This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation

Therapy Use and Discharge Outcomes for Elderly Nursing Home Residents

^a Cookingham Institute, Bloch School of Business and Public Administration, University of Missouri at Kansas City, MO
^b Currently at Medicalodges, Inc., Kansas City, KS

Correspondence: Greg Arling, PhD, Room 308, Bloch School of Business and Public Administration, University of Missouri at Kansas City, 5100 Rockhill Road, Kansas City, MO 64100. E-mail: arlingg{at}umkc.edu.

May 12, 1999 Accepted March 6, 2000 Decision Editor: Vernon L. Greene, PhD

	Abstract

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 
This study examines therapy use and discharge outcomes (community discharge, mortality, or remaining in the facility) over a 90-day period for 1,419 elderly, post-acute care nursing home admissions in South Dakota. Subjects met criteria as rehabilitation candidates (i.e., absence of serious behavioral or medical conditions that would limit rehabilitation potential). Receipt of therapies was related significantly to age (younger), Medicare coverage, hip fracture or stroke diagnosis, absence of cancer diagnosis, and resident or staff expectations for functional improvement. Therapy use was related positively to community discharge and negatively to mortality when controlling for covariates such as age, marital status, payment source, functional status, cognitive status, and major diagnoses. Also, community discharge was related positively to the facility's volume of therapy provision and percentage of Medicare-covered stays.

Key Words: Rehabilitation • Medicare • Mortality • Community discharge • Polytomous regression

During the 1990s increasing numbers of elderly nursing home residents had access to licensed therapies. Therapies and other rehabilitation services are intended to improve functional status, promote independence, and enhance quality of life. Even though therapies can be quite costly, they can have long-term benefits in the form of reduced service use or early discharge from the institutional setting. However, growth in access to therapies for elderly nursing home residents is likely to be slowed if not reversed by provisions of the 1997 Balanced Budget Act. These provisions place limitations on Medicare coverage and change the method of payment for rehabilitation services. Payment for licensed therapies is to be bundled into the daily per diem for Medicare skilled nursing facility (SNF) care; a limit is to be placed on the annual payment for Medicare Part B therapies; and facilities are to bill for all therapy use (and other Medicare services), whether provided by facility staff or an outside contractor. These measures are designed to simplify the administration of the Medicare payment system and to control costs. Although provisions of the Act can discourage inappropriate or inefficient provision of therapies to SNF residents, they also could limit access to care and have unintended negative impacts on resident health and functional status. Additional information about use of therapies and their effectiveness for nursing home residents has important implications in clinical decision making, care planning, and public policy.

	Previous Research

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 
Despite the importance of issues surrounding access to therapies, relatively little research has been reported concerning factors affecting therapy use and its impact on nursing home residents. Most studies of therapy use among nursing home residents have relied on small samples (Chiodo, Gerety, Mulrow, Rhodes, and Tuley 1992; Mulrow et al. 1994; Przybylski et al. 1996) or dealt with subgroups of the elderly population, such as hip fracture or stroke patients (Bonar, Tinetti, Speechley, and Cooney 1990; Fitzgerald and Dittus 1990; Kiel, Eichorn, Intrator, Silliman, and Mor 1994; Williams, Oberst, and Bjorklund 1994). These studies have found that more intense therapies (minutes per day or days per week) tend to produce better functional outcomes in mobility or other activities of daily living (ADLs) compared to less intense therapy (Chiodo et al. 1992; Przybylski et al. 1996) or no therapy at all (Mulrow et al. 1994). Discharge outcomes (home or community care vs remaining in the facility) also were associated with amount of therapy as well as other factors, such as younger age, lower levels of functional impairment at admission, or comorbidities (Fitzgerald and Dittus 1990; Kiel et al. 1994; Williams et al. 1994).

Two large prospective studies have examined functional or discharge outcomes for patients with hip fracture or stroke who received therapies in either rehabilitation facilities or nursing homes (Kane, Chen, Blewett, and Sangl 1996; Kane et al. 1998; Kramer et al. 1997). Both studies found that stroke patients had more favorable outcomes when they were treated in rehabilitation hospitals; however, the setting for care seemed to make little difference in outcomes for hip fracture patients. These studies also divided nursing facilities into "rehabilitative" versus "regular" nursing homes (Kane et al. 1996, Kane et al. 1998) or "subacute" versus "traditional" nursing homes (Kramer et al. 1997). Outcomes for stroke patients tended to be better in a rehabilitative or subacute nursing facility as opposed to a regular or traditional nursing facility, yet hip fracture patients fared about the same in either setting. Both studies adjusted for a series of covariates (e.g., ADL and other impairments, cognitive status, or medical diagnoses) when drawing comparisons between care in the different settings. Neither study looked specifically at mortality outcomes, although Kane and his colleagues (1996, 1998) included mortality as the endpoint (highest dependency score) in their functional outcome scale.

Large-scale studies of therapy effectiveness can be difficult to manage and costly to conduct. Yet, with the advent of the Minimum Data Set (MDS), a comprehensive assessment instrument required of all residents in certified nursing facilities, it is feasible to examine use of services and care outcomes with routinely collected data. The MDS has proven to be a reasonably reliable and valid instrument for assessing a wide range of resident conditions (Hawes, Phillips, Mor, Fries, and Morris 1992; Hawes et al. 1995), and it could have considerable value as a research tool (Teresi and Holmes 1992). To our knowledge, however, no published studies have reported findings on therapy use or outcomes using the MDS.

	Study Objectives

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 
In this study we draw upon data from MDS assessments performed on a cohort of elderly nursing home admissions in South Dakota. Our objectives are to describe factors related to receipt of therapies at admission to the facility, and to examine three outcomes within 90 days after nursing home admission: community discharge, death, or remaining in the nursing facility. Therapy use is hypothesized to have a direct positive relationship on community discharge and a negative relationship to mortality when controlling for covariates such as pay source, functional status, cognitive status, and major diagnoses. The sample consists of annual admissions to all nursing facilities in South Dakota. The sample has been restricted to post-acute care admissions, age 60 or older, who met our criteria as being likely candidates for rehabilitation.

	Methods

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 
Subjects
The sample comprised 1,419 elderly admissions (age 60 or older) to 105 South Dakota nursing facilities from April 1995 to March 1996. To qualify for inclusion in the study, subjects had to meet criteria for being a rehabilitation candidate. Criteria were established from review of published literature and clinical input from the physical therapist on our research team. They included admission from an acute care hospital, free of serious mental or behavioral problems, and having no diagnoses that might severely limit rehabilitation potential. Of the 4,523 admissions during the study period, 3,105 were excluded from the sample for one or more of the following reasons: not admitted from an acute care hospital (54%), severe behavioral problems (23%), actively resisted care (19%), chronic mental health history (12%), comatose (1%), quadriplegia (1%), stage 3–4 pressure sores (3%), explicit terminal prognosis (4%), or no limitations in mobility or other ADLs (6%). In addition, 7% were below age 60 and 7% had an indeterminate discharge status (i.e., they were transferred to another nursing home or hospital during the 90-day period following admission). There were 113 facilities in the original data set; however, only 105 facilities had admissions meeting our study criteria.

Major Variables
All study variables were taken from MDS+ assessments performed within 7 days of admission. The MDS+, which is a variant of the MDS, contains all of the standard MDS items. Development of the MDS, assessment procedures, and reliability of items are described elsewhere (Frederickson, Tariot, & De Jonghe, 1996; Hawes et al. 1995). Because South Dakota relies on the MDS data for Medicaid case mix reimbursement, state and nursing facility staff pay particular attention to data accuracy. For example, the state has a computerized tracking system to monitor admissions and discharges; it screens each MDS for data entry errors or clinically inconsistent items; and nurses employed by the Medicaid program perform periodic, on-site audits of MDS items. In addition, MDS records are reviewed routinely as part of the licensure and certification process. All facilities are required to maintain and transmit MDS data electronically, and most providers have computer software for error checking and data verification.

Therapy use was defined operationally as the number of minutes per day of licensed therapies (physical, occupational, or speech) that were either performed or scheduled at the time of the assessment. We categorized therapy use by intensity: 15–59 minutes/day and 60+ minutes/day. Another variable measured the number of nursing rehabilitation services received by the resident from nursing staff for at least 15 minutes per day, 5 days per week. These activities were active or passive range of motion and skill training in locomotion, eating, dressing, or transferring. Nursing rehabilitation was a complement to licensed therapies in some cases, and in other cases it was performed for the resident in the absence of therapies. We did not obtain information on the amount billed for therapies so we were not able to validate therapy times reported on the MDS. For purposes of our study, MDS data may be a more accurate measure of service use than would claims or billing data. During this period, before introduction of the Medicare Prospective Payment System (PPS), therapy charges varied considerably by pay source (e.g., Medicare Part A or Part B, Medicaid, private insurance), cost history of the facility, and whether therapies were provided by nursing home staff or under contract.

Discharge outcomes were obtained from a South Dakota supplement to the MDS+ that tracks resident stays. Outcomes were defined as community discharge (discharge to self-care, home care, or an assisted living facility), death, or remaining in the facility after 90 days. Deaths either occurred in the nursing home or were recorded from hospital records. Persons completing the MDS must indicate the discharge date and status for each resident. Deaths are recorded as "died in nursing facility" or "died outside facility." We combined the two categories to create our mortality variable. The biggest potential problem with mortality data would be for residents who are discharged to hospitals and die without returning to the facility. In order to close out the MDS record, facilities normally would record the death on the discharge supplement and submit this information to the state. However, we have no independent method of determining the accuracy of reporting on mortality for persons who are discharged to hospitals or other settings.

Study covariates included background variables, major diagnoses or conditions, and cognitive and functional status. Background variables were age, gender, marital status, and pay source (Medicare Part A, Medicaid, or private pay/other). Major diagnoses or conditions were stroke, aphasia, multiple sclerosis, Parkinson's disease, arthritis, osteoporosis, hip fracture (in the last 180 days), other fracture, daily joint pain, other reports of daily pain, cancer, emphysema/asthma/chronic obstructive pulmonary disease (COPD), diabetes, pneumonia, respiratory infection, fever, vomiting, congestive heart failure, and other cardiovascular diagnoses (arteriosclerotic heart disease, cardiac dysrhythmia, hypertension, peripheral vascular disease). The MDS+ instructions specify that diagnoses recorded on the assessment form must be related to the current health or functional status of the resident. However, the form does not identify a primary diagnosis or reason for admission. This is a study limitation in so far as we cannot pinpoint the exact reason a resident entered the facility or the specific health condition(s) that may have prompted therapy provision.

Cognitive status was measured using the algorithm developed by Fries and colleagues 1994 from MDS+ items dealing with memory, orientation, and decision making. We also used the depression scale developed by Fries and colleagues. Functional status was measured from MDS+ ADL items indicating the resident's self-performance of each task. Items were scored from 0–4 (0 = no assistance or supervision only, 4 = totally dependent on staff). A mobility index, ranging from 0–12, was constructed by summing scores on items for bed mobility, transfer, and locomotion. An index for other ADLs, ranging from 0 to 12, was constructed from dressing, eating, and toileting items. Additional variables measured the resident and direct care staff's expectations regarding potential for improvement in functioning (scored 1 = resident likely to improve, and 0 = resident not likely to improve).

Also, we constructed variables for the proportion of a facility's admissions that were covered by Medicare, and the proportion that received licensed therapies. The denominator for these proportions was the total number of admissions during the study period and not just those who were candidates for rehabilitation. Although these variables do not allow us to distinguish between rehabilitation or subacute versus traditional nursing homes (Kane et al. 1996, Kane et al. 1998; Kramer et al. 1997), they give an indication of the facility's experience in caring for post-acute residents and its volume or scale of operation for therapy services.

The 105 facilities in the study had a mean of 13.5 subjects (SD = 17.1) represented in the sample. The median number of subjects per facility was 9 and the range was 1 to 101, with four facilities having 50 or more subjects. In general, facilities contributing more subjects to the study had a higher proportion of therapy use, Medicare-covered stays, and community discharges. However, there was no significant difference in mortality rates between facilities with small and large numbers of subjects in the study. When facilities were divided into quartiles by number of subjects (ranked from low to high), the mean proportion of therapy use ranged from .28 for the first quartile to .63 for the fourth quartile. The mean proportion Medicare ranged from .12 to .69, and community discharge ranged from .28 to .43. Because facilities with higher therapy use, Medicare participation, and community discharges contributed disproportionately to the study sample, we applied "robust" estimators in our regression models involving facility-level variables (described below).

Analysis
We relied on contingency tables or one-way analyses of variance (ANOVAs), logistic regression, and ordinary least squares (OLS) regression to examine factors associated with therapy use. We then constructed a polytomous regression model for predicting discharge outcomes of community discharge, death, or remaining in the facility. The Statistical Analysis System (SAS Institute 1992) software package was used for all analyses except the polytomous regression, which was performed with STATA (1999). We used the model chi-square and pseudo R-square to assess significance, model fit, and explanatory power. Only those variables with statistically significant (p < .05) parameter estimates are reported in the tables. Because preliminary analysis indicated the importance of some facility effects, residents treated in a facility cannot be considered independent observations. Therefore, all standard errors are based on "robust" Huber/White/sandwich estimators.

Covariates were selected for the models because of their association with therapy use and outcomes of care. For example, hip fracture, stroke, and other impairments have been found to correlate with receipt of services and outcomes (Kane et al. 1996, Kane et al. 1998; Kramer et al. 1997). Also, mortality among nursing home residents has been associated with conditions such as congestive heart failure and other cardiovascular diagnoses (Breuer, Wallenstein, Feinberg, Camargo, and Libow 1998; Fried and Mor 1997; Wang, Mouliswar, Denman, and Kleban 1998), pneumonia or upper respiratory disease (Fried, Gillick, and Lipsitz 1997; Fried and Mor 1997; Marrie and Blanchard 1997; Mehr et al. 1998), and ADL dependency (Breuer et al. 1998; Fried and Mor 1997; Mehr et al. 1998). By entering these covariates into the multiple regression models, we were able to control statistically for competing risk factors and, thus, identify the independent effects of therapies on care outcomes.

	Results

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 
Characteristics of the study sample were what one might expect for a cohort of elderly, post-acute, nursing home admissions. Table 1 displays the sample characteristics overall and by level of therapy use. The residents in our study were relatively old, primarily unmarried females. About three fourths had Medicare (Part A) at admission to the facility. This contrasts with an average of 58% of all facility admissions on Medicare. Residents suffered from a variety of chronic conditions or major diagnoses. The most prevalent conditions were cognitive impairment, daily pain, arthritis, effects of stroke, congestive heart failure or other cardiovascular diagnosis, diabetes, pulmonary conditions (emphysema, asthma, or COPD), and cancer. The average mobility and other ADL dependence scores were 5.8 and 5.2, respectively. In only about one third of the cases did staff or the resident expect improvement in the resident's functional status, that is, greater independence in at least some ADLs. At 90 days after admission, approximately 17% of the residents had died, 41% were discharged to the community, and 42% remained in the facility.

Table 2 shows results from a logistic regression for therapy use (1 = receive and 0 = not receive). Therapy use and all covariates were entered into the model as a block. Statistically significant (p .05) variables from this model are reported in the table. The model chi-square was statistically significant although the pseudo R-square was a modest .169. Results from the logistic regression indicated significant independent effects for the resident's stay being covered by Medicare and being in a facility with a high percentage of admissions covered by Medicare. The facility's percentage of Medicare admissions seemed to have a particularly strong relationship to therapy use with an odds ratio of 4.22. Also, residents with hip fracture, other fracture, stroke, Parkinson's disease, arthritis, and contractures were more likely to receive therapy. Residents with cancer or vomiting were less likely to receive therapy. Finally, resident and staff expectations of improvement were positively related to therapy use.

Discharge Outcomes
Results from the polytomous regression for community discharge or death are shown in Table 3 . Remaining in the facility at 90 days serves as the reference category. All covariates were entered as a block into the model, although only significant variables (p .05) from this model are reported in the table. The model chi-square was statistically significant (p .001), and the pseudo R-square was .242.

We also wanted to determine if intensity of therapy use would be significantly related to discharge outcomes if we looked only at therapy recipients. Therefore, we carried out a separate set of analyses (not reported in the table but available from the authors on request) for community discharge and mortality among the subset of the sample that received therapies (n = 988). In this polytomous regression equation, we entered the number of therapy minutes per day as the independent variable rather than the dummy variables indicating level of therapy use. The mean therapy minutes per day were 44.2 (SD = 24.4). Results from this model were essentially the same as the regression carried out with the full sample. Minutes of therapy per day were positively related to community discharge (p .004) and negatively related to mortality (p .030). Percentage of facility admissions receiving therapy also was significantly related to community discharge (p .000). Among the covariates, marital status (married), Medicare coverage, and daily pain were positively related to community discharge, whereas mobility impairment, cognitive impairment, and stroke were negatively related. Impairment in ADL and diagnoses such as cancer, pulmonary conditions, fever, and congestive heart failure were significantly related to mortality. The model chi-square was statistically significant (p .001) and the pseudo R-square was a modest .227.

	Discussion

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 
The findings from the study tend to support our hypotheses. We focused on post-acute nursing home residents who were candidates for rehabilitation. Among these admissions, access to licensed therapies was greater for residents with Medicare (Part A)-covered stays and those in facilities with a high percentage of admissions on Medicare. Thus, as expected, Medicare played a key role in access to therapies. Furthermore, therapies were most likely to be received by residents with neurological or muscular-skeletal conditions such as hip fracture, other fracture, stroke, Parkinson's disease, and contractures. Persons with cancer were less likely to receive therapies. Finally, receipt of therapies was influenced by resident and staff expectations for improvement in the resident's functional status. It is noteworthy that expectations were so low even among therapy recipients. We infer from this finding that the treatment goal for most therapy recipients in the study was to avoid a further decline in health or functional status rather than achieving significant functional improvement.

Findings also tended to support our hypotheses regarding discharge outcomes. Intense therapy use (60 minutes/day) at nursing home admission was related significantly to community discharge, although lower levels of therapy (15–59 minutes/day) were not significantly related. With respect to mortality, both moderate and intense therapy use, as well as the receipt of nursing rehabilitation, were negatively related to mortality. In a separate analysis among persons receiving therapies, we found similar results. Greater intensity of therapies (minutes per day) was positively related to community discharge and negatively related to mortality. Although we cannot establish causality in this study, findings suggest that therapies may be facilitating community discharge and reducing risk of mortality. However, a relatively intense level of therapies may be required in order to achieve discharges to the community. Therapies also may increase the resident's chance of survival by improving physiologic or functional capacity and by ameliorating effects of acute or chronic conditions.

Several of the findings for study covariates also merit discussion. Married persons, who presumably have a spouse at home to care for them, also were more likely to be discharged to the community. Although mobility and other ADL impairment were unrelated to receipt of therapies, residents who were less mobility-impaired had a greater chance of community discharge. As expected, based on previous research, residents with greater impairment in nonmobility ADLs or diagnoses of cancer, pulmonary conditions, fever, and congestive heart failure had higher mortality. Report of daily pain was positively related to both community discharge and mortality. Some residents with daily pain may have had cancer, which may account for daily pain's relationship to mortality. Other residents with daily pain may have been in the nursing home recovering from surgery and, thus, they might have been candidates for community discharge.

Likewise, being a Medicare recipient was related to both community discharge and mortality. Because Medicare is targeted to post-acute patients, some Medicare-covered admissions may have been more acutely ill and, thus, might have had a higher mortality risk than admissions with other pay sources. Conversely, Medicare coverage requires a relatively high patient copayment after 20 days of a nursing facility stay. Therefore, residents whose conditions improved may have had a financial incentive to seek early discharge if they could arrange for alternative community placements.

Cognitively impaired residents were less likely to receive therapies or to be discharged to the community, although they had no greater risk of mortality. Neither age nor gender was significantly related to discharge outcome. At least among elderly nursing home admissions, the very old were no less likely than their younger counterparts to be discharged to the community. They were just as likely to survive for at least 90 days after admission. Finally, somewhat surprisingly, neither staff nor resident expectations for functional improvement were predictive of community discharge. The MDS variable, however, does not measure strength of expectations, that is, very high or only modest expectations for improvement, nor does it inquire specifically about expectations for discharge to the community.

Facility-level factors also seemed to play an important role in discharge outcomes. As we noted earlier, the facility's percentage of Medicare admissions was strongly associated with therapy use. Moreover, the volume or scale of the facility's therapy services (percentage of admissions receiving therapies) was strongly associated with community discharge, suggesting that a facility may require a minimum scale of operation or experience, as evidenced by therapy volume, in order to facilitate community discharge. These findings are consistent with results reported by Kane and colleagues 1996 and Kramer and colleagues 1997 regarding effectiveness of rehabilitation-oriented compared to traditional nursing homes. In addition, we discovered that residents with stroke diagnoses were less likely to be discharged to the community. This finding lends support to the observation that nursing homes may be less effective in achieving favorable outcomes for stroke patients than for residents with other conditions such as hip fracture (Kane at al., 1996; Kramer et al. 1997).

Limitations of the Study
Our findings must be qualified in several respects. First, we focused on candidates for rehabilitation or those residents admitted from acute care hospitals and without behavioral problems or other conditions that might impede the therapy process. Therefore, results may not be generalizable to the total nursing home population, particularly those admitted from community or other nonacute care settings. Second, the study must be qualified because it took place in a single, rural state (South Dakota) with a small population. Findings may not be generalizable to other states with larger, more concentrated populations and with more urban-based nursing facilities. On the other hand, South Dakota offers a "natural" contrast between facilities with very little Medicare participation or a low volume of therapy use and facilities with high Medicare participation and ready access to therapies.

Third, results about the impact of therapies may be called into question because of potential selection bias. We employed a nonexperimental, ex post facto design that relied on natural variation in access to therapies as opposed to conducting a controlled clinical trial with random assignment. We attempted to isolate the independent effects of therapies by narrowing the sample to rehabilitation candidates and by controlling statistically for a large number of covariates that may have been related to both therapy use and discharge outcomes. Nonetheless, statistical controls do not substitute for random assignment when it comes to internal validity threats (Mohr 1995). Selection criteria, such as Medicare coverage and diagnosis of stroke or cancer, seemed to play a role in both therapy provision and discharge outcomes. Yet, underlying factors, not measured in the study, may also have been responsible for the association between therapy use and discharge outcomes. Referring physicians or staff may have provided more intense therapies to residents who they believed to have greater community discharge potential, and they may have excluded from therapies residents who were medically unstable or at greatest risk of mortality. Validity problems are compounded by the absence of information on primary diagnosis or reasons for therapy provision.

On the other hand, a controlled trial that involved random assignment to a therapy or nontherapy condition is not feasible for this largely Medicare population. It would be difficult to gain approval from Medicare to conduct a randomized trial and, even if it were approved, many Medicare beneficiaries would no doubt opt out of the study, making results difficult to generalize. Potential selection bias is an unavoidable problem if one must rely on a nonexperimental design. As a check on selection bias, we had at least limited information on staff expectations regarding functional improvement. Although staff expectations were related significantly to therapy provision, neither staff nor resident expectations were significantly related to community discharge. Staff expectations and the process of selecting persons for therapies are complicated by the fact that many older nursing home admissions suffer from multiple chronic conditions which make it difficult to determine, in advance, how they will respond to therapy. Other nonclinical factors, such as Medicare participation or the availability of therapists (particularly in rural settings) may be just as influential in staff decision making.

Policy Implications
Despite these caveats, study findings have implications for Medicare policy. Medicare is moving from a cost-based to a prospective pricing approach for skilled nursing facilities. Both routine and ancillary services (including therapies) are being bundled into one payment, and facilities are being required to submit a consolidated bill for services. In addition, Medicare recipients now are subject to an annual limit on the amount of Part B therapies billable to Medicare. Medicare will pay for a maximum of $1,500 per year of occupational therapy and $1,500 of physical and speech therapy combined. These measures are intended to control costs, increase efficiency, and discourage inappropriate care. Some advocacy groups and members of the provider community have argued that these changes will discourage appropriate provision of services. That is, margins may be squeezed to the point that providers no longer are able to cover therapy costs for residents who need or could benefit from these services. Our study did not take into account the costs or amounts paid for therapies, so we cannot draw conclusions about the efficiency of providers in the study. Changes in Medicare reimbursement may encourage providers to lower unit costs without adversely affecting access to care. On the other hand, facilities may lower intensity of therapies, reduce the scale of their therapy operations, or be less willing to admit Medicare residents. Our study's findings suggest that these responses to the 1997 Balanced Budget Act could have a deleterious effect on community discharges and mortality. Medicare cost savings in payment for ancillary services may be offset by cost increases due to longer facility stays. These issues need to be closely monitored as new Medicare provisions are implemented.

Questions for Future Research
Finally, our study demonstrates the feasibility of using MDS data for assessing resident outcomes in nursing facilities. Although these data do not have the benefit of experimental control, they appear useful in studying outcomes longitudinally, on a large scale, and in natural settings. Additionally, MDS data represent a rich and readily available source of information for outcome studies by nursing facilities, health care plans, therapy organizations, geriatricians, and other providers. Our study raises a series of important questions for future research. Who is receiving therapies and what changes have taken place since the introduction of Medicare PPS? How does the intensity of therapies relate to functional change and discharge outcomes? Do specialized nursing facilities or those with a high volume of therapy provision achieve better outcomes? How are therapy outcomes influenced by clinical conditions, such as stroke, hip fracture, or cancer, and what types or levels of therapy should be provided to different residents? Do therapies reduce mortality risk in addition to their impact on functional status? What is the cost-effectiveness or benefit-cost of therapies? And, how should cost considerations enter into clinical decision making or policy making for nursing facilities?

	Acknowledgments

 
This research was supported by the South Dakota Department of Social Services as part of the Multistate Nursing Home Case Mix and Quality Demonstration, Health Care Financing Administration. We thank Carol Job, Damian Prunty, Elizabeth Cornelius, and the staff and residents of South Dakota nursing homes for their assistance in the study. The views expressed are those of the authors and not the sponsoring organizations.

	References

TOP Abstract Previous Research Study Objectives Methods Results Discussion References

 

• Bonar S.K., Tinetti M.E., Speechley M., Cooney L.M., 1990. Factors associated with short- versus long-term skilled nursing facility placement among community-living hip fracture patients. Journal of the American Geriatrics Society 38:1139-1144. [Medline]
• Breuer B., Wallenstein S., Feinberg C., Camargo M.F., Libow L.S., 1998. Assessing life expectancies of older nursing home residents. Journal of the American Geriatrics Society 46:954-962. [Medline]
• Chiodo L.K., Gerety M.B., Mulrow C.D., Rhodes M.C., Tuley M.R., 1992. The impact of physical therapy on nursing home patient outcomes. Physical Therapy 72:168-173. [Abstract/Free Full Text]
• Fitzgerald J.F., Dittus R.S., 1990. Institutionalized patients with hip fractures: Characteristics associated with returning to community dwelling. Journal of General Internal Medicine 5:298-303. [Medline]
• Fredericksen K., Tariot P., De Jonghe E., 1996. Minimum Data Set Plus (MDS +) scores compared with scores from five rating scales. Journal of the American Geriatrics Society 44:305-309. [Medline]
• Fried T.R., Gillick M.R., Lipsitz L.A., 1997. Short-term functional outcomes of long-term care residents with pneumonia treated with and without hospital transfer. Journal of the American Geriatrics Society 45:302-306. [Medline]
• Fried T.R., Mor V., 1997. Frailty and hospitalization of long-term stay nursing home residents. Journal of the American Geriatrics Society 45:265-269. [Medline]
• Fries B.E., Schneider D.P., Foley W.J., Gavazzi M., Burke R., Cornelius E., 1994. Refining a case-mix measure for nursing homes: Resource Utilization Groups (RUGS III). Medical Care 32:668-685. [Medline]
• Hawes C., Morris J.N., Phillips C.D., Mor V., Fries B.E., Nonemaker S., 1995. Reliability estimates for the Minimum Data Set for nursing resident assessment and care screening. The Gerontologist 35:172-175. [Abstract]
• Hawes C., Phillips C.D., Mor V., Fries B., Morris J.N., 1992. MDS data should be used for research. The Gerontologist 32:563-564.
• Kane R.L., Chen Q., Blewett L.A., Sangl J., 1996. Do rehabilitative nursing homes improve the outcomes of care?. Journal of the American Geriatrics Society 44:545-554. [Medline]
• Kane R.L., Chen Q., Finch M., Blewett L.A., Burns R., Moskowitz M., 1998. Functional outcomes of posthospital care for stroke and hip fracture patients under Medicare. Journal of the American Geriatrics Society 46:1525-1533. [Medline]
• Kiel D.P., Eichorn A., Intrator O., Silliman R.A., Mor V., 1994. The outcomes of patients newly admitted to nursing homes after hip fracture. American Journal of Public Health 84:1281-1286. [Abstract/Free Full Text]
• Kramer A.M., Steiner J.F., Schlenker R.E., Ellertsen T.B., Hrincevich C.A., Tropea D.A., Ahmad L.A., Eckhoff D.G., 1997. Outcomes and costs after hip fracture and stroke: A comparison of rehabilitation settings. Journal of the American Medical Association 277:396-404. [Abstract/Free Full Text]
• Marrie T.J., Blanchard W., 1997. A comparison of nursing home-acquired pneumonia patients with community-acquired pneumonia and nursing home patients without pneumonia. Journal of the American Geriatrics Society 45:50-55. [Medline]
• Mehr D.R., Zweig S.C., Kruse R.L., Popejoy L., Horman D., Willis D., Doyle M.E., 1998. Mortality from lower respiratory infection in nursing home patients: A pilot prospective community-based study. Journal of Family Practice 47:298-304.
• Mohr L.B., 1995. Impact analysis for program evaluation Sage Publications, Thousand Oaks, CA.
• Mulrow C.D., Gerety M.B., Kanten D., Cornell J.E., DeNino L.A., Chiodo L., Aguilar C., O'Neil M.B., Rosenberg J., Solis R.M., 1994. A randomized trial of physical rehabilitation for very frail nursing home residents. Journal of the American Medical Association 271:519-524. [Abstract/Free Full Text]
• Przybylski B.R., Dumont E.D., Watkins M.R., Warren S.A., Beaulne A.P., Lier D.A., 1996. Outcomes of enhanced physical and occupational therapy service in a nursing home setting. Archives of Physical Medicine and Rehabilitation 77:554-561. [Medline]
• SAS Institute1992. SAS-STAT manual (version 6) SAS Institute, Cary, NC.
• STATA1999. STATA release 6.0 Stata Press, College Station, TX.
• Teresi J.A., Holmes J., 1992. Should MDS data be used for research?. The Gerontologist 32:148-151. [Medline]
• Wang R., Mouliswar M., Denman S., Kleban M., 1998. Mortality of the institutionalized old-old hospitalized with congestive heart failure. Archives of Internal Medicine 158:2464-2468. [Abstract/Free Full Text]
• Williams M.A., Oberst M.T., Bjorklund B.C., 1994. Early outcomes after hip fracture among women discharged home and to nursing homes. Research Nursing and Health 17:175-183.

This article has been cited by other articles:

				F. H. Decker Dying in a Nursing Home: The Role of Local Bed Supply in Nursing Home Discharges J Aging Health, February 1, 2008; 20(1): 66 - 88. [Abstract] [PDF]

				J. Y. Yip, K. H. Wilber, and R. C. Myrtle The Impact of the 1997 Balanced Budget Amendment's Prospective Payment System on Patient Case Mix and Rehabilitation Utilization in Skilled Nursing Gerontologist, October 1, 2002; 42(5): 653 - 660. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation