The Artificial Heart: Prototypes, Policies, and Patients (1991)

MARJORIE FUNK

AN EXAMINATION OF THE EPIDEMIOLOGY of end-stage heart disease is a necessary first step in determining the nature and magnitude of the target population for fully implantable long-term mechanical circulatory support systems (MCSSs). In attempting to estimate the target population for these devices, it is important to focus on heart failure initially, because the purpose of MCSSs is either to support the failing heart (ventricular assist devices) or to replace it (total artificial hearts). To determine the number of individuals with the most severe illness (primary group), mortality statistics will be examined. Further, the circumstances of death (not sudden) and the presence of comorbid conditions that would limit the utility of the devices will be considered. To estimate the number of individuals who are less disabled by their heart disease (secondary group), prevalence statistics will be evaluated.

The National Center for Health Statistics (NCHS) reports that in 1988 there were 42,940 deaths in the United States for which congestive heart failure (CHF) was designated as the underlying cause of death (Thom, 1991). In the Framingham cohort, approximately 40 to 50 percent of CHF deaths

Marjorie Funk is Assistant Professor, Yale University School of Nursing, and doctoral candidate, Department of Epidemiology and Public Health, Yale University Graduate School, New Haven, Connecticut.

were sudden, which was defined as death within one hour in patients who were clinically stable (Kannel et al., 1988). Francis (1986) reviewed the incidence of sudden death in patients with CHF in 12 published reports and found that 8 cited sudden death rates of between 43 and 63 percent. Of 642 men with moderate failure (New York Heart Association Classes II and III) in the Veterans Administration Cooperative Study on Vasodilator Therapy of Heart Failure, 45 percent died suddenly (Cohn et al., 1986). Application of these percentages to the 42,940 CHF deaths in 1988 reveals that between 15,888 (0.37 × 42,940) and 25,764 (0.60 × 42,940) people did not die suddenly. This group could have benefited from MCSS.

The presence of medical and psychosocial comorbidities would likely preclude MCSS use in some of these individuals. Data, however, are not available regarding the presence of comorbid illness in individuals with end-stage heart failure. A very approximate estimate of the prevalence of comorbid conditions can be derived from the population-based study conducted by Kottke and associates (1990). In their attempt to estimate the need for long-term MCSSs, they reviewed all deaths in a five-year period in Olmsted County, Minnesota, using restrictive criteria for comorbid conditions, as well as age and time elapsed between onset of symptoms and death. They found that of the 248 people age 15 to 69 years old who died of cardiac disease, 52 percent died so quickly that intervention with an MCSS was not possible, 14 percent had comorbid conditions that would prevent them from benefiting from a device, and 20 percent died suddenly and had comorbidity. This left 35 individuals (14 percent) who were considered to be candidates for cardiac replacement.

Kottke and colleagues projected that there would be 16,500 potential candidates for cardiac replacement nationally each year. This estimate was based on assumptions that the 95,000 residents of Olmsted County were representative of all Americans, and that the population of the United States was 225 million at the time of the study, calendar years 1979 through 1983. Extrapolating the estimate based on the 35 candidates over five years yields approximately 16,500 annually.

With caution, the figure of 14 percent can be applied to the 42,940 CHF deaths to estimate the number of potential candidates for MCSS. This calculation reveals that 6,012 individuals per year may have benefited from these devices. This, however, may be misleading because only 31 percent of the Olmsted County deaths were attributed to chronic heart failure, whereas 54 percent were related to acute myocardial infarction (MI). The authors are not explicit in defining the diagnostic criteria for these conditions. They do, however, state that all of the 35 people designated as potential candidates for cardiac replacement had “severe depression of left ventricular function.” Acute MI, chronic heart failure, inability to wean from cardiopulmonary bypass, idiopathic cardiomyopathy, and cardiac tu-

mor were the “sources” of heart disease. It is unclear whether the term “heart failure” is used in the same way as in the NCHS mortality statistics.

If it is assumed that the Olmsted County deaths were indeed due to a broader range of heart problems than merely heart failure, then the 14 percent figure could be applied to NCHS mortality statistics for coronary heart disease (CHD), cardiomyopathy, and heart failure —the most frequent sources of ventricular dysfunction. The National Heart, Lung, and Blood Institute (NHLBI) (1990) reports that for 1987 there were 512,138 deaths due to CHD, 18,660 deaths from cardiomyopathy, and 41,210 attributed to heart failure, for a total of 572,008 deaths. Fourteen percent of this is 80,081, which is—as expected—considerably higher than the 6,012 calculated from the 1988 CHF mortality figure.

Because they do not take age or comorbidity into account, the calculations presented thus far are higher than are appropriate for the purpose of estimating a target population for MCSSs. It has been determined that these devices would generally be contraindicated in children who were still growing (less than 15 years old), as well as in the elderly (85 years and older). The above calculations, however, include all age groups.

To determine the number of people between the ages of 15 and 85 who might benefit from an MCSS, the numbers of deaths due to CHF and CHD in this age range were calculated. The most recent age-specific death rates available (Thom, 1991) were applied to age-specific projections of the United States population for 1990 (Spencer, 1989). These calculations reveal 29,649 deaths from CHF in the 15- to 85-year-old group and 384,102 deaths from CHD in the 25- to 85-year-old group. (Death rates from CHD for those less than 25 years old are not available but are presumed to be very low.)

Next, the 14 percent figure was applied to reflect the number who did not die suddenly and who did not have comorbid conditions which would preclude device use. For CHF 4,151 (0.14 × 29,649) individuals between the ages of 15 and 85 and for CHD 53,774 (0.14 × 384,102) individuals between 25 and 85 may have benefited from MCSS.

Application of the 14 percent figure from the study by Kottke and associates presupposes that the rates of sudden death and comorbid conditions for those between 70 and 85 years old are similar to rates for those between 15 and 69 (the age range in the work of Kottke et al.). No research could be found indicating that the percentage of deaths that are sudden is significantly different for the elderly. If anything, the rate of sudden death—defined as death within one hour in patients who were clinically stable—may be slightly lower in older people, because they are less likely to be clinically stable. One might also assume that older people have more comorbid conditions. Additionally, the phenomenon of “selective survivors” may be operating: people who live a long time do so because they

are healthy. The 14 percent figure, therefore, is probably appropriate for the entire age group considered (15 to 85). The potentially lower sudden death rate and higher comorbidity rate—if present—may negate each other.

It is tempting to examine statistics on cardiomyopathy, since currently it is the most common indication for heart transplantation (Kriett and Kaye, 1990). In the study by Kottke and associates, however, only 1 of the 35 potential candidates (3 percent) for cardiac replacement had cardiomyopathy. NHLBI (1990) reported 18,660 deaths from cardiomyopathy in 1987. This represented only 2.4 percent of all deaths from heart disease that year.

In summary, the best estimates for the current target population for MCSS are 4,151 for CHF and 53,774 for CHD, for a total of 57,925. Whereas these are the most precise figures obtainable, they must be viewed with much caution since they were calculated using percentages of sudden death and comorbidity derived from Olmsted County data. It is known that Olmsted County residents have better access to medical care, are more predominantly white (99 percent), and are of higher socioeconomic status than the U.S. population as a whole. The estimates for the target population for the primary group, thus, may be low.

The total of 57,925 is considerably higher than the estimates presented by the two most recent attempts at defining a target population for MCSSs. Lubeck and Bunker (1982) estimated that there would be 33,600 artificial heart candidates under the age of 65. They assumed, however, that there would be no implants in patients who were “stable” under medical management, regardless of the severity of their disease; it would almost always be an emergency procedure. They acknowledged that, should the artificial heart prove to be highly successful, this estimate could double to approximately 66,000. Conversely, if insurmountable problems are encountered, the number of candidates would be substantially lower, or approximately 16,000. As noted above, Kottke et al. (1990) estimated a target population of similar magnitude (16,500). The higher upper age limit (85 versus 65 and 69 in the other two analyses) likely accounts for the larger estimate in the present projection. If the upper age limit were reduced to 75 rather than 85, then the same calculations would produce a target population of 1,769 for CHF and 29,598 for CHD, for a total of 31,367.

Since long-term, fully implantable circulatory support devices are not yet in use clinically, it is essential to attempt to estimate the number of people who might benefit from these devices in the next 30 years. Weinstein and associates (1987) developed a computer simulation model which forecasts future mortality from CHD in the U.S. population. This model projected an

annual increase in mortality for each five-year interval from 1990 to 2010. Using the current CHD target population figure of 53,774 calculated above and applying the percentage increase in mortality for each five-year period as projected in the computer model results in the figures for CHD presented in Table D.1.

Because Weinstein and associates did not project CHF mortality, an alternative method for determining future trends in CHF mortality was developed. Assuming that trends in CHF mortality will continue into the immediate future, a simple linear regression using year and number of deaths was performed. The number of deaths per year was available as far back as 1950. It is apparent that the number of deaths attributable to CHF increased each year. After it was determined that a linear relationship existed, the parameter estimates were used to determine the number of deaths for future years.

To project the number of deaths in the relevant age group, the percentage of all current CHF deaths that occurred in the 15- to 85-year-old age group was calculated: 29,650/46,421 = 0.6387. Assuming that this percentage will be relatively constant, for each future year this figure was applied to the number of deaths derived from the linear regression equations. Then, to account for the effect of sudden death and comorbidities, 14 percent of the resulting numbers were computed and are displayed in Table D.1.

The total number of potential users of circulatory support devices is projected to increase over the next 30 years: from approximately 60,300 in 1995 to 72,869 in 2020. If the upper age limit were reduced to 75 instead of 85, then (as shown in Table D.2) there would be approximately 32,718 potential users in 1995. This figure would increase to about 39,334 by 2020. All of these projections assume that recent trends in cardiac risk factor levels and the development of new technologies and treatments for

heart disease will persist. The impact of primary prevention and medical care on mortality trends, however, is extremely difficult to determine. The increase in the target population for long-term MCSS reflects the anticipated growth of the general population during the next 30 years, and in particular the movement of the “baby boom ” generation born in the 1940s and 1950s into the age range of increased risk of heart disease.

The primary problem with using mortality data to estimate the target population for long-term implantable circulatory support devices is that some people with end-stage heart disease, but who are not yet dying, might also benefit from these devices. By definition, individuals in the secondary group are less disabled by their disease than the moribund patients who comprise the primary group. To determine the number of individuals in the secondary group, prevalence statistics will be examined.

Prevalence estimates can vary considerably depending on the source of the data. For example, data obtained by interview (e.g., the National Health Interview Survey) are apt to be very different from hospital discharge data (e.g., the National Hospital Discharge Survey). Whereas the former depend on the awareness and willingness of the respondent to report the condition, the latter are limited by the inclusion of only those whose condition is severe enough to require hospitalization. Because individuals with endstage heart disease who would be candidates for these devices are likely to be hospitalized, using hospital discharge data seems appropriate. It is important to note, however, that this source of data is limited by the fact that it is hospital discharges rather than individuals that are enumerated. Thus, one individual who is hospitalized on three occasions would be

counted three times, resulting in an overestimate of the need for these devices. Additionally, since most prevalence data do not discriminate according to the severity of illness, it is likely that a substantial proportion of those included would not be sick enough to require circulatory support devices.

The best available estimate of the prevalence of CHF in the United States is 3,037,665 (Thom, 1991). This figure was obtained by applying the rates observed in two community studies conducted in 1962 by Gibson and colleagues (1966) to 1988 census figures. Whereas age-specific data are not available, it is known that more than half (1,654,700) are 75 years old and over. Additionally, most of these individuals with CHF are not sufficiently disabled to benefit from device therapy.

Estimating prevalence from hospital discharge statistics yields a more severely ill group. National Hospital Discharge Survey data reveal that in 1988 there were 663,000 hospital discharges with CHF as the primary diagnosis, and 1,858,000 as one of all-listed diagnoses. Whereas the number in the relevant age group (15 to 84 years old) is not known, approximately 80 percent of both groups are discharges of individuals over 64 years old. Although this population is sick enough to require hospitalization, it is unclear how many are sick enough to warrant a circulatory support device. Additionally, information is lacking on the presence of comorbid conditions which would preclude benefit from device therapy. Lastly, as stated above, these numbers represent discharges rather than individual patients and thus constitute an inflated estimate of the need for these devices.

To obtain a more realistic estimate, age-specific data on the number of hospitalizations, excluding deliveries, for individuals interviewed in the National Health Interview Survey were analyzed to determine the proportion of discharges that represent individual patients. Percentages ranged from 69 percent for those over 74 years to 81 percent for those between 25 and 44 years old; older people are hospitalized multiple times more frequently than younger individuals. Application of a 70 percent figure to the 663,000 hospital discharges reveals that approximately 464,100 individuals were hospitalized with CHF in 1988. From this figure, the number who died of CHF (4,151) must be deducted, resulting in a prevalence of 459,949. Because people with CHF are hospitalized more frequently than the general population, the prevalence is probably closer to 400,000. This figure, however, does not take into account age, severity of illness, or comorbidity.

As with mortality, there has been an annual increase in the prevalence of CHF in the United States since 1971. Rates of hospitalization for CHF more than tripled from 1971 to 1988: from 8.2 to 27.2 per 10,000 popula

tion (Thom, 1991). Since 1980, rates increased more sharply for the 45- to 64-year-old group (70 percent) than for the 65-and-over group (33 percent). Because a majority of devices would probably go to those under 65, this trend is important to consider if the quality of prevalence data ever improves sufficiently to be used as the basis for future projections.

National Hospital Discharge Survey data indicate that there were over 2 million discharges for CHD in 1988. Of these, 716,000 were for acute MI, 411,000 were for atherosclerotic heart disease, and 921,000 were for other ischemic heart diseases. Seventy percent of the total of 2,048,000 minus the 53,774 who died of CHD results in a prevalence of 1,379,826. Again, because individuals with CHD are hospitalized more frequently than the general population, the prevalence is likely to be nearer to 1,000,000. Lack of information regarding age, severity of illness, and comorbidity, as outlined above for the CHF estimates, limits the usefulness of these data also.

In their review of trends in CHD, Higgins and Thom (1989) report that hospital discharges for CHD increased from 1970 to 1978, and then appeared to decline and rise again to a stable level. They point out that rates are influenced by admission policies, medical practice, and diagnosis-related groups. Hospital stays are shorter now, and more procedures are done on an outpatient basis. Despite this, the prevalence of CHD may be rising. As the population increases disproportionately at older ages and survival following acute MI improves, it is likely that there will be increasing numbers of elderly individuals with CHD. Some of these people may be young enough to benefit from a circulatory assist device.

Prevalence trends were also examined in the population-based Minnesota Heart Survey (Burke et al., 1989). In individuals age 30 to 74, the rate of discharge diagnoses for acute CHD decreased from 1970 to 1980, and then increased significantly from 1980 to 1985. There was no change in MI rates from 1970 to 1985.

In summary, the best available prevalence estimates are 400,000 for CHF and 1,000,000 for CHD. Since these numbers have not been adjusted for age, severity of illness, or comorbidity, they significantly overestimate the number of individuals in the secondary group. Although precise current numbers or future projections cannot be derived for this group, data indicate that the prevalences of both CHF and CHD are increasing.

The focus of this paper has been an epidemiologic analysis of heart disease in the United States. Although little is known about the incidence

and prevalence of heart disease internationally, mortality data are available and trends have been determined for a number of countries. It is known that countries such as Australia, Belgium, Canada, Finland, Israel, Japan, and New Zealand have experienced declining CHD mortality similar to the trend noted in the United States (Thom, 1989; Beaglehole, 1990). With considerable caution, the same calculations performed on U.S. mortality data to determine a target population for MCSS use might be undertaken for these countries. Potential differences in diagnostic customs and classification procedures, however, make international comparisons, and thus the validity of such calculations, highly speculative. Additionally, calculations using percentages of sudden death and comorbidity derived from Olmsted County—where socioeconomic status is higher and access to medical care is better —would result in spuriously low international estimates.

Mortality rates for CHD in selected industrialized countries are displayed in Table D.3. Countries such as Canada, England and Wales, and Germany have high enough mortality rates, as well as large enough populations, to produce a group of potential MCSS users of sufficient magnitude to warrant consideration when one is estimating the need for these devices.

Calculations based on mortality data (primary group) reveal that currently about 58,000 individuals might benefit from circulatory support sys-

tems. The target population is projected to increase over the next 30 years: to approximately 64,000 in 2000, 68,000 in 2010, and 73,000 in 2020. These estimates are considerably lower if the upper age limit is reduced from 85 to 75 years (31,000 currently, 35,000 in 2000, 37,000 in 2010, and 39,000 in 2020).

The inferior quality and lack of specificity that characterize prevalence data preclude their utility in estimating the potential application of long-term MCSSs. Although it is intuitively sensible to consider the number of individuals with end-stage heart disease who are currently alive, this information simply is not available. Population-based data concerning the natural history of heart disease in general, and heart failure in particular, could not be located.

Beaglehole, R. 1990. International trends in coronary heart disease mortality, morbidity, and risk factors. Epidemiologic Reviews 12:1-15.

Burke, G. L., J. M. Sprafka, A. R. Folsom, R. V. Luepker, S. W. Norsted, and H. Blackburn. 1989. Trends in CHD mortality, morbidity and risk factor levels from 1960 to 1986: The Minnesota Heart Survey. International Journal of Epidemiology 18:S73-S81.

Cohn, J. N., D. G. Archibald, S. Ziesche, et al. 1986. Effect of vasodilator therapy on mortality in chronic congestive heart failure: Results of a Veterans Administration Cooperative Study . New England Journal of Medicine 314:1547-1552.

Francis, G. S. 1986. Development of arrhythmias in the patient with congestive heart failure: Pathophysiology, prevalence and prognosis. American Journal of Cardiology 57:3B-7B.

Gibson, T. C., K. L. White, and L. M. Klainer. 1966. The prevalence of congestive heart failure in two rural communities . Journal of Chronic Disease 19:141-152.

Higgins, M., and T. Thom. 1989. Trends in CHD in the United States. International Journal of Epidemiology 18:S58-S66.

Kannel, W. B., J. F. Plehn, and A. Cupples. 1988. Cardiac failure and sudden death in the Framingham Study. American Heart Journal 115:869-875.

Kottke, T. E., D. G. Pesch, R. L. Frye, D. C. McGoon, C. A. Warnes, and L. T. Kurland. 1990. The potential contribution of cardiac replacement to the control of cardiovascular diseases: A population-based estimate. Archives of Surgery 125:1148-1151.

Kriett, J. M., and M. P. Kaye. 1990. The Registry of the International Society for Heart Transplantation: Seventh Official Report—1990. Journal of Heart Transplantation 9:323-330.

Lubeck, D. P., and J. P. Bunker. 1982. Case study #9: The Artificial Heart: Cost, Risks, and Benefits. In: The Implications of Cost-Effectiveness Analysis ofMedical Technologies: Background Paper #2: Case Studies of Medical Technologies. Washington, D.C.: U.S. Government Printing Office.

National Heart, Lung, and Blood Institute. 1990. Morbidity and Mortality Chartbook

on Cardiovascular, Lung, and Blood Diseases/1990. U.S. Department of Health and Human Services.

Spencer, G. 1989. Projections of the Population of the United States, by Age, Sex, and Race: 1988 to 2080.U.S. Department of Commerce.

Thom, T. J. 1989. International mortality from heart disease: Rates and trends. International Journal of Epidemiology 18:S20-S28.

Thom, T. J. 1991. Unpublished data supplied to the Institute of Medicine Committee to Evaluate the NHLBI Artificial Heart Program.

Weinstein, M. C., P. G. Coxson, L. W. Williams, T. M. Pass, W. B. Stason, and L. Goldman. 1987. Forecasting coronary heart disease incidence, mortality, and cost: The Coronary Heart Disease Policy Model. American Journal of Public Health 77:1417-1426.