The Last 2 Years of Life: Functional Trajectories of Frail Older People


Address correspondence to Kenneth Covinsky, MD, MPH, San Francisco VA Medical Center (111G), 4150 Clement Street, Bldg 1, San Francisco, CA 94121. E-mail:


OBJECTIVES: To characterize the functional trajectories during the last 2 years of life of patients with progressive frailty, with and without cognitive impairment, and to assess whether it was possible to identify discrete functional indicators that signal the end of life.

DESIGN: A retrospective analysis of functional trajectories during the last 24 months of life.

SETTING: Twelve demonstration sites of the Program of All-inclusive Care for the Elderly (PACE). PACE cares for frail older people who meet criteria for nursing home placement, with the goal of keeping the patient at home.

PARTICIPANTS: Nine hundred seventeen patients who died while enrolled in PACE.

MEASURES: At PACE entry and every 3 months thereafter, data were collected about the degree of dependence (none, partial, or full) in bathing, eating, and walking and the degree of incontinence (none, bladder, or bowel). Cognitive impairment was defined as six or more errors on the Short Portable Mental Status Questionnaire. To describe the end-of-life trajectories of patients, data were analyzed from observational windows of time, beginning with the patients' dates of death and extending backward in time to 24 months before death. Each analytical window was 3 months in duration. For each of the functional measures, the probability of functional deterioration in the last 2 years of life in patients with (64%) and without (36%) cognitive impairment was also compared.

RESULTS: The mean age at death was 84; 69% of patients were women. For patients with and without cognitive impairment, a prolonged, steady increase in the rates of functional dependence that were evident at least 1 year before death, rather than sudden increases in functional dependence shortly before death, characterized the functional trajectories. It was not possible for any of the four measures to detect a time point before death at which there was an abrupt decline in function likely to signal impending death.

For each measure, patients with cognitive impairment declined earlier, were more likely than patients without cognitive impairment to have the maximal level of dependence in the 0- to 3-month window before death (e.g., 56% vs 30% for mobility, P < .001), and were more likely to decline in the 2 years before death (e.g., 56% vs 36% for mobility, P < .001).

CONCLUSION: Patients with advanced frailty, with or without cognitive impairment, have an end-of-life functional course marked by slowly progressive functional deterioration, with only a slight acceleration in the trajectory of functional loss as death approaches. Patients with cognitive impairment have particularly high rates of functional impairment at the time of death. These results suggest that end-of-life care systems that are targeted toward patients with functional trajectories clearly suggesting impending death (such as the Medicare hospice benefit) are poorly suited to older people dying with progressive frailty.