P01 AcademyHealth Presentations

Transcription

1 June 25, 2016 P01 AcademyHealth Presentations Interest Group Poster Presentation: Gadbois, E.A., Tyler, D.A., & Mor, V. Patients Experiences Transitioning to Post-Acute Care in Skilled Nursing Facilities. June 25, 12:30-2:30PM. Poster: LTSS273. Research Objective: The purpose of this research was to understand the experience of patients during the hospital discharge planning and skilled nursing facility (SNF) placement process, including the information patients use to make decisions, which factors they view as important in determining their selection, and barriers and facilitators to the overall process. Study Design: We interviewed 97 SNF patients in five health care markets across the country. These five markets varied based on Medicare Advantage penetration and geographic distribution. We recruited three SNFs within each market (two in the smallest market), and in each facility conducted semistructured open-ended interviews with approximately 7 previously community-dwelling, newlyadmitted patients discharged from a hospital. These interviews were recorded, transcribed, and qualitatively coded to identify underlying concepts and themes. Population Studied: 97 SNF patients and/or their family members. Principal Findings: Despite different reasons for hospitalization and hospital lengths of stay, most respondents reported that they were required by hospital discharge planners to make decisions regarding SNF placement the day before their hospital discharge. Respondents generally only received a list of SNF options which included just the name and address of each facility. This list was often limited by location to just a few choices, but some respondents reported receiving a list of every facility in their geographic area (i.e. several pages of facilities). In most cases, hospital staff (i.e. discharge planners and physicians) were minimally involved in the decision-making process, but respondents reported appreciating such help when it was received. Common predictors of patients decisions included location (not surprising given that the address was often the only information patients were given) and previous experience (either a prior stay or knowing others who had prior stays). Some respondents reported relying on decision-making support from their family and friends (i.e. having family members make calls to facilities or go on tours); indeed, respondents without involved family seemed to have more difficult and stressful experiences. Most respondents reported being satisfied with their placement, but many stated that they would have been willing to travel further to a higher quality facility or one recommended by their doctor. Although respondents were generally satisfied with their placement, many described the process as overwhelming, stating that they did not know where or how to get help. In spite of market differences in how hospitals and SNFs collaborate, patients experiences in the discharge process were very similar. Conclusions: SNF placement is a stressful transition, occurring at a time when patients are physically vulnerable, and often without significant help or guidance from experienced professionals like hospital discharge planners. Most patients select a facility based on its location, simply because they are provided with no quality information or advice. Implications for Policy or Practice: Hospitals are increasingly responsible for the outcomes experienced by patients post-discharge, yet the discharge planning process remains rushed and chaotic. Hospitals could likely reduce readmissions and other adverse outcomes by helping patients choose better quality facilities most suited to their clinical needs.

2 Interest Group Poster Presentation: Tyler, D.A., Gadbois, E.A., & Mor, V. Flying Blind: Patient Choice of Post-Acute Care Provider & Hospital Accountability. June 25, 12:30-2:30PM, Poster: LTSS272. Research Objective: In this era of accountable care, hospitals have become responsible for care patients receive in the post-acute care (PAC) setting. Yet, little is known about how those needing PAC select a facility and even less is known about whether or how they use information on providers quality. Nor do we know much about the role of hospital discharge planners (DPs) in the PAC selection process. The purpose of this research was to explore how patients requiring PAC make decisions about which skilled nursing facility (SNF) to select and the role that hospital DPs play in the selection process. Study Design: We utilized a multiple case study approach. Eight markets throughout the US were chosen based on their managed care penetration rate (4 high/4 low). In each market we selected one hospital with a low readmission rate and one with a higher rate and 3 or 4 SNFs that received referrals from those hospitals. We interviewed approximately 20 clinical and administrative staff from the selected organizations in each market. In 5 markets, we returned to interview patients who had recently been admitted to one of our participating SNFs. All interviews were summarized immediately upon conclusion. Summaries were used to quantify participants responses in terms of the number of DPs who said they provided patients with quality data, the number who reported providing patients with only a list of facilities, the number of patients who reported receiving quality data, the number who reported using quality data and the number who would have gone elsewhere had their doctor recommended it or if they were told the quality was better. Population Studied: 97 patients recently admitted to SNF care and 16 hospital DPs. Principal Findings: Of the 77 patients who reported being involved in SNF selection, 57 received only a list of SNFs. Ten didn t receive lists because they already knew where they wanted to go; in all cases a SNF where they d previously been. In fact, 35 of the 97 patients reported previous stays at the same SNF. Only four patients reported receiving SNF quality data from DPs. This is consistent with responses from DPs, who reported not furnishing patients with data, but providing only lists of SNFs. DPs stated they had been told that patient choice statutes precluded them from providing any information beyond SNF addresses. Because the only data patients were given were addresses, this often resulted in patients simply selecting the SNF nearest their home and/or returning to a SNF where they d been. Yet, 73 patients said they would have traveled farther if a better SNF was recommended. Conclusions: Patient choice in PAC setting is apparently trumping hospital accountability. Further, belief that patient choice statutes preclude the sharing of quality data about SNFs means that patients choices cannot be based on quality. Implications for Policy or Practice: CMS must clarify what is meant by patient choice and allow hospitals to explicitly refer patients to SNFs they are willing to vouch for and work with to achieve better outcomes.

3 Interest Group Poster Presentation: Winblad, U., McHugh, J., Shield, R.R., Gadbois, E.A., & Tyler, D.A. How do hospitals that are part of ACOs lower readmission rates? June 25, 12:30-2:30PM, Poster: LTSS293. Research Objective: Many health care organizations in the U.S. are developing new integrated models of care in order to improve quality. One model, the Accountable Care Organization (ACO), consists of networks of care providers that voluntarily come together and are held accountable for the quality and costs for all care delivered to a given population under a contractual arrangement with Medicare. Recent studies demonstrate cost savings but also quality improvements, such as rapid reductions in readmissions to ACO hospitals. Still, the evidence is far from conclusive and more research, particularly on what hospitals do to decrease individual readmissions, is needed. This qualitative study investigates what measures ACO hospitals take to reduce readmissions for frail elderly with complex health care needs. Specifically, the study describes special programs and initiatives taken by the hospitals in connection to post-acute care settings, i.e. skilled nursing facilities (SNFs). Study Design: The study was conducted in eight different health markets throughout the U.S. Each market included two hospitals, one with low readmission rates and one with higher rates, and three or four SNFs that received patients from these hospitals. Altogether, 160 interviews were conducted with representatives from hospitals, SNFs and managed care organizations. In the present study, only data from the 64 hospital interviews were included, such as discharge planners, case managers and VPs of Strategy. Data from interviews with representatives in ACO hospitals were compared and contrasted with data from interviews with representatives in non-aco hospitals regarding measures taken to reduce readmissions from SNFs to hospitals. Population Studied: 16 hospitals; six participating in the Shared Savings program and/or the Pioneer Program and ten not participating in any ACO. Principal Findings: The results indicate that ACO hospitals, to a larger extent than non-aco hospitals, have introduced measures to improve care transitions and reduce readmissions. These measures can be summarized in three categories. First, ACO hospitals have created narrower networks with a few preferred SNFs that are selected based on specific requirements, such as geographic dispersion and star ratings. Second, within the preferred provider networks, ACO hospitals have increased communication and introduced formal meetings on a regular basis. At the formal meetings, overall statistics on readmission patterns and length of stay, as well as individual patient s readmissions, are discussed and assessed. Third, ACO hospitals have developed explicit programs and initiatives of varying types, such as nurse-to-nurse hand off, identifying high risk patients and getting patients to the right level of care (for instance through placing case managers at the ED). While some of these initiatives were also seen at non-aco hospitals they were not as common, and were not as structured and formalized. Conclusions: ACO-hospitals have taken active measures to lower readmissions through creating narrower networks of preferred providers and formalized programs. These developments were not as obvious in non-aco hospitals. Implications for Policy or Practice: The results indicate that an active approach can be taken by both ACO and non-aco hospitals to reduce readmissions. Additional research is needed to find out which of the initiatives and measures are most effective in reducing readmissions.

4 Interest Group Presentation: Kosar, C., & Thomas, K. Patients Admitted to Nursing Homes with Delirium are at High Risk for Adverse Outcomes. LTSS Interest Group, Oral Presentations Part 1, June 25, 3:15PM-4:15PM. Research Objective: Delirium is a serious clinical condition in older adults associated with high rates of adverse outcomes such as functional decline, high healthcare utilization, and mortality. Hospitalacquired delirium may last for several weeks, and ultimately some patients are discharged with the condition unresolved. Yet, delirium research is limited in settings other than acute care. Our goal was to estimate the rate in which patients with delirium are admitted to nursing homes for post-acute care and to determine whether rates of adverse outcomes were higher in patients admitted with delirium compared to those admitted without delirium. Study Design: Delirium was identified using the Confusion Assessment Method. Using a cross-sectional study design, we estimated the relative risk for 30-day mortality, 30-day readmission, and functional improvement associated with delirium presence on post-acute care admission using robust Poisson regression. Functional improvement was indicated by a decrease in the level of dependency in Activities of Daily Living. Models were adjusted for age, gender, comorbidity burden (measured with the Charlson Index), an indicator for whether the patient was terminally ill, and nursing home fixed effects. Analyses were repeated in a subgroup of patients with baseline dementia. Population Studied: 1,114,837 adults aged 65 years admitted for the first time to a nursing home for post-acute care during 2013, including 231,359 patients with baseline dementia. Principal Findings: Using the Minimum Data Set 3.0 resident assessments, we found that the rate of delirium on admission to nursing homes for post-acute care was 4%. Patients admitted with delirium had higher rates of 30-day mortality and readmission and a lower rate of functional improvement. Among patients admitted with delirium, the 30-day mortality rate was 11% compared with 4% in those admitted without delirium (adjusted Relative Risk [arr] = 2.24, 95% Confidence Interval [CI] = 2.17, 2.30). The 30-day readmission rate was 14% in patients admitted with delirium compared to 11.1% in those admitted without delirium (arr= 1.21, 95% CI= 1.24, 1.27). Among patients discharged to home, the rate of functional recovery was 61% in patients admitted with delirium and 71% in patients admitted without delirium (arr= 0.87, 95% CI= 0.86, 0.88). Results were similar in the subgroup of patients with baseline dementia. Conclusions: A substantial proportion of older adults are admitted to post-acute care with delirium. These patients were more likely to die or be readmitted to the hospital within 30 days and were also less likely to improve in physical function when discharged home. This finding was not accounted for by the presence of baseline dementia. Future research may be aimed to examine the care process for nursing home patients admitted with delirium. Implications for Policy or Practice: Our study raises the question of how vigorously delirium should be monitored and addressed in transitional care. Interventions to decrease the percentage of patients admitted with delirium may reduce the frequency of adverse outcomes, decrease healthcare utilization, and improve the goals of care.

5 Interest Group Presentation: McHugh, J., Shield, R.R., Winblad, U., Rahman, M., Gadbois, E.A., Zinn, J., Mor, V., & Tyler, D.A. Hospitals Response to Readmission Penalties Post-Discharge Care Management Approaches. LTSS Interest Group, Oral Presentations Part 1, June 25, 3:15PM-4:15PM. Research Objective: Readmission penalties have motivated hospitals to provide care management for 30 days post-discharge. However, the most effective solutions for a hospital to reduce its readmission rate are still unknown and the approach to care management varies by market and by hospital. While current readmission reduction research often focuses on patients discharged home, patients discharged to skilled nursing facilities (SNF) are at higher risk for readmission. The purpose of this research was to identify hospital post-discharge care management strategies in relation to patients discharged to home (with and without home health) and SNF. Study Design: We utilized a multiple case study approach. Eight markets throughout the US were chosen based on their Medicare managed care penetration rate (4 high/4 low). In each market we selected one hospital with a low readmission rate and one with a higher rate as well as three or four SNFs that received referrals from those hospitals. We also selected two of the top three MCOs in each market. We interviewed approximately 20 clinical and administrative staff from the selected organizations in each market, yielding about 160 interviews. This included approximately 50 interviews with hospital discharge planners, strategic planning executives, hospitalists and chief medical officers in 16 hospitals. We grouped hospitals based on whether their approaches to care management included SNF network development or was limited to patients discharged home only. Data from interviews with staff in hospitals with SNF networks were compared and contrasted with data from interviews with staff in hospitals without SNF networks. Population Studied: Sixteen hospitals from eight markets throughout the United States. Principal Findings: Twelve hospitals focused primarily on patients discharged home and four hospitals included SNF networks. The overall readmission rates for hospitals which included SNF network development decreased 4.1 percentage points from 17.7% in 2009 to 13.6% in 2013 compared to a 2.9 percentage point decrease from 17.5% to 14.6% for hospitals with limited care management approaches. More dramatically, when the rates were calculated for just those patients discharged to SNF, the rate decreased 6.9 percentage points from 21.1% to 14.2% for SNF network hospitals compared with only a 4.1 percentage point decrease from 21.2% to 17.1% for non-snf network hospitals. Interviews revealed hospitals with a limited approach to care management are highly focused on managing patients discharged to home or home health, often ignoring SNFs and sometimes suspending care management during SNF stays. In contrast, hospitals with SNF networks are recognizing the importance of SNF care in addition to managing patients discharged to the community. Interviews revealed that managing patients in SNF often included follow-up with hospital providers for SNF patients, shared electronic medical records and collaborative quality improvement initiatives. Conclusions: Hospitals with greater emphasis on care management for patients discharged to both home and SNF experienced greater reductions in their overall readmission rate, which appears to be driven by greater reductions in readmission rates of patients discharged to SNF. Implications for Policy or Practice: Hospitals should direct more resources to managing SNF patients post-discharge, including creating SNF networks. This will likely result in greater readmission reductions.

6 June 26, 2016 Podium Presentation: Gadbois, Tyler, Mor. Patients Experiences Transitioning to Post-Acute Care in Skilled Nursing Facilities. Presenting in Improving the Nursing Home Journey, from Facility Selection through Clinical Quality of Care, June 26, 10:30AM-12PM, Hynes Convention Center, 203. Research Objective: The purpose of this research was to understand the experience of patients during the hospital discharge planning and skilled nursing facility (SNF) placement process, including the information patients use to make decisions, which factors they view as important in determining their selection, and barriers and facilitators to the overall process. Study Design: We interviewed 97 SNF patients in five health care markets across the country. These five markets varied based on Medicare Advantage penetration and geographic distribution. We recruited three SNFs within each market (two in the smallest market), and in each facility conducted semistructured open-ended interviews with approximately 7 previously community-dwelling, newlyadmitted patients discharged from a hospital. These interviews were recorded, transcribed, and qualitatively coded to identify underlying concepts and themes. Population Studied: 97 SNF patients and/or their family members. Principal Findings: Despite different reasons for hospitalization and hospital lengths of stay, most respondents reported that they were required by hospital discharge planners to make decisions regarding SNF placement the day before their hospital discharge. Respondents generally only received a list of SNF options which included just the name and address of each facility. This list was often limited by location to just a few choices, but some respondents reported receiving a list of every facility in their geographic area (i.e. several pages of facilities). In most cases, hospital staff (i.e. discharge planners and physicians) were minimally involved in the decision-making process, but respondents reported appreciating such help when it was received. Common predictors of patients decisions included location (not surprising given that the address was often the only information patients were given) and previous experience (either a prior stay or knowing others who had prior stays). Some respondents reported relying on decision-making support from their family and friends (i.e. having family members make calls to facilities or go on tours); indeed, respondents without involved family seemed to have more difficult and stressful experiences. Most respondents reported being satisfied with their placement, but many stated that they would have been willing to travel further to a higher quality facility or one recommended by their doctor. Although respondents were generally satisfied with their placement, many described the process as overwhelming, stating that they did not know where or how to get help. In spite of market differences in how hospitals and SNFs collaborate, patients experiences in the discharge process were very similar. Conclusions: SNF placement is a stressful transition, occurring at a time when patients are physically vulnerable, and often without significant help or guidance from experienced professionals like hospital discharge planners. Most patients select a facility based on its location, simply because they are provided with no quality information or advice. Implications for Policy or Practice: Hospitals are increasingly responsible for the outcomes experienced by patients post-discharge, yet the discharge planning process remains rushed and chaotic. Hospitals could likely reduce readmissions and other adverse outcomes by helping patients choose better quality facilities most suited to their clinical needs.

7 Podium Presentation: Mitchell, S.L., Mor, V., Gozalo, P., Servadio, J., & Teno, J. The Decline in Feeding Tubes Insertion Rates in Nursing Home Residents with Advanced Dementia from Presenting in Improving the Nursing Home Journey, from Facility Selection through Clinical Quality of Care, June 26, 10:30AM-12PM, Hynes Convention Center, 203. Research Objective: Research conducted over two decades has failed to demonstrate any benefits of tube-feeding in advanced dementia. In recent years, experts and position statements by national organizations have advocated against this practice. This report s objective were to describe tube-feeding insertion rates among US nursing home residents with advanced dementia from and examine changes in state-level variation and racial disparities in this practice. Study Design: Cohort study using nationwide Minimum DataSet data from January 1, 2000 to December 31, Population Studied: In each year from , we identified nursing home residents with advanced dementia who had recently (prior 120 days) became totally dependent in eating. The proportions of these residents who did and did not get feeding tubes over the following 12 months were determined. Principal Findings: There was a gradual decline in feeding tube insertion rates among US nursing home residents with advanced dementia and eating problems from 2000 (11.1%) to 2014 (5.4%) (Table). In , insertions rates at the state-level categorized as high, intermediate, and low based on terciles were: high, 12.7%-29.4%; intermediate, 4.4%-11.9%; and low, 0%-4.2%. In , statelevel insertion rates decreased in each tercile: high, 4.8%-22.7%; intermediate, 1.2%-4.74; and low, 0%- 1.0%. From , feeding tube insertion rates were 8.2% and 34.7% in white and black residents, respectively. In , feeding tube insertion rates were 3.3% and 16.6% in white and black residents, respectively. Conclusions: Between 2000 and 2014, feeding tube insertion rates declined among nursing home residents with advanced dementia by 50%. Insertion rates decreased across states and for both black and white residents. However, considerable state-level variation and greater rates of feeding tube insertions in black versus white residents persisted over this time period. Implications for Policy or Practice: The findings suggest that the body of research and practice recommendations advocating against tube-feeding in advanced dementia is reflected in a marked overall decline in this practice. Federal policy initiatives that disincentivize the insertion of feeding tubes in these residents may be needed to reduce persistent regional and racial disparities.

8 Poster Presentation: McHugh, J., Shield, R.R., Winblad, U., Rahman, M., Gadbois, E.A., Zinn, J., Mor, V., & Tyler, D.A. Hospitals Response to Readmission Penalties Post-Discharge Care Management Approaches. Poster Session: Organizational Behavior and Management, June 26, 12:15PM-1:45PM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 529. Research Objective: Readmission penalties have motivated hospitals to provide care management for 30 days post-discharge. However, the most effective solutions for a hospital to reduce its readmission rate are still unknown and the approach to care management varies by market and by hospital. While current readmission reduction research often focuses on patients discharged home, patients discharged to skilled nursing facilities (SNF) are at higher risk for readmission. The purpose of this research was to identify hospital post-discharge care management strategies in relation to patients discharged to home (with and without home health) and SNF. Study Design: We utilized a multiple case study approach. Eight markets throughout the US were chosen based on their Medicare managed care penetration rate (4 high/4 low). In each market we selected one hospital with a low readmission rate and one with a higher rate as well as three or four SNFs that received referrals from those hospitals. We also selected two of the top three MCOs in each market. We interviewed approximately 20 clinical and administrative staff from the selected organizations in each market, yielding about 160 interviews. This included approximately 50 interviews with hospital discharge planners, strategic planning executives, hospitalists and chief medical officers in 16 hospitals. We grouped hospitals based on whether their approaches to care management included SNF network development or was limited to patients discharged home only. Data from interviews with staff in hospitals with SNF networks were compared and contrasted with data from interviews with staff in hospitals without SNF networks. Population Studied: Sixteen hospitals from eight markets throughout the United States. Principal Findings: Twelve hospitals focused primarily on patients discharged home and four hospitals included SNF networks. The overall readmission rates for hospitals which included SNF network development decreased 4.1 percentage points from 17.7% in 2009 to 13.6% in 2013 compared to a 2.9 percentage point decrease from 17.5% to 14.6% for hospitals with limited care management approaches. More dramatically, when the rates were calculated for just those patients discharged to SNF, the rate decreased 6.9 percentage points from 21.1% to 14.2% for SNF network hospitals compared with only a 4.1 percentage point decrease from 21.2% to 17.1% for non-snf network hospitals. Interviews revealed hospitals with a limited approach to care management are highly focused on managing patients discharged to home or home health, often ignoring SNFs and sometimes suspending care management during SNF stays. In contrast, hospitals with SNF networks are recognizing the importance of SNF care in addition to managing patients discharged to the community. Interviews revealed that managing patients in SNF often included follow-up with hospital providers for SNF patients, shared electronic medical records and collaborative quality improvement initiatives. Conclusions: Hospitals with greater emphasis on care management for patients discharged to both home and SNF experienced greater reductions in their overall readmission rate, which appears to be driven by greater reductions in readmission rates of patients discharged to SNF. Implications for Policy or Practice: Hospitals should direct more resources to managing SNF patients post-discharge, including creating SNF networks. This will likely result in greater readmission reductions.

9 Poster Presentation: Rahman, M., & Mor, V. The Role of Medicare Advantage Plans in Beneficiaries Switching to Traditional Medicare after Nursing Home Care Use. Poster Session: Organizational Behavior and Management, June 26, 12:15PM-1:45PM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 534. Research Objective: While policy changes fueled rapid growth of Medicare Advantage (MA) program during the last decade, high switching rate to traditional Medicare among MA beneficiaries with high cost care use remained a key concern. This paper focuses on nursing home care users, one of the most medically needy Medicare population, to examine two relationships: 1) how switching varies with MA contract characteristics and 2) whether a preferred provider relationship between a contract and a nursing home measured by the fraction of the contract s nursing home patients going to a specific nursing home affects switching. Study Design: Using Medicare enrollment file, we identified MA enrollees in January 2010 and compared their enrollment in January 2011 to identify switching. We identified the enrolled MA contract in 2010 using HEDIS data and merged MA contract characteristics in fall of 2009 reported by CMS. We used Minimum data set for 2010 to categorize patients based on length of nursing home stay: no stay, short stay and long stay. Since nursing homes may have incentive to influence patients to switch because of higher payment rate and more independence while treating patient, all our analyses incorporated nursing home fixed effects. Our first analysis examined association between switching and MA contract characteristics controlling for patient s demographic characteristics, Medicaid eligibility and nursing home fixed effects. Our second analysis examined the relationship between the fraction of a contract s nursing home admissions going to a nursing home and switching among patients in that nursing home contract pair adding MA contract fixed effects to our first model. Population Studied: Age 65+ enrollees in about 342 MA contracts in January 2010 who were still enrolled in Medicare as of the following January. Principal Findings: The switch rate to traditional Medicare was 4% among MA beneficiaries without any nursing home use, 8% among those with short nursing home stay and 18% among those with a long nursing home stay. Switching rate decreases with star-rating of the MA contract and increases with the size of the contract measured by the number of enrollees. For example, among the beneficiaries with a short nursing home stay, switching rate is 7% points lower for contracts with 4.5 or above star rating than for those in contracts with star rating 2.5 or below. This difference is 9% points for those with a long nursing home stay. Finally, if the share of contract s patients going to a nursing home increases by 1%, switch rate declines by 0.2% among short-stay and 0.6% among long-stay patients in that contract nursing home pair. Conclusions: Switching rate varies across MA contracts depending on star-rating, size and nursing home care management. Implications for Policy or Practice: High switch rate from low star contracts imply a questionable quality of service received by the medically needy beneficiaries and a transfer of cost from MA to traditional Medicare. MA contracts need to establish preferred care provider relationship with nursing homes to reduce switching.

10 Poster Presentation: Tyler, D.A., McHugh, J., Shield, R.R., Winblad, U., & Gadbois, E.A. The Unintended Consequences of Reduced Skilled Nursing Facility Length of Stay. Poster Session: Organizational Behavior and Management, June 26, 12:15PM-1:45PM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 543. Research Objective: Due in part to the emergence of accountable care organizations and bundled payment programs, as well as pressure from managed care organizations (MCOs), skilled nursing facilities (SNFs) have seen a reduction in median length of stay (LOS) in the past few years. The effects of this phenomenon are unknown. The purpose of this research was to identify the key challenges that reductions in LOS pose for post-acute care (PAC) providers, the unintended consequences of reduced LOS for PAC providers and patient and SNF responses to these. Study Design: We utilized a multiple case study approach. Eight markets throughout the US were chosen based on their Medicare managed care penetration rate (4 high/4 low). In each market we selected one hospital with a low readmission rate and one with a higher rate as well as three or four SNFs that received referrals from those hospitals. We also selected two of the top three MCOs in each market. We interviewed approximately 20 clinical and administrative staff from the selected organizations in each market, yielding about 160 interviews. This included approximately 70 interviews with SNF administrators, directors of nursing and admissions coordinators. We used data on risk adjusted SNF median LOS for created for another project to identify the SNFs with reductions in LOS and those with no reductions in LOS. Data from interviews with staff in SNFs with no reduction in median LOS were then compared and contrasted with data from interviews with staff in SNFs with reductions in LOS. Population Studied: 70 staff in 24 SNFs from eight markets throughout the US. Principal Findings: Among the 12 SNFs where median LOS dropped, the average reduction in LOS was 4.0 days. Staff reported several challenges related to reductions in LOS. They reported not having time to help patients unable to return to the community to find appropriate long term care and said they had to discharge patients who they felt were unsafe. Some overcame these challenges by developing programs to follow-up with patients post-discharge. While these programs served to ease concerns about patients, they also involved non-reimbursable costs for SNFs. Some SNF respondents also noted ways they avoid admitting patients who had the potential to become long-stay. Staff in SNFs that did not have reductions in LOS reported not being reimbursed by MCOs and other payers for patient stays that had surpassed their pre-determined length. Some also reported patients becoming unexpectedly responsible for the costs of part of their stay. Conclusions: The push for shorter SNF LOS has resulted in challenges for SNFs and possible unintended consequences for patients, including increased costs for care and potentially unsafe discharges. Patients may also have difficulty finding facilities willing to accept them if they may eventually need long-term care. Implications for Policy or Practice: Additional research is needed to determine if the challenges and unintended consequences identified by our exploratory research are widespread. If they are, policies are needed to respond to the issues caused by reduced SNF LOS.

11 Poster Presentation: Tyler, D.A., Gadbois, E.A., & Mor, V. Flying Blind: Patient Choice of Post-Acute Care Provider and Hospital Accountability. Poster Session: Medicare, June 26, 12:15PM-1:45PM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 436. Research Objective: In this era of accountable care, hospitals have become responsible for care patients receive in the post-acute care (PAC) setting. Yet, little is known about how those needing PAC select a facility and even less is known about whether or how they use information on providers quality. Nor do we know much about the role of hospital discharge planners (DPs) in the PAC selection process. The purpose of this research was to explore how patients requiring PAC make decisions about which skilled nursing facility (SNF) to select and the role that hospital DPs play in the selection process. Study Design: We utilized a multiple case study approach. Eight markets throughout the US were chosen based on their managed care penetration rate (4 high/4 low). In each market we selected one hospital with a low readmission rate and one with a higher rate and 3 or 4 SNFs that received referrals from those hospitals. We interviewed approximately 20 clinical and administrative staff from the selected organizations in each market. In 5 markets, we returned to interview patients who had recently been admitted to one of our participating SNFs. All interviews were summarized immediately upon conclusion. Summaries were used to quantify participants responses in terms of the number of DPs who said they provided patients with quality data, the number who reported providing patients with only a list of facilities, the number of patients who reported receiving quality data, the number who reported using quality data and the number who would have gone elsewhere had their doctor recommended it or if they were told the quality was better. Population Studied: 97 patients recently admitted to SNF care and 16 hospital DPs. Principal Findings: Of the 77 patients who reported being involved in SNF selection, 57 received only a list of SNFs. Ten didn t receive lists because they already knew where they wanted to go; in all cases a SNF where they d previously been. In fact, 35 of the 97 patients reported previous stays at the same SNF. Only four patients reported receiving SNF quality data from DPs. This is consistent with responses from DPs, who reported not furnishing patients with data, but providing only lists of SNFs. DPs stated they had been told that patient choice statutes precluded them from providing any information beyond SNF addresses. Because the only data patients were given were addresses, this often resulted in patients simply selecting the SNF nearest their home and/or returning to a SNF where they d been. Yet, 73 patients said they would have traveled farther if a better SNF was recommended. Conclusions: Patient choice in PAC setting is apparently trumping hospital accountability. Further, belief that patient choice statutes preclude the sharing of quality data about SNFs means that patients choices cannot be based on quality. Implications for Policy or Practice: CMS must clarify what is meant by patient choice and allow hospitals to explicitly refer patients to SNFs they are willing to vouch for and work with to achieve better outcomes.

12 Podium Presentation: Rahman, M., Grabowski, D., & Norton, E. Causal relationship between SNF s historical rehospitalization rate and patient s likelihood of 30 day rehospitalization. Presenting in Delivery and Payment Innovations, June 26, 4:00PM-5:30PM, Hynes Convention Center, 200. Research Objective: About 25% of all patients from hospital are discharged to Skilled Nursing Facilities (SNFs) for post-acute care. Rehospitalizations among these patients are frequent, costly and often preventable. Choosing the appropriate SNF to reduce rehospitalization is a difficult task because CMS reported SNF quality measures are generally uncorrelated with rehospitalization rates. CMS is currently seeking public opinion about the idea of implementing risk adjusted hospitalization rate as a quality measure. We are yet to know whether difference in rehospitalization rate across SNFs reflect the difference in patients treated or the difference in quality of care. This paper aims to void this gap by estimating the causal effect of SNF s 30 day historical adjusted rehospitalization rate (HARR) on the likelihood of 30 day rehospitalization of new SNF patients. Study Design: We uses data to calculate SNFs HARR which is a ratio of actual number of rehospitalization and expected number of rehospitalization based on patient characteristics multiplied the national rate of rehospitalization (20.4%). Our main objective is to estimate the effect of HARR of the treating SNF on rehospitalization of patients in 2013 controlling for patient s demographic characteristics and clinical characteristics and hospital fixed effects. We used an instrumental variable (IV) approach to control for patient selection bias. Our IV is the weighted mean HARR of SNFs in the choice set of SNFs faced by a patient. We used three different IVs depending on the weights received by a SNF in the choice set: 1) HARR of (i.e. all weight on) the nearest SNF from patient s residential zip code, 2) HARR of the nearest SNF of the patient if we consider both the distance from hospital and distance from residential zip code, and 3) Mean HARR of the SNFs in patients choice set where weights are calculated based on a SNF choice function and increase with proximity and available number of beds on the day before discharge. Population Studied: All Medicare fee-for-service patients admitted to SNF following an acute hospitalization between 2009 and 2013 and did not have any SNF stay in one year period before SNF admission. Principal Findings: HARR varies widely; about one quarter of the SNFs have HARR lower than 17% and one quarter of the SNFs have HARR higher than 23%. All the IV models result in a robust finding that an increase in HARR in treating SNF by 1 percentage point leads to an increase in patient s likelihood of rehospitalization by 0.5 percentage point. Conclusions: Treatment in SNFs with historically low rehospitalization reduces a SNF patient s likelihood of rehospitalization. There are differences across SNFs in terms of the ability to avoid hospitalizations and this ability, at least to some extent, is captured by HARR. Implications for Policy or Practice: HARR can serve as a good quality indicator of SNF care. Steering patients to low HARR SNFs can reduce rehospitalizations and may impact on patients welfare, Medicare s spending and hospital s lost revenue in terms of excess rehospitalization penalties.

13 June 27, 2016 Poster Presentation: Gozalo, P., Teno, J., Meltzer, D., & Mor, V. The Impact of Fragmentation of Physician Care during a Hospitalization on Patient Outcomes: Feeding Tube Insertion Among Nursing Home Patients with Advanced Dementia. Poster Session: Improving Safety, Quality, and Value, June 27, 8:00AM-9:30AM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 740. Research Objective: Physician handoffs during a hospitalization may be associated with patient outcomes. Each transfer of care responsibility from one physician to another may result in loss of clinical information, leading to inappropriate treatments. However, transfers of care-responsibility could occur because of greater clinical needs that require multiple physicians to properly address them. Our objective is to use a cohort of patients (with advanced dementia and high levels of physical impairment) with a design and methodology that controls for the level of clinical severity so that we can better isolate the impact of increased physician care fragmentation on care (feeding tube insertion) known to be inappropriate for these patients. Study Design: To reduce potential confounders, our cohort included Medicare beneficiaries with a nursing home stay within 120 days of their index hospitalization, with a diagnosis of advanced dementia and severe functional limitations (i.e. 4 impairments of the Activities of Daily Living) and cognitive impairment (Cognitive Performance Scale 5) in the MDS prior to the hospitalization. We excluded short-stay hospitalizations ( 2 days). Physician fragmentation was measured with the Sequential Continuity Index (SECON) which equals the average number of times where the physician (based on NPI) was different than the day before (=0 if same physician each day, =1 if different physician each day). To avoid associating fragmentation with patient severity, and noting that each year our frail dementia cohort has similar severity as a group for a given hospital, we used annual hospital average fragmentation levels. We used a (within) hospital fixed-effect logistic model to examine how variations in fragmentation within a hospital over time affected likelihood of feeding tube (FT) insertion. Population Studied: Hospitalizations during the ten-year period for a random 20% of all Medicare fee-for-service nursing home residents in the USA. Data used in the study included the Minimum Data Set (MDS) assessments of nursing home residents, all Medicare hospitalization claims plus the Part B Carrier file of doctor services billed during these hospitalizations. Principal Findings: During , we observed N=112,086 hospitalizations. FT insertions decreased over time (8.9% in 2000 versus 7.9% in 2009), while hospital-level average SECON fragmentation increased from 0.41 to Most hospitals increased their average fragmentation but there was wide variation across hospitals. Adjusting for many patient baseline risk factors, each 0.1 (10%) increase in hospital fragmentation was associated with a 7% increase in the likelihood of FT insertion (AOR 1.07; 95%CI ), but it was highly nonlinear (AOR 1.13 [95%CI ] for SECON range, but AOR 1.01 [95%CI ] for SECON range). Conclusions: While FT insertion decreased over time, patients in hospitals that increased their physician care fragmentation (particularly those that had low baseline fragmentation levels) had a significant increase in their risk of FT insertions. Implications for Policy or Practice: Payment and regulatory incentives are needed to enhance physician care coordination in the hospital setting to avoid the significant adverse effects that fragmented care has on frail vulnerable subpopulations, and to deliver care consistent with patients goals and values.

14 Poster Presentation: Rivera-Hernandez, M., Rahman, M., Mor, V., & Trivedi, A. Skilled Nursing Facilities That Disproportionately Serve Black and Hispanic Patients in the United States. Poster Session: Aging, Disability, and End-of-Life, June 27, 8:00AM-9:30AM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 614. Research Objective: Prior studies have found that racial and ethnic minority patients are often concentrated among a small number of providers with worse clinical performance, but this relationship has not been examined for post-acute care. We assessed the concentration and quality of skilled nursing facilities that care for black and Hispanic patients, including those in traditional Medicare and Medicare Advantage. Study Design: Cross-sectional study of skilled nursing facility (SNF) performance measures and facility characteristics associated with quality of care. Population Studied: 1,086,471 White, 127,421 black and 58,084 Hispanic Medicare enrollees in fee-forservice and Medicare Advantage aged 65 and older admitted to a SNF following a hospitalization in Principal Findings: Approximately 27% of SNFs accounted for 80% of all admissions for black patients. Care was even more concentrated for Hispanics, with 19% of all SNFs accounting for 80% of Hispanic patient care. Compared to SNFs with 15% blacks or less, SNFs with more than 15% blacks were more likely to have lower star ratings (2.8 vs. 3.1), less direct care per resident (3.7 hrs/day vs. 3.8 hrs/day), more beds (135.0 vs ), higher percentages of Medicaid patients (2.0% vs. 1.8%), be part of a chain (61.3% vs. 59.0%) and have for-profit tax status (80.7% vs. 67.6%); p<0.01 for all comparisons. Compared to SNFs with 15% Hispanics or less, SNFs with more than 15% Hispanics had lower star ratings (2.8 vs. 3.2), higher number of beds (136.9 vs ), higher percentage of Medicaid patients (2.1% vs. 1.7%), higher chain membership (64.2% vs. 59.0%) and greater for-profit tax status (83.0% vs. 68.2%); p<.01 for all comparisons. Conclusions: Racial and ethnic minorities in the Medicare program, particularly Hispanics, are concentrated within a small number of skilled nursing facilities with worse measured performance and characteristics associated with lower quality. Implications for Policy or Practice: The site of care may be an important mediator of racial and ethnic disparities in post-acute care. Our study suggests that efforts to reduce racial and ethnic disparities in post-acute care should focus on the relatively small number of SNFs that disproportionately care for minority Medicare beneficiaries.

15 Poster Presentation: Teno, J., Gozalo, P., Khandelwal, N., Curtis, J.R., Meltzer, D., Engelberg, R., & Mor, V. The Use of Step Down and ICU Beds Among Persons with Advanced Medical Illness and Severe Functional Impairment. Poster Session: Aging, Disability, and End-of-Life, June 27, 8:00AM-9:30AM, Hynes Convention Center, Exhibit Hall C/D, Poster Number: 624. Research Objective: Intensive care and step down beds have increased between 2000 and 2010, but that growth has been concentrated in a small number of hospitals. Our goal is to describe the changing use of intensive care and step-down beds among persons with advanced medical illness between 2000 and 2010 among hospitalized patients with recent nursing home stay and 4 or more ADL impairments. Study Design: Based on the Minimum Data Set completed within 120 days prior to hospitalization, we identified persons with advanced chronic illnesses and 4 or more ADL impairments. A hospital fixedeffect model examined changes in intensive care and intermediate bed supply (based on American Hospital Association data from 2000, 2005, 2007, and 2010) with the rate of ICU and/or intermediate bed admission, adjusting for patient characteristics, cognition, functional status and other medical conditions. Population Studied: Retrospective cohort of persons with advanced cognitive impairment, end stage renal disease, cancer, congestive heart failure and/or chronic obstructive pulmonary disease who were hospitalized between 2000 and Principal Findings: In the years examined, 2,958,444 persons (14.6% black, 36.0% female) were hospitalized with one-year mortality rate of 64.4%. Use of intensive care unit increased from 16.9% (2000) to 20.6% (2010) while use of step down unit without an intensive care unit stay increased from 7.7% to 18.6%. Persons with pre-existing DNR orders were more likely to use only a step-down unit or regular acute care bed compared to intensive care unit bed (32.5%, 44.2%, vs. 27.2% respectively, p<.001). Total ICU and step down beds increased from 27.1 (2000) to 35.6 (2010). In 2010, total beds varied with interquartile range from 13 to 46 beds. Using a hospital fixed effect multivariate model, an increase of total beds from 9 to 24 beds the adjusted relative risk = 1.12 indicating a 12% increase in the risk of hospitalization that involved the use of intensive care and/or step down unit. Conclusions: Use of step down beds has increased more than ICU beds. Although causality can t be inferred, we found increasing bed supply associated with greater likelihood of persons with advanced medical illness with severe functional impairment being admitted to an ICU and/or step down unit. Implications for Policy or Practice: The use of ICU and step down beds are not subject to regulatory oversight as the use of hospice, home health, or skilled service in a nursing home. Research is needed to understand whether the rapid growth of step down beds and continued growth of ICU beds improves the care of these functionally impaired persons. If evidence is lacking to effectiveness of this level of care for these patients, regulatory interventions may be needed to ensure persons receive care based on informed preferences and goals of care.