moore & P. ered, example, tеn one hundred for McCabe, to the Practice Introduction (0.1). Id. (2d 1993); ed. Page, & Statistics Munro may ap- be significance at The statement the re Different levels supra, 54. types of studies de- may propriate be true is stated as different suspects searcher willing to on much risk one is hypothesis. significant pending how the alternative If found, wrong. is accept null conclusion reached hypothesis is difference is examples by Munro Again, to offered rejected. significant is not take If a difference particular found, test for a hypothesis accepted. Page, assume that a the null Mun if the defect is genetic exists and that supra, concept defect at 54. This Page, ro & diagnosed early stage, repeti- at an a child (cid:127)with the risk between 1.3 and 3.8 95% the However, successfully if can be If the tions. the interval includes the defect treated. time, 1.0, study statistically sig- is not genetic diagnosed defect is not number or, way, is development severely child’s im- nificant said another inconclusive. -will be in- This is because the confidence interval paired. mistakenly diagnosed If a as child treated, risk values are having cludes relative both less the defect and there are no greater hypothesis than than the null agree harmful Most effects. would (1.0), leaving the researcher with results that preferable Type would be to make a I error suggest hypothesis both that the null should Type rather than a II error under these rejected. accepted be and that it should be Type Id. A circumstances. II error would See, 1; at n. e.g., Turpin, 959 F.2d failing diagnose be a child that had the Brock v. Merrell Dow genetic defect. (5th Cir.), reh’g, on modified hypothetical Contrast with one (5th Cir.1989); al., Bailey 884 F.2d 166 et which a federal is conducted to deter- Epidemiology, on Guide Ref Reference particular mine whether a method of teach- Evidence, erence on Manual Scientific ing underprivileged children increases their supra, concept explained at 173. This imple- success in school. Id. The cost of Glasser, jury in this сase Dr. one of menting teaching this in a nationwide method Thus, study may the Havners’ witnesses. program very great. A Type would be I 2.3, produce meaning risk of relative error would be to that the program conclude data, greater risk is 2.3 times based had an effect when it did not. Id. The 95%, but a confidence level of the confi- project significance level for this prob- would dence interval has boundaries of and 3.2. 0.8 ably higher be one used to screen insignificant The results are at the therefore genetic hypothetical. defects the other willing 95% level. If the researcher genetic example, In the prefera- defects it is accept greater risk of error and lowers the they ble to if may treat children even 90%, confidence level to the results disease, have but in teaching method statistically significant at that lower level example, preferable it is not to teach children range num because does include the considerable cost it has no effect. al., generally ber 1.0. See et Bailey Refer epidemio- confidence level can be used in Epidemiology, ence Guide on in Reference logical studies to establish the boundaries of Evidence, supra, at Manual on Scientific the relative risk. These boundaries are 151-55. in “[T]he narrower the confidence known as the terval, confidence interval. at greater See id. the confidence in the rela 59-63; Kaye see also David H. & A. study.” David tive in risk estimate found Id. at Freedman, Statistics, Guide on in 173. Reference

Reference Manual Evi- Scientific C 376-77, supra, 396; dence, at Moore & McCabe, supra, at 432-37. The confidence accepted generally significance level given study interval tells us if the of results a epidemiological confidence level studies statistically significant are particular 95%, at a meaning study that if the were re times, See peated confidence level. su numerous the confidence inter McCabe, Moore & pra, at A 432-33. interval range confidеnce shows val would indicate the of risk relative “range a of values within which the of results values that would result 95% the time. study sample Pharms., Inc., a likely would be to fall if the See DeLuca v. Merrell Dow study repeated F.Supp. (D.N.J.1992), were times.” Bai aff'd, numerous 791 (3d al., ley Cir.1993); al., et Epidemiology, Guide on Bailey F.3d 778 et Reference Refer in Reference Manual On Epidemiology, ence Guide on in Reference Evi Scientific If, supra, at 173. on a supra, dence, Evidence, based confi Manual on Scientific at 95%, study 153; Dore, Standard, Proposed dence level of supra showed relative A note 693; risk 2.3 and had at Thompson, supra, confidence interval 28 How. L.J. 3.8, that, we say Virtually publish 1.3 would if the N.C. 256. L.Rev. at all the ed, repeated, produce peer-reviewed were it would a relative studies on Bendectin have Although though appear there to be an increased confidence level at least 95%. witnesses, Swan, condition, activity of the Havners’ Dr. with an one risk associated of 90% relationship advocated the use confidence level this does not mean the is causal. (10 error), chance of other of she and original panel appeals, court of As the Havners’ conceded that 95% is witnesses ease, there observed this is demonstrable accepted generally level. association between summertime death drowning, not cause but summertime does witnesses, Another of the Havners’ Dr. drowning. at n. 8. 907 S.W.2d Glasser, any explained appli- that in scientific cation, kept very confidence interval many There are other factors to consider that'you high. He testified “don’t ever see evaluating reliability of a scientific 50% or in a [confidence 60%] intervals of to, certainly study including, but not limited study because that means we’re scientific study, power sample size of the going to miss a lot of and [scientists] times variables, study, confounding and whether willing to take that risk.” One com- are there was selection bias. These factors are level mentator advocates that сonfidence appeal, not central to a resolution of this admissibility epidemiological acknowledge that do no more than deter- we generally accepted higher should be than the mining many can reliability scientific have Dore, A and should See Pro- 95% be 99%. facets. Standard, posed supra note How. L.J. DeLuca, But 693-95. cf. VI (discussing statistics Kenneth Roth- principles of the basic Armed some predominate that the choice of a man’s view employed by community in con- the scientific arbitrarily 95% confidence level selected studies, ducting we turn an examination discipline); Longmore convention of his against case measured the evidence this F.Supp. Merrell Dow de See E.I. du Pont the Robinson factors. (D.Idaho 1990) (concluding Robinson, Nemours & v. 923 S.W.2d Co. determining that the scientific standard for (Tex.1995). upon by The evidence relied causation is much stricter than standard catego- experts Havners’ falls into four employed by the court and that confidence (1) studies; (2) in vivo ries: 95%, 90%, levels of or even 80% should (3) studies; studies; animal in vitro animal required). (4) analysis doxy- a chemical structure depart think it unwise to from the We succinate, compo- lamine the antihistamine methodology present generally that is at ac We consider each nent Bendectin. cepted among epidemiologists. general See turn. Black, ly The Supreme Bert Court’s View of Science: Has Daubert Exorcised the Cer Demon?, tainty CardozO L.Rev. “ Glasser, Dr. an associate J. Howard (1994) (stating all ‘[a]lmost that University of School professor at the Texas thoughtful agree scientists ... that [a would at Texas Medical Center of Public Health percent] is significance of five a reason level ” Houston, a epidemiologist in with Ph.D. is an general (quoting Amicus able standard’ Cu of experimental and a Master statistics Brief Alvan R. riae of Professor Feinstein gave jury He of Bio-Statistics. Science Respondent Daubert Support earlier, As of statistics. noted an overview Inc., 579, 113 Pharms., Merrell Dow 509 U.S. explained that statistiсs are used deter he (1993) (No. 2786, 125 L.Ed.2d 469 92- S.Ct. significant there is a association be mine if 102))). Accordingly, should not widen the we occurrences, but cau tween two events or acknowledge at which courts will boundaries a statistical association not tioned beyond statistically a significant association thing causation. same or lower values. the 95% level 90% epidemio- Glasser identified a number of that even a statisti-

It must reiterated found, from he concluded that significant logical studies cally association is as- likely is an equate was more than not there causation. Al- sociation does not exposure highest between and birth with the relative risk association Bendeetin de- fects, though coming exposure period one to even authors those stud- an three lunar months is counted.” The memo ies did not find such association. One evidence, itself was not introduced into and study done and a rela- was Cordero had confidence level at there is no evidence tive risk 1.18 and a confidence interval of found or of However, which the relative risk of 2.13was to 2.13. 0.65 the relative risk in- 2.0, interval. The confidence the confidence would need exceed and confidence may or 1.0. terval have contained 1.0, interval could not include for the results doubling to indicate more than a of the risk Finally, published testified about Glasser statistically significant association be- studies on Bendeetin that did show statisti- tween Bendeetin and limb reduction birth results, cally they significant but dealt with V; supra defects. See Part see also Daubert birth defects other than limb reduction de- v. Merrell Dow sup- These studies cannot course fects. (9th Cir.) (on remand) (noting that more port finding that Bendeetin causes limb likely requires, than not standard terms of Further, reduction later studies of defects. proof, doubling statistical a more than of the types these of birth defects did — risk), denied, -, cert U.S. 116 S.Ct. bear out an association with Bendeetin. 133 L.Ed.2d 126 None The other witness for the Havners other studies identified Glasser showed a epidemiological who testified about doubling of the risk. study The MeCredie Dr. She Shanna Swan. has doctorate had a risk 1.1 relative and a confidence Reproduc- in statistics and is the Chief interval of 0.8 to 1.5. The data in the Eskan- Epidemiological Program tive for the state of zi that considered limb reduction birth epidemiology California. She also teaches 4.18, defects resulted relative risk of but University Berkeley. of California at 36.3, the confidence interval was 0.48 to pub- Dr. Swan conceded none very large interval included 1.0. Dr. lished studies found an asso- agreed Glasser that results with a confidence ciation reduc- between Bendeetin and limb interval that included 1.0 or a lower number tion defects. She identified a number of statistically would be inconclusive insig- these confirmed that the confi- studies and nificant. in each 1.0. dence intervals of them included did, however, Dr. reanalzye Glasser some However, Dr. Swan testified about these data, data, called the Jiek that had been length studies at some criticized report included in a FDA. to the Glasser Then, methodology. relying same on these isolated information on women who had filled studies, opined she that Bendeetin more or prescriptions two more of Bendeetin and probably than not is associated with limb who spermicide, were not which reduction defects. Swan considered birth resulted in a relative risk of 13.0 of limb findings aggregate of these studies in the However, reduction birth defects. the confi- along testified that fall results Further, dence level he used was 90%. there “weight curve which the the curve” was is no regarding other evidence Yet, the direction of increased risk. the confidence interval. The confidence in- she also said these studies were consis- may may terval not have contained 1.0. tent with a risk that 0.7 relative was between doubling and 1.8. That is not a of the risk. point The Havners also to a memorandum may It her is more prepared within the FDA that was identified probable than an not there is associa- by Dr. Glasser. document indicates that tion Bendeetin and limb reduction between relative risk of the limb defects when Ben- defects, magnitude but the association the given dectin is within the lunar first three gleaned is she from these studies not more only pregnancy months of is 2.13. The con- 2.0, testimony. on her based own by Dr. clusion drawn Glasser from this mem- that, conjunction performed reanalysis is in Dr. orandum taken Swan also discussed, reanaly- the other articles he had there is data from at least two studies. One “importance “importance unpublished underlying an of time” and an sis was of raw data (D.D.C.1986), study F.Supp. 802 n. 10

the reduction birth de- Jiek limb (D.C.Cir.1988). fects, which Dr. aff'd, the same data about Glasser 857 F.2d 823 a relative testified. Dr. Swan derived risk shortcomings the In addition to statistical exposed to estimate of 2.2 for women Ben- evidence, epidemiological an- of the Havners’ during deetin the first trimester. She also reliability is that it against other strike its risk for who testified that the relative women published has or otherwise sub- never been exposed exposed were to Bendectin but not review, jected peer exception of with the to finally, that spermicide was 8.8 and abstract, acknowledges Dr. Swan’s which she women who were to two or more published paper. not equivalent is the considered, prescriptions Bendectin were papers published Dr. Swan has number of spermicide, regard exposure to without journals, study including in scientific risk was with a the relative confidence associated with concluded Bendectin is not not reveal interval from to 53. She did the has Although cardiac birth she been defects. obtaining in these re- confidence level used testifying in reduction birth Bendectin limb sults, no confi- there is the years, has many cases for Dr. Swan defect in record. dence level the attempted publish opinions her never reanalysis by Dr. was of The other Swan and limb conclusions about Bendectin reduc- study, in data the Cordero which was based Similarly, by Dr. tion defects. Glas- by the for on information collected Center journals, in published ser have been refereed in Atlanta. An abstract shе Disease Control publications 32 to mentions but none his prepared regarding published this data was or limb birth defects. Bendectin reduction Society Epidemio- Journal the the for discussed, thirty already are over As there logical or 1984 and states Research peer-reviewed published, study original Cordero found relationship Bendec- studies on the between ratio for limb reduction birth defects odds findings concluded, however, None of the tin and birth defects. that when 1.2. Swan experts in this selected, by Havners’ five group the rela- offered different control studied, repli- published, ab- case have been tive risk estimates are affected. Swan’s that, community. assump- by the relevant scientific stract stated “under certain cated identified, said, tions,” only “the Judge not “the odds As Kozinski has review which are high- plaintiffs’ experts’ are “a work has has ratio limb reduction defects” received ly explanation only is no significant” judges juries, place 2.8. There and the been publish- in Dr. the abstract or Swan’s their and studies have been theories significance obtain the 2.8 report- level used to state pages ed is federal and statistically Daubert, well be (commenting result. The result ers.” at a 95% confidence level. We inconclusive five called same witnesses Havners). simply not know from this record. With- do factor that should be related knowing significance or the out level study pre- considered whether the interval, scientifically there no confidence litigation. Has the been pared result is saying reliable basis for that the 2.8 upon the courtroom? used or outside relied Further, anything. her indication of recognized in methodology the scienti- Is group could have choice of control community? litigation spawned Has the fic Although results. her abstract skewed the “community” part that is its own group identify what control she does opinions community? The purely scientific used, trial that she chose Swan testified at testified Havners’ witnesses Syndrome of Downs babies. Swan’s births confines outside the have never beеn offered reanalysis Syndrome babies as using Downs a courthouse. Lynch group was considered the control peer review Richardson-Merrell, Publication and other courts and those reliability of scien significant indicia of Lynch v. it insufficient. See likewise found testimony is Labs., expert’s tific evidence when Merrell-National (D.C.Cir. publica (1st Cir.1987), peer review or in an area which aff'd, 823 857 F.2d Richardson-Merrell, Publication 1988); not be uncommon. tion would Richardson v.

727 reputable, journals established scientific ation between Bendectin and limb reduction peer other forms of review legally “increases not sufficient evi defects would be likelihood that substantive flaws methodol- The dence of causation. Havners’ witnesses ogy will v. be detected.” Daubert Merrell that a number of conceded studies have Pharms., Inc., 579, 593, done, Dow 509 U.S. good practice been not be would 2786, (1993). S.Ct. L.Ed.2d pick a support out one to conclusion. As the legal suggested One commentator has federal Manual on Evi Reference Scientific integrity expert the ultimate test of the of an out, points dence researchers are “[m]ost witness in the scientific arena is “her readi assessing conservative when it comes Daubert, publish ness to and be damned.” relationships, calling strong causal often for (quoting 43 F.3d at 1318 Peter W. Huber, er evidence and more research before a con Revenge: al., Galileo’s Junk Science Bailey clusion causation is drawn.” et (1991)). Further, “the exam Courtroom Epidemiology, Guide on in Ref- Reference study by ination of scientific cadre Evidence, erence on Scientific Manual lawyers is not the same as its examination supra, example, at 157. For Dr. ex- Swan others in the trained field of or medi science plained initially, some studies showed a Richardson-Merrell, cine.” Richardson statistically significant association between Inc., (D.C.Cir.1988) 857 F.2d 831 n. 55 pyloric Bendectin birth defect steno- States, (quoting Perry v. United However, ‍‌​‌‌​​‌​‌‌‌​​​​‌‌‌‌​‌‌‌‌​​‌‌‌‌‌‌‌​​‌‌​‌‌​​​​‌‌‌‌‍sis. subsequent, larger much stud- (11th Cir.1985)). association, ies did not bear out that and in fact, published Swan herself has studies that We publication pre do not hold is a failed find association between Bendec- requisite for reliability every scientific case, tin type and this birth defect. but courts must “especially skepti cal” of scientific that has not been Accordingly, methodology scientific published subjected to peer review. followed, single study would be viewed Inc., Brock v. Merrell Dow 874 as indicating probable that it is “more (5th Cir.), reh’g, as modified See, not” e.g., that an association exists. (5th Cir.1989); 884 F.2d 166 see also Bert Richardson-Merrell, Richardson v. al., Black et Science and the Law in the (D.D.C.1986) F.Supp. n. (noting Wake Daubert: New Search Scienti single study that no would be sufficient to Knowledge, 72 Tex. L.Rev. fic implicate exonerate or to cer Bendectin with peer Publication and review allow an tainty and that studies become “conclusive” opportunity for the relevant com scientific aggregate), aff'd, in the 857 F.2d 823 munity to comment on findings and conclu (D.C.Cir.1988). In affirming the district attempt sions replicate and to reported Richardsonr-Merrell, court in the District of using results populations different and differ recognized plain Columbia Circuit that the study designs. ent tiffs’ had recalculated statistically signifi data and had obtained replication need re Richardson, cant result. See F.2d at acknowledged by sults was wit Havners’ Moreover, 831. The court nevertheless held this was nesses. it must be borne mind evidence that would a verdict. discipline epidemiology associations, Id. should Courts not embrace inferences per not “causation” se. Particu good science would not But larly where, here, draw. experimentation direct cf. Lynch, 830 F.2d at 1194 (asserting that a conducted, has not been it is important that study coming new ato different conclusion any conclusions about causation be reached challenging consensus would be ad only after an association is observed in stud missible). among groups ies different and that the asso

ciation continues to hold when the effects of argument is sometimes made Sеe, other variables are into taken account. waiting until an association found in one e.g., supra, McCabe, at 202. & Moore study is confirmed others will mean that observed, already

As early recovery. we have an isolated will claimants be denied See, Green, finding statistically significant supra, e.g., associ- 86 Nw. U.L.Rev. at 680-81; Wendy limb birth defects in hu- Wagner, E. Trans-Science causes reduction (1986). Torts, 428, Yale L.J. presented mans. This Dr. 428-29 evidence was in veterinary Gross, history a veterinarian and a argument is that tells us that Adrian related from pathologist who had worked at the FDA community been slow at scientific has as the accept results. 1964 to served Chief times to valid research and its true, Toxicology Branch at the Environmental history are While these observations from and Agency also and research Protection tells us that-valid rehable accepted quickly was a Senior Science Advisor generally and are thereafter theories community EPA. Dr. Gross confirmed that within the scientific when suffi- provided empirical animal in explanation is and FDA and EPA consider studies cient al., potential response to supra, assessing human adequate. are See Black et data Galileo, drugs pesticides. He that what 72 Tex. Rev. at 779-82 L. testified (discussing drift). Pasteur, DNA, likely hu- an animal is to affect continental will affect way only in the same and that mans liability argued Others have that predict if reason animal studies are done is to only allocated on the basis of should drug at will have an adverse effect issue proof legal rehable of fault because rules on humans. goals spreading, of “risk should have the Dr. a number of animal Gross reviewed deterrence, cheapest ahocating costs to the on Bendec- studies that had been conducted cost-avoider, encouraging sociahy fa “ described on rabbits tin. He activities,” and because ‘consumers vored he mal- in which saw “a lot of to Bendectin justice people compensat American want ” kits.” another formed Gross testifiеd about Dreyfuss, Cooper ed.’ Rochelle Is Science statistically study he that found of rabbits Special Admissibility Case? The Scien opined significant. probability that the He Dow, Evidence Daubert v. Merrell tific After in occurred that the malformations this (1995) 73 Tex. L.Rev. (quoting 10,000. respect by chance were six With Feinberg, Litigation in Kenneth R. Civil rabbits, study on he stated to another animal Century: A Panel Discus Twentieth-First was harm- probability drug that the that the (1993)). sion, 59 Brook. L.Rev. 1,000,000. per He less was less than one It some has been contended that cases “[f]or rats, mice monkeys, listed studies compen very creating mean well significant harm- showing “highly deleterious satory mechanism even the absence are con- ful effects as far as birth defects proof of effect” and clear scientific cause and studies, Dr. Based on these animal cerned.” judgments “[d]eferring to scientific was was that Bendectin Gross policy about fault obscures the core humans, means that it teratogenic in by the questions that are addressed laws However, he conceded birth causes defects. applying.” expressly Id. the court is We dosage at which Bendectin levels system reject legal re these views. Our associated with birth defects rats became by a quires prove claimants their eases kilogram per day, milligrams per at 100 keeping preponderance of the evidence. daily equivalent aof which would be the proposition this at the heart of sound weighing dosage for a woman of 1200 tablets jurisprudence, law not be our should pounds. liability scientifically hasty impose Judge As evidence is unavailable. reliable briefing before their Havners assert said, science; lags it does Posner has “[l]aw technique for accepted this Court Corp., Cibar-Geigy not lead it.” Rosen determining teratogen if a substance — denied, (7th Cir.), cert. information, all includ- is to look at humans -, 136 L.Ed.2d 33 U.S. S.Ct. data, data, biologi- animal ing Dr. plausibility, in vitro studies. cal are the relevant that these Swan confirmed B determining terato- of information sources Brent, on Com- Comment genicity. animal See also relied on in vivo stud- The Havners Bendectin,” Update: “Teratogen that Bendectin ments on ies the conclusion

729 TeRatology (1985) (stating process 31:429-30 ed his data had Bendeetin was concluded determining teratogen: if a substance is a harmful. (1) consistent, reproducible findings in hu- The in vivo studies identified in this case (2) studies; man epidemiological develop- support jury’s cannot verdict. model; (3) toxicity ment of an animal embryo related; (4) is consistency that dose and with C basic, concepts recognized embryology of and Dr. Stuart Allen Newman also relied on Thus, development). fetal scientific method- support animal to his studies that ology rely studies, would not on animal teratogen Bendeetin a humans. Dr. alone, standing as conclusive evidence that phys- Newman holds a doctorate in chemical is a teratogen substance in humans. See professor ics and is a York Raynor Inc., New Medical v. Merrell F.3d (D.C.Cir.1997) articles, College. published fifty He has over (noting that only way although opinions test contain whether data nonhu- none or conclu- from man can extrapolated studies sions which he testified in this to humans case. would be to experiments conduct human The studies Newman reviewed were in data); epidemiological to use v. Rich Elkins studies, vitro which are on con- based tests ardson-Merrell, (6th petri ducted on cells a test dish. tube Cir.1993) (holding expert opinion that indicat Doxylamine placed directly succinate was on ing of basis animal studies limb including bud cells of animals chick- simply admissible inadequate permit but is development ens and cartilage mice. The of jury prob conclude that Bendeetin acknowledged was affected. Newman that in ably defects); than not Lynch, causes limb studies, these who had con- researchers 830 F.2d at 1194 that (asserting in vivo and doxylamine ducted them that concluded in vitro animal singly or in combina potentially succinate capable inducing of tion do not the capability proving have genetic damage that it and should be tested causation in beings human in the absence of on systems. But Newman testified any confirming data); see you prevails find an effect that across Brock, also 874 F.2d at (recognizing species, “you number of different can be animal very studies are of limited usefulness awfully thing prevail sure that the same will questions toxicity); confronted with in humans.” Pennsylvania

Allen v. Eng’g Corp., 102 F.3d 194, 197 (5th Cir.1996) (quoting following opined Kelly Newman Havner’s case). Brock tort toxic portions defect was due to loss certainty skeleton that could scientific respect We further note that with to the in teratogen a have been caused that affect testified, vivo studies about which Dr. Gross embryo. Similarly, ed the he testified that reliability their predictors of the effect of findings study, of one Bendeetin in Hassell/Hori- questionable humans is because gan Study, to him doxylamine indicated dosage levels. Dr. Gross offered no ehondrogenesis, succinate can interfere with explanation very of how the high dosages process which is the turning certain cells extrapolated could be to humans. Other cartilage. into We note that to the rejected courts have animal studies that re effect that a substance or “can” “could” cause high lied on dosage levels as evidence of See, disease or disorder not evidence that in e.g., Turpin causation in humans. v. See, Pharms., Inc., probability e.g., reasonable does. Par Merrell Dow 959 F.2d 1349 (6th Cir.1992) Co., Employers ker v. Mut. Liab. Ins. (reasoning that to eliminate (Tex.1969); drugs embryos S.W.2d Bowles Bour high dosage toxic levels don, many would eliminate 148 Tex. drugs most S.W.2d useful testified, however, society depends chemicals on which modem Newman that based on Wilson, heavily) (citing Hassell/Horigan James and other Current Status animal stud Teratology ies, Teratology, degree Handbook he concluded with reasonable (1977)). explain why certainty doxylamine Gross also failed to medical succinate published studies from extract- teratogen cartilage development which he is a between the 32nd doxylamine teratogen is a Bendectin somewhere succinate re- that he had day gestation, depending humans. He also testified how 42nd surrounding Marilyn calculated, viewed the records conception which was date to reasonable pregnancy Havner’s and that development of for period within the the any certainty, terato- she was not arm. Palmer ex- Kelly Havner’s hand and gen other than Bendectin. doxy- plained that the molecular structure succinate, components of the two cell lamine one The in vitro studies are similar the Pennsylva- Bendectin, placenta in v. biology permits data at issue Allen it to cross the fact Engineering, nia at 198. The body the fetus. from the and reach mother’s may that Bendectin have an adverse effect and on in vitro animal Based on this fact beginning, not the on limb bud cells is “the studies, studies, epidemio- intact animal inquiry proves noth- end of the scientific information, doxyla- logical he concluded ing without other scientific about causation in humans. teratogen mine succinate is Id.; Richardson, 857 see also evidence.” and on Relying on this information same (“Positive results from in vitro F.2d at 830 Havner, concerning Kelly includ- information provide signaling the need clue studies Bendectin, ingested ing the date her mother research, provide but alone not for further do Dr. to a reasonable Palmer concluded satisfactory opining about causa- basis certainty, Bendectin degree of medical al., context.”); Bailey in et tion the human Kelly Hav- in the birth defect seen caused Epidemiology, in Ref- on Guide Reference hand. ner’s Evidence, Manual on erence Scientific problem supra, (noting that at 130-31 However, testimony is Dr. Palmer’s based extrapolating with in vitro studies is just that conclude on findings “from in laboratories tissues he relied on opposite. To the extent that beings”). whole human Newman, Glasser, Swan, opinions of Drs. opinions Gross, scientifically Logical support for Dr. Newman’s reliable or there no is substances, lacking. A was also number of opinions, as we their evidence to G, have shown to dam- such as vitamin been no other seen. Palmer identified have directly placed tis- age animal cells support his body knowledge that would explanation no Dr. offered sue. Newman chemical structure opinion, other than the leap logical from the how he made the study done on doxylamine succinate and to his conclu- vitro studies on animal tissue antihistamines, The Sixth not Bendectin. birth defects sion that Bendectin causes of Dr. Palmer’s captured the essence Circuit testimony not humans. Dr. Newman’s said, testimony “no understandable when it of causation. evidence opinion, basis is stated. Personal scientific Turpin, science, testifying not here.” D further observed That court F.2d at 1360. testified on Of the five witnesses who overstated Dr. Palmer’s conclusions so causation, only who question of witness legitimately predicate it could not their Kelly opined the cause of that Bendectin was jury verdict. Id. We for a form basis defect, opposed to birth Havner’s birth based on agree with that observation John Davis Palm- general, was Dr. defects record this case. doctor er. Dr. Palmer is a licensed medical and holds a doctorate in pharmacology. He ^[*] v ^{[*] [*] [*] [*]} University of Arizona professor is a scientifically reliable There no acting head of College of Medicine and in this case. Accord- support the verdict opinion His Pharmacology Department. its the court judgment ingly, we reverse testimony of the part on the was based judgment part and render appeals Havners’ other witnesses. Merrell Dow. there is critical Dr. Palmer testified that gestation of a period during when the limbs J., BAKER, sitting. Marilyn took not forming. Havner fetus are GONZALEZ, Justice, concurring. techniques, ‘novel’scientific we do read requirements apply Rule 702 join judgment. I the Court’s I specifically exclusively to unconvention- separately guide write to reiterate that the Daubert, 509 al evidence.” U.S. 593 n. lines we in E.I. Pont Nem established du de Supreme 113 S.Ct. at 2796 n. 11. The Robinson, ours & Co. 923 S.W.2d 549 Rules, Court noted that “under the (Tex.1995), аre not limited to testimo judge any all trial must ensure ny theory. based on a novel scientific scientific or evidence admitted Robinson, we held that Texas Rule of relevant, but reliable.” Id. at requires Evidence proponent of sci- added). (emphasis at 2795 S.Ct. We testimony expert entific to the show that having likewise see no value in a different testimony is both relevant and Rob- reliable. admissibility standard of for scienti- novel inson, so, doing 923 556. In S.W.2d at we problems presented fic evidence. followed the lead of the United Su- States determining particular whether or not a preme Court and the Texas Court of Crimi- type of evidence would be considered “nov- Appeals nal adopted and of non-exclu list reject el” daunting enough applica- are to determining sive factors for such whether Moreover, tion of a dual standard. we testimony is admissible.1 See id. at 554-57 observe that the factors and criteria set (citing Daubert v. Merrell Dow Pharmaceu Kelly upon forth bearing as reliabili- the ticals, Inc., 2786, 509 U.S. 113 S.Ct. 125 ty proffered scientific evidence are ade- (1993); State, L.Ed.2d Kelly 469 v. 824 quate assuring measure for “novel” Here, (Tex.Crim.App.1992)). S.W.2d 568 the “junk scientific evidence which is science” applies Court the Robinson criteria to Mer- is excluded. These factors the “address legal rell sufficiency Dow’s challenge and underlying soundness of the scientific theo- concludes that the Havners’ expert testimony State, ry technique.” v. and Jordan 928 is no causation. 953 S.W.2d 706. (Tex.Crim.App.1996).... S.W.2d I agree with this approach. But I con- am State, (Tex. v. Hartman S.W.2d cerned litigants may that some misread Rob- Crim.App.1997). analysis applies equal This apply only inson to to novel scientific evi- ly before, to As I Robinson. have said we my writings dence applying because later “provide intended Robinson to the “junk R.V., exclusive it to science” cases. See S.V. v. evaluating reliability (Tex.1996) standard the of ex (Gonzalez, J., 933 S.W.2d pert anything about characterized concurring); Burroughs Wellcome v. Co. R.V., as (Tex.1995) science.” at 42 Crye, (Gonza- S.V. S.W.2d 907 S.W.2d (Gonzalez, J., rеhearing). lez, J., concurring on We concurring). not free grasp did intend to from Robinson’s Recently, the Court of Appeals Criminal might what be considered routine science. addressed a similar attack Kelly, on equivalent court’s In rejecting Robinson. attempted prove The Havners causation argument, this the court stated: primarily through expert testimony based on in Kelly Nowhere did we limit two- and animal studies. These pronged standard to by By ap- novel scientific evi- are no foundations means novel. Supreme dence. The plying [United States] to Merrell the Robinson factors Dow’s directly challenge, Court Daubert implicitly addressed no-evidence Court footnote, stating issue “[although applies holds that Robinson to scientific ex- Frye exclusively pert testimony decision itself focused across the board. The trial (5) 1. These underlying theory technique factors are: whether the (1) theory generally accepted by extent which the has been or has been as valid tested; be can community; relevant scientific (6) (2) technique the extent to which the relies non-judicial uses have been upon subjective interpretation ex- of the theory technique. made pert; Robinson, E.I. du de Pont Nemours & Co. v. (3) theory subjected whether the has been (Tex.1995) (citation S.W.2d foot- peer publication; review and/or omitted). note error; (4) technique's potential rate copies the Office only of those briefs to whether evi- forwarded court must determine of the Bar of Texas Robin- of General Counsel State dence relevant and rehable. See Conduct, son, pursuant to Texas Code of Judicial at 556. It need not decide S.W.2d 3(D)(2). the evidence is also novel. Canon Id. whether response assessing appropriate SPECTOR, Justice, concurring. rehearing now been that has the motion today to heed its own The Court fails Hilliard and his co-counsel this filed warning that of a scientific “the examination Court, of our courts of agree we with another study by lawyers is a cadre of same necessаry recently appeals found who its examination others trained as integrity court: on the of that address attacks or medicine.” 953 field science S.W.2d must drawn between distinction omitted). (internal agree I citations respectful advocacy judicial denigra- testimony is witness the Havners’ Although former is entitled to a tion. legally sufficient evidence causation. voice, can be con- protected the latter scientist, However, judge, not a I public’s confi- expense doned at the majority’s uncomfortable with the ambi- am judicial process. Even were in the dence unnecessarily analysis its tious scientific and personal to willing this court tolerate expansive application of the stan- Daubert [counsel], obligated we are insult levied majority’s opinion, replete with dard. The respect and maintain the due this Court to dicta, guidance gives practical no out- courts legal system we an oath to serve. took Ac- litigation. the context of Bendectin side cordingly, judgment I in the concur (Tex. Maloney, re S.W.2d the Court. (en banc) writ) App.—San Antonio no Johnson, curiam); (per also Johnson v. see Reheaeing FOR On Motion (Tex.App.—San An 948 S.W.2d ORDER (sanctioning requested)1 writ tonio rehearing for filed on behalf of The motion disparaging remarks about counsel for However, the ten- the Havners overruled. ap forwarding and the court trial court requires that we address or of that motion Coun peals’ opinion to the Office of General Respondents’ conduct of counsel. sel, concluding question substantial honesty, been about counsel’s had raised This not the first time this ease that trustworthiness, lawyer). fitness as or in less engaged counsel have the Havners’ Following the exemplary conduct. deci power inherent to disci- possess Courts panel of court of original of the sion pline attorney’s “‘Courts behavior. judgment appeals, which had reversed to justice universally acknowledged be are judgment trial court and rendered creation, vested, powеr to very their nothing, take Robert C. Hilliard the Havners decorum, silence, in their impose respect, and appeals court of filed two briefs with the ” NASCO, presence.’ Chambers v. court, sitting banc, found to be which that en 32, 43, 2123, 115 L.Ed.2d 27 U.S. S.Ct. unprofessional.” disrespectful, “insulting, (further (1991) observing court federal Pharmaceuticals, v. Inc. Hav Merrell Dow to its power bar to control admission has ner, (Tex.App.—Coipus 907 S.W.2d attorneys appear before who discipline banc) curiam). 1994) (en (per The Christi (6 Dunn, it) 19 U.S. v. (quoting Anderson further appeals court of concluded Wheat.) 204, (1821)); see also Public 227. a violation of Texas “evidence[d] briefs 121, 124 Cofer, S.W.2d v. Comm’n Util. Disciplinary Conduct Rules Professional Johnson, (Tex.1988); at 840-41. 948 S.W.2d question as raises a substantial the con- trustworthiness, Disciplinary governing Rules or fit The lawyer’s honesty, lawyer provide: of a accordingly duct appeals ness.” Id. court appeal. pending error merits of application An for writ of 1. Court, express and we no this Austin, lawyer City A respect day should demonstrate Done at the this 13th system November, legal those who serve 1997. it, including judges, lawyers public lawyer’s duty, officials. While it is a BAKER, J., sitting. necessary, challenge the rectitude action, lawyer’s duty

of official it is also a uphold legal process. pream- DISCIPLINARY R. PROf’l Conduct

Tex. ¶ 4, reprinted ble tit. Tex. Gov’t Code, (Vernon app. Supp.1997) subtit. G (Tex. 9). X, § art. R. BAR State 8.02(a) Disciplinary spe-

Rule Rules cifically MINNESOTA states: MINING AND MANUFAC- COMPANY, Petitioner, TURING

A lawyer shall not make statement lawyer that the knows to false with or disregard reckless falsity as to its truth or LTD., Corporation, NISHIKA Lentec qualifications concerning integrity or Ltd., American 3D and Nishika Manu- adjudicatory ‍‌​‌‌​​‌​‌‌‌​​​​‌‌‌‌​‌‌‌‌​​‌‌‌‌‌‌‌​​‌‌​‌‌​​​​‌‌‌‌‍judge, public or legal official (H.K.) facturing Ltd., Respondents. officer, of a candidate for election or appointment judicial legal office. No. 94-1124. 8.02(a). Id. Rule Supreme Court Texas. Legislature provided has also a mech- anism for courts to sanction counsel file who Argued March 1996. pleadings presented for an improper purpose 2, 1997. Decided Oct. or to harass. Tex. Civ. & PRAC. Rem.Code addition, §§ 10.001—10.005. In As one of the Corrected Oct. 1997. Nace,

lawyers Havners, Barry for the is a

non-resident attorney. appearance His subject

Texas courts the Rules Govern- Bar,

ing Admission to the including Rule

XIX. specific portions “Respondents’ Rehearing”

Motion for filed Court this particular

that raise concerns are the “State- 1-5), Rehearing”

ment (pages the Case for 8,14, Argument” (pages

the “Brief of the

16), “Prayer (pages Relief’ 19-

20). for Respondents Counsel Hil- Robert C. Muñoz, Barry

liard of the firm of Hilliard & Paulson, Nace, firm

J. Nace of the Nor- Sellinger,

&wind Rebecca E. Hamilton White, Hamilton, P.C.,

of the firm of White & hereby opportunity

are afforded the to re-

spond why as to should not Court

1) refer appropriate each them to the authorities;

disciplinary

2) prohibit attorney practicing Nace from courts;

in Texas

3) impose penalties monetary as sanctions.

Any response must filed this Court p.m., Monday, 5:00 November 1997.

Case Details

Case Name: Merrell Dow Pharmaceuticals, Inc. v. Havner

Court Name: Texas Supreme Court

Date Published: Nov 13, 1997

Citation: 953 S.W.2d 706

Docket Number: 95-1036

Court Abbreviation: Tex.