|
Activity Number:
|
203
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2003 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
ENAR
|
| Abstract - #300894 |
|
Title:
|
Modeling Health Trajectories Before and After a Sentinel Event
|
|
Author(s):
|
Qian-Li Xue*+ and Karen J. Bandeen-Roche and Cynthia Boyd
|
|
Companies:
|
The Johns Hopkins Center on Aging and Health and Johns Hopkins University and The Johns Hopkins Medical Institutions
|
|
Address:
|
2024 E. Monument St., Baltimore, MD, 21205-2217,
|
|
Keywords:
|
aging ; discontinuity ; longitudinal data ; multivariate outcome
|
|
Abstract:
|
Understanding change in health is important for designing interventions to ameliorate decline, while maintaining good function. However, modeling trajectories of health for older adults, particularly around the time of serious health events, presents at least four challenges: (1) characterizing individual and contextual differences in health change due to increased heterogeneity with age in terms of both life experience and host characteristics; (2) flexibly modeling the potential trajectory shift after a health event; (3) analyzing changes in health before and after the event; and (4) accounting for missing data due to skipped visits. We discuss and compare two approaches: a conventional growth curve model that characterizes patterns of change before and after the event separately and a joint analysis of pre- and post-event trajectories using a multivariate random effects model that addresses the challenges described above.The methods are applied to the Women's Health and Aging Study to characterize changes in physical function before and after hospitalization. It aims to guide inpatient and post-discharge care to reduce the risk of further decline in functional status.
|
