Conclusions
Our ability to draw any conclusions about the cost-effectiveness of cancer control interventions for low-and middle-income developing countries is limited, because most cost-effectiveness studies in this area have been conducted in high-income, developed countries. Cancer control interventions that appear to be cost-effective in high-income countries may not be cost-effective in low-income countries, even when the lower cost of providing health services is taken into account.
A useful way to draw inferences about the relative cross-country affordability of interventions is to translate cost-effectiveness ratios into percentage of per capita gross national product (GNP) per YLS (WHO 2001a). Our preliminary analysis of breast cancer screening in India, for example, suggesting an absolute cost-effectiveness level for screening mammography of about US$2,000 per YLS, compared with about US$3,000 per YLS in the Netherlands. At about 10 percent per capita GNP per YLS, screening mammography might be considered to be extremely cost-effective for the Netherlands. In India, however, we found a CE estimate equal to 400 percent per capita GNP per YLS suggesting that national policy makers would be much less likely to consider screening mammography as a viable intervention given India's health care budget constraints. However, they might well consider a CBE breast cancer screening program, at about 200 percent per capita GNP per YLS in India, to be moderately affordable if the program were definitively established to be effective.
For middle-income developing countries that have cancer incidence rates similar to those in high-income developed countries, the results of cost-effectiveness analyses from the developed countries may be more relevant, although further analysis clearly is needed. The case study of cervical cancer control that was cited earlier suggests that for low-income countries tailored cancer control interventions may need to be developed that would be both cost-effective and affordable. However, that suggestion does not imply that low-tech approaches should be uncritically embraced and assumed to be cost-effective. Until recently, education campaigns to promote breast self-examination were widely advocated as the low-tech alternative to screening mammography for breast cancer control in low-income countries; however, the best current evidence now indicates that such campaigns have no effect on breast cancer mortality (Semiglazov and others 1999; Thomas and others 2002).
In cancer treatment interventions, the cost-effectiveness of initial surgical treatment for treatable cancers, such as breast, cervical, and colorectal cancer, may be in the relatively favorable range of a few to several thousand dollars per YLS, which indicates that such interventions are likely to be cost-effective for middle-income countries and are possibly cost-effective for low-income countries. Although cost-effectiveness ratios for some of the approaches to adjuvant therapy that use conventional radiation and drugs also fall within this relatively favorable range, others are in the range of tens of thousands of dollars for each YLS. Thus, these forms of treatment would likely be considered potentially cost-effective and affordable in middle-income countries but not in low-income countries; however, more detailed examinations of specific cost conditions and available resource endowments for the delivery of cancer treatment services are needed to confirm these preliminary impressions. As with the case of cervical cancer control, treatment interventions that are tailored to the conditions of low-income countries might be shown to be efficacious and more economically attractive than treatment approaches that are transported directly from developed countries; however, research in this area is lacking.
Time Horizon and a Balanced Approach to Cancer Control Programs
The time horizon for cancer prevention and screening interventions is highly relevant to policy makers and health system planners, yet reports on the cost-effectiveness of such interventions often omit information about time horizons. For example, interventions that involve cancer control agents that prevent cancer cases that would have otherwise occurred many years after the preventive action, such as HPV vaccination, have a long time horizon. Similarly, the favorable cost-effectiveness of preventive screening for stomach cancer is not apparent until four decades following the initiation of the intervention. In the case of the 25-year program of CBE in India analyzed earlier, only about 10 percent of the benefits in terms of breast cancer deaths prevented would have been realized after 10 years of program operation. Decision makers must understand and take these time horizons into account when interpreting and acting on cost-effectiveness ratios; however, the long time horizon for cancer prevention and screening interventions is, in itself, not an argument against the application of such interventions. In some cases, countries that are more recent entrants into the field of cancer control may be able to benefit from the experience of developed countries and from the dynamic technical progress in this area to go directly to new innovations. For example, they might be able to implement HPV testing right away as the basis for cervical cancer screening, bypassing cervical cytology. Achieving the optimal temporal balance in comprehensive cancer control represents a daunting challenge to planning, evaluation, and implementation.
Start Small, Scale Up Smart
Because the current understanding of the effectiveness, optimal resource mix, and cost of many cancer control interventions is incomplete and uncertain, especially in relation to low-and middle-income countries, developing countries should start small. By starting small, they can gain knowledge from pilot programs that are well documented with regard to organizational and process factors; that are conducted in controlled settings, if possible; and that are monitored for efficiency, performance, and effectiveness. Thus, for example, new screening or treatment programs can be initiated in focused geographical areas or specific facilities with known and well-characterized target populations, and their performance and outcomes can be compared with matched control areas or facilities. Developing countries should consider scaling up their regional or national programs only after the pilot programs have been shown to perform well.
Starting small also might entail applying an initial pilot program to a limited age range that is estimated to yield the most benefits per resource use or to a limited group of high-risk individuals defined by various risk characteristics, such as first-degree relatives of people with cancer. Indeed, various versions of this approach have characterized the dissemination of many cancer control interventions in developed countries. Organized breast cancer screening programs in some European countries, for example, were first implemented as pilot programs in specific regions and evaluated against control communities (Fracheboud and others 2001; Olsson and others 2000; van der Maas 2001), and regional and national programs were initially limited to the age groups, screening procedures, and screening frequencies estimated to be the most cost-effective. The programs were later extended, in terms of more intensive procedures, more frequent screening intervals, and wider age groups, after monitoring and analysis of initial program performance indicated that the incremental cost-effectiveness of these extensions would be favorable (Boer and others 1995; Shapiro and others 1998). The United Kingdom has taken a similar approach to colorectal cancer screening (Steele and others 2001).
