Date of Award


Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)

First Advisor

Elizabeth Bradley


The purposes of this study were to evaluate the frequency of documented prognosis discussions among terminally ill cancer patients, to identify correlates of having documented prognosis discussions, and to describe the content of prognosis discussions as documented in patient medical records. Sample data were collected from the randomly selected medical records of inpatients (n=210) aged 65 years or older and admitted with diagnoses of brain, pancreas, liver, gall bladder, or inoperable lung cancer from six large Connecticut hospitals. A standardized instrument was used to extract data concerning patient demographics, hospital course, prognosis discussions, and evidence of advance care planning. Prognosis discussions were recorded in 79 (38%) of medical records and were correlated with emergency admission status (p=0.004) and longer length of hospital stay (p=0.003) on multivariate analysis. Of the documented prognosis discussions, 63% were within one week of admission but after the first day, and 57% included the patient, 76% included the family, 77 % included the doctor, and 69% did not include another health staff member (n=79). Life sustaining treatment discussions and DNR orders were both associated with prognosis discussions (p=0.001 and p=0.001, respectively) and were more often documented after the prognosis discussions. Prognosis discussions included planning for care and treatment in 33 (42%) of discussions documented. In conclusion, we found that prognosis discussions were infrequently documented during the hospitalization of terminally ill patients diagnosed with cancer. We also found that advance care planning, such as discussions of life sustaining treatment and DNR orders, was significantly associated with prognosis discussions and more often occurred after prognosis was discussed.