Performance Improvement 1997. Ventilator-Dependent Unit Demonstration: Outcome Evaluation and Assessment of Post-Acute Care



Chronic ventilator patients experience high mortality, and their care is exceedingly expensive. The purpose of this evaluation was to determine whether specialized units serving chronic ventilator-dependent patients deliver better clinical outcomes at a reasonable cost to Medicare. The evaluation compared chronic ventilator patients treated in ventilator-dependent units (VDU's) at four demonstration sites with similar patients treated in conventional hospital settings, usually in intensive care units. The evaluation found that patients treated in VDU's have improved clinical outcomes and lower hospital daily costs, but higher overall costs to Medicare because of greater longevity. For example, while the patients' median survival from the time of admission was about 5 months longer, and their functional status superior at discharge, the overall cost to Medicare was increased by 35 percent, due to longer life expectancy. The findings across the four demonstration sites varied greatly, however. Because of the variability in clinical outcomes and the difficulty of screening patients for placement in VDU'sþboth of which can lead to greater costs without commensurate clinical benefitsþthe evaluation concluded that nationwide implementation of VDU's is not recommended. However, the evaluation did find a small number of special VDU centers of excellence to be warranted. The analysis of costs and outcomes provides important information to policymakers and health providers about a seriously ill population with intensive resource needs.


The purpose of the evaluation was to assess whether chronic ventilator patients covered by Medicare were better served at a reasonable cost by specialized hospital units called ventilator-dependent units. These units, studied at four Medicare demonstration sites, have a rehabilitation focus and are staffed by a highly trained multidisciplinary team of health professionals. The daily cost of care in a VDU is about $500 to $700 less than that of an intensive care unit, where most hospitals care for chronic ventilator patients. The evaluation compared costs and clinical outcomes of 211 chronic ventilator patients treated in VDU's with 401 comparison cases treated in conventional hospital settings. The evaluation also projected the overall costs of providing care in VDU's to Medicare patients nationwide.


Mechanical ventilation is a life-sustaining technology for patients suffering from acute respiratory failure. The ventilator is an apparatus that delivers air to the lungs of patients unable to breathe on their own, usually because of failure of other organ systems or severe chronic illnesses. Because of the severity of their underlying illnesses, 49 percent of Medicare ventilator patients died in FY 1988. The 133,500 Medicare patients discharged in FY 1994 after an episode of mechanical ventilation had an average length of stay of 23.1 days, with total Medicare expenditures amounting to $3.5 billion. Thus, ventilator patients' high mortality, coupled with the high cost of treatment, warranted further scrutiny by the Health Care Financing Administration (HCFA), the agency that administers Medicare.

The focus of the evaluation was on chronic ventilator patients, that is, the small subset of ventilator patients who need ventilation for extended periods. Chronic ventilator patients are especially important to study because they are even more seriously ill and their care is correspondingly more costly. In this evaluation, a chronic ventilator patient was defined as one receiving at least 20 days of ventilation.

This evaluation was conducted under an exemption to Medicare's prospective payment system to allow demonstration hospitals to be paid for VDU care on a cost-reimbursement basis, with some incentives to control costs. Under Medicare's prospective payment system, Medicare normally pays a flat fee for chronic ventilator patients. Because of earlier reimbursement problems, patients are now grouped under one of three distinct diagnosis-related groups specifically tied to their receipt of chronic ventilation, rather than to their underlying illnesses. Yet problems still exist with the revised groupings, because hospitals and patients continue to have strong incentives to discharge patients to another setting. Understanding costs and effective treatments therefore permits further refinement to the prospective payment system for chronic ventilator patients.


The evaluation of costs and outcomes relied upon a constellation of data bases, including Medicare's enrollment and claims data and special clinical data sets. The four VDU demonstration sites were the Mayo Clinic; Temple University Hospital; RMS Health Providers in Hinsdale, Illinois; and Sinai Hospital of Detroit. New clinical instruments were developed for these sites to obtain a comprehensive portrait of the course of ventilator episodes. Claims data were used to track the VDU and the comparison group through 18 months following hospital admission.

The comparison group was drawn from a HCFA pilot data base known as the Uniform Clinical Data Set System, which is no longer in place. The data on chronic ventilator patients were derived from hospital records in five States. A major difference between VDU and comparison cases was that the former were screened for entry into VDU's, while the latter were not. Screening according to criteria set by HCFA excluded medically unstable patients and those with poor prospects for rehabilitation. Multivariate models were employed to control for differences between the VDU and comparison cases.


Three sets of findings were generated by the study: admissions, clinical and cost outcomes, and national implementation findings. The admissions findings revealed that, in many instances, patients were improperly admitted to VDU's. In other words, it was difficult for VDU sites to meet all of HCFA's complex criteria for entry, which relate to medical stability and reasonable potential for rehabilitation. Across VDU sites, there was great variability in the application of admission criteria.

With respect to outcomes, the evaluation analyzed 15 clinical outcomes and 17 expenditure outcomes. On many of the clinical outcomes, VDU cases fared better than comparison cases. Among the most important findings was that median survival from the point of hospital admission was 258 days for VDU cases and 106 days for comparison cases, a statistically significant increase of about five months. The mortality rate within the hospital was 34 percent for VDU cases and 48 percent for comparison cases. When patients were discharged, VDU cases were in better condition: 34 percent of VDU cases were discharged to their home, compared with 27 percent of comparison cases. Their functional status at discharge was superior, using a special index based on activities like bed mobility, locomotion, toileting, and eating. About 18 percent of VDU cases at discharge scored in the highest functional group, compared with 11 percent of comparisons. At two of the demonstration sites, however, some clinical outcomes were not significantly improved relative to the comparison group.

While daily Medicare expenditures for VDU cases were lower, overall expenditures were about 35 percent higher because patients lived longer. Total hospital expenditures averaged $123,000 per VDU case and $91,000 per comparison case. The mean daily spending within the hospital was $1,468 for VDU cases and $1,740 for comparison cases, because VDU's were generally less costly than alternatives.

In the national implementation analysis, costs were projected for covering VDU's for all eligible Medicare patients. The evaluation found that 67 percent to 80 percent of comparison cases would have been admitted to a VDU if one had been available, yielding an estimate of between 24,000 and 41,000 total admissions nationwide. Total Medicare expenditures for their care were estimated between $0.6 and $2.2 billion, depending on assumptions and duration of coverage. In light of the difficulty of controlling admissions and the variability in clinical outcomes, the evaluation concluded that nationwide implementation of VDU's would be very costly and of questionable benefit.

Use of Results

The evaluation recommends continuation of VDU's at selected centers of excellence to refine rehabilitative treatment for chronic ventilator patients. It also recommends further exploration of financing mechanisms that could encourage integration of care, rather than moving patients from one setting to another. Managed care, particularly through case management, is seen as holding promise for such integration. The evaluation also pointed to research questions that warrant further attention. Finally, by providing detailed information about costs and clinical outcomes for a population at high risk of death, the evaluation raises for policymakers and health care providers the difficult question of whether and how resources should be allocated to a seriously ill population in a manner that optimizes duration and quality of life, without being economically wasteful and prolonging suffering.

Agency sponsor:

Office of Research and Demonstrations

Federal contact:

Michael Henesch
PIC ID: 4590

Performer organization:

The Lewin Group, Fairfax, VA