This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

F U T U R E  E L D E R L Y
I N N O V A T I O N  

26 September 2005  
  

Consequences Of Health Trends 
And Medical Innovation For 
The Future Elderly

When demographic trends temper the optimism of
biomedical advances, how will tomorrow’s elderly fare?

By Dana P. Goldman, Baoping Shang, Jayanta Bhattacharya,
Alan M. Garber, Michael Hurd, Geoffrey F. Joyce,
Darius N. Lakdawalla, Constantijn Panis, and Paul G. Shekelle

ABSTRACT:

Recent innovations in biomedicine seem poised to revolutionize medical practice. At the same time, disease and disability are increasing among younger populations. This paper considers how these confluent trends will affect the elderly’s health status and health care spending over the next thirty years. Because healthier people live longer, cumulative Medicare spending varies little with a beneficiary’s disease and disability status upon entering Medicare. On the other hand, ten of the most promising medical technologies are forecast to increase spending greatly. It is unlikely that a “silver bullet” will emerge to both improve health and dramatically reduce medical spending.

The consequences for human health of recent breakthroughs in fundamental biology—including the landmark sequencing of the human genome—are impossible to predict with any certainty. However, the biomedical community appears confident that unprecedented advances in our ability to prevent, detect, and treat disease are within reach. If so, there will be striking improvements in population health and, if one believes the most optimistic forecasts, a concomitant decrease in the total resources devoted to medical care.

This optimism must be tempered by some pressing demographic trends. The number of Americans age sixty-five and older is projected to double by 2030. And the past few decades have witnessed alarming increases in obesity and diabetes among the young.¹ Disability rates for the young have risen within all demographic and economic groups. The volume and intensity of health services are also rising rapidly.² All of these trends could greatly increase health care spending.³

In this paper we consider how new medical advances will affect health and health care delivery for the elderly of the future. The type of innovation is important, and we consider several. Those that prevent disease—heart disease, diabetes, cancer, and Alzheimer’s are potentially the most promising—could protect large portions of the population and forestall costly complications. Better treatments for existing disease might also be developed, drawing on advances in gene therapy and bioengineering. But medical technology alone will not determine future outcomes; demographics and health trends also play a key role. So we also consider the implications for spending as successive cohorts age into Medicare.

Model Of The Elderly’s Health And Spending

We developed a demographic and economic model to predict costs and health status for the elderly. The future elderly model (FEM) is a microsimulation that tracks elderly, Medicare-eligible people over time to project their health conditions, their functional status, and ultimately their Medicare and total health care spending. Much more detail is provided in an online technical appendix; below we summarize the salient details.⁴

The FEM begins with a representative sample of approximately 100,000 aged Medicare beneficiaries from the 1992–1999 Medicare Current Beneficiary Surveys (MCBS) (Exhibit 1).⁵ We predicted health care spending for everyone in this cohort starting in 2000. These predictions came from pooled weighted least squares regressions of total health care spending on risk factors, self-reported conditions, functional status, and interactions of conditions and functional status.

We aged our cohort by simulating health and functional outcomes in the subsequent year. This process required knowledge of the underlying risk of changing health. We used the MCBS data to estimate the one-year probabilities (hazards) of dying, developing a new health condition, or entering a new functional state (Exhibit 2). Each hazard depended on risk factors, other health conditions and functional status where clinically warranted, and age. We treated all health conditions as “absorbing”—that is, once people got an illness, they had it forever and therefore could not get it again—and modeled transitions into these conditions. This assumption was consistent with the way the data were obtained (“Has a doctor ever told you…”) and with the course of most chronic diseases. Based on these hazard models, we then predicted each person’s probability of dying, getting a new disease, or entering a new functional state, using Monte Carlo techniques.

As our initial sample aged, it became less representative of the Medicare population. We annually replenished our sample through 2030 with a new cohort of sixty-five-year-olds using data on the health of younger cohorts from the 1982–1996 National Health Interview Surveys (NHIS) to predict the health of new Medicare entrants. For example, the health of sixty-five-year-olds in 2026 will depend on the health of thirty-five-year olds in 1996, appropriately trended. We then used this model to simulate the consequences of recent health trends.

Consequences Of Recent Health Trends

The health of the elderly has been improving in important ways since the early 1980s.⁶ However, the rising prevalence of diseases such as obesity and diabetes among the young and increases in disability suggest that future cohorts entering Medicare may be less healthy.⁷ The net effect of these trends is unclear. Improvements in health will allow the elderly to live longer and accrue more expenses and, as it is sometimes argued, ultimately incur more health care costs. The issue is further complicated because the effects depend largely on the mix of disease and disability, since not all conditions are equally expensive to treat over a lifetime.

We used our model to untangle these effects in three scenarios of the health of future entrants into Medicare. Scenario A, which is our preferred estimate, forecasts the constellation of disease, taking into account all of the information at our disposal, especially the health of the younger cohorts observed in the NHIS.⁸ Scenario B assumes that the entering cohorts will have a constellation of diseases and disabilities similar to the healthy cohorts observed in the 1990s. This scenario ignores what we know about the disease and disability in younger populations. Scenario C assumes continued improvement in preventing disability among the entire elderly population and the entering cohort.⁹

Exhibit 3 summarizes the differences among these scenarios using any limitations in activities of daily living (ADLs). Such limitations are important markers for health care spending, the likelihood of entering a nursing home, and mortality. Under our preferred scenario (A), by 2030, 27.3 percent of the elderly will be disabled. If one (naïvely) assumes that the entering cohort to Medicare always resembles recent entrants (Scenario B), then disability follows a similar pattern but ends up at 25.9 percent in 2030. This tells us that the demographic profile of the population—in particular, the baby boom—is driving disability more than the age profile of disease is. Under the strong assumption of continual improvement in disability among all elderly (Scenario C), disability rates will be 5.1 percentage points lower than the baseline in 2030.

Lower disability rates should translate into lower health care costs (Exhibit 4). Under Scenario A, per capita spending will be $8,759 per beneficiary in 2030 (measured in 1999 dollars). Under the most favorable assumptions (Scenario C), spending will be 8 percent lower by 2030 ($8,032 per beneficiary). However, any improvement in disability will also lead to mortality improvements and hence a larger elderly population. Exhibit 5 shows the impact on total spending by the elderly. There is little difference between Scenarios A and B; by 2030 spending under Scenario B is only 2 percent ($11.5 billion) less per year. So although the increasing burden of disease among the young might be a public health problem, it is not a trend that is likely to affect Medicare consequentially. Even in the best case, the potential savings are modest. By 2030, total spending would be $583 billion for Scenario C, compared with $621 billion for Scenario A, for a savings of 6 percent. The savings are lower for total spending (6 percent) than for per beneficiary spending (8 percent) because the attendant mortality improvement would result in approximately 1.7 million more elderly people by 2030. Thus, much as James Lubitz and colleagues found, cumulative Medicare spending is largely invariant to beneficiaries’ health status when they enter Medicare because healthier people live longer.¹⁰

Consequences Of Key Medical Technologies

The previous scenarios assume no medical innovations of consequence; medicine continues to be practiced as it was during 1992–1999, the years spanned by our Medicare data. Total spending rises only because of demographic forces and because of underlying changes in the health of the population. This raises the question: What impact would new medical technologies have on these trends?¹¹

Key technologies. To identify the technologies, we conducted systematic literature searches and then elicited consensus from several panels of distinguished experts. This process identified thirty-four technologies most likely to affect the health of the future elderly.¹² These innovations are classified into three clinical domains: cardiovascular disease, neurological disorders, and cancer and the biology of aging. For this analysis, we chose ten technologies most likely to be widely adopted: three addressing cardiovascular disease, three addressing cancer, two addressing neurological disease, one addressing diabetes, and one related to general aging. A more detailed technical report describes how these technologies were selected and summarizes the evidence on efficacy driving the assumptions below.¹³

Intraventricular cardioverter defibrillators (ICDs). These devices can be implanted in the heart to continuously monitor the heart rhythm and apply a therapeutic shock when life-threatening arrhythmias are detected. This simulation greatly expands this technology to primary prevention: Half of patients with either heart failure or acute myocardial infarction (AMI) would receive a surgical implantation at a one-time cost of $37,500 (1999 dollars). We assumed no annual maintenance costs; this generous assumption is partially offset by not allowing the cost of implants to decline over time, as one might expect with a mature technology. Patients with an ICD were assumed to have a 10 percent lower mortality rate.

Left ventricular assist devices (LVADs). These devices, similar to “artificial hearts,” are implanted in the chest to aid the left ventricle of the heart in pumping blood. This is a technology traditionally used as a bridge to heart transplantation; we modeled improvements to allow permanent implantation. About 10 percent of patients with heart failure would receive an implant at a cost of $120,000 and with subsequent mortality improvement of 15 percent. As with ICDs, we assumed no maintenance costs as a partial offset to no decline in the device’s cost over time.

Pacemakers to control atrial fibrillation. Atrial fibrillation is a disturbance of the heart rhythm that is common in older people and contributes to both heart failure and stroke. Our panel considered several possible breakthroughs for improved control: new generations of pacemakers or defibrillators, use of a catheter to interrupt the pathways by which disordered electrical currents are maintained, and new drugs. Drugs were considered unlikely candidates to dramatically improve outcomes for this condition. Consequently, we considered the adoption of pacemakers to treat this condition. All patients with chronic or paroxysmal atrial fibrillation would receive a pacemaker at a cost of $30,000. The risk of stroke would be reduced 50 percent.

Telomerase inhibitors. Telomerase inhibitors are molecules that prevent the expression of telomerase, an enzyme that allows cancer cells to replicate without limit. This treatment would target cancer patients with solid tumors. Our expert panels predicted that telomerase inhibitors could treat half of patients with local disease and one-tenth of those with disseminated disease (Exhibit 6). Using information on the types of cancer and the rate of metastasis, we assumed that approximately one-fourth of all cancer patients would be treated. The cost of this treatment would be similar to an antiretroviral HIV drug—$177 per month—and would be taken for the rest of one’s life. Half of patients would be cured, and the other half would have the effects of cancer mitigated, with the one-year probability of disability, nursing home entry, and mortality reduced by 25 percent.

Cancer vaccines. Attempts to stimulate the body’s immune system to fight cancer cells (analogous to vaccines to prevent viral disease) have been ongoing for more than twenty years. Active, nonspecific immune stimulants successfully treated bladder cancer and show promise for melanoma and renal cell carcinoma. Many vaccines directed against a tumor-associated antigen—and to which the host will respond—are in clinical trials. For this simulation, we used the same assumptions regarding cancer types and prevalence as in the telomerase inhibitor simulation. Treatment rates would vary by the type of cancer and whether it has metastasized (Exhibit 6). Half of patients with local disease would get treatment (21 percent of all cancer patients); 100 percent of patients with disseminated disease (41 percent of cancer patients) and 100 percent of patients with other cancers (18 percent of cancer patients) would get treatment. Because it is a vaccine, we assumed that patients would need only three doses at a cost of $195 apiece; this price is three times as much as for a hepatitis vaccine. Melanoma and renal cell carcinoma cases would be cured (2 percent of all cancer patients), and the other treated patients would have a 25 percent improvement, similar to that of noncured patients on telomerase inhibitors.

Anti-angiogenesis. These treatments use human antigrowth factors to inhibit the development of new blood vessels, a prerequisite for the growth of cancer masses beyond about 1 cm in size. Infusions (or injections under the skin) would be given as an adjuvant to existing therapy for 40 percent of patients with solid tumors (Exhibit 6). The therapy—similar to taking bevacizumab for colorectal cancer—would cost $4,800 per month. We assumed that patients would get the treatment for one year initially. At that time, 30 percent of patients would be determined to be responsive and would continue to receive the treatment for four more years. Therapy was assumed to be perfectly effective for these patients; that is, it would shrink the tumor to such a small size that the patient would be effectively cured. The remaining 70 percent would be nonresponsive and go off the treatment.

Treatment of acute stroke. It may be possible to limit the disability following acute stroke by decreasing the amount of programmed cell death that occurs in conjunction with ischemic cell death. This scenario assumed that the development of a neuroprotective drug would reduce the effect of stroke on disability by 50 percent and that everyone with an acute stroke would be eligible. The total cost of the medication would be $3,500 per year for each person treated.

Prevention of Alzheimer’s. Much effort has already been expended to find compounds that might delay the onset of Alzheimer’s disease. In our model, each person had a predicted rate of onset for Alzheimer’s disease that depended on age, sex, race, and education. We used this rate to simulate an age of onset for Alzheimer’s (if any). For this scenario, we added three years to the age of onset based on the consensus of our panel of neurological experts. So, for example, if a person developed Alzheimer’s at age seventy in our status quo simulation, then the disease appeared at age seventy-three (assuming the person survived that long). This delay would reduce the prevalence of Alzheimer’s by about one-third. The cost would be similar to a statin at $60 per month, and all elderly people without Alzheimer’s disease would take the drug.

Prevention of diabetes. Thirty percent of obese elderly people (defined as body mass index, or BMI, greater than 30) were assumed to be treated by insulin sensitization drugs, which would reduce the hazard of diabetes by 50 percent over fifteen years. Similar to rosiglitazone, the drug would cost about $108 per month, and patients would have to continue taking the medicine until they die, regardless of the effects.

Compound that extends life span. Restricting the caloric intake of mice and rats by 30 percent results in an approximate 25 percent extension in life expectancy. The mechanism underlying this effect is unknown. Our model considered a mythical compound that can reproduce this effect in humans. Only the incoming sixty-five-year-old cohorts would be eligible, and it was assumed that they had been taking the drug for thirty years with an annual cost of $365. Mortality would be reduced in such a way that life expectancy would increase by approximately ten years. We considered two scenarios: The first, consistent with biological evidence, assumed that these years of life would be healthy years, thereby compressing morbidity toward the end of the life cycle. The second (unhealthy) scenario assumed that the additional years would be added at the end of the life cycle, with a concomitant increase in disease and disability.¹⁴

Technologies’ impact. Exhibit 7 summarizes the impact of these technologies, assuming their full adoption by 2002. We chose a common adoption year to facilitate comparison between the technologies and to allow us to look ahead almost thirty years. Several striking patterns emerge. Some technologies, when fully adopted, are forecast to be extremely expensive. For example, the elderly could be spending $26 billion on ICDs by 2015, and—once one accounts for the attendant morbidity and mortality—the devices would increase overall elderly health care spending by 7 percent.

ICD impact. The ICD simulation is particularly germane in light of recent Medicare policy changes. As Mark Hlatky and colleagues note in their excellent discussion of ICDs, the devices are very effective for secondary prevention of sudden cardiac death. As a result, Medicare has covered ICDs for patients with a history of life-threatening arrhythmias (ventricular fibrillation) since 1986 and for less severe arrhythmias (ventricular tachycardia) since 1999. The cost-effectiveness for such secondary prevention is around $54,000.¹⁵

More controversial is the impact of ICDs for primary prevention. Medicare already covers prophylactic use in patients at high risk of sudden death from ischemic cardiomyopathy. But it is the most recent coverage decision that will greatly expand use to patients with heart failure or poor functioning in their left ventricle. Our simulation went further and assumed that half of elderly patients with new cases of heart failure or a heart attack would receive ICDs. Under these circumstances, there would be approximately 374,000 procedures performed annually in 2015 and 550,000 in 2030, at a total treatment cost of $14 billion and $21 billion, respectively. The cost for each additional life year would be approximately $100,000.

Cost effects. Some technologies are not expensive by themselves (such as cancer vaccines) but add to health care spending because of their efficacy. As people live longer, they incur more costs. Overall, though, cancer vaccines are promising because of the substantial potential gains in life. Treatment of stroke with neuroprotective drugs would add $3–$4 billion annually (0.4 percent of total spending) once one accounts for mortality and morbidity changes. The cost per additional life year is $22,000. Once one accounts for the potentially large morbidity effects—including reductions in nursing home entry—such technologies should be very cost-effective.

Other technologies achieve health improvements at a very high price. These include anti-angiogenesis, pacemakers for atrial fibrillation, and LVADs. All of these are costly in relation to the health benefits they are known to produce. Our findings for anti-angiogenensis are consistent with recent experience with bevacizumab (trade name Avastin), an anti-angiogenesis drug used to treat patients with advanced colorectal cancer. This drug extends median survival by about five months at a treatment cost of $50,000 or more.¹⁶ Our simulations show that if treatment is broadened, the cost per additional life year could go even higher. Without clear criteria for who will respond to anti-angiogenesis, and how long they will need to remain on these drugs, costs per additional life year are likely to be very high.

Telomerase inhibitors have the potential to save money to offset treatment costs; treatment is forecast to cost $6.4 billion by 2030, but total health care spending will rise modestly (0.5 percent). These savings occur because people are dying of less costly diseases. If the cost of the treatment could be reduced, it therefore could be very cost-effective. Attempts to prevent both diabetes and Alzheimer’s disease are not forecast to be cost-effective, in part because they involve treatment of large portions of the population, and the efficacy is limited. Better risk stratification in the future will allow for better targeting; however, these screening tests will also likely be costly at first. We also did not include any potential savings to families alleviated of the burden of illness, which can be very high with some diseases.

Life-extending compound. Finally, we simulated a compound to extend lifespan. Our expert panel suggested that this would be a pill taken over a lifetime. The consequences are dramatic; under our healthy scenario, it would increase health care spending by 14 percent in 2030, despite relatively little increase in morbidity and disability. The reason is simple: If such a pill had been available in 2002, the population of Medicare eligibles would be thirteen million greater in 2030 than the current forecast (seventy-one million). If such a pill could be developed with the efficacy profile we simulated, it would be worth $9,000 per additional life year.

Under the less optimistic scenario that the pill keeps people alive but increases disease and disability, the consequences are enormous. Total health care spending in 2030 would be 70 percent higher than under the status quo, since there would be more elderly people, and people would incur disease and disability at older ages. Still, this treatment is relatively inexpensive—$30,000 per additional life year. This scenario shows the inherent tension in medical improvements generally: We can keep people alive to incur more disease and disability, but the overall rate is one that many consider “worth it.”

Caveats. Simulations of this sort require certain caveats. First, we did not adjust our estimates—as many actuarial models do—to reflect historical trends in real health care costs. Including such projections would not change our conclusions about the relative impact of new innovations.¹⁷ Furthermore, our goal was to isolate the effects of new technologies, whereas historical trends subsume some level of technological improvement. Thus, we would in some sense be “double counting” if we had added a new technology on top of the trended projections.

More generally, changes in behavior are beyond the scope of our model, which is meant to highlight the effects of incremental improvements in medical technology. So, for example, our estimates of the effects of a drug that might improve longevity can be taken only as starting point for discussion. Such a technology is so transforming that it is well beyond the capacity of such a model to deal with it. As life expectancy moves beyond 100 years, people might take better care of themselves (or perhaps decide to “live it up”), work longer, and behave in fundamentally different ways. Society as a whole would be transformed. Our goal in presenting this scenario in particular was merely to demonstrate the technological risk embodied by current biomedical research and to consider its implications as best we can.

In addition, the ultimate effect of a technology depends on its timing and its price, both of which are difficult to forecast, are interrelated, and influence diffusion. But it is unclear how to forecast future prices in the context of our model. The panels recognized, but could not predict, that costs of a procedure will fall over time with higher rates of adoption. We assumed that these technologies were similar to treatments they resemble or replace.

Finally, it should be noted that we focus on the elderly. Of course, access to technology will not be restricted to older Americans. Many of the new treatments will be expensive and hence will raise the cost of health insurance for the nonelderly, and fewer people will be able to afford comprehensive coverage. In fragmented health insurance markets and incomplete health insurance coverage, the fruits of medical progress will be distributed unevenly. Furthermore, the benefits that any socioeconomic group derives from innovations will depend on the prevalence of treatable disease in that group. If we design cures for the diseases of “rich people”—as cardiovascular disease once was—then gradients in health are likely to widen.

Conclusion

A forecasting exercise is less important as a literal “reading of the future” than as an attempt to unpack the competing forces of the present day. Two issues are paramount in this discussion. First, the rising prevalence of chronic disease among the near-elderly will continue to drive gradual increases in morbidity among the elderly with only a modest increase in health spending. At the same time, advances in bioengineering, genetics, the life sciences, and clinical medicine will lead to rapid advances in medical care. These innovations will resemble their predecessors—modest improvements in health that are often worth the cost. Ultimately, society faces its greatest spending risk not from demographic and health trends, but rather from medical technologies.

These two trends are of course fundamentally related: Technological improvements and increases in chronic diseases such as cardiovascular disease and diabetes are both symptoms of widespread economic progress. Although biomedical research has been successful in combating many diseases, the development of a technological fix that makes us all behave in healthy ways does not appear to be on the near horizon.

Principal funding for this study came from the Centers for Medicare and Medicaid Services (CMS Contract no. 500-95-0056), with additional funding from the National Institute on Aging through its support of the RAND Roybal Center for Health Policy Simulation (P30AG024968), RAND Center for the Study of Aging (P30AG12815), and the Stanford Center for Demography and Economics of Health and Aging. The authors are solely responsible for the paper’s contents. No statement in this paper should be construed as being an official position of the CMS, the Department of Veterans Affairs, or the U.S. government.

NOTES

1. A.H. Mokdad et al., “Diabetes Trends in the U.S.: 1990–1998,” Diabetes Care 23, no. 9 (2000): 1278–1283; and A.H. Mokdad et al., “The Continuing Epidemic of Obesity in the United States,” Journal of the American Medical Association 284, no. 13 (2000): 1650–1651.
2. M.J. Buntin et al., “Increased Medicare Expenditures for Physicians’ Services: What Are the Causes?” Inquiry 41, no. 1 (2004): 83–94.
3. J. Lubitz et al., “Health, Life Expectancy, and Health Care Spending among the Elderly,” New England Journal of Medicine 349, no. 11 (2003): 1048–1055.
4. This online appendix can be found at content.healthaffairs.org/cgi/content/full/hlthaff.w5.r5/DC2.
5. The MCBS sample consists of aged and disabled Medicare beneficiaries. It attempts to interview each person twelve times over three years, regardless of whether the person resides in the community or in an institution. Each fall a new panel is introduced with a target sample size of 12,000 respondents, and each summer a panel is retired.
6. See, for example, E.M. Crimmins, Y. Saito, and S.L. Reynolds, “Further Evidence on Recent Trends in the Prevalence and Incidence of Disability among Older Americans from Two Sources: The LSOA and the NHIS,” Journals of Gerontology: Series B, Psychological Sciences and Social Sciences 52, no. 2 (1997): S59–S71; K.G. Manton, L. Corder, and E. Stallard, “Chronic Disability Trends in Elderly United States Populations: 1982–1994,” Proceedings of the National Academy of Sciences (U.S.) 94, no. 6 (1997): 2593–2598; and R.F. Schoeni, V.A. Freedman, and R.B. Wallace, “Persistent, Consistent, Widespread, and Robust? Another Look at Recent Trends in Old-Age Disability,” Journals of Gerontology: Series B, Psychological Sciences and Social Sciences 56, no. 4 (2001): S206–S218.
7. Mokdad et al., “Diabetes Trends”; Mokdad et al., “The Continuing Epidemic”; and D.N. Lakdawalla, J. Bhattacharya, and D.P. Goldman, “Are the Young Becoming More Disabled?” Health Affairs 23, no. 1 (2004): 168–176.
8. As a hypothetical example, if we find in the NHIS that fifty-year-olds in 2004 have a prevalence of diabetes of 10 percent, but forty-five-year-olds have a prevalence of 12 percent, then one would expect that when these cohorts enter Medicare in 2019 and 2024, respectively, the younger cohort would have a greater prevalence of diabetes by at least 2 percent. The actual difference is greater than 2 percent because we allow for the fact that there is an age-specific trend in diabetes. More details can be found in the technical appendix; see Note 4.
9. For this simulation, the one-year probabilities of one or two ADL limitations and three or more disabilities were cumulatively reduced by 1 percent per year. In addition, we assumed that the entering cohort to Medicare would experience a decline in disabilities of 0.56 percent per year, as Kenneth Manton and XiLiang Gu find using data from the National Long Term Care Survey. Otherwise, the forecast of disease and disability for the incoming cohort is based on the NHIS. K.G. Manton and X. Gu, “Changes in the Prevalence of Chronic Disability in the United States Black and Nonblack Population above Age Sixty-five from 1982 to 1999,” Proceedings of the National Academy of Sciences (U.S.) 98, no. 11 (2001): 6354–6359.
10. Lubitz et al., “Health, Life Expectancy, and Health Care Spending.”
11. Our baseline differs from the baseline produced by official estimates of future Medicare spending, which include a rate of technological advancement through improved mortality and higher spending. Such an assumption is inappropriate here because we want to isolate the consequences of a particular technology, and such technologies are implicitly embedded in the Medicare actuaries’ baseline forecast. Thus, our baseline is a standard of medicine practiced in the 1990s, whereas the actuaries’ baseline is their prediction of how medicine will be practiced in future years.
12. These technologies—and the process used to identify them—are described in more detail in D.P. Goldman et al., Health Status and Medical Treatment of the Future Elderly: Final Report, Pub. no. TR-169-CMS (Santa Monica, Calif.: RAND, 2004).
13. P.G. Shekelle et al., “Identifying Potential Health Care Innovations for the Future Elderly,” Health Affairs, 26 September 2005, content.healthaffairs.org/cgi/content/abstract/hlthaff.w5.r67.
14. We implemented the unhealthy scenario by reducing the mortality hazard for treated individuals by 80 percent; this corresponds to an increase in life expectancy at age sixty-five from nineteen to twenty-nine years. The healthy scenario reduced all hazards (mortality, disease, and functional limitations) by 63 percent, corresponding to a similar change in life expectancy.
15. M.A. Hlatky, G.D. Sanders, and D.K. Owens, “Evidence-based Medicine and Policy: The Case of the Implantable Cardioverter Defibrillator,” Health Affairs 24, no. 1 (2005): 42–51.
16. H. Hurwitz et al., “Bevacizumab plus Irinotecan, Fluorouracil, and Leucovorin for Metastatic Colorectal Cancer,” New England Journal of Medicine 350, no. 23 (2004): 2335–2342.
17. Real health care spending, after demographic changes were adjusted for, grew about 4 percent. Cost-effectiveness studies often discount future spending by 3 percent. If we included both, the resulting estimates would not be appreciably different.


Dana Goldman (dgoldman{at}rand.org) is corporate chair and director of health economics at RAND in Santa Monica, California. Baoping Shang is a fellow at the Pardee RAND Graduate School. Jay Bhattacharya is an assistant professor of medicine at Stanford University in Stanford, California. Alan Garber is a staff physician at the Veterans Affairs (VA) Palo Alto Health Care System and the Henry J. Kaiser Jr. Professor at Stanford. Michael Hurd directs the RAND Center for the Study of Aging. Geoffrey Joyce is an economist at RAND. Darius Lakdawalla is an associate economist at RAND. Constantijn Panis is a manager at Deloitte and Touche LLP in Los Angeles. Paul Shekelle is director of the Evidence-Based Practice Center at RAND Health and a staff physician at the Greater Los Angeles VA Medical Center.

Access the table of contents for this package

DOI: 10.1377/hlthaff.W5.R5
©2005 Project HOPE–The People-to-People Health Foundation, Inc.