Cost-Effectiveness of Cancer Control Interventions
There is a growing literature on the cost-effectiveness of interventions within each of the four categories above. In this section, we review published studies of the cost-effectiveness of health services-based cancer control interventions, and we present new analyses of the cost-effectiveness of screening interventions for cervical and breast cancer.
Primary Prevention
This subsection reviews studies of the effectiveness and cost-effectiveness of several interventions for the primary prevention of cancer.
Immunization against—or Treatment of—Infectious Agents That Cause Certain Cancers
Infectious agents are causally associated with three of the seven cancers that are the focus of this chapter—liver cancer (HBV), cervical cancer (HPV infection), and stomach cancer (H. pylori infection)—so eliminating these agents through immunization or other means offers hope for preventing such cancers.
The HBV vaccine was designed to prevent liver cancer and is currently the only such vaccine in widespread use. Long-term protection against acute and chronic infection has been demonstrated with the HBV vaccine in a wide range of settings (Coursaget and others 1994; Viviani and others 1999), and recent data support a reduction in hepatocellular carcinoma (Lee, Hsieh, and Ko 2003).
Infection with specific high-risk types of HPV plays a key role in causing cervical cancer. A double-blind placebo-controlled trial of an HPV 16 vaccine reported encouraging efficacy results in young female volunteers who had been fully vaccinated (three doses of vaccine or placebo) over a 1.7-year follow-up period (Koutsky and others 2002). In a more recent study, a bivalent HPV 16/18 vaccine prevented approximately 95 percent of persistent infections with HPV 16 and 18 (Harper and others 2004).
Several modeling studies have explored the potential benefits of HPV vaccination at the population level (Goldie and others 2003; Hughes, Garnett, and Koutsky 2002; Kulasingam and Myers 2003) and have elucidated several priorities for future research, including a better understanding of the heterogeneity of vaccine response and the effects of type-specific vaccination on other HPV types.
Hughes, Garnett, and Koutsky (2002) evaluate the potential effectiveness of HPV vaccination using a dynamic transmission model and find that, when both men and women were vaccinated—assuming 90 percent coverage, 75 percent effectiveness, and 10-year immunity—type-specific HPV prevalence was reduced by 44 percent. When only women were vaccinated, the reduction was 30 percent. The authors show that, if the vaccine targeted only certain types of high-risk HPV, cervical cancer incidence was not reduced proportionally because other high-risk types of HPV progressed to invasive cancer.
Goldie and others (2003) assess the impact of a type-specific HPV 16/18 vaccine calibrated to population-based data for Costa Rica. They find that a vaccine that prevented 98 percent of persistent HPV 16/18 was associated with an approximate equivalent reduction in HPV 16/18-associated cancer and a 51 percent reduction in total cervical cancer. The effect on total cancer was attenuated because of the competing risks associated with oncogenic types of HPV other than HPV 16/18.
Three studies have evaluated the potential cost-effectiveness of HPV vaccination in countries with cervical cancer screening programs (Goldie, Kohli, and others 2004; Kulasingam and Myers 2003; Sanders and Taira 2003). In general, these studies indicate that a program of HPV vaccination that permits a later age of screening initiation and a less frequent screening interval is likely to be a cost-effective use of health care resources in developed countries.
In Fujian province, China, a region of high mortality attributable to stomach cancer, a recently completed randomized controlled trial of H. pylori eradication with antibiotics provides some evidence that this approach may be effective in preventing stomach cancer in the subgroup of H. pylori carriers without precancerous lesions at the time of treatment (Wong and others 2004). A recent randomized trial of H. pylori eradication in Chiapas, Mexico, which used preneoplastic conditions as surrogate markers for the development of gastric cancer, found some evidence for the effectiveness of this treatment (Ley and others 2004).
Several studies, most of them in developed countries, have assessed the potential cost-effectiveness of screening individuals for infection with H. pylori and then eradicating H. pylori with antibiotic therapy as a means of preventing the later occurrence of stomach cancer. Roderick and others (2003) examine the cost-effectiveness of an H. pylori screening program conducted in the United Kingdom. Discounting costs and benefits at 6 percent, they find that the cost-effectiveness ratio for screening for H. pylori, initiated at age 40, is approximately US$28,000 per year of life saved (YLS). Optimal cost-effectiveness was not achieved until the H. pylori screening program had run for at least 40 years. Harris and others (1999) estimate the cost-effectiveness ratio associated with one-time screening for H. pylori at age 50 to be approximately US$50,000 per YLS (in 1995 dollars, 3 percent discount rate) when treatment for H. pylori infection results in a 15 percent reduction in stomach cancer risk. Assuming a 30 percent reduction, the figure was US$25,000 per YLS for the United States, but only a few hundred dollars per YLS in Colombia, which has a much higher rate of stomach cancer and lower health care costs.
Tobacco and Alcohol Control Programs
Tobacco consumption is the most important cause of lung and other cancers of the respiratory system, as well as of esophageal cancer, and may be a contributing factor for several other cancers. The most effective national tobacco control programs combine health promotion, education, and health service interventions with policies. Policy instruments include regulating tobacco advertising and promotion; enacting smoking bans in work-places, restaurants, and public buildings and on public transportation; and increasing excise taxes on tobacco products (Fiore, Hatsukami, and Baker 2002; WHO 2002). Decreased rates of smoking uptake by children and adolescents would result in the greatest potential gain in life years. The WHO Framework Convention on Tobacco Control (WHO 2003b) summarizes tobacco control policies and programs related to regulation, taxation, and education. Da Costa e Silva (2003) shows prioritized treatment approaches for tobacco cessation, based on countries' levels of resources.
Excessive alcohol use accounts for 20 to 30 percent of liver and esophageal cancer (WHO 2001b). Interventions to reduce excessive consumption of alcohol have many principles in common with tobacco control, including the effectiveness of regulatory and taxation measures along with health promotion and addiction treatment programs.
Dietary and Related Interventions
The dietary ingestion of substances produced by the mold Aspergillus flavus, specifically aflatoxin B1, is causally associated with hepatocellular carcinoma. Exposure to aflatoxins may be synergistic with HBV infection in the development of this cancer. Effective means are available for preventing the contamination of grains and other types of food with aflatoxin during the growth, harvest, storage, and processing of such products (Kensler and others 2003; Turner and others 2002). Furthermore, chlorophyllin supplements have been found to reduce the carcinogenic properties of aflatoxin. That finding provides additional evidence for current dietary guidelines that meals should contain foods rich in chlorophylls—for example, spinach and other green, leafy vegetables (Kensler and others 2003).
Among those infected with H. pylori, diet is thought to play a critical role in the progression of superficial gastritis to chronic atrophic gastritis. Prolonged consumption of foods rich in salted, pickled, and smoked products increases the risk of stomach cancer, and increased consumption of fresh fruit and vegetables likely decreases the risk. Obesity is also a well-established risk factor for several cancers (Vainio and Bianchini 2002b). For that reason, WHO recommends that governments seeking to ensure compliance with nutritional objectives conduct appropriate school and public education campaigns on diet and work with the food and agriculture sectors (WHO 2002).
Pharmacological Interventions
Chemoprevention is defined as the reduction of the risk of cancer development through the use of micronutrients or pharmaceuticals. Clinical trials among high-risk individuals to establish the efficacy of chemoprevention via micronutrients (for instance, carotenoids and retinoids) and dietary fiber have been mainly negative (Alberts and others 2000; ATBC 1994; Omenn and others 1996; Schatzkin and others 2000). However, several ongoing clinical studies are examining the potential cancer preventive effects of calcium, vitamin D, folic acid, selenium, and vitamin E (Christensen 2004).
Both case-control and cohort studies show a reduced risk for colorectal cancer after prolonged use of aspirin (Vainio and Morgan 1999). Additional evidence indicates that aspirin has a preventive effect on several other types of cancer, including hormone receptor-positive breast cancer (Terry and others 2004), but questions remain about the balance between the clinical benefits and adverse side effects of long-term aspirin therapy, including gastrointestinal bleeding and hemorrhagic stroke (Imperiale 2003).
Some evidence suggests that the antiestrogen drug tamoxifen may reduce the risk of breast cancer (Gail and others 1999), but there is also conflicting evidence (Powles and others 1998; Veronesi and others 1998). The potential for primary prevention using other selective estrogen receptor modulators is a topic of current clinical research (Lippman, Lee, and Sabichi 1998). Preliminary analyses indicate that the use of tamoxifen to prevent breast cancer could be cost-effective in the United States (T. Smith and Hillner 2000).
Early Detection and Secondary Prevention
This subsection looks at studies of the effectiveness and cost-effectiveness of several interventions for the early detection and secondary prevention of cancer.
Screening for Liver Cancer
Screening methods for early detection of liver cancer include serum assays for alpha-fetoprotein and, potentially, ultrasound. A recently completed randomized controlled trial of liver cancer screening in China evaluated the use of two or six alpha-fetoprotein assays over a period of four years among men age 30 to 69 with chronic HBV (Chen and others 2003). Screening resulted in earlier diagnosis of liver cancer, but because treatment for established liver cancer is largely ineffective, screening did not reduce overall mortality.
Randomized trials that include ultrasound screening for liver cancer and that incorporate recent advances in antiviral preventive treatment have yet to be conducted. Sarasin, Giostra, and Hadengue's (1996) model-based cost-effectiveness analysis explores whether biannual screening of patients with Child-Pugh class A cirrhosis, under a set of assumptions systematically favorable to screening, would be cost-effective. The authors conclude that, even under best-case conditions, screening for liver cancer is not likely to be cost-effective.
Screening for Stomach Cancer
Mass screening programs for the early detection of invasive stomach cancer using radiological or endoscopic techniques have been widely implemented in Japan, where incidence rates of stomach cancer are high.
Babazono and Hillman (1995) compare the cost-effectiveness of three methods for the early detection of stomach cancer in the context of mass screening programs in Japan: indirect radiology (barium meal plus photofluoroscopy), direct radiology, and endoscopy. When screening for stomach cancer was started late in life, indirect radiology was the most cost-effective screening method. This analysis supports an increase in the recommended age for initiating screening for stomach cancer from age 40 to 50.
Screening for Lung Cancer
Investigators have carried out several cost-effectiveness analyses of the screening of high-risk individuals, such as current and former smokers, for lung cancer using helical computed tomography (Chirikos and others 2002; Mahadevia and others 2003; Marshall and others 2001). The results of these studies vary widely from quite favorable (US$19,000 per YLS) to extremely unfavorable (more than US$100,000 per YLS). The main reason for the wide variation in these studies is different assumptions about the clinical nature of early lung lesions detected by helical computed tomography—specifically, whether a large proportion of these small lung nodules represents "pseudo-disease" that will never progress to clinical lung cancer (Marcus and others 2000). The National Lung Cancer Screening Trial, currently under way (van Meerbeeck and Tournoy 2004), hopes to answer this question. Until results from the trial are available, no definitive statement can be made about the effectiveness or cost-effectiveness of lung cancer screening.
Screening for Colorectal Cancer
Screening methods for early detection of colorectal cancer include fecal occult blood testing, sigmoidoscopy, barium enema, and colonoscopy. Several studies of the cost-effectiveness of colorectal cancer screening in developed countries have been published (Pignone and others 2002). Table 29.2 presents estimates of the cost-effectiveness of colorectal cancer screening in the United States. Cost-effectiveness ratios for various modalities of colorectal cancer screening range from almost US$6,000 to about US$40,000 per YLS. Using models closely linked to European trials of biennial fecal occult blood testing to screen for colorectal cancer, Whynes and Nottingham Faecal Occult Blood Screening Trial (2004) report favorable cost-effectiveness ratios ranging from US$2,500 to US$4,000 per YLS. Studies of the cost-effectiveness of colorectal cancer screening in developed countries consistently conclude that such screening is cost-effective, but they do not totally agree on the relative rankings of different colorectal screening strategies (Pignone and others 2002).
[Table .]
Screening for Cervical Cancer
Cytology-based screening using the Papanicolaou smear has been the main screening method used for the secondary prevention of cervical cancer worldwide. In many low-income countries, however, cytology screening has proved difficult to sustain because of its reliance on highly trained cytotechnologists; good-quality laboratories; and infrastructure to support up to three visits for screening, evaluation of cytologic abnormalities with colposcopy, and treatment (Sankaranarayanan, Budukh, and Rajkumar 2001). Two alternative screening approaches replace the Pap smear with simple visual screening methods, such as visual inspection after application of an acetic acid solution (VIA), or with HPV DNA testing (Denny and others 2000; Sankaranarayanan and others 1999; Schiffman and others 2000; Wright 2003; Wright and others 2000; Zimbabwe Project 1999). These newer options also eliminate colposcopy, potentially allowing screening and treatment to be performed during the same visit. In middle-income countries where cytology screening is available but cervical cancer mortality has not been reduced, key questions center around improving the quality of cytology-based programs; such improvement includes having adequate colposcopy and biopsy facilities and accessible treatment (Lazcano-Ponce and others 1999); making use of HPV DNA testing technology in a cost-effective manner; and targeting the appropriate age group for cervical cancer screening more accurately. The vast majority of published cost-effectiveness analyses of population-based cervical cancer screening performed during 1980-2003 focused on high-income countries. (A list of the 39 studies reviewed is available from the authors.) The detailed results of each study are somewhat difficult to compare. The types of costs included in each study varied substantially (patient time costs and programmatic costs often were omitted), studies frequently did not discount costs and benefits or did not note the discount rate used, and sensitivity analyses were not conducted consistently on all relevant variables. Despite those limitations, several themes emerge. The incremental cost-effectiveness of screening in the general population becomes increasingly less favorable as programs are intensified by shortening the screening interval. For example, Mandelblatt and others (2002) reported that for conventional cytology and HPV testing, compared with cytology alone, the incremental cost was more than US$300,000 when conducted annually compared to US$15,400 per YLS when conducted every 10 years. Maxwell and others (2002) reported that liquid-based cytology and HPV testing for equivocal results cost US$231,300 per YLS if conducted annually incremental to 14,300 per YLS if conducted every three years. Kim Wright, and Goldie (2002) reported similar results for this same strategy (US$20,300 per YLS conducted every five years, US$59,600 per YLS every three years, and US$174,200 every two years). The analyses, which included strategies that employed both frequent screening and screening tests with higher sensitivity, often found the cost-effectiveness of frequent screening to be even less attractive. For example, Goldie, Kim, and Wright (2004) reported annual screening with combined cytology and HPV DNA testing in women over age 30 exceeded US$1 million per YLS compared to every two years. Although many analyses find that extending the age range to the very young, the very old, or both can be less cost-effective, for certain women in high-risk groups, including older, uninsured women who have never been screened, screening for cervical cancer at older ages can be cost-effective.
The analyses conducted in low-income countries focused on assessing the cost-effectiveness of an expanded set of strategies that included alternatives to conventional cytology. In addition, these analyses—unlike those in developed regions—often raised issues of feasibility, affordability, cultural context, accessibility, and equity.
In one of the earliest stochastic modeling evaluations of cervical cancer screening programs in developing regions, Sherlaw-Johnson, Gallivan, and Jenkins (1997) explored the effectiveness of cytology and HPV testing in the context of infrequent screening. They reported that the most efficient use of resources would be to concentrate cervical cancer screening efforts on women age 30 to 59 at least once per lifetime, because such blanket screening would reduce the incidence of invasive cervical cancer by up to 30 percent.
In an analysis focused on cervical cancer control in Vietnam, Suba and others (2001) reported that, because of the low direct medical costs associated with Vietnam's cervical cytology program, such a program appeared to be attractive for that country. They found that total costs to establish a nationwide Pap screening program based on five-year intervals averaged less than US$148,000 annually during the 10 years the authors assumed would be necessary to develop the program. Assuming 70 percent participation in the program, the authors found the cost-effectiveness ratio for cervical cytology screening, compared with no screening, to be US$725 per discounted YLS.
Goldie and others (2001) assessed the cost-effectiveness of several cervical cancer screening strategies in previously unscreened 30-year-old South African women. Screening tests included VIA, cytology, and HPV DNA testing. Strategies differed by the number of clinic visits required, frequency of screening and individual's age at the time of screening, and response to a positive test result. The authors found that when all strategies were considered to be equally available and were compared incrementally, HPV DNA testing was always more effective and less costly than cytology and generally more effective but more costly than VIA. They found that, in comparison with no screening, a single lifetime VIA screen at age 35, coupled with immediate treatment of women with positive results, resulted in a cost saving of US$39 per YLS as compared with a two-visit HPV, although programmatic costs were not considered. Using sensitivity analysis, the authors find the choice between using HPV DNA testing or VIA depended on the relative costs and sensitivity of the two tests and on the percentage of women lost to follow-up between the first and second visit.
Mandelblatt, Lawrence, Gaffikin, and others (2002) used a simulation model to compare seven cervical cancer screening techniques in Thailand. Comparing each strategy to the next less expensive alternative, the authors found that VIA performed at five-year intervals in women age 35 to 55, followed by immediate treatment of abnormalities, was the least expensive option and saved the greatest number of lives.
The Alliance for Cervical Cancer Prevention used primary data from studies conducted in India, Kenya, Peru, South Africa, and Thailand to develop a series of standardized, country-specific cost-effectiveness analyses. The costs and benefits associated with alternative strategies to reduce cervical cancer mortality were estimated for these five countries with different epidemiological profiles by integrating country-specific data from each site and using a standardized set of assumptions agreed on by an expert panel with experience in each country (Goldie, Gaffikin, and others 2004). In all five countries, lifetime cancer risk was reduced by approximately 25 to 35 percent with a single lifetime screen using either one-visit VIA or two-visit HPV DNA testing targeted at women age 35 to 40. Risk was reduced by more than 50 percent if screening was performed two or three times per lifetime. Although the cost of screening differed considerably between the countries, strategies were identified that, when performed two or three times per lifetime, would be considered extremely cost-effective depending on the individual country's per capita gross domestic product.
We conducted an exploratory analysis to evaluate the potential cost-effectiveness of cervical cancer screening strategies in Brazil, Madagascar, and Zimbabwe using computer-based simulation models calibrated to age-specific cervical cancer incidence and mortality in each country, along with published data. We evaluated once-in-a-lifetime screening between age 35 and 40 with (a) one-visit VIA, with screening and treatment conducted during the same visit; (b) two-visit HPV DNA screening, with HPV DNA testing during the first visit followed by treatment of screen-positive women during the second visit; and (c) three-visit cervical cytology screening, with a cytology sample obtained during the first visit, colposcopy for screen-positive women conducted during the second visit, and treatment provided during the third visit. We assumed that for the one- and two-visit strategies, women who screened positive and were eligible for cryotherapy were treated immediately, but those ineligible for cryotherapy were referred for colposcopy and diagnostic workup.
We estimated direct medical costs using data from the literature and unit costs provided by the volume editors and WHO. All costs for the analysis are presented in 2000 dollars. We estimated patients' time costs and direct nonmedical costs using our own previous work and wage estimates based on World Bank data on per capita gross national income (WHO n.d.) and wage estimate regressions developed by the U.S. Department of Commerce. Table 29.3 presents the results of our analysis.
[Table .]
Lifetime costs per individual screened are given in international dollars. Cost-effectiveness ratios are provided in U.S. dollars as well as international dollars to facilitate comparison to other studies. The available data show that cervical cancer screening conducted once, twice, or three times in a lifetime can have a significant effect on the lifetime risk of cervical cancer compared with no screening. For countries with limited resources, screening efforts should target women age 35 or older; strategies should focus on screening all women at least once in their lifetime before increasing the frequency of screening; and countries should consider alternative approaches to the conventional three-visit cervical cytology screening techniques—for example, single-visit VIA, followed by immediate treatment, or HPV DNA testing or cervical cytology followed by treatment at a second visit. Note that all screening tests may not be equally available in low-resource settings and that certain screening tests may be selected because of cultural preferences or for programmatic reasons. Implementing cervical cancer screening programs on the basis of VIA, HPV DNA testing, or cytology requires different types of resources, and the relative availability of these resources in different settings will affect the choice of strategy.
Screening for Breast Cancer
Methods for early detection of breast cancer include screening by mammography, clinical breast examination (CBE), and breast self-examination. Screening by mammography, CBE, or both may decrease breast cancer mortality, but uncertainty about the magnitude of the benefit remains because the quality of the evidence varies and results are inconsistent (Humphrey and others 2002). Recent controlled studies of organized breast self-examination programs indicate that this approach is not effective (Semiglazov and others 1999; Thomas and others 2002).
A randomized controlled trial of CBE screening for breast cancer began in Manila in 1995, but the intervention was discontinued after the first round because compliance with referral among women who were found to have a breast lump was extremely low (21 percent) and attempts to improve compliance failed. Analysis of the incidence of cancer cases in 1999 shows that the screening intervention succeeded in detecting more localized breast tumors, but the low compliance with referral and low yield of early cancers meant that the early detection program could not succeed in preventing deaths from breast cancer (International Agency for Research on Cancer n.d.).
Numerous cost-effectiveness studies of breast cancer screening programs have been conducted in developed countries (Vainio and Bianchini 2002a). Most cost-effectiveness studies of mammography screening in Europe yield cost-effectiveness ratios in the range of US$3,000 to US$10,000 per YLS, whereas those in the United States yield far less favorable cost-effectiveness ratios, ranging from US$20,000 to US$100,000 per YLS (table 29.4).
[Table .]
To investigate the potential cost-effectiveness of CBE and mammography for India, we used a microsimulation model of breast cancer screening (van Oortmarssen and others 1990). The model simulates individual life histories of disease states, and consequences of screening are calculated by comparing the histories with and without screening intervention for each individual. For our purposes, we assumed a population of 1 million Indian women with the age distribution of the country in 2000 (United Nations Population Division 2003). We assumed that the screening program would last for 25 years and would have an attendance rate of 100 percent. We expressed the effects of screening as the reduction in the number of deaths caused by breast cancer and the number of life years gained because of the screening program. Costs and effects were discounted at a rate of 3 percent.
We estimated the model's parameters using data from Dutch screening projects (Collette and others 1992; Vervoort and others 2004). We used trial results to estimate the effectiveness of mammography in reducing breast cancer mortality (de Koning and others 1995). We based sensitivity estimates of CBE on data from Rijnsburger and others (2004) and based alternative (lower) estimates on data from Bobo, Lee, and Thames (2000) and Rijnsburger and others (forthcoming). We calibrated the model so that it would correctly predict the age-specific incidence and mortality of breast cancer in India (Ferlay and others 2001) and its stage distribution at clinical diagnosis (Sankaranarayanan, Black, and Parkin 1998). Details of these methods are available elsewhere (Lamberts and others 2004).
We calculated total costs by comparing the differential costs of breast cancer screening, diagnosis, initial therapy, adjuvant therapy, follow-up, and advanced disease in the case of screening versus no screening. We calculated component costs by multiplying the estimated resource use by the estimated costs per unit for each health care input. Reliable cost data for India were limited, so we extrapolated estimates from Dutch unit costs (Mulligan and others 2003). For the analysis discussed above, we calculated costs based on a market-basket approach.
The overall incidence of breast cancer is lower in India than in Western countries. The relationship between the incidence of breast cancer and age also differs: in Western countries, the incidence of breast cancer increases with age, whereas in India, it decreases with age, beginning at age 50. Investigators have generally attributed this finding to a cohort effect: breast cancer is more common among younger cohorts than older cohorts. The stage at which breast cancer is diagnosed is much less favorable in India than in Western countries.
Table 29.5 presents the results of our exploratory cost-effectiveness analysis of various breast cancer screening programs involving CBE or mammography for a population of 1 million women in India. As the table shows, biennial CBE from age 40 to 60 costs US$2.6 million, averts 358 breast cancer deaths, prevents the loss of 4,896 life years, and has a cost-effectiveness ratio of US$522 per YLS in comparison with no screening. Biennial CBE from age 50 to 70 is less favorable in terms of cost-effectiveness: US$582 per YLS.
[Table .]
The cost-effectiveness ratios for biennial mammography screening are not as favorable as those for biennial CBE screening. Annual CBE screening results in almost the same number of life years saved as biennial mammography screening at 36 percent of the cost.
Table 29.6 shows the results of our sensitivity analysis for the exploratory cost-effectiveness analysis of breast cancer screening in India. Cost-effectiveness ratios are lower when the incidence of cancer is higher, as in Bombay. Cost-effectiveness ratios are 32 and 16 percent higher, respectively, with a lower sensitivity of CBE and when the averted costs of palliative treatment are not included. Using alternative approaches to estimate screening program costs has a major effect, resulting in cost-effectiveness estimates 6 to 11 times higher than the base case analysis. This result underlines the need for economic studies that can obtain reliable data from primary sources on the true resource costs of cancer control interventions in developing countries. With data from such studies, researchers would not have to continue to rely on extrapolating cost estimates from data in developed countries.
[Table .]
These results depend critically on assumptions about the efficacy of CBE, for which the evidence is limited, highlighting the need for controlled studies of CBE in developing countries. Our estimates indicate that the cost-effectiveness of screening mammography in India compares favorably, in absolute terms, with breast cancer screening in developed countries. Nevertheless, screening mammography for breast cancer is likely to be less cost-effective in a country such as India than is screening for cervical cancer.
Cancer Treatment and Palliative Care
Barnum and Greenberg (1993) used an indirect approach to estimate the cost-effectiveness of initial cancer treatment in developing countries. They assumed that they could estimate the effectiveness of initial cancer treatment by comparing cancer survival in the United States for the period 1975-80 with the period 1940-45. The logic of such a comparison is that major advances in cancer diagnosis, surgery, radiation, and chemotherapy occurred during the intervening period, and thus survival in the 1940-45 period could be equated to outcomes expected to result from no treatment or ineffective treatment. Barnum and Greenberg's results indicated a cost-effectiveness ratio of the following:
-
US$1,300 to US$6,200 per YLS for initial treatment of the more treatable cancers, that is, cervical, breast, oral cavity, and colorectal cancer
-
US$53,000 to US$163,000 per YLS for initial treatment of the less treatable cancers, that is, liver, lung, stomach, and esophageal cancer.
The following subsections review cost-effectiveness studies performed on selected adjuvant or palliative cancer treatments that have been studied extensively in controlled clinical trials.
Breast Cancer Treatment Interventions
The following paragraphs review studies of the cost-effectiveness of adjuvant systemic therapy for early-stage breast cancer and of radiation therapy following mastectomy and chemotherapy to treat node-positive breast cancer in premenopausal women.
T. Smith and Hillner (2000), relying on results from the Early Breast Cancer Trialists' Collaborative Group (EBCTCG 1998), modeled the natural history of breast cancer in premenopausal 45-year-old women in the United States who were diagnosed with early-stage breast cancer and treated with tamoxifen, chemotherapy, or both. Table 29.7 summarizes the cost-effectiveness of various breast cancer treatments. Smith and Hillner's cost-effectiveness estimates for single-modality systemic adjuvant therapy for breast cancer are about the same order of magnitude as Barnum and Greenberg's (1993) estimates of cost-effectiveness for initial therapy of breast cancer (about US$7,300 per YLS in 2000 dollars). Other studies (Malin and others 2002; Norum 2000) have yielded cost-effectiveness estimates for chemotherapy and hormonal therapy two to three times more favorable than Smith and Hillner's estimates. The more favorable estimates are probably the result of the investigators' use of a discount rate of 3 percent instead of 5 percent and their assumption that the benefits of treatment continue over a longer period of time.
[Table .]
Two U.S. studies (Lee and others 2002; Marks and others 1999) have estimated the cost-effectiveness ratio for radiation therapy following mastectomy and chemotherapy for node-positive breast cancer in premenopausal women to be in the range of US$22,600 to US$43,000 per quality-adjusted life year (adjusted to 2000 U.S. dollars, with a discount rate of 3 percent). Results were sensitive to treatment costs, survival benefit, and patient time costs.
The clinical trials of postmastectomy radiation on which the two U.S. studies are based compared radiation following surgery plus chemotherapy with surgery plus chemotherapy alone. Love and others (2003), however, offer observational evidence that radiation treatment may also extend survival for Chinese and Vietnamese women when administered to patients with one to three positive nodes following mastectomy alone or mastectomy combined with oophorectomy and tamoxifen. If these benefits were confirmed, postmastectomy radiation might be cost-effective in developing countries, where the cost of radiation treatment is lower than in most developed countries.
Colorectal Cancer Treatment Interventions
As concerns colorectal cancer, investigators have carried out cost-effectiveness studies on surgical techniques, adjuvant treatment, follow-up monitoring for recurrence, and treatment of advanced disease (van den Hout and others 2002). Brown, Nayfield, and Shibley (1994) estimate that the cost-effectiveness of adjuvant chemotherapy for stage three colon cancer ranges from US$3,000 to US$7,000 per YLS (adjusted to 2000 U.S. dollars, with a discount rate of 6 percent). R. Smith and others' (1993) study conducted in the Australian health care setting obtains similar results in terms of cost per YLS but yields substantially higher costs per quality-adjusted life year.
Dahlberg and others' (2002) cost-effectiveness study, which relies on cost and clinical outcome data from the Swedish Rectal Cancer Trial (1997), demonstrates that rectal cancer patients receiving preoperative radiation therapy had improved cancer-specific and overall survival rates, as well as reduced local rectal cancer recurrence rates. They estimate the overall cost-effectiveness of preoperative radiation therapy for rectal cancer patients to be US$3,654 per YLS (in 2001 U.S. dollars, using a discount rate of 3 percent). In a sensitivity analysis, which varied the rates of local rectal cancer recurrence and the survival advantage with and without radiation treatment, cost-effectiveness ratios for preoperative radiation therapy for patients with rectal cancer ranged from US$908 to US$15,228 per YLS.
Cervical Cancer Treatment Interventions
Five recent phase 3 trials indicate that a new alternative therapy—cisplatin-based chemoradiation—is more effective than standard therapy using radiation alone in the treatment of advanced cervical cancer (Rose and Lappas 2000). Using an economic model, Rose and Lappas apply unit costs to resource allocation data derived from the cisplatin-based chemoradiation arms of the five randomized trials and examine the benefits in terms of increased median survival time. Costs per YLS for cisplatin-based chemoradiation regimens varied from US$2,384 to US$28,770 on the basis of published survival and from US$308 to US$3,712 on the basis of estimated survival. Although chemoradiation for advanced cervical cancer would probably be considered cost-effective in most developed countries, analyses that take local treatment settings into account are needed to determine if this result also holds for developing countries.
Palliative Care Interventions
The most basic approach to palliative care for terminally ill cancer patients, especially in low-resource settings, involves using inexpensive oral analgesics, ranging from aspirin to opiates, depending on individual patients' needs. Unfortunately, sufficient supplies of opioid drugs for use in palliative care are often not available in developing countries because of regulatory or pricing obstacles, ignorance, or false beliefs (for more information see http://www.medsch.wisc.edu/painpolicy/index.htm and chapter 52).
Appropriate palliative care for cancer patients may involve a variety of other treatment modalities, including antiemetic drugs to relieve the side effects of chemotherapy, radiation to effect temporary tumor regression, and physical therapy to alleviate disability related to lymphedema following breast cancer surgery. Berthelot and others' (2000) study combines information from several clinical trials and Canadian treatment cost information to perform cost-effectiveness analyses of different ambulatory chemotherapy regimens used for patients with metastatic non-small-cell lung cancer to palliate symptoms and modestly improve survival. They report that vinblastine plus cisplatin resulted in both better survival and lower health care expenditures than best supportive care because it resulted in fewer episodes of rehospitalization.
Van den Hout and others' (2003) study examines the cost-effectiveness of single-fraction versus multiple-fraction radiotherapy for palliative treatment of cancer patients with painful bone metastases. They find that overall medical and social costs for single-fraction radiotherapy for palliative therapy—US$1,144 per patient in medical costs and US$1,753 per patient in total social costs—were lower than comparable costs for multiple-fraction radiotherapy, despite the higher rate of retreatment associated with single-fraction radiotherapy. Whether those results are directly applicable to radiation treatment in developing countries, where single-fraction radiation treatment may be relatively less effective, is unknown. Nonetheless, the results strongly suggest that single-fraction radiotherapy may be an acceptable, if not preferred, choice of palliative treatment in settings where resources for radiation treatment are relatively scarce and the need for palliative treatment is relatively high.
