Corps de l’article

Introduction

It is no secret to anyone familiar with tort law that product liability has become an extensive, globally recognized area of private law.[1] The same can be said of the (in)famous development risk defence (DRD),[2] which precludes product liability whenever “the inherent risk of a defective product [is] undiscoverable at the time of supply by a manufacturer.”[3] Fuelled by the desire to enhance the “long-term social good”[4], the DRD is said to foster innovation by shielding industries from liability stemming from defective products born of research and development (R&D) in cases where the risk was not discoverable in light of accessible scientific knowledge at the time the product was put onto the market.[5] Conceptually distinct from the state-of-the-art defence[6] (which speaks the language of negligence),[7] the extension of such immunity from liability to producers amounts to a state in which victims of this type of defective product are, at best, dependent on other legal categories and causes of action in order to obtain compensation (contractual warranty, vice caché, negligence, faute, culpa, dangerous activity, etc.) or, at worst, left to themselves in a situation perhaps reminiscent of the pre-industrial caveat emptor paradigm fiercely criticized by PL scholars.[8] Despised and glorified, adopted and rejected, transplanted, transposed, and altered, much has been said—mostly in Europe[9] but also in Quebec[10]—about this immunity from liability that polarizes commentators along a left-right divide.

Beyond the civil law and common law divide, the exemption clause crystallizes the link between some conceptions of justice and our contemporary economy, as it suggests that too much liability chills innovation. Its features have been extensively documented, to the point that the aim of the present contribution may appear methodologically modest, as it merely seeks to provide a comparative overview of recent developments regarding the DRD, driven by expanding case law on the issue.[11] In which legal systems is the DRD to be found? A classical, horizontal comparative exercise (beyond mainstream EU/USA corresponding endeavours) allows us to map properly the territories it occupies after testifying of its projection, through vertical comparison, at the supranational level (Part I). Despite widespread adoption of the DRD, interpretations vary respecting its elements, including knowledge and its counterpart, uncertainty, which are at the very core of the defence (Part II). A comparative legal analysis provides the ability to highlight methodological and ethical concerns and ambiguity present in the DRD, while raising the delicate practice of melding law and science within the courtroom, which has long stimulated debate among evidence and causation experts.[12] Ultimately, indeed, the DRD’s boundaries are set by the judiciary, although “the courtroom is not the place for scientific guesswork, even of the inspired sort,” as Justice Posner once warned.[13]

I. The Territories of the Development Risk Defence

As product liability spread to many of the industrial world’s legal systems, the DRD, too, marked both national and international levels. Transported by circulating policy reasons and, possibly, path dependencies, the model spread rapidly beginning in the 1960s. Intertwining legal roots make it difficult to determine whether it originated in national laws (Section B) or supranational endeavours (Section A). The following section is, therefore, merely intended to offer greater clarity.

A. Supranational Moulds and Pathways

The 1960s were characterized by the spread of product liability maxi-models throughout the industrialized world, including those established in the Vienna Convention on Civil Liability for Nuclear Damage (1963);[14] the European Convention on Products Liability in regard to Personal Injury and Death (1977), adopted under the auspices of the European Council;[15] and the later, famous European Union PL Directive. In light of the general principles of monism and dualism that structure the law of treaties, some models are mandatory, while others are persuasive guidelines that emerge from academic or technocratic experiences (e.g., the Principles of European Tort Law (PETL)[16] or the Uniform Law Conference of Canada[17]); finally, some are of a hybrid nature, such as the American Restatement of Torts (Second)[18] and (Third)[19], as they may, through time, penetrate states’ legislation or case law. This last model has largely forged a field of law of its own.

In 1965, section 402A, entitled “Special Liability of Seller of Product for Physical Harm to User or Consumer,” gave birth to this new field known as product liability. Under this section, any professional vendor who sells a product in a defective condition unreasonably dangerous to its user or consumer is held liable even if he has “exercised all possible care in the preparation and sale of his product.”[20] This rule is also applied in civil law jurisdictions, mutatis mutandis, through either an extension of the contractual sphere to third parties, special consumer laws, or both (whereas common law product liability stems from an extension of the duty of care in tort law, applying Donoghue v. Stevenson[21]). The general rule in section 402A was tempered by an exception, which was added to the Restatement (Second) as a reporter’s comment—comment k—that explicitly addressed dangerous products.[22] This famous comment recognized that the market might include “some products which, in the present state of human knowledge, are quite incapable of being made safe for their intended and ordinary use.”[23] American Law Institute reporters were then undoubtedly particularly concerned with drugs and vaccines, as 1965 also marked the infamous thalidomide scandal.[24] After laying the foundation for proper risk-benefit analysis, the comment goes on to state that “[s]uch a product, properly prepared, and accompanied by proper directions and warning, is not defective, nor is it unreasonably dangerous.”[25] In this context, the comment argued that because “there [was] no assurance of safety,” economic actors should not be held liable for “unfortunate consequences.”[26] The model has been widely adopted in US case law and remains good law in many American states. The economic wind changed from sociodemocratic ideas to neoliberal ones, however, and the time came to update the rule, which had been designed, quite ironically, the same year John F. Kennedy asserted that “[c]onsumers, by definition, include us all.”[27] In 1997, a new version, although criticized by some commentators, was proposed by ALI members. Section 2, comment d, of the Restatement (Third) provides instead that

[t]he term “state of the art” has been variously defined to mean that the product design conforms to industry custom, that it reflects the safest and most advanced technology developed and in commercial use, or that it reflects technology at the cutting edge of scientific knowledge.[28]

Prima facie, this new rule bears little resemblance to its predecessor. Not only does it apply exclusively to certain products (drugs and medical devices), it also crystallizes the learned intermediary doctrine[29] while it states that

[a] prescription drug or medical device is not reasonably safe due to defective design if the foreseeable risks of harm posed by the drug or medical device are sufficiently great in relation to its foreseeable therapeutic benefits that reasonable health-care providers, knowing of such foreseeable risks and therapeutic benefits, would not prescribe the drug or medical device for any class of patients.[30]

Simply stated, neoliberal conceptions of the economy suggest that such a drug shall not be prescribed by any doctor (and should not, to begin with, ever be on the market).

The European Economic Community had earlier adopted the PL Directive, which defined product defect based on consumer expectations. It provides that a “product is defective when it does not provide the safety which a person is entitled to expect.”[31] It draws partly on the American experience, and article 7(e) states that a producer will not be liable if “the state of scientific and technical knowledge at the time when he put the product into circulation was not such as to enable the existence of the defect to be discovered.”[32] Commenting on the PL Directive in a European Court of Justice (ECJ) case, Advocate General Tesauro stated that

the Council opted for a system of strict liability which was no longer absolute, but limited, in deference to a principle of the fair apportionment of risk between the injured person and the producer, the latter having to bear only quantifiable risks, but not development risks which are, by their nature, unquantifiable.[33]

Bolstered by the Thatcherism then in vogue, the DRD had success in Brussels and other world capitals as well as academic endeavours.[34] The “culturally neutral” language of economic sciences unsurprisingly bloomed and flourished throughout the field of product liability, justifying just about any of the forms it took, be they closer to negligence-based models or risk-distribution anchored rationales.

Although in both the EU and the US harmonization of the DRD remains optional,[35] a fundamental difference exists between the two legal systems: the rule is merely persuasive authority in the latter but becomes truly normative in the former whenever a member state avails itself of the option to adopt the DRD.[36] The EU-style DRD harmonization process requires further explanation. The integration of the DRD in the PL Directive was so controversial that the European Council was politically forced in 1985 to leave member states a discretionary margin to either adopt or reject certain features of the directive including the DRD. For example, article 15(1)(b) provides that

by way of derogation [each member state may] maintain or provide in [its] legislation that the producer shall be liable even if he proves that the state of scientific and technical knowledge at the time when he put the product into circulation was not such as to enable the existence of a defect to be discovered.[37]

This discretionary freedom does not, however, allow member states to alter the wording of the DRD as they see fit, as the degree of harmonization sought is quite total given the sui generis, federal-inspired structure that now characterizes the EU.[38]

The EU learned its lesson after France subordinated the application of the DRD to a post-market duty of safety (obligation de sécurité).[39] The ECJ came to the conclusion that the wording of this national measure departed from the harmonized provision in article 7(e).[40] To summarize, the ECJ held[41] that the EU Council unanimity procedure that underlies “market-driven directives”[42] erected upon article 100 of the (former) Treaty Establishing the European Economic Community[43] (among them, the PL Directive), is incompatible with a bottom-up, minimum approach, because they seek to eliminate legal divergences that “may distort competition and affect the movement of goods within the common market and entail a differing degree of protection of the consumer against damage caused by a defective product to his health or property.”[44] Rather, the ECJ concluded that “the margin of discretion available to the Member States in order to make provision for product liability is entirely determined by the Directive itself and must be inferred from its wording, purpose and structure.”[45] Hence, it held that the discrepancies in the provisions of French transposition laws were contrary to the spirit of the directive, and although France aimed to increase protection for victims,[46] this nonetheless violated EU law, as establishing a “level playing field”[47] was the rationale underlying the PL Directive.[48] Therefore, EU harmonization law holds that inclusion of the DRD in a national legal panorama is optional, but that whenever a member state chooses to do so, adherence to its spirit (and conditions) must be quite total although interpretative flexibility is a tool explicitly recognized by national courts. Disharmony in the law addressing tainted blood, for example,[49] may well be explained by the divergent approaches of the civil and common law to judicial discourse.[50] Nonetheless, the limits of such flexibility remain to be defined.

B. Of National Clones, Mutants, and Hybrids

In a German case that addressed personal injury caused by an exploding glass bottle (the Mineralwasserflasche case),[51] for example, the Bundesgerichtshof held that the DRD does not apply to manufacturing defects, as

[w]hen the EC Directive on product liability was being fashioned it was agreed that the defence under art. 7 (e) should apply not to manufacturing defects, but only to defects of design and construction ... and the only dangers emanating from a product which the German legislator wished to exempt from the scope of the Product Liability Law were dangers, undetectable even with the exercise of all possible care, arising at the stage of design and construction.[52]

The court observed that the company’s bottle control system was inadequate despite that it may well have been “the best possible machinery.”[53] In fact, the court held that increased quality control by visual inspection could contribute to preventing the harm caused by ausreisser (isolated defective goods in the production line) and therefore insisted that the defect is specific to the individual product.[54] The judgment ultimately ruled in favour of the plaintiffs, anchoring its conclusion on the general principle of defect-based liability.[55]

Other member states availed themselves of the PL Directive’s discretionary window (a compromise clearly designed to satisfy public opinion), thereby depicting dissimilar scenarios in light of the very nature of the involved products. Because of these ad hoc responses to national—sometimes highly localized—consumer product crises (thalidomide, wine, oil, etc.), some products remain outside the DRD’s field of application today, including medicines in Germany;[56] foodstuffs in Spain;[57] or, in France, blood products and body parts.[58] The result is an unappealing, compartmentalized structure within liability law, which has shrunk, expanded, twirled, and been eradicated to the rhythm of political and moral sentiments, often oblivious to a principle that should—in my opinion[59]—always guide compensation law: that of equal treatment among victims (égalité horizontale). Despite this anti-aesthetic configuration, the model has rapidly spread beyond the EU to many other countries, including Australia,[60] Switzerland,[61] and Japan.[62]

Quebec, on the other hand, has adopted a dual product liability architecture rooted in both the Civil Code of Quebec[63] and the Consumer Protection Act.[64] Article 1473 of the 1991 reform of the CCQ mimics the EU’s DRD approach:

The manufacturer, distributor or supplier of a movable property is not liable to reparation for injury caused by a safety defect in the property ... if he proves that, according to the state of knowledge at the time that he manufactured, distributed or supplied the property, the existence of the defect could not have been known, and that he was not neglectful of his duty to provide information when he became aware of the defect.[65]

Much uncertainty remains regarding the application of the defence,[66] and it is still unclear whether it applies to contractual liability.[67] The defence has clearly not been adopted in section 53 of Quebec’s Consumer Protection Act, which allows victims to recover based on strict liability, as “[t]he merchant or the manufacturer shall not plead that he was unaware of the defect or lack of instructions.”[68]

Finally, some states never seriously considered adopting the DRD (e.g., Finland and Luxembourg); others have been unenthusiastic about judge-made product liability schemes to begin with (such as most Canadian common law provinces,[69] which tend to opt for the state-of-the-art standard that is better suited to the catch-all tort of negligence[70]). Moreover and by reason of a constitutional interference (pre-emption[71]), American victims are prevented altogether from initiating actions based on failure to warn (information defect) when it comes to certain products (e.g., medical devices and tobacco products).[72]

II. Knowledge or Uncertainty? The Methodology Maze

As already mentioned, the premise underlying the DRD is surprisingly simple. Overly broad liability chills innovation, threatens to make certain products entirely unavailable, and increases insurance premiums. To avoid these negative effects, the DRD aims to strike an acceptable compromise. It reintroduces the “knowledge variable” that is usually excluded by strict product liability schemes, which focus on product defects, evoking common law dialectics between foreseeable and unforeseeable, avoidable and unavoidable. Accordingly, producers of defective goods are legally excused if insufficient scientific knowledge was available at the time the product was commercialized. Recent product liability scholarship has distilled a two-prong DRD test out of the knowledge element and the discovery variables, which gives rise to a sui generis, EU-style product liability regime, distinct from the American approach.[73] Although helpful, this framework raises four difficult questions that will undoubtedly require judicial clarifications. First (Subsection A), how accessible must the knowledge be? Second (Subsection B), who must it be available to, and what criteria should be used to evaluate it? Third (Subsection C), how should the ethical questions be addressed that arise whenever producers of goods generate knowledge regarding the very products they put on the market? Finally (Subsection D), how should novel or marginal scientific opinions be evaluated?

A. The Domain of Knowledge: Existing Data and Accessibility

Although still scarce, recent comparative case law analysis has shed light on a possibility to avoid liability rooted in “scientific and technical” knowledge, and, by implication, the producer’s capacity to avoid risks. Knowledge entails two conditions: that relevant data indeed exists, and that it be accessible to the producer. The least worrying concern is accessibility, which may now be moot due to the proliferation of online databases and search engines (e.g., open access services, Google Scholar, etc.) and the hegemony of the English language within academia and the scientific community. Accessibility refers to the potential to identify the results of technical and R&D studies. By 1997, a well-known ECJ case[74] had already provided valuable insight as to the reach of the DRD. The ECJ’s case, coupled with Advocate General Tesauro’s opinion, began to establish the parameters of the knowledge element embedded in the DRD.[75] Indeed, the former had set the table for establishing the parameters of the knowledge element of the DRD with the predictable, academic “Manchurian hypothesis”: Does a tiny publication in the Mandarin language constitute accessible knowledge? Probably not. Is “knowledge” coterminous with peer-reviewed articles written in English? Probably not. Scientific and technical knowledge remains, reasonably enough, a factual question that is left to courts’ discretion.

Existing data refers to scientific information surrounding a given risk, founded on sources such as trials or experiments whose results are exposed and discussed in scholarly publications. A wide interpretation of the text of article 7(e) suggests that producers ought to know every scientific result, in every relevant technical or scientific field, that could enable them to discover the defect. This heavy informational burden has been criticized by invoking a hypothetical “lucky” plaintiff who “turn[s] up information from a totally unexpected field”[76] which precludes the defendant from availing himself of the DRD.

At least in Germany, the impossibility of improving a product due to lack of scientific knowledge is no defence to liability when, though unavoidable, a risk is known to the producer. This conclusion was reached by the Bundesgerichtshof in a seminal German case[77] that held that the explosion of a glass bottle containing mineral water does not fall within the scope of the development risk exemption, as such risk is known to producers. The court seemingly shifted its focus away from the defect toward the risk of harm itself. Although there were no technical means available to prevent the (rare) occurrence of this risk, the court stuck to a narrow interpretation of the knowledge element and discarded the arguments the defendant had put forward to exclude himself from liability. These thoughts echoed in London when Justice Burton, apparently seduced by the Germanic argument, held in the context of contaminated blood litigation that

[i]t would, in my judgment, be inconsistent with the purpose of the directive if a producer, in the case of a known risk, continues to supply products simply because, and despite the fact that, he is unable to identify in which if any of his products that defect will occur or recur.[78]

Seemingly, the boundaries of the DRD are interpreted narrowly, thereby confining the scope of an exoneration clause that seems at odds with the risk distribution rationale that underlies product liability.

B. The “Whom” Question

Another of the most critical issues was articulated by Newdick: “[T]o whom ought that information be available before it may be described as ‘discoverable’?”[79] Is knowledge to be evaluated in light of subjective or objective criteria? The “whose knowledge?” question was at the core of Commission v UK.[80] In their transplant of the PL Directive, English legislators adopted a formulation of the supranational instrument that rejected key features of the DRD. In their place, the UK substituted an assessment of scientific knowledge from the standpoint of “a producer of products of the same description as the product in question”[81] (i.e., the producer’s perspective). The UK provision uses the language of negligence: reference to classes of producers suggests reasonableness, objective standards, and other porous terms well known to common lawyers.[82] Despite this clear disparity in the wording of the provisions, the ECJ came to the conclusion that the action introduced by the Commission was premature, given that UK courts might still interpret the PL Directive’s section 7(e) national clone in conformity with EU law. According to Attorney General Tesauro, however, this knowledge standard is “not concerned with the practices and safety standards in use in the industrial sector in which the producer is operating.”[83]

As a result of the above-mentioned criterion, knowledge needs to be apprehended in light of a twin notion: that of discoverability, which in turn leads to the labyrinth of causation.[84]

C. “Enabling”, Discovery, and Business Ethics

Discoverability purports to establish the elusive link between knowledge and proof of a defect—the trail of breadcrumbs—allowing an inference that an improper R&D process caused the defect. While knowledge may be described as a bundle of objective information, discoverability refers to a manufacturer’s intellectual capacity to assemble the pieces of the puzzle to “enable the existence of the defect to be discovered.”[85] The key word is likely the verb “enable”—meaning permit[86]—which, despite lending itself to varying interpretations, clearly introduces a subjective element into the evaluation of the DRD. The question becomes whether, aside from dissenting or isolated opinions (discussed below), the knowledge available was such as to allow, scientifically speaking, the producer to discover the defect. Did he have the benefit of the interdisciplinary knowledge of a scientific, technical team[87] or rather consciously handpick from available information in a teleological fashion? In this context, evaluation of a producer’s behaviour may discover a partial merging of the DRD and the state-of-the-art defence.

Evaluation of the safety of everyday products usually calls for knowledge of technical standards (e.g., ISO, CEN, DIN). Assessing products with significant R&D cycles often entails the difficult task of combining two or more data sources. Airplanes may today be faster, safer, and ever more comfortable due to advances in metal alloys, physics, thermal radiation, electronic devices, meteorological science, etc. Would all this knowledge preclude the DRD, say, in the context of the Air France flight 447 accident over Brazil? Was Airbus capable of anticipating the risk of data distortion due to the presence of ice crystals in its airspeed measurement device, despite the fact that “[t]he obstruction of the Pitot probes by ice crystals during cruise was a phenomenon that was known but misunderstood by the aviation community at the time of the accident”?[88] In the case of experimental products—excepting GMOs, drugs, and medical devices—there might be little, contradicting, or no knowledge at all. Often, producers generate their own knowledge through the R&D they directly carry on or sponsor. They might know that a particular molecule cures a pathology, but remain unaware of its effects when combined with another. In such cases, does the “enabling” factor refer to the methodological capacity of the producer to gather sufficient knowledge (and if so, to what extent?), or is discoverability to be ascertained through existing knowledge stemming, say, from clinical trials? There is evidence of the discoverability loop in the thalidomide story, as the trials conducted on animals had failed to prove the drug’s toxicity. Rats did not metabolize the drug, and it was assumed the same would be true for humans, but this was ultimately true of many tests in which extrapolation of animal to human response is impossible.[89] In such circumstances the producer did not have knowledge, and thus, the DRD appears to apply, leaving victims without compensation. An alternative approach is to suggest that the producer was in fact able to discover the defect, had it combined the result of these trials (knowledge) with proper human clinical trials (discoverability).

In other words, what are the effects of traces or partial knowledge on the discoverability variable? Barker et al. ask the right question:

[W]hat if there is some evidence that the product may have been defective but the causal connection has not been established according to the usual scientific methods used to determine whether use of a product can cause a particular side effect (and hence be defective as not being as safe as persons generally are entitled to expect)?[90]

They add: “Can the [DRD] apply to a supply of the drug at the time the potential risk was discovered but before it had been established whether the risk was a real risk? It appears that it can.”[91] Until now, case law has remained focused on knowledge, neglecting discoverability. Nonetheless, a Dutch case handed down in the sensitive context of contaminated blood concluded that the DRD applied in the following context.[92] The plaintiff contracted HIV during the famous window period in which detection is impossible (that is, the period between the moment of infection and the actual development of HIV detectable antibodies); so that the risk, hence, was unavoidable. The court of first instance held that the blood supplier had acted in light of technological and scientific knowledge at the time of the transfusion. Was the defect discoverable with the help of proper research and development (and funding)? The court never really asked that question, as it rather relied on the consumer expectation test: “the general public is entitled to expect that blood products in the Netherlands have been 100% HIV-free for some time.”[93] Moving away from the prevailing objective criteria that the landmark case A and others had advocated, the Dutch case, precisely through the door left open by the discoverability element, seems to reintroduce the possibility of American-inspired interpretations.

Discoverability gains further force when the producer is capable of setting its own development risk boundaries through corporate strategies. Bureaucracies do not repeat R&D already carried out in laboratories and universities. Once invited to release a market authorization, for example, national regulators (e.g., Health Canada) and supranational agencies (e.g., European Medicines Agency, the Food and Drug Administration) evaluate the methodology that underlies scientific knowledge. Technocrats analyze the exactitude of scientific correlations between variables to assess the risks and potential benefits. However, the formulation of hypotheses and even the very selection of appropriate variables, through protocols, are among the prerogatives of the researchers, who will decide—whenever economically feasible[94]—what will be studied[95] in order to meet the regulators’ requirements. Ironically enough, it may not be in a producer’s best interest to know too much—including pharmacogenomic effects[96]—as too much knowledge could lead to the exclusion of the DRD, and, consequently, expose producers to liability. A group of Australian authors accordingly note this “incentive for a defendant not to conduct the necessary further testing to establish the causal connection, although the failure to do so may well amount to common law negligence.”[97]

Commenting on the UK blood contamination case A and others,[98] another author formulated similar comments, emphasizing that

[o]ne issue which was left unclear by Burton J’s judgment is whether the conduct of the producer, particularly in relation to efforts to discover the relative safety of a product by comparing it with another, is to be taken into account when assessing the discoverability of the defect.[99]

Although this would have properly addressed the discoverability issue, scrutinizing safety testing amounts to reintroducing fault into the product liability dialectic, and would certainly expand the length and complexity of trials, which strict liability sought, in the wake of the 1970s, precisely to circumvent.

A single example will suffice to illustrate these difficulties. In the recent 222-page Canadian common law negligence case Andersen v. St. Jude Medical,[100] the safety of heart valves coated with Silzone was the object of a class action after some patients suffered thromboembolic complications. Designed and manufactured by the defendant, the medical device had been authorized for distribution and sale in Canada in 1997.[101] After being regularly implanted in patients beginning in September 1997, a voluntary recall of the valves was issued on 21 January 2000. A clinical trial carried on while the device was on the market—under the name AVERT[102]—had brought to light “a small, but statistically significant increase in explants due to a medical complication known as paravalvular leak (PVL).”[103] The AVERT study, hence, ended prematurely in 1999. As 36,000 devices had already been sold, national regulators took varying action to manage the risk.[104]

The decision captured the spirit of many—if not all—actions involving defective products:

While it would be naïve to think that the company was unconcerned about profits or protecting its intellectual property, no valve manufacturer would be in business very long if it neglected patient safety and marketed products that didn’t work.[105]

The crux of the case materializes in the key issue surrounding the allegation of a design defect, which the court summarizes by asking whether “a Silzone coating on a mechanical heart valve [put] patients at a materially increased risk of experiencing one or more ... complications.”[106] The court answered this question negatively, holding that

the plaintiffs did not establish that the defendants failed to exercise a reasonable degree of care in the pre-market design and testing or in the post-market surveillance of Silzone-coated products that would be expected of a reasonable and prudent prosthetic heart valve manufacturer in similar circumstances.[107]

More specifically, the plaintiffs had argued that although the defendant St. Jude complied with the FDA’s Draft Heart Valve Guidance, ISO 5840[108] and ISO 10993[109] standards,[110] more testing should nonetheless have been done following pre-clinical (in vitro) and animal trials,[111] thereby contesting the American-style risk utility assessment[112] advocated by the defendant.[113]

St. Jude appears to have sought regulatory approval of its Silzone-coated valve despite gaps in knowledge about its clinical effectiveness:[114] cautionary labelling reflected the strategic step-by-step approach taken by the defendant. The plaintiff believed the coated valve offered no greater benefit than another one already marketed by the same defendant. On this issue, the court held that the risk utility test does not require defendants to assess

whether the benefits of the Silzone valve outweighed the benefits of the conventional valve relative to their risks. Rather, it was required to consider whether the potential benefits associated with the addition of Silzone outweighed the potential risks of Silzone.[115]

Methodologically, the court confined the analysis to the device itself, shifting away from the American idea suggesting that product comparison and safer alternatives be at the heart of asserting defectiveness. Accordingly, in the eyes of the court “the conventional valve did not meet the same need as the Silzone valve because it did not address the risk of [prosthetic valve endocarditis].”[116] Through these lenses, the coated valve brought about distinct risks, and different benefits, as “PVE was a known risk with the conventional valve that the Silzone valve had the potential to address.”[117] The case was therefore dismissed.

In the UK, scholarship suggests that “industry practice is less relevant than industry capability, or the potential for greater safety, in judging whether a ‘state of the art’ defence is made out.”[118] Albeit stringent, the hybrid nature of such a position appears more satisfactory, as some objectiveness in the criteria allows judges to leave aside dubious corporate choices designed to avoid specific risk testing.[119] From a conceptual point of view, though, asserting the centrality of such a paradigm may well amount to merging, as was already mentioned, the DRD with the distinct state-of-the-art defence and vocabulary.

D. Eppur si muove ... (Thou Shall Dissent)

This raises a final question, as R&D is largely advanced by initially marginal scientific voices: Were Galileo and Copernicus aware of the close intimacy between dissent and heresy? What are the “weight” and relevance of dissent and isolated scientific opinions? In his submission to the ECJ in Commission v. UK, Attorney General Tesauro rightly pointed out that “the progress of scientific culture does not develop linearly in so far as new studies and new discoveries may initially be criticized and regarded as unreliable by most of the scientific community, yet”—he added—“subsequently after the passage of time undergo an opposite process of ‘beatification’ whereby they are virtually unanimously endorsed.”[120] He then tackled the issue directly: “[W]here there is a risk that is not certain and will be agreed to exist by all only ex post, [may] the producer ... still rely on the [DRD]”?[121] In answer to his own query,[122] he stated that

the state of scientific knowledge cannot be identified with the views expressed by the majority of learned opinion, but with the most advanced level of research which has been carried out at a given time.[123]

The scientific community recently provided an interesting example, by investigating whether eating genetically modified maize causes health problems in laboratory rats. The answer suggested in a scholarly article published in Food and Chemical Toxicology by G.-E. Séralini et al. is that it does, as their research results

demonstrate that lower levels of complete agricultural glyphosate herbicide formulations, at concentrations well below officially set safety limits, induce severe hormone-dependent mammary, hepatic and kidney disturbances. Similarly, disruption of biosynthetic pathways that may result from overexpression of the EPSPS transgene in the GM NK603 maize can give rise to comparable pathologies that may be linked to abnormal or unbalanced phenolic acids metabolites, or related compounds.[124]

In sum, “the significant biochemical disturbances and physiological failures documented in this work confirm the pathological effects of these GMO and R treatments in both sexes, with different amplitudes.”[125] Many scientists fiercely opposed these results, arguing that the type of rats used in the study was predisposed to tumours, and that the statistical evidence presented was inconclusive.[126] These methodological gaps therefore allegedly contaminated the experiment and gave rise to results that—despite being subjected to appropriate peer review—were not scientifically sound.[127] Is Séralini’s article credible? When a reliable answer cannot be found within the scientific community, it is even more elusive to legal experts, who are intellectually ill equipped to second-guess scientists’ methodological choices and tactics. In such matters, speculation is always at best tentative and at worst empty rhetoric.

From a legal standpoint, then, what can possibly be done in the face of such scientific polemics? Coderch and Puig anticipated the problem triggered by “uncertain science” that may nonetheless influence the amount and quality of knowledge[128] for the purpose of making findings on a balance of probabilities.[129] Quite predictably, the debate leads to the junk science–good science dialectic, which underlies the landmark American case of Daubert,[130] as the criteria embedded in the DRD invite analysis of the scientific evidence presented by expert witnesses (whether appointed by the court or hired by the parties).[131] In a different context, the Australian decision in Peterson v. Merck Sharp & Dohme[132] was a pharmaceutical consumer class action involving the drug Vioxx. The court held in that case that the drug’s side effects were not sufficiently known to the producer so as to constitute scientific knowledge. The drug had been approved around 1999 by regulators but was withdrawn from the market in the fall of 2004 after a study revealed that its side effects included doubling the risk of adverse cardiovascular events.[133] A year later, the federal appeals division, known as the Full Court, rejected the imposition of a duty to warn on the defendant Merck toward the plaintiff of risks of heart disease, as scientific and technical knowledge was inconclusive.[134] Indeed, the result of one study made available in March 2000 named VIGOR was held insufficient to establish scientific knowledge, hence allowing recourse to the DRD, in force in section 142 of the Australian Consumer Law.[135]

Can this debate on the place of dissenting science be further nourished by scholarship in international trade law? The temptation is great, as minority opinions have caught the attention of WTO arbitrators in their interpretation of the Sanitary and Phytosanitary (SPS) Agreement,[136] and specifically determining the quality of scientific expertise underlying SPS measures adopted or maintained by member states. Article 2(1) of the SPS Agreement provides that such measures may be undertaken if “necessary for the protection of human, animal or plant life or health.” All such SPS measures must be “based on scientific principles and [shall not be] maintained without sufficient scientific evidence.”[137] In interpreting this provision, the WTO appellate body has mentioned that risk assessment is not bound by any “monolithic conclusion” of majoritarian scientific opinion, as these may hide the uncertainty within the community.[138] Nonetheless, SPS measures still must meet the “sufficiency” criterion[139] by demonstrating “the existence of a sufficient or adequate relationship between two elements, in casu, between the SPS measure and the scientific evidence.”[140]

The wording of the DRD probably opens the door to similar questions that flow from the definition of “scientific and technical knowledge” contained at article 7(e) of the PL Directive. A last, intriguing question will be left unexplored: To what extent does “science” include the humanities, epidemiology, or metaphysics?

Conclusion

The effects of the DRD on product liability are twofold. First, it predictably sends victims back to the labyrinth of fault-based liability (or other relevant causes of action) and invites them to search out other defendants. Focusing on different elements of the DRD may result in a more or less subjective analysis. Insisting that efforts could have been made by an industry to discover a defect, to carry on further research, etc., saddles plaintiffs with a burden that the directive seemingly sought to lift. As Howells and Mildred suggest, this situation is explained by the political compromise that was reached in the PL Directive. Even if it comes at the cost of further fractioning the interior market, flexibility as to the strictness of liability may well be available to member states willing to exercise their discretion. In this regard, strict liability and fault-based liability advocates will each long persist within the world of product liability, accompanied by “third-way”[141] proponents of hybrid solutions.

Second, the policy reasons underlying the exoneration clause have always been economic in nature: too much liability is said to chill innovation. However, little tangible proof of the correlation has been presented in empirical research. Ironically enough, although this aspect of the DRD merits further study, its proponents have never really been bothered by the lack of evidence surrounding the argument’s foundation in the alleged “chilling effect”. Indeed, one study conducted at the European Commission’s request remains utterly prudent in establishing such a cause-effect relationship.[142] A qualitative analysis was also carried out by the Fondazione Rosselli to illustrate the impact of the DRD on product liability. Published in 2004, the study cautiously concluded that

it can be said that the [DRD] has had the merit of providing industry with a clear-cut reference for evaluating product safety. At the same time, there is no evidence that the absence of the [DRD] in specific countries and/or industries have significantly hindered innovation.[143]

Indeed, participating companies[144] expressed the sentiment that if the DRD were to disappear, they would simply invest in more comprehensive insurance coverage or would deploy additional efforts to better assess the safety of their products.[145] There is a clear contradiction between this indifference toward the DRD on the part of economic actors and alarmism about its demise.[146] The shallowness of the rationale of chilling innovation probably explains the polarized positions found in contemporary scholarship and case law.

However, antibodies are present in most states’ veins that may allow them to ward off the DRD by allowing plaintiffs to institute their action on other grounds.[147] Somewhat by default, negligence, faute, and all of their variations have historically formed the basis of personal injury law, but have long been considered inadequate bases on which to ground compensation for damage caused by innovative products.[148] Nonetheless, it is to be anticipated that the costs and the uncertainty associated with such actions will render them decreasingly effective, as proof of negligence requires scientific evidence that is often outside the reach of plaintiffs. Pharmaceutical litigation, for example, revolves around a two-part strategy, which consists of proving a failure in product safety evaluation before the products’ introduction into the market and establishing a further failure to warn health professionals (in case of prescription drugs) or consumers (for over-the-counter medicines) of a risk. Proof of causation can also turn cases into fora for scientific debate,[149] as the waltz of variables endlessly shifts according to new hypotheses, theorems, deductions, and assumptions. Such an endeavour requires expert examination of the methodology and protocols underlying clinical trials and often involves a level of complexity that materializes in extended hearings.

Ultimately the benefit of the DRD lies not so much in its actual results for producers, but rather in the enrichment of judicial discourse through the use of comparative legal approaches that increase the circulation of ideas and solutions available within courtrooms.[150]