|Home | About | Journals | Submit | Contact Us | Français|
Nursing homes are increasingly serving short-stay rehabilitation residents under Medicare skilled nursing facility coverage, which is substantially more generous than Medicaid coverage for long-stay residents. In relation to increasing short-stay resident care, potential exists for beneficial or detrimental effects on long-stay resident outcomes. We employ panel multivariate regression analyses using facility fixed-effects models to determine how increasing the proportion of Medicare days in nursing homes relates to changes in quality outcomes for long-stay residents. We find increasing the proportion of Medicare days in a nursing home is significantly associated with improved quality outcomes for long-stay residents. Findings reinforce prior research indicating that quality outcomes tend to be superior in nursing homes with greater financial resources. This study bolsters arguments for financial investments in nursing homes, including increases in Medicaid payment rates, to support better care.
Achieving high-quality care outcomes for nursing home residents is a longstanding challenge,1 which was recently addressed by the federal Improving Medicare Post-Acute Care Transformation Act of 2014 (IMPACT Act).2 Long-stay nursing home residents are known to experience extreme pain, to acquire pressure ulcers, and to lose independence performing activities of daily living, such as eating and dressing.3 Changes in the types of residents served by nursing homes have resulted in a more diverse patient population with varying care needs, which adds to the complexity of ensuring nursing home quality.
The issues of nursing home quality are embedded within an industry that has been experiencing substantial change. Facilities are transforming from primarily providing long-term housing and chronic care services for the oldest adults to increasingly providing short-stay housing and rehabilitative care for younger groups.4 These two populations (long-stay residents and short-stay patients) may be defined by a nursing home stay longer or shorter than 100 days as Medicare provisions restrict coverage for skilled nursing facility care to 100 days after an inpatient hospital stay.5,6
Numerous operational and financial differences exist between serving short- and long-stay residents. Professional personnel, such as nurses, physical therapists, and occupational therapists, are required to provide skilled nursing facility care. Such staff are more expensive than those required for the custodial care services commonly provided to long-stay residents. Consequently, Medicare skilled nursing facility rates are substantially more generous than Medicaid rates for long-stay residents.
Although staffing for skilled care is more costly, greater profitability of Medicare versus Medicaid payments appears to be driving changes in the nursing home industry. Medicare covers skilled nursing facility care at approximately $500 to $600 per day while Medicaid covers long-term nursing home care at about $125 per day.7 Industry advocates contend that Medicaid rates are insufficient; “Medicaid does not cover the total costs of providing services to its beneficiaries.”8 In contrast, Medicare rates appear to drive revenue: In 2010, Medicare covered a median of 12% of resident days but encompassed 23% of revenue in freestanding nursing homes.6 Within this reimbursement context, the Medicare market has been, and continues to be, the most rapidly growing sector of the nursing home industry.4,10
Efforts to support nursing home quality require careful consideration of numerous factors, including the diversity of nursing home residents (e.g., long- and short-stay), differences in reimbursement, and distinct care processes that result in optimal outcomes for these different types of residents. Nursing home operators are keenly aware of the differences between serving short- and long-stay residents. Recent research on nursing home culture change efforts shows that organizations focus service reforms on attracting Medicare residents to support stronger financial performance.11 In the words of a nursing home administrator: “You have to attract the Medicare Part A resident because the Medicaid rates don’t even cover our cost.”12 No prior research has examined the effects of such focus on short-stay residents on outcomes for long-stay residents, but related studies point to the potential for such a focus in one direction to deteriorate quality in other directions. For example, warnings of a distortion or distraction effect have been voiced with regard to the design of pay-for-performance (P4P) schemes, whereby “healthcare performance included in the P4P scheme may discourage efforts on aspects of healthcare performance that are not included and rewarded by the scheme. As a result, P4P may result in reduced healthcare quality.”13 Alternatively, other research suggests that nursing homes with greater financial resources provide better care overall. For instance, nursing homes with high proportions of Medicaid residents have been found to be of poorer quality,14 but increases in Medicaid payment rates have been found to result in better clinical care for long-stay residents.3 These patterns point to the potential for beneficial spillover effects, whereby Medicaid residents in nursing homes with more Medicare residents may receive better care as a result of the facility’s greater financial resources.3,15 To shed light on these opposing possibilities, we sought to determine how increasing skilled nursing facility services in nursing homes relates to changes in quality outcomes for long-stay residents.
We used facility-level nursing home data for years 2007–2010 downloaded from LTCfocUS.org, a data repository developed as part of the Shaping Long-Term Care in America Project at the Brown University Center for Gerontology and Healthcare Research (www.ltcfocus.org).16 The data repository integrates data from three sources, including Online Survey Certification and Reporting data collected during annual nursing home inspections by state survey agencies; the Centers for Medicare & Medicaid Services (CMS) Nursing Home Compare repository; and aggregated resident-level data regarding clinical and functional status originating from the nursing home residential assessment instrument Minimum Data Set version 2.0. These individual-level data are collected for every nursing home resident on admission and at least quarterly thereafter.16
To create our study cohort, we started with 15,369 facilities included in the LTCfocUS.org dataset during at least one of the study years. We then excluded facilities that were not present in all four years of the dataset (n = 1,443). This exclusion ensures that changes in independent or dependent variables observed over time pertain to changes within nursing homes rather than to changes occurring to the cohort as a result of nursing homes entering or exiting the cohort during the study period (e.g., because of nursing homes closing or new nursing homes opening). For statistical modeling purposes, we also excluded facilities that had missing data on (1) our primary independent variable, the proportion of Medicare covered days for all residents (n = 610); (2) the covariates in the analytic model (n = 846); or (3) all three of the outcome variables (n = 292). Missing data on the outcome variables were the result of systematic censoring of the public data. The LTCfocUS.org dataset incorporates data, including the three outcomes examined in this paper, from CMS’s Nursing Home Compare website. Because CMS censors reporting a facility’s performance on outcomes when fewer than 30 long-stay residents have the outcome of interest, this censoring resulted in missing data and thus exclusions of facilities from our study. Implementing these exclusions resulted in a cohort of 12,178 nursing homes with at least two years of data on one or more of the dependent variables (percentage of long-stay residents experiencing daily pain, worsening pressure ulcers, activities of daily living decline) between 2007 and 2010. The 3,191 excluded facilities differed significantly from the included 12,178 facilities in that they had fewer female residents (69% vs. 72%), younger residents (79 vs. 81) and fewer beds (59 vs. 121), and included a higher proportion of hospital-based facilities (21% vs. 4%).
To conduct our analysis on each of the quality outcomes of interest we created three outcome-specific panels from the 12,178 facilities that met our inclusion criteria. Restricting the panels to facilities with at least 2 years of data on the respective dependent variables. Nineteen facilities were excluded for not having at least 2 years of publicly reported outcomes on pain, resulting in a final pain analytic sample of 12,159 facilities. The 19 excluded facilities had lower occupancy compared to the final sample (82% vs. 87%), and fewer beds (81 vs. 121). Similarly, 204 facilities had missing data on activities of daily living decline for a final sample of 11,974; the excluded facilities had lower occupancy rates (81 vs. 86), fewer beds (82 vs. 124 beds), and represented fewer for-profit facilities (68% vs. 71%). Additionally, 4,978 facilities were excluded from the pressure ulcer sample as a result of censored data on this outcome, resulting in a cohort of 7,200 nursing homes. In comparison to our pressure ulcer cohort, facilities excluded from this cohort were more likely to be part of a chain (60% vs. 52%), and had fewer beds (93 vs. 142 beds) and higher staffing levels (e.g., 2.24 vs. 2.11 certified nursing assistant hours per resident day).
The key payment variable we examined was the proportion of days during each study year that were Medicare skilled nursing facility covered days for all residents within each nursing home. For each study year, we also examined nursing home performance on three quality measures included in CMS’s 5-Star rating system including measures of long-term residents experiencing daily pain, decline in activities of daily living, and worsening pressure ulcers. Although CMS’s 5-Star rating system includes additional measures of quality, only these three measures are included in the LTCfocUS.org dataset.16
We used several control variables in our multivariate analysis, including aggregate resident characteristics and facility characteristics, based on their relationship to the Medicare payment variable or to the three quality outcomes found in other studies. Resident characteristics used as control variables include the proportions of residents in each nursing home by age, sex, and race18; the proportion of residents with a low, moderate, or high score on the Cognitive Performance Scale; and average Resource Utilization Group nursing case mix index (a measure of the relative intensity of care, whereby the higher the score, the more severe the average acuity profile of the residents in a facility) for all residents in each nursing home.19 Additionally, we incorporated nursing home characteristics as covariates, including number of beds and occupancy rate; indicators of for-profit status, ownership by a multi-facility chain, an indicator of a hospital-based nursing home; and several staffing features, including registered nurse, licensed professional nurse, and certified nursing assistant hours per resident day, and use of a medical extender. In nursing homes, certified nursing assistants constitute the majority of staff and provide direct hands-on care to residents, while licensed professional nurses and registered nurses are fewer in number and often serve in a supervisory capacity. A medical extender is a nurse practitioner or physician’s assistant. Furthermore, as a control for the quality of care processes, we included the percentage of residents in each nursing home who were restrained.
We performed panel multivariate regression analyses using facility fixed-effects models for each of the three outcome variables (proportion of long-term residents who experience daily pain, worsening pressure ulcers, activities of daily living decline). A panel fixed-effects model controls for unobserved time-invariant nursing home characteristics that may be correlated with the outcomes and allows for a difference-in-difference model causal interpretation of the results. We used the XTREG (cross-sectional time-series linear regression) procedure available in Stata statistical software version 12, which fits regression models to panel data. For each model, the value of the coefficient for the key independent variable, the proportion of days that were Medicare-covered skilled nursing facility days, was then plotted on a bar graph to present the relationship between the key independent variable and the three dependent variables.
Several study limitations are noted. First, because of censored or missing data in the LTCfocUS.org dataset, a portion of facilities were excluded from our full study sample and from each of the outcome panels, and statistically significant differences were found between our sample and the excluded nursing homes. Accordingly, our findings may not be generalizable to all U.S. nursing homes—such as hospital-based and smaller facilities, both of which were underrepresented in our sample—and the relationship of increasing Medicare days to long-stay resident outcomes may be different in the excluded facilities. The pressure ulcer sample was most limited by facility exclusions because of censored data. Second, our study attended only to three quality measures for long-stay residents: daily pain, decline in activities of daily living, and worsening pressure sores. These outcome measures may be particularly sensitive to the presence of a post-acute care team. Results may differ for other long-stay resident quality measures, such as those addressing urinary tract infections, falls, physical restraints, and catheters. These measures were not available in the dataset. Finally, our study relied on data from 2007 through 2010, and findings may be different in more recent years.
From 2007 to 2010 there was an aggregate and steady increase across study facilities in the proportion of days that were Medicare covered days (from 14.84 to 16.83; Exhibit 1). Simultaneously, across study facilities, there was improvement in all three long-stay resident quality outcome measures examined (Exhibit 1), specifically, decreases in the percentage of long-stay residents with daily pain (5.1% to 3.4%), with worsening pressure ulcers (2.5% to 2.0%), and with decline in activities of daily living (15.9% to 14.9%).
Exhibit 2 presents the characteristics of all study facilities, and the characteristics of the facilities in each outcome panel (activities of daily living decline, pain, pressure ulcers) in 2007, including the explanatory Medicare skilled nursing facility payment variable, aggregate resident characteristics, and nursing home characteristics. Across all facilities (n = 12,178), 14.9% of days were covered by Medicare, the majority of residents were female (71.25%), white (82.64%), and about 81 years of age. The full sample of facilities were mostly for-profit (71%), part of a chain (55%), and freestanding (not hospital-based). Facilities contributing to the activities of daily living decline and pain cohorts had residents who were predominantly female (83%) and 80 years of age, an average of 2.11 certified nursing assistant hours per resident day, and were predominantly for-profit (71%) and part of a chain (56%). Facilities in the pressure ulcer cohort had an average of 142.5 beds and an 86% occupancy rate, and 72% were for-profit. Staffing rates in the pressure ulcer cohort included 2.02 certified nursing assistant hours per resident day, 0.29 registered nurse hours per resident day, and 0.79 licensed professional nurse hours per resident day.
Multivariate analyses indicate that increasing the proportion of Medicare days in a nursing home is significantly associated with improved quality outcomes for long-stay residents. Specifically, with each 1 percent increase in the proportion of Medicare days there was a decrease in each of the outcomes—the proportion of long-stay residents with daily pain, residents experiencing a decline in performing activities of daily living, and worsening pressure ulcers (Exhibit 3).
In multivariate analyses, several additional nursing home characteristics were related to the resident outcome variables. All three outcomes were significantly associated with the use of restraints, the case mix, and the registered nurse and certified nursing assistant hours per resident day.20 In addition, the proportion of long-stay residents with daily pain was significantly associated with the number of beds in the facility, the occupancy rate, the age of the residents, and the proportions of residents who were female, white, and had low Cognitive Performance Scale scores.20 Additionally, the proportion of long-stay residents experiencing a decline in performing activities of daily living was significantly associated with the proportions of residents with a low or moderate Cognitive Performance Scale score.20
This study investigates whether changes in the proportion of days that are Medicare skilled nursing facility days in a nursing home help explain changes in long-stay resident outcomes reported on Nursing Home Compare and included in CMS’s 5-Star Quality Rating program. The objective was to address concerns about quality outcomes for long-stay residents in nursing homes with increasing proportions of Medicare days. Using panel multivariate regression analyses and facility fixed-effects models to examine data from 2007 through 2010, we find nursing homes increasing the proportion of Medicare days achieved improvement on all three of the quality outcomes for long-stay residents examined. These findings reflect those of previous research showing overall quality is better in nursing homes with greater financial resources,3,15 and alleviate concerns that heightened nursing home attention to more lucrative short-stay residents may deteriorate quality for less lucrative long-stay residents.11 Accordingly, this study supports the argument that beneficial spillover effects ensue for long-stay residents in nursing homes that increasingly serve more lucrative Medicare residents, and our findings oppose the suggestion that serving more short-stay residents may have detrimental distraction effects for long-stay residents within the nursing home. To contend with the persistently poor quality of nursing homes, these findings bolster the argument that greater financial investments, such as increases in Medicaid payment rates, will help improve quality.21 This contention is consistent with prior research showing higher Medicaid payment rates are associated with better care in nursing homes.22
In addition to increasing proportions of Medicare days, several other variables are significantly related to the examined long-stay resident quality outcomes (activities of daily living decline, worsening pressure ulcers, daily pain), including registered nurse and certified nursing assistant hours per resident day, restraint use, and case mix. In alignment with prior research showing that nursing homes with the highest levels of registered nurse staffing improve most quality outcomes, including pressure sores, fractures, and urinary tract infections,23 we find that increased registered nurse hours per resident day is correlated with significant improvements in the three quality outcomes we examine.20 These findings suggest that registered nurse time is a contributing mechanism by which the observed spillover effects are achieved, and policies and practice strategies affording nursing homes more registered nurse time may help improve nursing home care. Also, consistent with prior research that correlates restraint use with higher fall rates among nursing home residents,24 as well as lower ability to perform ADLs, and more walking dependence,25 we find increased use of restraints is significantly correlated with poorer performance on the three quality outcomes we examined.20 However, other findings offer more questionable implications. For example, increased certified nursing assistant hours per resident day is significantly correlated with poorer performance on the quality outcomes examined.20 In contrast, prior nursing home research shows that higher certified nursing assistant hours per resident day is associated with better performance on a wide range of quality indicators, including deficiency scores from inspections26 and resident falls.27 Accordingly, our findings should be interpreted in light of this prior research and call for inquiry regarding how the outcomes we examined are influenced by certified nursing assistants. Regarding resident case mix, we find increasing complexity within a nursing home is associated with significantly more decline in activities of daily living among long-stay residents, but also improved performance on measures of long-stay resident pain and pressure ulcers.20 Additional research to explain the clinical relevance of these empirical relationships is recommended.
Our findings support the perspective that nursing homes with more financial resources provide better care overall. However, the parameters of the association between increased Medicare days and improved quality outcomes, and the mechanisms driving this association, require additional research to address implications for practice and policy. Regarding the parameters of this association, during the observation period of this study the growth rate of Medicare days in nursing homes may be considered moderate (<2% growth from 2007 to 2010). Higher rates of Medicare growth in nursing homes may relate differently to long-stay resident outcomes. For example, there may be a rate of Medicare growth at which quality outcomes for long-stay residents deteriorate. Additional research is needed to specify these parameters, and to identify the mechanisms driving improved outcomes for long-stay residents. Research on hospice care in nursing homes indicates a beneficial spillover effect on the management of pain for other (non-hospice) residents in the nursing home.28 A similar diffusion of skills and competencies across nursing home staff may be associated with increasing skilled nursing facility care and support improved outcomes for long-stay residents. Because of the noted limitations of this study, future research is encouraged that examines the relationship between quality and resident payment sources with more recent and complete datasets and attending to additional quality outcomes.
Providing high-quality care and maintaining viable finances are both necessary for nursing home survival, and maintaining a supply of high-quality nursing homes is necessary to meet our national healthcare demands. Within the current reimbursement landscape, nursing homes shifting to provide proportionally more skilled nursing rehabilitation perform stronger financially.9 This shift is aligned with improvements in key quality outcomes for long-stay residents. Our findings help alleviate concerns that quality outcomes for long-stay residents decline as nursing homes shift to provide more skilled nursing facility care, and reinforce the argument that financial investment in nursing homes, such as increases in Medicaid payment rates, may be key to improving quality.21
Funding came in part from RTI International. Natalie Leland was funded by the National Center for Medical Rehabilitation Research (NCMRR) within the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) and the National Institute of Neurological Disorders and Stroke (Grant No. K12 HD055929).
Michael Lepore, Senior health policy and health services researcher in the Aging, Disability, and Long-Term Care program at RTI International, in Washington, D.C. and an adjunct assistant professor in the Department of Health Services, Policy, and Practice at Brown University.
Natalie E. Leland, Assistant professor with a joint appointment in the Mrs. T. H. Chan Division of Occupational Science and Occupational Therapy and the Davis School of Gerontology at the University of Southern California, in Los Angeles and an adjunct assistant professor in the Department of Health Services, Policy, and Practice at Brown University.