Currently, to help meet the expanded provisions for long term care mandated by the Millennium Act, VHA has used a simple planning model that has strengths, but many significant weaknesses, including biased measures of disability, ecologic associations of disabilities and resource use, and assumptions of uniformity among groups of elderly in the relationships between disability and long term care service needs.
We aim in this observational,decision-modeling study to construct a Long Term Care Healthcare Consumption Model based on an estimated 9100 veterans followed over the past 20 years in the National Long Term Care Survey which we hypothesize will produce more accurate LTC demand estimates at the national, medical center, and VISN-levels than the current LTC model.
To construct the dataset for these analyses, we will link the National Long Term Care Survey and the linked Medicare files for 1982-2001 with VA files that characterize the health and LTC services provided by VHA. Due to VHA data limitations, we'll limit the resource component to the 1994 and 1999 rounds of the NLTCS (resource consumption for years 1994-2001. We will test model validity, and compare accuracy of an updated version of the current model with the GoM-based Long Term Care Healthcare Consumption model using 3 additional data sources: the AHEAD/Health and Retirement Survey (6700 veterans) the Medicare Current Beneficiary Survey (3000 veterans) and the 2004 round of the NLTCS . We will produce the model in a user-friendly, desk-top format for use by VHA policymakers and researchers.
We have also geo-coded all priority group 1A veterans, matched to NHCU beds, and found that there is a steep geographic gradient for the proportion of veterans predicted to be in NHCU beds, declining from over 100% of those predicted for veterans within 30 minutes drive time to < 30% for those with greater than 90 minutes drive time. (Presented at AGS) Using regression weights from the NLTCS, we have estimated the number of catastrophically disabled veterans identifed by the US Census, in conjunction with VA Office of the Actuary. Estimates of severe cognitive impairment (MMSE<15) are 4.13% for enrolled veterans, compared to 3.89% for all males in the 1999 NLTCS; rates are 5.6% and 5.87% for a MMSE<21, respectively.
We plan to improve the effectiveness of Federal resources committed to community based and institutional long term care of veterans by demonstrating the value of a powerful health services research technique, Grade of Membership (GoM) analysis, and determining it's incremental value over a simplier static compoment model. We plan to use the comprehensive federal payment record to estimate a federal LTC health voucher, that involves all 3 programs (Medicare, Medicaid and VA).
External Links for this Project
- Kinosian B, Stallard E, Wieland D. Projected use of long-term-care services by enrolled Veterans. The Gerontologist. 2007 Jun 1; 47(3):356-64. [view]
- Kinosian B. Geographic variation in access to nursing home care among priority groups. Paper presented at: American Geriatrics Society Annual Meeting; 2004 May 1; Las Vegas, NV. [view]
- Kinosian B, Stallard E, Canamucio A, Polsky D, Manton K. Puzzling pieces: validating the VA long term care home and community based care projections. Paper presented at: American Geriatrics Society Annual Meeting; 2009 May 1; Chicago, IL. [view]
- Kinosian B, Stallard E, Field S, Manton K, Bergofsky L, Canamucio A, Chen W. VA long-term care planning model: validation and update of nursing home use. Paper presented at: VA HSR&D National Meeting; 2009 Feb 12; Baltimore, MD. [view]