Understanding costs and cost-effectiveness in critical care: report from the second American Thoracic Society workshop on outcomes research.

Economic evaluations are increasingly common in the critical care literature, although approaches to their conduct are not standardized. The American Thoracic Society convened a workshop to address methodologic and reporting issues for economic analyses in critical care and to determine how guidelines from the U.S. Public Health Service Panel on Cost-effectiveness in Health and Medicine (PCEHM) were applicable to critical care. We identified several issues that hamper cost-effectiveness analyses (CEAs) in the critically ill. Data on the effectiveness of intensive care unit (ICU) interventions are often lacking; ICU patients are complex, with multiple concurrent problems and interventions; most ICU therapies are only supportive, and therefore may not individually result in improved outcome; accurate cost data are not commonly available and are difficult to obtain; there is no standardized approach for measuring or valuing costs across countries; typical outcomes in ICU studies (e.g., short-term mortality) are not ideal for CEAs while preferred outcomes for CEAs (e.g., long-term quality-adjusted survival) are rarely collected; valuing the importance of appropriate end-of-life care, an important aspect of ICU care, is difficult, and the burden of critical illness on family members is not easily captured in a CEA. Nevertheless, many of these problems are not unique to critical care, and we believe the PCEHM guidelines can be adapted to the critical care setting. We recommend all CEAs in the critically ill include a PCEHM reference case, where the cost-effectiveness ratio is calculated by adopting a societal perspective, estimating long-term costs and quality of life after ICU care, applying a 3% annual discount rate to costs and effects, and conducting multiway sensitivity analyses. Because elements of the reference case, such as long-term costs and quality of life, may only be estimated using modeling and assumptions, we also recommend inclusion of a "data-rich" case, where the cost-effectiveness ratio is generated as closely as possible from data on actual patient outcomes and costs (e.g., hospital costs per hospital survivor). We recommend that investigators conducting a CEA concurrently with a randomized trial make the proposed model available (e.g., via the Internet) before unblinding of trial data to minimize bias. Adopting a standard approach to CEAs of ICU therapies will provide a valid and more transparent evidence base for health care policy with regard to care of the critically ill.