An economic evaluation of prolonged mechanical ventilation

Abstract

Objective: Patients who receive prolonged mechanical ventilation have high resource utilization and relatively poor outcomes, especially the elderly, and are increasing in number. The economic implications of prolonged mechanical ventilation provision, however, are uncertain and would be helpful to providers and policymakers. Therefore, we aimed to determine the lifetime societal value of prolonged mechanical ventilation.

Design and patients: Adopting the perspective of a healthcare payor, we developed a Markov model to determine the cost effectiveness of providing mechanical ventilation for at least 21 days to a 65-yr-old critically ill base-case patient compared with the provision of comfort care resulting in withdrawal of ventilation. Input data were derived from the medical literature, Medicare, and a recent large cohort study of ventilated patients.

Measurements and main results: We determined lifetime costs and survival, quality-adjusted life expectancy, and cost effectiveness as reflected by costs per quality-adjusted life-year gained. Providing prolonged mechanical ventilation to the base-case patient cost "dollars"55,460 per life-year gained and "dollars"82,411 per quality-adjusted life-year gained compared with withdrawal of ventilation. Cost-effectiveness ratios were most sensitive to variation in age, hospital costs, and probability of readmission, although less sensitive to postacute care-facility costs. Specifically, incremental costs per quality-adjusted life-year gained by prolonged mechanical ventilation provision exceeded "dollars"100,000 with age >or=68 and when predicted 1-yr mortality was >50%.

Conclusions: The cost effectiveness of prolonged mechanical ventilation provision varies dramatically based on age and likelihood of poor short- and long-term outcomes. Identifying patients likely to have unfavorable outcomes, lowering intensity of care for appropriate patients, and reducing costly readmissions should be future priorities in improving the value of prolonged mechanical ventilation.

Figure 1. The Decision Model

The solid box on the left represents the decision made between provision of prolonged mechanical ventilation or withdrawal of care. The circle represents a chance node. Triangles represent death. The encircled “M” represents entry into the Markov tree (see also Figure 2). MV = mechanical ventilation.

Figure 2. Markov Model

The Markov model represents clinical states in which a person could exist during each one-week period as they are followed until death: a person can live at home, be cared for at a post-acute care facility, be readmitted to an acute care hospital for complications, or die. The dashed lines extending from the post-acute care facility state depict in greater detail what patients may experience should they be discharged to one of these care locations. LTAC=long-term acute care facility, SNF=skilled nursing facility, NH=long-term nursing home.

Figure 3. Tornado Diagram

The tornado diagram represents the results of one-way sensitivity analyses. It shows how varying the range of each input variable (base-case values in parentheses) separately in the decision model affects incremental cost-effectiveness ratios (cost per QALYs) when provision of mechanical ventilation for 21 or more days is compared to ventilation withdrawal. The vertical red line represents the incremental cost-effectiveness ratio of the base-case analysis, $82,411 per QALY gained. LTAC=long-term acute care facility and SNF=skilled nursing facility.

Figure 4. Affect of Age on Cost-Effectiveness Ratios

This graph demonstrates both incremental costs per life-year (dashed line) and incremental costs per QALY (solid line) of PMV provision compared to withdrawal of ventilation, stratified by patient age.