The excellent review by Hanson and colleagues [1] reveals a striking omission in interventions designed to change care at the end of life. All the studies cited aimed to increase communication about the patient's preferences for care or to reduce the use of certain types of interventions. None apparently offered the patients or their caregivers meaningful and realistic alternatives to usual hospital care.

Terminal illness often brings with it disability; pain; and the need for assistance with eating, bathing, dressing, and toileting. In our society, the options for meeting these needs are very limited. Hospice care is not available in all communities; in areas where it is available, most is provided as home care. The hours of reimbursable professional services are limited; thus, unless relatives or friends can provide considerable personal care, hospice care at home is often not an option. Nursing homes are sometimes used for terminal care. However, very little care in nursing homes is reimbursible by third-party payers, and nursing home personnel are not usually well trained to provide good management of pain or dyspnea. Counselling for the emotional issues of the dying is rarely provided in this setting. Because of these limitations, patients whose terminal illness becomes too painful, frightening, or messy to be dealt with easily at home often come to acute hospitals for terminal care. When a patient is admitted to the hospital, there is immediate pressure, because of reimbursement policies, to discharge the patient quickly if aggressive interventional care is not provided.

Even if a patient chooses to have "comfort care only" at the end of life and expresses his or her wishes clearly to a physician, our health care system puts up substantial barriers to receiving such care. Until we can offer more practical resources for palliative care, we are unlikely to change how patients or their physicians behave at the end of life.