- Email: [email protected]
A Pilot Study: Post-Acute Geriatric Rehabilitation Versus Usual Care in Skilled Nursing Facilities
A Pilot Study: Post-Acute Geriatric Rehabilitation Versus Usual Care in Skilled Nursing Facilities Bong Kauh, MD, CMD, Tracy Polak, MSN, Susan Hazelett, MS, Keding Hua, MS, and Kyle Allen, DO Objectives: To compare discharge outcomes, postdischarge health care use, and death rates among patients treated in a postacute geriatric rehabilitation unit (GRU) housed within a skilled nursing facility (SNF) with those treated in a traditional SNF. Design: Retrospective observational pilot study. Setting: Two similar SNFs were compared. Participants: All patients were admitted from the acute hospital to either the GRU (n ⫽ 95) or to the usual care (UC) SNF (n ⫽ 55). Intervention: The GRU intervention consisted of comprehensive geriatric assessment and weekly interdisciplinary team rounds with a geriatrician and a geriatric nurse practitioner (GNP). The geriatrician visited the GRU twice a week and the GNP was present 4 to 5 times per week. On discharge, GRU patients were followed up with telephonic case management for 1 year. Measurements: Demographic data collected included age, gender, and race. Information collected from
Changes in the delivery of health care have been driven by increased emphasis on costs and dwindling financial resources,1,2 resulting in reduced hospital admissions, decreased lengths of stay, and decreased numbers of acute care beds.3 The decreased length of acute hospital stay has resulted in many patients (especially the elderly) being discharged prior to full functional recovery4 making them susceptible not only to a decline in physical health, but also to a decline in psychosocial well-being.3 Once initiated, this downward spiral can become self-perpetuating, culminating in decreased quality of
Division of Geriatric Medicine, Summa Health System, Akron, OH (B.K., T.P., S.H., K.A.); Office of Biostatistics, Northeastern Ohio Universities College of Medicine, Rootstown, OH (K.H.). Address correspondence to Bong Kuah, MD, CMD, 75 Arch Street, Suite 301, Akron, OH 44304. E-mail: [email protected]
Copyright ©2005 American Medical Directors Association DOI: 10.1016/j.jamda.2005.04.008 ORIGINAL STUDIES
each facility’s patient records included admitting diagnosis, length of stay, discharge disposition, and functional outcomes. Emergency department (ED) visits and hospital readmissions for 1 year after discharge from the nursing facility were obtained from our institutional database. The Rehabilitation Outcome Measure (ROM) was used by each facility to measure functional status on admission and at the time of discharge. Results: Baseline patient characteristics were comparable between the 2 facilities. At discharge from the nursing facility, GRU patients showed greater improvement in ADLs and mobility, had a significantly shorter length of stay, and were discharged to home more often. At 1 year, GRU patients had significantly fewer hospital readmissions. GRU patients also had fewer ED visits and days in the hospital at 1 year, however these results were not significant. Conclusion: These pilot results suggest that GRU may be an effective means to improve patient outcomes and reduce undesirable health care use after an acute illness. Further studies using a randomized design are needed. (J Am Med Dir Assoc 2005; 6: 321–326)
life, repeated hospital visits, avoidable institutionalization, and/or death. Patients requiring professional services after acute hospitalization are usually discharged to skilled nursing facilities (SNFs). Historically, postdischarge care delivered in nursing facilities has been fragmented, with relatively limited interdisciplinary input or active involvement of the primary care physician.5,6 Furthermore, there is usually limited knowledge of geriatric syndromes among SNF staff, which can result in poor preventive interventions, underdiagnosis of geriatric syndromes, and treatments that may or may not follow evidencebased guidelines and therefore have variable effectiveness.5,7,8 As a result, these patients often experience potentially avoidable increased SNF lengths of stay, increased rehospitalizations, decreased discharges to home, and increased visits to the emergency department upon returning home.2 Recently, greater urgency has been placed on finding innovative methods that improve care for the post–acute care frail elderly in a cost-effective manner.9 To save costs, some Kauh et al. 321
health maintenance organizations (HMOs) have invested in post–acute care in the form of transitional care units or geriatric rehabilitation units (GRUs). Generally, GRUs are staffed by an interdisciplinary team headed by a geriatrician and a geriatric nurse practitioner (GNP). GRUs have been described as “multidisciplinary units that provide comprehensive medical and social assessments and treatment, in addition to rehabilitation.”4 GRUs incorporate comprehensive geriatric assessment (CGA) by which the patient’s medical, psychosocial, and functional status are assessed to inform a comprehensive treatment plan and allow for appropriate postdischarge follow-up.7 The aim of a GRU is to maximize each patient’s rehabilitation potential by addressing physiological impediments to optimal rehabilitation outcomes (eg, shortness of breath, depression, nutrition). Studies of the effectiveness of GRUs have shown mixed results. Among the more consistent findings from these studies were improved functional outcomes, reduced length of stay, and decreased rehospitalization.7,10,11 The current observational pilot study compares the outcomes from the short-term stay skilled nursing units of 2 community-based facilities (1 with a geriatric rehabilitation unit and 1 with usual SNF care) that are operated by the same management company; use the same rehabilitation company; are in the same city; have similar patient populations; and use the same nursing policies, procedures, and practices. Admission to either the intervention GRU or the usual care (UC) SNF was based mainly on patient preference. The purpose of this pilot study was to determine how length of stay, function, and discharge disposition vary comparing patients treated on a GRU to patients on a similar unit with no GRU. Our GRU intervention also incorporated telephonic case management by a geriatric nurse practitioner for 1 year following discharge and examined how emergency department (ED) visits, hospital readmissions, and death differed comparing patients treated on a GRU plus 1-year telephonic case management with patients who underwent usual postdischarge SNF care. METHODS Permission to conduct this study was obtained from the Institutional Review Board at Summa Health System. All patients included in these analyses were admitted to our acute care hospital from home or an independent/assisted living facility between August 1, 1999, and July 31, 2000. We selected only patients discharged from the acute care facility with which the investigators are affiliated in order to ensure the availability of follow-up data. All included patients were certified as requiring a skilled admission to a SNF by Medicare criteria. All patients were of low acuity as described by Levenson.12 Patients discharged from the SNF or GRU to a long-term care facility were excluded since no team follow-up occurred for these patients. No exclusions were made based on diagnosis, comorbidities, functional status, age, gender, or race. This study involved retrospective chart review of 95 patients from the GRU and 55 patients from the UC SNF. Demographic data collected included age, gender, and race. Information collected from each facility’s patient records in322 Kauh et al.
cluded admitting diagnosis, length of stay, discharge disposition, and functional outcomes. ED visits and hospital readmissions for 1 year after discharge from the nursing facility were obtained from our institutional database. The Rehabilitation Outcome Measure (ROM)13 was used by each facility to measure functional status on admission and at the time of discharge. We used ROM data instead of Minimum Data Set (MDS)14 data because the MDS is not completed immediately on admission, nor is it routinely completed at discharge. In addition, the MDS is completed by nurses and the ROM is completed by the rehabilitation therapists. The ROM is a discipline-specific deficit measurement tool designed to meet the needs of the rehabilitation industry. Outcomes measured by the ROM include speech/language pathology, occupational therapy, and physical therapy. Deficits measured cover speech, hearing, writing, activities of daily living (ADLs), instrumental activities of daily living (IADLs), mobility, and gait. Function in each of these areas is measured on a 7-point scale. Use of the ROM allows clinicians to measure patients’ rehabilitation progress in an objective manner. All ROM users are required to obtain certification in its use to ensure consistency in the data collected. ROM data were obtained by the physical therapists and occupational therapists at the respective facilities. To obtain a rough relative comparison of medical conditions encountered at each facility, we categorized all admitting diagnoses into 1 of 6 categories: cardiac, neurovascular (mainly stroke), pulmonary, gastrointestinal (excluding incontinence), orthopedic, and other. These disease categories were scored 1 to 6. This scoring scheme roughly corresponds to the findings of Naughton et al.15 Scores were totaled for each group and the mean scores for each facility were compared to determine whether there were any important differences between the 2 facilities in the types of medical conditions encountered. Our GRU intervention consisted of comprehensive geriatric assessment by a geriatric nurse practitioner and geriatrician, and weekly interdisciplinary team meetings coordinated by a geriatrician and the GNP. Members of the interdisciplinary team included the rehabilitation director, a social worker, the unit coordinator, a dietician, an activity therapist, and the MDS nurse. Common issues discussed during team meetings included a review of medications, discharge needs, acute medical needs, and management of geriatric syndromes. The geriatrician visited the GRU at least 2 half-days per week, and the GNP was present 4 to 5 half-days per week. Under the direction of the geriatrician and/or the GNP, the team worked together to optimize each patient’s rehabilitation potential by minimizing physiological impediments to optimal functioning. The team addressed issues such as depression, sleep, incontinence, and nutrition, all of which can negatively impact on rehabilitation efforts. Medical conditions were also optimized for rehabilitation so that, for example, pain or shortness of breath would have the least possible impact on rehabilitation. Education of GRU staff regarding geriatric syndromes was ongoing. Upon discharge from the GRU, intervention patients were followed up with telephonic case management by the GNP JAMDA – September/October 2005
Table 1. Demographics on Admission to a Skilled Nursing Facility
Average age % White % Female Admitting diagnosis (mean) Admitting ADL scores (mean)‡ Admitting mobility scores (mean)‡
Intervention Group
Usual Care Group
Odds Ratio
P Value
95% CI
79.8 y 82 59 3.13 1.47 1.26
81.3 y 97 84 2.86 1.71 1.48
— 0.64 1.54 — — —
.26* .005† .0004† .30*§ .03* .051*
(⫺4.10, 1.12) (0.52, 0.78) (1.24, 1.93) (⫺0.24, 0.77) (⫺0.45, ⫺0.02) (⫺0.44, 0.001)
ADL, activity of daily living; 95% CI, 95% confidence interval. * Student t test. † Chi square. ‡ Higher scores mean better ADLs and mobility. § Diagnoses were coded as follows: 1 ⫽ cardiac, 2 ⫽ neurovascular, 3 ⫽ pulmonary, 4 ⫽ gastrointestinal, 5 ⫽ orthopedic, 6 ⫽ other.
for 1 year. This occurred at 24 to 48 hours postdischarge, and at 1, 3, 9, and 12 months. Issues discussed included medications, ADLs/IADLs, bowel/bladder problems, sleep, nutrition, mobility, mood, mental status, falls, and hospitalizations/ED visits. Problems discovered during these phone interviews were addressed by the GNP in collaboration with the geriatrician and the patient’s primary care physician. Suggestions about the patient’s plan of care were made directly to the patient or to the primary care physician. If necessary, the patient was also scheduled for a follow-up appointment with the GNP and geriatrician. UC patients received traditional SNF care that did use an interdisciplinary team but the team was not coordinated by a geriatrician and GNP. Their care in the SNF was directed by their own attending physician with no formal interaction with the SNF interdisciplinary team. There was no GNP following patients in the UC SNF. The nursing staff at the UC SNF received no special education regarding geriatric syndromes. Finally, there was no formal follow-up after discharge from the SNF, except as would have normally occurred with their primary care physician. The 2 groups were compared on various outcomes at the time of discharge from the skilled nursing facility as well as at 1-year postdischarge. Chi square analysis was performed to compare differences in proportions between the 2 groups. Student t tests were performed to determine the significance of the difference in mean scores between the 2 groups. Analysis of covariance (ANCOVA) was used to compare discharge outcomes for the 2 groups, adjusting for baseline status. Regression analysis was performed to identify significant inde-
pendent predictors of the primary outcomes of interest. A survival analysis was performed for 1-year health care use outcomes. Analyses were performed using SAS statistical software (SAS Inc., Cary, NC). RESULTS Ninety-five patients from the GRU and 55 from the UC SNF met the inclusion criteria and had outcome data at 1 year. There was no significant difference in age or admitting diagnosis in subjects comparing the 2 facilities at baseline (Table 1), however more white individuals and females were admitted to the UC SNF, and patients admitted to the GRU had significantly worse ADL and mobility scores. ADLs measured included grooming, oral care, dressing, bathing, toileting, bed transfers, toilet transfers, home maintenance, and feeding. Mobility measures included measurement of bed transfers, toilet transfers, tub transfers, and gait. At the time of discharge from the skilled facilities, patients from the GRU had significantly greater improvement in ADLs and mobility compared with those from UC (Table 2). Average improvement was calculated as the average of individual discharge ROM scores minus the average of individual admission ROM scores. No significant differences were found with respect to improvement in speech. Average length of stay was significantly shorter for patients treated on the GRU (17.63 days) compared with patients in usual care (26.47 days) (P ⫽ .005) (Table 3). A significantly greater proportion of GRU patients were discharged to home/ assisted living (vs long-term care) at the end of rehabilitation (76%) than in the usual care group (62%) (P ⫽ .0008).
Table 2. Functional Improvement at Discharge From GRU and SNF After Controlling for Baseline Scores
ADLs† Mobility‡ Speech
GRU Average Improvement§
Usual Care SNF Average Improvement
P Value
95% CI
0.83 0.99 1.85
0.42 0.70 2.28
⬍.001* .001* .416*
(0.23, 0.58) (0.12, 0.46) (⫺1.47, 0.62)
GRU, geriatric rehabilitation unit; SNF, skilled nursing facility; 95% CI, 95% confidence interval; ADLs, activities of daily living. * Analysis of covariance. † For ADLs, higher change scores indicate more improvement (range 0 –3). ‡ For mobility, higher change scores indicated more improvement (range 0 –3). § Improvement was calculated as the average of individual discharge scores minus the average of admission scores. ORIGINAL STUDIES
Kauh et al. 323
.005*
72 (76%)
34 (62%) .0008†
0.52† 0.23‡ ⫺0.003‡ 0.003‡ 0.28† 0.06†
26.47
.004* .01* .97 .53 .14 .82
17.63
Effect Size/Reg Coeff
P Value
P Value
Average length of stay, d Proportion of patients discharged to home/ assisted living (vs long-term care)
Intervention Usual Group Care Group
DC Disposition Home/Assisted Living
Table 3. Process Outcomes
324 Kauh et al.
0.59† ⫺2.20‡ ⫺5.80‡ ⫺1.80‡ 0.12† 0.07† ADL, activities of daily living; DC, discharge; LOS, length of stay; Reg Coeff, regression coefficient. * Statistically significant. † Effect size ⫽ the difference between the 2 groups/mean square root of the model. ‡ Regression coefficients.
.001* .39 .03* .17 .51 .78 0.63† 0.27‡ 0.46‡ ⫺0.002‡ 0.10† 0.54† 0.72† 0.75‡ 0.04‡ ⫺0.002‡ 0.33† 0.16† ⬍.0001* ⬍.0001* .69 .71 .08 .56
.001* .006* ⬍.0001* .73 .61 .053
Effect Size/Reg Coeff Effect Size/Reg Coeff Effect Size/Reg Coeff
P Value
DC Mobility
P Value
Dependent Variables DC ADL
P Value
Treatment group Higher admission ADL score Higher admission mobility score Increased age Male gender African American race
The results reported here support previous studies that showed a significant benefit of the GRU model and postdischarge case management by a GNP in improving patient outcomes. The GRU showed a significant positive effect on discharge disposition, length of stay, discharge ADL scores, and discharge mobility scores independent of admission ADLs, admission mobility, admitting diagnosis, age, sex, or
Independent Variables
DISCUSSION
Table 4. Regression Analysis for Independent Predictors of Primary Outcomes at Discharge
Simple t tests (Table 1) showed between-group differences with respect to race and gender, which could arguably account for the differences in outcomes observed (Tables 2 and 3); likewise for the differences in admitting ADL and mobility scores observed between groups. Therefore, regression analysis was performed to determine whether the effect of the intervention was significant, independent of these potentially confounding variables, and whether there were any interactions between variables (Table 4). Membership in the GRU treatment group was a significant independent predictor for better discharge ADL scores, better discharge mobility scores, shorter length of stay, and better discharge disposition (ie, home/assisted living vs long-term care facility) (Table 4). The effect size for the GRU intervention was 0.81, 0.56, 0.50, and 0.67 for discharge ADL scores, length of stay, discharge disposition, and discharge mobility score, respectively. In addition, higher admission ADL scores were significant independent predictors of discharge ADLs, discharge mobility, and discharge disposition. Higher admission mobility scores were independent predictors of discharge mobility and length of stay. Neither age, gender, nor race were significant independent predictors of the outcomes examined. Next we compared GRU and usual care patients with regard to their subsequent health care use (a rough indicator of health care costs) 1-year postdischarge (Table 5). Recall that GRU intervention patients received regular telephonic follow-up for 1 year postdischarge. No significant difference in mortality was found between the groups between discharge and 1 year (32% for intervention group vs 24% for usual care group, P ⫽ .29). The number of ED visits and the number of hospital admissions in the year following discharge approached significance in favor of the intervention. Information regarding long-term care admissions during the year postdischarge was not available.
LOS
* Student t test. † Chi square.
JAMDA – September/October 2005
Table 5. Rate of Health Care Use at 1 Year for Patients Discharged to Home/Assisted Living by Group
†
Hospital admission rate, mean ED visit rate, mean‡ Days in hospital, mean§
Intervention
Control
P Value*
0.0015 0.0010 0.0091 (⫻ 365 ⫽ 3.32 d)
0.0041 0.0032 0.0216 (⫻ 365 ⫽ 7.88 d)
.02 .16 .09
* Student t test. † Number of hospital admissions 1-year postdischarge/number of days alive 1-year post-discharge. ‡ Number of ED visits 1-year postdischarge/number of days alive 1-year postdischarge. § Number of days in the hospital 1-year postdischarge/number of days alive 1-year post-discharge.
race. After 1 year of case management with postdischarge telephone follow-up by the GNP, GRU patients had fewer ED visits, fewer acute hospital readmissions, and fewer days in the hospital. The most striking result of this study was the difference in the length of stay between the 2 facilities. One might expect that the GRU patients might have had a longer length of stay given the greater improvement in function observed at discharge,16 however their length of stay was significantly shorter. A shorter length of stay might also have been expected if patients in the GRU group had been so much more impaired as to make further therapy pointless or to require placement in a long-term care facility, however there is no evidence that this was the case. Recall that the 2 facilities are operated by the same management company and share similar policies and procedures, so it is unlikely that differences in reimbursement account for the shorter length of stay. Given the race and gender differences between the 2 facilities, however, we cannot rule out differences in social support as contributing to length of stay. Some might argue that inpatient GRU is too expensive for practical purposes.2 It is true that there are substantial costs associated with the operation of a GRU. These include the cost of the geriatrician and the advanced practice nurse, as well as the time spent by the members of the interdisciplinary team during team meetings. Future studies could test variations of our model to determine the essential elements for effective geriatric rehabilitation. The addition of the GRU resulted in the facility being able to secure more contracts with local and national managed care companies and an increase in referrals from local hospitals as the reputation for improved care spread in the community. The GRU is now one of the preferred sites for our own system-sponsored health plan, which also endorses that each of their facilities creates this model of care as a means of assuring performance outcomes by which the health plan measures each contracted facility. The GRU facility saw benefits in a change in their payer mix shifting more toward Medicare skilled reimbursement. However, the facility also started to have an increase in the long-term intermediate level of care census. Additional benefits for the facility resulting from the GRU team-building process included less employee turnover, improved staff morale, and the overflow of quality of care to other units in the facility. In addition to the benefits reaped by the nursing facility, our data indicate that insurance companies should also see ORIGINAL STUDIES
significant savings associated with GRU care since shorter lengths of stay mean significant savings. Futhermore, the postdischarge decrease in hospital admissions and emergency room visits seen in our study increases the savings associated with our intervention. The decrease in patients going to long-term care would also benefit Medicare, since patients in long-term care are at greater risk of hospitalization than their community-dwelling counterparts. It remains to be determined whether these benefits outweigh the costs associated with the GRU. The greatest beneficiaries of GRU interventions are the patients themselves who experience more discharges to the community and a higher level of function at discharge. This study was not designed as a randomized controlled trial, so some bias may be present due to self-selection into the respective treatment facilities. At the time, however, the difference between GRU and usual SNF care was poorly understood by patients and physicians and anecdotal evidence suggests that patients chose the respective facilities based mainly on geographic location and proximity to their home or family. Furthermore, even though the availability of the GRU unit was known, at the time neither patients nor physicians really understood the difference between GRU and usual SNF care (the data presented here represent admissions from the first year of the GRU’s operation). Thus, selection of facilities was unlikely to have been influenced by a perceived difference in the quality of services provided. However, gender and race differences were found between the 2 facilities. Some may argue that African Americans are more likely to be discharged to home due to greater social supports or that males are more likely to be discharged to home because they are more likely to have a spouse to care for them. However, the regression analysis showed that gender and race were not significant independent predictors of any outcomes examined. This result is similar to Naughton et al15 who showed that race and gender were not significant independent predictors of discharge to the community. The significantly better improvement in ADLs seen in the GRU patients could be because GRU patients were significantly worse in ADLs on admission, therefore they may have shown some regression to the mean. However, the differences in ADLs at admission were not clinically significant. Furthermore, the samples were relatively homogeneous with respect to other important variables such as the patients’ average age and their admitting diagnosis so that we are fairly confident that the 2 facilities were treating similar patient populations. The small sample size might acKauh et al. 325
count for some of the results that were not significant. Despite these limitations, the size of the impact of the intervention on length of stay alone warrants randomized trials of this model of post–acute care. CONCLUSION Our results suggest that increased attention to the medical needs of patients requiring rehabilitation in an SNF (ie, the GRU intervention) may have a significant positive impact on patient outcomes since patients whose medical issues were adequately addressed may have been able to tolerate more therapy for a longer period of time. This might then account for the shorter length of stay observed in the intervention facility. Such improvements could greatly benefit an increasingly overburdened health care delivery system. Further randomized trials that incorporate cost-effectiveness data are needed to substantiate a significant benefit on improvement in patient outcomes and GRU cost savings. If further studies support the benefits, the GRU intervention should become the standard of care in all SNFs in the future. ACKNOWLEDGMENTS The authors thank Joe VanNostran, Velva Spencer, Wyant Woods GRU interdisciplinary team members, CommuniCare Health Services, Tom Lehner, MD, David Jarjoura, PhD, and Judy Agner for their support with this project. REFERENCES 1. Basford J. Goals and goal setting in rehabilitation research. Arch Phys Med Rehabil 1999;80:479 – 480. 2. Boult C, Boult L, Pacala J. Systems of care for older populations of the future. J Am Geriatr Soc 1998;46:499 –505.
326 Kauh et al.
3. Hoenig H, Nusbaum N, Brummel-Smith K. Geriatric rehabilitation: State of the art. J Am Geriatr Soc 1997;45:1371–1381. 4. Applegate W, Miller S, Graney M, et al. D. A randomized controlled trial of a geriatric assessment unit in a community rehabilitation hospital. N Engl J Med 1990;322:1572–1578. 5. Kane R, Chen Q, Blewett L, Sangl J. Do rehabilitative nursing homes improve the outcomes of care? J Am Geriatr Soc 1996;44:545–554. 6. Reuben D, von Sternberg T, Hepburn K, et al. Post-hospital sub-acute care: An example of a managed care model. J Am Geriatr Soc 1997;45: 87–91. 7. Reubenstein L, Stuck A, Siu A, Wieland D. Impacts of geriatric evaluation and management programs on defined outcomes: Overview of evidence. J Am Geriatr Soc 1991;39(Suppl):15S. 8. Weinberg D, Levine J. Clinical areas of liability: Risk management concerns in long-term care. Ann Long-Term Care 2005;13:26 –32. 9. Reuben D, Schnelle J, Buchanan J, et al. Primary care of long-stay nursing home residents: Approaches of three health maintenance organizations. J Am Geriatr Soc 1999;47:131–138. 10. Reubenstein L, Josephson K, Wieland D, English P. Effectiveness of a geriatric evaluation unit. N Engl J Med 1984;311:1664 –1670. 11. Burl J, Bonner A, Rao M, Khan A. Geriatric nurse practitioners in long-term care: Demonstration of effectiveness in managed care. J Am Geriatr Soc 1998;46:506 –510. 12. Levenson S. The future of subacute care. Clin Geriatr Med 2000;16:683– 699. 13. Rehabilitation Outcome Measure for Physical Therapy. Copyright 1994 SCRS & Communicology, Inc. of Ohio. Revision 2, 1996. 14. Heaton WH, ed. Long-term care. Resident Instrument Users Manual, 2003 edition. Miamisburg, OH: Med-PASS Inc.; 2002. 15. Naughton B, Saltzman S, Priore R, Reedy K, Mylotte J. Using admission characteristics to predict return to the community from a post-acute geriatric evaluation and management unit. J Am Geriatr Soc 1999;74: 1100 –1104. 16. Chen C, Heinemann A, Granger C, Linn R. Functional gains and therapy intensity during subacute rehabilitation: A study of 20 facilities. Arch Phys Med Rehabil 2002;83:1514 –1523.
JAMDA – September/October 2005
Changes in the delivery of health care have been driven by increased emphasis on costs and dwindling financial resources,1,2 resulting in reduced hospital admissions, decreased lengths of stay, and decreased numbers of acute care beds.3 The decreased length of acute hospital stay has resulted in many patients (especially the elderly) being discharged prior to full functional recovery4 making them susceptible not only to a decline in physical health, but also to a decline in psychosocial well-being.3 Once initiated, this downward spiral can become self-perpetuating, culminating in decreased quality of
Division of Geriatric Medicine, Summa Health System, Akron, OH (B.K., T.P., S.H., K.A.); Office of Biostatistics, Northeastern Ohio Universities College of Medicine, Rootstown, OH (K.H.). Address correspondence to Bong Kuah, MD, CMD, 75 Arch Street, Suite 301, Akron, OH 44304. E-mail: [email protected]
Copyright ©2005 American Medical Directors Association DOI: 10.1016/j.jamda.2005.04.008 ORIGINAL STUDIES
each facility’s patient records included admitting diagnosis, length of stay, discharge disposition, and functional outcomes. Emergency department (ED) visits and hospital readmissions for 1 year after discharge from the nursing facility were obtained from our institutional database. The Rehabilitation Outcome Measure (ROM) was used by each facility to measure functional status on admission and at the time of discharge. Results: Baseline patient characteristics were comparable between the 2 facilities. At discharge from the nursing facility, GRU patients showed greater improvement in ADLs and mobility, had a significantly shorter length of stay, and were discharged to home more often. At 1 year, GRU patients had significantly fewer hospital readmissions. GRU patients also had fewer ED visits and days in the hospital at 1 year, however these results were not significant. Conclusion: These pilot results suggest that GRU may be an effective means to improve patient outcomes and reduce undesirable health care use after an acute illness. Further studies using a randomized design are needed. (J Am Med Dir Assoc 2005; 6: 321–326)
life, repeated hospital visits, avoidable institutionalization, and/or death. Patients requiring professional services after acute hospitalization are usually discharged to skilled nursing facilities (SNFs). Historically, postdischarge care delivered in nursing facilities has been fragmented, with relatively limited interdisciplinary input or active involvement of the primary care physician.5,6 Furthermore, there is usually limited knowledge of geriatric syndromes among SNF staff, which can result in poor preventive interventions, underdiagnosis of geriatric syndromes, and treatments that may or may not follow evidencebased guidelines and therefore have variable effectiveness.5,7,8 As a result, these patients often experience potentially avoidable increased SNF lengths of stay, increased rehospitalizations, decreased discharges to home, and increased visits to the emergency department upon returning home.2 Recently, greater urgency has been placed on finding innovative methods that improve care for the post–acute care frail elderly in a cost-effective manner.9 To save costs, some Kauh et al. 321
health maintenance organizations (HMOs) have invested in post–acute care in the form of transitional care units or geriatric rehabilitation units (GRUs). Generally, GRUs are staffed by an interdisciplinary team headed by a geriatrician and a geriatric nurse practitioner (GNP). GRUs have been described as “multidisciplinary units that provide comprehensive medical and social assessments and treatment, in addition to rehabilitation.”4 GRUs incorporate comprehensive geriatric assessment (CGA) by which the patient’s medical, psychosocial, and functional status are assessed to inform a comprehensive treatment plan and allow for appropriate postdischarge follow-up.7 The aim of a GRU is to maximize each patient’s rehabilitation potential by addressing physiological impediments to optimal rehabilitation outcomes (eg, shortness of breath, depression, nutrition). Studies of the effectiveness of GRUs have shown mixed results. Among the more consistent findings from these studies were improved functional outcomes, reduced length of stay, and decreased rehospitalization.7,10,11 The current observational pilot study compares the outcomes from the short-term stay skilled nursing units of 2 community-based facilities (1 with a geriatric rehabilitation unit and 1 with usual SNF care) that are operated by the same management company; use the same rehabilitation company; are in the same city; have similar patient populations; and use the same nursing policies, procedures, and practices. Admission to either the intervention GRU or the usual care (UC) SNF was based mainly on patient preference. The purpose of this pilot study was to determine how length of stay, function, and discharge disposition vary comparing patients treated on a GRU to patients on a similar unit with no GRU. Our GRU intervention also incorporated telephonic case management by a geriatric nurse practitioner for 1 year following discharge and examined how emergency department (ED) visits, hospital readmissions, and death differed comparing patients treated on a GRU plus 1-year telephonic case management with patients who underwent usual postdischarge SNF care. METHODS Permission to conduct this study was obtained from the Institutional Review Board at Summa Health System. All patients included in these analyses were admitted to our acute care hospital from home or an independent/assisted living facility between August 1, 1999, and July 31, 2000. We selected only patients discharged from the acute care facility with which the investigators are affiliated in order to ensure the availability of follow-up data. All included patients were certified as requiring a skilled admission to a SNF by Medicare criteria. All patients were of low acuity as described by Levenson.12 Patients discharged from the SNF or GRU to a long-term care facility were excluded since no team follow-up occurred for these patients. No exclusions were made based on diagnosis, comorbidities, functional status, age, gender, or race. This study involved retrospective chart review of 95 patients from the GRU and 55 patients from the UC SNF. Demographic data collected included age, gender, and race. Information collected from each facility’s patient records in322 Kauh et al.
cluded admitting diagnosis, length of stay, discharge disposition, and functional outcomes. ED visits and hospital readmissions for 1 year after discharge from the nursing facility were obtained from our institutional database. The Rehabilitation Outcome Measure (ROM)13 was used by each facility to measure functional status on admission and at the time of discharge. We used ROM data instead of Minimum Data Set (MDS)14 data because the MDS is not completed immediately on admission, nor is it routinely completed at discharge. In addition, the MDS is completed by nurses and the ROM is completed by the rehabilitation therapists. The ROM is a discipline-specific deficit measurement tool designed to meet the needs of the rehabilitation industry. Outcomes measured by the ROM include speech/language pathology, occupational therapy, and physical therapy. Deficits measured cover speech, hearing, writing, activities of daily living (ADLs), instrumental activities of daily living (IADLs), mobility, and gait. Function in each of these areas is measured on a 7-point scale. Use of the ROM allows clinicians to measure patients’ rehabilitation progress in an objective manner. All ROM users are required to obtain certification in its use to ensure consistency in the data collected. ROM data were obtained by the physical therapists and occupational therapists at the respective facilities. To obtain a rough relative comparison of medical conditions encountered at each facility, we categorized all admitting diagnoses into 1 of 6 categories: cardiac, neurovascular (mainly stroke), pulmonary, gastrointestinal (excluding incontinence), orthopedic, and other. These disease categories were scored 1 to 6. This scoring scheme roughly corresponds to the findings of Naughton et al.15 Scores were totaled for each group and the mean scores for each facility were compared to determine whether there were any important differences between the 2 facilities in the types of medical conditions encountered. Our GRU intervention consisted of comprehensive geriatric assessment by a geriatric nurse practitioner and geriatrician, and weekly interdisciplinary team meetings coordinated by a geriatrician and the GNP. Members of the interdisciplinary team included the rehabilitation director, a social worker, the unit coordinator, a dietician, an activity therapist, and the MDS nurse. Common issues discussed during team meetings included a review of medications, discharge needs, acute medical needs, and management of geriatric syndromes. The geriatrician visited the GRU at least 2 half-days per week, and the GNP was present 4 to 5 half-days per week. Under the direction of the geriatrician and/or the GNP, the team worked together to optimize each patient’s rehabilitation potential by minimizing physiological impediments to optimal functioning. The team addressed issues such as depression, sleep, incontinence, and nutrition, all of which can negatively impact on rehabilitation efforts. Medical conditions were also optimized for rehabilitation so that, for example, pain or shortness of breath would have the least possible impact on rehabilitation. Education of GRU staff regarding geriatric syndromes was ongoing. Upon discharge from the GRU, intervention patients were followed up with telephonic case management by the GNP JAMDA – September/October 2005
Table 1. Demographics on Admission to a Skilled Nursing Facility
Average age % White % Female Admitting diagnosis (mean) Admitting ADL scores (mean)‡ Admitting mobility scores (mean)‡
Intervention Group
Usual Care Group
Odds Ratio
P Value
95% CI
79.8 y 82 59 3.13 1.47 1.26
81.3 y 97 84 2.86 1.71 1.48
— 0.64 1.54 — — —
.26* .005† .0004† .30*§ .03* .051*
(⫺4.10, 1.12) (0.52, 0.78) (1.24, 1.93) (⫺0.24, 0.77) (⫺0.45, ⫺0.02) (⫺0.44, 0.001)
ADL, activity of daily living; 95% CI, 95% confidence interval. * Student t test. † Chi square. ‡ Higher scores mean better ADLs and mobility. § Diagnoses were coded as follows: 1 ⫽ cardiac, 2 ⫽ neurovascular, 3 ⫽ pulmonary, 4 ⫽ gastrointestinal, 5 ⫽ orthopedic, 6 ⫽ other.
for 1 year. This occurred at 24 to 48 hours postdischarge, and at 1, 3, 9, and 12 months. Issues discussed included medications, ADLs/IADLs, bowel/bladder problems, sleep, nutrition, mobility, mood, mental status, falls, and hospitalizations/ED visits. Problems discovered during these phone interviews were addressed by the GNP in collaboration with the geriatrician and the patient’s primary care physician. Suggestions about the patient’s plan of care were made directly to the patient or to the primary care physician. If necessary, the patient was also scheduled for a follow-up appointment with the GNP and geriatrician. UC patients received traditional SNF care that did use an interdisciplinary team but the team was not coordinated by a geriatrician and GNP. Their care in the SNF was directed by their own attending physician with no formal interaction with the SNF interdisciplinary team. There was no GNP following patients in the UC SNF. The nursing staff at the UC SNF received no special education regarding geriatric syndromes. Finally, there was no formal follow-up after discharge from the SNF, except as would have normally occurred with their primary care physician. The 2 groups were compared on various outcomes at the time of discharge from the skilled nursing facility as well as at 1-year postdischarge. Chi square analysis was performed to compare differences in proportions between the 2 groups. Student t tests were performed to determine the significance of the difference in mean scores between the 2 groups. Analysis of covariance (ANCOVA) was used to compare discharge outcomes for the 2 groups, adjusting for baseline status. Regression analysis was performed to identify significant inde-
pendent predictors of the primary outcomes of interest. A survival analysis was performed for 1-year health care use outcomes. Analyses were performed using SAS statistical software (SAS Inc., Cary, NC). RESULTS Ninety-five patients from the GRU and 55 from the UC SNF met the inclusion criteria and had outcome data at 1 year. There was no significant difference in age or admitting diagnosis in subjects comparing the 2 facilities at baseline (Table 1), however more white individuals and females were admitted to the UC SNF, and patients admitted to the GRU had significantly worse ADL and mobility scores. ADLs measured included grooming, oral care, dressing, bathing, toileting, bed transfers, toilet transfers, home maintenance, and feeding. Mobility measures included measurement of bed transfers, toilet transfers, tub transfers, and gait. At the time of discharge from the skilled facilities, patients from the GRU had significantly greater improvement in ADLs and mobility compared with those from UC (Table 2). Average improvement was calculated as the average of individual discharge ROM scores minus the average of individual admission ROM scores. No significant differences were found with respect to improvement in speech. Average length of stay was significantly shorter for patients treated on the GRU (17.63 days) compared with patients in usual care (26.47 days) (P ⫽ .005) (Table 3). A significantly greater proportion of GRU patients were discharged to home/ assisted living (vs long-term care) at the end of rehabilitation (76%) than in the usual care group (62%) (P ⫽ .0008).
Table 2. Functional Improvement at Discharge From GRU and SNF After Controlling for Baseline Scores
ADLs† Mobility‡ Speech
GRU Average Improvement§
Usual Care SNF Average Improvement
P Value
95% CI
0.83 0.99 1.85
0.42 0.70 2.28
⬍.001* .001* .416*
(0.23, 0.58) (0.12, 0.46) (⫺1.47, 0.62)
GRU, geriatric rehabilitation unit; SNF, skilled nursing facility; 95% CI, 95% confidence interval; ADLs, activities of daily living. * Analysis of covariance. † For ADLs, higher change scores indicate more improvement (range 0 –3). ‡ For mobility, higher change scores indicated more improvement (range 0 –3). § Improvement was calculated as the average of individual discharge scores minus the average of admission scores. ORIGINAL STUDIES
Kauh et al. 323
.005*
72 (76%)
34 (62%) .0008†
0.52† 0.23‡ ⫺0.003‡ 0.003‡ 0.28† 0.06†
26.47
.004* .01* .97 .53 .14 .82
17.63
Effect Size/Reg Coeff
P Value
P Value
Average length of stay, d Proportion of patients discharged to home/ assisted living (vs long-term care)
Intervention Usual Group Care Group
DC Disposition Home/Assisted Living
Table 3. Process Outcomes
324 Kauh et al.
0.59† ⫺2.20‡ ⫺5.80‡ ⫺1.80‡ 0.12† 0.07† ADL, activities of daily living; DC, discharge; LOS, length of stay; Reg Coeff, regression coefficient. * Statistically significant. † Effect size ⫽ the difference between the 2 groups/mean square root of the model. ‡ Regression coefficients.
.001* .39 .03* .17 .51 .78 0.63† 0.27‡ 0.46‡ ⫺0.002‡ 0.10† 0.54† 0.72† 0.75‡ 0.04‡ ⫺0.002‡ 0.33† 0.16† ⬍.0001* ⬍.0001* .69 .71 .08 .56
.001* .006* ⬍.0001* .73 .61 .053
Effect Size/Reg Coeff Effect Size/Reg Coeff Effect Size/Reg Coeff
P Value
DC Mobility
P Value
Dependent Variables DC ADL
P Value
Treatment group Higher admission ADL score Higher admission mobility score Increased age Male gender African American race
The results reported here support previous studies that showed a significant benefit of the GRU model and postdischarge case management by a GNP in improving patient outcomes. The GRU showed a significant positive effect on discharge disposition, length of stay, discharge ADL scores, and discharge mobility scores independent of admission ADLs, admission mobility, admitting diagnosis, age, sex, or
Independent Variables
DISCUSSION
Table 4. Regression Analysis for Independent Predictors of Primary Outcomes at Discharge
Simple t tests (Table 1) showed between-group differences with respect to race and gender, which could arguably account for the differences in outcomes observed (Tables 2 and 3); likewise for the differences in admitting ADL and mobility scores observed between groups. Therefore, regression analysis was performed to determine whether the effect of the intervention was significant, independent of these potentially confounding variables, and whether there were any interactions between variables (Table 4). Membership in the GRU treatment group was a significant independent predictor for better discharge ADL scores, better discharge mobility scores, shorter length of stay, and better discharge disposition (ie, home/assisted living vs long-term care facility) (Table 4). The effect size for the GRU intervention was 0.81, 0.56, 0.50, and 0.67 for discharge ADL scores, length of stay, discharge disposition, and discharge mobility score, respectively. In addition, higher admission ADL scores were significant independent predictors of discharge ADLs, discharge mobility, and discharge disposition. Higher admission mobility scores were independent predictors of discharge mobility and length of stay. Neither age, gender, nor race were significant independent predictors of the outcomes examined. Next we compared GRU and usual care patients with regard to their subsequent health care use (a rough indicator of health care costs) 1-year postdischarge (Table 5). Recall that GRU intervention patients received regular telephonic follow-up for 1 year postdischarge. No significant difference in mortality was found between the groups between discharge and 1 year (32% for intervention group vs 24% for usual care group, P ⫽ .29). The number of ED visits and the number of hospital admissions in the year following discharge approached significance in favor of the intervention. Information regarding long-term care admissions during the year postdischarge was not available.
LOS
* Student t test. † Chi square.
JAMDA – September/October 2005
Table 5. Rate of Health Care Use at 1 Year for Patients Discharged to Home/Assisted Living by Group
†
Hospital admission rate, mean ED visit rate, mean‡ Days in hospital, mean§
Intervention
Control
P Value*
0.0015 0.0010 0.0091 (⫻ 365 ⫽ 3.32 d)
0.0041 0.0032 0.0216 (⫻ 365 ⫽ 7.88 d)
.02 .16 .09
* Student t test. † Number of hospital admissions 1-year postdischarge/number of days alive 1-year post-discharge. ‡ Number of ED visits 1-year postdischarge/number of days alive 1-year postdischarge. § Number of days in the hospital 1-year postdischarge/number of days alive 1-year post-discharge.
race. After 1 year of case management with postdischarge telephone follow-up by the GNP, GRU patients had fewer ED visits, fewer acute hospital readmissions, and fewer days in the hospital. The most striking result of this study was the difference in the length of stay between the 2 facilities. One might expect that the GRU patients might have had a longer length of stay given the greater improvement in function observed at discharge,16 however their length of stay was significantly shorter. A shorter length of stay might also have been expected if patients in the GRU group had been so much more impaired as to make further therapy pointless or to require placement in a long-term care facility, however there is no evidence that this was the case. Recall that the 2 facilities are operated by the same management company and share similar policies and procedures, so it is unlikely that differences in reimbursement account for the shorter length of stay. Given the race and gender differences between the 2 facilities, however, we cannot rule out differences in social support as contributing to length of stay. Some might argue that inpatient GRU is too expensive for practical purposes.2 It is true that there are substantial costs associated with the operation of a GRU. These include the cost of the geriatrician and the advanced practice nurse, as well as the time spent by the members of the interdisciplinary team during team meetings. Future studies could test variations of our model to determine the essential elements for effective geriatric rehabilitation. The addition of the GRU resulted in the facility being able to secure more contracts with local and national managed care companies and an increase in referrals from local hospitals as the reputation for improved care spread in the community. The GRU is now one of the preferred sites for our own system-sponsored health plan, which also endorses that each of their facilities creates this model of care as a means of assuring performance outcomes by which the health plan measures each contracted facility. The GRU facility saw benefits in a change in their payer mix shifting more toward Medicare skilled reimbursement. However, the facility also started to have an increase in the long-term intermediate level of care census. Additional benefits for the facility resulting from the GRU team-building process included less employee turnover, improved staff morale, and the overflow of quality of care to other units in the facility. In addition to the benefits reaped by the nursing facility, our data indicate that insurance companies should also see ORIGINAL STUDIES
significant savings associated with GRU care since shorter lengths of stay mean significant savings. Futhermore, the postdischarge decrease in hospital admissions and emergency room visits seen in our study increases the savings associated with our intervention. The decrease in patients going to long-term care would also benefit Medicare, since patients in long-term care are at greater risk of hospitalization than their community-dwelling counterparts. It remains to be determined whether these benefits outweigh the costs associated with the GRU. The greatest beneficiaries of GRU interventions are the patients themselves who experience more discharges to the community and a higher level of function at discharge. This study was not designed as a randomized controlled trial, so some bias may be present due to self-selection into the respective treatment facilities. At the time, however, the difference between GRU and usual SNF care was poorly understood by patients and physicians and anecdotal evidence suggests that patients chose the respective facilities based mainly on geographic location and proximity to their home or family. Furthermore, even though the availability of the GRU unit was known, at the time neither patients nor physicians really understood the difference between GRU and usual SNF care (the data presented here represent admissions from the first year of the GRU’s operation). Thus, selection of facilities was unlikely to have been influenced by a perceived difference in the quality of services provided. However, gender and race differences were found between the 2 facilities. Some may argue that African Americans are more likely to be discharged to home due to greater social supports or that males are more likely to be discharged to home because they are more likely to have a spouse to care for them. However, the regression analysis showed that gender and race were not significant independent predictors of any outcomes examined. This result is similar to Naughton et al15 who showed that race and gender were not significant independent predictors of discharge to the community. The significantly better improvement in ADLs seen in the GRU patients could be because GRU patients were significantly worse in ADLs on admission, therefore they may have shown some regression to the mean. However, the differences in ADLs at admission were not clinically significant. Furthermore, the samples were relatively homogeneous with respect to other important variables such as the patients’ average age and their admitting diagnosis so that we are fairly confident that the 2 facilities were treating similar patient populations. The small sample size might acKauh et al. 325
count for some of the results that were not significant. Despite these limitations, the size of the impact of the intervention on length of stay alone warrants randomized trials of this model of post–acute care. CONCLUSION Our results suggest that increased attention to the medical needs of patients requiring rehabilitation in an SNF (ie, the GRU intervention) may have a significant positive impact on patient outcomes since patients whose medical issues were adequately addressed may have been able to tolerate more therapy for a longer period of time. This might then account for the shorter length of stay observed in the intervention facility. Such improvements could greatly benefit an increasingly overburdened health care delivery system. Further randomized trials that incorporate cost-effectiveness data are needed to substantiate a significant benefit on improvement in patient outcomes and GRU cost savings. If further studies support the benefits, the GRU intervention should become the standard of care in all SNFs in the future. ACKNOWLEDGMENTS The authors thank Joe VanNostran, Velva Spencer, Wyant Woods GRU interdisciplinary team members, CommuniCare Health Services, Tom Lehner, MD, David Jarjoura, PhD, and Judy Agner for their support with this project. REFERENCES 1. Basford J. Goals and goal setting in rehabilitation research. Arch Phys Med Rehabil 1999;80:479 – 480. 2. Boult C, Boult L, Pacala J. Systems of care for older populations of the future. J Am Geriatr Soc 1998;46:499 –505.
326 Kauh et al.
3. Hoenig H, Nusbaum N, Brummel-Smith K. Geriatric rehabilitation: State of the art. J Am Geriatr Soc 1997;45:1371–1381. 4. Applegate W, Miller S, Graney M, et al. D. A randomized controlled trial of a geriatric assessment unit in a community rehabilitation hospital. N Engl J Med 1990;322:1572–1578. 5. Kane R, Chen Q, Blewett L, Sangl J. Do rehabilitative nursing homes improve the outcomes of care? J Am Geriatr Soc 1996;44:545–554. 6. Reuben D, von Sternberg T, Hepburn K, et al. Post-hospital sub-acute care: An example of a managed care model. J Am Geriatr Soc 1997;45: 87–91. 7. Reubenstein L, Stuck A, Siu A, Wieland D. Impacts of geriatric evaluation and management programs on defined outcomes: Overview of evidence. J Am Geriatr Soc 1991;39(Suppl):15S. 8. Weinberg D, Levine J. Clinical areas of liability: Risk management concerns in long-term care. Ann Long-Term Care 2005;13:26 –32. 9. Reuben D, Schnelle J, Buchanan J, et al. Primary care of long-stay nursing home residents: Approaches of three health maintenance organizations. J Am Geriatr Soc 1999;47:131–138. 10. Reubenstein L, Josephson K, Wieland D, English P. Effectiveness of a geriatric evaluation unit. N Engl J Med 1984;311:1664 –1670. 11. Burl J, Bonner A, Rao M, Khan A. Geriatric nurse practitioners in long-term care: Demonstration of effectiveness in managed care. J Am Geriatr Soc 1998;46:506 –510. 12. Levenson S. The future of subacute care. Clin Geriatr Med 2000;16:683– 699. 13. Rehabilitation Outcome Measure for Physical Therapy. Copyright 1994 SCRS & Communicology, Inc. of Ohio. Revision 2, 1996. 14. Heaton WH, ed. Long-term care. Resident Instrument Users Manual, 2003 edition. Miamisburg, OH: Med-PASS Inc.; 2002. 15. Naughton B, Saltzman S, Priore R, Reedy K, Mylotte J. Using admission characteristics to predict return to the community from a post-acute geriatric evaluation and management unit. J Am Geriatr Soc 1999;74: 1100 –1104. 16. Chen C, Heinemann A, Granger C, Linn R. Functional gains and therapy intensity during subacute rehabilitation: A study of 20 facilities. Arch Phys Med Rehabil 2002;83:1514 –1523.
JAMDA – September/October 2005