View larger version (21K): [in this window] [in a new window]   Figure 1. Estimated direct and indirect costs (in billions of dollars) of cardiovascular disease (CVD) and stroke, United States, 2004. All estimates prepared by Thomas Thom, National Heart, Lung, and Blood Institute.

Complicating the formidable challenge of CVD is the unprecedented increase in obesity and in type 2 diabetes (5). If current trends continue, 1 in 3 Americans born in 2000 will develop diabetes sometime during their lifetimes, and this lifetime probability is even higher in racial and ethnic minority groups (6). Indeed, patients receiving a diagnosis of diabetes between the ages of 40 and 60 years lose a decade or more of life expectancy (6). Among persons with diabetes, CVD is the leading cause of mortality, morbidity, and loss of quality of life (7).

Although age-specific rates of CVD may continue their long-term downward trend, the total prevalence of CVD, like that of diabetes, is likely to increase in the future because of the aging of the "baby-boomer" generation and because of the increasing rates of obesity and diabetes (8). The percentage of the U.S. gross national product devoted to overall health care is expected to increase from 8.8% in 1980 to a projected 17% in 2011 (9, 10), placing a substantial economic burden on society. Cardiovascular disease is also becoming an increasingly severe global burden (11, 12).

Currently, it is estimated that approximately 95% of health care spending is for treatment and only 5% is targeted to earlier prevention strategies (Figure 2) (13). Too often, unfavorable social and environmental conditions support unhealthy lifestyles, often characterized by smoking; lack of regular physical activity; increased consumption of fats, salt, and carbohydrates, and inadequate consumption of fruits and vegetables, whole grains, and healthy oils. These factors, together with individual genetic determinants, challenge efforts in both primary and secondary prevention directed to CVD. However, although secondary therapeutic and behavioral preventive efforts for extant clinical illness will always be needed, more effort should be directed toward innovative and effective approaches for prevention. A recently proposed, novel preventive approach for CVD is the "polypill."

View larger version (60K): [in this window] [in a new window]   Figure 2. Action framework for a comprehensive public health strategy to prevent heart disease and stroke. The figure is split diagonally in halves, representing opportunity, not current spending, for the spectrum of preventive and therapeutic care. CHF = congestive heart failure; CVD = cardiovascular disease; DM = diabetes mellitus; HBP = high blood pressure; MI = myocardial infarction; TC = total cholesterol.

The Concept of the Polypill

In 2003, Wald and Law (14) suggested a public health approach for preventing CVD by using a polypill. The original formulation proposed for the polypill contained a statin, 3 antihypertensive agents at half doses (a ß-blocker, a diuretic, and an angiotensin-converting enzyme inhibitor), aspirin (75 mg), and folic acid (0.8 mg).

All components of the proposed polypill, except folic acid, have been shown individually to decrease cardiovascular events by 20% to 35% in secondary prevention interventions, primary prevention interventions, or both. However, Wald and Law also estimated that most of the benefit of the polypill could be retained without folic acid, with only 2 antihypertensive agents, and even without aspirin. For patients who cannot tolerate angiotensin-converting enzyme inhibitors, a formulation with an angiotensin-receptor blocker may be appropriate. Instead of 1 polypill for all populations, different combination pharmacotherapies (CPs) could be developed for different populations; for primary prevention and for secondary prevention; for different countries; or for special subgroups, such as patients with diabetes.

Wald and Law (14) proposed that a single polypill be given to all adults 55 years of age and older and to adults of any age with diabetes or CVD, regardless of risk factors. They reasoned that randomized trials with pharmacologic therapy have decreased levels of low-density lipoprotein (LDL) cholesterol and blood pressure and that aspirin blocks platelet activity, thereby substantially reducing the incidence of CVD. Observational studies have also shown that lowering levels of serum homocysteine with folic acid may reduce the incidence of CVD. Wald and Law (14) estimated the efficacy and adverse effects of the proposed polypill formulation from meta-analyses of randomized trials and cohort studies and suggested that relative to no treatment, a polypill could reduce the incidence of coronary heart disease and stroke by 88% and 80%, respectively. They concluded that "no other preventive method would have so great an impact on public health in the western world." The editor of the British Medical Journal called the issue in which the article was published "the most important BMJ in 50 years" (15). In response to the proposals of Wald and Law, the 2 major concerns that were expressed were that CP will cause people and policymakers to forgo support for healthy behaviors and lifestyle improvements and may result in undertreatment of individuals at higher risk for CVD (16).

The strategy proposed by Wald and Law (14) has at least 2 unique aspects: simultaneous intervention on 4 potentially causal risk factors for CVD (lipids, blood pressure, platelet aggregation, and homocysteine) with a single pill and intervention in the entire population at high risk on the basis of age, history of CVD, or diabetes. Wald and Law argued that because 96% of deaths from CVD occur in people older than 55 years of age, in people with extant CVD or diabetes mellitus, or in those older than 55 years of age who have extant CVD or diabetes mellitus, efforts should therefore be redirected from screening individuals for CVD risk factors or levels of CVD risk factors to population-wide treatment and primary prevention with a single pill.

Wald and Law (14) assert that serum cholesterol level, blood pressure, and serum homocysteine level individually are poor predictors of future CVD events and that no clear separation exists in the distributions of these risk factors between people who have CVD events during a specified period and those who do not. In other words, no threshold exists for the common risk factors. Thus, it would be beneficial to intervene on risk factors regardless of their absolute levels.

Workshop on Combination Pharmacotherapy

Although the polypill is new and controversial, it shares a familiar strategy with other traditional public health interventions: focusing on a common condition, applying the plan broadly to the population, addressing prevention, and emphasizing simplicity. The Centers for Disease Control and Prevention (CDC) convened a group of experts with varying areas of interest to objectively assess the pros and cons of CP and to examine approaches to rigorous scientific testing of its public health applications. This workshop involved collaboration among the CDC; the Constella Group, LLC, which provided intellectual and logistic support; and the expert panel. The workshop discussed many issues, including the role of screening for risk factors, the potential efficacy and problems with CP, how CP should be studied with clinical trials, the impact of health disparities, behavioral and lifestyle issues, and the potential of CP in the developing world. This article is a product of these discussions.

Potential Efficacy and Effectiveness

Broad implementation of CP has the potential to reach more people with greater efficiency than traditional care based on risk factors (17). Experts agree that with blood pressure and LDL cholesterol lowering, larger reductions than those currently achieved in broad populations will reduce the risk for major vascular events to a greater extent. Combining several antihypertensive drugs at low doses is likely to be more effective and have fewer side effects than high-dose therapy with a single drug (18). Several drug combinations, such as statins and cholesterol absorption inhibitors, statins and calcium-channel blockers, statins and aspirin, and 2 or more classes of antihypertensive agents, are currently on the market. Little is known, however, about interactions between drugs, which are mostly assumed to be additive (for example, 1 + 1 = 2). Subadditive effects (1 + 1 > 1 but < 2), antagonism (1 + 1 < 1), and synergism (1 + 1 > 2) are also plausible. Thus, future efficacy trials should evaluate the type and the magnitude of interactions.

Wald and Law's (14) estimates of the benefit of drug combinations may seem overly optimistic. For example, 10 mg of atorvastatin or 40 mg of simvastatin in dose-ranging studies reduced events by approximately 35% (19-22), far lower than the 61% assumed by Wald and Law (14). Similarly, aspirin trials have indicated a 25% reduction in coronary heart disease (23-25). Several clinical trials have shown that pharmacologic control of hypertension will decrease the incidence of stroke. Finally, for Alzheimer disease and vascular dementia, intervention studies of modifiable risk factors (for example, blood pressure and hypercholesterolemia) (26) have yielded mixed results (26-32). Although the incremental absolute risk reduction with subsequent medications is likely to diminish as the residual risk becomes smaller, when lifetime risk is considered (as opposed to the typical 5-year interval of clinical trials), even low-risk patients may derive substantial benefit.

Combination pharmacotherapy may also improve medication adherence. One study showed that only approximately half of persons with known CVD reported taking aspirin, and of those who believed they were taking aspirin, 15% were actually taking acetaminophen (33). Combination drugs for diabetes, hypertension, and HIV and AIDS improved adherence compared with drugs taken as separate tablets (34, 35).

Potential Problems with Combination Pharmacotherapy

Perhaps the biggest challenge with population-wide CP for primary prevention is that some people will receive therapy without benefit, whereas in the setting of primary prevention and or secondary prevention, others may not receive sufficient therapy to reduce risk factors to within current clinical guidelines. Because of the issue of whether CP is appropriate in either primary or secondary prevention strategies, these populations will need to be studied separately. Finally, the use of CP without screening for CVD risk factors is an approach to control risk factors in the population as a whole, but it may not control risk factors on an individual basis as comprehensively as screening and individualized therapy.

The true rates of side effects of CP are unknown. Wald and Law (14) estimated these rates by taking "rates attributable" to the individual drugs from clinical trials. How CP should be monitored for therapeutic efficacy and side effects (if at all) is uncertain. This question is important because it matters whether CP is prescribed or is available over the counter. That more people may obtain CP over the counter can be inferred from studies of other pharmaceuticals, such as nicotine replacement therapy, where pills previously available only by prescription became available over the counter, probably with greater reach but without greater adherence (17).

It is not clear whether the rate of discontinuation of multicomponent pills depends on the number of components. A study comparing a calcium-channel blocker with a thiazide noted that the incidence of "adverse events" was similar in the amlodipine (52.1%), chlorthalidone (52%), and placebo (59.6%) groups (36). Discontinuation rates because of adverse events were similar for both active treatment groups: 6.3% for amlodipine and 4.0% for chlorthalidone. We do not know what the discontinuation rates would be for a combination pill.

Problems with adherence are well known. Of all written prescriptions, 14% are never filled and an additional 13% are filled but never taken (37). Among patients prescribed statins, reported discontinuation rates at 1 year ranged from 15% to 60% (38). Similarly, for blood pressure medications, fill rates are approximately 50% and full adherence is only approximately 20% (39). Concomitant adherence to statins and blood pressure drugs in a managed care study was only 36% at 12 months, and adherence was better in symptomatic patients than in asymptomatic patients (40). Patients eligible for CP may be "healthy" and asymptomatic and may have high rates of discontinuation.

Many other questions must be considered. What happens when patients discontinue CP because of minor side effects? How often will this occur? Will the patient seek medical care at this point? How often will patients with contraindications, such as renal insufficiency, take CP on their own? What constitutes informed consent for use of CP? How can the key drug information be provided to patients in a concise manner?

Evaluating Efficacy

Although studies have shown that all components of CP, except folic acid, independently prevent CVD, it is necessary to establish the efficacy of the components in combination. The question is whether providing CP broadly offers an advantage over prescribing each component separately in response to elevated risk factors. To address this question, 4 groups of studies need to do the following: Evaluate the impact and tolerance of CP compared with the individual components on blood pressure, lipoproteins, and platelet function; determine whether CP improves adherence, as well as surrogate end points (for example, blood pressure and serum lipid levels) and economic outcomes; assess whether CP secondarily improves access to care in resource-poor settings (CP may be less expensive than prescribing each component separately, with a reduction in the costs of preventive strategies); and determine the impact of CP on clinical outcomes, which requires large trials of individuals or communities randomly assigned to usual care versus CP.

Several factors should be considered when designing larger efficacy trials. The first factor is the patient population. Combination pharmacotherapy could be studied for secondary prevention in patients who have had a cardiovascular event. In this population, the components will largely be indicated, so a trial comparing CP with usual care (with management titrated to aim for risk factor target levels, as appropriate) could be done with surrogate end points, safety, and health care costs as the outcomes. A primary prevention trial may be feasible in people at moderate risk (for example, men >55 years of age) in whom the component interventions would not be mandated by guidelines, with cardiovascular events as the primary outcome. Such a study would also evaluate the impact on intermediate surrogates (blood pressure control and lipid levels), adherence, safety, and cost-effectiveness.

Investigators may encounter substantial regulatory challenges when designing a trial. For instance, the most appropriate design may be to compare CP with usual care rather than comparing CP with a placebo. This change from the classic placebo-controlled design requires an evolution in thinking.

A major efficacy trial could be done with random assignment of individual patients. It would also be possible to randomly assign several hundred communities (if over the counter) or medical practices (if by prescription) to CP versus usual care, thereby allowing assessment of efficacy and effectiveness. A meta-analysis of several small trials addressing similar questions could also be valuable.

Demonstrating Effectiveness

Because the proposal is to implement CP on a broad population basis, questions concerning generalizability, implementation, and adoption in real-world circumstances are critical. Estimates of efficacy from narrowly based randomized trials may not be easily generalizable because of differences in populations, availability of resources, adherence, costs, and implementation issues. Thus, the true benefit of CP, in addition to assumptions about additivity of effects (multiplication of relative risks), may be lower than the estimate by Wald and Law (14) because their assumptions about risk reduction were derived from selected groups of volunteers. Some experts have argued that efficacy trials should be conducted under real-world conditions, whereas others maintain a more traditional view that efficacy trials should precede effectiveness studies. The expert panel agreed that the first study should consider efficacy. A recommendation such as the use of CP has such important ramifications for cost, risk, and potential benefits that the standard of evidence should be high. Either a commitment should be made to pursue efficacy and effectiveness studies, or a hybrid study that incorporates a diverse, heterogeneous study sample should be considered.

Establishing Cost-Effectiveness

For CP to gain wide acceptance, it will need to show value and efficacy; thus, clinical and population trials should include economic analyses (41). Several issues should be considered in economic evaluations of CP. First, cost-effectiveness evaluates absolute change rather than relative change. Thus, a therapy with a 20% relative effect will be much more cost-effective if the standard therapy is associated with a 25% absolute event rate rather than a 5% absolute event rate (42). In the former case, the rates of events are decreased by 5%, but in the latter case, the rates of events are decreased by just 1%, at the same cost. This issue leads to the conclusion that secondary prevention is generally more cost-effective than primary prevention. However, it is unreasonable to focus on preventing CVD only in people who have already had a CVD event, because the consequences of first events can be irreversible or fatal. Discounting future costs and events decreases the cost-effectiveness of prevention, where costs are incurred now to prevent future events. In addition, economic evaluations generally take a societal perspective, and it is not always the payer who derives the benefit (43). The "rule of rescue" is also a problem; for example, we will spend anything to save the little girl who fell down the well, but we will not spend the money to build fences around wells (42). Thus, people's own interests and attitudes can limit investment in prevention (44).

Social and Behavioral Issues

For CP to achieve full benefit, patients must take it reliably and indefinitely and should not abandon heart-healthy lifestyles. Unlike the predictable patterns of medication use, the uptake, adherence, behavioral, and lifestyle issues for such a public health intervention are uncertain. Long-term use of CP will require numerous steps to evaluate the reach, efficacy, adoption, implementation, and maintenance of this therapy (45). Many barriers must fall for the polypill to have the large impact suggested by Wald and Law (14).

Adherence will probably be higher than it would be if all of the components needed to be taken as individual pills (35), but the absolute rate of side effects and symptoms from previous trials, not just the increment in side effects compared with placebo, will be a better predictor of population response because users are likely to attribute all symptoms to CP.

The best evidence available from which to extrapolate a pattern of adoption, adherence, and consequent impact (reach x effectiveness of adoption and adherence) could be the experience with tobacco control when nicotine replacement therapy moved from prescription to availability over the counter (17). The effectiveness of this therapy remained 14%, but the reach increased from 2.5 million to 6.3 million smokers. This reach-times-effectiveness increased successful attempts at smoking cessation from 350 000 with physician counseling and prescription to 882 000 without recourse to physician screening, counseling, and monitoring.

Research on adherence to therapeutic and preventive medications indicates that cost is the most frequent reason for lapses in maintaining either type of regimen (46). Unanticipated side effects are also important. Strategies to improve adherence could include cost control or subsidy for the drugs and anticipatory counseling about potential side effects, the benefits of adapting to less severe side effects, the possibility of reducing the dosage or changing the regimen if side effects are intolerable, and minimizing the cost. How these goals can be accomplished with widespread use is uncertain.

The simplicity of administering a multiple-component pill to an at-risk population might reduce the gap between treatment guidelines and actual practice. In doing so, CP may change the current health care model, in which physicians identify, treat, and monitor, to a public health model, in which population-based delivery relies on nonphysician practitioners (47). In this model, physician expertise would be reserved for people who cannot tolerate CP prevention or those who do not respond to CP prevention. The challenge will be to avoid undertreatment for people who truly require intensification to reach desirable levels of risk factors.

Health Care Disparities and Combination Pharmacotherapy

Cardiovascular disease is the leading cause of death for all ethnic groups (48), but socioeconomic and racial and ethnic disparities exist for control of risk factors and outcomes in the United States (49-51). Non-Hispanic black and Mexican-American persons, for example, have poorer control of blood pressure than other groups (52). Members of minority groups are more likely to be uninsured, are more likely to have poor access to care (53), and are also more likely to be generally underrepresented in clinical trials. Combination pharmacotherapy may help improve health care delivery for underserved people. Studies evaluating CP must actively recruit racial and ethnic minority groups and people with low socioeconomic status, and specifically attend to cultural sensitivity, strategies to promote health literacy, adherence to medications, and attendance at follow-up examinations.

Combination Pharmacotherapy in Developing Countries

The number of people with CVD is increasing faster in developing countries than in the developed world, where CVD also occurs earlier in life (11, 12). The INTERHEART study has suggested that 9 modifiable risk factors account for 90% of the risk for myocardial infarction in all populations in developing countries and developed countries (54). Therefore, interventions that reduce risk factors should have similar benefits across all countries and ethnic groups.

In the developing world, the major concern is affordable interventions. Combination pharmacotherapy may thus be of great interest in the developing world if it can reduce costs. It may, however, be necessary to target CP to those people at highest risk, in whom cost-effectiveness will be highest (for example, secondary prevention and high-risk primary prevention). In much of the developing world, risk differs between rural and urban areas. In urban Delhi, India, for people older than 55 years of age, 22% have more than a 20% risk for CVD over 10 years, whereas in rural Haryana, India, only 8% have more than a 20% risk (Reddy S. Personal communication). Countries may also differ in the burden of coronary heart disease and stroke or in the incidence of hemorrhagic and thrombotic strokes. Combination pharmacotherapy may have to be customized to include different components for different countries.

Developing countries may manufacture and distribute variations of CP without waiting for the developed world. Clinical trials focusing on bioavailability, safety, tolerability, and adherence may happen soon. The price is expected to be low, aiming at large volumes, to make CP available to those people who need it the most.

We think CP offers the potential to decrease the incidence of CVD worldwide, perhaps especially in people who have never had a cardiovascular event. An editorial welcomed the development of CP but questioned its use for global CVD prevention (55). This expert panel believes that the concept of CP shows sufficient promise to justify the additional scientific testing of its potential public health applications. Specifically, we recommend further evaluation to address the following questions: Which components should be included in 1 or more formulations of CP? What can be determined about the safety, efficacy, and effectiveness of CP using surrogate and hard CVD end points? What characteristics can identify suitable subpopulations for evaluation of CP? Can CP be realistically considered for primary prevention and secondary prevention? What is the cost-effectiveness of CP in primary prevention and secondary prevention, and in disparate ethnic and socioeconomic populations within the United States and internationally? What is the role of screening for risk factors in the context of a population-based CP strategy? What is the benefit versus risk for over the counter availability of CP? What impact will CP have on healthful behaviors and associated prevention programs? Can strategies be developed that promote a healthful lifestyle among people who take CP? What are the optimal roles of physicians and nonphysician practitioners in the delivery of CP?

We hope this paper stimulates an international debate about the pros and cons of the polypill (15), and an associated spectrum of research proposals that address the issues noted.

Appendix

Participants in the Combination Pharmacotherapy and Public Health Research Working Group

Planning and Coordinating Group

Lawrence Green, DrPH, University of California at Berkeley (Berkeley, California); William S. Weintraub, MD, Emory University (Atlanta, Georgia); K.M. Venkat Narayan, MD, MPH, MBA, David F. Williamson, PhD, Michael Engelgau, MD, MS, Frank Vinicor, MD, MPH, Centers for Disease Control and Prevention (Atlanta, Georgia); Andrew Friede, MD, MPH, Constella Group, LLC (Durham, North Carolina).

Expert Panel

Virgil Brown, MD, Veterans Administration; Jeffrey A. Cutler, MD, MPH, National Heart, Lung, and Blood Institute (Bethesda, Maryland); Brent Eagan, MD, Medical University of South Carolina (Charleston, South Carolina); Russell Glasgow, PhD, MS, Kaiser Permanente of Colorado; Philip Greenland, MD, Northwestern University; Richard Grimm, Jr., MD, PhD, Berman Center for Outcomes & Clinical Research (Minneapolis, MN); Harlan M. Krumholz, MD, MSc, Yale University (New Haven, Connecticut); Thomas Marciniak (by telephone), MD, Food and Drug Administration (Rockville, Maryland); David Maron, MD, Vanderbilt University (Nashville, Tennessee); David M. Nathan, MD, Harvard University (Cambridge, Massachusetts); Elizabeth Ofili, MD, MPH, Morehouse School of Medicine (Atlanta, Georgia); Thomas Pearson, MD, MPH, PhD, University of Rochester (Rochester, New York); Srinath Reddy (by telephone), MD, All India Institute of Medical Sciences (New Delhi, India); Salim Yusuf, DPhil, Hamilton General Hospital; Peter W.F. Wilson, MD, Medical University of South Carolina (Charleston, South Carolina).

Other Attendees

Carl Caspersen, PhD, MPH, CDC (Atlanta, Georgia); Lawrence Fine, MD, MPH, National Heart, Lung, and Blood Institute (Bethesda, Maryland); Linda Giess, CDC (Atlanta, Georgia); Cynthia Mulrow, MD, MSc, University of Texas, Health Science Center of San Antonio (San Antonio, Texas); Steven Sorensen, PhD, CDC (Atlanta, Georgia); Myrlene Staten, MD, National Institute for Diabetes and Digestive and Kidney Diseases (Bethesda, Maryland); Elizabeth Yetley, PhD, MS, National Institutes of Health (Bethesda, Maryland).

Support Staff

Beverly Jean Cambridge, Kristen Copes, Stacie Deaton, Andrew Hammond, Billie Lester, Gina Lewis, Shelley Sheremata, Ikeysha Tucker, Constella Group, LLC (Durham, North Carolina).