What is Cost-Effectiveness Analysis?
Cost-effectiveness analysis is a method for assessing the gains in health relative to the costs of different health interventions. It is not the only criterion for deciding how to allocate resources, but it is an important one, because it directly relates the financial and scientific implications of different interventions. The basic calculation involves dividing the cost of an intervention in monetary units by the expected health gain measured in natural units such as number of lives saved. For example, using volunteer paramedics and trained lay people as first responders to accidents costs about US$128 per life saved in South Asia and US$283 in the Middle East and North Africa, whereas using a community-based ambulance costs about US$1,100 and US$3,500 per life saved in the same two regions, respectively. By measuring cost-effectiveness in terms of lives saved, all lives are treated equally regardless of whether the person is an infant who might live another 80 years or a middle-aged person who can expect only another 40 years of life.
". . . using volunteer paramedics and trained lay people as first responders to accidents costs about US$128 per life saved in South Asia and US$283 in the Middle East and North Africa, whereas using a community-based ambulance costs about US$1,100 and US$3,500 per life saved . . . respectively."
Some studies calculate cost-effectiveness using years of life lost as the natural unit for measuring the effect of interventions (box 3.2). This measure treats each additional year of life gained from an intervention as equal. It sums the number of years of life that would be saved by an intervention. Hence an intervention that saved an infant's life (for example, preventing dehydration from diarrhea) would count more than one aimed at saving an older person's life (for instance, preventing recurrence of a stroke).
[Box 3.2]
Because the future is uncertain, common (but not universal) practice is to discount both health gains and costs in distant years. DCP2 uses a discount rate of 3 percent per year, which has the effect of making 80 years of life expectancy at birth worth about 30 discounted years. With discounting, saving an infant's life still gains more years than saving that of a middle-aged person, but the difference shrinks considerably. Interventions that incur costs now but provide gains only years later look less cost-effective under discounting than when gains accrue immediately, but interventions whose costs and health benefits follow the same time pattern are all affected equally and their relative cost-effectiveness is unchanged.
Nevertheless, averting death or prolonging life is not the only goal of health interventions. Investigators have proposed other measures to differentiate between a year of life in perfect health and a year of life with some health impairment. One of the more commonly used measures that addresses this issue is the disability-adjusted life year. A DALY measures not only the additional years of life gained by an intervention but also the improved health that people enjoy as a consequence. It assigns a value of 1 to a single year lived in perfect health. Any health impairment or disability is assigned a disability weight that describes the magnitude of the impairment, with a larger weight if the impairment is severe and a smaller one if the disability is modest. The value of a year lived with a disability then gets a value of 1 minus the disability weight, which measures the remaining degree of health. Researchers have assigned disability weights to various chronic conditions, pain, disability, and loss of bodily functions using a variety of methods, including international surveys that ask individuals to compare the quality of life under different health conditions. DCP2 relied on disability weights calculated by WHO's disease burden studies, sometimes using these to estimate disability from conditions that WHO had not explicitly considered.
". . . a cost-effectiveness analysis that measured health gain by the number of averted deaths would find little value in preventing onchocerciasis, but measuring health gain in DALYs assigns a high value to preserving people's vision . . ."
DALYs are useful for policy makers because they are a more comprehensive measure of population health than merely counting deaths and because they allow comparisons among a wide range of health interventions. Some health interventions are aimed directly at reducing mortality, but many are aimed at reducing the severity of illness and improving the quality of life. With DALYs, these different interventions can be compared against a common standard. For example, a cost-effectiveness analysis that measured health gain by the number of averted deaths would find little value in preventing onchocerciasis, but measuring health gain in DALYs assigns a high value to preserving people's vision because the disability weight of blindness is large.
One of the advantages of using cost-effectiveness ratios is that they avoid some ethical dilemmas and analytical difficulties that arise when attempting cost-benefit analyses. Applying the alternative analytical technique of cost-benefit analysis requires assigning a monetary value to each year of life. By foregoing this step, cost-effectiveness analysis draws attention exclusively to health benefits, which are not monetized. When an intervention leads to health savings, the costs should be subtracted from intervention costs when compared to health outcomes. Many health interventions yield benefits beyond the immediate improvement of health status. For example, healthier parents will be able to provide better care for their children, healthier workers will be more productive in the workplace, and healthier families may avoid falling into poverty. Some health interventions can induce virtuous cycles. For instance, preventing the death of a parent may mean that a family has more income to provide nourishment for growing children. Other health interventions provide important ancillary benefits that are valued independently. For example, the cost-effectiveness of water and sanitation services in reducing gastrointestinal diseases is low, but piped water and sanitation services are valued in and of themselves as a convenience and an environmental improvement.
The values people place on nonhealth benefits are quite high as demonstrated by their willingness to pay for such services, but cost-effectiveness will not measure additional nonhealth-related benefits. Therefore comparing interventions according to cost-effectiveness criteria must be done with a clear understanding that it compares interventions only in terms of their efficiency at improving health, and if nonhealth benefits are going to be introduced into a debate, then they should be considered for all the interventions under discussion and not for a select few.
Cost-effectiveness analysis also requires comparable units for measuring costs. For domestic studies, the cost units in domestic currency will have a clear meaning. In the absence of unit prices of the inputs into interventions, for comparison across countries, DCP2 authors were provided costs for each World Bank region in a widely used currency, usually U.S. dollars. The main question involves whether to use market foreign exchange rates to convert domestic currency costs and compare them to the value of imported and importable inputs expressed in dollars, or whether to use a different conversion factor based on studies of the relative purchasing power of the domestic currency. Because market exchange rates are easier to understand and correspond better to actual financial constraints, DCP2 has used such rates for such conversions.
Cost estimates are affected by prices and prices can vary considerably between, and even within, countries. The authors of DCP2 were unable to collect unit prices of the inputs into interventions in every country, so instead they were provided with average unit prices in each of six developing regions: East Asia and the Pacific, Europe and Central Asia, Latin America and the Caribbean, the Middle East and North Africa, South Asia, and Sub-Saharan Africa (previously published analyses, however, sometimes used WHO regional groupings). In the most complete analyses, the authors multiplied these regional unit prices by the estimated quantities of inputs required for each intervention and then divided by the estimated health effect to derive the cost-effectiveness ratios. In cases where the authors could not find disaggregated information on inputs but some cost-effectiveness measures were reported, they made extrapolations. In some cases, input ratios were available for one region and the authors extrapolated these to other regions (see, for example, DCP2, chapter 30).
To conduct a cost-effectiveness analysis, researchers also need to specify the health intervention in some detail. A health intervention is a deliberate activity that aims to improve someone's health by reducing the risk, the duration, or the severity of a health problem. Such interventions can be defined relative to adverse health events, such as being involved in an accident, contracting an infection, or suffering from a malignant tumor. Primary prevention seeks to avert an adverse health event, while secondary prevention aims to keep an adverse health event from recurring or causing a related problem once it has occurred. Following an adverse health event, interventions can also fall into several categories of case management, including cures, acute care, chronic care, rehabilitation, and palliation (box 3.3).
[Box 3.3]
Characterizing an intervention fully also requires defining the level of care at which it is delivered; the particular supplies and processes involved; and the types of health care workers and any associated services required, such as laboratory tests. The more detailed and accurate the analysis, the more readily investigators can assess whether it is similar to or diverges from how that intervention is characterized in other contexts. For example, health interventions might be provided by a less specialized facility or involve more visits in one country than another.
The scope of the costs included will also affect the cost-effectiveness analysis. Researchers may choose a narrow definition of costs and focus exclusively on the direct variable costs of providing a service; that is, they may only include the costs of additional materials and staff that are required and exclude costs associated with the use of existing infrastructure or installed capacity. In other cases researchers may use wider definitions of costs by apportioning some share of the fixed costs of facilities and administration to the costs of the service. The DCP2 authors were asked to follow the latter approach.
In some studies, researchers include other costs, such as the value of the time patients and family members spend in obtaining a service or the cost of transportation to reach facilities. When more costs are included, the cost per unit of health gain will be higher and the intervention will appear to be less cost-effective. If the interventions that are being compared have similar characteristics, such as all being offered at a similar facility, then including these other costs will not alter the ranking of interventions, but comparisons across interventions that are dissimilar could yield different results if the ratios are otherwise close. To be consistent, DCP2 chapters use only direct costs, because estimates of these other costs are both difficult to obtain and rarely consistent across studies. An ethical problem is also involved if poor people's time is valued only on the basis of their low wages or incomes.
"An ethical problem is involved if poor people's time is valued only on the basis of their low wages or incomes."
