The Large Cost of Critical Care: Realities and Challenges

In this issue of Anesthesia and Analgesia, Bloomfield (1) writes a provocative article about costs and cost drivers in critical care medicine in the United States (US); he focuses specifically on the role of new and existing technologies through the prism of market economics. A free market health care system allows providers of health insurance (privately owned companies) and health care services (hospitals, clinics, and physicians) to compete for patients and operate on a for-profit basis. In this paper, we will make two observations about the economics of critical care. The first is that we need to look beyond the free market for the optimal model of health care financing. The second is that the potential to reduce intensive care unit (ICU) costs, either by limiting care at the end of life or by rigorously evaluating expensive technologies, may be limited. Bloomfield (1) identifies several crucial determinants of technology use and development in the US that have contributed to exploding health care costs. For patients, there is no natural economic barrier; insurance companies have a frequent willingness-to-pay, and the insured are largely shielded from the true costs of the interventions they receive. For physicians, adequate prognostic information is frequently lacking, and faced with prognostic uncertainty, the treatment recommendation is usually to continue life support. Moreover, the duty to rescue and preserve life is embedded in society and in medical culture. A closely related theme is the cultural representation of disease as an enemy subject to defeat or delayed advancement, given adequate resources and skills. Bloomfield suggests that technology assessment should play a greater role in the definition and deployment of effective health technologies. In particular, he identifies information technology, conventionally thought to improve efficiency at the point of care (by providing instant access to patient notes and test results), as a means to enable technology evaluation through observational research using data on patient outcomes and technology use stored in large data repositories. Let us examine some of these arguments. Doomsday predictions of soaring and unsustainable health care costs abound, particularly in countries such as Canada in which most of the health care funding comes from government sources. However, as Guyatt et al. (2) remind us, discussions of raw cost data can mislead if the effects of inflation, population growth, and available resources are not considered. For example, Figure 1 shows that per capita health care expenditures in the US are increasing more rapidly than the Organization of Economic Cooperation and Development (OECD) average of 30 industrialized countries (3). The interpretation changes when these expenditures are expressed as a fraction of gross domestic product, a measure of societal resources and, therefore, ability to pay. Figure 2 reveals that US health care expenditures, whereas 50% more than the OECD average, have remained remarkably stable since 1990 (4).Figure 1: Trend in per capita health care costs (in 1995 US dollars) for Canada, the United States, and the average of countries in the Organization for Economic Cooperation and Development (OECD).Figure 2: Health care costs as a percentage of gross national product (GNP) for Canada, the United States, and the average of countries in the Organization for Economic Cooperation and Development (OECD).The US’s approach to health care expenditures is unique in the industrialized world: it spends the most but with the smallest contribution (44.3% in 2000) from government taxation revenues. The implication is that the US relies heavily on private sources for health care financing. As Bloomfield (1) points out, large health care expenditures in the US have not consistently produced the best rankings for common health measures; life expectancy is less and infant mortality more frequent than in other industrialized countries that spend less on health care. Therefore, given comparative performance data within the OECD, a health care financing system dominated by the private sector may not achieve the optimal combination of expenditures and patient outcomes. Returning to critical care costs, we contend that framing the discussion in terms of an American model will limit due consideration of the optimal system of health care financing. Bloomfield (1) also argues that providing health insurance increases the demand for health services because patients are not directly responsible for the costs of services they receive in this model. There is limited supporting evidence for this argument from the Rand Health Insurance Experiment (5,6), a randomized controlled trial designed to determine the effect of varying levels of personal co-payment on the use of health services. Methodologic limitations of this study that have previously been raised (7) include the exclusion of elderly and disabled patients (who use more health services), the income-related caps on personal health expenditures featured in all co-payment arms, and the limited measurement of health outcomes. Although total expenditures were higher in the zero co-payment arms, driven by increased ambulatory visits and hospital admissions, there were few differences in health outcomes. The application of these results to the critical care setting is questionable because of the nonelective nature of most admissions and the commonly held ethical principle that saving endangered lives is part of providing best care. Although the concept of a crisis in health care costs may thus be open to argument, the contribution of critical care services to total expenditures is clearly high (8). Given the current differences in health care financing models for critical care, what is the prospect of reducing costs in this hospital service? A crucial principle in any discussion is that cost containment requires control over patient flow. Critical care services, such as those provided by emergency and radiology departments, do not directly generate demand. Admissions come from emergency departments, hospital wards, and operating rooms through activities and decision-making that are not regulated by intensivists. For example, intensivists are generally not involved in the selection of candidates for high-risk surgery or in discussions with patients about resuscitation on the medical wards. In addition, as the critical care team has become more successful in supporting life through episodes of acute illness, the demand for its services has increased. Although there may be room to decrease the cost-per-admission through more efficient care, definitive cost reductions would require fewer admissions by limiting ICU resources (beds and personnel). The major consequence of resource limitation would be to delay or deny admission to patients who would potentially benefit. This approach would entail radically changed attitudes about the appropriate role of advanced supportive care, and physician and patient acceptance would be questionable. One intuitively attractive option for reducing ICU costs would be to identify patients who would not benefit from critical care before or early in their admission and transfer them to a more appropriate setting for palliative care. As Luce and Rubenfeld (9) have recently argued, there are several reasons why this strategy is unlikely to produce significant cost savings: (a) most hospital costs are fixed and relatively insensitive to reductions in ICU length of stay; (b) current prognostic systems lack specificity in the prediction of death, limiting their usefulness in decision-making for indeterminate prognosis patients who are the costliest; and (c) limiting care even to those patients with a dismal prognosis would not substantially reduce costs, because these patients constitute a minority of ICU admissions. What is the potential role of technology assessment in reducing ICU costs? It is important to cast the net broadly when evaluating technologies. Nonlabor costs account for 38%–56% of total ICU costs (10,11). Of those, routine diagnostic tests (laboratory and plain radiographs), medications, and supplies, including technological devices, each contribute substantially, although the attributable cost of all routine tests is more than that of any single technological device. Therefore, the greatest potential for cost reduction may be in reducing the use of individually inexpensive tests rather than focusing on expensive (but less often used) devices, such as the pulmonary artery catheter. A significant methodologic challenge in the assessment of ICU diagnostic technologies is that effects on important clinical outcomes depend not only on the quality and timeliness of the information, but also on the subsequent strategy of information use. The purpose of monitoring equipment is to provide physiologic data to support short-term and evolving decisions. Even if it can be shown that clinical outcomes are not inferior when a strategy of information restriction (fewer diagnostic tests or less monitoring) is used, clinicians may be reluctant to treat patients when there is less information available, especially when it can be easily obtained with few adverse effects. Furthermore, in the current medico-legal culture emphasizing responsibility to provide the best care for each individual patient, there is a strong disincentive for physicians to practice in a more cost-effective manner if doing so would potentially increase the risk of harm and thus liability. Closely related to ICU diagnostic tools are technological devices supporting organ function such as ventilators, intraaortic balloon pumps, and renal replacement machines. There are no widely accepted guidelines for the assessment of life-support technologies, but important components include engineering, clinical effectiveness, and costs. The focus on technology must be paralleled by continued efforts to define best practices including pharmacologic therapies (12), fluid and nutritional interventions (13,14), and ICU structures and staffing models (15–17). As Bloomfield (1) points out, information technology promises to consolidate and present existing information so that clinical efficiency improves and medical errors decrease, especially for common complex conditions for which evidence-based clinical practice guidelines may be developed. Clinical informatics applications include physician order entry systems, electronic medical records with laboratory and radiology data, and computerized clinical decision support systems (CDSSs). A recent systematic review of CDSSs showed improved physician performance and patient outcomes in a variety of non-ICU settings (18). More recently, studies of a CDSS for mechanical ventilation for patients with acute respiratory distress syndrome have shown improved patient morbidity (19,20). Whether this approach will reduce ICU costs per case (for example, by allowing the implementation of diagnostic pathways and standardizing medication use) remains to be tested. In summary, Bloomfield (1) has described the convergence of several important factors exerting upward pressure on health care costs in the US, including rapid development of health care technologies, inadequate technology assessment and prognostic information, and lack of fiscal restraints on patient demand. We have argued that other models of health care financing may offer lessons on how to contain costs without sacrificing health outcomes. In addition, there may be opportunities to reduce the cost per ICU admission by eliminating unnecessary diagnostic tests and using information more effectively. However, it is questionable whether robust assessments of life-support technologies will substantially reduce ICU costs if the demand for ICU services remains high and continues to grow.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.