Printer Friendly

Factors associated with hospitalization in a sample of chronic hemodialysis patients.

This study examines the hospitalization experience of a sample of chronic hemodialysis patients, using primary data sources. There were multiple causes of hospitalization over the six-month tracking period, with stays extending from 1 to 87 days. Patients were more likely to be hospitalized if they had a negative hepatitis antigen, lower functional status scores, lower phosphate and protein levels, repeated access procedures, other cardiovascular conditions, arthritis, psychiatric disorders, ischemic peripheral vascular disease, lung disease, or larger households. Hospitalization for access-related problems was associated with arthritis, previous access procedures, and blood pressure levels. Sociodemographic and treatment characteristics did not have a significant influence on the risk of hospitalization. Improved management in these clinical areas may improve the quality of life of chronic hemodialysis patients and reduce the high level of expenditures associated with delivering inpatient services to this segment of the Medicare population.

The Medicare end-stage renal disease (ESRD) program has increased enormously in size, cost, and political importance since its beginning in July 1973 (Blagg, Bovbjerg, and FitzSimmons 1989). Cost increases have resulted from general price inflation, a tenfold increase in the size of the patient population, and the substantial percentage of enrolled patients who are being dialyzed in the most expensive care setting - the dialysis center or facility. Smith et al. (1983) claim that center dialysis has become the preferred mode of therapy. They attribute this trend to an increasingly aged dialysis patient population, the presence of systemic disease in a growing number of dialysis patients, and a decline in the socioeconomic and educational status of ESRD patients.

Additional factors might also be contributing to the large percentage of patients dialyzing in facilities rather than at home. These may include patient and family preferences (architectural, emotional, or other deterrents to home treatment) and provider preferences (higher reimbursement/payment levels for treating in-center dialysis patients than for supervising home patients). The combination of incentives and case-mix changes has resulted in substantial variations in percentages of home dialysis patients across types of facilities (hospital based versus freestanding), systems of care (private sector versus Department of Veteran Affairs), and states (Washington, DC versus Washington state). Such case-mix and home percentage differences help explain variations in outcomes and costs across modalities and settings.

The ESRD program has received much publicity and scrutiny from federal policymakers, primarily because of its rapid increases in spending due to the unexpected growth in number of patients served. In 1983, ESRD beneficiaries accounted for 3.7 percent of total Medicare expenditures (Parts A and B), while accounting for only 0.25 percent of all Medicare beneficiaries (Eggers 1984). Medicare pays for hospital inpatient care and physician services as well as for the costs of dialysis itself and transplantation. A person qualifying for Medicare under the renal disease provision is eligible for the full range of benefits available under the health insurance program, not just those services relating to renal care. Inpatient reimbursement is a sizable and growing portion of total ESRD expenditures. It increased from 26 percent in 1980 to 33 percent in 1983 (Health Care Financing Administration 1987). The increasing average age of hemodialysis patients and the presence of more debilitating conditions may serve to further increase the proportional amount being spent on hospital episodes for ESRD hemodialysis patients.

The escalating ESRD budget led to attempts to control these expenditures through the use of a composite payment methodology and actual reductions in the amount of reimbursement per treatment. These cost control measure brought about changes in the hemodialysis treatment process, primarily the practices of dialyzer reuse and shortened treatment sessions. Concerns about the possible effect of such changes on the well-being of patients have been expressed by policymakers, medical professionals, and hemodialysis patient interest groups.

Quality of care is an important consideration in the treatment of Medicare patients and other patient groups. Patient outcomes are now frequently used alternatives to structure and process measures of quality assessment (Brook, Avery, Greenfield, et al. 1977; Wennberg, Bunker, and Barnes 1980; Williamson, Hudson, and Nevins 1982). Hospitalization is a possible outcome measure appropriate to the ESRD patient experience (Eggers 1982). Outcome analysis is difficult, however, because of the elusive relationship between outcomes and potential explanatory variables such as patient characteristics and treatment approach (Health care Financing Administration 1987). Several studies have documented substantial case-mix differences among patients on hemodialysis, but little work has been done thus far on the direct effects of specific case-mix measures on the hospitalization experience of these patients (Health Care Financing Administration 1987).

Information about the outcome measures of hospital episodes and hospital days of care is also deficient because of the lack of detailed information about morbidity and resource use within the ESRD program. This situation may be remedied soon by the availability of the U.S. Renal Data System (USRDS). The government and other researchers will then be able to examine outcomes associated with ESRD treatment as well as responses of patients to new treatment modalities and approaches (Blagg, Bovbjerg, and FitzSimmons 1989).

This article presents the results of an analysis of hospitalization in a sample of chronic hemodialysis patients, using primary data collected from medical records and unit flow chart reviews. Its findings provide the basis for further study of morbidity outcomes in a more representative sample of hemodialysis patients. It is an important avenue of research to pursue because of its cost reduction and quality enhancement implications. Identification of specific risk factors for hospitalization might enable better targeting of medical interventions or treatment processes useful for reducing the incidence of hospitalization in this patient population. Such results might also improve the quality of life of these chronic patients. The study also provides some preliminary evidence of the effects of treatment variations on patient well-being.

The Process of Hemodialysis

Hemodialysis is a classic "halfway technology" for the treatment of end-stage renal disease. It prolongs survival, but does not cure the disease that caused the kidney failure. Chronic hemodialysis also allows the evolution of disease that coexist with chronic renal failure, such as complications of diabetes mellitus and cerebrovascular diseases. The hemodialysis treatment process itself creates a new clinical pathology which includes uremic cardiomyopathy, renal osteodystrophy, and dialysis dementia (Funck-Brentano 1980). The hemodialysis process itself is costly, while the associated morbidity adds to the costs.

In the course of their years of dialysis therapy, patients may require hospitalization for treatment of infection, cardiovascular instability, hypertension, myocardial infarction, gastrointestinal bleeding, or a vast array of other conditions (Gutman and Amara 1978). In addition, revision or replacement of the vascular access for hemodialysis treatments is required from time to time.

The literature provides a substantial number of published research papers concerned with mortality in hemodialysis patients, but few quantitative studies on morbidity have been done on this population. The several potential uses for the results of studies done in this area of analysis included estimations of probabilities that individual patients will develop major morbid events; estimation of the economic burden of illness created by maintaining patients on chronic hemodialysis; and control of quality across and within individual dialysis centers, as differences in morbidity rates may be found to relate to patient characteristics, dialysis technique, staffing mix and levels, and other factors.

Hospitalization in the

Chronic Hemodialysis

Population

Hospitalization is a relatively frequent event in the chronic hemodialysis population. In 1979, there were 15,966 days of hospital care per 1,000 enrollees in the program, with an average length of stay of 9.2 days (Eggers 1984). Eggers found no relationship between patient sex, race, or diagnosis and the reimbursement level per hospital episode.

Evans, Manninen, Garrison, et al. (1984) reported that ESRD patients experienced on average 11.6 inpatient days and 1.96 admissions in 1981. Case-mix measures were not found to be related to hospital utilization except for diabetes mellitus and number of comorbid conditions, which generated higher hospital expenses per patient. Pollak et al. (1986) reported an annual average of 14.3 days for all hemodialysis patients and 18.9 days for diabetics. Delano (1989), in a study of anemia in hemodialysis patients, noted an annual hospitalization experience of 14.7 days.

The National Kidney Dialysis and Kidney Transplantation study (Health Care Financing Administration 1987) reported that two-thirds of its sample were hospitalized at least once during the year. Patients being dialyzed in centers were hospitalized on average 2.4 times per year with an average of 6.6 days per episode and a mean total of 13.8 days per year.

Given the high cost of hospital care, any progress made in identifying and reducing the risk associated with hospitalization in this population could result in substantial cost savings, especially as the ESRD population grows and becomes increasingly elderly and debilitated.

Factors Associated with

Hemodialysis Outcomes

Psychological problems have been identified with the risk of hospitalization in chronic hemodialysis patients. Numan, Barklind, and Lubin (1981) reported that depression is significantly associated with frequency of hospitalization as well as death. Noncompliance may also lead to hospitalization (Wolcott et al. 1986). Compliance may be better in older patients (Czaczkes and DeNour 1978), but cardiac instability, frequently observed in elderly patients, would tend to increase the hospitalization rate in the older groups. Kutner and Cardenas (1981) did not find greater compliance among older patients, while Procci (1978) reported that unemployed single males living away from their families are the most noncompliant group among hemodialysis patients, and may

therefore experience a greater number of hospitalizations.

The elderly do experience a greater number of medical problems per patient, with lower blood pressure and serum phosphate levels and more episodes of gastrointestinal bleeding, which may lead to more frequent hospitalizations (Chester, Rakowski, Argy, et al. 1979). Lowrie et al. (1981) found that higher blood urea nitrogen (BUN) levels and shorter dialysis times were associated with hospitalization, perhaps indicating dietary abuses. This study excluded hospitalization for vascular access problems. The authors also pointed out that patients who have restricted protein intake and are undernourished may also have a greater risk of medical complications.

The recent National Kidney Dialysis and Kidney Transplantation study (Health Care Financing Administration 1987) found that having a primary diagnoses of diabetes and an increasing number of comorbid conditions increased hospital admissions and hospital days, while an increasing level of education reduced admissions and days. Patient age and duration of dialysis therapy had negative, but nonsignificant relationships with these outcomes.

Functional status assessments have been frequently used to characterize the effect of disease on patients, and should be considered in an analysis of outcomes in ESRD patients. Loss of function is related to the cumulative physical, physiological, and psychological effects of the disease process on the patient (Mor et al. 1984). Stewart, Greenfield, Hays, et al. (1989) point out that measures of functioning and well-being have been shown to predict health expenditures and death (Manning, Newhouse, and Ware 1982; Bergner, Hallstrom, Bergner, et al. 1985). Functional status has been used frequently as an outcome measure in comparisons of the different ESRD treatment modalities.

Gutman, Stead, and Robinson (1981) used the Karnofsky Performance Status scale (KPS) as a measure of current physical performance or activity level in a study of rehabilitation of hemodialysis patients dialyzing in 18 different facilities. They concluded that a larger proportion of dialysis patients than previously suspected may be severely debilitated.

Evans, Manninen, Garrison, et al. (1985) also used the Karnofsky Scale, employing it as an objective indicator of quality of life in a comparison of home dialysis, in-center dialysis, continuous ambulatory peritoneal dialysis (CAPD), and transplant patients. The study found that transplant recipients had the least functional impairment and that in-center hemodialysis patients were the most limited in physical activity. Hart and Evans (1987) used the Sickness Impact Profile (SIP) to measure functional statuts in ESRD patients. They found that in-center hemodialysis and CAPD patients were less functional than home patients. They also discovered that less educated patients; diabetics; and patients with angina, other cardiovascular problems, respiratory diseases, neurologic disorders, gastrointestinal problems, and musculoskeletal disorders were significantly less functional than were comparatively well educated patients and patients without these co-morbid conditions.

The National Kidney Dialysis and Kidney Transplantation study (Health Care Financing Administration 1987) reported on functional limitations of ESRD patients using both the Karnofsky Scale and the SIP. The Karnofsky scores were related to treatment modality, revealing that transplant patients experience less functional impairment than home, CAPD, and in-center patients. Large differences in functional status as measured by the SIP were also found to exist across treatment modality groups, but it was concluded that the differences may be partly a function of case-mix differences among the modality groups. Patients receiving the different ESRD treatment modalities differed significantly in terms of age, sex, race, education, primary diagnosis, and comorbidity, which in turn could influence the patient's functional abilities. No studies were found that used a measure of functional status itself to predict the risk of hospitalization or to explain resource use variation in chronic hemodialysis patients. There were also no studies located that examined functional status variations among center hemodialysis patients or across hemodialysis units.

In a recent study of synthetic erythropoietin, Delano (1989) used the Karnofsky score as a measure of rehabilitation in ESRD patients. Delano noted that the anemia of chronic renal failure leads to a loss of the stamina needed to perform ordinary daily activities. Administration of the drug r-Hu-EPO (recombinant Human Erythropoietin) to an experimental group of patients led to an increase in the group's mean Karnofsky score, from 76 before treatment to 86.6 after administration of the drug. If functional status is determined to be a significant risk factor for hospitalization of the chronic hemodialysis patient, the administration of synthetic erythropoietin may lead to substantial cost savings from reduced morbidity.

Much of the previous research in the hemodialysis population has focused on mortality and case-mix variations by type of unit (Held, Pauly, and Diamond 1987; Degoulet, Legrain, Reach, et al. 1982; Volmer, Wahl, and Blagg 1983). Eggers (1982) reported that hospital-based units had lower survival rates, higher hospital rates of admission, greater percentages of patients with diabetes mellitus, and more newly initiated patients than freestanding units, suggesting a more complex case mix. Five-year survival rates have been found to be significantly associated with the primary causes of renal failure (Eggers, Connerton, and McMullan 1984). Those with polycystic kidney disease had the best survival rates, while those with diabetic nephropathy had the worst five-year survival rates.

Hutchinson, Thomas, and MacGibbon (1982) found that patient age, the presence and duration of diabetes mellitus, and left-sided heart failure as evidence by pulmonary edema were significantly associated with the probability of death in hemodialysis patients. The findings suggested that the relative risk of death increased exponentially with each additional ten years of age. Garcia-Garcia, Deddens, DiachiardiRey, et al. (1985), in a similar study, found that patient age, lack of private health insurance, a primary diagnosis of diabetes mellitus, and existence of comorbidities significantly increased the risk of death for hemodialysis patients. Sex and race were not found to be significant risk factors. These authors concluded that socioeconomic factors rather than race played a significant role in determining survival of patients with end-stage renal disease.

Plough and his colleagues (Plough, Salem, Shwartz, et al. 1984; Plough, Shwartz, and Sale 1984; Plough, Shwartz, Salem, et al. 1985) examined case-mix differences between independent and hospital-based units. They developed a range of severity groupings (clusters) based on primary and secondary diagnosis, race, and age of the patient, and measured severity by estimating the probability of death in the first year of treatment and over the course of treatment. Plough concluded that since hospital-based units had a significantly higher percentage of patients within the highest severity groupings with the highest probability of death, they may be taking care of a greater portion of more severe cases. Plough also reported a higher mortality rate within each severity grouping for patients dialyzing in hospital-based units. They viewed this as another indicator of a more severe case mix in those units.

Radecki, Mendenhall, Nissenson, et al. (1988) examined differences in case mix and therapy for patients on different types of renal dialysis therapy and in different treatment settings. Their comparison of hospital-based versus independent units suggested that hospital-based units may be caring for a medically sicker patient population. However, they found no substantial differences in number or nature of comorbid conditions, and based their conclusion on the findings that hospital-based patients were somewhat older, more likely to be hospitalized, and "requiring" more physician time, testing, and therapy. Radecki admitted that many of the variables used in the comparison were either subjective or done at the discretion of the physician, and that the physicians in the hospital-based units were providing a more resource-intensive treatment regimen than physicians practicing in independent units.

Plough's study was criticized by representatives of the proprietary dialysis industry (Lowrie and Hampers 1984), who claimed that the severity groups were actually only surrogates for mortality rate and therefore demonstrated that hospital-based units have higher death rates than independent units, controlling for patient severity. Lowrie and Hampers also pointed out that a correlation does not necessarily exist between mortality and treatment costs, and that more important factors for determining cost variations would be items such as number of access procedures and compliance with treatment restrictions. Jones, Nickerson, Kilpatrick, et al. (1986) demonstrated that case-mix differences as measured by duration of dialysis therapy, presence of comorbid conditions, primary renal diagnosis, and race do exist among units, but they were unable to relate these case-mix differences to resource consumption variations.

Even fewer studies have examined the relationship between patient clinicial characteristics and costs of treatment, primarily because cost is very difficult to determine. Eggers (1984) used treatment reimbursement as a proxy for cost, and found that patient age, sex, race, and primary diagnosis were not associated with per capita rates of reimbursement. Females and whites did have slightly higher reimbursement levels, while a slight inverse relationship was found between reimbursement and age.

Patient outcomes might also vary according to specific facility and treatment characteristics. For example. Schlesinger, Cleary, and Blumenthal (1989) suggested that physicians may treat patients under different regimens in order to increase a facility's profits. However, studies comparing outcomes of care in for-profit and not-for-profit settings have revealed no consistent differences in outcomes related to ownership. Schlesinger's analysis found that ownership-based incentives did affect clinical decision making, but the authors were unable to determine whether observed differences in treatment were related to health outcomes.

The flat-or fixed-rate method of reimbursement for ESRD dialysis treatments also establishes incentives for reducing the costs per dialysis treatment. Lack of patient-specific data, however, makes it difficult to evaluate the quality of care being delivered and to monitor patient responses to shorter dialysis times and reuse of dialyzers and tubes. No standard treatment protocol exists, and thus far few if any studies have examined the relationship between treatment pattern styles and patient outcomes in the ESRD population. It is possible that morbidity and mortality vary with reuse practices, staff-patient ratios, and percentage of home patients, but at the point little knowledge exists on this issue (Maxwell and Sapolsky 1987).

The controversy surrounding the selection of patients for dialysis and transplantation has served to focus attention on patient outcomes. Survival is a reliable indicator of patient outcome, but satisfactory benefits to patients need to be judged on more than survival alone. The National Kidney Dialysis and Kidney Transplantation study, mentioned earlier, made a strong case for the further systematic studies to interrelate treatment, costs, and outcomes. The Omnibus Budget Reconciliation Act of 1987 called for the Institute of Medicine (IOM) to examine the quality of care provided to end-stage renal disease patients, as measured by clinical indicators, functional status, patient satisfaction, and the effect of reimbursement on quality of life. The results of the research described in this article provide some additional evidence to consider as these important issues are further explored.

METHODS

The study was conducted on a sample of 527 chronic hemodialysis patients who dialyzed seven different facilities in northern and central Florida from December 1985 to June 1986. The dialysis units were both hospital based and freestanding (independent), proprietary and not-for-profit, and they were located in both urban and rural areas of the state. In order to capture all available data on the maximum number of patients, all chronic hemodialysis patients dialyzing more than one month in each of the units were enrolled in the study.

Detailed patient case-mix information was collected on each patient by unit hemodialysis nurses, who were reimbursed for their efforts. Training sessions were conducted to ensure accurate, consistent completion of the data forms, and to instruct the staff participants on the assessment of patient functional ability using the Karnofsky Performance Status Scale (KPS). This was the only objective measure used in the study, requiring staff nurse assessment of the appropriate score. The remaining data were abstracted from unit flow sheets and patient treatment records.

The KPS is an 11-point rating scale which ranges from normal functioning (100) to dead (0). It is a complex measure of a patient's activity level, and is viewed as a concise approach to classifying and ranking patients whose individual clinical status is far more complicated (Mor et al. 1984). Use of the KPS has been criticized because of its subjective nature and variability in scoring among observers (Hutchinson, Boyd, Feinstein, et al., 1979). However, the KPS was found to achieve at least a moderate level of interrater agreement in more recent applications (Yates, Chalmers, and McKegney 1980). Interrater reliability coefficients of over 0.97 were achieved in a national hospice study (Mor et al. 1984). The hospice researchers also used analysis of variance (ANOVA) with KPS as the independent variable and survival in days as the dependent variable, and showed a direct, monotonic relationship between KPS and survival at the .001 probability level. They concluded that although KPS is a subjective rating, it is based on objective factors and is a reliable measure of functional status.

Major categories of information collected were patient sociodemographic characteristics; primary renal diagnosis; current level of functional ability; specific cardiovascular, neurologic, gastrointestinal/metabolic, and other comorbid conditions; compliance behaviors; specific risk factors and elements; and physiologic values. In addition, staff kept track of unusual events occurring during the hemodialysis treatment process (bleeding episodes, dressing changes); number, cause, and length of hospitalization episodes; cause and occurrence of death; and treatment characteristics, including medications and transfusions. These measures were generated from extensive literature review, recommendations of a national advisory committee, and interviews with local practicing nephrologists and hemodialysis nurses.

Clinical, sociodemographic, and treatment variables were collected at an initial time period, and outcome measures were collected six months later. Efforts to complete a second case-mix, treatment, and outcomes measurement were unsuccessful, due to high nursing staff turnover and stress within most of the hemodialysis units. The analysis was conducted on the sample of patients for whom both the initial casemix measurement and six-month outcomes follow-up were completed.

VARIABLES

Sociodemographic

Age. Older patients were expected to have increased risk of hospitalization because of greater physiologic instability.

Race. Nonwhites were expected to have greater risk of hospitalization because of greater severity of the underlying renal disease process.

Sex. Males were expected to be at greater risk of hospitalization because of more noncompliance behaviors.

Education. More highly educated patients were expected to be at lower risk of hospitalization because of greater compliance with treatment restrictions.

Household size, Medicaid status, private health insurance, housing status. Larger household, being a Medicaid recipient and a lacking private health insurance, and living with others or renting were included as proxies for socioeconomic status, and were expected to be associated with greater risk of hospitalization.

Marital status. Married patients were expected to be at lower risk of hospitalization because of greater compliance and presence of a support person in the environment.

Functional Status

Patients at higher levels of the Karnofsky Performance Status Scale were expected to be at lower risk of hospitalization than patients at the lower functional status levels.

Comorbid Conditions

Cardiovascular. A greater number of cardiovascular conditions were expected to be associated with greater risk of hospitalization.

Neurologic. A greater number of neurologic conditions were expected to be associated with a higher risk of hospitalization.

Gastrointestinal. A greater number of gastrointestinal and metabolic conditions were expected to be associated with a higher risk of hospitalization.

Other. A greater number of other comorbid conditions were expected to be associated with a higher risk of hospitalization.

Specific comorbid conditions. The presence of each was expected to place a patient at greater risk of hospitalization, as compared to a patient without that comorbid condition.

Risk Factors and Elements

Age and Treatment History. Increasing age and duration of treatment were expected to have a negative association with hospitalization, due to higher compliance and increased stability on treatment.

Duration of Home Hemodialysis, CAPD, CCPD, and Diabetes, and History of Alternative Treatment Modality. Longer treatment and diabetic histories, and experience of an alternative treatment modality, were expected to be associated with an increased risk of hospitalization.

Positive Hepatitis Antibody or Antigen, Non-A Non-B Hepatitis. These were expected to have a positive association with hospitalization.

Residual Urine Formation. The capacity to excrete water was expected to reduce the risk of hospitalization due to the need for fewer restrictions on fluid and for less intense dialysis treatments.

Symptomatic Cardiac Arrhythmias. A positive finding was expected to increase the risk of hospitalization due to cardiac instability.

Accesses. An increasing number of vascular accesses was expected to increase the risk of hospitalization.

Primary Renal Disease

Diabetes Mellitus, Hypertensive Nephropathy, Collagen Disease, and Unknown Diagnosis. These diagnoses were all expected to be associated with an increased risk of hospitalization.

Polycystic Kidney Disease. This diagnosis was expected to be associated with a reduced risk of hospitalization.

Behavioral Factors

No-Shows and Early Sign-offs. An increasing number of no-shows and early sign-offs was expected to be associated with increased risk of hospitalization.

Physiological Factors

Potassium, Sodium, Phosphate, Glucose, BUN, Creatinine, Serum Glutamic-Oxaloacetic or -Pyruvic Transaminase (SGOT/SGPT). Higher values were expected to be associated with an increased risk of hospitalization.

Albumin, Protein, Hematocrit. Higher values were expected to be associated with reduced risk of hospitalization.

Blood Pressures. Higher values pretreatment and lower values post-treatment were expected to be associated with an increased risk of hospitalization.

Interdialytic Weight Gain. Higher values were expected to be associated with an increased risk of hospitalization.

Hemodialysis Intensity Events

Intensity events. A greater number of unusual occurrences during hemodialysis treatments was expected to increase the risk of hospitalization.

Treatment Variables

Central Dialysate Delivery System, Reuse of Dialyzer, Shorter Treatment Times. These factors were expected to increase the risk of hospitalization.

Special Dialysate Solutions, Special Artificial Kidneys, Individual Dialysate Delivery Systems, Longer Dialysis Times. These were expected to reduce the risk of hospitalization.

Results

DESCRIPTION OF SAMPLE

The sample consisted of 527 patients entered into the study from seven different hemodialysis units. A total of 240 patients were from independent or freestanding units while 287 patients were from hospital-based units. Sixty-seven percent of the patients were black, while 30.9 percent were white and the rest were primarily Asian or Hispanic. Males accounted for 41.5 percent of the cases. The patients demonstrated a relatively low educational level, with 59.2 percent having less than a high school education. A substantial number of patients were married (43.4 percent), but the majority of patients were either single, divorced, or widowed. Half of the patients owned their own home, while the other half either rented or lived with other people, including residence in a nursing home. A substantial portion of patients were receiving Medicaid benefits (40.4 percent), while a slightly lower portion of patients reported that they had additional private health insurance. A very small percentage of the patients worked, and only 58 percent of them were judged to be able to carry on normal activity or to do active work. This may be related, at least in part, to the fact that 31.8 percent of the sample were over age 65. These results can be seen in Table 1. [TABULAR DATA OMITTED]

FREQUENCY AND CAUSES OF HOSPITALIZATION

As shown by other studies, hospitalization is a frequent event in his patient population. The 527 patients in our sample experienced 500 hospital episodes over the six-month follow-up period. Fifty percent of the patients required at least one hospitalization, with an average 1.9 episodes per patient hospitalized. The total number of days of hospitalization per patient averaged 15.2 days, with a range extending from 1 to 87 days.

Over half of the episodes involved a single discharge diagnosis. The greatest portion of hospital admissions was for access-related problems. Over a quarter of the hospital episodes were for declotting or replacement of the access device, for treatment of infection or rupture of the access, or for placement of a Tenchoff catheter in preparation for peritoneal dialysis treatment.

The second most frequent reason for admission to the hospital was for circulatory or cardiovascular-related problems, which is consistent with the high incidence of cardiovascular comorbid conditions in these patients. The hospitalizations involved episodes of congestive heart failure, hypertension management, digitoxin toxicity, pericarditis, and myocardial infarction. Gastrointestinal or metabolic disorders were the third most frequent cause of hospitalization, with diagnoses including gastrointestinal bleeding, pancreatitis, ulcers, and peritonitis.

The remaining hospital episodes were distributed fairly evenly across neurologic disorders, pulmonary and orthopedic problems, electrolyte imbalances, other major diagnoses (sepsis, removal of kidney graft, cancer therapy) and other minor diagnoses (biopsies, anemia, fever, and cataract repair).

Slightly more than half of the admissions were for five days or less. Almost a quarter were from six to ten days in length. The remaining episodes ranged from 11 to 87 days.

Bivariate Relationships

Sociodemographic characteristics. Only a few sociodemographic variables achieved statistical significance when compared across hospitalized and nonhospitalized patients, using Student t-tests or Mann-Whitney U tests as appropriate, and chi-square analysis. Hospitalized patients came from larger households, and were less likely to be homeowners. Marital status approached statistical significance, with hospitalized patients less likely to be married. Race, sex, education, Medicaid recipient, and private health insurance showed no significant differences across groups.

Risk Factors/Elements. Hospitalized patients had significantly lower functional ability and were found more frequently in the functional status categories below 70. They had required more accesses over their dialysis history. Hospitalized patients were also less likely to have a positive hepatitis antigen test result. There were no differences in patient age, duration of diabetes, duration of dialysis treatment, hepatitis antibodies, non-A, non-B hepatitis, residual urine formation, nor in presence of a symptomatic cardiac arrhythmia.

Primary Renal Disease. No significant association was found between any primary renal disease and hospitalization.

Comorbid Conditions. Similar to results of other studies, hospitalized patients had significantly more total comorbid conditions, plus a higher number of cardiovascular, neurologic, metabolic/gastrointestinal, and other comorbid conditions. The specific comorbid conditions that were found significantly more often in the hospitalized group were ischemic peripheral vascular disease, other cardiovascular conditions, psychiatric disorders, arthritis, and systemic infection. Comorbid conditions that approached statistical significance were carotid artery obstruction, insulin-dependent diabetes mellitus, and lung disease.

Physiologic Variables. Serum protein was lower in the hospitalized group, and interdialytic weight gain was also surprisingly lower. Serum phosphate and creatinine, and predialysis diastolic blood pressures, approached statistical significance.

Behavioral Factors. Neither behavioral measure varied significantly across the hospitalized and nonhospitalized groups.

Treatment Measures. Hospitalized patients did not experience a greater number of unusual events during dialysis treatments; neither did they come from one type of dialysis facility more often than the other. The hospitalized patients did receive more blood transfusions. There were no significant differences in hours of treatment per week, routine or dialysis medications, psychiatric referrals, dialyzer reuse, or special dialysate administration.

Multivariate Analysis

Variables identified in the previous section as having a significant relationship with hospitalization, or one approaching significance, were entered into multivariate models using stepwise logistic regression. The SAS logistic regression procedures was used to identify which variables had significant associations with the risk of hospitalization, holding other factors constant. Given that hospitalizations can be quite diverse in nature, several different models were estimated. The first identified risk factors associated with being hospitalized at least once during the six-month follow-up period for any reason. Then hospitalizations were divided into two groups: one group contained all of the single hospitalizations for an access-related problem; the second group consisted of non-caused single hospitalizations or patients with multiple hospitalizations for any reason in the follow-up period. It was believed that the second group would have a higher severity of illness and require a more intensive hospitalization. In addition to these basic models, comorbidities were entered into the models in two different ways. First, they were entered as group counts; second, the specific individual comorbid conditions were used. Three models will be discussed.

Risk of Hospitalization using Grouped Comorbid Conditions. As seen in Table 2, this model predicted correctly 63.2 percent of the cases, with a 68.5 percent specificity and 57.3 percent sensitivity. Seven variables had a significant association with being hospitalized. Higher functional status scores were associated with a lower risk of hospitalization. Higher serum protein and phosphate levels were also associated with a reduced risk, as was having a positive hepatitis antigen test result. Increasing household size, increasing number of other comorbid conditions, and increasing number of accesses all were associated with a greater risk of hospitalization. The list of other comorbid conditions included lung disease, arthritis, infection, and bone disease.
Table 2: Logistic Regression Results for Hospitalization
Using Grouped Comorbid Conditions
 Standard Odds
Variables Beta Error [X.sup.2] P Ratio
Inyercept -.5528 1.543 0.12 .7753 -
Karnofsky -.0219 .006 14.67 .0001 1.02
Household size .2378 .063 14.21 .0002 1.27
Number other .2394 .089 7.23 .0072 1.27
 comorbids
Poditive -1.7057 .593 8.28 .0040 5.51
 hepatitis
 antigen
Phosphate -.0952 .045 4.39 .0361 1.10
Protein -.3071 .154 3.98 .0461 1.36
Accesses .1479 .076 3.80 .0513 1.16
Model chi-squre = 53.37 Specificity = 68.5%
 Sensitivity = 57.3% R = .251
 P = .0 Correct = 63.2%


The magnitude of influence exerted by these variables can be determined through calculation of the variable's odds ratio. The odds ratio indicates the relative risk of hospitalization associated with a given measure. For a dichotomous measure, the odds ratio represents the extent to which the chance of hospitalization is greater for a hemodialysis patient with the trait compared to a patient without the trait. For example, a hemodialysis patient with a positive hepatitis antigen has a 5.51:1 greater chance of not being hospitalized than a patient with a negative hepatitis antigen. For a continuous variable, the odds ratio represents the increased risk of hospitalization for each additional unit. A patient with a history of three accesses has a 1.16:1 greater chance of being hospitalized than a patient who has had two accesses. The odds ratios can be used to rank-order the model variables in terms of decreasing influence on risk of hospitalization. In the first model the ordering would be positive hepatitis antigen, serum protein level, number of other comorbid conditions, household size, number of accesses, serum phosphate level, and functional status.

The model for a single access-related hospitalization, seen in Table 3, predicted correctly 90.7 percent of the cases. However, its specificity (99.6 percent) was vastly superior to its sensitivity (3.6 percent). Only four variables has a statistically significant relationship with an access-related hospital episode. Having arthritis increased the risk, as did an increasing number of accesses in the past and higher postdialysis diastolic blood pressures. On the other hand, an increasing predialysis diastolic blood pressure reduced the risk.
Table 3: Logistic Regression Results for Hospitalization for
Access-Related Cause, Using Specific Comorbid Conditions
 Standard Odds
Variable Beta Error [X.sup.2] p Ratio
Intercept 1.089 1.993 0.30 .5848 -
Arthritis: -1.206 0.422 8.15 .0043 3.34
 absent
Number accesses 0.296 0.116 6.51 .0117 1.35
Predialysis -0.107 0.039 7.86 .0056 1.11
 diastolic
 blood pressure
Postdialysis 0.084 0.045 3.54 .0601 1.09
 diastolic
 blood pressure
Model chi-square = 27.59 Sensitivity = 3.6%
 Correct = 90.7% r = .302
 p = .0 Specificity = 99.6%


Looking at the odds ratios, a patient with arthritis had a 3.34:1 chance of requiring and access-related hospitalization as compared to a patient without arthritis.

The logistic regression model for multiple or non-access-related admissions, seen in Table 4, predicted correctly 68.3 percent of the cases, with a specificity of 77.6 percent and a sensitivity of 56.8 percent. This model had 12 predictor variables with a significance level p < .100. Five specific comorbid conditions had a positive association with the risk of multiple hospitalizations: other cardiovascular conditions, arthritis, psychiatric disorders, lung disease, and ischemic peripheral vascular disease. An increasing functional status, positive hepatitis antigen, higher serum phosphate and protein levels, and higher interdialytic weight gains all were associated with reduced risk of hospitalizations. Increasing household size and number of accesses were associated with an increased risk of multiple hospitalizations. Positive hepatitis antigen results, other cardiovascular conditions, psychiatric disorders, lung disease, and ischemic peripheral vascular disease had the greatest impact on risk, as reflected by the calculated odds ratios.
Table 4: Logistic Regression Results for Hospitalization for
Non-Access = Caused or Multiple Episodes Using Specific
Comorbid Conditions
 Standard Odds
Variable Beta Error [X.sup.2] p Ratio
Intercept 5.953 2.334 6.50 .0108 -
Karnofsky -0.024 .006 14.02 .0002 1.02
Other cardio- -1.367 .474 8.34 .0039 3.92
 vcular
conditions:
absent
Household size 0.297 .067 8.34 .0000 1.35
Positive hepatitis -2.019 .678 8.87 .0029 7.53
antigen
Psychiatric -0.976 .416 5.50 .0190 2.65
disorder:
absent
Phosphate -0.086 .045 3.64 .0565 1.10
Protein -0.315 .162 3.79 .0517 1.37
Lung disease: -0.616 .338 3.32 .0683 1.85
absent
Number accesses 0.150 .082 3.35 .0673 1.16
Interdialytic -0.111 .051 4.76 .0291 1.12
weight gain
Ischemic -0.565 .317 3.17 .0750 1.76
peripheral
vascular
disease:
absent
Model chi-square = 76.06 Specificity = 77.6%
 Sensitivity = 56.8% R = .298
 p = .01 Correct = 68.3%


Risk Estimation. Chronic hemodialysis patients experience a wide variety of comorbid, physiologic, and treatment-induced problems that contribute in varying extents to the probability of being hospitalized. The previous analyses identified specific factors that have a significant association with risk of being hospitalized. It would be useful to assess the overall risk of hospitalization given the presence or absence, or specific value, of each of the significant risk factors for any individual patient. The beta coefficients from the logistic regression models are used to estimate a risk score for individual patients and to calculate the probability of hospitalization. This score could be useful in targeting medical and nursing interventions to reduce, when possible, the risk of hospitalization. The probability of hospitalization is calculated by inserting values for each variable in the model, multiplying by the coefficient of that variable, summing up those values and calculating scores, then solving for p using the formula p = (exp Z/1 + exp Z).

Table 5 shows an example of the estimated probability of hospitalization using the model for multiple hospital episodes. A patient with none of the specific comorbid conditions, a good level of functioning, a two-access history, five pounds of interdialytic weight gain, and a three-member household has a .1275 risk of hospitalization. However, a patient with all five comorbid conditions, a lower functional status, and more abnormal physiologic values has a .9343 chance of being hospitalized - a high certainty in the near future.
Table 5: Estimation of Hospital Risk Using Multiple Hospital
Episodes Model
Variable Bi Xi Bi Xi Xi Bi Xi
Karnofsky -0.024 70 -1.680 40 -0.96
Other cardiovascular -1.367 1 -1.367 0 0
 conditions
Household 0.297 3 0.891 5 1.485
Positive hepatitis -2.019 0 0 0 -2.019
Arthritis -0.587 1 -0.587 0 0
Psychiatric -0.976 1 -0.976 0 0
Phosphate -0.086 6 -0.516 4 -0.344
Protein -0.315 7 -2.205 3 -0.945
Lung -0.616 1 -0.616 0 0
Access 0.150 2 0.300 1 0.150
Weight gain -0.111 5 -0.555 6 -0.666
Ischemic peripheral -0.565 1 -0.565 0 0
 vascular disease
Intercept 5.953 5.953 5.953
 p = .1275 p = .9343


Table 6 shows the risk estimation for a single, access-related hospitalization. A patient without arthritis, with a two-access history, and predialysis and postdialysis diastolic blood pressures of 70 and 60, respectively, has a .1218 chance of hospitalization. A patient with arthritis and twice as many accesses, but the same blood pressure values, has a .5441 chance of a single, access-related hospital episode.
Table 6. Estimation of Risk of Hospitalization for
Access-Related Problem
Variable Bi Xi Bi Xi Xi Bi Xi
Arthritis -1.206 1 -1.206 0 0
Accesses 0.296 2 0.592 4 1.184
Predialysis -0.107 70 -7.490 70 -7.490
blood pressure
Postdialysis 0.084 60 5.040 60 5.040
blood pressure
Intercept 1.089 1.089 1.089
 p = .1218 p = .5441


Discussion

Hospitalization is a frequently experienced event among patients receiving chronic hemodialysis treatments (Health Care Financing Administration 1987). These hospitalizations are related in part to the treatment process itself, especially the requirement for vascular access. Both internal and external access devices periodically require replacement or revision because of infections, clotting, aneurysms, or other problems. It is an important risk factor for ESRD outcomes, as pointed out by Lowrie and Hampers (1984). Not only are problems with access maintenance a major cause of hospitalization for these patients, but repeated access procedures and revisions are also associated with hospitalizations for non-access-related problems.

Focusing attention on access maintenance might lead to reduced access problems and lower hospitalization rates for chronic hemodialysis patients. The analysis of access-related hospitalizations indicates that increased attention to two factors may lead to more positive outcomes for these patients. One aspect is improved blood pressure control. The ability to avoid abnormally low diastolic blood pressures between dialysis treatments, while removing a sufficient amount of fluid and adequately lowering diastolic blood pressures during treatment, may preclude the occurrence of access problems in some patients. However, the low blood pressures may also be related to patient age. Chester, Rakowski, Argy, et al. (1979) found in a small sample of patients that diastolic blood pressure declined with age, a decline that accelerated in the elderly. In fact, lower diastolic blood pressures, interdialytic weight gains, and serum phosphate levels are all found more frequently in older patients (Chester, Rakowski, Argy, et al. 1979), and may explain both the somewhat counterintuitive results that occurred in this study (higher phosphate and weight gains associated with reduced risk), and why age itself was not a significant risk factor for hospitalization.

Arthritis patients, who were found to be at greater risk of both access-related and non-access-related hospitalizations, may also require increased attention. These patients experience problems with access maintenance as well as additional problems, with, for instance, drug therapy and joint replacement procedures. Arthritic patients may require special consideration in selection of access sites, type of access, monitoring during and between treatments, physical therapy, and intensity of teaching and self-care programs. Improved blood pressure control and management of patients with arthritis may lead to reduced costs to the Medicare program through avoided hospitalizations, as well as an improved quality of life for these patients. The finding that access problems are also associated with multiple or non-access-related hospitalizations may indicate a lack of knowledge of care of access devices by non-dialysis hospital personnel such as staff nurses and phlebotomists, and appropriate in-service sessions may need to be conducted.

Similar to studies of other outcomes, this study found cardiovascular problems to be a major reason for hospitalization, although the specific cardiovascular conditions differed from those implicated in other studies (Kutner and Cardenas 1981; Lindner et al. 1974; Lowrie et al. 1974; Hutchinson, Thomas and MacGibbon 1982; Plough and Salem 1982). This may have been due to our more extensive list of comorbid conditions, our short follow-up period, or characteristics of this particular sample. Patients with other cardiovascular conditions (heart block, pacemakers, cardiomyopathy, and valvular heart disease) and ischemic peripheral vascular disease (IPVD) were at significantly greater risk of imminent hospitalization. IPVD is frequently associated with advanced diabetes mellitus as well as advanced age, and may help explain why these two variable failed to achieve statistical significance in this analysis. Consistent with other studies, psychiatric disorders also increased the risk of hospitalization (Numan, Barklind, and Lubin 1981). The specific diagnoses range from anxiety to depression to paranoia. The results suggest that increased attention to preventing advanced cardiovascular disease and to managing psychiatric, arthritic, and heart problems may reduce the hospitalization risk associated with these comorbidities.

The results also suggest that improvements in two nonmedical areas, nutritional status and functional status/rehabilitation, might also lead to better outcomes for these patients. The need for more emphasis on rehabilitation of these patients has been mentioned in several studies (Gutman, Stead, and Robinson 1981; Evans, Manninen, Garrison, et al. 1984). Higher functional status levels might reduce the risk of hospitalization and improve the quality of life. Physical therapy, routine exercise and fitness programs, and administration of the r-Hu-EPO might elevate level of functioning and ultimately reduce the morbidity costs associated with hemodialysis patients. In addition, the results suggest that avoidance of protein depletion is also required, perhaps through extensive dietary counseling or nutritional supplements.

It is equally interesting to identify those variables that did not achieve a significant association with risk of hospitalization. None of the primary renal diseases predicted hospitalization, nor did patient age, race, education level, length of time of time on treatment, or directly measured compliance behaviors. This might have been due to the more extensive list of variables included in this data set. Age and diabetes mellitus may not be important predictors per se, but specific manifestation of age and diabetes mellitus, such as lower functional status, ischemic peripheral vascular disease, pacemaker implantation, arthritis, lower blood pressure and phosphate levels, and lower fluid gains between treatments, do appear to have an influence on this outcome and may be more appropriate risk measure than general categories of age and primary diagnosis.

This study found few variables previously to be related to death in chronic hemodialysis patients to be significantly associated with hospitalization. Patient age, heart failure, diabetes mellitus, no private health insurance, and treatment carried out in a hospital-based facility have all been associated with higher death rates, but none of these were significantly risk factors for hospitalization. Only "comorbid conditions," reported to be a risk factor for death and hospitalizations in previous studies, were also significant in this analysis of the outcome hospitalization.

A separate analysis of risk factors associated with death in this sample found some overlap of predictor variables as well as some differences across the different outcomes. Lower functional status, psychiatric problems, low serum albumin and protein levels, and increasing number of accesses were all associated with a higher risk of death as well as hospitalization. However, race (African-American) and more hours of treatment per week reduced the risk of death, while increasing age, increasing number of gastrointestinal comorbid conditions, and having active hepatitis increased the risk of death but did not significantly influence risk of hospitalization. It should also be noted that survival rates and hospital rates did not differ across the units in this analysis, although treatment pattern did vary substantially.

Moreover, none of the treatment variables examined in this study were found to have a significant association with risk of hospitalization. Reuse of dialyzers, centralized dialysate delivery systems, and fewer hours of treatment per week did not increase a patient's risk of hospitalization. It is possible that treatment variations might have influenced blood pressure or physiologic values, but this was not detected in the analysis. Cost-containment activities, such as reductions in use of dietitians, physical therapists, psychologists, and clinical nurse specialists, could be associated to some degree with the high number of low-functioning patients, the nutritional problems, and the psychiatric problems. In addition, reduced staffing levels in the unites and lower skill mix might be related to less intensive monitoring of patients during hemodialysis and to reduced emphasis on counseling, education, and training, leading to inadequate blood pressure control, inadequate consumption of needed nutrients, and access problems. These relationships are speculative and were not measured in this study, but should be explored in future research.

The results of this study may not be generalizable to the rest of the chronic hemodialysis population. The sample of patients differed substantially in sociodemographic characteristics and other measures from figures reported by HCFA. Of patients in the ESRD population, 23.2 percent are over age 65; the study sample contained 31:8 percent in this age group. The national sample is 56 percent male and 68 percent white; this group was 42 percent male and 31 percent white. The national sample has been on dialysis an average of 51.4 months, the study patients for 38 months. The study patients are less educated (most patients nationally have completed high school) and are less likely to be working. However, the percentage of diabetics is similar across groups, at about 23 percent.

In summary, hospitalization, a frequent outcome in the ESRD hemodialysis population, might be reduced in frequency through improved medical and nonmedical management of the patient. Specific interventions that might lead to avoided hospitalizations include administration of r-Hu-EPO; dietary counseling and nutritional supplements to avoid low protein and albumin values; avoidance of low diastolic blood pressure levels between schedule treatments; psychological logical counseling and psychiatric referrals; closer monitoring of pacemaker, arthritis, and IPVD patients; and improved phosphate and fluid management in relevant patients.

This study found no evidence that hemodialysis treatment practice differences have a negative influence on hospitalization risk. There was also no indication that type of dialysis unit (freestanding or hospital-based) made a difference for this outcome. The study results do suggest that clinical factors, especially functional status, number of accesses, and specific comorbid conditions, rather than sociodemographic or treatment factors, make a difference in the prediction of hospitalization of hemodialysis patients. It also suggests that explanatory or risk factors for outcomes in the ESRD population may vary by specific outcome being considered. Therefore, additional research needs to be done to analyze which factors are most important in explaining resource use or cost variations per patient, since it is highly likely that risk factors for death or hospitalization are not appropriate explanatory variables for cost differences across patients or units. The development of effective program policies demands an understanding of the specific causes of each of the outcomes of this expensive treatment.

REFERENCES

Bergner, L., A. P. Hallstrom, M. Bergner, M. S. Eisenberg, and L. A. Cobb.

"Health Status of Survisors of Cardiac Arrest and of Myocardial Infarction

Controls." American Journal of Public Health 75, no. 11 (1985):

1321-23. Blagg, C. R., R. R. Bovbjerg, and S. C. FitzSimmons. "Here Are (almost all)

the Data: Evaluation of the U.S. Renal Data System." American Journal of

Kidney Diseases 14, no. 5 (1989): 347-53. Brook, R. H., A. D. Avery, S. Greenfield, L. J. Harris, L. Tova, N. E.

Solomon, and J. R. Ware, Jr. "Assessing the Quality of Medical Care

Using Outcome Measures: An Overview of the Method." Medical Care

15, no. 9 supplement (1977): S1-S165. Chester, A. C., T. A. Rakowski, W. P. Argy, D. Giacalone, and G. E.

Schreiner. "Hemodialysis in the English and Ninth Decades of Life."

Archives of Internal Medicine 139, no. 9 (1979): 1001-1005. Czaczkes, J. W., an A. K. DeNour. Chronic Hemodialysis as a Way of Life. New

York: Brunner and Mazel, 1978. Degoulet, P., M. Legrain, I. Reach, F. Aime C. Devries, P. Rojas, and C.

Jacobs. "Mortality Risk Factos in Patients Treated by Chronic Hemodialysis."

Nephron 31, no. (1982): 103-10. Delano, B. G. "Improvements in Quality of Life Following Treatment with

r-Hu-EPO in Anemic Hemodialysis Patients." American Journal of Kidney

Diseases 14, no. 25 (1989): 14-18. Eggers, P. W. "Analysis of Indicators of Case-Mix Differences between Freestanding

Facility and Hospital-based Medicare ESRD Patients." Working

Paper no. OR38. Baltimore, MD: Health Care Financing

Administration, Office of Research and Demonstrations, 1982. * Trends in Medicare Reimbursements for End-Stage Renal Disease: 1974-1979.

Baltimore, MD: Health Care Financing Administration, 1984. Eggers, P.W., D.L. Manninen, L.P. Garrison, L.G. Hart, C.L. Blagg, R.A. Gutman, A.R. Hull and E.G. Lowrie. "The Quality of Life of Patients with End-Stage Renal Disease." New England Journal of Medicine 312, no. 9 (1985): 553-59. * A Comparative Assessment of the Quality of Life of End-Stage Renal Disease Patients on Four Treatment Modalities: Results from the National Kidney Dialysis and Kidney Transplantation Study. Seattle, WA: Batelle Human Affairs Research Centers, 1984. Funck-Brentano, J.L. "New Clinical Pathology Created by Chronic Dialysis."

Advances in Nephrology 9 (1980): 3-5. Garcia-Garcia, G., J.A. Deddens, R. Diachiardi-Rey, M.R. First, S.J. Samuels, K.S. Kant, and V.E. Pollak. "Results of Treatment in Patients with End-Stage Reanl Disease: A Multivariate Analysis of Risk Factors and Survival in 341 Successive Patients." American Journal of Kidney Disease V, no. 1 (1985): 10-18. Gutman, R. A., and A. H. Amara. "Outcome of Therapy for End-Stage

Uremia: An Informal Predictions of Survival Rate and Degree of Rehabilitation."

Pos-graduate Medicine 64, no.5 (1978): 183-94. Gutman, R. A., W. W. Stead, and R. R. Robinson. "Physical Activity and

Employment Status of Patients on Maintenance Dialyses." New England

Journal of Medicine 304, no. 6 (1981): 309-13. Hart, L. G., and R. W. Evans, "The Functional Status of ESRD Patients as

Measured by the Sickness Impact Profile." Journal of Chronic Diseases 40

(1987, Supplement 1): 117S-130S. Health Care Financing Administration. End-Stage Renal Disease Program Highlights,

1983. Baltimore, MD: U.S. Department of Health and Human

Services, 1984.

* Health Care Financing Special Report: Findings from the National Kidney

Dialysis and Kidney Transplantation Study. Baltimore, MD: U. S. Department

of Health and Human Services, 1987. Held, P.J., M. V. Pauly, and L. Diamond. "Survival Analysis of Patients

Undergoing Dialysis." Journal of the American Medical Association 257, no. 5

(1987): 645-50. Hutchinson, T. A., N. F. Boyd, A. R. Feinstein, A. Gorda, D. Hollomby,

and B. Rowat. "Scientific Problems in Clinical Scales, as Demonstrated

ithe Karnofsky Index of Performance Status." Journal of Chronic Diseases 32, no. 9/10 (1979): 661-66. Hutchinson, T. A., D. G. Thomas, and B. MacGibbon. "Predicting Survival

in Adults with End-Stage Renal Disease: An Age Equivalence Index."

Annals of Internal Medicine 96, no. 4 (1982): 417-23. Jones, K. R., D. Nickerson, K. E. Kilpatrick, W. Pfaff, M. Conlon, W. B.

Vogel, and D. D. Bradham. "Analysis of Resource Consumption Variations

among ESRD Hemodialysis Facilities." Final Report Prepared for

the Health Care Financing Administration. Gainesville, FL: University

of Florida, Center for Health Policy Research, 1986. Kutner, N. G., and D. D. Cardenas. "Rehabilitation Status of Chronic Renal

Disease Patients Undergoing Dialysis: Variations by Age Category."

Archives of Physical Medicine and Rehabilitation 62, no. 12 (1981): 626-30. Linder, A., B. Charra, D. J. Sherrand, and B. H. Scribner. "Accelerated

Atherosclerosis in Prolonged Maintenance Hemodialysis." New England

Journal of Medicine 290, no. 13 (1974): 697-701. Lowrie, E. G., and C. L. Hampers, "ESRD Case-Mix Differences: Fact vs.

Fantasy." Dialysis & Transplantation 13, no. 10 (1984): 735-40. Lowrie, E. G., N. M. Laird, T. F. Parker, and J. A. Sargent. "Effect of the

Hemodialysis Prescription on Patient Morbidity." New England Journal of

Medicine 305, no. 20 (1981): 1176-81. Lowrie, E. G., J. M. Lazarus, C. L. Hampers, and J. P. Merrill "Cardiovascular

Disease in Dialysis Patients." New England Journal of Medicine 210,

no. 13 (1974): 737-38. Manning, W. G., J. P. Newhouse, and J. E. Ware. "The Status of Health in

Demand Estimation; or, Beyond Excellent, Good, Fair, and Poor." In

Economic Aspects of Health. Edited by V. R. Fuchs. Chicago , Il: University

of Chicago Press, 1982. Maxwell, J., and H. Sapolsky. "The First DRG: Lessons from the End-Stage

Renal Disease Program for the Prospective Payment System." Inquiry

24, no. 2 (1987): 57-67. Mor, V., L. Laliberte, J. N. Norris, and M. Wiemann. "The Karnofsky

Performance Status Scale: An Examination of Its Reliability and Validity

in a Research Setting." Cancer 53, no. 9 (1984): 2002-2007. Numan, I. M., K. S. Barklind, and B. Lubin. "Correlates of Depression in

Chronic Dialysis Patients: Morbidity and Mortality." Research in Nursing

and Health 4, no. 9 (1981): 295-97. Plough, A. L., and S. Salem. "Social and Contextual Factors in the Analysis

of Mortality in End'Stage Renal Disease Patients: Implications for

Health Policy." American Journal of Public Health 72, no. 11 (1982):

1293-95. Plough, A. L., M. Shwartz, and S. Salem. "Greater Efficiency or Case-Mix

Differences? Fact vs. Fantasy in ESRD Reimbursement Policy." Dialysis

& Transplantation 13, no. 10 (1984): 735-40. Plough, A. L., G. R. Salem, M. Shwartz, J. M. Weller, and C. W. Ferguson.

"Case Mix in End-Stage Renal Disease: Difference between Patients in

Hospital-Based and Freestanding Treatment Facilities." New England

Journal of Medicine 310, no. 2 (1984): 1432-36.

" "Severity Analysis in End-Stage Renal Disease: A Risk Group

Approach." ASAIO 8, no. 1 (1985): 33-39. Pollak, V. E., K. S. Kant, S. L. Parnell, and N. W. Levin. "Repeated Use of

Dialyzers is Safe: Long-term Observations on Morbidity and Mortality

in Patients with End-Stage Renal Disease." Nephron 42, no. 3 (1986): 21717-33. Procci, W. R. "Dietary Abuse in Maintenance Hemodialysis Patients." Psychosomatics 19, no. 1 (1978): 16-24. Radecki, S. e., R. C. Mendenhall, A. R. Nissenson, R. B. Freeman, C. R.

Blagg, J. P. Capelli, D. E. Gentile, and E. I. Feinstein. ""Case-Mix and

Treatment in ESRD: Hemodialysis vs. Peritoneal Dialysis." American

Journal of Kidney Diseases 11, no. 1 (1988): 7-14. Schlesinger, M., P. D. Cleary, and D. Blumenthal. "The Ownership of Health

Facilities and Clinical Decision Making: The Case of the ESRD Industry."

MEdical Care 27, no. 3 (1989): 244-57. Smith, M., B. Hong, J. Michelman, and A. Robson. "Treatment Bias in the

Management of End-State Renal Disease." American Journal of Kidney

Diseases 3, no. 1 (1983): 21-26. Stewart, A. L., S. Greenfield, R. D. Hays, K. Wells, W. H. Rogers, S. D.

Berry, E. A. McGlynn, and J. E. Ware. "Functional Status and Well-Being

Bf Patients with Chronic Conditions: Results from the Medical

Outcomes Study." Journal of the American Medical Association 262, no. 7

(1989): 907-13. Vollmer, W. M., P. W. Wahl, and C. R. Blagg. "Survival with Dialysis and

Transplantation in Patients with ESRD." New England Journal of Medicine

308, no. 26 (1983): 1553-58. Wennberg, J. E., J. P. Bunker, and B. A. Barnes. "The Need for Assessing the

Outcome of Common Medical Practices." Annual Review of Public Health 1 (1980): 277-95. Williamson, J. W., J. I. Hudson, and M. M. Nevins. Principles of Quality

Assurance and Cost Containment in Health Care. San Francisco: Jossey-Bass,

Inc., 1982. Wolcott, D. L., C. A. Maida, R. Diamond, and A. R. Nissenson. "Treatment

Compliance in End-Stage Renal Disease Patients on Dialysis." American

Journal of Nephrology 6, no. 5 (1986): 329-38. Yates, J. W., B. Chalmers, and F. P. McKegney. "Evaluation of Patients with

Advanced Cancer Using the Karnofsky Performance Status." Cancer 45,

no. 8 (1980): 2220-24.

Researh for this article was supported by HCFA Contract No. 14-C-98696/4-01.

Address correspondence and requests for reprints to Katherine R. Jones, Ph.D., School of Nursing, Division of Nursing and Health Care Systems Administration, University of Michigan, 400 N. Ingalls, Ann Arbor, MI 48109. This article, submitted to Health Services Research on December 12, 1988, was revised and accepted for publication on January 14, 1991.
COPYRIGHT 1991 Health Research and Educational Trust
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Jones, Katherine R.
Publication:Health Services Research
Date:Dec 1, 1991
Words:10007
Previous Article:Using medical care: the views and experiences of high-risk mothers.
Next Article:Supply and demand factors in the determination of Medicare expenditures.
Topics:

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters