Minimizing Autonomy Errors when Stopping Life Support

It has long been recognized that errors will be made when it comes to stopping life support.  On the one hand, some patients will have life support stopped even though they still wanted it.  This might result from, among other things: (a) an error in prognosis, (b) an error in determining the patient's preferences, or (c) a failure to follow the patient's preferences.  On the other hand, some patients will have life support continued even though they did not want it.  This might result from, among other things: (a) an error in prognosis, (b) an error in determining the patient's preferences, or (c) a failure to follow the patient's preferences.  

It struck me in Baltimore, on Tuesday, that amending the law to permit unilateral refusal probably does raise at lease the "risk" of erroneous stopping of life support.  But such an acknowledgment is not fatal to the effort.  First, significant efforts can and should be made to implement safeguards to minimize that risk.  Second, the risk of erroneous stopping is surely significantly outweighed by the benefits of reducing erroneous prolongation.  But for the latter argument to have persuasive force, the latter type of error must be perceived as serious as the former type of error.