Causation in clinical negligence is decided in two stages. Factual causation asks whether the breach of duty actually caused the injury, usually by applying the but for test on the balance of probabilities. Legal causation asks whether the law should hold the defendant responsible for that injury, looking at intervening events, remoteness and the scope of the duty the defendant owed. Claims can fail at either stage, and many fail on causation after breach has been proved. Our earlier piece on the but for test and structuring a causation opinion covers the basics. This guide covers the wider law that an expert's evidence has to serve.
The practical bottom line
- Factual causation comes first. But for the breach, would the injury have happened? The answer is given on the balance of probabilities, and it is all or nothing.
- A lost chance is not enough. A reduced chance of a better outcome is not, on its own, recoverable in personal injury (Gregg v Scott).
- Material contribution has limits. It can succeed where science cannot separate cumulative causes, and it fails where the possible causes are distinct.
- Legal causation is for the court. The expert supplies the medical facts it turns on: timing, mechanism and the effect of each cause.
What is the difference between factual and legal causation?
Factual causation is a question of fact: did the negligence make a difference to what happened? Legal causation is a question of law and policy: given that it made a difference, should the defendant answer for this particular loss? A claimant has to establish both. A late diagnosis can be a factual cause of harm while the law still limits recovery, for example because a later event broke the chain of causation, or because the harm falls outside the risk the doctor was meant to guard against.
Medical experts give evidence on the facts that both questions depend on. The court draws the legal conclusions.
How does factual causation work in clinical negligence?
The starting point is the but for test. In Barnett v Chelsea and Kensington Hospital Management Committee [1969] 1 QB 428, a casualty officer negligently failed to examine a night watchman who had drunk tea laced with arsenic. The claim failed because the evidence showed he would have died even with proper treatment.
The standard of proof is the balance of probabilities, and the result is all or nothing. In Hotson v East Berkshire Area Health Authority [1987] AC 750, a boy's hip injury was diagnosed late, and the trial judge found a 75% chance that the damage would have happened anyway. He awarded a quarter of the damages for the lost chance. The House of Lords reversed him: on the balance of probabilities the delay was not the cause, so nothing was recoverable for it. In Gregg v Scott [2005] UKHL 2 the House of Lords, by a majority, refused damages where a delayed cancer diagnosis reduced the claimant's chance of cure from 42% to 25%. A lost chance is not, on its own, recoverable damage in clinical negligence.
Omissions add a step. Where the breach is a failure to act, the court asks what would have happened if the doctor had acted properly. Bolitho, better known for its point on breach, is a causation case of this kind: the question was whether a doctor who failed to attend would have intubated, and if not, whether failing to intubate would itself have been negligent.
When does material contribution apply?
Where the medical evidence cannot show that the negligence was a but for cause, the claimant may still succeed by proving that it made a material contribution to the injury, meaning a contribution that was more than negligible. The principle comes from Bonnington Castings Ltd v Wardlaw [1956] AC 613, an industrial disease case.
Two clinical cases show how it works. In Bailey v Ministry of Defence [2008] EWCA Civ 883, a patient weakened partly by negligent post-operative care and partly by pancreatitis, which was not negligent, aspirated her vomit and suffered a cardiac arrest and brain damage. Medical science could not say which cause was decisive, and because the negligent weakness contributed more than negligibly, she recovered in full. In Williams v Bermuda Hospitals Board [2016] UKPC 4 the Privy Council took the same approach where a delay in treating appendicitis let sepsis develop, holding that it made no difference that the causes operated one after another.
The doctrine has limits. In Wilsher v Essex Area Health Authority [1988] AC 1074, a premature baby's sight loss had five possible causes, only one of them negligent. The House of Lords held that the claimant had to prove the negligence caused or materially contributed to the injury, and that adding to the risk was not enough. The special rule allowing recovery for a material increase in risk, developed for mesothelioma in Fairchild v Glenhaven Funeral Services Ltd [2002] UKHL 22, is confined to a narrow class of cases and has not been applied to ordinary clinical negligence claims.
How does causation work in consent cases?
Where the negligence is a failure to warn, the question is what the patient would have done if properly advised. Usually the claimant must show that they would have declined or deferred the treatment and so avoided the injury. In Chester v Afshar [2004] UKHL 41 the House of Lords went further. A neurosurgeon failed to warn of a 1 to 2% risk of cauda equina syndrome. The claimant could not say she would never have had the operation, only that she would have taken time to consider it. The majority allowed her claim as a narrow modification of ordinary causation principles, to protect her right to make an informed choice.
Since Montgomery the duty to warn is wider, so expert evidence on what the material risks and reasonable alternatives were carries more weight than it used to. Our guide to the Bolam test and Montgomery covers the duty side.
What is legal causation?
Legal causation asks three questions:
- Was the chain of causation broken? A new intervening act, known as a novus actus interveniens, can relieve the original wrongdoer. Later negligent medical treatment rarely does. In Webb v Barclays Bank plc and Portsmouth Hospitals NHS Trust [2001] EWCA Civ 1141, negligent advice to amputate a leg injured at work did not break the chain from the employer's original negligence, and liability was shared between the employer and the hospital.
- Is the damage too remote? The harm must be of a kind that was reasonably foreseeable (The Wagon Mound (No 1) [1961] AC 388). Its extent need not be: a defendant takes the claimant as they find them, which is known as the thin skull rule (Smith v Leech Brain and Co Ltd [1962] 2 QB 405).
- Is the loss within the scope of the duty? In Khan v Meadows [2021] UKSC 21, a mother who asked whether she carried the haemophilia gene was negligently led to believe she did not. Her son was born with haemophilia and, separately, autism. She recovered the extra costs of his haemophilia, but not those of his autism, because the advice concerned the risk of haemophilia only.
What should a causation expert put in the report?
- The counterfactual, in full. What non-negligent care would have looked like, when it would have happened, and what the outcome would probably have been.
- Timing. Most clinical causation disputes turn on when: when the condition became treatable or untreatable, and how long each step of proper care would have taken. Tie every time point to the records.
- Probability in the court's terms. Say whether an outcome was more likely than not, and explain how any population statistics apply to this patient, including where they do not.
- Each cause and its effect. Where there is more than one cause, say what each contributed, whether the contributions can be separated, and whether the injury is divisible, meaning it worsens with dose or duration, or indivisible. The law treats the two differently.
- The medical facts, with the legal conclusions left to the court. Whether the chain was broken or the loss falls within the scope of the duty is for the judge. The expert explains the mechanism and sequence the judge needs in order to decide.
Most causation disputes turn on a single question the records can answer: when. When did the window for treatment close, and would proper care have arrived before it did?
Whatever the conclusion, the report has to meet Part 35, including a summary of any range of opinion. Our guide to CPR 35 sets out the requirements.
How ALLDOQ fits
A causation opinion is only as strong as its timeline. ALLDOQ builds a chronology from the bundle with every entry linked to its source page, and ALLDOQ Verify summarises records with a citation for each statement, so the time points a causation argument rests on can be checked in one click. Book a walkthrough to see it on a real case.
