Trends among gynecologic oncology inpatient deaths: Is end-of-life care improving?

Objective. The objective was to describe trends over time in key factors surrounding end-of-life care on a gynecologic oncology service at a tertiary cancer center. Methods. Patients with gynecologic cancers who were hospitalized and died at our institution between 1992 and 1997 were identified using institutional databases. The study group consisted of 176 patients. For analysis, patients were divided into two groups: those who died between 1992 and 1994 and those who died between 1995 and 1997. Data were abstracted from medical records regarding the interval between placement of do-not-resuscitate (DNR) orders and death; the interval between the patient's being informed of her terminal-stage disease and death; the types of interventions performed near the end-of-life; and other factors related to end-of-life care. Results. The average patient age on last admission to the hospital was 56 years, and the most common disease sites were ovary (47%), cervix (30%), and uterus (17%). Most deaths (82%) occurred on medical or surgical units, with the remainder occurring in the intensive care unit (12%) or emergency room (6%). The average interval between placement of DNR orders and death was longer among patients who died in 1995-1997 than among patients who died in 1992-1994 (49.6 days vs 19.2 days, P = 0.027). The average annual number of deaths (42.0 vs 16.7, P = 0.061) and the average length of hospital stay (13.3 vs 8.8 days; P = 0.079) decreased between 1992-1994 and 1995-1997, but the differences did not reach statistical significance. Changes between patients who died in 1992-1994 and those who died in 1995-1997 in the interval between the last cycle of chemotherapy and death (87 days and 49 days, respectively; P = 0.29), the proportion of patients with a DNR order on admission (39 and 45%, respectively; P = 0.59), the proportion of patients admitted for terminal care only (22 and 19%, respectively; P = 0.47), and the proportion of patients who died awaiting transfer to hospice (32 and 22%, respectively; P = 0.24) likewise did not reach statistical significance. Conclusions. Our data indicate that discussions about DNR orders are occurring earlier in relation to terminal events. However, we could not detect significant changes in the other outcome measures we studied. Major opportunities remain for further enhancements in the realm of advance planning for end-of-life care. Educational opportunities should be offered to physicians regarding communicating with patients about disease progression and end-of-life decision-making.