Cost-effectiveness is getting more attention these days, and the burgeoning number of publications on this subject as well as the increased attention to medical accountability in terms of effectiveness and cost make it important that we all understand how to interpret cost-effectiveness analyses. This is easier said than done because reading the literature on cost-effectiveness is like navigating through the Tower of Babel.
Cost-effectiveness boils down to value for money. Some interventions may be very expensive but well worth it (eg, bypass grafting in patients with 3-vessel obstructive coronary artery disease and depressed left ventricle function), whereas others may be cheap but wasteful because they simply do not work (eg, taking vitamin E for the prevention of coronary artery disease). In mathematical terms, the key quotient is the incremental cost per incremental clinical benefit (Δ cost/Δ life expectancy or preferably quality-adjusted life expectancy; ie, comparing 2 alternative treatment strategies modeled out to a typical lifetime from a societal perspective). But how do we determine which interventions are truly worth the cost? The most basic conventional litmus test of this in the United States is what payers such as Centers for Medical and Medicaid Services or other financers of health care (such as Blue Cross Blue Shield) are willing to pay, and this is in the range of $50 000 to $100 000 additional cost per additional quality-adjusted life-year (QALY). Even this willingness-to-pay threshold is affected by many other (including political) factors. From a global perspective, it depends on a nation's wealth, which directly affects willingness to pay. A poor nation, for example, may only be able to afford cost-saving interventions such as vaccination programs or water treatment interventions, whereas certain sectors of the United States are even willing to pay for expensive screening programs that, using conservative assumptions, likely cost over $1 million per additional life-year saved, or possibly even infinite costs in situations in which the marginal benefit is unknown and possibly nil.