Observations Ethics Man

“The patient would have died anyway”

BMJ 2013; 346 doi: https://doi.org/10.1136/bmj.f285 (Published 15 January 2013) Cite this as: BMJ 2013;346:f285
  1. Daniel K Sokol, medical ethicist and barrister, Inner Temple, London
  1. daniel.sokol{at}talk21.com

Determining causation of harm is crucial in establishing whether medical negligence has occurred

William Barnett was a nightwatchman at the Chelsea College of Science and Technology in London. On the morning of 1 January 1965 he and two colleagues had tea. Twenty minutes later they started to vomit. They drove to the nearby hospital, where they were seen by a nurse. The nurse spoke to a doctor on the phone, who advised the men to “go home and call in their own doctors.” They left the hospital. A few hours later Barnett was rushed to the hospital and died from arsenic poisoning. His widow sued the hospital for negligence.1

The court found that the doctor failed in his duty of care. He should have examined the patient. Yet the claim failed because it could not be shown that Barnett would have survived even with proper care. The doctor was ethically culpable, but to establish negligence in law the widow had to prove, on the balance of probabilities, that the doctor’s breach of duty caused her husband’s death. In medicine this can prove challenging, as patients are often unwell even before they visit a doctor. Was it the doctor’s negligence or the pre-existing condition that caused the damage?

No doubt the doctor in the Barnett case told himself what I have heard some doctors say after fatal mistakes: “The patient would have died anyway.” This can lead to non-disclosure of the error. Prognostication, a highly fallible exercise one minute, suddenly becomes, in the doctor’s mind, an exact science. Yet ethically it is wrong for the doctor who made the mistake to decide what would have occurred “but for” the error. Not only might there be a lack of necessary expertise, but the doctor would hardly be an impartial judge. No one should be a judge of his own case. This is why there are lay people on the disciplinary panels of the General Medical Council.

The doctor who justifies non-disclosure by saying that the patient would have died anyway falls foul of the General Medical Council’s Duties of a Doctor, which requires doctors to “be honest and open and act with integrity” and “never to abuse your patients’ trust in you or the public’s trust in the profession.”2

The doctor in the Barnett case, on hearing the nurse’s account, could not believe that a cup of tea caused the vomiting. But what is causation? The complexities of causation were exposed in the impassioned debates on the link between smoking and cancer in the 1950s and beyond. It is only in the past 10-15 years that tobacco manufacturers have publicly admitted that smoking causes cancer and other diseases.3

Philosophy has also grappled with causation. The editors of the Oxford Handbook of Causation note in their introduction that, in spite of the best efforts of philosophers, “there is still very little agreement on the most central question concerning causation: what is it?”

The law has developed an increasingly sophisticated—some say confusing—approach to causation, and causation arguments continue to appear before judges in the highest courts.

Readers may have heard of the “but for” test of causation in law: would the injury have occurred “but for” the defendant’s breach of duty? If a general practitioner fails to refer a patient in time, and that patient requires an amputation, a key question is whether a timely referral would have made any difference. This scenario would require an expert in general practice to establish whether the GP was negligent (“Would a reasonably competent GP have referred at that particular time?”)4 and perhaps a separate orthopaedic expert to deal with the causation issue (“What would have been the patient’s chances of avoiding an amputation if he had been referred by the GP on time?”).

The law adopts a black and white approach to hypothetical scenarios. If the chances of an event occurring are more than 50%, then it would have happened. In Gregg v Scott,5 Dr Scott negligently assessed a cancerous lump as benign. As a result Mr Gregg’s treatment for non-Hodgkin’s lymphoma was delayed by nine months. The cancer had spread, and his prospects of recovery dropped from 42% to 25%. The House of Lords ruled, controversially, that because the initial prospects were less than 50% the outcome would have been the same. He would not have survived either way. The claim failed on causation.

The “but for” test was modified in the case of Bailey v Ministry of Defence.6 Ms Bailey was admitted to hospital for a gall stone operation. She was treated negligently during her postoperative stay at hospital, became extremely weak, and developed pancreatitis. The pancreatitis was unrelated to the negligent care. She was transferred to another hospital, where she aspirated on her vomit, had a cardiac arrest, and sustained brain damage. The judge found that the patient’s extreme weakness, caused by the negligence of the first hospital, materially contributed to her inability to protect her airway from the vomit and to the subsequent hypoxic injury. This was sufficient to establish causation.

To recap my last three columns on clinical negligence, doctors cannot be liable in negligence unless (i) they have a duty of care to the patient, (ii) they have breached that duty, and (iii) the breach caused or materially contributed to the patient’s injury (BMJ 2012;345:e7858, doi:10.1136/bmj.e7858; 2012;345:e6804, doi:10.1136/bmj.e6804).


Cite this as: BMJ 2013;346:f285


  • Many thanks to Ronald P Sokol for helpful comments on an earlier draft.


View Abstract