We rely on them to know what's best for us, but many doctors have lied to their patients at times. Can such behaviour ever be justified? Daniel K Sokol explores one of the great, unspoken dilemmas of the medical profession

The Italian poet Dante is not kind to deceivers. He condemns them to the very depths of hell, where they are whipped by horned demons and swim in the excrement they spouted to others whilst on earth. But do all deceivers deserve such punishment, in particular doctors who deceive patients with their best interests at heart? Are there instances when a doctor's deception is virtuous, rather than wicked? The use of deception in medicine is a sensitive subject. No one denies that it happens, but few are willing to tackle the issue full on.

Two years ago, an 18-year-old girl was admitted to a Canadian hospital, wishing to donate a kidney to her father. Despite six months on haemodialysis, her father's condition had declined rapidly and a kidney transplant was his last chance. Blood tests were requested to determine whether the two were compatible. When the results arrived, doctors discovered that the father and daughter were not biologically related, although they were compatible. The doctors were faced with an unsettling moral dilemma: should they tell their patients the truth or simply conceal the information?

In this instance, arguments can be found to support both disclosure and non-disclosure. On the one hand, informing the patients might destroy a close-knit family, dissuade the daughter from donating the life-saving kidney or even anger the patients who, after all, never asked to know this. On the other hand, doctors might invoke a duty to inform, or argue that the truth will out eventually and thus undermine the trust in the hospital and doctors generally. Finally, those in favour of disclosure could play one of the strongest cards in the medical ethics deck: the respect for patient autonomy. People are autonomous if they can make decisions based on their own beliefs, free from coercion and misleading information. In this case, the daughter is not making an autonomous decision as she clearly holds a false belief - that she is biologically related to her father. On this view, the doctors ought to correct the daughter's false belief by telling her the truth, thereby allowing her to make an informed decision.

In the end, the doctors agreed to inform the father and daughter of their unexpected finding. Although shaken by the revelation, the patients were glad that they were told, and the kidney transplant went ahead as planned.

The case just mentioned is rare, but situations involving deception are common in medical practice. There is a mismatch between the advice of the General Medical Council, which instructs doctors to be "honest and trustworthy", and what goes on in the private consultation rooms of GPs and in hospital wards. At times, doctors do not tell elderly patients that they have terminal cancer. "The relatives beg us not to tell their darling mother that she has cancer, insisting that she'll just fall apart, and usually we comply," says a senior palliative care registrar who works in a major London hospital. Even in teaching hospitals, some doctors encourage students to use deception. In last April's issue of theBMJ, a medical student wrote to the editor: "For the second time our student group has been advised to lie when reporting patient information." Last January, the BMJ revealed that a quarter of rectal and vaginal examinations on anaesthetised patients were performed by medical students without patient consent. When asked about the practice, a fourth-year medical student commented: "I was told in the second year that the best way to learn to do rectal examinations was when the patient was under anaesthetic. That way they would never know." Some doctors, to evade the awkward questions of patients, use carefully crafted language. One cancer specialist, when asked by a patient how long she had to live, replied: "We deal with the living here." The patient interpreted this as several months, possibly years. A few days later, the patient met another doctor and asked him the same question. When the doctor informed her that she had only days to live, the patient was distraught. She died that same day.

Last week, a medical student on a hospital placement told me that one of the surgeons introduces his students as "junior doctors" to patients. Deceiving patients in this way erodes public trust in the medical profession, and undermines the laudable work of those other doctors who hold in higher esteem the relationship between doctors and patients. If the duty to tell the truth is not set in stone, it should at least be the default option.

Although in some cases deception is clearly wrong, there are times when it seems justified. The lies and deceptions of doctors need not lead to the lower circles of Dante's "Inferno". What if a patient, the sole survivor of a fatal car crash, asks on her expiring breath about the fate of her young child who was also in the car? Should the doctor say that her daughter suffered gruesome injuries and died a slow death, or can he or she lie to reassure his dying patient? Here the desire to comfort a patient in her final moments arguably trumps the respect of a fleeting autonomy. What about lying to family members who enquire about the death of their loved one? "We often lie to relatives when the patient died horribly, especially if the relatives are clearly very distressed," comments the palliative care registrar. "Today, for example, a patient died really badly, vomiting litres of blood, and I told her relatives that she passed away peacefully in her sleep. I don't think it's wrong to lie in such cases." What about a patient with spinal injuries who, on the way to the operating theatre, asks the doctor whether he will ever walk again? Or the elderly patient, diagnosed with incurable metastatic cancer, about to go on a long-awaited holiday to Barbados? Can the doctor wait until the patient's return before breaking the bad news? Right or wrong, more research needs to be done on the use of deception in medicine, on its prevalence, its justifications, and in particular the beliefs of doctors, patients and members of the public on the matter. Only then can we aim to improve current practice and policy in ways that would be acceptable to all parties involved.

Daniel Sokol is an instructor in medical ethics at the Alliance for Lifelong Learning at the University of Oxford. He is also conducting doctoral research on truth-telling and deception in medicine at Imperial College, London, funded by the Wellcome Trust (daniel.sokol@talk21.com)