Residential and Health Care Transition Patterns Among Older Medicare Beneficiaries Over Time

Abstract

Purpose: To describe annual care transition patterns across residential and health care settings and assess consistency in care transition patterns across years.

Design and Methods:  This retrospective cohort study used the Medicare Current Beneficiary Survey (2000–2005). The sample comprised beneficiaries aged 65 years and older (N = 57,684 person-years of observation). We defined annual care transition patterns by combining 4 types of settings: C (community), F (facility), S (skilled nursing facility—SNF), and H (hospital). We compared weighted frequencies of transition patterns across years. We counted repeated/multiple transitions that involved movement into hospital and SNF settings and compared them by demographic characteristics.

Results:  Care transition patterns remained consistent from year to year. Approximately 22% of the study population experienced a transition annually. The most frequent transition pattern was transition to the hospital and back. Care transition patterns were enormously heterogeneous with more than 230 unique patterns; approximately 1 in 4 community-dwelling (∼23%) and most facility-dwelling (∼60%) beneficiaries with at least one transition had a unique transition pattern. Beneficiaries residing in a facility were more likely to undergo multiple transitions to hospitals and SNFs compared with community-dwelling beneficiaries.

Implications: The study provides a description of annual care transition patterns across six years. Knowledge of the consistency of care transition patterns may serve as a baseline from which to compare future patterns and aid in designing interventions targeted at specific transitions.

Transitions between different residential and health care settings represent vulnerable periods for older adults (Coleman, 2003; Coleman & Berenson, 2004; Naylor et al., 2004). The subsequent care transitions that result from the movement of persons from one setting to another require effective coordination as care is transferred across providers (Coleman, 2003). There is growing evidence that suboptimal care transitions may generate unnecessary spending (e.g., emergency department utilization, avoidable rehospitalizations, duplicative laboratory tests; Coleman, 2003; Jencks, Williams, & Coleman, 2009), medical errors (e.g., medication discrepancies; Coleman, Smith, Raha, & Min, 2005; Moore, Wisnivesky, Williams, & McGinn, 2003), and poor health outcomes (e.g., adverse events, relocation stress, uncontrolled chronic conditions, avoidable rehospitalizations; Arbaje et al., 2008; Boockvar et al., 2004; Forster, Murff, Peterson, Gandhi, & Bates, 2003; Intrator, Schleinitz, Grabowski, Zinn, & Mor, 2009).

Care transitions are common among older adults (Coleman, Min, Chomiak, & Kramer, 2004). Coleman and colleagues identified 46 distinct types of care transition patterns within the first 30 days after hospital discharge of older adults and that more than 30% experience at least two or more care transitions during that time frame (Coleman et al., 2004). Between 13% and 25% of care transition patterns after hospital discharge were “complicated”—representing movement from a lower to a higher level of care. A recent study found that a fifth of Medicare patients who had been discharged from the hospital were rehospitalized within 30 days after discharge and a third were rehospitalized within 90 days (Jencks et al., 2009). The number and complexity of transition patterns in Coleman's posthospital discharge study suggest that older adults do not experience a single predominant transition pattern but rather a collection of a few primary patterns followed by many unique and complex ones.

Policymakers, health care organizations, and providers are paying increased attention to care transitions because they represent an essential element in optimal continuity of care and subsequent reduction of rehospitalizations. Programs to improve care transitions are difficult to implement under the current payment system. A 2008 recommendation by the Medicare Payment Advisory Commission (MedPAC)—that hospitals with relatively high rehospitalization rates should be subject to reduced payments as part of a plan to increase efficiency and improve quality of care—also illustrates the interest in this area (Jencks et al., 2009; Medicare Payment Advisory Commission, 2009). The Centers for Medicare & Medicaid Services (CMS) has initiated a Care Transitions Project that involves 14 Quality Improvement Organizations to improve care coordination and reduce unnecessary hospitalizations and rehospitalizations (CMS, 2009; Colorado Foundation for Medical Care, 2010). The recent passage of the Patient Protection and Affordable Care Act (PPACA) includes numerous examples of efforts intended to increase care coordination and improve care transitions (PPACA, 2010; Kaiser Family Foundation, 2010). Guidance on where to target such efforts is hampered by the scarcity of information on national care transition patterns of older adults.

Prior work examining care transition patterns among older adults has focused solely on care transitions to and from hospitals or on care transitions that occur shortly after discharge. The earliest depiction used 1992–1994 data from the National Long-Term Care Survey and presented the proportion of individuals having distinct types of transitions across four settings: home, hospital, skilled nursing facility (SNF), and other institutional settings (Murtaugh & Litke, 2002). The study by Coleman and colleagues (Coleman et al., 2004) examined transitions in 1997 and 1998 originating after a hospitalization and occurring during the first 30 days after discharge. The most recent study by Jencks (Jencks et al., 2009) also examined postdischarge transitions but specifically focused on rehospitalizations and only those occurring during the first 30 days after discharge. It did not describe residential settings encountered after the initial hospital discharge. Though these studies examined transition patterns nationally, they did not describe (a) transitions experienced by community-dwelling beneficiaries without hospitalizations, (b) transitions occurring beyond 30 days after discharge, (c) transition patterns over several years, or (d) transitions involving long-term care settings other than SNFs. In summary, the extent to which transitions happen and how they are encountered is found in the literature but is often restricted to specific sub-populations, specific transitions, or limited time periods.

Knowledge of care transition patterns across residential and health care settings can help provide a broader context not only to understand health care utilization among older beneficiaries, but also potentially inform health care organizations on the types of scenarios regarding transitions of older individuals across different settings. Such understanding also may help identify potential issues related to quality of care and identify areas for improvement in patient safety and outcomes. To address the knowledge gap of the patterns of transitions for individuals over the course of a year, we designed a study to (a) describe the natural history of care transitions among Medicare beneficiaries across residential and health care settings, (b) assess whether annual care transition patterns are consistent over time, (c) characterize repeated/multiple transitions, and (d) report variations in transition patterns across demographic characteristics.

Methods

Data Source

We obtained data from 6 years of the Medicare Current Beneficiary Survey (MCBS), a continuous multipurpose survey of a nationally representative sample of the Medicare population sponsored by CMS (CMS, 2010). Beneficiaries are surveyed three times annually for a maximum of 4 years. A key feature of the MCBS is a rotating-panel design where approximately one quarter of the sample is replenished annually. Unique to the MCBS is the ability to follow survey respondents across all settings, including movement into and out of long-term care (Adler, 1994). The data also include demographic and socioeconomic characteristics, health status, physical functioning, self-reported health care use and expenditures, health insurance coverage, living arrangements, and a wide range of other information. The survey-reported data are linked to Medicare administrative claims data. We included all respondents in the MCBS who were either community or facility dwelling during years 2000–2005, aged 65 years and older, and not lost to follow-up.

Measures of Care Transition Patterns

To operationalize the patterns of residential living and health care settings over time, we obtained information on living situations and hospital utilization from the Residence Time Line file and the Inpatient Hospital Events file. The Residence Time Line file contains dated summaries on the movement of individuals between community, facility, and SNF settings through the year. The Inpatient Hospital Event file includes data about all inpatient hospital stays of the MCBS population and is based on administrative and survey records.

From these two sources, we coded day-level care transition patterns as follows: days when individuals stayed in the community, facility, SNF, or hospital were coded as “C,” “F,” “S,” or “H,” respectively. The facility setting (F) is an amalgam of individual residential settings, including nursing homes, assisted living facilities, continuing care retirement communities, board and care homes, adult/group homes, personal care, rehabilitation, and psychiatric facilities. Within this composite category, nursing homes are the predominant type of facility setting (∼65%), followed by assisted living (∼10%). We retained the single all-encompassing facility setting for ease of presentation of results because the addition of new settings increases the number of possible permutations exponentially.

We created the study's transition variable as the concatenation of all unique settings encountered by the beneficiary during the year. This variable thus sequentially summarized the movement of the beneficiary across the four study settings of community, facility, SNF, and hospital in the order in which they happened. For example, if the individual stayed in the community during the entire year, the transition variable was expressed by “C.” If the individual was in the community at the start of the year, then hospitalized at some point, and subsequently discharged back to the community, the transition variable was expressed by “CHC.”

To account for the cross-sectional nature of the data that starts on January 1 and ends on December 31, we classified transitions as being “complete” or “active.” Complete transitions reflect movement to a final residential setting of community or facility by the end of the year. Active transitions, in contrast, represent movements into health care settings (e.g., SNF or hospital) and have not resolved to a final residential setting. This classification was necessary to describe transitions that were still “in process” at the time the survey started or ended.

The MCBS naturally distinguishes SNF from other facility settings, even though the SNF may be located in the same physical building as another facility. We retained this distinction because, despite being relatively short in nature, a SNF stay is a distinct care setting that resolves to a more permanent setting (either community or facility) and thus important to track when describing the continuum of residential transitions an individual can encounter. We also retained this as a unique setting because it also introduces a change in type of care provided. A SNF stay is defined as a transitional postacute stay (and Medicare reimbursable up to 100 days) that follows a minimum 3-day hospitalization per Medicare regulations (CMS, 2011).

Because the focus of the study is on describing the counts and patterns of care transitions across settings, we did not consider the length of stay in each setting. Counts of transitions were created as the number of different settings encountered sequentially (e.g., CHC = 2 transitions, CHSC = 3 transitions). Individuals who encountered no changes in settings during the year were defined as having zero transitions. Because the purpose of this study was to provide a descriptive natural history of the movement from one setting to another and across levels of care, lateral moves across the same type of setting (e.g., facility-to-facility or hospital-to-hospital transfers) were not considered to be a transition.

Statistical Analysis

Frequency distributions of annual transition patterns were generated to assess if transition patterns were consistent across the 6 years of the study. We characterized repeated/multiple transitions that involved movement into hospital and SNF settings and compared them by residential status as of January 1 of each year. We also looked at the number of transitions by age, gender, race, and region. All analyses were weighted and standard errors adjusted to account for the complex survey design of the MCBS and were performed using SAS 9.1.

Results

The final analytic unweighted sample consisted of 57,684 person-years of data, representing 28,103 unique individuals, who were either community or facility dwelling during the years 2000–2005, aged 65 years and older, and not lost to follow-up. The unweighted sample size ranged from 9,239 to 10,020 across the study years; the weighted number ranged from 30 to 32 million.

Approximately 95% of the sample beneficiaries were community dwelling and 5% were facility dwelling as of January 1 in each year. The mean age differed by residential setting and was 77.2 years for the community dwelling and 85.7 years for the facility dwelling. On average across the 6-year study period, 57.5% of the community-dwelling beneficiaries and 73.1% of the facility-dwelling beneficiaries were female and were predominantly White. On average across the 6 years, most beneficiaries (93.8% of community-dwelling beneficiaries and 85.2% of facility-dwelling beneficiaries) had supplemental insurance.

Care Transition Patterns

Table 1 shows the predominant community and facility annual transition patterns for the 6 years of the study. We found that care transition patterns were consistent over this time and were independent of the composition of the sample, which was continuously refreshed every year. Approximately 78% of beneficiaries per year stayed “in place” over the course of the year (74% in the community and 4% in a facility) and did not experience any transition. Of the 22% of the population who experienced a transition, 50% involved a single hospitalization and return to their original residential setting. The remaining 50% involved complex repeated/multiple transitions with only a few predominant patterns present. There were approximately 240 unique care transition patterns for any given year with many of the complex transition patterns consisting of either an individual or a very small number of individuals. Care transition patterns among older adults who experienced more than one transition were extremely heterogeneous and not easily summarized due to the unique nature of the patterns themselves, which defied simple aggregation. Details of transition patterns are available upon request. For example, in 2000, unique transition patterns were observed in 4.8% of community and 1.3% of facility dwellers (i.e., the sum of “all other complete transitions” and “all other active transitions” in Table 1). When redenominated as a proportion of the transitioning populations for each setting, these proportions were 23.7% and 60.6% among beneficiaries who experienced at least one transition, respectively.

Table 1.

Predominant Community and Facility Annual Transition Patterns in 2000–2005

Type of transition	2000		2001		2002		2003		2004		2005
	%	SE	%	SE	%	SE	%	SE	%	SE	%	SE
Sample N	10,020		9,823		9,751		9,609		9,242		9,239
Weighted N	30,714,456		30,991,141		30,991,141		31,832,934		31,687,553		32,134,123
Community (∼95%)
C	74.00	0.45	74.16	0.46	73.62	0.46	74.41	0.46	75.17	0.46	74.81	0.46
CHC	11.16	0.33	10.63	0.33	10.81	0.33	10.56	0.33	10.87	0.34	10.7	0.34
CHCHC	2.27	0.15	2.56	0.17	2.72	0.17	2.74	0.17	2.24	0.16	2.75	0.18
CHSC	0.79	0.09	0.88	0.09	0.84	0.10	1.00	0.10	0.94	0.10	0.77	0.09
CHCHCHC	0.81	0.10	0.80	0.11	0.78	0.09	0.97	0.11	0.74	0.09	0.68	0.09
CF	0.42	0.06	0.44	0.06	0.43	0.06	0.37	0.06	0.47	0.07	0.37	0.06
All other complete transitionsa	2.79	0.16	2.82	0.17	3.04	0.18	2.68	0.17	2.47	0.16	2.78	0.17
All other active transitionsb	2.02	0.14	2.16	0.15	2.29	0.16	2.07	0.14	2.07	0.15	2.16	0.15
Facility (∼5%)
F	3.63	0.16	3.39	0.16	3.49	0.16	3.09	0.15	3.14	0.15	2.91	0.15
FHF	0.45	0.06	0.46	0.06	0.40	0.06	0.39	0.06	0.40	0.06	0.40	0.06
FHSF	0.25	0.04	0.42	0.06	0.31	0.05	0.36	0.05	0.26	0.05	0.35	0.05
FHFHF	0.14	0.03	0.15	0.03	0.10	0.03	0.07	0.02	0.08	0.02	0.07	0.02
All other complete transitionsa	0.61	0.07	0.57	0.07	0.62	0.07	0.68	0.08	0.73	0.08	0.78	0.08
All other active transitionsb	0.68	0.07	0.55	0.06	0.53	0.06	0.62	0.07	0.43	0.06	0.48	0.06
	100		100		100		100		100		100

Notes: Numbers are weighted.

^aThe annual transitions ended with residential settings (C or F).

^bThe annual transitions ended with health care settings (H or S).

Repeated/Multiple Care Transitions

Characterizations of repeated/multiple transitions that involved movement into hospital and SNF settings by residential status are presented in Table 2. Patterns of hospitalizations were also relatively consistent over time. Beneficiaries who were facility dwelling at the beginning of the year had higher rates of repeated/multiple transitions to SNFs and hospitals. There was a slight increase in SNF use over time in facility-dwelling beneficiaries.

Table 2.

Hospital and SNF Utilization by Residential Status by Year

Community	2000		2001		2002		2003		2004		2005
	%	SE	%	SE	%	SE	%	SE	%	SE	%	SE
Sample N	9,187		9,043		8,988		8,887		8,549		8,553
Weighted N	28,942,008		29,273,225		29,599,206		30,173,689		30,090,906		30,531,781
No hospitalization
No SNF	78.95	0.44	78.99	0.45	78.36	0.45	78.9	0.45	79.66	0.45	79.13	0.45
1+ SNF	0.17	0.04	0.14	0.04	0.06	0.03	0.1	0.03	0.12	0.04	0.08	0.03
1 hospitalization
No SNF	13.18	0.37	12.61	0.36	12.78	0.36	12.23	0.36	12.63	0.37	12.61	0.37
1 SNF	1.47	0.13	1.47	0.12	1.57	0.13	1.73	0.14	1.56	0.13	1.45	0.12
2+ SNF	0.03	0.02	0.04	0.02	0.03	0.02	0.07	0.03	0.03	0.02	0.04	0.02
2 hospitalizations
No SNF	2.92	0.18	3.16	0.20	3.37	0.20	3.36	0.20	2.75	0.18	3.34	0.2
1 SNF	0.62	0.08	0.67	0.08	0.72	0.09	0.64	0.08	0.61	0.09	0.81	0.09
2 SNFs	0.23	0.05	0.25	0.05	0.32	0.06	0.24	0.05	0.24	0.05	0.31	0.06
3+ SNFs	0		0.01	0.01	0.01	0.01	0.01	0.01	0.04	0.02	0.01	0.01
3+ hospitalizations
No SNF	1.59	0.13	1.68	0.15	1.66	0.14	1.74	0.15	1.39	0.13	1.39	0.13
1 SNF	0.38	0.06	0.47	0.08	0.55	0.08	0.5	0.07	0.48	0.07	0.40	0.06
2 SNFs	0.33	0.06	0.32	0.06	0.29	0.06	0.29	0.05	0.28	0.06	0.24	0.05
3+ SNFs	0.13	0.04	0.18	0.05	0.27	0.06	0.19	0.04	0.19	0.05	0.22	0.05
	100		100		100		100		100		100
Facility
Sample N	833		780		763		722		693		686
Weighted N	1,772,449		1,717,917		1,708,108		1,659,245		1,596,648		1,602,342
No hospitalization
No SNF	63.35	1.74	61.94	1.18	64.63	1.80	60.23	1.90	62.94	1.91	59.64	1.95
1+ SNF	2.42	0.59	2.35	0.60	2.43	0.61	2.99	0.71	2.73	0.69	3.03	0.67
1 hospitalization
No SNF	11.07	1.09	11.19	1.16	11.43	1.22	11.9	1.25	11.97	1.28	11.2	1.23
1 SNF	7.81	0.99	10.49	1.16	9.25	1.05	12.3	1.26	7.62	1.04	9.76	1.20
2+ SNF	0.93	0.36	0.68	0.28	1.10	0.39	1.79	0.56	1.32	0.48	0.96	0.39
2 hospitalizations
No SNF	4.08	0.69	3.25	0.64	2.42	0.58	1.63	0.48	2.24	0.57	1.59	0.47
1 SNF	2.85	0.64	2.93	0.61	2.15	0.56	2.88	0.68	2.69	0.69	2.77	0.63
2 SNFs	2.45	0.64	2.16	0.56	2.29	0.59	1.60	0.49	2.04	0.53	3.94	0.85
3+ SNFs	0.45	0.23	0.45	0.28	0		0.51	0.30	0.34	0.25	0.57	0.29
3+ hospitalizations
No SNF	1.15	0.35	1.23	0.43	1.05	0.36	0.69	0.41	1.94	0.52	0.97	0.37
1 SNF	1.38	0.42	0.92	0.35	0.91	0.35	0.84	0.35	0.28	0.20	1.70	0.50
2 SNFs	1.50	0.57	0.52	0.27	1.26	0.43	1.35	0.44	1.27	0.48	1.45	0.49
3+ SNFs	0.56	0.26	1.89	0.51	1.08	0.37	1.26	0.43	2.62	0.68	2.43	0.65
	100		100		100		100		100		100

Note: Numbers are weighted. SNF = skilled nursing facility.

Number and Patterns of Care Transitions by Demographic Characteristics

Table 3 shows the number of annual transitions by age, gender, race, and region for a representative year of 2003. There were statistically significant differences seen by age (p < .001), with those who were 85 years and older experiencing more frequent movements compared with younger age groups. Although statistically significant differences also were observed by gender (p = .0280), we did not observe substantially meaningful differences in the number of annual transitions between male and female beneficiaries. There were no significant variations by race (p = .0985) or region (p = .1036). These findings were similar for other years. We also examined transition patterns stratified by age, gender, race, and region. Females and beneficiaries in the oldest age group (≥85 years) had more repeated/multiple transitions. There were no significant differences in transition patterns by race. Those in the South had higher repeated/multiple transitions compared with those in other regions of the United States (data not shown).

Table 3.

Number of Annual Transitions by Demographics, Medicare Current Beneficiary Survey 2003 (weighted N = 31,832,934)

Characteristic	Number of transitions								p value
	None		1		2		3 or more
	Row %	SE	Row %	SE	Row %	SE	Row %	SE
Sample N	7,326		185		1,136		962
Weighted N	24,698,812		511,024		3,689,608		2,933,490
Age (years)
65–74	82.1	0.66	0.7	0.13	10.8	0.54	6.5	0.42	<.0001
75–84	75.9	0.70	1.6	0.20	12.4	0.54	10.1	0.49
85+	68.9	1.05	4.5	0.47	11.7	0.73	14.9	0.81
Gender
Male	77.1	0.68	1.5	0.18	12.7	0.54	8.8	0.45	.0280
Female	78.0	0.58	1.7	0.16	10.8	0.44	9.5	0.40
Race
White	77.6	0.47	1.6	0.13	11.8	0.37	9.0	0.32	.0985
Non-White	77.2	1.22	1.4	0.32	10.5	0.89	10.9	0.91
Region
East	78.0	1.03	1.9	0.30	10.7	0.79	9.4	0.69	.1036
Midwest	77.2	0.89	1.4	0.23	11.9	0.70	9.5	0.60
South	76.3	0.74	1.6	0.20	12.5	0.58	9.6	0.51
West	80.2	0.94	1.6	0.28	10.4	0.71	7.9	0.64

Discussion

This study provides a natural history of care transition patterns for the years 2000–2005 in a nationally representative sample of older Medicare beneficiaries. We evaluated overall annual care transition patterns in older adults, and provide a benchmark for care transition patterns and number of care transitions across residential and health care settings over time. Similar to the findings reported by Coleman and colleagues (2004), there were a few predominant patterns followed by many heterogeneous patterns of care transitions. This study adds to the current literature by providing population-level transitions that include the following previously unreported data: (a) transitions experienced by community-dwelling beneficiaries without hospitalizations, (b) transitions occurring beyond 30 days after discharge, (c) transition patterns over several years, nor (d) transitions involving long-term care settings other than SNFs.

The unique rotating-panel design of the MCBS, which introduces new beneficiaries into the survey annually, means that the sample is entirely replaced by every fourth year resulting in two pairs of years that comprised totally different individuals (2000 + 2004 and 2001 + 2005). Because these pairs of annual samples comprise entirely new beneficiaries, this study suggests that the consistency of the care transition patterns is a stable population characteristic rather than an artifact of following the same individuals across years during their MCBS tenure. Using 6 years of information, we found that care transition patterns were remarkably stable and consistent across years, and this observed stability was neither due to measurement error nor due to the persistence of the same persons being followed across the years of the study period. Although the 4-year rotating-panel design means that beneficiaries can remain across several samples of our study, the transitions they can experience for each year are unique and not a repeat of the pattern from the prior year. The consistency of these patterns can serve as a stable baseline from which to compare future patterns or provide evidence of the prevalence of specific types of transition patterns.

These results show that the majority of beneficiaries stayed in their residential settings (e.g., community or facility) for an entire year and did not experience a transition. And when there was a transition, the most frequent instance was a transition to the hospital and back to the original residential setting (i.e., CHC or FHF). These results concur with our expectations that as a population, there is overall stability with moderate levels of individuals experiencing transitions (∼22%) and that the most common transition comprised a hospital admission. Facility-dwelling beneficiaries were more likely to undergo multiple transfers to hospitals and SNFs compared with community-dwelling beneficiaries. Despite policy efforts to reduce potentially avoidable hospitalizations of nursing home residents (Grabowski, Stewart, Broderick, & Coots, 2008), we did not observe a significant decrease in hospitalization rates across years for the facility-dwelling population.

There was significant heterogeneity of transition patterns across individuals. Beyond the predominant patterns shown in Table 1, patterns quickly became more unique and did not easily lend themselves to a simple or concise typology. When expressed as a proportion of the entire Medicare population, these heterogeneous unique transitions appeared small (4.81% for community and 1.29% for facility in 2000), but when redenominated as a proportion of the transitioning populations for each setting, these proportions were 23.7% and 60.6% for the community and facility dwelling, respectively. Stated another way, community-based beneficiaries with at least one transition will have a one-in-four chance of experiencing a “unique” transition. Similarly, for facility-based beneficiaries with at least one transition the probability of a “unique” transition rises to almost two out of three. This has profound implications for organizations or individuals involved with care provision. The heterogeneity of transition patterns of older adults will challenge approaches to improve care transitions because it is difficult and organizationally inefficient to plan for all potential care patterns when so many apply only to a limited number of individuals. The formation of multidisciplinary teams can be one way to deal with the variety that transitions that will be encountered (Arbaje et al., 2010; Pham, Grossman, Cohen, & Bodenheimer, 2008).

The recent passage of the PPACA (2010) is only in the beginning stages of its implementation, but within it are a number of provisions for which our findings may be useful: (a) the Independence at Home demonstration to allow teams of health professionals to share in savings if they prevent rehospitalizations (Title III, Part III, Section 3024); (b) the creation of an Innovation Center at CMS to test, evaluate, and expand payment structures to foster care coordination across treatment settings (Title III, Section III, Part 3021); (c) the improvement of care coordination for dual-eligible individuals through the newly established Federal Coordinated Health Care Office (Title II, Subtitle H, Section 2602); (d) establishment of the Community First Choice Option (Title II, Subtitle E, Section 2401) to allow for provision of home and community-based attendant services and supports for persons who otherwise qualify for an institutional level of care, and loosening of some of the fiscal constraints on states to allow state programs to cover persons who need less than institutional care under Medicaid home and community-based services waiver programs (Title II, Subtitle E, Section 2402); (e) the establishment of a Medicare shared savings program through Accountable Care Organizations to coordinate services and redesign care processes for efficient service delivery across care settings (Title III, Part III, Section 3022); (f) the creation of a pilot program to develop and evaluate a bundled payment program for an episode of care that begins 3 days prior to a hospitalization and lasts until 30 days following discharge (Title III, Part III, Section 3023); (g) implementation of the Community-Based Care Transitions Program demonstration, which aims to reduce hospital readmissions for high risk Medicare beneficiaries and test sustainable funding streams for care transition services (Title III, Part III, Section 3026); and (h) the creation of community-based interdisciplinary interprofessional teams to support the patient-centered medical home and provide improved care across settings (Title III, Subtitle F, Section 3502).

Our study findings provide a broader understanding of the extent of the transitions patterns older adults experience annually. How the implementation of the many specific items contained within the PPACA will play out remains to be seen, but annual transition patterns provided by this study give a sense of the types of transitions occurring during the years of the study before the implementation of new policies under PPACA. The new Center for Medicare and Medicaid Innovation within CMS (Title III, Part III, Section 3021) can be used as a vehicle to develop transition programs that are community based and potentially better poised to follow an individual across settings and assist with care management and coordination. The results presented here suggest that a specific transition population that could be targeted for intervention comprises those experiencing multiple community and hospitalization sequences that, on average, affect 3.3% of the Medicare population per year (∼1 million beneficiaries).

Several limitations of the study should be noted. Multiple institutional settings were collapsed into a single setting of “facility.” To address the potential implications of this classification, we reanalyzed the data for one representative year (2003), using an alternative definition of the transition variable that separated out both nursing homes and assisted living facilities from the original facility setting. The predominant patterns across years were not affected by examining these two new settings separately. However, the variety and heterogeneity of transition patterns increased by 30% (from 242 to 327 unique patterns), making it difficult to meaningfully report summarized results. The consolidation of all residential settings into a single facility setting was retained in order to achieve a balance between comprehensiveness and parsimony.

Another limitation is that the definition of “transition” is open-ended and can include other types of health care settings, such as assisted living facilities, emergency departments, physician offices, and home with home care services. Because the focus of this study was to describe changes across broad residential settings, we did not take into account describing the natural histories of these more finely defined types of transitions. These remain as areas for further research as does further characterization of factors related to transitions, such as the effect of prescription drug use or the impact of family and other informal caregivers. Finally, transitions that were lateral moves and did not involve changes in levels of care (e.g., H-H, S-S, F-F) were also excluded, but we found that only a small number (∼2%–3%) of the total population experienced such moves during a reanalysis looking at this phenomenon. Despite these limitations, this study provides a description of the natural history of transitions across both residential and health care settings in a nationally representative sample of older adults over an extended period of time. This broad view of of transitions highlights the heterogeneity of individual transition patterns to both formal and informal care providers, and is of interest to a wide range of stakeholders interested in understanding and improving care transitions.

Funding

Dr. I. H. Zuckerman was supported by a career development grant from the National Institute on Aging (K01AG22011). The authors declare that they have no competing interests. Dr. A. I. Arbaje is a Harold Amos Medical Faculty Development Program Scholar, a program funded by the Robert Wood Johnson Foundation.