Cost-Effectiveness of Interventions
Costs include expenditures required to identify and treat risk factors as well as expenditures for treating CVD when it is not prevented. Where possible, this chapter deals with the separate sources of costs for several reasons. First, the costs for identifying those requiring treatment vary significantly by level of economic development and by urban versus rural location. In many situations in developing countries, such costs will make most or all forms of screening beyond a determination of CVD history unaffordable. Second, in some countries, such as India, that are large producers of generic drugs, prices are reported to be lower than in most other drug-producing or -importing countries. Third, this approach allows researchers and policy makers to understand the constituent costs so that they can examine where cost reductions may be most beneficial. Fourth, it clarifies what expenditures may be required as a result of changes in decisions about the treatment of risk factors. Finally, many people in developing countries do not have access to hospitals for acute management of CVD events. Nonetheless, increased expenditure on treating risk factors may lead to significant reductions in the costs of treating subsequent CVD events for many countries. Ultimately, the net effect is reflected in cost-effectiveness analyses. Unless otherwise stated, costs are in 2001 U.S. dollars.
The costs of personal interventions include the costs of patient screening (identifying high-risk patients), drugs and their acquisition, clinic visits, health care workers' time, laboratory tests, and travel. Annual drug costs for medications to lower blood pressure and cholesterol vary widely by country and depend on whether generics are available and used. For example, according to the International Drug Price Indicator Guide (Management Sciences for Health 2004), annual costs in 2002 of generic 40 mg lovastatin ranged from US$14 in Barbados to US$217 in Costa Rica, and on-patent statins can cost almost a US$1,000 a year in the United States. Because drug costs vary by up to two orders of magnitude across countries, results of cost-effectiveness analyses are particularly sensitive to their input prices. Table 45.3 presents some sample prices. The costs of these medications have dropped considerably in recent years, and now the annual costs for hydrochlorothiazide (25 mg), atenolol (50 mg), and captopril (50 mg), are US$2, US$4, and US$9, respectively (Management Sciences for Health 2004). Statins will become increasingly affordable as simvastatin joins lovastatin in coming off patent (2006 in the United States and already off patent in Germany and the United Kingdom).
[Table .]
The estimated number of visits to manage high blood pressure and cholesterol, under traditional paradigms, ranges from two to six per year at costs ranging from US$3 to US$20 per visit across the six regions assessed, but note that generally many fewer tests and less follow-up is required with a strategy based on absolute risk. Diagnostic testing for cholesterol in the United States using a general laboratory is reimbursed at US$6 for total cholesterol, US$16 for a complete lipoprotein cholesterol fractionation analysis, and US$6 for triglycerides (Xact Medicare Services 2003). Point-of-care one-step enzymatic strips that require only a few drops of blood from a finger stick and that can process total cholesterol in minutes cost less than US$3 per test (Greenland and others 1987). A basic metabolic panel for those on diuretics or for measuring renal function is US$12. The costs attributed to patient time and travel for visits have not been estimated for many countries, but they were recently estimated at US$12 to US$26 per visit in the United States, depending on age and sex (Prosser and others 2000).
A review of studies to date highlights several issues regarding cost-effectiveness analyses, including the significant variations in terms of calculations of cost per life year saved. The two most important aspects of the cost-effectiveness of any primary intervention are the future risk for CVD of the population treated and the costs of the medications.
Population-based Interventions
Given the strong association between CVD and high blood pressure, cholesterol, and body mass, most guidelines for those risk factors begin by recommending lifestyle modifications. Although these benefits can lead to changes in risk factors, their effect on CVD events is not well documented. However, on the basis of assumptions about cholesterol and blood pressure reduction from population-based lifestyle education programs and given the relatively low cost of the interventions—US$5 to US$17 per person per year (Tosteson and others 1997)—the cost-effectiveness of such programs may be reasonable. However, the cost-effectiveness ratios of these interventions were sensitive to the cost of the intervention as well as to the expected reduction in the risk factor. For example, a community intervention that expects a 4 percent reduction in total serum cholesterol and costs US$5 per person annually targeted would save more than US$2 billion over 25 years of the program. When the North Karelia (Puska 1999) estimates were used in a cost-effectiveness analysis in the United States (Tosteson and others 1997), the cost-effectiveness ratios ranged from being cost saving to US$88,000 (in 1985 U.S. dollars) per life year saved, depending on the percentage reduction in cholesterol (1 to 4 percent).
Personal Interventions to Lower Blood Pressure or Cholesterol in Developed Countries
A common finding of cost-effectiveness analyses of primary prevention of CVD by means of lowering blood pressure and cholesterol is the wide variability in cost-effectiveness ratios, depending on underlying risk, age, and costs of medications. For personal interventions using drug treatment for lowering blood pressure and cholesterol levels, no single cost-effectiveness analysis adequately summarizes experience in the developed countries. Collectively, the studies evaluating hypertension treatment in Australia, New Zealand, the United States, and the Scandinavian countries suggest a range of cost-effectiveness ratios from US$4,600 to more than US$100,000 per life year gained when applied to the entire adult population without further risk stratification (Kupersmith and others 1995). Compared with the entire population, for those at high risk with diastolic blood pressures over 105 mmHg and older than 45, hypertension treatment can cost as little as a few hundred dollars per life year gained or can even be cost saving (Johannesson and others 1991).
Investigators have reported that primary prevention with cholesterol-reducing medications is less attractive overall than other interventions, such as hypertension treatment, from a cost-effectiveness perspective, although once again this finding is likely to differ considerably now that statins are off patent. Reported cost-effectiveness ratios have ranged from US$10,000 to US$2 million per life year gained (Hay, Yu, and Ashraf 1999), whereas dietary interventions for cholesterol reduction are more favorable, with ratios of around US$2, 000 per quality-adjusted life year (QALY) (Prosser and others 2000). For cholesterol treatment, Prosser and others (2000) find cost-effectiveness ratios of US$50,000 per QALY for on-patent statins among those at highest risk (high cholesterol levels and multiple risk factors) and up to US$1.4 million per QALY among low-risk females when compared with dietary strategy alone. The cost per life year gained in the primary prevention trial of the West of Scotland Coronary Prevention Study among high-risk individuals treated with pravastatin was about US$30,000 (Caro and others 1997). Using the same criteria, Downs and others (1998) find that the cost per life year saved in the Air Force/Texas Coronary Atherosclerosis Prevention Study cohort with average cholesterol levels was more than US$100,000. In general, younger and older age groups tend to have the least favorable cost-effectiveness ratios. For younger groups, this finding probably reflects their overall lower risk and the many years of treatment required before realizing a benefit. For the elderly, high cost-effectiveness ratios may reflect other competing causes of death and the delay of up two years between treatment and benefit seen in most primary prevention trials.
Personal Interventions to Lower Blood Pressure or Cholesterol in Developing Countries
No trials of blood pressure, cholesterol, or body mass lowering have been conducted solely in developing countries. As a result, we have derived assessments of cost-effectiveness by extrapolating from results in developed countries presented earlier. Goldman and others (1991) report that a decline in the cost of lovastatin by 40 percent, once generic, would result in a roughly 30 percent reduction in the cost-effectiveness ratio. However, this finding does not take into account that both the underlying epidemiology and the costs can be quite different across and within countries and regions.
Murray and others (2003) compare 17 nonpersonal and personal health service interventions or combinations of interventions in the 14 epidemiological subregions defined by the World Health Organization (WHO) as part of its Choosing Interventions That Are Cost-Effective (WHO-CHOICE) initiative. The nonpersonal interventions included health education through the mass media and legislative efforts to reduce salt intake, improve blood pressure generally, and reduce cholesterol and obesity levels. The personal interventions included treatment with statins of those above two different cholesterol-level thresholds (greater than 6.2 mmol/l or greater than 5.7 mmol/l), treatment with beta-blockers and diuretics of those above two different hypertension thresholds (greater than 160 mmHg or greater than 140 mmHg), and treatment of individuals with both hypertension and increased cholesterol with all three medications. Finally, the effects of combination treatment with a beta-blocker, diuretic, statin, and aspirin were modeled for four groups defined on the basis of absolute risk (10-year probability of a cardiovascular event of 5, 15, 25, or 35 percent).
Intervention effects were based on systematic reviews of randomized trials or meta-analyses. Population health effects caused by the interventions were based on stochastically simulating populations on the basis of age, sex, and risk factor distribution of smoking, hypertension, cholesterol, BMI, and smoking in the 14 subregions, both with and without the various treatments to determine the effect. The effects of the intervention were then translated into DALYs using a standard multistate modeling tool. Costs include both program-level costs (media, training, and administration) and patient-level costs (medicines, health care visits, diagnostic tests). All costs were based on a standard ingredients approach and on regional estimations. The costs of CVD events were not included.
The results are summarized in table 45.4. The incremental cost-effectiveness ratios for the strategy assessing absolute risk and using the triple combination of beta-blocker, statin, and aspirin with or without the addition of health education and salt legislation ranged from US$138 per DALY saved (absolute risk greater than 35 percent) in the Africa E region to US$4,319 per DALY saved (absolute risk greater than 5 percent) in the Latin America and the Caribbean B region. These estimates are in international or purchasing-power parity dollars (see chapter 15 for an explanation). Table 45.4 also shows the approximate equivalent costs in U.S. dollars and explains the conversion from the WHO-CHOICE estimates. The nonpersonal interventions, including efforts to reduce salt intake in processed foods, were less costly than the personal interventions. Personal interventions based on treatment guidelines were cost-effective; however, when the strategies for treating high cholesterol or hypertension were compared with the absolute-risk approach, they were not favorable and were dominated by the latter, meaning that the absolute-risk approach of treating those with a greater than 35 percent risk averted more DALYs and cost less than either the blood pressure or cholesterol strategies. For an example of a country-specific analysis, see box 45.1.
[Table .]
[Box 45.1]
Several recent publications have suggested that combination treatments of medications to lower blood pressure, statin, aspirin, and perhaps other agents such as folate could more than halve cardiovascular risk (Wald and Law 2003; WHO 2002; Yusuf 2002). This suggestion is especially relevant for developing countries, given that suitable components are all now off patent. Good evidence indicates that single-pill combinations increase adherence to drug regimens (Connor, Rafter, and Rodgers 2004) and reduce supply and transport costs. We used a Markov model to evaluate the cost-effectiveness of such a hypothetical pill or combination packaging of the individual medications. We modeled the effect of a pill that included half of the standard doses of hydrochlorothiazide, atenolol, lisinopril, lovastatin, and aspirin on overall morbidity and mortality in treating those without prior CVD. We did not include folate because no randomized trials had shown that it reduced CVD events at the time of the analysis. The assumptions of the relative risk reductions were based on those of Wald and Law (2003). The strategies compared were for treating various high-risk populations (absolute risk for CVD greater than 15, 25, and 35 percent over 10 years. We applied the model to a population of 1 million adults over the age of 35, with the costs and benefits seen from a societal perspective and with the intervention run for 10 years. We calculated estimates for one representative country from each region where demographic and risk factor data existed. Unlike the WHO-CHOICE analysis, this analysis separated use of the intervention according to those with and without established CVD. Table 45.5 presents the results.
[Table .]
Table 45.6 shows the breakdown of events averted and costs for India. Even though the absolute numbers differ for other countries, the relative differences between the different groups receiving the "polypill" compared with the groups not receiving it are similar for all countries. Although the total costs for treating lower-risk patients increase, so do the benefits, and the overall incremental cost-effectiveness ratio remains relatively favorable. The proportion of costs shifts away from those attributable to hospital care when no primary prevention is initiated to costs attributable to ambulatory care and pharmaceuticals when more lower-risk patients are treated.
[Table .]
Interventions to Reduce Bodyweight
No large-scale randomized trials of weight reduction as an isolated intervention are available on which to base estimates of the benefits of weight loss in lowering the risk of coronary heart disease. Thus, costs per life year saved would have to be modeled to project benefits. In one such analysis, a school-based educational program to reduce obesity among middle school students reported a cost of US$4,300 per QALY (L. Wang and others 2003). However, this analysis assumed that the weight loss would be maintained throughout adulthood, but the high relapse rates found in weight reduction studies do not bear out this assumption (Serdula and others 1999). Further research is needed to evaluate the benefits of weight reduction in relation to reducing CVD events and the long-term sustainability of weight loss before reliable cost-effectiveness estimates can be made.
Distributional and Equity Consequences
Failing to translate available evidence from industrial countries about CVD prevention strategies into practicable solutions for developing countries would have clear equity implications, especially when CVD is a large and growing problem in developing countries and when safe and effective interventions that were once extremely expensive are now available for a few cents a day. Because access to cardiovascular health care in developing countries often depends on patients' ability to pay, the poor would stand to benefit the most from a low-cost intervention such as a polypill.
Some see CVD as exclusively a disease of the affluent in developing countries; however, in many developing countries, the transition of CVD to becoming a disease of the poor has already begun—a transition already seen in developed countries around the world. A recent analysis of the distribution of major cardiovascular risks by poverty levels has shown that many cardiovascular risks already affect the poor in the world's poorest countries (Ezzati and others 2004; WHO 2002). Combating the trend requires highly effective, low-cost solutions relevant for most or all of those at risk in developing countries, in contrast to the investments in high-tech treatment interventions that have commonly occurred to date.
