A blind teenager with a brain tumour is at the centre of a UK court case that pits the hopes of his parents against medical opinion.
In February, doctors argued that the 18-year-old had no more than two weeks to live and that active treatment including chemotherapy and brain surgery would be futile. If his heart were to stop beating, he should not be resuscitated.
The high court authorised the unnamed hospital trust to discontinue treatment against his parents' wishes. But the teenager is still alive, more than three months later. His mother says: "I am fighting for my child's life. He is a sick child, but he is coping. He is not dying."
It's difficult to predict exactly when a patient is going to die, or, sometimes, if they are going to die at all.
The convicted Lockerbie bomber Abdelbaset al-Megrahi , diagnosed with prostate cancer, was freed on compassionate grounds after doctors predicted he had less than three months to live. Megrahi died two years and nine months after his release.
One of the largest reviews, published in the British Medical Journal, systematically reviewed survival predictions in terminally ill patients with cancer. Eight studies were analysed in three countries over 30 years.
Overall, doctors' predictions were correct to within one week in 25% of cases, correct to within two weeks in 43%, and correct to within four weeks in 61%. The study found that doctors tended to overestimate survival.
The very measure of a doctor lies in their predictive abilities, their grasp of the crystal ball: "How long have I got, doctor?" The Corpus Hippocratum of early Greek medicine underlined just that: "I hold that it is an excellent thing for a doctor to practise forecasting. For indeed, if he discover and declare unaided by the side of his patients their present, past and future circumstances, he will be able to inspire greater confidence that he knows about illness, and thus people will decide to put themselves in his care."
Why is it so difficult to prognosticate?
Every patient is different, every disorder is different, every disorder within a disorder is different. People are unpredictable, their illness even more so. But there exist other subtleties that are harder to admit to.
It takes experience to know that sometimes you don't know
In my first week as an intern, I spoke to the family of an 85-year-old patient, Nora. She lay gasping, racked by sepsis, her skin bruised from intravenous drips, her legs swollen from heart failure, her consciousness clouded from all of it.
"How long has she got, doctor?"
"We'll be lucky if she's here in the morning," I replied.
We were indeed lucky the next morning. We were also lucky for the next week and the month after that, at which point Nora went home, happy and healthy.
It takes experience to know that sometimes you don't know.
Assuming the diagnosis is right in the first place, prognostication is further complicated by medical advances - the very advances designed to improve prognosis. Algorithms used following cardiac arrest are no longer universally valid. Walk into an intensive-care unit and you might see a comatose patient swaddled in cooling blankets, surrounded by ice packs, a drip running cold fluids through their veins.
Therapeutic hypothermia aims to protect the brain from a lack of blood flow. Before this technique, neurologists could start to ascertain prognosis in a comatose patient even by day three. Watch if the pupil dilates with light, see if the eye blinks as a wisp of cotton wool touches the cornea. But for comatose patients who have been rendered hypothermic, everything changes. Dropping core temperature by just five degrees entirely alters the brain's reaction to these tests. The old algorithm falls apart. A new one takes its place. Don't pull the plug.
Surgeon and author Dr Atul Gawande writes about how even the duration of dying has changed: "As for last words, they hardly seem to exist any more. Technology sustains our organs until we are well past the point of awareness and coherence. Besides, how do you attend to the thoughts and concerns of the dying when medicine has made it almost impossible to be sure who the dying even are? Is someone with terminal cancer, dementia, incurable congestive heart failure dying, exactly?"
Professor Dominic Wilkinson, director of medical ethics and consultant neonatologist at the University of Oxford, offers an example of patients who seemingly defy the odds: "Imagine a doctor estimates that there is a 99% chance that a patient will die within a short period of time. The patient returns some time later - still very much alive - and says to the doctor: 'You were wrong.'"
Wilkinson says to look beyond the 99%. "Based on the doctor's prediction, we would expect one out of 100 patients to survive for longer. Even if there is only a one-in-a-million chance of prolonged survival, there will still be very occasional patients who live longer. Those patients have not 'defied predictions' - they are exactly what was predicted."
Stephen Jay Gould, himself a scientist, was diagnosed with an abdominal mesothelioma at the age of 40. He faced a median survival of eight months.
But then Gould looked at the word "median" and looked again. This did not simply mean he probably had eight months left to live. No, it meant 50% of patients might live longer than eight months.
Mesothelioma statistics vary from many other types of cancer. Whilst most patients do have a life expectancy close to the median, a very small number do far better. Could he be one of them? In an essay titled The Median Isn't the Message, he wrote : "I looked at the mesothelioma statistics quite differently - and not only because I am an optimist who tends to see the doughnut instead of the hole, but primarily because I know that variation itself is the reality. I had to place myself amidst the variation."
Gould was able to place himself as someone who might survive for a reasonable amount of time because all the features of his specific presentation (his symptoms, his investigation results) suggested that was the case. He went on to have surgery and experimental chemotherapy. He died 20 years later from lung cancer, unrelated to his original mesothelioma.
Not all survival curves for cancer look like the one for mesothelioma. But there are always outliers who are said to have beaten the odds when really there was a small chance of survival all along.
Science writer Penny Sarchet , in a Wellcome award-winning essay , begins by asking if just telling a man he has cancer can kill him? In 1992, the Southern Medical Journal described a case of a man diagnosed with liver cancer and given just months to live. After his death, an autopsy showed that his tumour had not grown or spread. His doctor wrote: "Could it be that, instead of the cancer, it was his expectation of death that killed him?"
It's a fascinating question. How can we work fighting spirit into our prognostic algorithms. Is it our inability to do so that makes us poor forecasters?
In AD200, Galen hypothesised that melancholic women were predisposed to breast cancer and, in 2003, depression was reported to predict mortality in heart disease. Of course, someone with depression might be less likely to engage with treatment, or to address risk factors such smoking or lack of exercise.
But studies linking psychological state to prognosis are fraught with methodological issues, false-positive results, reporting of associations rather than causality, and manipulation of statistics.
Having a positive attitude convincingly helps quality of life and resilience in chronic illness. The power of those pink wristbands, the spirit of Movember, the support groups fuelled by coffee or soaked in wine - these are not to be underestimated. But the effect of positive attitude on fortitude has sadly been extrapolated to survival.
Could it be that, instead of the cancer, it was his expectation of death that killed him?
Among others, a 2007 study of more than 1,000 patients with cancer showed no impact of emotional wellbeing on survival. The author, James Coyne, later wrote : "We urge positive psychologists to rededicate themselves to a positive psychology based on scientific evidence rather than wishful thinking."
Psychologists from the University of British Columbia also did not hold back : "It has become accepted in popular culture that cancer patients need to maintain a positive attitude to heroically defeat cancer. [The] magnitude of the effect of depression on mortality does not seem to warrant the assignment of responsibility and blame to cancer patients."
The impacts of over- or underprognostication are all too clear: the parliamentary and health service ombudsman has reported that "too many people are dying without dignity".
Failure to recognise that the end might be near can lead to patients being denied palliative care, denied a compassionate end, denied the chance to say goodbye to their loved ones.
Conversely, delivering an unnecessarily grim prognosis (or, at least, not thinking beyond the median) could result in patients not receiving ongoing treatment that could prolong their survival.
When Stephen Jay Gould looked at that survival curve, his world shifted: "I had obtained, in all probability, the most precious of all possible gifts in the circumstances - substantial time. I didn't have to stop and immediately follow Isaiah's injunction to Hezekiah: 'Set thine house in order for thou shalt die, and not live.' I would have time to think, to plan, and to fight."
Professor Jane Plant, emeritus professor of applied geochemistry at Imperial College London, underwent several operations, 35 sessions of radiotherapy treatments and multiple rounds of chemotherapy, but in 1993 her breast cancer returned for the fifth time. She had two young children and needed to know what lay ahead. The first doctor she spoke to was circumspect about her chances but another speculated that she could have only a few months left.
She is still alive more than 20 years later and attributes her survival to an integrated approach that combined conventional chemotherapy and an overhaul of her diet. How does she feel about the doctor who said she had months to live? "I feel no bitterness," she says. "Doctors give us time frames so that we can put our affairs in order. That doctor was just trying to be kind."
The Megrahi case highlights consequences that sweep their way far beyond the family circle. Professor Karol Sikora had been director of cancer services at Hammersmith hospital in the early 1990s and chief of the World Health Organisation's cancer programme later that decade. He was asked by the Libyan Ambassador in London to assess Megrahi ahead of an appeal for release.
Sikora concluded that Megrahi would likely die within three months. The Scottish government denied that evidence from three doctors paid by the Libyan government (including Sikora) influenced its decision. But their evidence was shared with the Greenock prison doctor who submitted the final report.
The prediction made headlines. Speaking to the Observer in 2010, Sikora said : "What I find difficult is the idea I took the key and let him out. I provided an opinion, others provided an opinion, and someone else let him out. That decision of compassionate release is nothing to do with me. No one asked me: 'Should we let him out?' All they said was when do you think he will die?"
Sikora now attributes Megrahi's survival to his subsequent treatment: "[Megrahi] had everything including several expensive drugs still not available in the NHS when he was released back to Tripoli." Abiraterone, a drug Megrahi received in the Libyan capital, has since been made available on the NHS for men with advanced prostate cancer.
Sikora wrote in 2012 : "I feel very sorry for any distress the role of the various doctors' opinions, including mine, may have caused to the families of the victims of the Lockerbie bombing. Maybe now, 23 years on, they can close a very painful and emotionally fraught chapter in their lives."
Has the Megrahi experience changed how Sikora prognosticates, or, indeed, how he delivers that prognosis. "Absolutely. I've had people living with metastatic cancer for many years. I am much more cautious now."
Meanwhile, an 18-year-old with a brain tumour is alive longer than expected, his prognosis deemed futile. Could he have had chemotherapy, surgery, experimental drugs in the past three months? Or has he been spared the potential burden, harm and indignity of aggressive and needless interventions.
His mother's words are heartbreaking: "Give him a chance. Maybe chemo will work. If [he] goes, he goes. But he may not. He may have months to live. [He] may not be able to do martial arts but he talks, he walks. He is a pleasure to be with. He is a miracle child and has survived things you thought he would not. He deserves a chance."
His case will now go through the court of protection. Complex questions will be asked and hopefully answered. His story will go beyond the binary, outside the algorithms, away from the median. Increasingly, we know what we don't know.