An estimated 500,000 or 16.5% of enrolled veterans over 65 years of age have difficulty performing basic and instrumental activities of daily living (ADL), but the majority of healthcare costs are concentrated among a small proportion of medically complex older adults usually defined on the basis of a large number of chronic conditions. The work from this project spans these two topics of disability and of multiple chronic conditions. Although medical complexity is often defined primarily by the number of chronic conditions, not all patients with chronic conditions are complex and patients with a single serious illness may have significant disability and can have complex and costly medical needs. This project examines the relationships among disability, chronic conditions, and healthcare utilization and costs.
The proposed project bridges several lines of scientific research on disability and chronic conditions in aging and on veterans' use of health services in and outside of the VHA -- to address the following research questions:
1. Are older veterans at increased risk of incident or new-onset disability compared to older non-veterans? Do older veterans who develop incident disability differ in their likelihood of recovery compared to older non-veterans?
2. What is the role of chronic conditions in the development of incident disability in veterans and in non-veterans?
3. Do chronic conditions, previous healthcare utilization, and functional disability predict future health care costs and survival?
4. How do different definitions of complexity (based on chronic conditions, healthcare utilization, and/or functional measures) compare in their predictive value for healthcare costs and survival?
We hypothesize that complexity is associated with increased healthcare costs and increased mortality.
This is an observational cohort study using the Medicare Current Beneficiary Survey (MCBS), a continuous, longitudinal survey of a representative rotated sample of 12,000 Medicare beneficiaries. In the first phase of the analysis covering questions 1 and 2, we merged data from the 1996-2002 MCBS with data from the VA's National Patient Care Database (NPCD) to identify a sample of 6637 males, age 65 and older, with no disability when they entered the MCBS. The cohort includes 4432 (67%) veterans. The MCBS provides data on disability over time, medical conditions, other baseline demographic characteristics and longitudinal data on non-VHA health care use from linked Medicare claims. VHA use is identified from the NPCD. Our outcomes were death and incident disability any time after entry in the MCBS. We estimated discrete time hazard models with follow up from 1 to 3 years, adjusting for baseline age, race/ethnicity, education, income, secondary insurance, self-reported health, smoking, comorbid conditions, instrumental ADLs, hospitalization in past 12 months, use of antidepressants, use of anxiolytics, use of antipsychotics, follow-up year and veteran status.
In the second phase of this analysis (occurring after Medicare data became available to VA investigators again in 2010), we used the MCBS data from 2000 (year 1) & 2001 (year 2). We excluded those under age 65, managed care participants, & decedents in 2000. Using year 1 data, we developed 3 definitions of complexity from candidate variables including: chronic conditions from claims; healthcare utilization patterns associated with high costs in year 1; and self-reported functional status (i.e. performance of activities of daily living). For each definition, we measured prevalence, year 1 & 2 costs, and mortality in year 2 compared to the full sample.
Our initial analyses focused on the risk of incident disability and death in veterans compared to non-veterans. Overall, 9.7% of the cohort died and 23% became disabled between 1996 and 2002. There were no significant differences in risk of death or disability between veterans (n = 4432) and non-veterans (2205). We then used the VA NPCD to further categorize veterans as: non-users of VHA (n = 3873) and users of VHA (n=559). Older male veterans using VHA services were 1.3 times as likely to become disabled as veterans not using VHA and non-veterans after controlling for demographic factors and health status. There was no difference in the risk of death for veteran users, non-users and non-veterans. We also analyzed data on adults above age 64 with no activity of daily living (ADL) difficulties at baseline (n = 14,226). Five ADLs were measured annually and recovery was defined as regaining complete ADL function at follow-up. The strongest correlates of ADL difficulty were use of antipsychotic medications (adjusted odds ratio [AOR] = 1.93, 95% confidence interval [CI] = 1.44 to 2.58), instrumental ADL difficulty (AOR = 1.90, 95% CI = 1.74 to 2.07), and fair-poor general health (AOR = 1.59, 95% CI = 1.42 to 1.78). Selected specific chronic conditions (arthritis, asthma, cancer, diabetes, and stroke) were associated with a moderately increased risk. Only the number of incident ADL difficulties was associated with recovery (AOR = 0.02, 95% CI = 0.01 to 0.02).
In the second phase, we found that older adults with 4 or more chronic conditions accounted for 24% of the Medicare population; although they accounted for 46% of health care costs in year 1, this decreased to 36% of health care costs in year 2. Alternatively, those who had a major chronic condition (i.e. high cost conditions such as diabetes, dementia) and either 2 or more acute hospitalizations or 3 or more other conditions accounted for a smaller fraction of the population (14% of the sample) but accounted for 29% of year 2 costs. Adding functional limitation in bathing to the earlier definition (using major chronic condition and either 2 or more hospitalizations or 3 or more other chronic conditions) identified 23% of the sample which accounted for 48% of costs in year 2. Thus, the two methods of identifying complexity (either 4 or more chronic conditions or a definition that combines chronic conditions with acute hospitalizations, other chronic conditions, and disability) identified an approximately equal fraction of the population as being complex (approximately a quarter); however, the definition that combines chronic conditions with hospital utilization and disability identified a subgroup with a greater concentration of costs in the subsequent year. All 3 methods of defining complexity identified individuals with a significantly increased risk of mortality in the subsequent year.
This project has advanced the field by examining incident disability among veterans using a longitudinal nationally-representative data set. Such studies have not been conducted among veterans in the past because no VHA dataset is available with information on functional status of veterans that is both nationally-representative and longitudinal. Our phase 2 analyses have shown that using chronic conditions alone identifies a population that has a moderate concentration in future costs and identification of medically complex adults is improved when prior hospitalizations and functional status are also considered. These measures can be used to identify patients as appropriate targets for interventions to improve care and reduce the costs of serious illness in the VA.
- Hung WW, Ross JS, Boockvar KS, Siu AL. Association of chronic diseases and impairments with disability in older adults: a decade of change? Medical care. 2012 Jun 1; 50(6):501-7.
- Hung WW, Ross JS, Boockvar KS, Siu AL. Recent trends in chronic disease, impairment and disability among older adults in the United States. BMC geriatrics. 2011 Aug 18; 11:47.
- Federman AD, Penrod JD, Livote E, Hebert P, Keyhani S, Doucette J, Siu AL. Development of and recovery from difficulty with activities of daily living: an analysis of national data. Journal of aging and health. 2010 Dec 1; 22(8):1081-98.
- Siu AL, Spragens LH, Inouye SK, Morrison RS, Leff B. The ironic business case for chronic care in the acute care setting. Health affairs (Project Hope). 2009 Jan 1; 28(1):113-25.
- Keyhani S, Ross JS, Hebert P, Dellenbaugh C, Penrod JD, Siu AL. Use of preventive care by elderly male veterans receiving care through the Veterans Health Administration, Medicare fee-for-service, and Medicare HMO plans. American journal of public health. 2007 Dec 1; 97(12):2179-85.
- Siu AL, Morano B. Interventions to improve transitions of care and dual use. Presented at: VA Geriatric Research Education and Clinical Center (GRECC) Annual Directors Meeting; 2010 Jun 29; Washington, DC.
- Ross JS, Zhu CW, Livote E, Siu AL. VA Use and Reliance Among Dual Users with Multiple Chronic Diseases. Poster session presented at: VA HSR&D National Meeting; 2009 Feb 12; Baltimore, MD.
- Penrod J, Zhu CW, Dellenbaugh C, Keyhani S, Hebert P, Siu AL. New or increased use of VA services by older veterans. Poster session presented at: AcademyHealth Annual Research Meeting; 2008 Jun 8; Washington, DC.
- Zhu CW, Kim H, Luo X, Penrod J, Ross J, Siu AL. Predicting high medical cost using administrative claims and survey data. Poster session presented at: AcademyHealth Annual Research Meeting; 2008 Jun 8; Washington, DC.
- Penrod J, Zhu CW, Dellenbaugh C, Hebert P, Keyhani S, Doucette J, Federman A, Siu AL. New onset disability in older veterans. Poster session presented at: VA HSR&D National Meeting; 2008 Feb 13; Baltimore, MD.
- Keyhani S, Ross JS, Hebert P, Dellenbaugh C, Penrod JD, Siu AL. Does the VA Provide Veterans with Better Quality Preventative Care Compared to Medicare HMO Plans? Paper presented at: AcademyHealth Annual Research Meeting; 2007 Jun 3; Orlando, FL.
- Penrod JD, Dellenbaugh C, Doucette J, Federman AD, Hebert P, Zhu CW, Keyhani S, Siu AL. Risks of Incident Disability and Death in Older Veterans and Non-veterans. Poster session presented at: VA HSR&D National Meeting; 2007 Feb 23; Arlington, VA.
Aging, Older Veterans' Health and Care, Health Systems
Data Merge, Disability, Utilization patterns, VA/non-VA comparisons