- Split View
-
Views
-
Cite
Cite
Justin Blackburn, Julie L. Locher, Meredith L. Kilgore, Comparison of Long-term Care in Nursing Homes Versus Home Health: Costs and Outcomes in Alabama, The Gerontologist, Volume 56, Issue 2, 1 April 2016, Pages 215–221, https://doi.org/10.1093/geront/gnu021
- Share Icon Share
Abstract
To compare acute care outcomes and costs among nursing home residents with community-dwelling home health recipients.
A matched retrospective cohort study of Alabamians aged more than or equal to 65 years admitted to a nursing home or home health between March 31, 2007 and December 31, 2008 (N = 1,291 pairs). Medicare claims were compared up to one year after admission into either setting. Death, emergency department and inpatient visits, inpatient length of stay, and acute care costs were compared using t tests. Medicaid long-term care costs were compared for a subset of matched beneficiaries.
After one year, 77.7% of home health beneficiaries were alive compared with 76.2% of nursing home beneficiaries (p < .001). Home health beneficiaries averaged 0.2 hospital visits and 0.1 emergency department visits more than nursing home beneficiaries, differences that were statistically significant. Overall acute care costs were not statistically different; home health beneficiaries’ costs averaged $31,423, nursing home beneficiaries’ $32,239 (p = .5032). Among 426 dual-eligible pairs, Medicaid long-term care costs averaged $4,582 greater for nursing home residents (p < .001).
Using data from Medicare claims, beneficiaries with similar functional status, medical diagnosis history, and demographics had similar acute care costs regardless of whether they were admitted to a nursing home or home health care. Additional research controlling for exogenous factors relating to long-term care decisions is needed.
With an aging U.S. population, finding the optimal setting for long-term care is an issue many older Americans and their families must confront. Nine million older Americans were estimated to have needed long-term care in 2012, and the number is projected to increase 33% by 2020 to 12 million (U.S. Department of Health and Human Services, 2012). In the United States, Medicaid is the primary means for financing long-term care services, and traditionally, most spending is on institutional care (Doty, 2000; Health Policy Institute, 2007). Policymakers in the future must not only increase overall capacity of long-term care systems but also address the issue of resource allocation to provide long-term care in other settings than nursing homes.
For some adults, nursing home care is unavoidable, but many older adults prefer to remain in the community as long as possible (Bayer & Harper, 2000; Nishita, Wilber, Matsumoto, & Schnelle, 2008). Driven by the U.S. Supreme Court’s Olmstead decision, recent federal legislation, including the Deficit Reduction Act of 2005 and the Affordable Care Act of 2010, has attempted to reduce states’ institutional bias for long-term care and permit more individuals to remain or transition to the community (Arling, Kane, Cooke, & Lewis, 2010; Denny-Brown & Lipson, 2009; Irvin & Ballou, 2010; Kasper & O’Malley, 2006a, 2006b; Miller, Ramsland, Goldstein, & Harrington, 2001).
Federal initiatives are increasing states’ incentives to increase access and utilization of Medicaid home- and community-based services (HCBS) waiver programs (Wenzlow & Lipson, 2009). One such example is the Money Follows the Person Rebalancing Demonstration Program administered by the Centers for Medicare and Medicaid Services, which provides funds to transition eligible individuals from nursing homes back into the community. From 2008 through 2011, approximately 20,000 people in 42 states transitioned back into community settings (Centers for Medicare and Medicaid Services, 2013). However, the evidence for cost-effectiveness of care provided through HCBS waivers as compared with nursing homes is equivocal (Chen & Berkowitz, 2012; Miller et al., 2001). Within federal guidelines, states have the freedom to allocate their Medicaid spending on long-term care as they see fit; for example, by capping the number of individuals who can receive HCBS waivers (Kitchener, Ng, & Harrington, 2004). In general, HCBS allocations have increased in the last decade; however, there is variability among states, and over time, in the amount for monies allocated to programs designed to prevent or reduce institutional long-term care.
The state of Alabama distributes relatively little of the Medicaid long-term care budget to HCBS compared with other states with 31.5% allocated toward HCBS and 61.8% on nursing home care (Kaiser Family Foundation, 2012). As of 2012, the number of aged individuals eligible to receive HCBS is capped at 9,205 (Alabama Medicaid Agency, 2012). As Alabama and policymakers from other states consider programs to increase HCBS in a time of state budget crises, determining the comparative effectiveness of different long-term care policies is imperative.
The purpose of this study was to compare outcomes and associated costs for acute care among patients receiving care in nursing homes with those receiving care in the community by home health agencies. A propensity-score-matched cohort design was used to compare nursing home residents and older adults receiving care from home health agencies with similar functional limitations and patterns of health service utilization.
Methods
Study Design and Sample
This matched, retrospective cohort study focused on Alabama residents aged 65 years or older who were admitted to either a nursing home or a home health care between March 31, 2007 and December 31, 2008. Although the ideal study is to compare costs and outcomes among nursing home residents and individuals receiving HCBS, information such as comprehensive assessments and cost of care received under Medicaid waiver programs does not currently exist. However, a related question can be addressed with existing data by comparing nursing home residents to home health recipients who have comparable data available.
Beneficiaries were required to have enrollment in Medicare Parts A and B, and no enrollment in a Medicare Advantage plan. The study period follow-up was the one-year period beginning with the initial admission assessment. A 90-day look-back period prior to follow-up was required in which individuals must have had Medicare Parts A and B, no Medicare Advantage coverage, and no record of assessments in either home health or a nursing home. Data from this 90-day period were used to define the cohort. Ensuring complete coverage during the look-back period permitted the inclusion of health status information in the propensity score generation that occurred prior to follow-up. Outcomes and cost data from this time period were not included in the analyses beyond defining the cohort. Some beneficiaries were dually eligible for Medicare and Medicaid. Medicaid long-term care costs were compared among pairs within the matched cohort wherein both beneficiaries were dual eligible. Beneficiaries’ Medicare claims data were included for the one year after the initial qualifying assessment, or the date of the beneficiary’s death. For the main analysis, costs and outcomes were compared among all beneficiaries, regardless if their follow-up was censored at death; a supplemental analysis stratified by death is presented as Supplementary Data.
Data Sources
Data for the analyses come from the Minimum Data Set 2.0 (MDS) assessment for beneficiaries receiving nursing home care and the Outcome and Assessment Information Set (OASIS) for beneficiaries receiving home health care. Nursing homes are required to complete an MDS survey on all nursing home residents at the time of admission, with quarterly and annual estimates used to develop a plan of care and assess patient status. Home health agencies are required to complete an OASIS assessment on all patients at the time of admission, at least every 60 days thereafter, or when patients’ conditions change, to develop a plan of care and assess patient status. Data collected by nursing home operators and home health agencies are maintained by Centers for Medicare and Medicaid Services. Medicare enrollment and claims data were linked to assessment data to provide cost and health status outcome data. Data from all sources, except Medicaid claims, were obtained to cover a period from January 1, 2007 through December 31, 2009. Medicaid long-term care claims information were obtained for a subset of dual-eligible beneficiaries from the Medicaid Analytic Extract data files covering the period from January 1, 2007 through December 31, 2008.
Basic demographic information, sex, age, race, and marital status were derived from assessment data. Assessments conducted for both home health and nursing home beneficiaries have a high degree of overlapping information. Data from MDS and OASIS were mapped to determine comparable variables. Functional status variables, including mobility and toileting, and the presence of pressure ulcers were derived in this way. Variables related to health services utilization were derived from Medicare claims; claims during the 90-day period prior to admission to nursing home or home health for matching variables, and up to one year during follow-up for outcome variables. For diagnosed illnesses in claims data, the Clinical Classification Software groupings (Elixhauser, Steiner, & Palmer, 2012) were used based on International Classification of Diseases 9th Edition primary diagnoses.
Data Analysis
Nursing home residents and home health recipients were matched based on high-dimensional propensity scores (Schneeweiss et al., 2009). The high-dimensional propensity score is an extension of traditional propensity score methods wherein it utilizes claims history of the population to reduce residual confounding. The model that produced propensity scores contained age, gender, race, marital status, functional status, prior health services utilization, and up to 200 diagnosis codes from Medicare claims. The matching procedure was a modified version of the Parsons greedy matching technique contained in the Pharmacoepi Toolbox version 2.3.0 (Rassen, Doherty, Huang, & Schneeweiss, 2011).
Standardized differences were used to compare the nursing home and home health samples before and after matching with propensity scores, with greater than an absolute value of 10 being indicative of imbalance. All outcomes were annual rates, based on one year of claims data. Of interest were survival rate, utilization of health services (hospital stays, days in hospital, and emergency department [ED] visits), and cost of health services (inpatient hospital stay, outpatient hospital visit, physician visit or lab, durable medical equipment, home health, nursing facility, and hospice care). Survival after one year was compared using McNemar’s test. Annual mean costs, hospital and ED visits, and days in hospital were compared using paired sample t tests.
Results
From January 1, 2007 through December 31, 2009, there were 45,655 Alabama Medicare beneficiaries with at least one assessment in either home health or nursing facility. During that time period, 27,245 beneficiaries had at least one admitting assessment. A total of 7,120 beneficiaries were excluded for not having a 90-day look-back period without an assessment, or the clean period was unobservable (i.e., admissions prior to March 31, 2007). Among the remaining 20,125 beneficiaries, 6,907 did not have full Medicare benefits (Parts A and B) or had some form of Medicare Advantage during the clean or follow-up period and thus were excluded. Also excluded were 584 beneficiaries with missing data; 580 were missing information about ability to bathe, 2 toileting, and 2 race. The remaining sample included 12,634 beneficiaries, 3,419 with admissions in a nursing facility and 9,215 into home health care.
Suitable matches could not be found for 10,052 of the 12,634 (79.6%) of eligible beneficiaries (62% among nursing home and 86% among home health), leaving a sample of 1,291 beneficiaries from each group. Table 1 presents the descriptive statistics comparing the sample prior to and after matching, using the standardized difference to demonstrate balance. The unmatched samples differed substantially on nearly all characteristics measured. A greater proportion of home health beneficiaries demonstrated greater mobility and toileting independence, had a lower presence of pressure ulcers, and had fewer inpatient stays. On average, home health beneficiaries had fewer diagnosed illnesses. Matching successfully created a balanced sample, as standard differences were all less than 5, except for cognitive disorders which was 9.3. The matched sample was a predominately women (74%), Caucasian (77%), currently unmarried (85%), and on average aged 80 years. Most of the beneficiaries in the cohort were independent or needed minimal assistance with mobility and toileting. A notable 71% of the sample had an inpatient hospital stay in the 90 days prior to their admission assessment.
. | Unmatched sample . | Matched sample . | Dual-eligible sample . | |||||
---|---|---|---|---|---|---|---|---|
Nursing home, N = 3,419 (%) . | Home health, N = 9,215 (%) . | Standard difference . | Nursing home N = 1,291 (%) . | Home health, N = 1,291 (%) . | Standard difference . | Nursing home, N = 426 (%) . | Home health, N = 426 (%) . | |
Female | 74.3 | 78.3 | 9.4 | 74.2 | 74.3 | 0.2 | 72.1 | 75.1 |
Average age (years) | 81.6 | 79.6 | 25.5 | 80.4 | 80.5 | 1.2 | 80.7 | 80.7 |
Caucasian | 77.4 | 70.7 | 15.3 | 76.5 | 76.6 | 0.2 | 75.5 | 73.2 |
African American | 22.2 | 27.5 | 12.2 | 22.9 | 23.0 | 0.2 | 23.9 | 25.4 |
Other | 0.4 | 1.8 | 0.14 | 0.5 | 0.4 | 0.1 | 0.5 | 1.4 |
Married | 15.6 | 17.4 | 4.8 | 15.4 | 15.2 | 0.6 | 14.8 | 14.1 |
Mobility | ||||||||
Independent | 18.0 | 67.2 | 114.7 | 36.2 | 35.7 | 0.1 | 33.8 | 35.0 |
Walks with assistance | 49.2 | 26.7 | 47.6 | 47.3 | 48.3 | 1.9 | 48.8 | 46.9 |
Wheelchair-dependent without assistance | 7.1 | 3.4 | 17.0 | 6.4 | 5.9 | 2.3 | 6.6 | 6.6 |
Wheelchair-dependent with assistance | 19.2 | 2.2 | 57.0 | 8.1 | 8.1 | 0 | 9.6 | 9.2 |
Bedbound (can reposition self) | 4.0 | 0.2 | 27.2 | 1.0 | 1.1 | 0.8 | 0.5 | 0.9 |
Bedbound (cannot reposition self) | 2.5 | 0.3 | 18.5 | 1.0 | 1.0 | 0 | 0.7 | 1.4 |
Toileting | ||||||||
Independent | 13.2 | 45.5 | 76.1 | 29.0 | 28.0 | 2.1 | 27.5 | 26.8 |
Some assistance required | 22.6 | 50.7 | 60.9 | 49.0 | 50.0 | 2.0 | 47.7 | 48.1 |
Complete assistance required | 44.6 | 1.6 | 118.6 | 9.8 | 9.9 | 0.5 | 11.0 | 11.3 |
Cannot toilet | 19.7 | 2.2 | 58.3 | 12.2 | 12.0 | 0.7 | 13.8 | 13.8 |
At least one pressure ulcer | 14.2 | 2.7 | 42.0 | 6.7 | 6.3 | 1.6 | 8.0 | 7.3 |
Prior inpatient stay (90 days) | 82.7 | 54.7 | 63.4 | 71.2 | 70.4 | 1.7 | 67.4 | 70.7 |
Prior hospice use (90 days) | 2.5 | 0.3 | 18.1 | 1.1 | 1.1 | 0 | 2.1 | 0.7 |
Delirium, dementia, and amnestic and other cognitive disorders (Clinical Classification Software 653) | 2.5 | 0.5 | 16.8 | 2.2 | 1.0 | 9.3 | 2.1 | 0.7 |
Mean number of diagnosed illnesses | 14.3 | 11.1 | 50.3 | 12.5 | 12.5 | 0 | 12.3 | 12.7 |
. | Unmatched sample . | Matched sample . | Dual-eligible sample . | |||||
---|---|---|---|---|---|---|---|---|
Nursing home, N = 3,419 (%) . | Home health, N = 9,215 (%) . | Standard difference . | Nursing home N = 1,291 (%) . | Home health, N = 1,291 (%) . | Standard difference . | Nursing home, N = 426 (%) . | Home health, N = 426 (%) . | |
Female | 74.3 | 78.3 | 9.4 | 74.2 | 74.3 | 0.2 | 72.1 | 75.1 |
Average age (years) | 81.6 | 79.6 | 25.5 | 80.4 | 80.5 | 1.2 | 80.7 | 80.7 |
Caucasian | 77.4 | 70.7 | 15.3 | 76.5 | 76.6 | 0.2 | 75.5 | 73.2 |
African American | 22.2 | 27.5 | 12.2 | 22.9 | 23.0 | 0.2 | 23.9 | 25.4 |
Other | 0.4 | 1.8 | 0.14 | 0.5 | 0.4 | 0.1 | 0.5 | 1.4 |
Married | 15.6 | 17.4 | 4.8 | 15.4 | 15.2 | 0.6 | 14.8 | 14.1 |
Mobility | ||||||||
Independent | 18.0 | 67.2 | 114.7 | 36.2 | 35.7 | 0.1 | 33.8 | 35.0 |
Walks with assistance | 49.2 | 26.7 | 47.6 | 47.3 | 48.3 | 1.9 | 48.8 | 46.9 |
Wheelchair-dependent without assistance | 7.1 | 3.4 | 17.0 | 6.4 | 5.9 | 2.3 | 6.6 | 6.6 |
Wheelchair-dependent with assistance | 19.2 | 2.2 | 57.0 | 8.1 | 8.1 | 0 | 9.6 | 9.2 |
Bedbound (can reposition self) | 4.0 | 0.2 | 27.2 | 1.0 | 1.1 | 0.8 | 0.5 | 0.9 |
Bedbound (cannot reposition self) | 2.5 | 0.3 | 18.5 | 1.0 | 1.0 | 0 | 0.7 | 1.4 |
Toileting | ||||||||
Independent | 13.2 | 45.5 | 76.1 | 29.0 | 28.0 | 2.1 | 27.5 | 26.8 |
Some assistance required | 22.6 | 50.7 | 60.9 | 49.0 | 50.0 | 2.0 | 47.7 | 48.1 |
Complete assistance required | 44.6 | 1.6 | 118.6 | 9.8 | 9.9 | 0.5 | 11.0 | 11.3 |
Cannot toilet | 19.7 | 2.2 | 58.3 | 12.2 | 12.0 | 0.7 | 13.8 | 13.8 |
At least one pressure ulcer | 14.2 | 2.7 | 42.0 | 6.7 | 6.3 | 1.6 | 8.0 | 7.3 |
Prior inpatient stay (90 days) | 82.7 | 54.7 | 63.4 | 71.2 | 70.4 | 1.7 | 67.4 | 70.7 |
Prior hospice use (90 days) | 2.5 | 0.3 | 18.1 | 1.1 | 1.1 | 0 | 2.1 | 0.7 |
Delirium, dementia, and amnestic and other cognitive disorders (Clinical Classification Software 653) | 2.5 | 0.5 | 16.8 | 2.2 | 1.0 | 9.3 | 2.1 | 0.7 |
Mean number of diagnosed illnesses | 14.3 | 11.1 | 50.3 | 12.5 | 12.5 | 0 | 12.3 | 12.7 |
. | Unmatched sample . | Matched sample . | Dual-eligible sample . | |||||
---|---|---|---|---|---|---|---|---|
Nursing home, N = 3,419 (%) . | Home health, N = 9,215 (%) . | Standard difference . | Nursing home N = 1,291 (%) . | Home health, N = 1,291 (%) . | Standard difference . | Nursing home, N = 426 (%) . | Home health, N = 426 (%) . | |
Female | 74.3 | 78.3 | 9.4 | 74.2 | 74.3 | 0.2 | 72.1 | 75.1 |
Average age (years) | 81.6 | 79.6 | 25.5 | 80.4 | 80.5 | 1.2 | 80.7 | 80.7 |
Caucasian | 77.4 | 70.7 | 15.3 | 76.5 | 76.6 | 0.2 | 75.5 | 73.2 |
African American | 22.2 | 27.5 | 12.2 | 22.9 | 23.0 | 0.2 | 23.9 | 25.4 |
Other | 0.4 | 1.8 | 0.14 | 0.5 | 0.4 | 0.1 | 0.5 | 1.4 |
Married | 15.6 | 17.4 | 4.8 | 15.4 | 15.2 | 0.6 | 14.8 | 14.1 |
Mobility | ||||||||
Independent | 18.0 | 67.2 | 114.7 | 36.2 | 35.7 | 0.1 | 33.8 | 35.0 |
Walks with assistance | 49.2 | 26.7 | 47.6 | 47.3 | 48.3 | 1.9 | 48.8 | 46.9 |
Wheelchair-dependent without assistance | 7.1 | 3.4 | 17.0 | 6.4 | 5.9 | 2.3 | 6.6 | 6.6 |
Wheelchair-dependent with assistance | 19.2 | 2.2 | 57.0 | 8.1 | 8.1 | 0 | 9.6 | 9.2 |
Bedbound (can reposition self) | 4.0 | 0.2 | 27.2 | 1.0 | 1.1 | 0.8 | 0.5 | 0.9 |
Bedbound (cannot reposition self) | 2.5 | 0.3 | 18.5 | 1.0 | 1.0 | 0 | 0.7 | 1.4 |
Toileting | ||||||||
Independent | 13.2 | 45.5 | 76.1 | 29.0 | 28.0 | 2.1 | 27.5 | 26.8 |
Some assistance required | 22.6 | 50.7 | 60.9 | 49.0 | 50.0 | 2.0 | 47.7 | 48.1 |
Complete assistance required | 44.6 | 1.6 | 118.6 | 9.8 | 9.9 | 0.5 | 11.0 | 11.3 |
Cannot toilet | 19.7 | 2.2 | 58.3 | 12.2 | 12.0 | 0.7 | 13.8 | 13.8 |
At least one pressure ulcer | 14.2 | 2.7 | 42.0 | 6.7 | 6.3 | 1.6 | 8.0 | 7.3 |
Prior inpatient stay (90 days) | 82.7 | 54.7 | 63.4 | 71.2 | 70.4 | 1.7 | 67.4 | 70.7 |
Prior hospice use (90 days) | 2.5 | 0.3 | 18.1 | 1.1 | 1.1 | 0 | 2.1 | 0.7 |
Delirium, dementia, and amnestic and other cognitive disorders (Clinical Classification Software 653) | 2.5 | 0.5 | 16.8 | 2.2 | 1.0 | 9.3 | 2.1 | 0.7 |
Mean number of diagnosed illnesses | 14.3 | 11.1 | 50.3 | 12.5 | 12.5 | 0 | 12.3 | 12.7 |
. | Unmatched sample . | Matched sample . | Dual-eligible sample . | |||||
---|---|---|---|---|---|---|---|---|
Nursing home, N = 3,419 (%) . | Home health, N = 9,215 (%) . | Standard difference . | Nursing home N = 1,291 (%) . | Home health, N = 1,291 (%) . | Standard difference . | Nursing home, N = 426 (%) . | Home health, N = 426 (%) . | |
Female | 74.3 | 78.3 | 9.4 | 74.2 | 74.3 | 0.2 | 72.1 | 75.1 |
Average age (years) | 81.6 | 79.6 | 25.5 | 80.4 | 80.5 | 1.2 | 80.7 | 80.7 |
Caucasian | 77.4 | 70.7 | 15.3 | 76.5 | 76.6 | 0.2 | 75.5 | 73.2 |
African American | 22.2 | 27.5 | 12.2 | 22.9 | 23.0 | 0.2 | 23.9 | 25.4 |
Other | 0.4 | 1.8 | 0.14 | 0.5 | 0.4 | 0.1 | 0.5 | 1.4 |
Married | 15.6 | 17.4 | 4.8 | 15.4 | 15.2 | 0.6 | 14.8 | 14.1 |
Mobility | ||||||||
Independent | 18.0 | 67.2 | 114.7 | 36.2 | 35.7 | 0.1 | 33.8 | 35.0 |
Walks with assistance | 49.2 | 26.7 | 47.6 | 47.3 | 48.3 | 1.9 | 48.8 | 46.9 |
Wheelchair-dependent without assistance | 7.1 | 3.4 | 17.0 | 6.4 | 5.9 | 2.3 | 6.6 | 6.6 |
Wheelchair-dependent with assistance | 19.2 | 2.2 | 57.0 | 8.1 | 8.1 | 0 | 9.6 | 9.2 |
Bedbound (can reposition self) | 4.0 | 0.2 | 27.2 | 1.0 | 1.1 | 0.8 | 0.5 | 0.9 |
Bedbound (cannot reposition self) | 2.5 | 0.3 | 18.5 | 1.0 | 1.0 | 0 | 0.7 | 1.4 |
Toileting | ||||||||
Independent | 13.2 | 45.5 | 76.1 | 29.0 | 28.0 | 2.1 | 27.5 | 26.8 |
Some assistance required | 22.6 | 50.7 | 60.9 | 49.0 | 50.0 | 2.0 | 47.7 | 48.1 |
Complete assistance required | 44.6 | 1.6 | 118.6 | 9.8 | 9.9 | 0.5 | 11.0 | 11.3 |
Cannot toilet | 19.7 | 2.2 | 58.3 | 12.2 | 12.0 | 0.7 | 13.8 | 13.8 |
At least one pressure ulcer | 14.2 | 2.7 | 42.0 | 6.7 | 6.3 | 1.6 | 8.0 | 7.3 |
Prior inpatient stay (90 days) | 82.7 | 54.7 | 63.4 | 71.2 | 70.4 | 1.7 | 67.4 | 70.7 |
Prior hospice use (90 days) | 2.5 | 0.3 | 18.1 | 1.1 | 1.1 | 0 | 2.1 | 0.7 |
Delirium, dementia, and amnestic and other cognitive disorders (Clinical Classification Software 653) | 2.5 | 0.5 | 16.8 | 2.2 | 1.0 | 9.3 | 2.1 | 0.7 |
Mean number of diagnosed illnesses | 14.3 | 11.1 | 50.3 | 12.5 | 12.5 | 0 | 12.3 | 12.7 |
After one year of follow-up, 77.7% of the home health group was alive compared with 76.2% of the nursing home (p = .3746). The average length of stay in nursing homes was 232 days, and the average length of time services were received from home health was 185 days (p < .001). Crossover and attrition among beneficiaries among each care setting were apparent, as shown in Table 2. A small proportion of nursing home residents received care in home health (8%) by the end of follow-up, while a larger proportion of home health beneficiaries (11%) were admitted to a nursing home. Another transition that occurred was among 828 (32%) beneficiaries who no longer received services in either the nursing home (22% of study group) or the home health (42% of study group; p < .0001).
Status or care setting . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | p-Value . |
---|---|---|---|
Died | 307 (24%) | 288 (22%) | .3746 |
Home health | 109 (8%) | 325 (25%) | <.0001 |
Nursing home | 589 (46%) | 136 (11%) | <.0001 |
Home (no Medicare home health) | 286 (22%) | 542 (42%) | <.0001 |
Status or care setting . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | p-Value . |
---|---|---|---|
Died | 307 (24%) | 288 (22%) | .3746 |
Home health | 109 (8%) | 325 (25%) | <.0001 |
Nursing home | 589 (46%) | 136 (11%) | <.0001 |
Home (no Medicare home health) | 286 (22%) | 542 (42%) | <.0001 |
Status or care setting . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | p-Value . |
---|---|---|---|
Died | 307 (24%) | 288 (22%) | .3746 |
Home health | 109 (8%) | 325 (25%) | <.0001 |
Nursing home | 589 (46%) | 136 (11%) | <.0001 |
Home (no Medicare home health) | 286 (22%) | 542 (42%) | <.0001 |
Status or care setting . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | p-Value . |
---|---|---|---|
Died | 307 (24%) | 288 (22%) | .3746 |
Home health | 109 (8%) | 325 (25%) | <.0001 |
Nursing home | 589 (46%) | 136 (11%) | <.0001 |
Home (no Medicare home health) | 286 (22%) | 542 (42%) | <.0001 |
One-year health service utilization rates are shown in Table 3. Beneficiaries in the home health sample were observed to have more frequent hospital stays (+0.3 visits per beneficiary per year) than beneficiaries in the nursing home sample (p < .001). Home health beneficiaries also had a higher rate of ED visits (+0.1 visits per beneficiary per year) that was statistically significant at the p < .05 level.
Average utilization . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Days in care setting | 232 | 185 | 47 | <.001 |
Hospital stays | 1.1 | 1.3 | −0.2 | <.001 |
Hospital length of stay (days) | 7.8 | 8.7 | −0.9 | .1222 |
Emergency department visits | 0.7 | 0.8 | −0.1 | .0021 |
Average utilization . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Days in care setting | 232 | 185 | 47 | <.001 |
Hospital stays | 1.1 | 1.3 | −0.2 | <.001 |
Hospital length of stay (days) | 7.8 | 8.7 | −0.9 | .1222 |
Emergency department visits | 0.7 | 0.8 | −0.1 | .0021 |
Average utilization . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Days in care setting | 232 | 185 | 47 | <.001 |
Hospital stays | 1.1 | 1.3 | −0.2 | <.001 |
Hospital length of stay (days) | 7.8 | 8.7 | −0.9 | .1222 |
Emergency department visits | 0.7 | 0.8 | −0.1 | .0021 |
Average utilization . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Days in care setting | 232 | 185 | 47 | <.001 |
Hospital stays | 1.1 | 1.3 | −0.2 | <.001 |
Hospital length of stay (days) | 7.8 | 8.7 | −0.9 | .1222 |
Emergency department visits | 0.7 | 0.8 | −0.1 | .0021 |
Annual costs of care between nursing home residents and home health recipients are shown in Table 4. Average total medical expenditures for home health beneficiaries was $31,423; $816 greater than the average costs for nursing home residents, a difference that was not statistically significant. Statistically significant differences of the annual average cost of care were detectable for several categories. Among 426 pairs of dual-eligible beneficiaries for whom Medicaid data were available, long-term care costs were $4,582 greater for nursing home residents than home health recipients (p < .001).
Cost of care . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Hospital stays | $8,902 | $9,995 | $ −1,093 | .0692 |
Outpatient hospital visit | 2,401 | 2,043 | 357 | .1032 |
Physician visit or lab | 3,990 | 4,416 | −427 | .0462 |
Durable medical equipment | 609 | 1,101 | −492 | <.0001 |
Home health | 1,853 | 5,851 | −3,998 | <.0001 |
Nursing facilitya | 12,340 | 5,626 | 6,714 | <.0001 |
Hospice care | 2,144 | 2,389 | −245 | .3935 |
Total costs | $32,239 | $31,423 | $816 | .5046 |
Cost of care . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Hospital stays | $8,902 | $9,995 | $ −1,093 | .0692 |
Outpatient hospital visit | 2,401 | 2,043 | 357 | .1032 |
Physician visit or lab | 3,990 | 4,416 | −427 | .0462 |
Durable medical equipment | 609 | 1,101 | −492 | <.0001 |
Home health | 1,853 | 5,851 | −3,998 | <.0001 |
Nursing facilitya | 12,340 | 5,626 | 6,714 | <.0001 |
Hospice care | 2,144 | 2,389 | −245 | .3935 |
Total costs | $32,239 | $31,423 | $816 | .5046 |
Note:aThese Medicare-incurred costs are associated with skilled nursing care or rehabilitation—not long-term care.
Cost of care . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Hospital stays | $8,902 | $9,995 | $ −1,093 | .0692 |
Outpatient hospital visit | 2,401 | 2,043 | 357 | .1032 |
Physician visit or lab | 3,990 | 4,416 | −427 | .0462 |
Durable medical equipment | 609 | 1,101 | −492 | <.0001 |
Home health | 1,853 | 5,851 | −3,998 | <.0001 |
Nursing facilitya | 12,340 | 5,626 | 6,714 | <.0001 |
Hospice care | 2,144 | 2,389 | −245 | .3935 |
Total costs | $32,239 | $31,423 | $816 | .5046 |
Cost of care . | Nursing home (N = 1,291) . | Home health (N = 1,291) . | Difference . | p-Value . |
---|---|---|---|---|
Hospital stays | $8,902 | $9,995 | $ −1,093 | .0692 |
Outpatient hospital visit | 2,401 | 2,043 | 357 | .1032 |
Physician visit or lab | 3,990 | 4,416 | −427 | .0462 |
Durable medical equipment | 609 | 1,101 | −492 | <.0001 |
Home health | 1,853 | 5,851 | −3,998 | <.0001 |
Nursing facilitya | 12,340 | 5,626 | 6,714 | <.0001 |
Hospice care | 2,144 | 2,389 | −245 | .3935 |
Total costs | $32,239 | $31,423 | $816 | .5046 |
Note:aThese Medicare-incurred costs are associated with skilled nursing care or rehabilitation—not long-term care.
Discussion
With the direction of Medicaid spending on long-term care seemingly shifting from institutional to home-based services, evaluating the outcomes and costs of each must be considered. This study sought to compare the acute care costs and outcomes among Medicare eligible nursing home residents with beneficiaries receiving home health, after matching on levels of functional impairment. In such situations, the optimal care setting may be unclear. Many beneficiaries in nursing homes had impairment levels too great to be considered for home health, while many home health recipients were functioning at a level that would not require nursing home care. Using a propensity-score-matched cohort, home health recipients were observed to have more frequent hospital stays and ED visits than the nursing home beneficiaries. However, no differences were observed in the one-year survival rate or the total acute care costs after one year.
Although Medicaid is the primary financier of long-term care in the United States, Medicare covers primarily acute care costs, including qualified short-term nursing home stays that are less than 100 days (e.g., postacute care). Medicare coverage of home health is based on physician-certified needs for intermittent skilled nursing care and physical therapy including occupational therapy and speech-language pathology services. Therefore, the outcomes of this study were primarily for acute care costs. The majority of the beneficiaries (71%) in the matched sample were observed to have had an inpatient stay in the 90 days before nursing home admission or home health initial assessment. Thus, the costs attributed specifically to home health or skilled nursing facility categories were for short-term rehabilitation services because they were paid by Medicare. In fact, among the nursing home group, this expenditure was the largest expense, on average and was much greater than the home health group. It is likely that some beneficiaries would not fully rehabilitate and nursing home placement could become long term. It was beyond the scope of this study to address long-term costs and outcomes because follow-up was limited to one year. The results of this study should be interpreted in light of its limitations. As a whole, the nursing home study group had worse functional status than home health recipients, and therefore, unobserved factors may have introduced selection bias with respect to choice of long-term care setting. Other studies have used techniques for deriving matching samples wherein an overlap of health needs could be identified from heterogeneous groups (Weiland, Kinosian, Stallard, & Boland, 2013). Data pertaining to social support were not available among recipients in home health that was comparable to what is available for nursing home residents via MDS. Social support could affect the placement to nursing home or home health. Such bias would likely favor outcomes and costs among home health recipients. Additionally, variables in MDS and OASIS do not completely map to one another and thus were limited in what pre-defined variables that could be specified. For example, cognitive deficits may not be captured equally in assessments. Despite inclusion of activities of daily living variables, such as toileting and mobility, there could remain differences in the cognitive functioning of the matched groups. On the basis of the claims data, the nursing home group had a slightly larger proportion of delirium, dementia, and amnestic and other cognitive disorders (Clinical Classification Software 653) than the home health group, even after matching (Table 1). It is also possible that the groups differed on availability of community services, home environments, and other unobserved factors not available in our data set. Finally, crossover between groups occurred among 245 beneficiaries (9%) at some point during follow-up; 136 (5% of study group) from home health to nursing home and 109 (4%) from nursing home to home health. In the interest of providing comparisons to inform policy on optimal placement, data were analyzed with an “intention to treat” design, where costs and outcomes for crossover beneficiaries were included in their original setting.
The idea that HCBS were a viable alternative to nursing homes began to gain attention in the 1970s, and the first evaluation of this idea was conducted on a national scale through the National Long-term Care Channeling Demonstration (Kemper et al., 1982). The channeling demonstration and subsequent studies found that services to keep older adults in the community required greater financial costs than comparable services provided in a nursing home, but it is associated with increased quality of life (Amaral, 2010; Dale & Brown, 2007; Thornton & Dunstan, 1986). Recently, states have begun experimenting with integrating care to align financing among dual-eligible beneficiaries. Seven states with different HCBS waiver allocations (California, Illinois, Massachusetts, New York, Ohio, South Carolina, and Virginia) have implemented demonstration projects aimed at reducing costs, primarily by reducing inpatient and emergency room visits (Kaiser Family Foundation, 2013). Evidence is not yet available on whether such cost-savings have been realized. Although Alabama is not among the seven states with a demonstration project, the lack of available data for integrating care (Gold, Jacobson, & Garfield, 2012) makes this study of interest. It is beyond the scope of this study to determine the overall cost-effectiveness between settings of long-term care in Alabama. However, the findings represent evidence that beneficiaries with similar levels of functioning have similar overall acute care expenditures whether in a nursing home or utilizing community services in the short term. It should be highlighted that although overall costs were not statistically different, there were differences in the categories of costs, such as for hospital visits.
This study highlights the need for future research using data from more sources, including more states. Future research is needed to further clarify the factors affecting outcomes associated with care in the community relative to care in the nursing facilities, such as comparing patients in HCBS wavier programs versus those who applied for the program but were denied and subsequently admitted to nursing homes.
Supplementary material
Please visit the article online at Supplementary Data to view supplementary material.
Acknowledgments
None of the authors have any proprietary interests or conflicts of interest related to this submission. This manuscript has not been published previously and is not simultaneously under consideration by any other publications.
References
Author notes
Decision Editor: Rachel Pruchno, PhD