Causation — factual and legal

Causation is the hinge between breach and liability. A claimant who has established duty and breach has not yet established that the defendant must pay. The law must still decide whether the breach caused the relevant damage, and whether the damage is legally attributable to the defendant rather than to coincidence, another cause, the claimant’s own decision, or an abnormal sequence of events.

For Part IA Tort, the subject is best divided into two questions. First, factual causation asks whether the defendant’s breach made a difference to the occurrence of the damage. The orthodox test is the but-for test: would the damage have occurred but for the defendant’s breach? If the answer is yes, factual causation is not proved. If the answer is no, it is. Secondly, legal causation asks whether the causal connection is sufficiently close for legal responsibility. This includes remoteness, intervening acts, scope of risk, and rules about the type of damage recoverable.

The separation is analytically useful, but not absolute. Many modern cases resist clean classification. The Fairchild exception, Chester v Afshar, and cases on intervening acts all show courts reasoning about factual connection, policy, and the normative limits of liability at once. Do not therefore treat causation as a mechanical afterthought. It is a doctrinal site where corrective justice, evidential uncertainty, distributive policy and practical adjudication meet.

The basic exam structure is as follows. Identify the damage with precision. Ask what counterfactual world must be constructed: the world in which the defendant did not breach the duty, not necessarily a world in which the defendant did nothing at all. Apply the but-for test. If the test fails because there are multiple sufficient causes, evidential gaps, or indivisible disease cases, consider whether an exceptional rule applies. Then address legal causation: was the kind of damage reasonably foreseeable; was the causal chain broken; did the claimant’s response or a third party’s act fall within the risk created by the defendant; and is the damage within the scope of the defendant’s duty?

For Cambridge supervisions, the critical skill is not listing cases but explaining what work each case does. Barnett illustrates the ordinary counterfactual test. McGhee and Fairchild concern evidential uncertainty in industrial disease. Wilsher marks the limit of that reasoning where there are multiple possible agents. The Wagon Mound anchors remoteness in reasonable foreseeability of the kind of damage. Hughes, Jolley and similar cases warn that the precise manner of occurrence need not be foreseeable. Corr and Empress illustrate that later acts may not break the chain when they are the very risks created by the defendant’s wrongdoing.

An excellent answer will also distinguish causation from breach. Breach asks whether the defendant’s conduct fell below the required standard. Causation asks whether that breach mattered to this loss. It will distinguish damage from loss: the actionable damage in negligence is often physical injury or property damage, while financial consequences are losses flowing from that damage. It will also avoid the common mistake of using causation to smuggle in a conclusion about fairness. The law does sometimes make policy-sensitive exceptions, but they must be named, confined and justified.

The history of causation in English tort is the history of a movement from metaphysical causation towards controlled legal responsibility. Early negligence cases could speak as if causation were simply a matter of common sense. Once industrial accidents, medical uncertainty and mass exposure cases reached appellate courts, that confidence became untenable. The law needed rules capable of operating where science could not identify a single operative cause, and where several actors or processes contributed to an indivisible harm.

The orthodox starting point remains the but-for test. Its attraction is obvious. It expresses a counterfactual inquiry in simple terms and fits the compensatory structure of tort. If the claimant would have suffered the same damage anyway, compensation from the defendant would appear to overcompensate and to impose liability for a loss not caused by the defendant. Barnett v Chelsea and Kensington Hospital Management Committee is the classic classroom example. The hospital was negligent in failing to examine the claimant, but the evidence showed that he would have died even with proper treatment. Breach without causation meant no liability.

The difficulty is that but-for causation is not a universal solvent. It works where one alleged cause is tested against one outcome. It falters where there are cumulative causes, multiple sufficient causes, lost chances, or scientific uncertainty about which exposure caused an indivisible disease. Industrial disease litigation forced this issue into the open. In McGhee v National Coal Board the claimant developed dermatitis after being required to cycle home covered in brick dust because the employer failed to provide washing facilities. Medical science could not prove whether the additional period of exposure caused the dermatitis, but the House of Lords treated the material increase in risk as sufficient. The case remained controversial because it was uncertain whether it modified the evidential burden, relaxed the causal test, or merely drew an inference of causation from the facts.

Wilsher v Essex Area Health Authority reined in broad readings of McGhee. A premature baby suffered blindness after negligent excess oxygen, but several non-negligent factors could also have caused the condition. The House of Lords refused to treat a material increase in risk as proof of causation in an ordinary multiple-agent case. That restraint matters. Without it, proof of risk creation would too easily become proof of causation, converting negligence liability into insurance against bad outcomes.

Fairchild v Glenhaven Funeral Services then created the modern exceptional rule. Where an employee had been tortiously exposed to asbestos by several employers, and medical science could not show which exposure caused mesothelioma, each employer who materially increased the risk could be liable. The decision was driven by a perceived injustice: every employer had breached a duty designed to protect against the very disease suffered, yet conventional proof would leave the claimant uncompensated because of the state of medical science. Barker v Corus refined and limited that liability as several rather than joint and several, but Parliament reversed that consequence for mesothelioma through section 3 of the Compensation Act 2006.

In parallel, legal causation developed through remoteness and intervening act cases. Re Polemis had once suggested that a defendant was liable for all direct consequences of negligence, however unforeseeable. The Privy Council in The Wagon Mound rejected that approach and substituted reasonable foreseeability of the kind of damage. That shift aligns negligence with fault: a defendant should not generally be liable for a wholly unforeseeable type of harm merely because it happened to follow directly from a breach.

The modern law therefore contains two tendencies. The first is orthodox and restrictive: liability tracks proved causal contribution and foreseeable kinds of damage. The second is corrective and exceptional: where conventional proof would defeat meritorious claims in narrow settings, the courts or Parliament may adjust the rules. A Cambridge answer should expose both tendencies without collapsing them into an unprincipled discretion.

1. Start with the damage alleged. Causation cannot be answered in the abstract. The claimant must identify the relevant damage: death, personal injury, property damage, psychiatric injury, economic loss consequential upon physical injury, or some other recognised harm. The question is not whether the defendant caused something, but whether the defendant’s breach caused the legally relevant damage for which compensation is claimed.

1. The primary test is but-for causation. The court asks whether the damage would have occurred but for the defendant’s breach. The correct counterfactual is a world in which the defendant complied with the relevant duty. In a medical case, that may mean timely diagnosis, warning or treatment. In an employer’s liability case, it may mean safe equipment, protective systems, or proper supervision. The but-for test is usually applied on the balance of probabilities. A claimant must show that it is more likely than not that the breach made the relevant difference.

1. Breach and causation must be kept distinct. A finding that the defendant acted carelessly does not show that the careless conduct caused the damage. Barnett is indispensable for this proposition. Conversely, a causal connection alone does not establish liability unless duty, breach and actionable damage are independently satisfied. In problem questions, students often spend too long proving breach and then assume causation. That is fatal in medical negligence and delayed-diagnosis problems.

1. The all-or-nothing effect of the balance of probabilities is important. If it is 51 per cent likely that breach caused the harm, the claimant ordinarily recovers in full. If it is 49 per cent likely, the claimant recovers nothing. English law generally rejects damages for a lost chance of a better medical outcome in personal injury cases, as Gregg v Scott confirms. That rule is often defended by reference to certainty and coherence, but it can operate harshly where negligence deprives a claimant of a substantial but sub-probable prospect of recovery.

The leading cases form a map of recurring causal problems. Barnett is the starting point. The claimant’s husband attended hospital after arsenic poisoning but was negligently sent home. He died. The evidence showed that, even with proper care, death would probably have occurred. The case is indispensable because it separates negligence from causation. A negligent failure to treat is not actionable unless proper treatment would probably have avoided the injury or death.

Bonnington Castings supplies the core material-contribution principle. The claimant inhaled dust from two sources, one involving breach and one not. The disease resulted from the cumulative inhalation. The defendant was liable because the negligent dust made a material contribution to the disease. This is not merely risk creation: the dust was part of the causal process producing the harm.

McGhee is more difficult. The claimant could prove that the employer’s breach increased the period during which brick dust remained on his skin, and that this increased the risk of dermatitis. He could not prove that the additional exposure caused the dermatitis. The House of Lords allowed recovery. Later cases have struggled to classify the decision. It can be treated as permitting an inference of causation on the facts, or as an early version of the Fairchild exception. In examination answers, it is safest to say that McGhee bridges material contribution and material increase in risk, but that Wilsher prevents it becoming a general rule.

Wilsher is the limiting case. Excess oxygen negligently administered to a premature baby was one possible cause of blindness, but several other non-negligent causes were also possible. The House of Lords refused liability. The defendant had increased a risk, but not necessarily the risk in a situation where all possible causes were tortious exposures of the same agent. Wilsher prevents Fairchild from swallowing the ordinary burden of proof.

Fairchild is the exceptional industrial disease case. Several employers had negligently exposed employees to asbestos. The employees developed mesothelioma. Medical science could not show which exposure caused the disease. The House of Lords held that a material increase in risk could suffice. The decision is explicitly exceptional, founded on the special features of mesothelioma and the injustice of denying recovery where each defendant breached a duty aimed at the very harm suffered.

Barker limited Fairchild by making liability proportionate to risk. Parliament then reversed that limitation for mesothelioma by section 3 of the Compensation Act 2006. Sienkiewicz later confirmed the exceptional character of the Fairchild rule and rejected attempts to dilute the requirement of a material increase in risk into an ordinary statistical balance.

The Wagon Mound is the foundational remoteness authority. It rejects liability for all direct consequences and substitutes reasonable foreseeability of the kind of damage. Hughes complements it by showing that the precise manner of injury need not be foreseeable. Corr shows that an apparently deliberate act, suicide, will not necessarily break the chain where it flows from the injury caused by the defendant’s breach. Chester v Afshar remains controversial: it gives causal significance to failure to warn in circumstances where orthodox but-for causation was not fully satisfied, reflecting the importance of patient autonomy but raising concerns about coherence.

See practice questions section below.

See further reading section below.