The statistical possibility of factual causation less than a probability may be enough for general causation in products liability class actions, even if it means the compensated class includes persons whose injury was not, in fact, caused by the product.  See  Bartram v. GlaxoSmithKline Inc., 2012 BCSC 1804 at paras. 27-35.

In product liability actions, general causation is the inquiry into whether the product is capable of causing the type of harm the plaintiff alleges the plaintiff sustained. See Harrington v. Dow Corning Corp., 2000 BCCA 605 at para. 42, 193 DLR (4th) 67

[42] … the first step in every products liability case alleging negligent design, manufacture, or marketing is the determination of whether the product is defective under ordinary use or, although non-defective, has a propensity to injure. Some American authorities refer to this step as “general causation”, whether a product is capable of causing the harm alleged in its ordinary use.

The test for causation in product liability class actions may not be the but-for test but some version of the material contribution to injury test. See Clements v. Clements, 2012 SCC 32 at para. 44

[44]  This is not to say that new situations will not raise new considerations. I leave for another day, for example, the scenario that might arise in mass toxic tort litigation with multiple plaintiffs, where it is established statistically that the defendant’s acts induced an injury on some members of the group, but it is impossible to know which ones.

This leads to Bartram v. GlaxoSmithKline Inc., a class action certification motion. The action was certified. The first of the proposed common issues was causation.

1) Did Paxil cause or increase the likelihood of birth defects?

[27]  The parties have put forward conflicting expert evidence on the issue of causation. The defendant relies on the evidence of Dr. Edward Lammer, a pediatrician and medical geneticist, who says that every woman has a three per cent chance of giving birth to a baby with a congenital malformation. He adds that the causes of such malformations are diverse, that no single agent can cause all of them and only about one per cent of all major congenital malformations are caused by exposure to chemicals, medications or radiation. He says the causative role of Paxil, if any, must be determined on a case-by-case basis.

[28] The plaintiffs rely on the evidence of an epidemiologist, Shira Kramer, who says there is a “consistent body of epidemiological research” that establishes that Paxil causes cardiovascular birth defects.

[29] It is neither necessary nor appropriate on a certification application to weigh that evidence or to consider the limitations of each witness’s expertise. In any case, GSK’s own published material has acknowledged that epidemiological studies suggest that the use of Paxil during pregnancy is associated with at least an increased risk of cardiovascular defects in newborns.

[30]  Authorities such as Harrington v. Dow Corning Corp.,  2000 BCCA 605, establish the distinction between general and individual causation. In the context of this case, the general causation question is whether Paxil is capable of causing cardiac birth defects and, if so, which ones. That will depend on expert evidence that will be applicable to the claim of all class members.

[31]  If the plaintiffs fail to prove general causation, that will be the end of the matter. If they succeed, it will then be up to each individual plaintiff to show that the injury that occurred was of a kind that can be caused by Paxil and was in fact one that would likely not have occurred but for the use of Paxil.

[32]  In an individual action, a plaintiff probably could not succeed by merely showing that the use of Paxil increased the risk of injury. In Clements v. Clements, 2012 SCC 32 (CanLII), 2012 SCC 32, the Supreme Court of Canada re-affirmed the primacy of the “but for” test in proving causation and confined the alternate “material contribution” test to cases involving multiple negligent defendants where it is not possible to prove which one caused the injury. However, dicta in Clements may leave open an argument that different considerations apply in cases involving multiple plaintiffs, such as class actions.

[44]  This is not to say that new situations will not raise new considerations. I leave for another day, for example, the scenario that might arise in mass toxic tort litigation with multiple plaintiffs, where it is established statistically that the defendant’s acts induced an injury on some members of the group, but it is impossible to know which ones.

[33]  Depending on what findings the court makes on some of the other common issues, each individual adult plaintiff may also have to prove that a reasonable person in her position, having been informed of the risk of taking Paxil and of the countervailing risks of changing or discontinuing treatment, would have stopped taking Paxil.

[34]  One should not minimize the difficulties each plaintiff may face in proving individual causation, but those issues will be irrelevant without a finding on general causation, which is clearly a common issue.

[35] I would, however, narrow the question to whether Paxil causes or increases the likelihood of cardiovascular birth defects. That is the type of defect alleged in the case of the proposed representative plaintiff and is the only type referred to in the proposed class definition.

Anybody care to suggest that Justice Smith did not  rule that the test for general causation as a common issue would be material contribution to risk, not but-for?

Canadian practitioners interested in the manner in which causation arguments based on statistics become relevant in class action product liability cases will find find useful looks at the subject in recent articles such as Lynda M. Collins and Heather McLeod-Kilmurray, “Material Contribution to Justice? Toxic Causation after Resurfice Corp v. Hanke (2010) 48 Osgoode Hall L.J. 411; and, Craig Jones, “Reasoning Through Probabilistic Causation in Individual and Aggregate Claims: The Struggle Continues” (2011) 39 Adv. Q. 18. Both articles predate Clements but that is irrelevant to the point made by the authors.