Case finding in palliative care: development and evaluation of a resident prognostication tool

Steckart MJ, Sverdlik A, Riopelle D, Wagner G, Rosenfeld KE. Case finding in palliative care: development and evaluation of a resident prognostication tool. Paper presented at: VA HSR&D National Meeting; 2006 Feb 1; Arlington, VA.

Related HSR&D Project(s)

• IIR 02-294 – Randomized Trial of Care Management to Improve End of Life Care

Objectives:Improving end-of-life care is of critical importance to the VA as it faces an increasingly aging and dying veteran population. Currently only a minority of patients with palliative needs receive palliative services, partly due to the difficulty of identifying patients appropriately early in the course of an ultimately-fatal illness. Although various methods for patient identification have been proposed, none has been rigorously, prospectively tested for accuracy. The purpose of this study was to evaluate a tool utilizing medical resident prognostication to identify patients at high risk of dying, who might benefit from additional palliative services.Methods:A prognosis rating scale was implemented to assign one-year mortality risk for patients admitted to the General Medicine inpatient service. Residents estimated one-year mortality on a 6-point scale ranging from 0 ('minimal' / < 5 percent) to 5 ('terminal' / > 75 percent). Primary diagnoses were also collected for those patients with a greater than 25 percent mortality risk. One-year survival data was tabulated and an accuracy scoring system applied to measure resident accuracy based on prognostic category and diagnosis. Results:Prognostic data was collected on 1055 of 1120 admissions (94%) over 158 admission days. At one year follow-up, overall survival was 71%. Actual one-year mortality based on prognostic category showed that 9% (n = 173) of patients rated as having a minimal risk of death (ROD) were deceased; 18% (n = 329) of patients with low ROD were deceased; 29% (n = 194) of patients with moderate ROD were deceased; 39% (n = 139) of patients with high ROD were deceased; 44% of patients rated as having a very high ROD were deceased; and 61% (n = 98) of patients rated as terminal were dead at one year follow-up. Prognostic estimates for cancer diagnoses were found to be more accurate than for non-cancer diagnostic categories.Implications:Resident prognostication is a reasonable method to identify patients at risk of dying in the intermediate term. Impacts:Resident prognostication can be a valuable tool to identify patients for additional supportive care services to improve the quality of end-of-life care.