A comprehensive prognostic index to predict survival based on multiple comorbidities: a focus on breast cancer.

BACKGROUND: The presence of comorbidities often influences clinical decision-making, although many studies exclude patients with comorbid disease for the sake of analysis. OBJECTIVES: The purpose of this study was to develop a Comprehensive Prognostic Index (CPI), designed specifically for breast cancer patients. RESEARCH DESIGN: This study linked Medicare claims with the Kentucky Cancer Registry and developed two models based on 1 year survival; one focused on deaths caused by breast cancer and the other on deaths from all causes. Comorbidities were derived from inpatient and ambulatory claims for up to 2 years before the diagnosis of breast cancer. SUBJECTS: Subjects included a cohort of 848 elderly women first diagnosed with breast cancer in the state of Kentucky in 1993. MEASURES: Each model identified the comorbidities specific to breast cancer that were detrimental to survival, and generated a refined comorbidity index. The CPI integrated these measures with age and stage of cancer into a comprehensive prognostic index. RESULTS: Nearly two-thirds of the patients had evidence of at least one comorbidity. Survival rates decreased with age, more advanced stage, and increased comorbidity burden, as expected. The interaction of comorbidity burden with either age or stage was particularly strong for the older and more advanced stage of cancer. CONCLUSIONS: The CPI could be a useful tool in breast cancer intervention studies and a prognostic aid for clinicians.

Linking claims and cancer registry data: is it worth the effort?

Insurance claims and cancer registries represent different sources of data, each with advantages and limitations insofar as describing severity of illness, cost, utilization, and outcome of care for cancer patients. Although registries typically are able to identify stage of disease and initial course of treatment, claims-based data include more detailed information on the cost and utilization of medical-care services. Moreover, claims data can provide critical information about the use and location of medical services prior to diagnosis of cancer. The purpose of this paper is to argue the advantages of merging claims with registry data by considering evidence and insights from the literature. We discuss the advantages and disadvantages of claims data, describe studies that compare claims and registry data, and highlight the advantages of linking both sources of data. The latter is accomplished by considering a major advantage of each source of data. Registry data contain useful measures of severity of illness (cancer stage, tumor size, and sites of metastasis), whereas claims data include a detailed history of the cost of medical service.

Complications, adverse events, and iatrogenesis: classifications and quality of care measurement issues.

Accountability in the healthcare system demands the development of valid and reliable measures of quality, particularly outcome measures that have been risk-adjusted for factors that increase the probability of a poor outcome. Although the literature documents the existence of complications, adverse events, and iatrogenic illness, these concepts have not been compared and discussed thoroughly. This article ponders complications as a measure of quality of care by proposing a three-level classification scheme and by examining the incidence, consequence, and determinants of these events.

Breakeven under capitation: pure and simple?

This article addresses the issue of breakeven analysis as applied to managed care under capitation. The traditional two-dimensional breakeven analysis is expanded into three dimensions: cost, enrollment, and utilization. This issue is examined and then visualized through the use of three-dimensional graphics.

Rural hospital closures and access to services.

This article examines the implications resulting from the closure of 25 rural hospitals during 1990. The implications are evaluated by estimating travel distance and time to the nearest open hospitals. In addition, the types of services offered in the hospitals studied were measured to provide a view of potential change in access to services. The average travel distance and time to the nearest hospital after closure was 25.7 miles and 30.2 minutes, respectively. In most cases, the remaining hospitals offered a broader scope of services than did the hospitals that closed. A possible interpretation is that the hospital closures resulted in a tradeoff between breadth of services and rapid access for emergency conditions.

Primary care, avoidable hospitalization, and outcomes of care: a literature review and methodological approach.

The purpose of this article is to review the literature on the relationships between primary care, potentially avoidable hospitalizations, and outcomes of care and to develop a methodology to study these relationships. The methodological approach includes developing criteria to select medical conditions, aggregating patient claims files of both ambulatory and acute care records, and delineating episodes of care. A taxonomy of physician visits is proposed that classifies visits on the basis of type of care, type of illness, and linkage to hospital episodes. A structural model of use and outcomes is specified that includes hazard rate models to estimate the likelihood of a potentially avoidable hospitalization, primary care and ad hoc physician visits, and mortality; the latter suggests a modification of the Health Care Financing Administration methodology that includes physician visit variables.

Interpreting the Health Care Financing Administration's mortality statistics.

In this article, the methodology used by the Health Care Financing Administration in the 1992 release of 1990 mortality statistics is described, and the performance of one outlier hospital is evaluated as a case study. The study hospital is compared to all other hospitals, and to a smaller cohort of 200-to-299-bed minor teaching hospitals, in terms of predicted and observed mortality rates and mortality model determinants. Proportionately more patients treated in the study hospital were women and had cerebrovascular degeneration or chronic renal disease; fewer patients had cardiovascular disease. Substantially more patients from this hospital were transfers from a skilled nursing facility. Fewer patients were admitted through the emergency department. Although patients tended to be more seriously ill overall compared with other hospitals in the country, observed mortality rates were still higher than predicted. Possible explanations for the discrepancy were coding inconsistencies, inability to control adequately for the severity of illness of transfers from skilled nursing facilities, or quality of care problems.

Multiple sclerosis as a comorbidity: a study of resource utilization and outcomes of care.

The purpose of this paper is to compare the resource utilization and outcomes of care of elderly hospitalized patients with multiple sclerosis (MS) as a comorbidity to a comparison group without multiple sclerosis, matched by age, gender, and diagnosis related group (DRG) using the 1989 Quality Care (QC) MEDPAR file. The results of this paper demonstrate some differences in resource use. MS discharges incur lower average charges ($8698 for MS discharges, $8977 for controls), although the results are not statistically significant. Proportionately fewer MS discharges use intensive or coronary care services (14.7% versus 18.5%, P < .05). MS patients had a lower, but not statistically significant, 30-day mortality rate than the comparison group matched by age, sex, and DRG (7.6% versus 8.8%), a lower rate of readmission (13.9% versus 16.7%, P < .05), and a lower rate of complications (2.6% versus 4.7%, P < .05), although the results are not consistent across DRGs. Charges, length of stay, and intensive/coronary care utilization were lower, as expected, for a subset of less comorbid MS discharges and controls with, at most, four diagnoses. In this subset, MS discharges and controls were not statistically different, with the exception of overall complication rates (1.0% for MS versus 4.1% for controls, P < .05) and surgical complication rates (2.1% for MS versus 10.0% for controls, P < .05). It is unclear whether these results are due to differences in disease severity, case-mix within DRG, surgical risk, patient surveillance, or quality of care.

Patterns of comorbidity in elderly patients with multiple sclerosis.

This study explored the prevalence of comorbid conditions in hospitalized patients with multiple sclerosis (MS) who were 65 years of age or older. Using 1989 data from the Quality of Care Medicare Provider Analysis and Review (MEDPAR) file, hospitalized MS patients were compared with respect to discharge diagnoses to an age- and sex-matched group of hospitalized patients without MS. As expected, the following discharge diagnoses were more common (P < 0.05) for MS patients: urinary tract infection, pneumonia, septicemia and cellulitus. In contrast, MS patients were less likely (P < 0.05) to have discharge diagnoses of acute myocardial infarction, heart failure, hypertension, angina pectoris, cerebrovascular disease, diabetes mellitus and chronic obstructive pulmonary disease. Possible explanations include under-reporting of certain comorbid conditions on discharge records of MS patients, a protective effect of MS or its treatment, reduced prevalence of risk factors, disproportionate mortality in younger MS patients with comorbidity and the benefits of medical surveillance.

Financial and clinical performance: bridging the gap.

The article explores the relationship between the cost and the quality of hospital care by elaborating a conceptual model of hospital performance. The model relates the financial health status of an organization (financial integrity) to the quality of care provided by that organization (clinical integrity) within an environment that is characterized by various forms of risk. The model suggests that both concepts determine the corporate destiny (success, bankruptcy, or merger) of the organization. If this model proves valid empirically, the results could be used as an early warning system to identify hospitals that might experience financial or clinical distress.

Profitable and unprofitable DRGs: the implications for access.

Given a choice, hospitals would prefer to admit a patient with the potential to contribute to an accounting profit and prefer not to admit a patient with the potential to contribute to an accounting loss. It is suggested that if all hospitals found the same DRGs to be unprofitable, access to inpatient care would be denied those patient types. A set of 509 hospitals was stratified according to bedsize, Medicare load, type of control, teaching status and geographic location. The 10 most and 10 least profitable DRGs were identified for each hospital category and a Spearman's rank order correlation was used to determine the similarity or dissimilarity across hospital category. The results indicate that the more alike hospitals are in terms of bedsize, Medicare load and teaching status, the more alike are the DRGs that are determined to be unprofitable (or profitable). Conversely, the less alike they were on these characteristics, the less alike were the unprofitable (or profitable) DRGs. There were no differences evident when the hospitals were classified according to type of control or geographic location. These results are generally encouraging in terms of potential access but disturbing in terms of possible further financial threat to rural hospitals.

Spreading the "good news" of total quality management: faith, conversion, and commitment.

In many ways the spread of total quality management (TQM) across this country can be compared to a religious conversion. Both cases are characterized by a philosophical shift with far-reaching changes in responsibilities and incentives for the people involved. This article bridges the disciplines of theology and health services management by elaborating a metaphor in which TQM is compared to various aspects of the Judeo-Christian faiths, such as the role of laws and standards; the importance of miracles, prophets, and evangelists; and the practical applications of living out the faith.

The impact of diagnosis related group profitability on the skimming and dumping of psychiatric diagnosis related groups.

Psychiatric DRGs are identified in terms of their relative profitability within each hospital of a 386 hospital cohort. It is then determined whether hospitals admitted more of the more profitable and fewer of the less profitable patients over the period 1983-1987 (skimming). Also determined is whether hospitals discharged more of the less profitable to other short term hospitals over the same period of time (dumping). The findings generally indicate that this did not happen.

Outcomes of care for anesthesia services: a pilot study.

This paper describes a pilot study involving 4347 patient abstracts collected by anesthesia providers during the first quarter of 1991 under the auspices of the American Association of Nurse Anesthetists. Descriptive statistics are presented on surgical site, type of provider and utilization of anesthetic agents as well as the prevalence of 87 preexisting conditions and the incidence of 103 adverse events.

Toward the development of integrative risk-adjusted measures of quality using large clinical data bases. The case of anesthesia services.

This article describes a process to risk-adjust multiple outcomes of care and aggregate them into integrative measures of quality. A methodology is outlined for anesthesia services which is designed to use the new data base that is being constructed by the American Association for Nurse Anesthetists. Most of the methods should apply to other health professions as well, if outcomes of care and risk factors can be identified. The basic approach is to choose either exemplary or adverse outcomes of care which are under the control of the provider and to standardize these outcomes to take into consideration multiple risk factors.

The measurement of mortality. A risk-adjusted variable time window approach.

This paper describes the development of risk-adjusted mortality indices (RAMI) using 1985 MEDPAR data from 657 hospitals. The RAMI methodology is adopted from the Commission on Professional and Hospital Activities, however, both inhospital and post-discharge deaths are counted within time windows that vary by clinical condition. Five different RAMI measures (expected deaths/observed deaths) are developed, compared, and aggregated into various hospital characteristic strata. These measures vary by which discharge is held responsible for deaths within a time window, and whether or not inhospital deaths that occur beyond the time window are included. The RAMIs using varying time windows are compared with the RAMIs based upon inhospital deaths only. The inhospital RAMI was higher for the nonteaching hospitals (.95) as compared with the major and minor teaching institutions (.91 and .89). The RAMIs using the varying time windows, on the other hand, tend to be higher for the teaching institutions (e.g., 1.07 for major teaching hospitals; 0.99 for nonteaching hospitals).

The relationship between quality and cost: pure and simple?

This study focuses on the relationship between the quality and the cost of hospital care. I report the findings of variable cost regressions that include outcome indicators of hospital quality: risk-adjusted mortality and readmission indices. Hospital level data are aggregated from a modified 1985 Medicare Provider Analysis File (MEDPAR) and the Commission of Professional and Hospital Activities. Results show several quality measures to be statistically significant determinants of cost. With each measure, the cost-quality relationship is nonlinear but not monotonically increasing throughout the entire range of quality.

A reassessment of hospital product and productivity changes over time.

Were the changes found in the first year of the prospective payment system (PPS) one-time changes that attenuated as hospitals gained familiarity with the system? The results of this research show that, over time, discharges to home (self-care) continued to decrease, discharges to home health agencies continued to increase, but transfers and discharges to skilled nursing facilities or intermediate care facilities accounted for an increasing share of total discharges. After a dramatic decrease in the first year, the use of laboratory tests, diagnostic tests, and X-rays returned, over time, almost to pre-PPS levels.

Were hospitals selective in their product and productivity changes? The top 50 DRGs after PPS.

Five separate hospital products are identified based on the concept of the amount of disease remission achieved by the hospital. The parameters of this concept are illness level on admission and discharge location. In a cohort of 646 nonfederal, short-term hospitals over the period 1980-1984, changes in the hospital product are examined separately in the 50 diagnosis-related groups (DRGs) with the greatest volume of Medicare discharges. Productivity changes, as defined by the number of certain inputs, are also examined. In both sets of analyses, patient severity level is controlled for by indexing to the base year (1980) case mix. The purpose of this study was to examine whether the dramatic product and productivity changes following implementation of the prospective payment system, as found in our earlier work, were across-the-board changes or the result of selective changes, specific to certain DRGs or products. The results suggest that the changes were an across-the-board phenomenon. Policy implications are discussed.