CPR or DNR? End-of-life decision making on a family practice teaching ward

OBJECTIVE To determine the proportion of patients on a family practice ward who had "code status" orders and end-of-life discussions documented on the ir charts in the first week of admission. To examine the correlation betwee n a tool predicting the likelihood of benefit from cardiopulmonary resuscit ation (CPR) and actual end-of-life decisions made by family physicians and their patients. DESIGN Cross-sectional descriptive study using a retrospective chart review . SETTING A 14-bed teaching ward where family physicians admit and manage the ir own patients in an urban tertiary care leaching hospital. PARTICIPANTS Patients admitted to the ward for 7 or more days between Decem ber 1, 1995, and August 31, 1996. MAIN OUTCOME MEASURES Frequency of documented "do not resuscitate" (DNR) or "full code" orders and documented end-of-life discussions. Prognosis-after -resuscitation (PAR) score. RESULTS In the 103 charts reviewed, code status orders were entered within 7 days for 60 patients (58%); 31 were DNR, and 29 were full code. Discussio n of code status was documented in 25% of charts. The PAR score for 40% of patients was higher than 5, indicating they were unlikely to survive to dis charge from hospital should they require CPR There was a significant associ ation between PAR scores done retrospectively and actual code status decisi ons made by attending family physicians (P<.005). CONCLUSIONS End-of-life discussions and decisions were not fully documented in patients' charts, even though patients were being cared for in hospital by their family physicians. A PAR score obtained during the first week of admission could assist physicians in discussing end-of-life orders with the ir patients.