Can you really predict a prisoner's death?
Last August, two of the UK's most notorious prisoners were released from prison.
They were Great Train Robber Ronnie Biggs and Libyan bomber Abdelbaset Ali al-Megrahi and in each case the short-term prognosis was poor.
Biggs, who was severely ill with pneumonia, was released by the then Justice Secretary Jack Straw after being told he was unlikely to recover.
He is still alive but is reportedly in very poor health, following a series of strokes and heart attacks. The former gangster is now living in a care home.Time frame
Megrahi was released from jail on compassionate grounds after medical reports suggested he had about three months to live from prostate cancer.
He too is still alive, although there are no current reports about his state of health.
So how could doctors get it so wrong?
Leading oncologist Professor Karol Sikora, who examined Megrahi, said one of the big problems with predicting death is that it is impossible to be precise.
"There are no set rules," he said.
"There is nothing written anywhere that says you have to die within a set time frame and, with cancer, the difficulty is that what we call the survival curve is so indistinct to start with.
End Quote Dr Theodore Dalrymple
If you don't release them and they die you can be regarded as unduly harsh”
"On the balance of probability you would say that both those customers would have died within three months, but you can't be absolutely sure.
"They do not have to die and what I find very disheartening is people wanting to put them back in prison because they have managed to survive."Appeal grounds
He said he thought these high-profile cases and the media reaction to them might make doctors more cautious in future about committing to a firm survival rate.
In England and Wales there are two possible grounds for compassionate release - a prisoner's medical condition and "tragic family circumstances".
Prison Service rules state: "Early release may be considered where a prisoner is suffering from a terminal illness and death is likely to occur soon. There are no set time limits, but three months may be considered to be an appropriate period."
In Scotland, the regulations are set out in the Prisoner and Criminal Proceedings (Scotland) Act 1993, which also says a life expectancy of less than three months would appear appropriate.
Reggie Kray - serving life for murder - was released in August 2000, when it became clear he had inoperable bladder cancer. Others, such as Moors murderer Myra Hindley, died in jail.
However, compassionate release is not a power that is used very often. Last year's figures show that 48 people were granted permanent early release in England and Wales over five years, with no figures available for the number of applications rejected.
In Scotland, the 2010 figures show 26 prisoners have been released on medical grounds in the last 10 years, with seven requests denied.
Former prison doctor Theodore Dalrymple said that, because medicine 'is not an exact science', doctors can be left looking "silly".
"If you don't release them and they die you can be regarded as unduly harsh, but if you do release them and they go on living for a long time then you look a fool."Doctors not deciding
Dr Emma Cuzner, medico-legal adviser for the Medical Defence Union said making a prediction in such cases was difficult.
"Of course, anticipating the course of an illness for any patient is fraught with difficulty and, in our experience, doctors are much more likely to give a general indication of survival rates for a particular condition than to give a definitive prediction of life expectancy.
"It's important to stress that doctors are not responsible for deciding whether or not a prisoner should be released early on health grounds.
"They can only be asked to give their medical opinion about a prisoner's condition which the authorities can refer to when making a decision. "