ABSTRACT MEMORY CARE UNITS IN OHIO LONG-TERM CARE FACILITIES by Nathan David Sheffer The prevalence of Alzheimer s disease (AD) is growing in the United States. Many adults with AD will require long-term care services. Memory care units (MCUs) are a popular type of special care unit within long-term care facilities. Previous research is inconclusive regarding quality of life and quality of care outcomes for MCU residents. MCU regulations are a growing topic in Ohio and the United States. Ohio Governor John Kasich signed House Bill 470 into law effective March 21, 2017, requiring the development of MCU recommendations within six months. Using the 2005 and 2013 Ohio Biennial Survey of Long-Term Care Facilities, this study examined the number of MCUs, developed a profile of facilities with and without a MCU, and determined facility characteristics significantly associated with operating a MCU. Logistic regression analysis determined that profit status, facility size, proportion of days paid for by Medicaid, county poverty rate, and specialized activities were significantly associated with an Ohio nursing facility operating a MCU in 2013. Recommendations are presented for consumers, Ohio policy makers, and future researchers.
MEMORY CARE UNITS IN OHIO LONG-TERM CARE FACILITIES Thesis Submitted to the Faculty of Miami University in partial fulfillment of the requirements for the degree of Master of Gerontological Studies by Nathan David Sheffer Miami University Oxford, Ohio 2017 Advisor: Jonathon Vivoda, Ph.D Reader: Robert Applebaum, Ph.D Reader: Jane Straker, Ph.D 2017 Nathan David Sheffer
This Thesis titled MEMORY CARE UNITS IN OHIO LONG-TERM CARE FACILITIES by Nathan David Sheffer has been approved for publication by The College of Arts and Science and Department of Sociology and Gerontology Jonathon Vivoda, Ph.D Robert Applebaum, Ph.D Jane Straker, Ph.D
Table of Contents Chapter I: Introduction..1 Background..2 Memory Care Units..3 Regulations..4 Research Questions..6 Chapter II: Methods 6 Measures..6 Analysis 9 Chapter III: Results 10 Analysis 1: Change of Number of MCUs...10 Analysis 2: Facilities with a MCU vs Facilities without a MCU..10 Missing Data Analysis...13 Analysis 3: Logistic Regression 16 Chapter IV: Discussion..17 Limitations.20 Recommendations..20 For Consumers...20 For Policy Makers.21 For Future Research...23 References..25 iii
List of Tables Table 1: Measures and Descriptions 9 Table 2: 2005 vs. 2013 Ohio MCUs..10 Table 3: Descriptive Statistics Comparing Nursing Homes (NH) with and without a MCU...12 Table 4: Descriptive Statistics of Continuous Measures- NH with MCU vs. NH without MCU...13 Table 5: Analysis of Missing Data.15 Table 6: Results of Logistic Regression Model.17 iv
Dedication This project is dedicated to my grandmother, Lillian Steele, and her identical twin sister, Lucille Deter. v
Acknowledgements I would like to thank my committee for their help, advice, support, and willingness to serve on my committee. Thank you, Dr. Vivoda, for your continued support as my committee chair, academic advisor, and graduate assistant supervisor. I appreciate the relationship we built and everything you have done for me. I would also like to acknowledge Matt Nelson for his willingness to help me throughout this process. I would like to recognize all the graduate students in Gerontology, and specifically my cohort for working through our issues together. I am blessed to have family and friends who continue to support me. Finally, I would like to thank Sara Perkins for listening, insisting I never give up, and supporting me. I greatly appreciate everyone at Miami University and will miss Oxford, Upham, the students, faculty, and researchers dearly. Love and Honor. vi
Chapter I Introduction Memory care units are a type of special care unit (SCU) within a nursing home designed for adults with Alzheimer s disease or dementia. Dementia is a disease of the brain, and often does not affect the body or physical functioning. It occurs in stages and is frequently slow in its progression. Dementia and Alzheimer s disease are not the same. Dementia is the broad, overarching term that includes multiple diseases, one of which is Alzheimer s disease (Alzheimer s Association, 2016). While there are over 100 different types of dementia, Alzheimer s disease is the most common and a projected 60 to 80% of dementia cases are of the Alzheimer s type (Alzheimer s Association, 2016). In both research and practice, Alzheimer s and dementia are often used in combination or interchangeably. Most statistical reporting is related to Alzheimer s disease and often does not differentiate Alzheimer s from other forms of dementia. For this study, memory loss of any kind is the main concern, however the terms dementia and Alzheimer s disease, AD for short, will both be used. Similarly, there is a lack of consensus on a singular name for special care units for this population. SCUs were originally created to provide tailored, disease-specific care to a group of residents with a particular condition or need (Smith, 1995; Rowles & Teaster, 2015). Dr. Powell Lawton created the first nursing home setting specifically for people with dementia in the 1970s, and many others have followed in his footsteps (Calkins, 2012). SCUs for people with AD exist under many names, Alzheimer s Special Care Unit, Dementia Special Care Unit, and Memory Care Unit among others (Gruneir, Lapane, Miller, & Mor, 2008; Kelsey, Laditka, & Laditka, 2008; Alzheimer s Association, 2016). Due to the growth of research on AD and specifically on long-term care for AD, the name of SCUs for this population has changed multiple times over. Currently, academics and researchers argue that memory care unit is an inappropriate name due to the notion that these units only provide memory care. In reality, these units provide (or should provide) behavioral management, skilled care and many other aspects of care unrepresented by the memory care unit title (Stone, 2014). Nonetheless, all attempt to provide targeted, enhanced care to adults with memory loss. Although memory care units, or MCUs, may not be the most appropriate term, it is currently the most widely used name and will be the term used throughout this study. In addition, the term SCU will also be used when discussing previous literature conducted on SCUs, before the term MCU came about. Two problems with MCUs are the lack of knowledge about what these units provide and variation in what exactly is provided from one facility to another. Despite considerable growth in the number of SCUs, and specifically MCUs, over the past 15 years throughout the United States, there have been limited guidelines or regulations directing the assessment of these units. Previous research is inconclusive regarding quality of care and outcome measures for adults with AD or dementia living in SCUs/MCUs compared to the same population in traditional (non- SCU) nursing home units. In addition, advocates have been calling for increases in staffing ratios, consistent assignment, and dementia-specific staff training in MCUs since the early 1990 s (Grant, Potthoff, Ryden, & Kane, 1998). This combination has led to questions regarding care in MCUs (Phillips, Sloane, Hawes, Koch, Han, Spry, & Williams, 1997; Grant et al., 1998; Buchanan, Choi, Wang, Ju, & Graber, 2005; Gruneir, Lapane, Mill, & Mor, 2008; Luo, Fang, Liao, Elliot, & Zhang, 2010). Are memory care units a good idea? Are they just a marketing and money making strategy? Should memory care units have different regulations? Although this 1
study may not specifically answer each of these questions, the goal is to gain a better understanding of MCUs in the state of Ohio and raise questions for future researchers to explore. Prior to December 2016, and the inception of this study, Ohio had no regulations or requirements for MCUs. However, before the 2017 change in government, House Bill 470 was signed into law and Senate Bill 283 was introduced, sponsored by Ohio Senator Capri Cafaro (Ohio House Bill 470, 2016; Ohio Senate Bill 283, 2016). Both Bills are related to MCU regulations and will be discussed in sections to follow. A joint presentation by Senator Cafaro and former Director of the Ohio Department of Aging, Bonnie Burman, at the 2016 Ohio Association for Gerontology and Education Conference discussed the potential need for MCU regulations and increased overall awareness on the need for research. The results of this study will provide needed background research for House Bill 470 and Ohio Senate Bill 283. The current lack of regulations to guide MCUs in Ohio allows for potential inconsistencies among facilities in terms of what is provided (or not provided) to adults with AD. Although regulations may not be necessary, quantitative analysis of state-wide nursing facility data is needed to assess whether expectations are being met and if inconsistencies exist. The purpose of this research study is to compare Ohio nursing facilities with MCUs to those without MCUs, and determine what attributes are significantly associated with a nursing facility operating a MCU. This study will also assess any change in the number of MCUs in Ohio over an eight-year period. A review of previous research was conducted to understand the purpose of MCUs for nursing home residents with memory loss, how MCUs compare to traditional units in care outcomes, and what characteristics are associated with a facility operating a MCU. It is important for Ohio policy makers to understand the characteristics of Ohio memory care units and what they provide to determine if regulations are needed. Additionally, consumers need to know more about MCUs in order to determine if this type of care is best suited or necessary. Describing characteristics of nursing facilities with MCUs, and comparing characteristics of nursing facilities with and without MCUs, will provide a needed profile for consumers, policy makers, and researchers. This study will determine what characteristics are significantly associated with operating a MCU, while controlling for other factors. Recommendations for consumers, policy makers (particularly regarding House Bill 470 and Senate Bill 283), and future research are provided. Background Older adults (65+) in the United States are the fastest growing segment of the population (Werner, 2011). With Americans living longer, their risk of disease and comorbidities increases (Niccoli & Partridge, 2012). Alzheimer s disease is becoming increasingly prevalent in the United States, with 1 in 9 older Americans having AD or another type of dementia (Alzheimer s Association, 2016). In 2016, there were approximately 5.4 million Americans with Alzheimer s disease (Alzheimer s Association, 2016). Due to the prolonged nature of AD, the costs to the healthcare system are staggering. Estimates suggest that $236 billion were spent in 2016 on healthcare and long-term care costs for people with AD or other types of dementia alone (Alzheimer s Association, 2016). The Alzheimer s Association (2016) projects that the number of individuals with the disease may nearly triple by 2050, a projected 13.8 million people over the age of 65 with the disease. In other words, almost 16% of the older population (65+) are estimated to have AD by 2050 (Vincent & Velkoff, 2010; Alzheimer s Association, 2016). 2
Ohio has the seventh largest older population in the United States, home to over 1.7 million people over the age of 65 (Ritchey, Mehdizadeh, & Yamashita, 2012). Long-term care costs in Ohio continue to rise, and Medicaid, a primary payer of long-term care, continues to account for a growing percentage of Ohio s total state expenditures. In 2012, approximately $2.6 billion of Ohio s Medicaid spending was allocated to long-term care services and supports (Nelson, Applebaum, Mehdizadeh, & Straker, 2015). This is a result of the estimated two-thirds of Ohio s nursing home residents relying on Medicaid (Nelson et al., 2015). In 2016, Ohio s Medicaid spending for adults with AD almost exceeded the entire 2012 Medicaid spending on long-term care services and supports, nearly $2.3 billion (Alzheimer s Association, 2016). AD is a growing concern nationally and in Ohio. It is becoming an increasing portion of Medicaid spending and total state expenditures in both Ohio and the rest of the United States. High healthcare costs are due in part because people with AD are much more likely to spend some amount of time living in a nursing facility than adults without AD. It is estimated that nearly two-thirds of people who die from dementia do so in a nursing facility (Mitchell, Teno, Miller, & Mor, 2005). In addition, over 50% of nursing home residents in the United States have dementia and approximately 61% have mild to severe cognitive impairment (Harris- Kojetin et al., 2016; Centers for Medicare and Medicaid Services, 2015). Ohio has the third most certified nursing facilities in the country, fewer than only Texas and California (Centers for Medicare and Medicaid Services, 2015). In terms of the number of certified nursing home beds, Ohio has the fifth most beds of any state (American Health Care Association [AHCA], 2015). In summary, Ohio has a large population of older adults with AD or dementia living in long-term care facilities, requiring both Medicaid coverage and many specific care needs. Memory Care Units MCUs were designed to provide specialized, dementia-specific skilled nursing care to adults with AD or dementia. Researchers have stated that specialized care in these units is considered fundamental to providing care (Grant et al., 1996). In order to provide specialized care, these units are expected to provide aspects such as disease-specific staff training, higher staffing ratios, and activities for residents with dementia (Gruneir, Lapane, Miller, & Mor, 2008). Most states with MCU regulations require additional staff training and some require specialized activities (U.S. Congress, 1992). In the early 1990 s, a national study of medical expenditures found that 74% of nursing facilities with SCUs provided special training for unit staff (U.S. Congress, 1992). In addition, challenges such as resident outburst or actions may be caused by insufficient activities and inappropriate staff responses (U.S. Congress, 1992). These problems could be reduced or mitigated by specialized activities and additional staff training for dementia or behavioral management. In order to provide enhanced memory care, it is important for MCU staff to have additional dementia-specific training and specialized activities for residents with memory problems (Grant et al., 1996; Gruneir et al., 2008). Researchers have compared residents with dementia in SCUs to those in traditional nursing facilities in terms of many indicators and outcomes, typically finding mixed results. Multiple studies found that residents with dementia in SCUs are less likely to have bed rails, and more likely to have toileting/incontinence plans than those in traditional units (Gruneir et al., 2008; Luo et al., 2010). However, Gruneir and colleagues (2008) found that residents with dementia in SCUs are more likely to be prescribed psychotropic medications. Similarly, another study found that SCU residents with AD were significantly more likely to receive anti- 3
depressant, anti-psychotic, or anti-anxiety medication on a daily basis compared to non-scu residents with AD (Buchanan et al., 2005). These studies suggest that there are inconsistencies between residents with AD in SCUs/MCUs and residents with AD in traditional nursing home units. Characteristics of SCUs have also been explored. In 1987, less than 10% of nursing facilities had a SCU (Leon, Potter, & Cunningham, 1991). In the same study, 60% of nursing homes with a SCU were for-profit, with a majority of those as a part of a multi-facility ownership. Forty-five percent of nursing homes with SCUs had less than 100 total beds, and 26% had over 150 beds. The 1987 study found that only 40% of all nursing facilities were certified, however, 75% of nursing homes with SCUs were certified nursing facilities (Leon et al., 1991). Additionally, researchers determined what nursing home characteristics lead to innovations such as Alzheimer s special care units. Banaszak-Holl and associates (1996) found that competition creates incentive for nursing facilities to innovate and create SCUs, specifically Alzheimer s special care and subacute care units. In addition, larger facility size, non-profit ownership status, being chain affiliated, and fewer Medicare patients were predictors of nursing facilities with Alzheimer s special care units (Banaszak-Holl, Zinn, & Mor, 1996). Interestingly, research on early adopters (first 20%) of innovations, such as SCUs, shows similar findings. Facilities with large bed size, high levels of private-pay residents, chain affiliation, in competitive markets, with high average income in the county are more likely to adopt innovations such as special care units (Castle, 2001). In addition, Castle (2001) suggested that facilities with an Alzheimer s SCU may attract more private-pay residents, although this was not examined in their study. One explanation is that facilities may innovate to create an Alzheimer s SCU with the expectation to make more money by increasing the cost for private-pay residents. These studies aid in understanding what factors may be associated with a nursing home operating a MCU. The American Health Care Association has collected data on the number of SCU and MCU beds both in Ohio and nationally. SCU beds are a rather small percentage of all nursing facility beds, however; MCU beds make up the majority of all SCUs beds. In the United States in 2014, MCU beds accounted for only 4% of all nursing home beds, yet accounted for 71% of all SCU beds (AHCA, 2014). In Ohio, MCU beds account for about 5.3% of all nursing home beds but nearly 80% of all SCU beds (AHCA, 2015). Interestingly, the overall number of MCU beds in the United States decreased by 3% from 2013 to 2014; but returned with a 1.8% increase from 2014 to 2015 (AHCA, 2014; AHCA, 2015). The number of MCU beds in Ohio increased by nearly 29%, from 3,751 beds in 2014 to 4,836 beds in 2015 (AHCA, 2014; AHCA, 2015). Regulations Regulations for MCUs vary by state. Prior to 1993, only six states had established regulations for SCUs, and five states were in the process of developing regulations (U.S. Congress, 1992). A few other states established mandates and committees for creating regulations for SCUs. Ohio was on the other end of the regulation discussion. Ohio and at least three other states were strongly opposed to the development of regulations in the 1990s (U.S. Congress, 1992). In addition, Ohio was one of six states who eased the process for creating SCUs by modifying the certificate of need process which is required for creating additional nursing home beds (U.S. Congress, 1992). The climate of regulations has changed since the 4
1990 s, and now more states have developed, or are in the process of developing, MCU regulations (Burke & Orlowski, 2015; Oregon Administrative Rules, 2010; The Commonwealth of Massachusetts, 2014; Ohio House Bill 470, 2016). Prior to the inception of this study, Ohio had no regulations specific to SCUs or MCUs. The recently retired director of the Ohio Department of Aging (ODA), Bonnie Burman, discussed the need for MCU requirements at the 2016 Ohio Association of Gerontology and Education Conference. Prior to 2017, Ohio policy makers were discussing the potential need for regulations of MCUs, but this was not a pressing issue for ODA or the Ohio Government. However, this is now a pressing issue for the newly appointed ODA director, Stephanie Loucka, as Ohio Governor John Kasich signed House Bill 470 into law in December 2016. The bill requires the development of MCU recommendations within six months of March 21, 2017, but does not state any specific recommendations/regulations (Ohio House Bill 470, 2016). Concurrently, Ohio Senate Bill 283, presenting specific MCU requirements, is in the committee action stage after introduction in April 2016 (Ohio Senate Bill 283, 2016). Senate Bill 283 proposes regulations in the following areas: approval process and applications, physical space, activities, ADL (activities of daily living) and IADL (instrumental activities of daily living) assistance, admission process, and inspections and disciplinary actions (Ohio Senate Bill 283, 2016, p. 4-8). In addition, Senate Bill 283 defines a memory care unit as a unit within a nursing home that provides or proposes to provide specialized care and services for residents with Alzheimer s disease or other dementia (Ohio Senate Bill 283, 2016, p. 4). With the passing of House Bill 470 and the introduction of Senate Bill 283, it is now only a matter of time until Ohio has MCU regulations. Many other states currently have MCU regulations in place. For example, Oregon has rather extensive regulations for MCUs that went into effect in 2010. The purpose of these regulations is to ensure that residents living in memory care [units] have positive quality of life, consumer protection, and person directed care (Oregon Administrative Rules, 2010, p.1). Regulations in Oregon include topics regarding applications for a MCU, advertising, administration responsibilities, extensive staffing and staff training requirements, services provided, and environment/design and safety, among others (Oregon Administrative Rules, 2010). Other states, such as Massachusetts, have less comprehensive requirements for MCUs. Massachusetts regulations emphasize additional staff training and activities, along with physical space and advertising requirements. Staff training requirements include at least eight hours of specialized training upon hire, and an additional four hours of dementia-specific training annually. In addition, at least eight hours of therapeutic activities must be provided daily, with the goal of maintaining physical and emotional wellness (The Commonwealth of Massachusetts, 2014). In 2015, 14 states had MCU laws for dementia-specific staff training and an additional nine states had these laws for all nursing facility staff (Burke & Orlowski, 2015). MCU marketing and advertising regulations were developed to ensure that MCUs are providing the specialized care they are claiming (and expected) to provide. Many MCUs across the country mislead consumers about the type of care that can be provided (Lazar, 2015). The president of the Massachusetts and New Hampshire chapter of the Alzheimer s Association, James Wessler, spoke out about the danger of this misinformation. People s lives are at risk this is misinformation to the general public, and it s one of the things we wanted to stop having nursing homes claiming they have special care and not be in compliance with minimal standards of dementia care stated Mr. Wessler (Lazar, 2015, p.1). This demonstrates the need for ensuring honest marketing by MCUs, and the need for monitoring facilities and applying sanctions for noncompliance. 5
Numerous contributing factors have led to the situation we have today. Many states have MCU regulations or requirements in place, other states, like Ohio, are in the process of developing MCU regulations, and some states have no MCU regulations. The results of this study will inform consumers about Ohio MCUs, make recommendations to Ohio policy makers about MCU regulations, and highlight potential future research projects. This study will attempt to answer the following research questions: Research Questions: 1. Did the number of memory care units in Ohio change between 2005 and 2013? 2. What do nursing facilities with and without a memory care unit look like in terms of location, profit status, chain affiliation, facility size, occupancy rate, days paid for by Medicaid, private pay reliance, private pay daily rate, surrounding poverty, amount of competition, staff training, and specialized activities? 3. What characteristics are significantly associated with an Ohio nursing facility operating a memory care unit? Chapter II Methods Data from the 2005 and 2013 Ohio Biennial Survey of Long-Term Care Facilities were used in this study. This is an online survey developed and administered by Miami University s Scripps Gerontology Center, supported by the Ohio Department of Aging. Ohio law requires participation in this survey, leading to a high response rate. This survey was completed by 95% of all nursing facilities in the state of Ohio in 2013 (N=919) and 94% (N=899) in 2005. The survey is now in its 12th wave and is conducted every two years. It is emailed to nursing home administrators and relies on self-reported data. Survey data are at the provider (nursing home) level, and the survey does not include any information on individual residents. Ohio Census data from 2010 and 2015 were also incorporated in this study to determine the poverty rate and 2013 estimated population 65+ of each Ohio county. Measures The variables used in this study are presented and summarized in Table 1. Further variable descriptions follow. Memory Care Unit (MCU) The memory care unit question of the 2013 Biennial Survey was asked among a list of questions related to special care units. Facilities were asked to indicate what type of special care unit (if any) that their nursing home provides, with dedicated dementia or memory-care unit as one of ten options. Nursing homes located within a hospital were removed from the data because these units typically provide rehabilitation or transitional care and were not the population of interest. Memory care units in the 2005 survey were defined as a secured and locked unit for residents with Alzheimer s or dementia. 6
Location Location was determined by the Scripps Gerontology Center based on the county in which the facility was located. Each location was identified as urban or rural. Profit Status Profit status categories included for-profit, non-profit, and government owned. Government owned facilities were combined with non-profit facilities for multiple reasons. First, government owned facilities are not for profit. Second, there were only 22 government owned facilities in 2013 and therefore the sample size warranted combination with non-profit. Last, previous research often combines government owned and non-profit facilities (Holmes, 1996; Nelson et al., 2015). Chain Affiliation To assess chain affiliation, facilities were asked to indicate if they were owned or leased by a multi-facility organization (two or more nursing homes in different locations). Response categories were yes or no. Facility Size Facility size, or number of beds, was reported as the number of Ohio Department of Health licensed beds on the last day of 2013. Facility size was broken into three categories as suggested by previous literature (Banaszak-Holl et al., 1996; Castle, 2001): small (1-50 beds), medium (51-100 beds), and large (101 or more beds). Occupancy Rate Facility occupancy rate was calculated using several other variables in the survey. The survey asked the size of the facility (number of beds) and calculated total resident days (number of resident days per year). To calculate occupancy rate, the number of beds in the facility was first multiplied by 365 and rounded to the nearest whole number (total bed days). Next, the number of total resident days (rounded to the nearest whole number) was divided by the total bed days to obtain the ratio of occupied beds to total beds. Finally, values greater than 1.00 (100%) were recoded to equal 1.00 and values below 0.50 were set to missing due to facility reporting issues. Medicaid Proportion The total number of resident days per year paid for by Medicaid was included in the dataset. This measure was created by calculating the proportion of days paid for by Medicaid to the number of total resident days. Therefore, Medicaid days is the facilities percentage of total resident days paid for by Medicaid. Private Pay Proportion Private pay days were the total number of resident days per year paid for privately by the resident (out-of-pocket payment). This measure was asked in the payment source section of resident days. Private pay days were also the proportion of resident private pay days to total resident days, and are presented as a percentage of each facilities days paid for privately. Private Pay Daily Rate Facilities were asked to indicate the average daily rate for a private pay (non-insurance) single and shared rooms in the last half of 2013. The averages for single and shared rooms were combined and averaged. The rate is rounded to the nearest whole dollar value. Percentage of County in Poverty This measure was not asked in the Biennial survey but was derived from the 2010 United States Census Data. It was the percentage of the population in each Ohio County that was below 100% of poverty in 2010. The 100% poverty income level in 2010 was $14,218 for a household 7
of two (increasing for each additional person or related child), therefore, this is the percentage of the population in each county under the rate based on family size (U.S. Census Bureau, 2010). These data were merged with the Biennial survey data based on county name, and the percentage of the county in poverty was assigned to facilities in each respective county. County Bed Competition County bed competition was conceptualized to represent competition in the area surrounding the nursing facility. Ohio has a limited number of available nursing home beds in each county, therefore, a county-level ratio of the number of beds to the older adult population in the county serves as a proxy measure of competition. This measure was calculated using the 2010 and 2015 Ohio Census data and the number of nursing home beds per county available through Scripps Gerontology Center. County populations over 65 in 2013 were calculated by creating a one year estimated population increase between 2010 and 2015, and then adding the one year increase three times (to the 2010 numbers) to get the 2013 estimate. The estimated bed to population ratio was calculated by dividing the number of nursing home beds in the county by the estimated 2013 population, and then multiplying by 1,000. This gives the number of nursing home beds per 1,000 population 65 plus in each Ohio County, which is an established proxy for nursing home competition (Mehdizadeh & Applebaum, 2009). Higher values indicate more competition. Behavioral Management Staff Training Behavioral management staff training for residents with dementia was asked in a module of training questions. Facilities were asked to indicate which type of staff, RNs/LPNs, STNAs, other direct-care staff, and all other staff, have received training for behavioral management of residents with dementia. Facilities may check all that apply or check none. Specialized Activities for Residents with Memory Loss A question about specialized activities for residents with memory problems was asked of all facilities in an individualized care module. Facilities were asked to indicate if they offered activities designed for residents with memory problems, with the following selection options: yes, this describes our facility, this partially describes our facility; we have this in progress, or no, this does not describe our facility. 8
Table 1. Measures and Descriptions Variable Definition Memory Care Unit (MCU) Did the facility have a Dementia/Memory Care Unit Location Was the facility located in an urban or rural area Profit Status Proprietary (For-profit), Non-Proprietary (Non-Profit and Government owned) Chain Affiliation Was the facility owned or leased by a multifacility organization Facility Size Number of beds in each nursing facility that were state licensed nursing home beds Occupancy Rate Ratio of number of occupied beds to the number of total beds Medicaid Proportion Total Medicaid day per year divided by total resident days per year Private Pay Proportion Total private pay days per year divided by total resident days per year Private Pay Daily Rate The facilities average private pay daily rate (average of single and shared rooms) Percentage of County in Poverty The percentage of the county in which the nursing home is located that was in poverty (2010) County Bed Competition Number of nursing home beds per 1,000 population 65+ by county (2013) Behavioral Management Staff Training Were the facilities RNs/LPNs, STNAs, other direct care staff, and all other staff trained for behavioral management of residents with dementia Specialized Activities for Residents with Did the facility offer specialized activities for Memory Loss residents with memory loss Analysis This study was divided into three primary analyses and one supplemental analysis. All statistical analyses were performed using SAS software version 9.4. In the first primary analysis, the change in the number of MCUs over eight years is described. Frequencies were used to determine the number and proportion of Ohio nursing facilities with MCUs in 2005 and in 2013. This description allows for understanding a potential change in the number and proportion of nursing facilities who operated a MCU. For the second analysis, descriptive statistics were employed to compare nursing facilities with a MCU to facilities without a MCU. The goal of this analysis is to understand and develop a profile of nursing facilities with MCUs compared to those without. Frequencies were used to describe the number and proportion of nursing facilities with MCUs on all key variables. The third primary analysis supplements the second analysis by controlling for all factors simultaneously. Logistic regression was used to determine which independent variables were 9
significantly associated with the presence of a MCU. A supplemental post hoc missing data analysis is also presented prior to analysis three. That analysis describes facilities excluded from analysis three (the logistic regression model) due to missing data and list-wise deletion. Two hundred and fifty-seven facilities were excluded from the model for having missing information for at least one of the variables in the model. The dependent variable in the logistic regression model was the presence of a MCU within the nursing facilities. This model assessed all the variables from the second analysis as independent variables for the third analysis. The analysis also examined the issue of multicollinearity, and excluded private pay proportion for this reason. Significance was set at p<0.05. Analysis 1: Change in Number of MCUs Chapter III Results Results (Table 2) indicate a decline in the number of MCUs between 2005 and 2013. Among the 899 Ohio nursing homes with data in 2005, 320 (35.60%) facilities had a MCU. In 2013, 302 of the 859 (35.16%) Ohio nursing homes with data operated a MCU. The number of facilities in 2013 was reduced due to non-response for the MCU question and the removal of hospital based facilities. Overall, there was a 1.24% decrease in the number of MCUs between 2005 and 2013, or an average decrease of about 0.15% each year. Table 2. 2005 vs. 2013 Ohio MCUs Year Total Valid NHs Number of MCUs % of Included 2005 899 320 35.60% 2013 859 302 35.16% Note. NH= nursing home. Analysis 2: Facilities with a MCU vs. Facilities without a MCU The results from the 2013 descriptive comparison of Ohio nursing homes with MCUs to nursing homes without MCUs are presented in Table 3. As shown in this table, 557 (64.8%) nursing facilities did not operate a MCU, and 302 (35.2%) facilities operated a MCU. The majority of all nursing facilities in Ohio were located in urban areas, 77.8% of facilities with MCUs were in urban areas and 74.7% of facilities without MCUs were in urban areas. Similarly, the majority of nursing facilities were for-profit. For facilities with MCUs, 73.2% were for-profit and 26.8% were non-profit. As for non-mcu facilities, 82.2% were for-profit and 17.8% were non-profit. Chain organizations own 69.0% of nursing homes with MCUs, and 70.5% of nursing homes without MCUs were owned by a chain. Facility size was analyzed in three categories, small (1-50 beds), medium (51-100 beds) and large (101 or more beds). Small facilities make up the smallest proportion of facilities with MCUs; only 19 small facilities operate a MCU (6.3% of MCUs). Large facilities make up the largest proportion, 163 facilities or 54% of all facilities with a MCU were large. In fact, there were more large facilities with a MCU than large facilities without, even though there were almost twice as many facilities without a MCU. There were 163 large facilities with a MCU, and only 140 large facilities without a MCU. Of the facilities without a MCU, medium sized 10
facilities were most common, making up over 50% of these facilities. Larger facilities appear to be more likely to operate a MCU, compared to medium or small facilities. Results for behavioral management staff training for residents with dementia indicated that most facilities, with a MCU or without, did not have staff trained in this manner. For facilities with a MCU, over 95% did not have RNs/LPNs trained, 93% did not have STNAs trained, 73% did not have direct care staff trained, and 57% did not have all other staff trained. In other words, the majority of nursing homes with a MCU did not have their staff trained in behavioral management for residents with dementia. These findings were similar to nursing facilities without a MCU. In facilities that did not operate a MCU, 95% did not have RNs/LPNs trained, 93% did not have STNAs trained, 74% did not have direct care staff trained, and 63% did not have all other staff trained. A higher proportion of nursing homes with a MCU offered specialized activities for residents with memory loss compared to facilities without a MCU, with the majority of both offering special activities. Ninety-five percent of nursing homes in Ohio with a MCU offered these specialized activities, and all facilities with a MCU offered some level of specialized activities. About 77% of nursing homes without a MCU offered specialized activities, and 15 facilities did not offer specialized activities for residents with memory loss. 11
Table 3. Descriptive Statistics Comparing Nursing Homes (NH) with and without a MCU N NHs with MCU Freq % Of MCU NHs without MCU Ohio NHs 859 302 100 859 557 100 Location Urban 302 235 77.8 557 416 74.7 75.8 Rural 302 67 22.2 557 141 25.3 24.2 Profit Status For-Profit 302 221 73.2 556 457 82.2 79.0 Non-Profit 302 81 26.8 556 99 17.8 21.0 Chain Affiliation Yes 300 207 69.0 553 390 70.5 70.0 No 300 93 31.0 553 163 29.5 30.0 Facility Size <50 beds 302 19 6.3 557 123 22.1 16.5 51-100 beds 302 120 39.7 557 294 52.8 48.2 >100 beds 302 163 54.0 557 140 25.1 35.3 Behavioral MGMT Training RNs/LPNs Yes 299 14 4.7 554 30 5.4 5.2 No 299 285 95.3 554 524 94.6 94.8 STNAs Yes 297 21 7.2 536 36 6.7 6.8 No 297 276 92.9 536 500 93.3 93.2 Direct Care St Yes 299 80 26.8 554 147 26.5 26.6 No 299 219 73.2 554 407 73.5 73.4 Other Staff Yes 299 128 42.8 554 203 36.6 38.8 No 299 171 57.2 554 351 63.4 61.2 Specialized Activities Yes 298 283 95.0 540 415 76.8 83.3 Partial 298 15 5.0 540 110 20.4 14.9 No 298 0 0.0 540 15 2.8 1.8 Note. NH= nursing home; N= sample size; Freq= Frequency. The means of all continuous variables were also calculated and are presented in Table 4. Average occupancy rates were lower in homes without a MCU than facilities with a MCU. Nursing homes without a MCU had a mean occupancy rate of 85.51% (standard deviation [SD] = 10.81%), whereas nursing homes with a MCU had a mean occupancy rate of 86.43% (SD= 10.26%). The proportion of days paid for by Medicaid was higher for nursing homes with MCUs compared to those without. The mean days paid for by Medicaid for nursing homes with a MCU was 62.88% (SD= 15.00%), compared to 61.96% (SD= 16.65%) for nursing homes without a MCU. All nursing homes with a MCU had at least some day paid for by Medicaid (min= 6.98%), but at least one facility without a MCU had no days paid for by Medicaid (min= 0%). The proportion of private pay days was also higher for nursing homes with MCUs compared to those without. The mean proportion of private pay days was 26.73% for nursing facilities with a MCU (SD= 14.26%) and 25.24% for nursing homes without a MCU (SD= N Freq % Of Non % Of Valid NHs 12
14.27%). Both types of facilities (with and without a MCU) had at least one facility with only private pay residents (max= 100% for both). On the other hand, there were facilities without MCUs that did not have any private pay residents (min= 0%), but all nursing homes with a MCU had some private pay residents (min=3.47%). Nursing homes with a MCU had a higher daily private pay rate than facilities without a MCU. The mean private pay rate for facilities with a MCU was $231.15 a day (SD= $42.27), compared to $224.07 a day (SD= $42.77) for facilities without a MCU. The maximum private pay daily rate was over $400 for both facilities with and without a MCU. Consequently, nursing homes with a MCU may charge a higher daily private pay rate than nursing homes without a MCU. Higher values for county bed competition indicate greater competition, whereas lower values indicate less competition. As described earlier, this variable does not take into account individual facilities, it assigns a bed ratio to facilities based on the county within which they are located. The mean value for county bed competition for nursing homes with a MCU was 53.50 nursing home beds per 1,000 population 65 plus (SD= 11.61 beds), compared to 53.24 nursing home beds per 1,000 population 65 plus for nursing homes without a MCU (SD= 10.54). It appears that facilities with and without a MCU face comparable levels of bed competition. County poverty rate is also not specific to each facility, but is based on the county in which they are located. The mean poverty rate for facilities with a MCU was 11.68% poverty (SD= 2.51%), whereas the mean poverty rate for nursing homes without a MCU was 11.91% poverty (SD= 2.37%). Both groups of nursing homes had a maximum county poverty rate of 16.27% and a minimum of 6.50%. Table 4. Descriptive Statistics of Continuous Measures- NH with MCU vs. NH without MCU NHs with MCU NHs without MCU N Mean/% SD Min Max N Mean/% SD Min Max Occupancy Rate 302 86.43% 10.26 54.16 100.0 557 85.51% 10.81 50.92 100.0 Medicaid Prop. 302 62.88% 15.00 6.98 91.56 577 61.96% 16.65 0.00 97.43 Private Pay Prop. 302 26.73% 14.26 3.47 100.0 577 25.24% 14.27 0.00 100.00 Private Pay Daily Rate 302 231.15 42.27 150.0 405.0 557 224.07 42.77 135.0 401.0 County Bed Competition 302 53.50 11.61 19.35 109.36 577 53.24 10.54 15.92 109.36 County Poverty 302 11.68% 2.51 6.50 16.27 577 11.91% 2.37 6.50 16.27 Note. NH= nursing homes; N= sample size; SD = standard deviation; Prop= Proportion. Missing Data Analysis Logistic regression utilizes list-wise deletion and removes any cases with missing data for any variable in the model. A description of which measures had missing data is provided first, followed by statistical comparisons. Four facilities chose not to complete the survey at all. Seventy-four nursing homes did not report if they operated a MCU. Four facilities also had missing data for location. In regards to profit status, five facilities had missing data. Forty-one nursing homes did not have data for chain affiliation. Only four facilities did not report their facility size. Thirty-five facilities had missing data for occupancy rate. Twenty-seven facilities 13
had missing data for proportion of days paid for by Medicaid and only four had missing data for private pay proportion. The measure with the most missing data was private pay daily rate. One hundred and sixty-five nursing facilities did not report a private pay daily rate. The four facilities that had missing county data subsequently had missing data for county poverty rate and county bed competition. Eighty-five facilities had missing data for the specialized activities measure. Finally, a total of 91 nursing homes had missing data for the staff training in behavioral management measure. In total there were 257 facilities excluded from analysis three. Missing data analysis determined whether any differences existed between facilities included in the logistic regression model and those excluded from the model due to missing data. Missing data were analyzed using chi-square tests for categorical variables and independent samples t-tests for numeric variables. This analysis indicated that statistically, facilities excluded from the model were significantly different from facilities in the model for the presence of a MCU, profit status, chain affiliation, facility size, behavioral management training, and specialized activities. Results are presented in Table 5. Significantly more facilities without a MCU were excluded from the model than facilities with a MCU (p<0.01). One hundred thirty eight facilities without and 45 facilities with an MCU were excluded from the logistic regression model due to missing data. Most facilities with a MCU, 257 of the 302 (85.1%), were included in the final model. 14
Table 5. Analysis of Missing Data Missing (Freq) Percent of Missing Included (Freq) Percent of Included Total NHs 257 100% 676 100% MCU Yes 45 24.59% 257 38.02% Sig. (p-value) <0.001*** No 138 75.41% 419 61.98% Location 0.064 Urban 203 80.24% 503 74.41% Rural 50 19.76% 173 25.59% Profit Status For-Profit 181 71.83% 556 82.25% Non-Profit 71 28.17% 120 17.75% Chain Affiliation Yes 136 62.96% 482 71.30% <0.001*** 0.021* No 80 37.04% 194 28.70% Facility Size <0.001*** <50 beds 70 27.67% 90 13.31% 51-100 beds 116 45.85% 329 48.67% >100 beds 67 26.48% 257 38.02% Behavioral MGMT Training RNs/LPNs Yes 10 6.02% 34 5.03% 0.606 No 156 93.98% 642 94.97% STNAs 0.882 Yes 11 6.63% 47 6.95% No 155 93.37% 629 93.05% Direct Care St 0.034* Yes 55 33.13% 169 25.00% No 111 66.87% 507 75.00% Other Staff 0.072 Yes 74 44.58% 250 36.98% No 92 55.42% 426 63.02% Specialized Activities Yes 135 78.49% 568 84.02% 0.023* Partial 30 17.44% 100 14.79% No 7 4.07% 8 1.18% Missing- (Mean) Percent of Total (%) Included (Mean) Percent of Total (%) Sig. (p-value) Occupancy Rate 84.64% 24.72% 85.72% 75.28% 0.197 (N=222) (N=676) Medicaid Prop. 63.60% 25.39% 62.26% 74.61% 0.277 (N=230) (N=676) Private Pay Prop. 26.32% 27.23% 25.05% 72.77% 0.220 (N=253) (N=676) Private Pay Daily Rate 231.20 11.98% 226.10 88.02% 0.282 (N=92) (N=676) County Bed Competition 53.34 27.23% 53.35 72.77% 0.993 (N=253) (N=676) County Poverty 11.89% (N=253) 27.23% 11.84% (N=676) 72.77% 0.775 Note. Freq=Frequency; Prop= Proportion. *p < 0.05. **p < 0.01. ***p < 0.001. 15
Analysis 3: Logistic Regression Logistic regression was utilized to determine which nursing home characteristics were significantly associated with the presence of a MCU, while controlling for all other variables in the model. The model included all variables from analysis two, except those creating multicollinearity in the model. The proportion of private pay days was removed due to a high correlation with the proportion of days paid for by Medicaid (r= -0.85). The final model included 676 Ohio nursing homes (70.03% of all Ohio NHs; 73.64% of valid NHs), and 257 MCUs (85.01% of all MCUs), who responded to every question represented by a variable included in the model. Results are presented in Table 6. Multivariate logistic regression demonstrated that profit status, facility size, proportion of days paid for by Medicaid, county poverty rate, and specialized activities for residents with memory loss were significantly associated with an Ohio nursing facility operating a MCU in 2013. Overall, the regression model was statistically significant (Max-rescaled generalized R 2 = 0.236). Profit status was a significant characteristic of a facility operating a MCU. For-profit facilities had 0.33 times the odds of operating a MCU compared to non-profit facilities, while controlling for all other variables in the model (OR= 0.33, p<0.001). Large facility size (more than 100 beds) and medium facility size (51-100 beds) had significantly higher odds of operating a MCU compared to small facilities (less than 50 beds). Large nursing homes had 7.83 times the odds of operating a MCU compared to small nursing facilities (OR= 7.83, p<0.001). Medium size facilities had 3.02 times the odds of operating a MCU when compared to small facilities (OR= 3.02, p<0.001). In addition, large sized facilities had 2.56 times the odds of operating a MCU compared to medium facilities (OR=2.56, p<0.001). For each percentage point increase in the proportion of days paid for by Medicaid in the facility, the odds of operating a MCU were 6.01 times higher (OR= 6.01, p<0.01). Additionally, for each percentage point increase in county poverty rate, the odds of operating a MCU were 0.08% lower (OR= 0.92, p<0.05). Compared to facilities that offered specialized activities for residents with memory loss, those that stated that they only partially offered these activities had 0.21 times the odds of operating a MCU (OR= 0.21, p<0.001). The odds of having a MCU did not significantly differ between those that did offer specialized activities, and those that did not (p= 0.98). Similarly, the odds of having a MCU did not significantly differ between facilities that partially offered and did not offer specialized activities (p= 0.99). None of the other variables in the model, location, chain affiliation, staff training in behavioral management, county bed competition, private pay daily rate, or occupancy rate demonstrated significance. 16
Table 6. Results of Logistic Regression Model Param. Est. Sig (P-value) Presence of MCU SE OR OR 95% CI Lower Upper Location (ref=urban) Rural 0.18 0.418 0.23 1.20 0.77 1.87 Profit Status (ref=non-profit) For-Profit -1.12 <0.001*** 0.26 0.33 0.20 0.54 Chain Affiliation (ref=no) Yes -0.09 0.675 0.20 0.92 0.62 1.37 Facility Size (ref= <50 beds) 51-100 beds >100 beds 1.10 <0.001*** 0.33 3.02 1.59 5.75 2.06 <0.001*** 0.35 7.83 3.98 15.42 Occupancy Rate 0.33 0.709 0.88 1.39 0.25 7.74 Medicaid Proportion 1.79 0.009** 0.69 6.01 1.57 23.12 County Poverty -0.08 0.044* 0.04 0.92 0.85 1.00 County Competition 0.01 0.410 0.01 1.01 0.99 1.03 Private Pay Daily Rate 0.00 0.295 0.00 1.00 1.00 1.00 Behavioral MGMT Training (ref=yes) RNs/LPNs STNAs Direct Care Staff -0.01 0.06 0.10 0.988 0.921 0.688 0.71 0.60 0.25 0.99 1.06 1.11 0.25 0.33 0.68 Other Staff -0.36 0.080 0.21 0.70 0.47 1.04 Specialized Activities (ref=yes) Partially -1.58 <0.001*** 0.31 0.21 0.11 0.38 No -14.91 0.983 715.50 <0.01 <0.01 >999.99 Note. SE = standard error; OR = odds ratio; CI= confidence interval; Generalized R 2 = 0.173, and max-rescaled generalized R 2 = 0.236. *p < 0.05. **p < 0.01. ***p < 0.001. 3.95 3.40 1.80 Chapter IV Discussion The purpose of this study was to develop a profile of Ohio MCUs in order to provide insight about whether or not regulations are appropriate and needed for these facilities. This was determined by examining three research questions: the change in number of Ohio MCUs over time, a comparison of nursing facilities with and without MCUs, and determining which facility characteristics were significantly associated with a nursing home operating a MCU while controlling for all other factors. Using the 2005 and 2013 Ohio Biennial Survey of Long-Term Care Facilities data sets, this study examined common nursing home characteristics along with variables found to be important in previous research. This study found results that both supported and differed from previous research on MCUs. First, the number of MCUs declined by 18 units from 2005 to 2013. As a result of missing data, the number of valid nursing homes for analysis one decreased between 2005 and 2013 by 40 facilities. However, there were actually more nursing homes in 2013. The survey was sent to 962 facilities in 2013 and 949 facilities in 2005. It is likely that the number of MCUs decreased by an even greater percentage between 2005 and 2013, but this is not definitive due to missing data. Aggregate national level data from the American Health Care Association (2014) 17