Hospitalization of Elderly Medicaid Long-Term Care Users
Who Transition from Nursing Homes
Andrea Wysocki, PhD,* Robert L. Kane, MD,
†Bryan Dowd, PhD,
†Ezra Golberstein, PhD,
†Terry Lum, PhD,
‡and Tetyana Shippee, PhD
†OBJECTIVES: To compare hospitalizations of dually eli-gible older adults who had an extended Medicaid nursing home (NH) stay and transitioned out to receive Medicaid home- and community-based services (HCBS) with hospi-talizations of those who remained in the NH.
DESIGN: Retrospective matched cohort study using Medic-aid and Medicare claims and NH assessment data.
SETTING: Community (receiving Medicaid HCBS) or NH.
PARTICIPANTS: Dually eligible fee-for-service beneficia-ries aged 65 and older in Arkansas, Florida, Minnesota, New Mexico, Texas, Vermont, and Washington from 2003 to 2005. Individuals who had a Medicaid NH stay of at least 90 days and transitioned to Medicaid HCBS (N= 1,169) were matched to individuals who had a Medic-aid NH stay of at least 90 days and remained in the NH (N= 1,169).
MEASUREMENTS: Potentially preventable hospitaliza-tions (defined according to ambulatory-care-sensitive conditions) and all hospitalizations were examined. RESULTS: Cox proportional hazards models were used to compare the risk of hospitalization between the groups, accounting for the differing time at risk and censoring. Being a NH transitioner increased the hazard of experienc-ing a potentially preventable hospitalization by 40% (95% confidence interval (CI)= 1.01–1.93) over remaining in
the NH. NH transitioners had a 58% (95% CI=
1.32–1.91) greater risk of experiencing any type of hospi-talization than NH stayers.
CONCLUSION: Individuals who transitioned from the NH to HCBS had a greater risk of hospitalization. Most of the attention in long-term care transition programs has been focused on NH readmission, but programs encouraging NH
transition should recognize that individuals may be at greater risk for hospitalization after returning to the commu-nity. Planning for the medical needs of individuals who tran-sition from an extended NH stay may improve their posttransition outcomes. J Am Geriatr Soc 62:71–78, 2014. Key words: home- and community-based services; nurs-ing home; transition; hospitalization; dually eligible
E
nthusiasm for increasing access to community-based long-term care (LTC) is growing. Individuals of all ages with LTC needs have expressed a preference to be served in home and community settings whenever possi-ble.1–4 In addition to the preference for noninstitutional care, the per-person costs are generally lower for home-and community-based services (HCBS) than for nursing home (NH) services.5–7 Many state Medicaid programs have been working to change the institutional bias in Medic-aid by providing more HCBS to individuals with LTC needs, in part by implementing HCBS waivers and state plan options.7,8 Some states have specific diversion pro-grams, which aim to identify individuals at risk of long-term NH placement and to provide HCBS to keep them out of a NH, and many states also have transition pro-grams to move long-stay NH residents into home and community settings.3Specifically, the national Money Fol-lows the Person (MFP) program, which began in 2008, represents a major effort to assist states in moving long-stay individuals out of institutions.9Most individuals are discharged within 90 days of a NH admission, but individuals who remain longer than 90 days have a low likelihood of being discharged and often end up living in the NH for many months or years.10–12 Some of these individuals who become long-stay NH residents may prefer to be in a home or commu-nity setting and could reasonably be cared for in one of those settings. Moreover, some individuals’ conditions, preferences, and circumstances change over time, so even if they needed NH care initially, they may wish to move to a home or community setting after a period of time. From the *Center for Gerontology and Healthcare Research, Brown
University, Providence, Rhode Island;†School of Public Health, University of Minnesota, Minneapolis, Minnesota; and‡Department of Social Work and Social Administration and Sau Po Center on Ageing, The University of Hong Kong, Pokfulam, Hong Kong, SAR, China.
Address correspondence to Andrea Wysocki, Center for Gerontology and Healthcare Research, Brown University, Box G-S121–6, Providence, RI 02912. E-mail: [email protected]
DOI: 10.1111/jgs.12614
JAGS 62:71–78, 2014 © 2013, Copyright the Authors
Previous studies have focused on the factors predicting whether an individual will transition to the community after being admitted to the NH and on the likelihood of, and factors related to, readmission to the NH after an individual is discharged to the community.9,12–18 Less attention has been paid to how individuals fare once they have moved into home and community settings after an extended NH stay. Understanding this is critical for assess-ing how effectively HCBS programs and policies are meet-ing the goals of carmeet-ing for individuals with disabilities in the least-restrictive and most-cost-effective way possible.
Hospitalization of individuals who move out of NHs into home and community settings is an important out-come to understand. The organization and emphasis of medical care is different in the NH setting than in home and community settings, which may affect hospital use. NHs use professional nursing staff to provide care to resi-dents, and they are on site 24 hours per day. Nursing pro-fessionals are trained to provide medical care and assistance with activities of daily living (ADLs), and NHs are required to perform periodic standard assessments, in collaboration with a physician, that are intended to be used to update care plans.19HCBS programs primarily use direct-care workers or paraprofessional staff, often with assistance from informal caregivers, to provide most sup-port services. These caregivers generally do not have medi-cal training. Unlike NHs, there are no standard requirements for HCBS programs to complete updated assessments or to involve physicians in care planning.20–22
Not only does the model of care change when an indi-vidual transitions out of the NH, but the set of care pro-viders, including LTC and medical propro-viders, also changes. HCBS users must become familiar with a new set of pro-viders in the community. Individuals who remain in the NH will likely have a more-continuous set of providers and a set schedule of visits and assessments. The continuity of care may not be adequate to meet HCBS users’ needs to avoid hospitalization.
This analysis compares hospitalizations of dually eligi-ble Medicaid beneficiaries who transition to home and community LTC settings after an extended NH stay with dually eligible Medicaid beneficiaries who remain in the NH.
METHODS Data Sources
The data for this analysis were available through a data reuse agreement from a Centers for Medicare & Medicaid Services (CMS) contract to examine states’ progress toward “rebalancing” their LTC programs to include more HCBS options. Seven states (Arkansas, Florida, Minnesota, New Mexico, Texas, Vermont, and Washington) were chosen as part of the original CMS contract because of their variation in financing for institutional and commu-nity LTC programs, as well as variation in other character-istics including size, population, demographics, geography, management, policy, and county government structure.
States may offer Medicaid HCBS through three pro-grams: (i) optional 1915(c) waivers, (ii) optional personal care state plan benefits, and (iii) mandatory home health
state plan benefits. The 1915(c) waivers, or HCBS waivers, permit CMS to “waive” certain statutory requirements of the Medicaid program for states to provide services that are not normally available to beneficiaries in order to pre-vent institutionalization. These waivers are targeted to spe-cific populations. The Medicaid offices in each state identified HCBS waiver recipients and HCBS and NH state plan recipients in their Medicaid programs for each month from 2002 to 2005 and provided finder files with these LTC recipients categorized on a monthly basis. Waiver recipients included enrollees in aged and physically dis-abled waivers, and LTC state plan recipients included Medicaid beneficiaries who used home health, personal care, or NH services. The state finder files contained the CMS Eligible Identifier Number and the CMS Health Insurance Claim number to link individuals to their Med-icaid and Medicare claims files.
Centers for Medicare and Medicaid Services provided the Medicaid and Medicare claims data for the seven states for 2002 to 2005. The Medicaid files included the Medicaid Analytic extract personal summary file, inpatient file, LTC file, and other therapy file. These files were linked to the state finder files for the study population. Medicare claims for the study population were also extracted and linked to the finder files and Medicaid files. The Medicare files included the Medicare Denominator file; the Medicare Provider Analysis and Review file; the Outpatient, Home Health, and Hospice Institutional stan-dard analytical files; and the Carrier Non-Institutional standard analytical files.
Centers for Medicare and Medicaid Services provided the Minimum Data Set (MDS) files for individuals in these seven states for 2003 to 2005. All residents in Medicare-or Medicaid-certified nursing facilities are required to have standardized assessments that are part of the MDS; the assessments occur at admission and quarterly thereafter, with additional assessments performed when there is a sig-nificant change in status. The MDS measures health status and physical, cognitive, psychological, and social function-ing of residents and includes information on resident demographic characteristics.
Study Population
The study population included dually eligible fee-for-service (FFS) beneficiaries aged 65 and older in Arkansas, Florida, Minnesota, New Mexico, Texas, Vermont, and Washing-ton from 2003 to 2005. To obtain as close to an admis-sion sample as possible, individuals had to be admitted to a NH during 2003, 2004, or 2005, with no prior NH stay recorded within the previous 5 years. These individuals had to have a length of stay in the NH of longer than 90 days, and Medicaid entirely paid all stays (as indicated from the state finder files); 48,881 individuals who met these inclusion criteria were identified. “Stayers” were then defined as individuals who were not discharged to the community during the study period, and “transitioners” were defined as individuals who were discharged at any time after 90 days to the community and received Medic-aid HCBS (waiver or community state plan services) begin-ning within the month after discharge. Individuals who were discharged to the community but did not receive
Medicaid HCBS within the month after discharge were not included in the analysis in order to focus on LTC users. After defining these groups, there were 32,504 stayers and 1,942 transitioners.
To obtain the final study sample, individuals in the transitioner group were matched with individuals in the stayer group using propensity score matching. Variables used to obtain the propensity score included the long-form ADL score, Cognitive Performance Scale (CPS) score, Chronic Illness and Disability Payment System score (Medic-aid risk score based on diagnoses), year of NH admission, age, sex, urban versus rural residence, reason for Medicaid eligibility, and race.23,24All characteristics used for match-ing were calculated at the time of NH admission. To implement the propensity score matching, one-to-one nearest neighbor matching with no replacement and com-mon support was used. After matching, there was a final study sample of 1,169 NH transitioners and 1,169 NH stayers.
Variables
Dependent Variable
The outcome of interest in this analysis was the first poten-tially preventable hospitalization experienced during the observation period and the first hospitalization of any type (preventable and nonpreventable). Potentially preventable hospitalizations were defined as hospitalizations with an ambulatory-care-sensitive (ACS) condition as the primary diagnosis on the inpatient claim. Although definitions can vary, the ACS conditions in this analysis included angina pectoris; asthma; cellulitis; chronic obstructive pulmonary disease (COPD); congestive heart failure (CHF); dehydra-tion; diabetes mellitus; gastroenteritis; epilepsy; hyperten-sion; hypoglycemia; urinary tract infection; pneumonia; and severe ear, nose, and throat infections.25–28
Independent Variables
The independent variable of interest for this analysis was an individual’s group (stayer or transitioner). A number of control variables were also used in the regression model. Variables reflecting demographic characteristics were age, sex, race and ethnicity, the reason for Medicaid eligibility, urban or rural residence, and state of residence. Age was categorized as 65 to 70, 71 to 75, 76 to 80, 81 to 85, 86 to 90, and 91 and older. Race and ethnicity was classified as white, black, Hispanic, or other. The reason for Medic-aid eligibility was classified as poverty versus medical need or another reason. Beneficiary county of residence was used to determine urban or rural residence based on Metropolitan Statistical Area classifications of counties. State of residence was also included as a control variable.
Dummy variables for a number of diseases and condi-tions (anemia, anxiety, arthritis, cancer, chronic kidney disease, COPD, dementia, depression, diabetes mellitus, heart failure, hypertension, ischemic heart disease, stroke) were included to control for clinical characteristics that may be associated with potentially preventable hospitaliza-tions in this population. These were identified from diag-nosis codes on claims with a look-back period of 1 year,
so individuals were coded as having a disease or condition if they had a diagnosis on any claim file from the previous year; 2002 claims were used for individuals with censoring in 2003.
Activities of daily living and CPS scores from the most-recent MDS assessment were also included to control for physical and cognitive function, respectively.23,24 For transitioners, these scores were calculated at the time clos-est to their discharge date.
Analysis
The start of analysis time for transitioners began at the date of discharge to Medicaid HCBS. The start of analysis time for stayers was set to the start date of their matched individual in the transitioner group. The average length of stay in the NH up to the start of analysis time was 214 days for transitioners and 320 days for stayers. The event of interest in this analysis was the first potentially preventable hospitalization experienced during the obser-vation period, and censoring of data could occur because of death or end of follow-up. Any type of hospitalization was also examined.
To adjust for the differing time at risk for experienc-ing a potentially preventable hospitalization and for the censoring of data, a Cox proportional hazards regression model was used to estimate the effect of the independent variables on the time to first hospitalization. The Cox pro-portional hazards model is a semiparametric survival model, which accounts for time at risk and censoring events. The hazard ratio is obtained from this model.29In addition to the control variables used in the Cox propor-tional hazards model, inverse probability weighting with the estimated propensity score was also used in the model to address selection problems related to an individual’s group (stayer vs transitioner). All control variables were defined at the time closest to right censoring.
All analyses were performed using Stata, release 11 (StataCorp LP, College Station, TX). The cumulative haz-ard function estimated from the Cox proportional hazhaz-ards model for a potentially preventable hospitalization for stayers and transitioners was plotted using the Stata com-mand stcurve. The hazard estimates were calculated at the means of the other covariates.
RESULTS
The final matched sample included 1,169 individuals in each group. There were no significant differences between stayers and transitioners in the characteristics used for matching, which were defined at NH admission (not shown). Analysis sample characteristics (defined at the time closest to censoring) are presented in Table 1. There were some differences between the groups in the analysis characteristics, including several diagnoses and the ADL and CPS measures.
There were 113 potentially preventable hospitaliza-tions among the stayers (9.7% of the sample) and 133 potentially preventable hospitalizations among the transi-tioners (11.4% of the sample). Table 2 shows the frequency distribution of the potentially preventable hospitalization conditions for each group. Transitioners were hospitalized
more frequently for COPD, CHF, and dehydration than stayers. Of the individuals in the stayer group who did not experience a potentially preventable hospitalization, 12 were censored because of death, compared with 61 transitioners.
Results from the adjusted Cox proportional hazards model comparing potentially preventable hospitalizations of the stayers and transitioners are found in Table 3. After adjusting for other characteristics, being in the transitioner group increased the hazard of experiencing a potentially Table 1. Sample Description of Elderly Medicaid Nursing Home Stayers and Transitioners After Matching
Variable Stayers, n= 1,169 Transitioners, n= 1,169
Event
Potentially preventable hospitalization 113 hospitalizations/385,608 days at risk;a9.7% experienced a hospitalization
133 hospitalizations/335,577 days at risk;a11.4% experienced a hospitalization
Any type of hospitalization 297 hospitalizations/334,076 days at risk;a25.4% experienced a hospitalization
419 hospitalizations/274,696 days at risk;a35.8%b experienced a hospitalization Characteristic Age, % 65–70 8.2 9.8 71–75 15.6 16.5 76–80 22.3 22.9 81–85 24.9 20.9b 86–90 15.91 19.0 ≥91 13.1 11.0 Sex, % Male 25.0 27.3 Female 75.0 72.7 Race, % White 77.3 73.3b Black 9.7 10.2 Hispanic 10.4 13.7b Other 2.7 2.8
Reason for Medicaid eligibility, %
Poverty 13.5 16.3
Medically needy or other 86.5 83.8
Residence, %c Urban 72.6 71.9 Rural 27.4 28.1 State, % Arkansas 9.5 2.0 Florida 21.1 8.0 Minnesota 8.4 5.1 New Mexico 2.8 2.0 Texas 49.9 63.4 Vermont 1.3 0.3 Washington 7.0 19.2 Diagnosis, % Anemia 52.3 51.1 Anxiety 13.5 14.7 Arthritis 41.9 51.0b Cancer 10.0 13.5
Chronic kidney disease 15.8 24.2b
Chronic obstructive pulmonary disease 24.2 32.9b Dementia 58.8 53.8b Depression 42.6 38.7 Diabetes mellitus 40.3 41.2 Heart failure 41.5 51.2b Hypertension 76.4 85.6b
Ischemic heart disease 42.9 55.2b
Stroke 30.0 36.4b
Most recent activity of daily living score, mean SD (0–28)
11.6 8.4 9.2 8.1b
Most recent cognitive performance scale score, mean SD (0–6)
2.1 1.5 1.7 1.4b
SD= Standard Deviation.
aDays at risk is the total cumulative time that the sample is at risk for experiencing a hospitalization. b
Significantly different at P < .05.
c
preventable hospitalization by 40% compared with being in the stayer group (hazard ratio (HR)= 1.40, 95% confi-dence interval (CI)= 1.01–1.93). Figure 1 shows the esti-mated cumulative hazard function for a potentially preventable hospitalization for stayers and transitioners. Other characteristics in the full sample that were signifi-cant predictors of experiencing a potentially preventable hospitalization included the reason for Medicaid eligibility, COPD, and heart failure.
When any type of hospitalization (preventable or nonpreventable) was examined, there were 297 hospital-izations among the stayers (25.4% of the sample had an event) and 419 hospitalizations among the transitioners (35.8% of the sample had an event). Results from the adjusted Cox proportional hazards model for all hospital-izations (Table 4) indicate that being in the transitioner group significantly increased the risk of experiencing a hos-pitalization. Specifically, being in the transitioner group increased the risk of experiencing a hospitalization by 58% (HR= 1.58, 95% CI = 1.32–1.91) compared with being in the stayer group.
The predictors of experiencing a potentially prevent-able hospitalization were also separately examined for stayers and transitioners (data not shown). Most character-istics, with the exception of a few diagnoses, were not significant predictors of experiencing a preventable hospi-talization. For the stayer group, individuals with chronic kidney disease and COPD had a greater risk of experienc-ing a potentially preventable hospitalization compared with individuals without these conditions. Among the NH transitioners, having COPD increased the risk of experi-encing a potentially preventable hospitalization.
DISCUSSION
Transitioners had a greater risk of experiencing a poten-tially preventable hospitalization and a greater risk of experiencing any type of hospitalization. These results
suggest that transitioning to home and community-based LTC after an extended stay in a NH may increase an indi-vidual’s risk of hospitalization.
It is not clear whether the medical and nursing care or the continuity of care in the NH determines the lower risk of hospitalization of individuals who remain in the NH, but it is likely that both of these factors are important. Older adults with LTC needs typically have complex medi-cal conditions in addition to functional limitations. Receiv-ing timely medical care is important for these individuals because they can decline rapidly if health conditions are not addressed promptly.30 To avoid going to the hospital unnecessarily, it is important that HCBS users have a con-sistent set of providers who are familiar with their medical needs and are accessible when health problems arise. Other work focusing on transitions between the hospital and home have found that continuity of care, follow-up care, communication with patients and caregivers, and outpatient visits within 30 days of discharge are impor-tant factors in improving postdischarge outcomes, including hospital readmission, of older adults.31–37 Just Table 2. Hospitalizations for Each
Ambulatory-Care-Sensitive Condition According to Group
Condition Nursing Home Stayers Nursing Home Transitioners Hospitalizations, n Angina pectoris 0 2 Asthma 3 3 Cellulitis 9 4 Chronic obstructive pulmonary disease 9 15
Congestive heart failure 13 23
Dehydration 9 14 Diabetes mellitus 5 2 Gastroenteritis 3 7 Epilepsy 5 4 Hypertension 0 1 Hypoglycemia 0 0
Urinary tract infection 22 24
Pneumonia 35 34
Severe ear, nose, and throat infections
0 0
Total 113 133
Table 3. Cox Proportional Hazards Model Results for Potentially Preventable Hospitalizations—Nursing Home (NH) Transitioners Compared with NH Stayers (N = 2,338)
Variable
Hazard Ratio (95% Confidence Interval) NH transitioner 1.40a(1.01–1.93) Age (reference 65–70) 71–75 1.22 (0.64–2.34) 76–80 1.37 (0.74–2.54) 81–85 1.25 (0.63–2.45) 86–90 0.77 (0.35–1.71) ≥91 0.58 (0.23–1.44) Female 1.11 (0.68–1.83)
Race (reference white)
Black 1.08 (0.57–2.01)
Hispanic 0.93 (0.51–1.70)
Other 5.06a(1.64–15.65)
Medically needy or other 2.93a(1.46–5.88)
Rural residence 0.84 (0.55–1.30) Diagnosis Anemia 0.80 (0.54–1.17) Anxiety 0.95 (0.59–1.56) Arthritis 0.71 (0.47–1.06) Cancer 0.73 (0.40–1.34)
Chronic kidney disease 1.27 (0.80–2.01)
Chronic obstructive pulmonary disease 1.96a(1.29–2.97) Dementia 0.99 (0.68–1.44) Depression 1.11 (0.74–1.66) Diabetes mellitus 1.02 (0.67–1.56) Heart failure 1.38 (0.91–2.09) Hypertension 0.67 (0.39–1.17)
Ischemic heart disease 1.28 (0.88–1.86)
Stroke 1.10 (0.73–1.66)
Activity of daily living score (0–28)
1.03a(1.00–1.06) Cognitive performance scale
score (0–6)
0.92 (0.81–1.05)
Model also controlled for state of residence.
a
like individuals transitioning from the hospital, those who transition from NHs into the community face a frag-mented healthcare system, and these factors may also improve posttransition outcomes from the NH.
As more investment is made in transition programs, policy-makers and program staff need to recognize the potential for adverse outcomes after a transition. Most of the attention in LTC transition programs has been focused on NH readmission,9,16but individuals who transition out of NHs may have greater risk of other outcomes such as hospitalizations. In addition to examining a more-comprehensive set of outcomes for transitioners, transition programs should focus on ways to improve and ease indi-viduals’ transitions over time and not just at the time of the move. Housing and LTC services need to be arranged for individuals who transition.38 Coordinating medical care providers should also be a priority to ensure that indi-viduals receive the care they need.30,31,35 Counseling or case management at the time of transition and frequent follow-up with transitioners could help address access problems or unmet needs before they lead to the use of more-intensive services or to decline.
This study focused on dually eligible older adults. The sample population was limited to individuals with Medic-aid-covered LTC services, but Medicare covers these bene-ficiaries’ hospital stays. Each program has an incentive to limit spending, so neither program takes full responsibility for the care of dually eligible individuals. This system not only creates a tension between Medicare and Medicaid for the management and quality of care of dually eligible indi-viduals, but it may also lead to cost-shifting between the programs, particularly for individuals in the community.39 Incorporating programs that integrate LTC and medical care as part of transition initiatives may lead to better overall care for transitioning individuals.40
For stayers, chronic kidney disease and COPD were significant predictors of experiencing a potentially prevent-able hospitalization, whereas for transitioners, having COPD was a significant predictor of experiencing a
potentially preventable hospitalization. This suggests that there are few differences between the groups in the predic-tors of preventable hospitalizations, and the significant predictors seem to relate to diagnoses. Focusing on individ-uals with specific conditions such as COPD may decrease hospitalizations in both of these populations.
There are a few limitations of this analysis to note. First, only the first hospitalization was examined. It is pos-sible that individuals may cycle in and out of the hospital, NH, and home and community settings multiple times, but no conclusions can be made about these patterns. This study relied on claims data, and the results depend on the quality of these data. The data did not permit differentia-tion between specific home or community service settings (e.g., adult day center vs assisted living) and how this may affect the risk of hospitalization. The states examined were based on available data and cannot be said to be represen-tative of all state Medicaid LTC and HCBS programs. To obtain more-comparable groups, the sample was narrowed down by a number of characteristics, which limits the gen-eralizability of the results. Last, it was possible to include only a limited set of control variables that were available Table 4. Cox Proportional Hazards Model Results for Any Type of Hospitalization—Nursing Home (NH) Transitioners Compared with NH Stayers (N = 2,338)
Variable
Hazard Ratio (95% Confidence Interval) NH transitioner 1.58a(1.32–1.91) Age (reference 65–70) 71–75 0.90 (0.62–1.31) 76–80 0.91 (0.64–1.31) 81–85 0.95 (0.65–1.38) 86–90 0.79 (0.53–1.18) ≥91 0.55a(0.35–0.88) Female 0.79 (0.61–1.01)
Race (reference white)
Black 0.80 (0.54–1.18)
Hispanic 0.85 (0.59–1.21)
Other 1.64 (0.80–3.36)
Medically needy or other 1.43a(1.01–2.04)
Rural residence 1.13 (0.88–1.45) Diagnosis Anemia 1.34a(1.06–1.69) Anxiety 1.42a(1.06–1.89) Arthritis 0.93 (0.74–1.17) Cancer 0.95 (0.68–1.33)
Chronic kidney disease 1.35 (1.03–1.78)
Chronic obstructive pulmonary disease 1.80a(1.43–2.28) Dementia 0.95 (0.76–1.19) Depression 0.95 (0.75–1.19) Diabetes mellitus 1.14 (0.91–1.43) Heart failure 1.34a(1.06–1.68) Hypertension 0.85 (0.63–1.15)
Ischemic heart disease 1.41a(1.14–1.75)
Stroke 1.25 (0.98–1.59)
Activity of daily living score (0–28)
1.02a(1.00–1.04) Cognitive performance scale
score (0–6)
0.93 (0.85–1.02)
Model also controlled for state of residence.
a P < .05. 0 .0 5 .1 .1 5 .2 C u mu la ti ve H a za rd 0 200 400 600 800 1000 Analysis Time
Nursing Home Stayers Nursing Home Transitioners
Estimated Cumulative Hazard
Figure 1. Plot of estimated cumulative hazard function from Cox proportional hazards model using Stata command stcurve for a potentially preventable hospitalization for nursing home stayers and transitioners. Potentially preventable hospitaliza-tions are defined according to ambulatory care sensitive conditions.
in the claims data. It was not possible to examine other characteristics, such as the presence of do-not-hospitalize and do-not-resuscitate orders, which may affect hospital-izations. A matched sample with inverse probability weighting was used in the analysis, but no further correc-tions were made for selection, so there is some potential for selection bias in the results.
CONCLUSION
As more states implement transition programs and focus on providing HCBS to older adults with LTC needs, it is critical that medical needs are incorporated in transition planning. Ensuring that an individual has LTC and medi-cal providers and a care plan at the time of transition may keep them out of the hospital and result in more successful long-term outcomes.
ACKNOWLEDGMENTS
This paper was presented at the AcademyHealth Annual Research Meeting, Baltimore, Maryland, June 23 to 25, 2013.
Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Robert Kane consults for the SCAN Health Plan. This paper was made possible through a data reuse agreement from Centers for Medicare and Medicaid Services Contract MRAD HHSM-500–2005–000271 Task Order #1 “Monitoring Chronic Disease Care and Outcomes Among Elderly: Extending the Use of MAX to Examine Rebalancing.”
Author Contributions: All authors participated in study design, analysis, and manuscript preparation and approved the final version. Dr. Wysocki was primarily responsible for study design, data analysis, and drafting the manuscript. Dr. Kane acquired the data and provided extensive contributions to study design, analysis, and manuscript preparation.
Sponsor’s Role: The Centers for Medicare and Medic-aid Services played no role in the study design, methods, analysis, or preparation of the paper.
REFERENCES
1. Eckert JK, Morgan LA, Swamy N. Preferences for receipt of care among community-dwelling adults. J Aging Soc Policy 2004;16:49–65.
2. Kane RL, Kane RA. What older people want from long-term care, and how they can get it. Health Aff 2001;20:114–127.
3. Reinhard SC. Diversion, transition programs target nursing homes’ status quo. Health Aff 2010;29:44–48.
4. Wolff JL, Kasper JD, Shore AD. Long-term care preferences among older adults: A moving target? J Aging Soc Policy 2008;20:182–200.
5. Kaye HS, Harrington C, LaPlante MP. Long-term care: Who gets it, who provides it, who pays, and how much? Health Aff 2010;29:11–21. 6. Kitchener M, Ng T, Miller N et al. Institutional and community-based
long-term care. J Health Soc Policy 2006;22:31–50.
7. Ng T, Harrington C, Kitchener M. Medicare and Medicaid in long-term care. Health Aff 2010;29:22–28.
8. Kaiser Family Foundation. Medicaid and Long-Term Care Services and Supports. Report #2186–08. Kaiser Commission on Medicaid and the Uninsured, 2011 [on-line]. Available at www.kff.org Accessed May 22, 2012.
9. Schurrer J, Wenzlow A. A First Look at How MFP Participants Fare After Returning to the Community. The National Evaluation of the Money Fol-lows the Person (MFP) Demonstration Grant Program: Reports from the Field. Mathematica Policy Research, Number 7, 2011 [on-line]. Available at http://www.mathematica-mpr.com/publications/pdfs/health/MFPfieldrpt7. pdf Accessed June 20, 2012.
10. Gill TM, Gahbauer EA, Han L et al. Functional trajectories in older persons admitted to a nursing home with disability after an acute hospital-ization. J Am Geriatr Soc 2009;57:195–201.
11. Mehdizadeh SA. Health and long-term care use trajectories of older disabled women. Gerontologist 2002;42:304–314.
12. Mor V, Zinn J, Gozalo P et al. Prospects for transferring nursing home residents to the community. Health Aff 2007;26:1762–1771.
13. Arling G, Abrahamson KA, Cooke V et al. Facility and market factors affecting transitions from nursing home to community. Med Care 2011;49:790–796.
14. Arling G, Kane RL, Cooke V et al. Targeting residents for transitions from nursing home to community. Health Serv Res 2010;45:691–711. 15. Chapin R, Wilkinson DS, Rachlin R et al. Going home: Community
reen-try of light care nursing facility residents age 65 and over. J Health Care Finance 1998;25:35–48.
16. Howell S, Silberberg M, Quinn WV et al. Determinants of remaining in the community after discharge: Results from New Jersey’s Nursing Home Transition Program. Gerontologist 2007;47:535–547.
17. James ML, Wiley E, Fries BE. Predicting nursing facility transition candi-dates using AID: A case study. Gerontologist 2007;47:625–632.
18. Newcomer R, Kang T, Graham C. Outcomes in a nursing home transition case-management program targeting new admissions. Gerontologist 2006;46:385–390.
19. Ouslander JG, Osterweil D, Morley J. Medical Care in the Nursing Home, 2nd Ed. New York: McGraw-Hill, 1997.
20. Harahan M, Stone RI, Shah P. Examining Competencies for the Long-Term Care Workforce: A Status Report and Next Steps. U.S. Department of Health and Human Services, Assistant Secretary for Planning and Evaluation, Office of Disability, Aging and Long-Term Care Policy, 2009 [on-line]. Available at http://aspe.hhs.gov/daltcp/reports/2009/examcomp. pdf Accessed June 6, 2012.
21. Institute of Medicine. Retooling for an Aging America: Building the Health Care Workforce. Washington, DC: The National Academies Press, 2008. 22. Stone R, Harahan MF. Improving the long-term care workforce serving
older adults. Health Aff 2010;29:109–115.
23. Morris JN, Fries BE, Mehr DR et al. MDS cognitive performance scale. J Gerontol 1994;49:M174–M182.
24. Morris JN, Fries BE, Morris SA. Scaling ADLs within the MDS. J Gerontol A Biol Sci Med Sci 1999;54A:M546–M553.
25. Carter MW. Factors associated with ambulatory care-sensitive hospitaliza-tions among nursing home residents. J Aging Health 2003;15:295–331. 26. Culler SD, Parchman ML, Przybylski M. Factors related to potentially
pre-ventable hospitalizations among the elderly. Med Care 1998;36:804–817. 27. Grabowski DC, O’Malley AJ, Barhydt NR. The costs and potential savings
associated with nursing home hospitalizations. Health Aff 2007;26: 1753–1761.
28. Intrator O, Zinn J, Mor V. Nursing home characteristics and potentially preventable hospitalizations of long-stay residents. J Am Geriatr Soc 2004;52:1730–1736.
29. Cleves M, Gould W, Gutierrez RG et al. An Introduction to Survival Analysis Using Stata, 3rd Ed. College Station, TX: Stata Press, 2010. 30. Wenger NS, Young RT. Quality indicators for continuity and coordination
of care in vulnerable elders. J Am Geriatr Soc 2007;55:S285–S292. 31. Coleman EA. Falling through the cracks: Challenges and opportunities for
improving transitional care for persons with continuous complex care needs. J Am Geriatr Soc 2003;51:549–555.
32. Coleman EA, Berenson RA. Lost in transition: Challenges and opportuni-ties for improving the quality of transitional care. Ann Intern Med 2004;141:533–536.
33. Coleman EA, Parry C, Chalmers S et al. The care transitions intervention: Results of a randomized controlled trial. Arch Intern Med 2006;166: 1822–1828.
34. Kripalani S, Jackson AT, Schnipper JL et al. Promoting effective transitions of care at hospital discharge: A review of key issues for hospitalists. J Hosp Med 2007;2:314–323.
35. Misky GJ, Wald HL, Coleman EA. Post-hospitalization transitions: Exam-ining the effects of timing of primary care provider follow-up. J Hosp Med 2010;5:392–397.
36. Naylor MD. Transitional care: A critical dimension of the home healthcare quality agenda. J Healthc Qual 2006;28:48–54.
37. Naylor MD, Kurtzman ET, Pauly MV. Transitions of elders between long-term care and hospitals. Policy Polit Nurs Pract 2009;10:187–194. 38. Irvin CV, Bohl A, Peebles V et al. Post-Institutional Services of MFP
Partic-ipants: Use and Costs of Community Services and Supports. The National Evaluation of the Money Follows the Person (MFP) Demonstration Grant Program: Reports from the Field: Mathematica Policy Research,
Number 9, 2012 [on-line]. Available at http://www.mathematica-mpr.com/ publications/PDFs/health/mfpfieldrpt9.pdf Accessed July 25, 2013. 39. Grabowski DC. Medicare and Medicaid: Conflicting incentives for
long-term care. Milbank Q 2007;85:579–610.
40. Meyer H. The coming experiments in integrating and coordinating care for ‘dual eligibles’. Health Aff 2012;31:1151–1155.
