U.S. Department of Health and Human Services

Medicare Post-Acute Care: Quality Measurement Final Report

Appendix A. 1998 Reports

REPORT 1: Selecting and Evaluting Eight Targeted Conditions

REPORT 2: Selecting the Targeted Conditions

SELECTING AND EVALUATING EIGHT TARGETED CONDITIONS

Korbin Liu and Barbara Gage, The Urban Institute

Andrew Kramer, University of Colorado Health Sciences Center

June 1998

This report was prepared under contract #HHS-100-97-0010 between the U.S. Department of Health and Human Services (HHS), Office of Disability, Aging and Long-Term Care Policy (DALTCP) and The Urban Institute. For additional information about the study, you may visit the DALTCP home page at http://aspe.hhs.gov/daltcp/home.htm or contact the ASPE Project Officer, Jennie Harvell, at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, SW, Washington, DC 20201. Her e-mail address is: Jennie.Harvell@hhs.gov.

Conclusions and opinions in this paper are solely those of the authors and do not necessarily reflect the views of the Urban Institute, University of Colorado or ASPE.

I. INTRODUCTION

The Administration has signaled its intent to move towards an integrated Medicare post-acute care financing and service delivery system. In order to integrate post-acute care services, information is needed on the cost-effectiveness of different post-acute care providers for different conditions. This project will design a set of quality of care measures for Medicare post-acute care services provided by SNFs, HHAs, rehabilitation hospitals and long-term care hospitals. The project will focus on four clinical conditions that are found in the post-acute care population. As a preliminary step to selecting the four conditions, the range of possibilities will first be narrowed to eight conditions. This report describes our analysis leading to the selection of the eight conditions.

We think that, as a general rule, the criteria used to identify the eight conditions at this stage should have policy-related implications. This orientation helps to make the findings from the eventual study more valuable to BBA prescribed and other reform initiatives directed at Medicare post-acute care services and the growth of the Medicare risk HMO program. This orientation also complements, rather than duplicates, our current plans to employ clinically-related criteria to narrow the eight conditions to the four conditions that will be the focus of the eventual outcomes study.

In light of the goal of this project to examine the relative quality outcomes of treating specific conditions across post-acute care providers, a necessary criterion for the selection of the conditions is that they are being treated in more than one post-acute care provider modality. For this study, we are examining the types of post-acute care cases that receive services from skilled nursing facilities (SNFs), home health agencies (HHAs) and rehabilitation facilities (RFs), and to a lesser extent long-term care hospitals. Although it would be particularly desirable to select conditions that are found in the caseload of all three provider types, the nature of RFs implies that only the subset of post-acute care patients that require intensive, inpatient rehabilitation will be found in this setting. A substantial proportion of post-acute care patients do not need this level of therapy services, but do need skilled nursing, medical social work and aide assistance. These services can be provided either in SNFs or by HHAs. Hence, we will not restrict the selection to only conditions that can be observed in all three settings. Rather, we will include some conditions that are treated largely in RFs or SNFs and others that are treated primarily in SNFs or by HHAs (or in some cases long-term care hospitals).

As has been noted in prior reserch (Steiner and Neu, 1993; ProPAC, 1996), post-acute care use of SNFs, HHAs and RFs is concentrated in relatively few types of cases (at least in terms of broad categorizations of patients, such as by DRGs). This situation persisted in 1995. We examined the 39 most prevalent DRGs among post-acute care users of SNFs, HHAs, and RFs, and found that those selected DRGs accounted for 53% of all SNF episodes, 36% of all HHA episodes and 68% of all RF episodes. Hence, we anticipate that selection of conditions based on some variant of those DRG categories will provide information on a sizeable proportion of Medicare’s post-acute care population.

We envisioned three steps in the process to arrive at the eight conditions that will be proposed for further analysis. The first step is to conceptualize methods for defining post-acute care conditions and identify selected conditions using this framework. The second is to establish criteria for choosing eight conditions among the unlimited number of possible conditions. The final step is to apply the criteria to narrow down the range of possibilities. The following sections address the three steps.

II. APPROACHES FOR DEFINING CONDITIONS

Various approaches are possible to define conditions for studying outcomes of post-acute care services. The commonly used approach in past studies is the selection of conditions on the basis of hospital DRGs or diagnosis (e.g. hip fracture, stroke, congestive heart failure). Two major studies of outcomes of post-acute care services across different modalities were conducted by Kramer and colleagues (1997) and Kane and colleagues (1997a, 1997b). Kramer examined the outcomes of hip fracture and stroke patients, selected on the basis of hospital diagnoses representing the two conditions. Kane selected cases based on the hospital DRGs. Along with hip fracture (DRG 210) and stroke (DRG 14) cases, Kane examined outcomes of people discharged from hospitals with congestive heart failure (DRG 127), chronic obstructive pulmonary disease (DRG 88) and hip replacement procedures (DRG 209). The selection of those hospital diagnoses or DRGs was not surprising because hip fractures and strokes are two of the most prevalent conditions treated by Medicare post-acute care providers. Moreover, the five DRGs selected by Kane accounted for almost 30 percent of Medicare-sponsored post-acute services (ProPAC, 1996).

Another approach for selecting conditions is to define them by functional measures (e.g., disability prior to the acute event, ADL disability after the acute event, changes in ADL disabilities before and after the acute event). This approach reflects the need for rehabilitation services, which is one of the major causes for people to receive post-acute care. Because many post-acute care patients require more skilled nursing care than rehabilitation services, they may be underrepresented if this approach is used. On the other hand, medical post-acute care patients may also be functionally dependent at various times during the episode of acute and post-acute care. Another potential drawback of this approach for selecting conditions is the variation in medical conditions that is associated with functional disability. Because disabled populations have such varied medical problems, it may be difficult to predict outcomes on the basis of functional status. Moreover, outcomes based on disabilities may be influenced by the accessibility and use of informal care; comparisons between post-acute care providers may be confounded if such factors are not controlled for in analyses.

A third option is to define conditions by care requirements (e.g., ventilator dependent, wound care, rehabilitation). As with the preceding approach, this strategy for selecting groups of post-acute care patients may be constrained by the diversity of underlying conditions of people requiring particular types of care. For quality measurement purposes, it is is important that study groups are relatively homogeneous so that differences in outcomes are related to site of care and not underlying medical needs. Thus, if similar quality measures can be used for all patients that are ventilator-dependent or that require wound care, it would not be critical to distinguish the exact diagnosis underlying the need for a ventilator or for wound care. On the other hand, we might have to select, for the study, subgroups of ventilator-dependent people by underlying medical conditions and levels of comorbidity (See Kramer, et al. 1990 for a further discussion of patient classification and quality measure development).

III. SELECTION CRITERIA

We reflected on various criteria that could be used to aid in the selection process and have arrived at four categories: (1) Overlap across providers of potential users of post-acute care; (2) High volume cases, either as a proportion of hospital discharges or as a high post-acute care admission group, (3) Medicare service use patterns; (4) Analytical practicality.

Overlap across post-acute care providers. This criterion, as noted above, is a requirement of the study. The aim is to develop quality measures to determine if treatment modality is responsible for differences in outcomes, all other things being equal. As a first step, it is important to be able to make comparisons between the different Medicare post-acute care providers. According to this criterion we will identify conditions that are found in substantial numbers across the different providers. Because rehabilitation facilities can only admit patients who require and can tolerate three hours of therapy per day, some conditions will be those that are treated largely in either RFs or SNFs. Conditions will also be chosen that may be treated either in SNFs or HHAs (and if possible long-term care hospitals).

High volume. Because of the policy-orientation of condition selection at this stage, we include the criterion of high Medicare volume of post-acute care services. For the eventual demonstration project to have program impact, the conditions selected will have to occur in relatively high proportions among Medicare’s post-acute care users or be a relatively high proportion of Medicare’s costs for post-acute care services. This approach is consistent with most of the prior research on post-acute care services, starting with the Rand studies in the 1980's (Neu and Harrison, 1989, Steiner and Neu, 1993).

In addition, the BBA mandated that, for selected hospital DRG’s, cases involving transfers from PPS hospitals to post-acute care providers would result in adjustments in the payment amount of those DRGs. The DRGs were chosen because they represent the highest volume of discharges to post-acute care or a disproportionate use of post-acute care services. The 10 specific DRG’s that have been selected by the Health Care Financing Administration are: strokes (DRG 14), amputations for circulatory system disorders (DRG 113), hip and limb procedures (DRGs 209-211), hip fractures (DRG 236), skin graft/debridement for skin ulcers (DRGs 263-264), organic disturbances and mental retardation (DRG 429), and tracheostomies (DRG 483). The adjustment in DRG payments could have an effect on the pattern of care for patients with those DRGs. Because of the potential change in practice patterns, it would be useful to select conditions in this group so more information will be available to determine if quality is negatively affected by the change in payment method.

Utilization patterns. This criterion covers a range of Medicare service use issues that may be important to consider in the selection of specific conditions to track across post-acute care episodes. Four issues illustrate this criterion.

First, a substantial proportion of post-acute care episodes ends with a hospital readmission. It can be shown, for example, that the rate of hospital readmission varies by DRG. Hospital readmission could reflect severity of patients’ condition and/or quality of care. Regardless, hospital readmission probably reflects higher patient costs. While hospital readmission can be used as an outcome measure, it might also be useful in selecting types of conditions to review because it provides an indication of relatively higher (lower) Medicare episode costs.

Second, some post-acute care patients receive care, in sequence, from more than one type of post-acute care provider. Selecting conditions for which patients frequently use more than one type of post-acute care provider can be useful in not only targeting relatively high cost cases, but also ones in which quality of care may be more questionable.

Third, because of the emergence of “subacute” care, which refers to treatment for patients with more complex needs, it may be important to select conditions that are characteristically viewed as requiring this level of care. Lewin-VHI (Manard, Bieg, Cameron, et al., 1995) concluded that the term subacute care tends to refer to a level of care that some nursing facilities, home care providers and others have been providing for years under a variety of different names (e.g., high end skilled care). Ventilator-dependent patients, for example, are often cited as illustrative of subacute care patients. Other “types of illnesses” treated by subacute care providers include stroke, hip fracture, spinal cord injury, pulmonary disease, and cardiovascular disease. Learning about the quality of subacute care is also motivated by the growing use of this level of care in the managed care environment, and particularly in situations where subacute care replaces inpatient hospital care.

Fourth, when the casemix-adjusted prospective payment systems for SNFs, HHAs and RFs are implemented, there may be some mismatches between payments and costs of specfic casemix categories. Conditions that are potentially subject to such mismatches would be important to consider in light of possible changes in utilization that reflect access and quality problems.

Analytical practicality. The final criterion refers to the practical aspects of selecting particular conditions for analysis. Assuming that a subset of the conditions that we select at this stage will eventually be the ones that are analyzed for the design of the demonstration, it is important to consider ways to minimize subsequent problems even at this stage of selecting the initial eight conditions. We identified two issues that fall under the criterion of analytical practicality.

First, a condition that is selected should include cases that are relatively homogeneous, thereby enhancing our ability to compare outcomes across provider modalities. Although risk adjustment strategies will be used in the eventual study, that process cannot fully compensate for potential variability of a given condition.

Second, the conditions that are chosen should be ones that are relatively recognizeable and understandable to a policy audience as well as to clinical ones. Applying this selection criterion will help to enhance the dissemination of study findings and the likelihood that results may be incorporated into future policy analyses or policies governing post-acute care services.

IV. APPLICATION OF CRITERIA IN THE SELECTION OF EIGHT CONDITIONS

We examined different approaches for defining conditions and employed the criteria discussed in the previous section to identify likely candidates for the eight conditions. This section discusses the rationale behind our selection.

A. Approach for Defining Conditions

As noted above, few studies have addressed patient outcomes when making comparisons across provider types. The conditions selected in those studies were either hospital DRGs or hospital diagnoses. We started with this strategy, partly because of the literature and partly because such DRG and diagnosis information is available from Medicare claims data. In current research using 1995 Medicare claims data, we have already constructed episodes of hospital and post-acute care use. Hence, it is posssible for us to examine expeditiously conditions based on this information source. Moreover, the claims data also allow us to examine conditions based on use of particular Medicare services (e.g., ventilators).

A major shortcoming of the Medicare claims data, for this project, is that it provides no information on functional status (e.g., ADLs, IADLs). Hence, using this data source precludes the possibility of establishing conditions based on disability. Although we considered examing other sources of information, such as the Medicare Current Beneficiary Survey, which collect information on functional status of Medicare enrollees, we focused our efforts on the claims data because of limitations in those sources. Notably, the number of cases of post-acute care use of survey samples is very small compared to the number of cases when using Medicare claims. The sample size problem of surveys means that relatively few post-acute care events can be measured. Sample size limitations are exacerbated when we focus on subgroups of beneficiaries (e.g., persons with more than 3 ADL dependencies) or when we focus on patients using particular types of post-acute care providers (e.g., rehabilitation facilities).

Episodes of post-acute care use. We employed data from the 5-percent sample of 1995 Medicare claims data to examine episodes of post-acute care use. By creating episodes, in which SNF, HHA, and RF use was linked to hospital stays and to each other, it was possible to identify nationally representative patterns of post-acute care when a single or multiple providers were involved. We could also relate the post-acute use to the DRGs or diagnoses of the qualifying hospital stay, and any other information on the hospital claims, including services used (e.g., ventilator). We could also measure the duration of the total episode of post-acute care use.

Initial selection of DRGs. Because of the importance to this study of identifying conditions that are found in multiple post-acute care settings and occur in substantial numbers, we used an initial screening process to select cases that met the two criteria. We sorted post-acute care use of SNFs, HHAs, and RFs, respectively, by the DRG of the preceding hospital stay. We selected the 20 most prevalent DRGs for each of the three provider types. Because certain DRGs (e.g., hip fracture, stroke) were found in the top 20 lists of more than one provider type, we ended up with 32 unique DRGs from the original list of 60. Finally, we created a file containing all hospital stays with those 32 DRGs, regardless of whether it was followed by use of any of the post-acute care providers. These DRGs accounted for two-thirds of all Medicare post-acute care episodes. The distribution by discharge destination (including no post-acute care use) of these high volume DRGs is presented in Table 1.

Based on the 5-percent file, we identified a total of 157,280 episodes involving the 32 DRGs. For the 32 DRGs as a whole, only 51.6 percent of the episodes did not have post-acute care. The most common type of post-acute care service was home health care (24.4 percent), followed by SNF care (12.6 percent). Rehabilitation facilities accounted for almost 2 percent of the episodes involving the DRGs, and 9.2 percent of those episodes involved the use of more than one of the three post-acute care providers.

The numbers of post-acute care episodes by DRG show that many of the DRGs from previous studies continued to be important in 1995. These include stroke (DRG 14 with 12,002 cases), chronic obstructive pulmonary disease (DRG 88 with 7,569 cases), heart failure (DRG 127 with 14,165 cases), major joint replacement procedures (DRG 209 with 12,969 cases), and hip and femur procedures (DRGs 210 and 211, with a combined 5,926 cases).

B. Application of Criteria

High volume and multiple providers. With the data from Table 1, we derived the DRG distributions of discharge locations only for cases that did use post-acute care; effectively, we excluded the cases that did not use any of the three post-acute care modalities. These distributions are presented in Table 2. From this table, we can identify specific DRGs that meet the high volume and/or multiple providers criteria.

The number of post-acute care episodes by DRG in Table 2 ranged from 10,950 for major joint procedure (DRG 209) to 337 for back and neck procedures without complications (DRG 215). As an initial step toward identifying 8 conditions for further consideration, we made a decision rule to focus on DRGs with at least 1,500 episodes in Table 2.¹ Given that we were using the 5 percent claims file, this minimum number is equivalent to 20,000 episodes in 1995. This decision rule defines 17 DRGs in Table 2 as high volume conditions: DRG 14, DRG 15, DRG 79, DRG 88, DRG 89, DRG 106, DRG 121, DRG 127, DRG 138, DRG 148, DRG 174, DRG 182, DRG 209, DRG 210, DRG 296, DRG 320, and DRG 416.

Looking at the distributions of the DRGs by provider type, it is apparent that rehabilitation facilities are not frequent providers for many of them. Some of the DRGs with the highest percentages using RFs are stroke (DRG 14; 10.6 percent), major joint procedure (DRG 209, 9.8 percent), hip procedure (DRG 210, 4.9 percent), and tracheostomy (DRG 483; 10.3 percent). It is interesting to note that, although back and neck procedures (DRGs 214 and 215) did not make the high volume cut-off, a very high proportion (11.1 percent, 10.1 percent) of their episodes involved using rehabilitation facilities The DRGs with high proportions of patients using rehabilitation facilities also tended to have lower than average proportions using HHAs. HHA use was the lowest for DRG 210, but the total number of episodes for DRG 210 is very large.

Given the decision rules that we employed, there are 17 DRGs that are high volume conditions and 3 DRGs (i.e., DRG 214, DRG 215, DRG 483) that are exceptionally interesting for this study. Of those 20 conditions, 5 are relatively well distributed across SNFs, HHAs, and RFs. Because those 5 DRGs also use RFs in relatively high proportions, the remaining 15 high volume DRGs are best suited for paired comparisons between SNFs and HHAs.

Variations in utilization patterns. For the 20 DRGs, we derived summary statistics to examine variations in the patterns of Medicare service utilization, such as hospital readmissions and lengths of stay. These statistics are presented in Table 3.

Hospital readmissions are high for most of the DRGs, ranging from 19.0 percent for back and neck procedures without complications (DRG 215) to 51.7 percent for chronic obstructive pulmonary disease (DRG 88). Most of the DRGs have approximately one-third of cases being readmitted to hospitals within the year. Other DRGs with relatively low proportions of cases readmitted to hospitals are major joint replacement (DRG 209), hip procedure (DRG 210), and back and neck procedure with complications (DRG 214). Other DRGs with relatively high hospital readmission rates include congestive heart failure (DRG 127), acute myocardial infarction (DRG 121), and heart arrythmia (DRG 138).

For all episodes, the median length of stay (LOS) was 33 days. Hip procedures (DRG 210) had a relatively long LOS of 44 days. In contrast, back and neck procedures without complications (DRG 215) had a relatively short median LOS of 23 days. It is important to note that one cause of episodes to end is a hospital readmission. Hence, for DRG 215, when a readmission occurred, it occurred relatively quickly.

Stroke (DRG 14), major joint procedure (DRG 209), hip fracture without major procedure (DRG 210), and back and neck procedure with complications (DRG 214) had among the highest post-acute care episode costs, with stroke, for example, having a median cost of $5,399. This finding is consistent with these cases being rehab DRGs which require considerable amounts of therapy services. Tracheostomy (DRG 483) also had very high costs, with a median of $5,303. In contrast, DRGs with relatively low costs included congestive heart failure (DRG 127), pneumonia (DRG 89), and chronic obstructive pulmonary disease (DRG 88).

The use of different types of post-acute care providers during an episode was highest among patients with stroke (DRG 214), hip fracture procedures (DRG 209, DRG 210), or back and neck procedures (DRG 214, DRG 215). About one-third of the episodes for those DRGs involved use of some combination of SNF, HHA and RF care. In contrast, the DRGs with the lowest proportions using multiple providers of care were arrythmia (DRG 138), esophagitis (DRG 182), and congestive heart failure (DRG 127). These DRGs also had relatively low costs.

V. RECOMMENDATIONS

A summary of characteristics of the 20 high volume or exceptional DRGs is presented in Table 4. The rehabilitation DRGs and the medical DRGs are separated and each group is ranked according to the number of episodes of post-acute care.

The two hip procedure DRGs (DRGs 209 and 210), along with stroke (DRG 14), are very high volume post-acute care cases and are among the 10 “transfer” DRGs. They are also served in relatively high proportions by all three provider types (SNFs, HHAs, and RFs). Moreover, many stroke and hip fracture episodes involve relatively high use of multiple post-acute care providers. Because many hip fractures are treated with a joint replacement (or arthroplasty), we will not define the hip fracture group exclusively by DRG assignment. Rather, we will define one group as hip fracture reflecting the emergent nature of the procedure and include patients from DRG 209, DRG 210, and DRG 211 who meet this criterion. A subgroup of the patients will have had arthroplasty. We do not plan to include elective hip replacements in the eight conditions. Thus, we will include hip fracture from DRG 209, DRG 210, DRG 211 and stroke from DRG 14 for further review. Other strokes coded under DRGs 1 and 5 (craniotomy and extravascular procedures) will also be included with those coded under DRG 14.

Back and neck procedures (DRGs 214 and 215) are interesting conditions that have not been widely studied. Our analysis indicated that people with those DRGs make substantial use of rehabilitation facilities, as well as SNFs and HHAs, so that three site comparisons are possible. In addition, utilization characteristics of back and neck procedure DRGs are somewhat similar to those of hip fracture cases (e.g., low hospital readmissions, high costs). Hence, back and neck procedures present an interesting alternative to such cases as we attempt to study “new” conditions. We recommend including back and neck procedures for further consideration.

Congestive heart failure (DRG 127) and pneumonia (DRG 89) are the highest volume conditions among the medical DRGs. CHF has a relatively high hospital readmission rate and a relatively low likelihood of being treated by multiple providers. Pneumonia is a condition that is easily recognizeable by policy makers and lay people, and while it is a relatively low cost DRG, it is treated in large numbers both in SNFs and by HHAs. More pneumonia cases are likely to be identified in other DRGs such as respiratory infections (DRG 79) and pleural effusions (DRGs 85 and 86). We recommend including these two conditions for further study.

Two other medical conditions that are treated in high volume by post-acute care providers are chronic obstructive pulmonary disease (DRG 88) and circulatory disorders with acute myocardial infarction (DRG 121). Both DRGs have very high use rates in HHAs, but are also treated in SNFs. Both are also relatively low cost and have high hospital readmission rates. Thus, we are also recommending these for further study. For both of these conditions, we anticipate that other cases can be identified by ICD9 codes included under different DRGs. For example, COPD cases may be identified under lung diseases (DRGs 92 and 93) or acute bronchitis (DRGs 96 and 97).

The tracheostomy DRG (DRG 483) appears in all three of the post-acute care settings and is relatively high cost. Moreover, it is a DRG that involves ventilator care, and ventilator care is often noted as a characteristic of “subacute care.” The increasing importance of subacute care in managed care is a significant policy-oriented criterion for supporting the study of ventilator-related conditions. We recommend including tracheostomy (and other ventilator-related conditions) for further study.

In sum, our review resulted in the recommendation of three rehab DRGs, four medical DRGs, and ventilator-related conditions for further study. Several of the recommended conditions are “transfer” DRGs and many of them can be examined in comparisons involving SNFs, HHAs, and RFs.

VI. CONCLUSION

The analysis to recommend 8 conditions for further study focused on DRGs. Although DRGs have been used to select conditions for post-acute care outcome studies, they were examined here principally to identify broad areas for more detailed consideration. In particular, we think that some of the DRGs might be heterogeneous in terms of diagnoses that are included. More precise specification of conditions for further study may involve subsetting of diagnoses in any given DRG or combining diagnoses from multiple DRGs. On the other hand, this initial screen of potential conditions for further study, based on DRG-related conditions, provides a policy-oriented group of conditions that policy and clinical experts can recognize and understand.

REFERENCES

Kane, R.L., Q. Chen, M. Finch, et al., 1997. “Functional Outcomes of Post-Hospital Care for Stroke and Hip Fracture Patients Under Medicare.” Manuscript, University of Minnesota School of Public Health.

Kane, R.L., Q. Chen, M. Finch, et al., 1997a. “The Optimal Outcomes and Costs of Post-Hospital Care Under Medicare.” Manuscript, University of Minnesota School of Public Health.

Kramer, A.M., J.F. Steiner, R.E. Schlenker, et al., 1997b. “Outcomes and Costs After Hip Fracture and Stroke.” Journal of the American Medical Association 277 (5): 396-404.

Kramer, A.M., P. W. Shaughnessy, M. K. Bauman, and K. S. Crisler, 1990. “Assessing and Assuring the Quality of Home Care: A Conceptual Framework.” The Milbank Quarterly 68 (3): 413-442.

Lewin-VHI (Manard, B., K. Bieg, R. Cameron, et al.), 1995. Subacute Care: Policy Synthesis and Market Area Analysis. Submitted to Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation. [Full Report]

Neu, C.R. and S. C. Harrison, 1989. “Medicare Patients and Postacute Care.” RAND Report R-3780-MN. Santa Monica, California: The RAND Corporation.

Steiner, A. And C.R. Neu, 1993. “Monitoring the Changes in Use of Medicare Posthospital Services.” RAND Report MR-153-HCFA. Santa Monica, California: The RAND Corporation.

Prospective Payment Assessment Commission (ProPAC), 1996. Medicare and the American Health Care System: Report to the Congress. Washington, D.C.

NOTES

1. Exeptions to this general rule were DRG 15 (n = 1,477) for which almost 1,500 episodes were recorded, DRG 214 (n = 883) and DRG 215 (n=337) because they are an alternative to widely studied hip procedures, and DRG 483 (n=386) because it is likely to involve ventilator care and is a “transfer” DRG.

[Top of Page]

SELECTING THE TARGETED CONDITIONS

Andrew Kramer, Theresa Eilertsen, Danielle Holthaus, Marie Johnson
University of Colorado Health Sciences Center

Barbara Gage
The Urban Institute

July 22, 1998

This report was prepared under contract #HHS-100-97-0010 between the U.S. Department of Health and Human Services (HHS), Office of Disability, Aging and Long-Term Care Policy (DALTCP) and The Urban Institute. For additional information about the study, you may visit the DALTCP home page at http://aspe.hhs.gov/daltcp/home.htm or contact the ASPE Project Officer, Jennie Harvell, at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, SW, Washington, DC 20201. Her e-mail address is: Jennie.Harvell@hhs.gov.

Conclusions and opinions in this paper are solely those of the authors and do not necessarily reflect the views of the Urban Institute, University of Colorado or ASPE.

I. INTRODUCTION

A. Purpose of the Project

In this project, we will develop a set of quality of care measures for Medicare post-acute care (PAC) services provided by SNFs, HHAs, rehabilitation hospitals or units (referred to as rehabilitation hospitals) and long-term hospitals. To the extent that ambulatory care is provided for these conditions during the post-acute period, services such as outpatient physician care will also be considered part of PAC. The measures will be designed for evaluating the quality of PAC services in both fee-for-service and HMO settings.

The project will focus on four clinical conditions for which PAC is frequently provided. For each of the four targeted conditions, outcome measures reflecting quality of PAC will be identified. For each outcome measure, the necessary data items, the data sources, approaches to applying the measures across episodes of care, and risk factors will be specified. In addition, we will develop methods to stratify patients within a condition for the purpose of comparing similar patients across settings, and also approaches to testing the validity and reliability of the outcome measures that are proposed.

B. Framework for Selecting Tracer Conditions

The possible clinical conditions for further study were narrowed down to eight based on policy considerations (Liu, Gage, and Kramer, 1998). The selection criteria to choose eight conditions included: 1) conditions for which PAC is provided in different PAC settings; 2) high volume conditions for PAC either as a proportion of hospital discharges or in terms of the number of PAC admissions; 3) Medicare utilization criteria such as readmission rates to the hospital, conditions where more than one PAC setting is used in an episode, or where prospective payment systems for PAC providers are likely to have their greatest impact; and 4) DRGs that include a homogeneous group of clinical conditions from the perspective of outcome measurement.

Using Medicare claims analyses of hospitalizations occurring in the thirty days prior to PAC episodes, eight DRGs were chosen meeting these eligibility criteria. These included: (1) hip fracture/replacement, (2) stroke, (3) pneumonia, (4) congestive heart failure, (5) acute myocardial infarction, (6) chronic obstructive pulmonary disease, (7) back and neck procedures, and (8) ventilator-dependent patients. In this report, we discuss the framework for further evaluating these eight conditions, and then review the conditions using this framework. Finally, we recommend four conditions for inclusion in the project based on this framework.

The framework for selecting four conditions from these eight emphasizes representativeness of the conditions as a group as much as the characteristics of the individual conditions. We want the conditions included in the project to represent different types of illnesses (chronic versus acute); different PAC settings; different PAC services; and different domains of outcome measures (e.g., functional status, utilization, quality of life). Recognizing that all eight of these conditions meet the criteria of policy relevance, the criteria applied in this paper relate more to clinical issues.

We evaluated each of the conditions based on the literature and further analysis of Medicare claims data. We were interested in the PAC settings in which patients with each condition are treated and the types of PAC services that patients receive in these settings (e.g., physical therapy, IV care). We also classified the conditions in terms of whether PAC is provided following a single acute event or an exacerbation of an underlying chronic disease. With respect to outcomes, we were interested in the types of outcome measures used for the different conditions (e.g., functional status, mortality, quality of life, symptoms). Finally, we were interested in whether there was evidence that outcomes for the condition differed between in PAC settings, to the extent that such information was available.

The results of our literature review and claims analysis are synthesized in two tables. However, a more extensive presentation of selected literature for each condition is provided in the Appendix. The literature included in this paper is intentionally not exhaustive -- our goal was to describe PAC services and the types of outcomes measured. However, after selecting the four tracer conditions, a more comprehensive literature summary will be provided in a subsequent paper with more detail on specific outcome measures.

II. METHODS

A. Literature Search

We conducted a systematic review of published studies describing outcomes in skilled nursing facilities, rehabilitation hospitals, home health agencies and long term hospital settings for each of the eight conditions. Because published literature evaluating outcomes in PAC was extremely limited for six of the eight conditions, we broadened our search to include studies describing outcomes in acute care hospitals and outpatient programs using home health or rehabilitation services that may be in the continuum of PAC. We searched the MEDLINE database for articles published in English between January 1990 and March 1998. We supplemented this literature with papers published in the 1980s when limited assessment studies were available. We used each of the eight conditions as a title word or key word combined with the following title and key words: aftercare, PAC, hospital discharge, skilled nursing facility, rehabilitation hospitals, rehabilitation, home health care, and outcomes. To identify additional articles, we searched the reference lists of the selected articles.

B. Definition of the Criteria

Our criteria for selecting targeted conditions included the following service-related criteria: variation in PAC settings, and variation in type of services provided. PAC settings that we included are rehabilitation hospitals (RH), skilled nursing facilities (SNF), home health agencies (HHA), and long term care hospitals (LTH). Of secondary interest were outpatient-based programs and specialized hospital units (SHU). The latter include extended hospital stays in acute hospitals in a specialized program for which PAC may substitute. Type of services referred to the specific services provided within the PAC setting. From claims and the literature, we were only able to obtain meaningful data corresponding to: physical therapy, occupational therapy, respiratory therapy, speech therapy, skilled nursing, and IV care.

One criterion for selecting targeted conditions pertained to the domains of outcome measures used for assessing quality of care. We collapsed individual measures into domains (general categories of measures) that represented similar constructs. However, this taxonomy of measures is preliminary and will be refined in the subsequent paper and throughout the project. Currently, we have identified the following 11 outcome domains: function, mortality, return to community, psychological, physiologic, symptom resolution, quality of life, patient satisfaction, utilization, weaning from mechanical ventilation, and disease-specific outcomes (endurance, re-infarct rate, and motor assessments).

Another criterion related to the chronicity of the condition. This referred to whether the condition occurs acutely without the prior active diagnosis, or as an acute exacerbation of a chronic disease, or as an acute episode that may become a chronic problem. The final criterion referred to whether there was evidence that differences in PAC settings and services actually impact outcomes for the condition. While information on outcomes of PAC was not available for all conditions, we wanted to include conditions for which available evidence suggests that outcomes are influenced by the PAC services. If PAC services do not alter outcomes significantly for a condition, the condition will not be a useful tracer for comparing treatment options.

C. Analysis of Medicare Claims Data

Service utilization was determined using HCFA’s 1995 5% sample Standard Analytical Files for Inpatient, SNF and HHA Medicare claims. Only the Medicare Part A files were available for this analysis. Patients were selected from these files if all the following conditions were met:

• age >65,
• enrolled in Medicare Part A and Part B,
• not ESRD patient,
• not HMO enrollee, and
• received PAC in a RH, SNF, or HHA following an acute episode (data were not available for LTH).

The majority of these patients’ admissions occurred in 1995, though there were some 1994 admissions. The claims records indicate what services were provided during the post-acute stay; we compiled statistics for physical therapy, occupational therapy, speech therapy, respiratory therapy, skilled nursing, and IV care. Because skilled nursing is not a separate cost center for SNFs and RHs, we could not determine the percent of patients in these settings receiving skilled nursing.

It is important to note that the file contains claims for services billed under Medicare Part A only; services provided through Part B are not included. This affects SNFs more than RHs or HHAs because some therapy services are billed under Part B for SNFs. This may somewhat underestimate the percent of patients receiving therapy services in SNFs. We have some concern that the claims data are not 100% accurate, as facilities are ultimately reimbursed based on year-end cost report data rather than interim claims data. Because these claims are not used for payment, all services may be somewhat underreported. However, we believe that any error in reporting services is likely to be approximately equivalent across facility types and across services, and thus should not introduce large biases in any particular direction.

D. Application of Criteria

The evaluation of the conditions involved applying the above criteria, synthesizing the information, and considering different combinations of conditions. To apply our criteria, we rated each condition based on setting, chronicity, services, and outcomes. The purpose was to identify similarities and differences among conditions such that four conditions could be chosen that would represent a broad spectrum of PAC. If two or three conditions were found to have similar profiles, only one of them was included in the final set of four conditions.

For each of the eight conditions, the predominant settings in which PAC occurs was determined, based on Medicare billing data. For most conditions, this was two settings (e.g. PAC for pneumonia is received largely in skilled nursing facilities and home health agencies).

Using Medicare claims data, the eight conditions were ranked based on the percentage of patients with that condition receiving each of the following five services: physical therapy, occupational therapy, speech therapy, respiratory therapy and the use of intravenous medications. We did not rank facilities on provision of any skilled nursing care because we only had HHA data for this service. Because the percentage of patients receiving services were very similar for some conditions, the conditions were ranked in quartiles from highest to lowest percentage of patients receiving each service. When the same percentage of patients received a service for more than two conditions, they were all assigned to the same quartile.

Each condition was labeled as either acute or chronic based on the predominant clinical presentation of such conditions. In the case of ventilator-associated conditions, both acute and chronic conditions are included.

Outcome measurement criteria were applied by identifying the outcome domains with some supporting literature and for which validated measures exist. The predominant outcome domains studied for each condition was determined from the literature review. We also identified conditions where outcomes are known to be sensitive to PAC setting, when such information was available.

For each of the four criteria, conditions with a similar profile were assigned to the same group (A, B, C, etc.). A final overall grouping was then determined based on similarities in the individual criteria groups. We then considered various combinations of the conditions to select the final four conditions.

III. RESULTS

A. Service Utilization

The proportion of patients receiving six services in different types of PAC facilities is presented in Table 1. The data presented indicate the percent of patients who received any amount of each service during their PAC stay. The PAC stay was defined by the period after hospital discharge until death, hospitalization or no PAC services for at least 30 days. For example, 83% of hip fracture patients in rehabilitation hospitals (RHs) received physical therapy during their stay, and 46% of pneumonia patients in skilled nursing facilities (SNFs) received physical therapy during their stay. The percentages listed for RH, SNF, and HHA include patients who received PAC only in the specified setting during an episode that involved PAC. Data for patients admitted to more than one type of care (for example, RH followed by SNF, or SNF followed by home health) are combined under >1 PAC.

For some conditions, very few patients (less than 1% of discharges) were admitted to RHs, making utilization statistics unreliable. The RH row is therefore not presented for pneumonia, CHF, AMI, or COPD. In addition, skilled nursing is not discernable for patients in RHs and SNFs, all of whom presumably receive skilled nursing care. Therefore, skilled nursing figures are presented only for patients receiving home health care. However, the percent of patients receiving IV care (a type of skilled care) is listed separately for all provider types. Service intensity figures in parentheses for hip fracture and stroke patients were obtained from previous national research conducted by Kramer, et al., 1997. No other studies for other conditions provided intensity figures.

For all eight conditions, it is apparent that the predominant services provided in PAC facilities are therapies, particularly physical and occupational therapy. That there is relatively little IV care provided for conditions such as pneumonia suggests that most patients with these conditions who are discharged to PAC may have completed the acute phase of their illness. Recognizing that the claims may underestimate the percent of patients receiving IV care as well as other services, the important point is that relatively more therapy services were provided for these patients. Thus, the focus of PAC services even for most patients with medical conditions appears to be functional recovery designed to help patients return home, rather than management of extremely ill, medically complex patients.

Across conditions, there is variation in the amount of each therapy provided. For example, the condition with the most patients receiving physical therapy (hip fracture) is also the condition with the fewest patients receiving respiratory therapy. This variation in service mix will enable us to divide the eight conditions into groups with similar patterns of service utilization.

B. Outcomes

The availability of literature on outcomes for the eight conditions varied substantially by condition. With respect to PAC outcomes, only hip fracture and stroke have been studied in some depth. Congestive heart failure and chronic obstructive pulmonary disease have been studied to a lesser extent after hospitalization; while pneumonia, back and neck procedures, and ventilator care have been studied largely with respect to the acute care outcomes. Acute MI has been studied in outpatient treatment and cardiac rehabilitation occurring in the home, but not generally involving home health nurses. The availability of extant research is summarized in the second column of Table 2a. The specific studies that were most relevant to outcomes of these conditions in PAC settings are summarized in the Appendix by condition.

From the studies reviewed, we identified outcomes and classified them into the domains shown in Table 2a. While this is only a preliminary taxonomy of outcomes that will be refined both in the subsequent paper to the clinical panel and clinical panel reviews, it provides a simple way of classifying the various types of outcome measures that we found in the literature. The relative use of measures from the different domains across conditions is rated from +++, denoting that an outcome measure was used in almost all the studies reviewed, to a blank in the table representing that no outcome measure from that domain was found in the literature for that condition. The types of measures that we included in each domain are listed in Table 2b. However, for comparative purposes at this stage of the study we are focusing on outcome domains and not specific measures.

Function was the most widely studied outcome domain for the eight conditions. However, different outcome measures were used depending upon the condition and study. While our intent is not to review the specific measures in this paper, illustrative measures that we identified from the literature are listed in Table 2b. For example, ambulation and walking distance were emphasized for hip fracture, measures of neurological functioning was emphasized for stroke, the New York Heart Association (NYHA) functional classes were used for congestive heart failure, exercise capacity was used for acute MI and independence from mechanical ventilation was used for ventilator-dependent patients. Activity of daily living (ADL) and instrumental activity of daily living (IADL) measures were used for multiple conditions, with substantial variation in the scales used (e.g., RUGS III, FIM).

Mortality was the next most frequent outcome domain for these post-acute conditions, in part because so many of the outcome studies to date examined the hospital stay where keeping patients alive is often the focus of care.

For some conditions, disease-specific measures of physiology and symptom-specific measures were often found. Various quality of life measures were used in studying outcomes for many conditions, but not in the context of PAC. While re-hospitalization is a PAC issue, many studies examined re-hospitalization for chronic conditions over extended periods of time. In the Other category, we included condition-specific outcome measures that were not applied across conditions.

C. Profiling the Eight Conditions

Table 3 classifies each condition based on the settings in which PAC was frequently received following the acute stay, the chronicity of the disease, major outcome domains which are supported in the published literature, and the percentage of patients receiving selected services.

Using the information from Table 3, we grouped the eight conditions into categories according to their similarities with respect to each of the four criteria. Table 4 depicts these categories. Under the Setting criterion, we assigned hip fracture and stroke to the same category because both are primarily treated in rehabilitation hospitals, skilled nursing facilities and home health agencies, and secondarily in long term hospitals. Back and neck procedures differed only slightly because treatment occurs largely in home health agencies but secondarily in rehabilitation hospitals and skilled nursing facilities. Pneumonia, CHF, acute MI, and COPD were all in the same group for setting. Ventilator was in its own category due to the large proportion of patients on ventilators who are treated in long term care hospitals, making this a unique condition.

All conditions were either acute or chronic, with the exception of ventilator, which can be an acute and/or chronic condition. This does not refer to disease severity, rather the difference between a rapid onset event or illness without prior existence of the specific diagnosis (acute) versus an ongoing disease that becomes worse requiring immediate attention (chronic). We recognize that those diseases that we classified as acute often occur in patients at higher risk due to underlying disease processes (e.g., osteoporosis and balance impairment for hip fracture), but the person does not have the diagnosis of hip fracture until the actual fracture occurs, which is not the case for CHF or COPD. These classifications were carried over directly from Table 3 to Table 4.

The conditions were divided into five groups for outcome domains, with hip fracture and stroke grouped together -- having similar outcome domains except that stroke includes additional outcome domains. CHF, Acute MI, and COPD were grouped together because they all included five similar domains (function, mortality, physiology, quality of life, and utilization) and selected other domains (e.g., return to community, symptom resolution). The remaining three conditions had unique combinations of outcome domains based on the literature, and therefore were assigned individual categories.

For the Services criterion, we matched conditions to one another by the overall patterns of quartile designations across five services. We did not include the skilled nursing service for selecting conditions because data for skilled nursing was only available for home health patients. Hip fracture and back and neck procedures are grouped together because a high proportion of patients with these conditions received physical therapy and occupational therapy, and a low proportion of these patients receive respiratory therapy and IV services. Stroke is also similar, but differs slightly by the high proportion of stroke patients who receive speech therapy as well. Ventilator constitutes its own category due to the high proportion of ventilator patients who receive all five services. Among the remaining four conditions, pneumonia and COPD have similar patterns across services -- low proportions of patients receiving physical therapy and occupational therapy, but a high proportion of patients receiving respiratory therapy -- and are therefore grouped together. Finally, low proportions of CHF and AMI patients receive physical therapy, occupational therapy, and respiratory therapy, creating a separate category.

Overall, Table 4 illustrates several points. First, hip fracture, and stroke patients receive similar types of PAC services, both in rehabilitation hospitals and skilled nursing facilities. Both conditions are acute and studied outcome domains are largely functional, return to community living and mortality, with some additions for stroke (i.e., psychological and quality of life).

Second, pneumonia, CHF, COPD and AMI patients all receive PAC largely in skilled nursing facilities and home health agencies, but vary mostly in chronicity of the conditions, the types of outcome domains studied, and in use of respiratory therapy. COPD and CHF are both chronic conditions and share many outcome domains. Pneumonia and AMI are both acute. COPD and pneumonia use more respiratory therapy.

Third, back and neck patients receive similar services to hip fracture and stroke patients, and many similar outcome domains are studied. While the chronicity of illness is different, setting and services are not.

Finally, ventilator patients are different from all other conditions in setting of care, services received (particularly respiratory therapy), combination of both chronic and acute illnesses, and outcome measures related to ventilator weaning.

IV. DISCUSSION/CONCLUSION

While there is no scientific method for selecting tracer conditions, we have tried to follow an unbiased and systematic approach to evaluate the eight conditions. Our overall objective is to identify conditions that represent the range of patients, settings, services, and outcomes that constitute PAC. We used explicit criteria, collecting data pertinent to each criterion.

The final step is to examine the profiles of the different conditions to choose a combination of conditions that exhibits the greatest variation in these criteria. This final step involves making selections based on the profiles in Table 4, that summarizes the results from all the previous tables. While all eight of these conditions would be interesting to study for one reason or another, we are restricted to four due to project limitations. Thus, our proposed four tracer conditions are contained below, including the rationale.

Either hip fracture or stroke should be included in the study because they are the major conditions treated in both rehabilitation hospitals and skilled nursing facilities. However, we see little benefit in including both of these conditions because they are both acute, require similar services, and can be studied with similar outcome measures. If our sole criterion were number of PAC admissions or PAC cost, then both of these conditions would probably be chosen for this project. However, for developing quality measures, we are interested in a range of conditions so that we address different clinical situations, different settings, different services, and different outcomes during the developmental process.

We recommend choosing stroke over hip fracture for two reasons. First, with the frequency of speech therapy as a service and the multiple types of neurologic and mobility impairments that result from stroke, to measure outcomes will require a wider range of measures for stroke than hip fracture. This is supported in part by the additional outcome domains that have been studied for stroke. Second, stroke outcomes have been shown repeatedly to be influenced by PAC setting (Kramer et al., 1997, Kane et al., 1996). For hip fracture patients, evidence is mixed at best that the PAC setting influences outcomes (Koval et al., 1998, Kramer et al., 1997, Kane et al., 1996). If this is the case, then hip fracture will not be a very sensitive tracer for comparing care in HMO and fee-for-service settings or examining the impacts of prospective payment on quality.

The four medical conditions -- pneumonia, CHF, Acute MI, and COPD -- exhibit similarities and differences. While they are all treated frequently in both SNFs and home health agencies, two of them are acute (pneumonia and acute MI) and the other two are chronic. PAC services are similar for the two cardiac conditions (acute MI and CHF), while the pulmonary conditions have more respiratory therapy (pneumonia and COPD). By selecting two of these four conditions, we can choose one that is an acute medical problem (either pneumonia or acute MI) and one chronic medical problem (CHF or COPD). We can also choose one that is a cardiac problem and one that is a pulmonary problem. Thus, to maximize variability we should select either pneumonia and CHF, or acute MI and COPD.

In view of the extremely high PAC utilization and cost of both CHF and pneumonia, we recommend including these two in the study. By choosing COPD instead of pneumonia, we would be selecting a second chronic medical condition -- an acute condition such as pneumonia introduces different outcome measurement issues (e.g., measurement time points). Furthermore, pneumonia has more patients treated in SNFs than COPD (Liu et al., 1998), while both have substantial numbers of home health patients. Because there is significant clinical overlap among these conditions, however, we will not be entirely eliminating any of the four. Pneumonia patients receiving PAC often have underlying COPD, while CHF patients receiving PAC often have underlying coronary artery disease and/or a previous acute MI.

Both back and neck procedures and ventilator-dependent patients have unique profiles. However, patients with back and neck procedures use fairly similar settings and services to hip fracture and stroke patients. While it is a chronic condition, often the treatment for which someone is receiving PAC will be in response to an acute repair of a back or neck injury. While an interesting condition, we do not find it as compelling for studying rehabilitation as either hip fracture or stroke. The latter have so many more PAC cases.

Ventilator care offers the addition of long-term hospitals as a major setting, a unique mix of services with very intense respiratory care, unique outcomes, and patients who may in some cases be acute while in others will go on to become chronically dependent on the ventilator. This type of care can be extremely expensive and the choice among post-acute settings will be driven substantially by new payment policies. Controversy exists regarding the use of long-term hospitals, and subacute care reimbursement for treating very complex patients such as those who are dependent upon ventilators.

Thus, the four conditions we recommend for development of quality measures include stroke, pneumonia, congestive heart failure, and ventilator-dependent patients. As a group, they represent both acute and chronic conditions, cover all PAC settings, and the range of post-acute services in these settings. Most importantly for this project, these conditions will require a range of quality measures to assess quality of care across PAC settings.

REFERENCES

Atlas SJ, Deyo RA, Keller RB, Chapin AM, Patrick DL, Long JM, Singer DE. The Maine lumbar spine study, part II: 1-year outcome of surgical and nonsurgical management of sciatia. Spine 1996;21(5):1777-86.

Atlas SJ, Deyo RA, Keller RB, Chapin AM, Patrick DL, Long JM, Singer DE. The Maine lumbar spine study, part III: 1-year outcomes of surgical and nonsurgical management of lumbar spinal stenosis. Spine 1996;21(15):1787-95.

Ciol MA, Deyo RA, Howell E, Kreif S. An assessment of surgery for spinal stenosis: time trends, geographic variations, complications, and reoperations. JAGS 1996;44:285-90.

DeBusk RF, Haskell WL, Miller NH, Berra K, Taylor CB, Burger WE III, Lew H. Medically directed at home rehabilitation soon after clinically uncomplicated acute myocardial infarction: A new model for patient care. American Journal of Cardiology 1985;55(4):251-7.

Dennis LI, Blue CL, Stahl SM, Benge ME, Shaw CJ. The relationship between hospital readmissions of Medicare beneficiaries with chronic illness and home care nursing interventions. Home Healthcare Nurse 1996;14:303-8.

Deyo RA, Ciol MA, Cherkin DC, Loeser JD, Bigos SJ. Lumbar spinal fusion: a cohort of complications, reoperations, and resource use in the Medicare population. Spine 1993;18(11):1463-70.

Deyo RA, Cherkin DC, Loeser JD, Bigos SJ, Ciol MA. Morbidity and mortality in sssociation with operations on the lumbar spine. Journal of Bone & Joint Surgery 1992;74A(4):536-43.

Dressendorfer RH, Franklin, Cameron JL, Trahan KJ, Gordon S, Timmis GC. Exercise training frequency in early post-infarction cardiac rehabilitation. Journal of Cardiopulmonary Rehabilitation 1995;15(4):269-76.

Dubach P, Myers J, Dziekan G, Goebbels U, Reinhart W, Vogt P, Ratti R, Muller P, Miettunen R, Buser P. Effect of exercise training on myocardial remodeling in patients with reduced left ventricular function after myocardial infarction. Circulation 1997. 95(8):2060-7.

Elpern EH, Lurson R, Douglass P, Rosen RL, Bone RC. Long-term outcomes for elderly survivors of prolonged ventilator assistance. Chest 1989;96(5):1120-4.

Fine MJ, Smith MA, Carson CA, Mutha SS, Sankey SS, Weissfeld LA, Kapoor WN. Prognosis and outcomes of patients with community-acquired pneumonia. JAMA 1996;274:134-141.

Gracey DR, Naessens JM, Viggiano RW, Koenig GE, Silverstein MD, Hubmayr RD. Outcome of patients cared for in a ventilator-dependent unit in a general hospital. Chest 1995;107(2): 494-9.

Gracey DR, Viggiano RW, Naessens JM, Hubmayr RD, Silverstein, MD, Koenig, GE. Outcomes of patients admitted to a chronic ventilator-dependent unit in an acute-care hospital. Mayo Clinic Proceedings 1992;67(2):131-6.

Gulanick, M. Is phase 2 cardiac rehabilitation necessary for early recovery of patients with cardiac disease? A randomized, controlled study. Heart & Lung 1991;20(1):9-15.

Hamalainen H, Luurila OJ, Kallio V, Knuts L-R. Reduction in sudden deaths and coronary mortality in myocardial infarction patients after rehabilitation. European Heart Journal 1995;16(12):1839-44.

Hanumanthus S, Butler J, Chonsky D, Davis S, Wilson JR. Effect of a heart failure program on hospitalization frequency and exercise tolerance. Circulation 1997;96:2842-48.

Harjai KJ, Mehra MR, Ventura HO, Lapeyre YM, Miugo JP, Stapleton DD, Smart FW. Home inotropic therapy in advanced heart failure. Chest 1997;112:1298-1303.

Hood SA, Weigl K. Lumbar spinal stenosis: surgical intervention for the older person. Israel Journal of Medical Sciences 1983;19:169-72.

Kane RL, Chen Q, Blewett LA, Sangl J. Do rehabilitative nursing homes improve the outcomes of care. JAGS 1996;44:545-554.

Kane RL. A study of post-acute care. 1994. Institute for Health Services Research.

Katz JN, Lipson SJ, Lew RA, Grobler LJ, Weinstein JN, Brick GW, Fossel AH, Lean MH. Lumbar laminectomy alone or with instrumented or noninstrumented arthrodesis in degenerative lumbar spinal stenosis. Spine 1997;22(10):1123-31.

Keith RA, Wilson DB, Gutierrez P. Acute and subacute rehabilitation for stroke: a comparison. Arch Phys Med & Rehab 1995;76:495-500.

Kennie DC, Reid J, Richardson IR, Kiamari AA, Kelt C. Effectiveness of geriatric rehabilitative care after fractures of the proximal femur in elderly women. BMJ 1998; 297:1083.

Ketelaars CAJ, Abu-Saad HH, Schlosser MAG, Mostert R, Wouters EFM. Long-term outcome of pulmonary rehabilitation in patients with COPD. Chest 1997;112(2):363-9.

Kopacz MA, Moriarty-Wright R. Multidisciplinary approach for the patient on a home ventilator. Heart & Lung 1984;13(3):255-62.

Kornowski R, Zeeli D, Averbuch M, Finkelstein A, Schwartz D, Moshkovitz M, Weinrieb B, Hershkovitz R, Eyal D, Miller M. Intensive home-care surveillance prevents hospitalization and improves morbidity rates among elderly patients with severe congestive heart failure. Am Heart J 1995;129:762-6.

Koval KJ, Aharonoff GB, Su ET, Zuckerman JD. Effect of acute inpatient rehabilitation on outcome after fracture of the femoral neck or intertrochanteric fracture. Journal of Bone and Joint Surgery 1998;80-A(3):357-64.

Kramer AM, Steiner JF, Schlenker RE, Eilertsen TB, Hrincevich CA, Tropea DA, Ahmad LA, Eckoff DG. Outcomes and costs after hip fracture and stroke. JAMA 1997;277:5:396-404.

Leitch JW, Newling RP, Basta M, Inder K, Dear K, Fletcher PJ. Randomized trial of a hospital-based exercise training program after acute myocardial infarction: Cardiac autonomic effects. Journal of the American College of Cardiology 1997;29(6):1263-8.

Make B, Gilmartin M, Brody JS, Snider GL. Rehabilitation of ventilator-dependent subjects with lung diseases: the concept and the initial experience. Chest 1984;86(3):358-65.

McLean RF, McIntosh JD, Kung GY, Leung DMW, Byrick RJ. Outcome of respiratory intensive care for the elderly. Critical Care Medicine 1985;13(8):625-9.

Meehan TP, Fine MJ, Krumholz HM, Scinto JD, Galushi DH, Mockalis JT, Weber GF, Petullo MK, Hauch PM, Fine JM. Quality of care process, and outcomes in elderly patients with pneumonia. JAMA 1997;278:2080-2084.

Metlay JP, Fine MJ, Schulz R, Marrie TJ, Coley CM, Kapoor WN, Singer DE. Measuring symptomatic and functional recovery in patients with community acquired pneumonia. JGIM 1997;12:423-430.

Miller NH, Haskell WL, Berra K, DeBusk RF. Home versus group exercise training for increasing functional capacity after myocardial infarction. Circulation 1984;70(4):645-9.

Muir J, Girault C, Cardinaud J, Polu J. French Cooperative Study Group. Survival and long-term follow-up of tracheostomized patients with COPD treated by home mechanical ventilation: a multicenter French study. Chest 1994;106(1):201-9.

Munin MC, Rudy TE, Glynn NW, Crossett LS, Rubash HE. Early inpatient rehabilitation after elective hip and knee arthroplasty. JAMA 1998;279(11):847-52.

Niederman MS, Henerson Clemente P, Fein AM, Feinsilver SH, Robinson DA, Ilowite JS, Bernstein MG. Benefits of a multidisciplinary pulmonary rehabilitation program. Chest 1991;99(4):798-804.

North RB, Kidd DH, Campbell JN, Long DM. Dorsal root ganglionectomy for failed back surgery syndrome: A 5-year Follow-Up Study. J Neurosurg 1991;74:236-42.

North RB, Campbell JN, James CS, Conover-Walker MK, Wong H, Piantadosi S, Rybock JD, Long DM. Failed back surgery syndrome: 5 year follow-up in 102 patients undergoing repeated operation. Neurosurgery 1991;28:685-91.

Oldridge N, Guyatt G, Jones N, Crowe J, Singer J, Feeney D, McKelvie R, Runions J, Steiner D, Torrance G. Effects on quality of life with comprehensive rehabilitation after acute myocardial infarction. American Journal of Cardiology 1991;67(13):1084-9.

Ortqvist A, Hedlund J, Gilner L, Jalonen E, Kallings I, Leinonen M, Kalin M. Aetiology, outcome and prognostic factors in community-acquired pneumonia requiring hospitalization. Eur Resp J 1990;3:1105-1113.

Pehrsson K, Olofson J, Larsson S, Sullivan M. Quality of life of patients treated by home mechanical ventilation due to restrictive ventilatory disorders. Respiratory Medicine 1994;88:21-26.

Reardon J, Awad E, Normandin E, Vale F, Clark B, ZuWallack RL. The effect of comprehensive outpatient pulmonary rehabilitation on dyspnea. Chest 1994;195(4):1046-52.

Retchin SM, Brown, RS, Yeh SJ, Chu D, Moreno L.Outcomes of stroke patients in Medicare fee for service and managed care. JAMA 1997;278(2):119-24.

Rich MW, Beekham V, Wittenberg C, Leven CL, Freedland KE, Carney RM. A multidisciplinary intervention to prevent readmission of elderly patients with congestive heart failure. N Engl J Med 1995;333(18):1213-4.

Rich MW, Vinson JM, Sperry JC, Shah AS, Spinner LR, Chung MK, Davila-Roman V. Prevention of readmission in elderly patients with congestive heart failure. JGIM 1993;8:585-590.

Smith EB, Hanigan WC. Surgical results and complications in elderly patients with benign lesions of the spinal canal. JAGS 1992;40(9):867-70.

Stapleton DC, Kaplan SJ, Manard, B. Ventilator dependent unit demonstration: outcome evaluation and assessment of post acute care. 1996. Report prepared for Office of Research and Demonstrations, Health Care Financing Administration.

Steiner T, Mendoza G, DeGeorgia M, Schellinger P, Holle R, Hache W. Prognosis of stroke patients requiring mechanical ventilation in a neurological critical care unit. Stroke 1997;28(4):711-15.

Strijbos JH, Postma DS, van Atena R, Gimeno F, Koeter GH. A comparison between an outpatient hospital-based pulmonary rehabilitation program and a home-care pulmonary rehabilitation program in patients with COPD. Chest 1996;109(2):366-72.

Stineman MG, Fiedler RC, Granger CV, Maislin G. Functional task benchmarks for stroke rehabilitation. Archives of Physical Medicine and Rehabilitation 1998;79(5):497-504.

Swinburne A, Fedullo AJ, Bixby K, Lee DK, Wahl GW. Respiratory failure in the elderly: analysis of outcome after treatment with mechanical ventilation. Arch Inter Med 1993;153:1657-62.

Trzcieniecka-Green A, Steptoe A. Stress management in cardiac patients: a preliminary study of the predictors of improvement in quality of life. Journal of Psychosomatic Research 1994;38(4):267-80.

Vale F, Reardon JZ, ZuWallack RL. The long-term benefits of outpatient pulmonary rehabilitation on exercise endurance and quality of life. Chest 1993. 103(1):42-5.

Votto J, Bowen J, Scalise P, Wollschlager C, ZuWallack R. Short-stay comprehensive inpatient pulmonary rehabilitation for advanced chronic obstructive pulmonary disease. Archives of Physical Medicine & Rehabilitation 1996;77(11):1115-8.

Wijkstra PJ, van der Mark Th. W, Kraan J, van Atena R, Koeter GH, Postma DS. Effects of home rehabilitation on physical performance in patients with chronic obstructive pulmonary disease (COPD). European Respiratory Journal 1996;9(1):104-10.

Wijkstra PJ, Van Atena R, Kraan J, Otten V, Postma DS, Koeter GH, Postma DS. Quality of life in patients with chronic obstructive pulmonary disease improves after rehabilitation at home. European Respiratory Journal 1994;7:269-73.

Williams DN. Reducing costs and hospital stay for pneumonia with intravenous cefotaxime treatment: results with a computerized ambulatory drug delivery system. Amer J Med 1994;97:50-55.

Young, JB, Forster A. The Bradford community stroke trial: results at 6 months. BMJ 1992;304:1085-9.

Zuckerman JD, Sakales SR, Fabian DR, Frankel VH. Hip fractures in geriatric patients: results of an interdisciplinary hospital care program. Clinical Orthopeadics 1992; 274:213.

TECHNICAL ADVISORY GROUP RECOMMENDATIONS

At the conclusion of this phase of the study, we distributed this report and the report entitled "Selecting and Evaluating Eight Targeted Conditions" to our Technical Advisory Group (TAG) for review. Feedback from members of the TAG revealed a high level of concern over the inclusion of ventilator-dependent patients as one of our four selected conditions. Although many TAG members agreed that ventilator-dependent patients represent a unique and interesting condition for study, they expressed concerns about whether such a group is representative of post-acute care patients. The low prevalence of ventilator-dependent patients in post-acute care, as well as a lack of homogeneity among these patients, were among the specific concerns expressed. Furthermore, the large majority of ventilator-dependent patients are treated in long-term care hospitals, while relatively few are treated in more traditional types of post-acute settings.

In light of these concerns, we decided to include back and neck medical and surgical conditions instead of ventilator dependency as our fourth condition for study. Back and neck patients are similar to ventilator-dependent patients in that they both have somewhat unique profiles across treatment settings, chronicity, outcome domains, and services required (see Table 4). Back and neck patients are also relatively prevalent in the post-acute care population, are treated in multiple settings, and represent a population that has not been well studied.

APPENDIX

[Top of Page]