Causation in Medical Liability

Overview

Fault (faute) draws most of the attention in a medical liability file, yet the (lien de causalité) is the element on which most claims are won or lost. A physician may have breached the standard of the reasonably prudent practitioner and still owe nothing, because the patient's condition was already beyond recovery, because the injury flowed from the underlying disease, or because the evidence cannot separate the consequences of the fault from the consequences of the accident that brought the patient to hospital. This note, the last in the medical liability series, examines how Quebec courts establish the existence of the causal link, how it is proven, and which presumptions may or may not relieve the victim of a burden that medical science frequently cannot satisfy. It draws on Snell c. Farrell, Laferrière c. Lawson, St-Jean c. Mercier, and Benhaim c. St-Germain, together with the Quebec case law that applies them.

Learning Objectives

Explain why (causalité juridique) differs from and why the distinction favours the victim in medical matters.
Distinguish the that may flow from a breach of the from the absence of any corresponding presumption of causation after St-Jean c. Mercier.
Apply the conditions of art. 2849 CCQ to a case in which the physician's fault deprived the patient of the evidence needed to prove causation, and explain the discretionary character of the confirmed in Benhaim c. St-Germain.

Preliminary Remarks on Causation in Medical Matters

Legal Causation and Scientific Causation

In most areas of civil liability the causal question answers itself. The Superior Court in Succession d'Ex-Mari et al. c. Bernier, 2021 QCCS 1475, gave the standard example: someone throws a baseball carelessly and shatters a neighbour's window. Medicine is different. The patient arrived already ill or injured, the body reacts to treatment in ways that vary from one person to the next, and the same clinical picture may have several possible origins. The court in Bernier therefore described causation as a question of fact left to the trial judge on the evidence adduced, and insisted on the threshold that matters: the fault must have participated in a significant way in the materialization of the injury. A fault that is merely one element in the background does not suffice.

The same decision identified the principle that governs the rest of this note. Legal causation is distinct from scientific causation. The court does not pursue scientific certainty; it pursues legal truth, and it draws or declines to draw factual conclusions on the . A connection that a scientist would label uncertain may nevertheless be probable enough to satisfy the legal standard. Professor Pierre Nicol has put the contrast in a sentence: scientific causation requires quasi-mathematical certainty, whereas legal causation is recognized on a simple balance of probabilities.

That formulation reproduces, in civil-law language, the position taken by the Supreme Court of Canada in Snell c. Farrell, [1990] 2 SCR 311, where Justice Sopinka observed that much of the dissatisfaction with the traditional approach to causation stemmed from courts applying it too rigidly, and that causation does not have to be determined with scientific precision. Although Snell was a common-law appeal, Quebec courts have adopted its central teaching. In Laforce c. Dumont, 2003 CanLII 17139 (QC CA), the Court of Appeal held that the Supreme Court's warning against placing an exaggerated burden on plaintiffs in medical matters applies in Quebec civil law.

The practical consequence appears in Véronneau c. Girouard, 2021 QCCS 1704, an obstetrical case. The question was whether the plaintiffs had shown, on the balance of probabilities, that the failure to deliver the child before a 26-minute bradycardia, and the occurrence of that bradycardia, constituted the direct, logical, and immediate cause of the child's cerebral palsy. The court recalled that the burden is not one of scientific or medical certainty, and that the existence of other possible medical causes does not, by itself, prevent the plaintiffs from meeting it. The Court of Appeal had taken the same position in Brousseau c. Laboratoires Abbott limitée, 2019 QCCA 801.

Two Illustrations of the Difficulty

Two Quebec decisions show how a claim fails at the causation stage even where fault is arguable or established.

In Zanchettin c. De Montigny, [2000] R.R.A. 298 (C.A.), a 20-month-old child was found motionless in a pond next to the family's country home. The parents drove the child to a physician's residence; the physician, observing no vital signs, advised them to go to the hospital so that the death could be recorded. At the hospital the child was resuscitated but was left with severe neurological sequelae. The Court of Appeal upheld the dismissal of the action against the physician: the anoxia had already damaged the child by the time the child reached the physician's door. The deficit had been produced by the drowning, not by the delay.

Berlinguet c. Lamson-Engo, 2009 QCCS 910, illustrates a different failure. The patient had attended the emergency department for significant bleeding and had been discharged. At home she fell against the glass of a French door, severed her carotid artery, and died. Her relatives argued that the discharge had been premature. The court identified two possible causes of the fall: a syncope brought on by hypovolemia, which would have connected the fall to the discharge, or a simple trip while walking through the kitchen, which would not. The judge could not find the relatives' account more convincing than the defendants' on the balance of probabilities. Even if the early discharge was a fault, the plaintiffs had not shown that the fault, rather than an unrelated misstep, caused the fatal fall.

Establishing the Causal Link

The Balance of Probabilities

The rules of evidence in the Civil Code of Quebec (Code civil du Québec, CCQ) set the frame. Under art. 2803 CCQ, the person who wishes to assert a right must prove the facts on which the claim is based, and the person who claims that a right is null, modified, or extinguished must prove the facts on which that contention rests. Under art. 2804 CCQ, evidence that makes the existence of a fact more probable than its non-existence is sufficient, unless the law requires more convincing proof.

Applied to medical liability, these provisions produce three consequences:

The patient bears the burden of proving fault, injury, and causation. Nothing in the Code shifts that burden to the physician because the defendant is a professional or because the facts lie within the physician's knowledge.
The plaintiff need not exclude every other explanation, but must make the explanation involving the fault the more probable one.
Expert testimony that a link is "possible" will rarely suffice, but expert testimony need not reach certainty for the court to find the link "probable".

The substantive test applied within this frame is expressed in the language of art. 1607 CCQ, which entitles the creditor to damages for the injury that is a of the debtor's default. Véronneau asked whether the fault was the direct, logical, and immediate cause of the injury; Bernier asked whether the fault participated in a significant way in the materialization of the damage. These formulations describe the same inquiry: among the antecedents of the injury, the court identifies those that made it probable rather than merely possible, and asks whether the physician's fault belongs among them.

Loss of Chance and the Laferrière Decision

Medical causation problems frequently present themselves as claims for . The argument runs as follows: the physician's fault cannot be shown to have caused the patient's death or disability, but it deprived the patient of a statistical chance of avoiding that outcome, and that lost chance is itself a compensable injury. Quebec law's answer was given in Laferrière c. Lawson, [1991] 1 SCR 541.

The physician failed to tell the patient that she had cancer and therefore failed to disclose that she retained a small chance of survival, estimated at 3 percent. Even had she been told in 1975, the state of medicine at the time meant that nothing could have altered the outcome. After her death, her succession sued the physician for having deprived her of that chance.

The trial judge held that loss of chance is not compensable in Quebec civil law, and in medical liability in particular. The test was the balance of probabilities, and the scientific evidence was overwhelming: the patient had almost no chance of survival, so she had not died because of the physician's fault. The Supreme Court confirmed that the death could not be attributed to the physician and that no damages could be awarded for the lost chance. If the plaintiff must show that the fault more probably than not caused the death, a 3 percent chance of survival means that the fault, more probably than not, did not cause it. Recharacterizing the injury as the loss of the chance would allow recovery where the ordinary test has failed and would in substance lower the standard of proof.

The Court did not leave the succession without a remedy. Two heads of damage survived because they were causally connected to the fault on the ordinary standard:

The patient had suffered a direct emotional shock from not having learned of her condition earlier. The Court awarded $10,000 for this injury.
Had she known, she would have enjoyed a better quality of life in her final months, in particular through earlier treatment. The Court awarded a further $7,500.

The injuries the fault caused were compensated in full; the death was not compensated at all, and could not be compensated by fractions under the label of a lost chance. The causation inquiry was conducted head of damage by head of damage, and that is the method the student should adopt.

No Special Regime for Medical Victims

A recurring temptation is to argue that the difficulty of proof justifies a distinct regime for victims of therapeutic accidents. Snell c. Farrell addressed the argument directly. The Court acknowledged that it is often difficult for the patient to prove causation in cases of medical fault, but declined to adopt either of the approaches derived from the English decision in McGhee, since either would compensate the plaintiff in the absence of a significant connection between the injury and the defendant's conduct. A plaintiff should not be compensated, through a reversal of the , for an injury that may well result from factors unrelated to the defendant and from no one's fault.

Quebec case law has received this reasoning without reservation. Medical victims benefit from no special regime; their burden is the same as that of any other plaintiff. What the courts have granted instead, as Laforce c. Dumont confirms, is a recognition that the standard must not be applied with unrealistic rigour. The difference is one of application, not of rule.

Proof of Causation

Expert Evidence and the Role of the Judge

Causation in medical matters is proven, in practice, through expert testimony. The plaintiff's expert explains the mechanism by which the fault produced the injury and gives an opinion on probability; the defence expert proposes alternative mechanisms or contends that the outcome was inevitable. The judge does not choose between them on scientific grounds but decides which account is more probable in light of all the evidence, including the medical record, the timeline, and the credibility of lay witnesses.

Three consequences follow:

The expert's vocabulary must be translated. An expert who describes a link as "uncertain" in scientific terms may be describing something that is legally probable.
Alternative causes must be weighed, not merely listed. Following Véronneau and Brousseau, the defence must make its alternative explanation as probable as, or more probable than, the plaintiff's. Where two mechanisms are equally plausible and the evidence cannot separate them, as in Berlinguet, the plaintiff fails regardless of fault.
The judge may prefer one expert. In Benhaim c. St-Germain, 2016 SCC 48, discussed below, the trial judge relied on a statistical opinion, and the Supreme Court held that she was entitled to believe that expert and decide the case on that basis.

Presumptions of Fact under Article 2849 CCQ

Direct evidence of causation is often unavailable, and the Code permits the court to reason by inference. Article 2849 CCQ provides that presumptions not established by law are left to the discretion of the court, which shall take into consideration only those that are serious, precise, and concordant. In the causation context this is the only presumption that exists: the burden rests on the victim, attempts have been made to reverse it, and the Supreme Court has refused to recognize any presumption beyond the ordinary .

Three characteristics of the presumption of fact structure the discussion that follows:

It is discretionary. The court may draw the inference; it is never obliged to. A trial judge who declines to draw it commits no reviewable error, provided the refusal is grounded in the evidence as a whole.
It is conditioned. The known facts must be serious, precise, and concordant. A single suggestive fact will not do.
It is a mode of proof, not a new standard. A presumption of fact is a way of reaching the balance of probabilities by inference. It does not lower the standard and does not by itself reverse the burden.

Everything that follows under the heading of "presumptions" is, on analysis, an application of art. 2849 CCQ to particular fact patterns.

Discussion of Certain Presumptions

Breach of a Statute or Code of Ethics

The first question is whether a physician's breach of a statutory or regulatory norm, in particular the Code de déontologie des médecins, gives rise to any presumption. The answer differs for fault and for causation.

On fault, Morin c. Blais, [1977] 1 SCR 570, and the later decision in Kosoian established that the breach of a norm found in a statute or regulation may be used to establish fault. Where the physician has contravened a rule of the Code de déontologie, the court may treat the contravention as evidence that the standard of conduct was not met. The weight given to the Code varies from one judge to another, but it is a legitimate element in the proof of fault.

On causation, the position is settled by St-Jean c. Mercier, [2002] 1 SCR 491. The Court explained that in Morin it had been correcting an approach of the Quebec Court of Appeal that appeared to see only two mutually exclusive possibilities, the fault of one party or the fault of the other, and that the principle it had formulated about presuming causation was confined to a precise situation. St-Jean thereby sets aside Morin c. Blais as authority for any general presumption of causation flowing from a statutory breach. A breach of the Code de déontologie may help to prove fault, but it is of no assistance in proving the causal link.

A Fault Carrying a Manifest Danger

The second and most debated presumption arises where the physician's fault created an obvious danger and that danger materialized. In Laferrière c. Lawson, Justice Gonthier wrote that in certain cases where the fault carries a for the patient's health and safety, and where that danger has materialized, it may be logical for the judge to presume the existence of a causal link between the fault and the injury, subject to a demonstration or a strong indication to the contrary. Read in isolation, the passage seemed to announce a new rule: a physician who creates a risk by performing a manipulation is liable once fault is shown, and the causal link is presumed.

St-Jean c. Mercier closed that opening. The Court of Appeal had held that it is not enough to show that the defendant created a risk of injury and that the injury then materialized within the zone of risk so created, and the Supreme Court agreed. To the extent that the notion of a materialized risk was being treated as a distinct means of proof carrying a less demanding standard, Snell and unquestionably Laferrière ought to have put an end to those attempts to bypass the traditional rules of proof. Justice Gonthier addressed his own earlier words directly: the statement in Laferrière had perhaps been misread. It merely restated the traditional rule applicable to presumptions and created no additional means of proof in Quebec civil law for establishing causation.

The result is a discretionary power, not a rule of reversal. The judge may presume causation where a fault carrying a manifest danger has been followed by the materialization of that danger, but is not obliged to draw an adverse inference against the physician. Whether the inference is drawn depends on whether the facts, taken together, are serious, precise, and concordant under art. 2849 CCQ. Creating a risk within which the injury later occurred is not the same as causing the injury.

The outcome in St-Jean shows the discretion at work. The patient had been injured in a motor vehicle accident and was later paralyzed. The court, choosing between the accident and the physician's risky manipulation, concluded that the paralysis was attributable to the accident and found no fault on the physician's part, so that compensation fell to the public automobile insurance scheme. The course materials contrast this with the later Godbout decision, which also involved a patient injured in an automobile accident and then treated in hospital. In St-Jean, no one relied on the fact that the initial injury was an automobile accident; in Godbout, the court took the accident into account to dismiss the action and refer the victim to the public scheme. Godbout reflects the current state of the law on that interaction.

When the Physician's Fault Prevents Proof

The third presumption concerns a narrower situation: the physician's fault not only caused (or may have caused) the injury but also deprived the patient of the means of proving what caused it. The typical example is the physician who fails to perform a test or record an observation that would have fixed the patient's condition at a particular moment, so that no one can later say whether the injury preceded or followed the treatment.

The Court of Appeal addressed this pattern in two decisions from the 1980s. In Houde c. Côté, 1987 CanLII 387 (QC CA), the court asked whether, when the patient cannot prove the cause of an injury because the physician omitted to provide a complete report of the intervention, there is a presumption against the physician, and answered that there is. In Gburek c. Cohen, REJB 1988-63095 (C.A.), the court went further. Even though the physician's obligation is generally one of means, where the physician, in addition to providing faulty care, prevents the patient from being in a position to prove the causal relation between the faulty care and an injury that may have been caused by it, the burden shifts to the physician to prove the absence of causation. Because the physician had not made that proof, the court took it as established that abnormal concentrations of gentamicin had been tolerated in the patient's blood.

Leduc c. Têtu, 2010 QCCS 623, applied the Gburek principle. The patient arrived at hospital with an injury to one eye. The physician performed no test to determine whether the patient still had vision in that eye, and the patient later became blind in it. To succeed, the patient had to show that the blindness resulted from the physician's fault rather than from the original accident, which required proof that the eye could still see on arrival. That proof was impossible, because the physician had omitted the very examination that would have established it. The court held that it fell to the physician to demonstrate the absence of causation on the balance of probabilities. The defendants argued both that the patient was already blind and that the intervention might not have saved the eye; the court answered that they bore the burden on both points and had not discharged it. The plaintiff succeeded.

The Supreme Court then had to decide whether this line of authority created a rule. In Benhaim c. St-Germain, 2016 SCC 48, a small spot was detected on the patient's lungs in November 2005. By December 2006 the spot had grown, and in January 2007 lung cancer was diagnosed. Everyone agreed that the physician had not acted as a prudent and diligent practitioner. The difficulty was causation. If the spot detected in 2005 had been a stage one cancer, earlier treatment would have extended the patient's life considerably and the claim was valued at approximately $1.7 million. If it had already been a stage four cancer, the patient would most probably have died within months in any event, and the claim was worth approximately $70,000. The fault had made it impossible to know which.

The trial judge considered the statistical evidence that 78 percent of cancers detected in comparable circumstances are at stage one, but declined to apply the Leduc rule, held that the victim retained the burden, and was not persuaded that the cancer had been at stage one. The Court of Appeal reversed: it drew an adverse inference, presumed a stage one cancer, placed on the physician the burden of proving stage four, and awarded $1.7 million.

The Supreme Court restored the trial judgment. The adverse inference remains discretionary. The trial judge was entitled to believe one expert and to decide not to apply the presumption. The presumption does not apply automatically as soon as the victim is unable to prove causation because of the physician's fault; it must be assessed in light of the whole of the evidence. The judge may draw the inference where the evidence as a whole shows facts that are serious, precise, and concordant, but the judge has the last word. The family received $70,000.

Benhaim thus reconciles the earlier case law with art. 2849 CCQ. Houde, Gburek, and Leduc remain good illustrations of situations in which a court may reasonably infer causation against a physician whose fault destroyed the evidence; they do not establish a mandatory reversal of the burden. Counsel for a patient must establish the fault that prevented proof and persuade the trial judge, on the entire record, that the inference should be drawn; counsel for the physician must offer evidence that makes the alternative explanation at least as probable.

A Statutory Exception for Tobacco Litigation

The rule that causation cannot be established by statistics alone has one legislative exception, which confirms the rule by contrast. Section 15 of the Tobacco-related Damages and Health Care Costs Recovery Act (Loi sur le recouvrement du coût des soins de santé et des dommages-intérêts liés au tabac) provides that, in an action brought on a collective basis, the causal link between a defendant's fault and the cost of health care whose recovery is sought, or between exposure to a tobacco product and the disease or deterioration of health of the recipients of that care, may be established on the sole basis of statistical, epidemiological, sociological, or other relevant studies, including information obtained by sampling. The same applies to proof of the cost of the care claimed.

Outside that statutory field, the individual medical liability plaintiff must prove causation in the particular case, and statistical evidence is one element among others for the judge to weigh under art. 2849 CCQ, as Benhaim demonstrates.

Worked example: sorting out fault, causation, and presumptions

A 58-year-old patient attends a walk-in clinic complaining of intermittent chest discomfort. The physician takes a brief history, does not order an electrocardiogram, and sends the patient home. Four days later the patient suffers a myocardial infarction and is left with permanent cardiac impairment. The patient sues, alleging that the failure to order the electrocardiogram was a fault and that timely referral would have avoided the infarction.

Fault. The plaintiff's expert testifies that a reasonably prudent physician would have ordered an electrocardiogram. A contravention of the Code de déontologie could support that finding under Morin and Kosoian, but says nothing about causation after St-Jean.
Causation on the ordinary burden. The plaintiff must show that timely referral would more probably than not have prevented the infarction or reduced its consequences. The defence expert testifies that the coronary lesion was already advanced and lists other possible mechanisms. Following Véronneau, the mere existence of alternatives does not defeat the claim; the judge decides which account is more probable.
Loss of chance. If the evidence shows only that referral would have given the patient a 20 percent chance of avoiding the infarction, Laferrière forecloses recovery for the infarction itself. The plaintiff may still recover for any distinct injury the fault probably caused, such as four days of untreated pain and anxiety.
Presumption from a manifest danger. Failing to investigate chest pain carries an obvious danger which materialized. After St-Jean, that argument invites a presumption of fact; it does not compel one and does not shift the burden.
Fault preventing proof. If the missing electrocardiogram is the only evidence that could have shown whether the cardiac event was already under way, Gburek and Leduc support an inference against the physician. Benhaim reminds the plaintiff that the trial judge may decline to draw it if the defence account remains at least as probable.

A judge who accepts that the lesion was already advanced will find fault without causation, on the model of Zanchettin. A judge persuaded that the missing test deprived the patient of the only means of proof, and that the facts are serious, precise, and concordant, may infer causation on the model of Leduc.

Practice Checklist

Framing the causal question

Identify each head of injury separately and ask, for each, whether the fault more probably than not produced it (the Laferrière method).
Confirm that the plaintiff bears the burden under art. 2803 CCQ and that the standard is the balance of probabilities under art. 2804 CCQ.
Ask whether the fault was the direct, logical, and immediate cause of the injury, or participated in a significant way in its materialization.

Working with expert evidence

Translate the expert's scientific language into the legal standard.
Weigh alternative causes rather than treating their existence as fatal (Véronneau, Brousseau).
Assess whether the record can separate the consequences of the fault from those of the pre-existing condition (Zanchettin, Berlinguet).

Presumptions

Use a breach of the Code de déontologie to support fault, never to establish causation (St-Jean).
Where the fault carried a manifest danger that materialized, invite a presumption of fact but do not plead a reversal of the burden (Laferrière as clarified in St-Jean).
Where the physician's fault destroyed the evidence of causation, plead the Gburek and Leduc inference and marshal the whole record so that the facts are serious, precise, and concordant (Benhaim).
Remember that the decision to draw or refuse the inference is discretionary and will rarely be disturbed on appeal.

Special situations

Where the patient was first injured in an automobile accident, consider whether the public scheme, rather than the physician, bears the loss (Godbout).
Treat statistical causation as sufficient only where a statute so provides, as in collective tobacco actions.

Glossary

adverse inferenceinférence négative: An inference drawn against a party whose own conduct has made proof impossible; in medical causation it remains discretionary rather than an automatic reversal of the burden.
balance of probabilitiesprépondérance des probabilités: The civil standard of proof under art. 2804 CCQ; a fact is proven when its existence is more probable than its non-existence.
burden of prooffardeau de la preuve: The obligation of the party asserting a right to prove the facts on which it rests under art. 2803 CCQ; the victim bears this burden for fault, injury, and causation alike.
causal linklien de causalité: The connection required between the fault and the injury under art. 1457 CCQ; the fault must have participated in a significant way in the materialization of the injury.
Code de déontologie des médecins: The regulation governing the professional conduct of Quebec physicians; its breach may support a finding of fault but does not create a presumption of causation.
direct and immediate consequenceconséquence directe et immédiate: The formula of art. 1607 CCQ describing the injury for which reparation is owed, echoed when courts ask whether the fault was the direct, logical, and immediate cause of the harm.
legal causationcausalité juridique: Causation as a court assesses it, on the balance of probabilities, as a question of fact, as distinct from scientific causation, which demands near mathematical certainty.
loss of chanceperte de chance: The argument that a fault deprived the patient of a statistical possibility of a better outcome; rejected as a distinct head of damage in medical liability by Laferrière c. Lawson.
manifest dangerdanger manifeste: An obvious risk to the patient's health created by the fault; its materialization may support, but does not compel, a presumption of causation.
presumption of factprésomption de fait: An inference the court may draw from known facts to establish an unknown fact; it is discretionary and admissible only if serious, precise, and concordant under art. 2849 CCQ.
presumption of faultprésomption de faute: The inference that the standard of conduct was breached, which may follow from a contravention of a statutory or regulatory norm such as the Code de déontologie des médecins.
scientific causationcausalité scientifique: A causal connection established to a near mathematical degree of certainty; not the standard applied in civil liability.
therapeutic accidentaccident thérapeutique: An injury arising in the course of medical care, the causation of which is often difficult to prove.

References and Further Reading

Civil Code of Quebec: arts. 1457, 1607, 2803, 2804, 2849.
Tobacco-related Damages and Health Care Costs Recovery Act (Loi sur le recouvrement du coût des soins de santé et des dommages-intérêts liés au tabac): s. 15.
Code de déontologie des médecins.
Supreme Court of Canada: Snell c. Farrell, [1990] 2 SCR 311; Laferrière c. Lawson, [1991] 1 SCR 541; St-Jean c. Mercier, [2002] 1 SCR 491; Benhaim c. St-Germain, 2016 SCC 48; Morin c. Blais, [1977] 1 SCR 570; Kosoian.
Quebec Court of Appeal: Zanchettin c. De Montigny, [2000] R.R.A. 298 (C.A.); Laforce c. Dumont, 2003 CanLII 17139 (QC CA); Brousseau c. Laboratoires Abbott limitée, 2019 QCCA 801; Gburek c. Cohen, REJB 1988-63095 (C.A.); Houde c. Côté, 1987 CanLII 387 (QC CA).
Superior Court of Quebec: Succession d'Ex-Mari et al. c. Bernier, 2021 QCCS 1475; Véronneau c. Girouard, 2021 QCCS 1704; Berlinguet c. Lamson-Engo, 2009 QCCS 910; Leduc c. Têtu, 2010 QCCS 623.
Related decision: Godbout, on the interaction between automobile accidents and subsequent medical treatment.
Doctrine: Pierre Nicol, "Faute médicale : preuve par présomptions de fait et exonération".

Disclaimer

This article is for informational purposes only and does not constitute legal advice. Quebec civil law may evolve through legislation and judicial interpretation. For advice on a specific situation, consult a qualified Quebec lawyer or notary.