INTRODUCTION

Racial and ethnic differences in the age patterns of incidence and prevalence of chronic morbidity and disability are important because they provide data on (1) biological mechanisms, endogenous and exogenous, of age-related morbidity, (2) how the quality of life for groups changes over age and time, and (3) lead indicators of future mortality and disability trends that aid in the development of health policy by improving mortality, health service, and population forecasts.

We have previously examined epidemiological, demographic, and other data on racial and ethnic health differences (Manton et al., 1987). Here we examine more recent biomedical and epidemiological research on African-American, white, and Hispanic differences in health and disability. Though data on racial and ethnic health differences have improved and have generated many new hypotheses, there are many areas in which no definitive studies have been done for different racial and ethnic groups. Consequently, there is often little consensus on the magnitude or age dependence of racial and ethnic differences in health or on the biological mechanisms underlying many differences.

Since the scientific record on racial and ethnic health differences, especially at late ages, is incomplete, we must assemble available data into a coherent description to compare age-related health differences across racial and ethnic groups. For this purpose, a life-table model of the age relation of three health processes—disability, morbidity, and mortality—is useful. Figure 3-1 portrays changes in two hypothetical cohorts, one with high mortality and one with low mortality, as linked life-table functions.

FIGURE 3-1

Interrelations of morbidity, disability, and mortality in two hypothetical populations. SOURCE: Duke University Center for Demographic Studies.

The area between curves for each cohort represents the person-years lived free of morbidity and disability, with morbidity but free of disability, or with both morbidity and disability. The high-mortality cohort reflects a population with more time spent in morbid states but less time spent with disability (e.g., in developing countries, where acute and long-term medical care for disabled elderly persons is scarce, or in socioeconomically disadvantaged groups). The low-mortality cohort spends less time in morbid states, but more time with disability.

The first scenario suggests that in populations with low life expectancy, people spend time in clinically latent morbid states and that once disability occurs, the lack of social and long-term care infrastructure causes rapid terminal declines due to a loss of physical homeostasis (e.g., Colantonio et al., 1992). It is uncertain whether this scenario holds for U.S. African Americans and Hispanics who, often socioeconomically disadvantaged over much of the life course, may have greater social resources and relatively equitable access to medical and postacute services at age 65 through Medicare—and to long-term care services through Medicaid. In populations with high life expectancy, health care may defer chronic morbidity to later ages, with disabled people able to live a long time with adequate long-term care. This scenario, however, may not occur even in economically advanced countries. In Japan, one of the world's richest economies, only 6 percent of gross national product was spent on health services in 1990 even though, because of high life expectancy and declining fertility, the population is aging. Consequently, the health services provided to the elderly are limited, with many elderly enduring stays of 6 months or more in acute care hospitals because of a lack of long-term care and rehabilitation facilities (Okamoto, 1992).

The two hypothetical models represent scenarios against which the health experience of U.S. racial and ethnic groups can be compared. Race-related physiological differences (e.g., hypertension; Cooper and Rotimi, 1994) may produce more complex scenarios than the ones shown in Figure 3-1, as may differences in access to health care.

Ideally, longitudinal data on levels and types of disability, on flows into and out of morbidity and disability states, and on ageand gender-specific effects of diseases on mortality are used to construct such models. For example, a disabling event prevalent in U.S. postmenopausal women is hip fracture. The natural history of hip fractures, like that of other events and conditions, has evolved. The mean age for hip fractures in Britain increased from 67 years in 1944 to 79 years in 1990 (Keene, 1993). This 12-year shift marked a change in hip fracture's natural history. At age 67, hip fractures often do not involve the hip socket and are due to exogenous forces. Hip fractures at later ages often involve the hip socket, are due to osteoporosis, and require more health care. Such age changes in disease presentation could affect U.S. racial and ethnic differences in disability and mortality because African-American and Hispanic females have lower risks of hip fracture than white females.

Unfortunately, longitudinal data are not equally available for the groups reviewed. Consequently, we need to assess the quality and scope of the literature on racial and ethnic differences on specific diseases and conditions and then integrate these various findings. This is a multistep process.

First, we review data on the racial and ethnic differences in the risks of specific diseases and in the rates at which the diseases progress. When possible, we make gender-specific comparisons because racial and ethnic health differences often vary in degree, and sometimes in direction, by gender. Since many studies of racial and ethnic groups were done for individual diseases, studies of racial and ethnic differences have to be reviewed for a number of areas. Thus, the first stage of the review provides insight into racial and ethnic differences in disease-specific mechanisms and their age and gender dependence.

Since most disease-specific studies are of select populations, as a second stage we review the age relation of racial and ethnic differences in health, disability, and mortality in two sets of national data. First, we examine age patterns of total and cause-specific mortality by race to determine if they are consistent with the epidemiological data on racial and ethnic differences in the age dependence of disease processes. This is logical since racial and ethnic differences in disease in large part cause racial and ethnic differences in mortality age patterns. Second, we examine the National Long Term Care Surveys, where race-specific disability and mortality trajectories can be linked.

Thus, an assemblage of data is used to identify (1) the age dependence of disease mechanisms, (2) mortality and morbidity linkages, and (3) disability and mortality linkages. These relations, from different data sets, provide information to construct models like Figure 3-1.

MAJOR MORBIDITY PROCESSES AND THEIR AGE DEPENDENCE ACROSS RACIAL AND ETHNIC GROUPS

There are few studies that describe a number of diseases in a longitudinally followed elderly population. One such study (Guccione et al., 1994), though representing only whites, provides a context in which to discuss racial and ethnic differences in disease. In the study, 2,731 people were followed for 38 years to determine which of 10 conditions caused one or more of seven disabilities in people aged 64 to 98. The proportion of disability due to each condition, adjusted for age, sex, and comorbidity, is presented in Table 3-1. Stroke and heart disease caused disability on all seven functions. Diabetes, osteoarthritis, and hip fracture were also responsible for significant disability.

TABLE 3-1

Percentage of Disability Attributable to Specific Conditions After Adjustment for Age, Sex, and Comorbidity, by Condition and Activity for 2,731 Whites Followed for 38 Years in the Framingham Heart Study.

These results suggest that as a minimum, we should examine studies on hip fractures (osteoporosis), stroke, and heart disease in African Americans and Hispanics, and that it would be good to examine atherosclerosis, hypertension, diabetes, and anemia as well. In the review we look for differences between racial and ethnic groups (e.g., hip fractures are more prevalent among white females).

We examine age differences in disease as well as risk factors and disease mechanisms. We examine other prevalent diseases (notably cancer) that are not listed in Table 3-1. Also, to describe possible physiological mechanisms, we occasionally refer to cross-national data to distinguish social from physiological factors in racial and ethnic differences in disease. A complication in reviewing disease mechanisms is that there are racial and ethnic differences in disease interactions (e.g., atherosclerosis and osteoporosis, diabetes and stroke), especially at later ages.

BLACK, HISPANIC, AND WHITE TOTAL AND CAUSE-SPECIFIC MORTALITY DIFFERENCES

There are two reasons for analyzing black, white, and Hispanic age differences in disease mechanisms. One is to study the age-specific effects of the risks, incidence, and management of diseases in different racial and ethnic groups. The second is to discover whether it is plausible that racial and ethnic differences in the age trajectory of mortality are due to the different age risks of specific diseases in the three groups and to the effects of mortality selection on the age prevalence of specific conditions. The morbidity data suggest that black, white, and Hispanic morbidity risks are multifactorial and complex and are unlikely to be proportional across age. For example, hypertension and diabetes, two risk factors for stroke, may raise the mortality of blacks relative to whites in middle age. At older ages, whites (especially males) are at greater risk for atherosclerosis than blacks; this suggests that whereas blacks susceptible to stroke die at earlier ages, white circulatory mortality risks may accelerate more rapidly at later ages. A similar pattern may occur for white females; their greater osteoporosis risk relative to black females may cause their risks of atherosclerotic heart disease to increase more rapidly postmenopausally.

The need to examine the mechanisms of specific diseases and subsequently their contribution to cause-specific (and total) mortality at specific ages is amplified by the greater uncertainty about the accuracy of age reporting at later ages in population data for U.S. blacks than for whites. Uncertainty about data quality also exists for Hispanics because their mortality is advantaged relative to whites and because Hispanics are a racially mixed group that is hard to define in population data. Thus, the morbidity review provides independent information that can be used to assess the plausibility of the race- or ethnic-specific age trajectory of mortality rates calculated from cause-specific and total mortality data. This is especially important for studies in the United States, which does not have long-standing population registries (like Sweden) or vital statistics programs (like Great Britain) to provide high-quality population and mortality data at late ages. In the United States the problem is compounded because the level and structure of age-reporting errors in census and mortality data may differ; error structures may also differ across censuses. To avoid the problem of using data sources with different error structures in life-table calculations, we calculated extinct cohort tables for U.S. whites and nonwhites using only the age at death reported on death certificates. To see if this is a reasonable strategy, we first examined the quality of age reporting on death certificates relative to other sources in matched record studies.

Table 3-2 compares ages at death reported on death certificates (grouped into 5-year age categories) for three racial groups and ethnic groups with ages at death reported on Medicare data from the Master Beneficiary Record (MBR) for 124,507 matched cases in 1987 for Massachusetts and Texas. This allows us to quantify the size and direction of errors in age reporting on death certificates for blacks, whites, and Hispanics for ages 65 to 100 and above (Kestenbaum, 1992). The rows in Table 3-2 refer to age groups in the MBR files. In column 1, the mean age within each MBR age group is assumed to be the midpoint of the 5-year age interval. Columns 2, 5, and 8 present the mean ages in the death certificate files for the same people. The death certificate files may classify some of these people in younger or older age groups, leading to discrepancies in the estimated mean ages. Columns 3, 6, and 9 display the discrepancies. For whites, the differences are small, except for the last two age groups (-0.18 and -0.51). For Hispanics, the differences are also small, with only the last two age groups exceeding ±0.2 year. For blacks, all but the first two of the differences exceed ±0.2 year with what appears to be a substantial downward bias at older ages.

TABLE 3-2

A Comparison of the Mean Age at Death and Ratio of the Number of Deaths Calculated from Death Certificates for Persons Assigned to a Given Medicare Master Beneficiary Record Age Group.

Misclassification of age in the death certificate files may lead to underestimates and overestimates of the actual number of deaths in the various age groups. Columns 4, 7, and 10 display the ratios of (1) the number of deaths for matched cases in each age category for ages at death reported on the death certificate to (2) the number of deaths in the age categories generated from the MBR. For whites, all but two of the age groups are within 1 percent. For Hispanics, all but three are within 2 percent. For blacks, only one is within 1 percent, and two are within 2 percent. For blacks there were substantially more deaths reported on death certificates up to age 79 and fewer from ages 80 to 99. Deficits in the three older age categories from 85 to 99 were large for the death certificate data (i.e., -6.1%, -14.7%, -11.8%). These cases apparently were in the three lower age categories due to the underreporting of age at death on death certificates for blacks. The number of blacks (92) reported as age 100 and above on death certificates was similar to that in the MBR (91).

Also in Table 3-2 are two sets of ratios for blacks from Preston et al. (1996:Figure 1). Columns 11 and 12 refer to a sample of 2,657 deaths among African Americans in 1985 that were linked both to Social Security Administration (SSA) records and to the 1900-1920 U.S. censuses. The similarity of the census and SSA age distributions was used to argue for their accuracy relative to that of death certificate ages. SSA age was assumed to be most reliable because it was one of two sources where age agreed in 91 percent of 1,087 two-way agreements in the sample.

Preston's results differ from Kestenbaum's (1992). Both showed that the number of deaths reported at ages 65 to 69 was overstated on death certificates. For ages 70 to 84, the number of deaths recorded from death certificates may be overstated or understated. The number of deaths recorded from death certificates at ages 85-94 is understated, but the two studies disagreed on magnitude. For ages 95-99 the number of deaths recorded from death certificates may be overstated or understated. For ages 100 and above, the number of deaths from the death certificates is overstated, but the two studies disagreed on magnitude.

A difficulty in assessing mortality rates above age 100 is that the number of cases is small. Thus, it is important to assess the statistical precision of the results of match studies above age 100 to see if real differences are being identified. In the data in Preston et al. (1996), there appeared to be 26 death certificate, 20 SSA, and 16 census deaths above age 100 (calculated from the ratios in Table 3-2 and the 52 deaths reported at age 100 and older on death certificates, and with a three-way match rate of 50.5% assumed). If the number of deaths above 100 recorded by the SSA is assumed to be the most accurate and we assume that the 20 deaths are Poisson distributed, then the standard error of the estimate is 4.5, and a 2-standard error confidence interval ranges from 11 to 29 deaths. This confidence interval includes both the death certificate (26) and the census estimates (16) of deaths. Thus, the numbers of deaths reported above age 100 in the three sources are not significantly different.

In Kestenbaum (1992), 92 blacks had ages at death greater than 100 on death certificates, compared with 91 blacks in the MBR. The standard error for the 91 persons with ages at death over 100 in the MBR is 9.5, with a 2-standard error confidence interval of 72 to 110 deaths. This clearly includes the 92 deaths above age 100 recorded on death certificates. Again there is no statistically significant difference in the number of deaths above age 100 recorded in the death certificates and the MBR. The standard error for the 42 Hispanic deaths reported at age 100 and older from the MBR is 6.5, with a confidence interval of 29 to 55 deaths. This range includes the 49 deaths above age 100 recorded on the death certificates. Again, the difference is not statistically significant.

Thus, it can be concluded that the accuracy of death certificate ages differs over age and by race. For blacks, deaths at ages up to 84 were overreported and at ages 85 to 94 (and possibly later) underreported on death certificates. Given the differences between Kestenbaum (1992) and Preston et al. (1996), adjusting the data by using either ratio estimate is problematic. Neither match study showed that age reporting above age 100 on the death certificate is statistically different from reporting in the other data components.

The analysis of the two match studies suggests that it is not unreasonable to use death certificate data to calculate age-specific probabilities of death (q_x's) for U.S. white and nonwhite males and females using extinct cohort methods. In this method, the numerator of each q_x is the number of deaths recorded on death certificates for that age. The denominator is the sum of deaths reported on the death certificate at age x, plus deaths reported on death certificates for each subsequent year for each subsequent age. This specification of the denominator is not standard, but it can reduce the effects of certain types of errors in death certificate reports of age at death. For example, if there is a constant ratio of the number of deaths recorded on death certificates to the true number of deaths beyond age x, then the numerator and denominator used to calculate q_x have the same relative error; errors will cancel out, and an accurate estimate of q_x will be produced. This may be the case for nonwhite deaths for ages 85 to 99 recorded on death certificates. In Table 3-2 the downward bias in the denominator for ages 80 to 84 will produce overestimates of q_x's even though the numerator is correct. The near constancy of ratios from ages 70 to 84 means that bias will decline over this range. Thus, using extinct cohort methods to calculate life expectancy should show a slight downward bias in life expectancy for nonwhites relative to whites (if white data are not subject to error, say, to age 95). The q_x's for male cohorts aged 70, 75, and 80 in 1962, followed to 1990, are given in Figure 3-2. Female results are shown in Figure 3-3.

FIGURE 3-2

Mortality rates (q_x) for three male cohorts. SOURCE: Duke University Demographic Studies, analysis of 1960-1990 U.S. death certificate files.

FIGURE 3-3

Mortality rates (q_x) for three female cohorts. SOURCE: Duke University Demographic Studies, analysis of 1960-1990 U.S. death certificate files.

For both genders, nonwhite q_x drops below that for whites at age 81 (i.e., about 5 years below the crossover age observed by Kestenbaum, 1992, in Medicare data for 1987; Figures 3-2 and 3-3 were not adjusted to reflect data in Table 3-2). At late ages, rates are variable owing to small numbers. Since we used death certificate data from 1960 to 1990, we had to estimate the number of deaths above ages 98, 103, and 108, the highest ages observed for the three cohorts. We did this by using the data from the cohort 5 years older to fill in the mortality experience at ages higher than we observed in the younger cohort. We excluded deaths above age 118 from all computations. This minimizes between-cohort differences for the younger cohorts, since the experience used to close out tables is identical to that of older cohorts, cohorts likely to be initially smaller and to have higher mortality to late ages (Manton and Stallard, 1996; Stallard and Manton, 1995).

Table 3-3 shows gender-specific ratios of white to nonwhite mortality at every 5^th year of age for cohorts aged 70, 75, 80, 85, 90, and 95 in 1962. For both genders the gap for younger cohorts in the age range 80 to 90 narrows owing to declining white, and relatively stable nonwhite, mortality. For the cohort aged 70 in 1962, the white mortality excess at 85 is estimated to be 1.1 percent for females and 3.9 percent for males; at 95 the excess is 28 percent for both genders. The 28 percent white excess converts to a 22 percent nonwhite deficit, too large to be accounted for by any of the death certificate undercount ratios in Table 3-2. The ratios are sufficient to account for the 5-year difference in the crossover age in Kestenbaum (1992) and in the extinct cohort mortality rates in Figures 3-2 and 3-3.

TABLE 3-3

Mortality Ratios at Specific Ages for Each Cohort.

Having examined racial differences in cohort mortality, we then examined cross-sectional (1992) data on several causes of death to see at what ages the peak differences in mortality relative risks occur. Figure 3-4 shows the black to white relative risks for males and females for total and cause-specific mortality. The peak black risk for cancer occurs at age 35 to 44 for females (1.5 to 1) and 45 to 54 for males (2 to 1). For heart disease the peak occurs at age 25 to 34 for males (2.8 to 1) and 35 to 44 for females (3.9 to 1). For both males and females the peak black stroke risk occurs at age 35 to 44 (about 4.5 to 1 and 4.0 to 1, respectively). For total mortality in 1992 the peak black risk occurs at age 35 to 44 (2.5 to 1 for males and 2.7 to 1 for females). For all causes (except cancer) and total mortality, there is a decline by age 85 to a mortality ratio less than 1.0. Thus, the mortality patterns show declines in black-white ratios beginning no later than age 45 to 54 (National Center for Health Statistics, 1994).

FIGURE 3-4

Ratios of black to white mortality rates, by age, cause of death, and gender. SOURCE: Data from Kochanek and Hudson, 1994:Table 8.

Because of concerns about data quality, no U.S. analysis of racial survival based solely on population data is likely to resolve the issue of whether there is a black-white mortality crossover. However, there are several ways to increase the credibility of results. First, one can determine if population mortality analyses are consistent with differences in the age dependence of mechanisms causing morbidity (and subsequently, mortality) in Hispanics, blacks, and whites in epidemiological studies. Thus, the validity of the population results must be assessed in terms of the substance of the disease mechanisms whose mortality outcomes are observed in the population. As discussed in prior sections, there is ample evidence that the age trajectories of specific diseases for racial groups are not proportional and that certain mechanisms favor the largest mortality disadvantage for blacks in middle and late-middle ages, as found in Figure 3-4. Both mortality selection of frail persons and differences in the age dependence of the mechanisms of disease (which are documented in the section on morbidity) could cause the age trajectory of mortality for whites and nonwhites to cross over. A second way to validate population results is to compare them against national surveys where cause-specific and total mortality risks are assessed through independent data collection mechanisms.

The National Longitudinal Mortality Survey (NLMS) examined black-white differences in mortality. The NLMS used 10 samples (for 1978 to 1985) from the Census Bureau's Current Population Survey, which has a response rate of 96 percent. The NLMS samples covered 500,000 whites and 50,000 blacks aged 25 and above. Linkage to the National Death Index identified 32,508 deaths. The relative (black to white) risks for total mortality declined with age. For males the relative risks were 2.13, 1.67, and 1.04 for ages 25 to 44, 45 to 64, and 65 and above. For women the relative risks were 2.33, 1.82, and 1.16 for the same ages. Adjustments for income reduced male relative risks to 1.74, 1.30, and 0.98 and reduced female relative risks to 1.93, 1.48, and 1.16. For cause-specific mortality, cardiovascular disease showed the same age pattern of relative risk as total mortality (2.06, 1.59, and 0.93 for males; 4.07, 2.09, and 1.16 for females). For cancer there was no convergence for blacks up to age 65; there was a moderate convergence for black females after age 65. For all other causes of death there was a convergence for both genders. Income-adjusted relative risks for total mortality declined by 5-year categories from age 25 to age 85 and above, with convergence, then crossover, by age 60 for males and by 80 to 85 for females (Sorlie et al., 1992).

A similar study was done of Hispanic mortality using 12 Current Population Surveys. Of 700,000 persons aged 25 and above, 40,000 were Hispanic with 1,562 Hispanic deaths. The age patterns are different for Hispanics in that the age-adjusted standardized risk ratios for Hispanics to non-Hispanics were 0.74 for males and 0.82 for females. Age-specific risk ratios for total mortality were 0.97 (age 25 to 44), 0.73 (age 45 to 64), and 0.83 (age 65 and above). Risk ratios for cardiovascular disease declined with age for men (0.77 to 0.59) and increased moderately for females (0.72 to 0.84). Hispanic cancer ratios were lower at all ages (0.46, 0.52, and 0.76 for males; 0.46, 0.68, and 0.55 for females). This was confirmed in age-adjusted rates of cancer incidence, where the relative risk for white Hispanic males was 88 percent of that for white non-Hispanic males. The risk for black Hispanic males was only 71 percent of that for black non-Hispanic males. The cancer risk for black Hispanic males was only 85 percent of that for white Hispanic males (Trapido et al., 1994a). For females, both black and white Hispanic cancer risks were about 70 percent of that for black and white non-Hispanics, with black Hispanic risk about 78.5 percent that of white Hispanics (Trapido et al., 1994b). When the age-specific total mortality risks were adjusted for income, they dropped further: to 0.81 (age 25 to 44), 0.67 (age 45 to 64), and 0.66 (age 65 and above) for males; 0.64 (age 25 to 44), 0.74 (age 45 to 64), and 0.83 (age 65 and above) for females. One explanation of the Hispanic mortality advantage is their recent migrant status. Other explanations may involve differences in risk factors for specific diseases (e.g., while diabetes prevalence is high in Hispanics, so are HDL levels).

A final source of validating data is longitudinal studies of mortality in closed cohorts. Lew and Garfinkel (1990) assessed the mortality of 49,469 people passing age 75 from 1960 to 1987. There were 7,911 person-years of exposure for black males and 7,412 person-years of exposure for black females. Black male mortality rates dropped below white rates at about age 80 to 84. For females the crossover occurred between ages 85 to 89 and 90 to 94. In 4,107 people aged 65 and above in 1986 and followed for 5 years. Guralnik et al. (1993) found higher life and active life expectancies for black men and women above age 75.

WHITE-NONWHITE DIFFERENCES IN DISABILITY AND ACTIVE LIFE EXPECTANCY

Morbidity and disability are both powerful determinants of the age trajectory of mortality. As discussed above, hypertension, stroke, diabetes, renal disease, certain types of heart disease, and some cancers may create higher mortality for blacks in middle age. We compared the age dependence of race-specific disease mechanisms to total and cause-specific age patterns of mortality in several types of data in the preceding section. These represent two of three pairwise relations needed to evaluate the model in Figure 3-1. Here we evaluate the third linkage: the age relation of disability and mortality. We examined this using the 1982, 1984, and 1989 National Long Term Care Surveys (NLTCS) linked to Medicare mortality records for the period 1982 to 1991.

The NLTCS was used because large list samples of individuals were drawn from Medicare enrollment files and followed for up to 9.5 years. There are 30,308 distinct persons, not households, in the 1982, 1984, and 1989 longitudinal file; a total of 60,232 assessments were attempted, with 57,290 completed, an overall response rate of 95.1 percent. The Medicare list sample (which represents 98.4% or more of the U.S. elderly population) allows 100 percent of the sample to be followed for mortality. Both community and institutional persons are represented. In the three surveys, 16,485 detailed community interviews were completed. In 1984 and 1989, 3,100 detailed institutional interviews were done. About 11,000 deaths were recorded from 1982 to 1991 so there are a large number of deaths and person-years of exposure to assess—especially since people had to be aged 65 and older to be in the sample. Replenishment samples of about 4,900 people were drawn in 1984 and 1989 to replace the 65- to 66-year-olds (1984) or the 65- to 69-year-olds (1989) who would otherwise be missing owing to the aging of the original sample of those aged 65 and older. Age is recorded both from Medicare records and as reported by the sample person. If a sample person is too impaired to complete an in-person interview, the interview is conducted with a proxy. The date of birth is also recorded on the interview. Ages can be checked for temporal consistency for people who were interviewed on more than one date. A problem identified by Kestenbaum (1992), timely reporting of age at death, was handled by updating the Medicare mortality records 6 to 12 months after the time of the interview. Age reporting in NLTCS records should be of better quality than standard Medicare records because of validation by direct survey contact and the possibility of tracking people over time.

We analyzed these data in two steps. First, we examined the distribution of disability for whites and nonwhites, stratified by education to control for differences in socioeconomic status. These results are presented in Table 3-4; stratifications by age are found in Table 3-5. After examining white-nonwhite differences in the distribution of disability at specific ages, we modeled the dependence of mortality on disability.

TABLE 3-4

Disability Status of White and Nonwhite High School Graduates and Nongraduates Living in the Community, 1982-1989.

TABLE 3-5

Disability Status of White and Nonwhite Community Residents, Stratified by Age and Education, for 1982 and 1989.

As Table 3-4 shows, the proportion of nondisabled community residents declined 1.03 percent for white nongraduates, but increased 0.61 percent for nonwhite nongraduates from 1982 to 1989. For graduates the proportion of nondisabled whites increased 2.41 percent; for nonwhites it declined 0.41 percent. In general the proportion receiving help with three or more activities of daily living is higher for nongraduates. Stratified by education, race, and age, sample sizes for some estimates are small.

There were sizable changes from 1982 to 1989 in the age distribution of the population aged 65 and older. In Table 3-5, we present age-specific changes in disability for the education groups. There are large changes in the age distribution of disability across education. For whites aged 65 to 74 in 1982 there was little difference in disability between education groups. For nonwhites, the disabled group was 5.0 percent larger in 1982 for nongraduates. From 1982 to 1989, disability declined for white graduates. Whereas white nondisabled nongraduates increased 0.4 percent at age 65 to 74, white nondisabled graduates increased 2.9 percent from 1982 to 1989. Nonwhite, nondisabled graduates increased 1.3 percent.

For age 75 to 84, education effects were larger. For white graduates, the nondisabled proportion was 5.0 percent higher in 1982, 5.1 percent in 1989. For nonwhite graduates the nondisabled proportion was higher: 14.5 percent higher in 1982; 6.7 percent in 1989.

For whites aged 85 and older, disability increased 2.1 percent for graduates versus nongraduates in 1982—based on few (eight) cases. By 1989, with 30 persons in the disabled group, the effect reversed; disability declined 6.8 percent for white graduates, and there was a larger proportion (55.3%) nondisabled. For nonwhites, 85 and older who were graduates, sample sizes are small and estimates unstable.

To link disability to survival, we first identified dimensions of disability and then estimated individual scores on these dimensions. We identified the dimensions from multivariate analyses of 27 items in the 1982, 1984, and 1989 NLTCS (Manton et al., 1995). In the analysis, all years were pooled so that the dimensions identified describe the distribution of cases for any survey year. The analysis produced two types of coefficients. The coefficients describing the relation of the 27 variables to the six disability dimensions are listed in Table 3-6. In addition, we estimated scores for each person in the community sample each time he or she was interviewed. For persons with more than one interview, changes in the scores on the six dimensions can be assessed directly. Scores are estimated so that they sum to 1.0 across the six dimensions for each person at each time; that is, a person may be a partial member of multiple groups.

TABLE 3-6

Prevalence Rates for Select Functional Disabilities for Six Groups Identified in the 1982, 1984, and 1989 National Long Term Care Survey.

The six dimensions define six groups that can be characterized by examining which coefficients in Table 3-6 are large. The six groups are as follows:

1. Nondisabled, with modest physical limitations, a group with no impairment in activities of daily living or instrumental activities of daily living (except heavy housework) and only moderate difficulty doing a few physical tasks.
2. Active, a group with no impairments in activities of daily living or instrumental activities of daily living and few physical impairments.
3. IADL impaired with performance limitation, a group with a number of impairments in instrumental activities of daily living (but none in the activities of daily living except bathing) and significant physical impairment.
4. IADL impaired, a group with many impairments in the instrumental activities of daily living, including those related to cognitive performance (but none in the activities of daily living). Physical performance, however, is less impaired than group 3.
5. ADL impaired, a group with impairments in four of the six activities of daily living, impairment in many of the instrumental activities of daily living, and moderate physical impairment.
6. Frail, a group with impairments in almost all activities of daily living and instrumental activities of daily living.

This analysis produced six scores for each of the 16,485 community interviews conducted in 1982, 1984, and 1989. To generate scores for all 30,308 people in the longitudinal file, we made two additional calculations. First, there were about 35,700 people who screened out on one or more of the three surveys. These people had no chronic disabilities in activities of daily living or instrumental activities of daily living (the reason for screening out) for that time. Thus, they were assigned a score of 1.0 on the first dimension (i.e., were put into the nondisabled group). A plausible alternative would be to assign them a score of 1.0 on the second dimension (i.e., put them into the active group). We dealt with this uncertainty by combining the first two groups into a single “active/non-disabled” group, which appears later in Table 3-9 and Figure 3-7. Second, there were about 5,100 people in institutions in total in 1982, 1984, and 1989. Because the 1984 and 1989 detailed institutional surveys indicated that the average resident had 4.8 impairments in activities of daily living (no institutional interview was done for 1,992 institutionalized people in 1982), we created a seventh category for institutionalized people. Institutionalized persons received a 1.0 on this seventh category (and 0.0 on all other dimensions).

TABLE 3-9

Projected Survival Function (l_t), Life Expectancy (e_t), Active Survival (l_t[ḡ₁[t]+ḡ₂[t]]), and Group Prevalence Rates (ḡ_k[t]) by Age, Race, and Gender.

FIGURE 3-7

Estimates of race-specific prevalence rates for active/nondisabled groups (groups 1 and 2 combined) at each age by gender. SOURCE: Duke University Demographic Studies, analysis of 1982, 1984, and 1989 National Long Term Care Surveys.

With these scores and mortality records for 1982 to 1991, we estimated two types of equations. First, we calculated the changes in the disability scores by (1) interpolating monthly values of the disability scores for people with interviews at the beginning and end of an interval, and (2) using a “nearest neighbor” match for people who died in an interval to determine the monthly rate of change for the person most like the decedent. In addition, we weighted cases by replicating each case by the ratio of its sample weight to the base weight. When this was done we calculated a monthly transition matrix,

A second equation describes mortality as a quadratic function of the current scores (g_i(t)) and a term reflecting the dependence of mortality on age net of the changes in disability, that is,

The 7 × 7 matrix B contains hazard coefficients describing how the pairwise interactions of the g_ik(t),k = 1, …, 7, affect mortality. Those effects are multiplied by the exponential term eθ ^t, which represents the average effects of unobserved variables on mortality over age. As the g_ik(t) become more informative the effects of unobserved factors will decrease. This will decrease the value of θ. The matrix B_t is calculated for age t, as B_t = B · eθ ^t, making the hazard function age dependent.

With the dynamic and hazard equations estimated, difference equations can use the coefficients from these equations to calculate life-table parameters, for each age t, by applying first the mortality equation to the distribution of the g_ik(t), and then the dynamic equations (Manton et al., 1994).

Mortality coefficients for disability scores estimated for white and nonwhite females are given in Table 3-7. Disability-specific mortality is roughly similar for white and nonwhite females (the confidence bounds of nonwhite coefficients were broader). There is a consistently higher level of mortality for nonwhite institutionalized females, and there are fairly large differences for terms involving group 4. The exponent θ, indicating the annual percentage increase in mortality, conditional on the time-variable disability profile, is higher (6.9%) for white females than for nonwhite females (4.3%). The higher age rate of mortality increase for white females will produce a convergence of white and nonwhite female mortality although white females start with lower mortality at age 65.

TABLE 3-7

White and Nonwhite Female Quadratic Mortality Coefficients.

Table 3-8 shows mortality coefficients for white and nonwhite males. There are more differences in the coefficients for white and nonwhite males (e.g., 12.5 vs. 6.4 for the effect of group 3, with whites disadvantaged; 11.9 vs. 26.0 for group 4, with nonwhite males disadvantaged; 3.75 vs. 4.62 for active males) than for females. The θ for males differs for whites and nonwhites (i.e., conditional on disability, white male mortality increases 7.14% per year vs. 3.73% per year for nonwhites). The ratio of 1.91 for the male θ's (vs. 1.59 for females) suggests that the convergence of mortality with age is more rapid for males.

TABLE 3-8

White and Nonwhite Male Quadratic Mortality Coefficients.

Figure 3-5 plots the white and nonwhite, male and female, age-specific probabilities of death calculated from equation (2). Female probabilities of death converge. A crossover is evident only at late ages. For males the crossover occurs about age 83—or roughly halfway between the extinct cohort estimate (age 81) and Kestenbaum's (1992) estimate (age 86).

FIGURE 3-5

Estimates of age-specific mortality rates for whites and nonwhites, by gender. SOURCE: Duke University Demographic Studies, analysis of 1982, 1984, and 1989 National Long Term Care Surveys.

Figure 3-6 shows male and female survival probabilities (l_t's) normed to 100,000 at age 65. For males, l_t's converge at age 92-93. For females, although l_t's converge, there is no crossover.

FIGURE 3-6

Race-specific numbers of survivors at estimates of each age by gender if we assume 100,000 males and 100,000 females alive at age 65. SOURCE: Duke University Demographic Studies, analysis of 1982, 1984, and 1989 National Long Term Care Surveys.

The age trajectories of the mean scores (i.e.,ḡ_k(t)) for the first two groups (active and nondisabled with modest physical limitations) for white and nonwhite males and females are shown in Figure 3-7. For both nonwhite males and females, disability (lower functioning) is greater. This is reasonable if (1) the survival of higher proportions of nonwhites to late ages implies their greater survival at higher disability levels, and (2) acute diseases cause mortality to increase more rapidly with age for whites. This may be explicated by examining Table 3-4, where the functioning of white high school graduates improved from 1982 to 1989. In contrast, the disability of nonwhite graduates increased while that of nonwhite nongraduates decreased. This could be due to differences in mortality; that is, nonwhite graduates could appear to be less functional if greater proportions of them survive to higher ages.

Such differences suggest the existence of acute disease mortality, not represented by disability dynamics, but by the age dependence coefficients (θ). The average score in the active/nondisabled group is higher for whites than for nonwhites and, after a long decline to age 95, begins to increase, owing to whites' higher θ's (6.9% and 7.3%), which may cause impaired whites to die off at late ages more rapidly than disability sets in. In nonwhites, θ is smaller (i.e., 4.3% and 3.7%) so there is a relatively slower mortality increase in impaired groups. Another possibly important factor in the higher disability of nonwhites is body mass index, which is higher in black women and is strongly related to mobility disability in the NHANES I follow-up study (Launer et al., 1994). High past and current body mass indexes (> 27) were significant risk factors for incident disability in the young old, possibly because of effects on osteoarthritis. Current weight loss and past high body mass index ( 28) (but not current high body mass index) were significant for the incidence of disability in the oldest old. Thus, different age changes in function for whites and nonwhites can result from mortality interactions with several age-dependent factors.

Table 3-9 presents life tables and average disability scores for white and nonwhite males aged 65, 75, 85, 95, and 105 calculated using the dynamic (1) and mortality (2) equations. Whites have higher life expectancies and higher proportions active/nondisabled at age 65 than nonwhites. Nonwhite survival rates eventually converge with those for whites. At late ages the average active/nondisabled score is lower for nonwhites than for whites. Mortality is higher for whites at late ages owing to their larger θ's so that mortality selection at late ages tends to preserve, or slightly increase, the aggregate level of functioning for whites. A larger θ generates higher mortality for nondisabled white persons producing life expectancies close to those in Social Security Administration (1992) cohort life tables. The column marked l_t × [ḡ₁(t) + ḡ₂(t)] gives survival-weighted activity levels for the g_ik(t) summed for the first two groups; that is, survival and functioning are summarized in a single trajectory. Higher white activity levels occur in disability changes observed over a total of seven years, (i.e., from 1982 to 1989). During that time, while the overall prevalence of chronic disability declined in the United States (Manton et al., 1993), the declines in disability were larger for whites than for blacks (Clark and Wolinsky, 1996; Clark et al., 1996). Since the disability declines over the seven years of follow-up occur for most age groups represented in 1982, the life-table calculations in Table 3-9 represent the splicing together of disability and survival experience of distinct cohorts. The trajectories of disability score changes are representative of the changes manifest in the distinct cohort trajectories—as well as the changes between the cross-sectional distributions of disability observed in 1982 and 1989.

SUMMARY

We examined race-specific and ethnic-specific age patterns of health and mortality changes in three ways. First, we examined epidemiological data on racial and ethnic differences in the age dependence of the risk of several diseases. The diseases examined were osteoporosis, heart disease and stroke, cancers, and diabetes; they represent 70 percent of all U.S. mortality above age 65. Not represented were chronic pulmonary conditions, pneumonia and influenza, externally caused deaths (except hip fracture), and liver and selected other conditions. Since a number of major risk factors (diabetes, hypertension, atherosclerosis, anemia, osteoporosis) were examined, the conditions examined probably affect more than 70 percent of deaths above age 65. For example, pneumonia and influenza often cause mortality in persons with other chronic diseases; cardiac problems may affect pulmonary function, and diabetes and hypertension can affect many organs such as the kidney, liver, and brain. There are over 35 types of cancer, so we selected several to represent racial and ethnic differences in risk. For example, prostate and breast cancer are the most frequently diagnosed U.S. cancers; black males have a 20 percent greater incidence of prostate cancer than whites, and white females have a 30 percent greater incidence of breast cancer (NCHS, 1994).

First, we reviewed disease mechanisms. Both the accelerating effect of hypertension and diabetes on black mortality and the stimulating effect of atherosclerosis and other factors on white mortality at late ages suggest the plausibility of a physiological basis for mortality convergences and crossover. We did not discuss specific nonlethal diseases (e.g., osteoarthritis), although their chronic effects on health at later ages are reflected in the disability and mortality linkages.

Second, we examined both the quality of data used to determine the age trajectory of total and cause-specific mortality and the ages of maximum black-white mortality differences. For cancer, heart disease, and stroke, the peak relative risk occurred in middle age, with declines in relative risk beginning before age 55. This was consistent with the morbidity review.

Third, we examined the linkage of disability with mortality at late ages. In a model describing the interaction of disability and mortality, there was a more rapid increase in mortality with age for whites than for blacks. This occurred for both males and females and was consistent with the higher disability found in blacks at later ages. The age at crossover in the NLTCS data was similar to ages found in extinct cohort analyses of death certificate data and Kestenbaum's (1992) matched record study.

Thus, we were able to examine three data elements (age trajectory of morbidity, mortality-morbidity linkages, and disability-mortality linkages) to develop a picture of U.S. racial differences in the age dependence of mortality, morbidity, and disability to extreme ages.

REFERENCES

• Aronson MK, Ooi WL, Morgenstern H, Hafner A, Masur D, Crystal H, Frishman WH, Fisher D, Katzman R. Women, myocardial infarction, and dementia in the very old. Neurology. 1990;40:1102–1106. [PubMed: 2356012]
• Ayanian JZ. Race, class, and the quality of medical care. Journal of the American Medical Association. 1994;271(15):1207–1208. [PubMed: 8151880]
• Barker DJP, Godfrey KM, Osmond C, Bul A. The relation of fetal length, ponderal index and head circumference to blood pressure and the risk of hypertension in adult life. Pediatric and Perinatal Epidemiology. 1992;6:35–44. [PubMed: 1553316]
• Barker DJP, Meade TW, Fall CHD, Lee A, Osmond C, Phipps K, Stirling T. Relation of fetal and infant growth to plasma fibrinogen and factor VII concentrations in adult life. British Medical Journal. 1992;304:148–152. [PMC free article: PMC1881173] [PubMed: 1737158]
• Becker TM, Wheeler CM, McGough NS, Stidley CA, Parmenter CA, Dorin MH, Jordan SW. Contraceptive and reproductive risk for cervical dysplasia in Southwestern Hispanic and non-Hispanic white women. International Journal of Epidemiology. 1994;23(5):913–922. [PubMed: 7860171]
• Bild DE, Fitzpatrick A, Fried LP, Wong ND, Hann MN, Lyles M, Bovill E, Polak JF, Schulz R. Age-related trends in cardiovascular morbidity and physical functioning in the elderly: The cardiovascular health study. Journal of the American Geriatrics Society. 1993;41:1047–1056. [PubMed: 8409149]
• Boult C, Kane RL, Louis TA, Ibrahim JG. Forecasting the number of future disabled elderly using Markovian and mathematical models. Journal of Clinical Epidemiology. 1991;44:973–980. [PubMed: 1832442]
• Bowden M, Crawford J, Cohen HJ, Noyama O. A comparative study of monoclonal gammopathies and immunoglobulin levels in Japanese and United States elderly. Journal of the American Geriatrics Society. 1993;41:11–14. [PubMed: 8418116]
• Browner WS, Pressman AR, Nevitt MC, Cauley JA, Cummings SR. Association between low bone density and stroke in elderly women: The study of osteoporotic fractures. Stroke. 1993;24:940–946. [PubMed: 8322393]
• Browner WS, Seeley DG, Vogt TM, Cummings SR. Non-trauma mortality in elderly women with low bone density. Lancet. 1991;338:355–358. [PubMed: 1677708]
• Campbell AJ, Busby WJ, Robertson C. Over 80 years and no evidence of coronary heart disease: Characteristicsof a survivor group. Journal of the American Geriatrics Society. 1993;41:1333–1338. [PubMed: 8227916]
• Carmelli D, Robinette D, Fabsitz R. Concordance, discordance and prevalence of hypertension in World War II male veteran twins. Journal of Hypertension. 1994;12:323–328. [PubMed: 8021487]
• Cauley JA, Gutai JP, Kuller LH, Scott J, Nevitt MC. Black-white differences in serum sex hormones and bone mineral density. American Journal of Epidemiology. 1994;139(10):1035–1046. [PubMed: 8178783]
• Caulfield M, Lavender P, Farrall M, Munroe P, Lawson M, Turner P, Clark A. Linkage of the angiotensinogen gene to essential hypertension. New England Journal of Medicine. 1994;330(23):1629–1633. [PubMed: 8177268]
• Centers for Disease Control and Prevention. Mortality from congestive heart failure—United States 1980-1990. Journal of the American Medical Association. 1994;271(11):813–814. [PubMed: 8114224]
• Chan J, Cockram C, Nicholls M, Cheung C, Swaminathan R. Comparison of enalapril and nifedipine in treating non-insulin dependent diabetes associated with hypertension: One year analysis. British Medical Journal. 1992;305:981–985. [PMC free article: PMC1883996] [PubMed: 1458144]
• Chapuy MC, Arlot ME, Delmas PD, Meunier PJ. Effect of calcium and cholecalciferol treatment for three years on hip fractures in elderly women. British Medical Journal. 1994;308:1081–1082. [PMC free article: PMC2539939] [PubMed: 8173430]
• Clark DO, Stump T, Wolinsky F. Race and gender specific replication of five dimensions of health and disability. American Journal of Public Health. 1996 [PubMed: 10182409]
• Clark DO, Wolinsky F. Trends in disability and institutionalization among older blacks and whites. 1996 In review at American Journal of Public Health.
• Colantonio A, Kasl SV, Ostfeld AM. Level of function predicts first stroke in the elderly. Stroke. 1992;23(9):1355–1357. [PubMed: 1519292]
• Cooper R, Rotimi C. Hypertension in populations of West African origin: Is there a genetic predisposition? Journal of Hypertension. 1994;12:215–227. [PubMed: 8021474]
• Corder EH, Guess HA, Hulka BS, Friedman GD, Sadler M, Vollmer RT, Lobaugh B, Dreznar MK, Vogelman JH, Orenteigh N. Vitamin D and prostate cancer: A prediagnostic study with stored sera. Cancer Epidemiology Biomarkers Prevention. 1993;2:467–472. [PubMed: 8220092]
• Couderc R, Mahieux F, Bailleul S, Fenelon G, Mary R, Fermanian J. Prevalence of apolipoprotein E phenotypes in ischemic cerebrovascular disease: A case-control study. Stroke. 1993;24(5):661–664. [PubMed: 8488520]
• Cowie CC, Harris MI, Silverman RE, Johnson EW, Rust KF. Effect of multiple risk factors on differences between blacks and whites in the prevalence of non-insulin-dependent diabetes mellitus in the United States. American Journal of Epidemiology. 1993;137(7):719–732. [PubMed: 8484363]
• DeStefano F, Merritt RK, Anda RF, Casper ML, Eaker ED. Trends in nonfatal coronary heart disease in the United States, 1980 through 1989. Archives of Internal Medicine. 1993;153:2489–2494. [PubMed: 8215754]
• Dzau V. Cell biology and genetics of angiotensin in cardiovascular disease. Journal of Hypertension. 1994;12(suppl. 4):S3–S10. [PubMed: 7965271]
• Dzau V, Re R. Tissue angiotensin system in cardiovascular medicine: A paradigm shift. Circulation. 1994;89(1):493–498. [PubMed: 8281685]
• Eastell R, Yergery AL, Vieira NE, Cedel SL, Kumar R, Riggs BL. Interrelationship among vitamin D metabolism, true calcium absorption, parathyroid function, and age in women: Evidence of an age-related intestinal resistance to 1-25 dihydroxyvitamin A action. Journal of Bone Mineral Research. 1991;6:125. [PubMed: 2028834]
• Edelstein SL, Barrett-Connor E. Relation between body size and bone mineral density in elderly and women. American Journal of Epidemiology. 1993;138(3):160–169. [PubMed: 8356959]
• Eggen DA, Solberg LA. Variation of atherosclerosis with age. Laboratory Investigations. 1968;18:571–579. [PubMed: 5681199]
• Eggen DA, Strong JP, McGill HC. Coronary calcification: Relationship to clinically significant coronary lesions and race, sex, and topographic distribution. Circulation. 1965;32:948–955. [PubMed: 5845254]
• Elledge RM, Clark GM, Chamness GC, Osbourne CK. Tumor biologic factors and breast cancer prognosis among white, Hispanic, and black women in the United States. Journal of the National Cancer Institute. 1994;86(9):705–712. [PubMed: 7908990]
• Ferrières J, Sing CF, Roy M, Davignon J, Lussier-Cacan S. Apolipoprotein E polymorphism and heterozygous familial hypercholesterolemia: Sexspecific effects. Arteriosclerosis and Thrombosis. 1994;14:1553–1560. [PubMed: 7918304]
• Feskens EJM, Kromhout D. Hyperinsulinemia, risk factors, and coronary heart disease: The Zutphen Elderly Study. Arteriosclerosis and Thrombosis. 1994;14:1641–1647. [PubMed: 7918315]
• Flegal K, Ezzati TM, Harris MI. Prevalence of diabetes in Mexican Americans, Cubans, and Puerto Ricans from the Hispanic Health and Nutrition Examination Survey, 1982-1989. Diabetes Care. 1991;14(suppl. 3):628–638. [PubMed: 1914812]
• Fogel RW. Economic growth, population theory, and physiology: The bearing of long-term processes on the making of economic policy. American Economic Review. 1994;84:369–395.
• Fontbonne A. Insulin: A sex hormone for cardiovascular risk? Circulation. 1991;84(3):1442–1444. [PubMed: 1884467]
• Fontbonne A, Tchobroutsky G, Eschwège E, Richard JL, Claude JR, Rosselin GE. Coronary heart disease mortality risk: Plasma insulin level is a more sensitive marker than hypertension or abnormal glucose tolerance in overweight males. The Paris Prospective Study. International Journal of Obesity. 1988;12:557–565. [PubMed: 3069769]
• Gann PH, Hennekens CH, Sacks FM, Grodstein F, Giovannucci EL, Stampfer MJ. Prospective study of plasma fatty acids and risk of prostate cancer. Journal of the National Cancer Institute. 1994;86:281–286. [PubMed: 8158682]
• Gayton JR, Dahlen G, Patsch W, Kautz JA, Gotto AM. Relationship of plasma lipoprotein Lp(a) levels to race and to apolipoprotein B. Arteriosclerosis. 1985;5:265–272. [PubMed: 3158297]
• Ghali JK, Cooper R, Ford E. Trends in hospitalization rates for heart failure in the United States, 1973-1986. Archives of Internal Medicine. 1990;150:769–776. [PubMed: 2327838]
• Griendling K, Murphy T, Alexander R. Molecular biology of the renin-angiotensin system. Circulation. 1993;87(6):1816–1828. [PubMed: 8389259]
• Grisso JA, Kelsey JL, Strom BL, O'Brien LA, Maislin G, LaPann K, Samelson L, Hoffman S. Risk factors for hip fracture in black women. New England Journal of Medicine. 1994;330(22):1555–1559. [PubMed: 8177244]
• Guccione AA, Felson DT, Anderson JJ, Anthony JM, Zhang Y, Wilson PWF, Kelly-Hayes M, Wolf PA, Kreger BE, Kannel WB. The effects of specific medical conditions on the functional limitations of elders in the Framingham Study. American Journal of Public Health. 1994;84:351–358. [PMC free article: PMC1614827] [PubMed: 8129049]
• Guralnik JM, Land KC, Blazer D, Fillenbaum GG, Branch LG. Educational status and active life expectancy among older blacks and whites. New England Journal of Medicine. 1993;329(2):110–116. [PubMed: 8510687]
• Guzmán MA, McMahan CA, McGill HC, Strong JP, Tejada C, Restrepo C, Eggen DA, Robertson WB, Solberg LA. Selected methodologic aspects of the International Atherosclerosis Project. Laboratory Investigations. 1968;18:479–497. [PubMed: 5681192]
• Haffner SM, Hazuda HP, Stern MP, Patterson JK, Van Heuven WAJ, Fong D. Effect of socioeconomic status on hyperglycemia and retinopathy levels in Mexican-Americans with NIDDM. Diabetes Care. 1989;12:128–134. [PubMed: 2702895]
• Harris MI, Klein R, Welborn TA, Knuiman MW. Onset of NIDDM occurs at least 4 to 7 years before clinical diagnosis. Diabetes Care. 1992;15:815–819. [PubMed: 1516497]
• Heller DA, de Faire U, Pedersen NL, Dahlén G, McClearn GE. Genetic and environmental influences on serum lipid levels in twins. New England Journal of Medicine. 1993;328(16):1150–1156. [PubMed: 8455681]
• Helmhold M, Bigge J, Muche R, Mainoo J, Thiery J, Seidel D, Armstrong VW. Contribution of the apo(a) phenotype to plasma Lp(a) concentrations shows considerable ethnic variation. Journal of Lipid Research. 1991;32:1919–1928. [PubMed: 1840066]
• Hunter CP, Redmond CK, Chen VW, Austin DF, Greenberg RS, Correa P, Muss HB, Forman MR, Wesley MN, Blacklow RS, Kurman RJ, Dignam JJ, Edwards BK, Shapiro S. Breast cancer: Factors associated with stage at diagnosis in black and white women. Journal of the National Cancer Institute. 1993;85(14):1129–1137. [PubMed: 8320742]
• Jeunemaitre X, Soubrier F, Kotelevstev YV, Lifton RP, Williams CS, Charru A, Hunt SC, Hopkins PN, Williams RR, Lalouel JM, Corvol P. Molecular basis of human hypertension: Role of angiotensinogen. Cell. 1992;71:169–180. [PubMed: 1394429]
• Joffe BI, Panz VR, Wing JR, Raal FJ, Seftel HC. Pathogenesis of non-insulin-dependent diabetes mellitus in the black population of southern Africa. Lancet. 1992;340:460–462. [PubMed: 1354791]
• Kahn KL, Pearson ML, Harrison ER, Desmond KA, Rogers WH, Rubenstein LV, Brook RH, Keeler EB. Health care for black and poor hospitalized Medicare patients. Journal of the American Medical Association. 1994;271(15):1169–1174. [PubMed: 8151874]
• Kaul L, Heshmat MY, Kovi J, Jackson MA, Jackson AG, Jones GW, Edson M, Enterline JP, Worrell RG, Perry SL. The role of diet in prostate cancer. Nutritional Cancer. 1987;9:123–128. [PubMed: 3562290]
• Keene GS. Mortality and morbidity after hip fractures. British Medical Journal. 1993;307:1248–1250. [PMC free article: PMC1679389] [PubMed: 8166806]
• Keil JE, Sutherland SE, Knapp RG, Lackland DT, Gazes PC, Tyroler HA. Mortality rates and risk factors for coronary disease in black as compared with white men and women. New England Journal of Medicine. 1993;329:73–78. [PubMed: 8510705]
• Kellie SE, Brody JA. Sex specific and race specific hip fracture rates. American Journal of Public Health. 1990;80:326–380. [PMC free article: PMC1404674] [PubMed: 2305917]
• Kestenbaum B. A description of the extreme aged population based on improved Medicare enrollment data. Demography. 1992;29(4):565–581. [PubMed: 1483542]
• Kiechl S, Aichner F, Gerstenbrand F, Egger G, Mair A, Rungger G, Spögler F, Jarosch E, Oberhollenzer F, Willeit J. Body iron stores and presence of carotid atherosclerosis: Results from the Bruneck Study. Arteriosclerosis and Thrombosis. 1994;14:1625–1630. [PubMed: 7918313]
• Knapp RG, Sutherland SE, Keil JE, Rust PF, Lackland DT. A comparison of the effects of cholesterol on CHD mortality in black and white women: Twenty-eight years of follow-up in the Charleston Heart Study. Journal of Clinical Epidemiology. 1992;45(10):1119–1129. [PubMed: 1474408]
• Ko W, Gold JP, Lazzaro R, Zelano JA, Lang S, Isom OW, Kreiger KH. Survival analysis of octogenarian patients with coronary artery disease managed by elective coronary artery bypass surgery versus conventional medical treatment. Circulation. 1992;86(suppl. II):191–197. [PubMed: 1423999]
• Kochanek MA, Hudson BL. Monthly Vital Statistics Report. 6. Vol. 43. Hyattsville, MD: National Center for Health Statistics; 1994. Advance report of final mortality statistics, 1992.
• Lakka TA, Nyyssonen K, Salonen JT. Higher levels of conditioning leisure time physical activity are associated with reduced levels of stored iron in Finnish men. American Journal of Epidemiology. 1994;140(2):148–160. [PubMed: 8023803]
• Launer LJ, Harris T, Rumpel C, Madans J. Body mass index, weight change, and risk of mobility disability in middle-aged and older women. Journal of the American Medical Association. 1994;271(14):1093–1098. [PubMed: 8151851]
• Lew EA, Garfinkel L. Mortality after age 76 and older in the Cancer Prevention Study (CPSI). CA: A Cancer Journal for Clinicians. 1990;40:211–224. [PubMed: 2114200]
• Lin RS, Lee WC, Lee YT, Chou P, Chung-Fu C. Maternal role in type 2 diabetes mellitus: Indirect evidence for a mitochondrial inheritance. International Journal of Epidemiology. 1994;23(5):886–890. [PubMed: 7860167]
• Lindpaintner K. Genes, hypertension, and cardiac hypertrophy. New England Journal of Medicine. 1994;330(23):1678–1679. [PubMed: 8177273]
• Lipton RB, Liao Y, Cao G, Cooper RS, McGee D. Determinants of incident non-insulin-dependent diabetes mellitus among blacks and whites in a national survey: The NHANES I epidemiologic follow-up study. American Journal of Epidemiology. 1993;138(10):826–839. [PubMed: 8237971]
• Löwick MRH, Odink J, Kok FJ, Ockhuizen T. Hematocrit and cardiovascular risk factors among elderly men and women. Gerontology. 1992;38:205–213. [PubMed: 1427119]
• MacMahon B, Cole P, Brown J. Etiology of human breast cancer: A review. Journal of the National Cancer Institute. 1973;50:21–42. [PubMed: 4571238]
• Manton KG, Corder L, Stallard E. Estimates of change in chronic disability and institutional incidence and prevalence rates in the U.S. elderly population from the 1982, 1984, and 1989 National Long Term Care Survey. Journal of Gerontology Social Sciences. 1993;47(4):S153–S166. [PubMed: 8315240]
• Manton KG, Patrick CH, Johnson K. Health differentials between blacks and whites: Recent trends in mortality and morbidity. Milbank Quarterly. 1987;65:129–199. [PubMed: 3327005]
• Manton KG, Stallard E. Demographics (1950-1987) of breast cancer in birth cohorts of older women. Journal of Gerontology. 1992;47(special issue):32–42. [PubMed: 1430880]
• Manton KG, Stallard E. Longevity in the U.S.: Age and sex specific evidence on life span limits from mortality patterns, 1960-1990. Journal of Gerontology: Biological Sciences. 1996;51A:B362–B375. [PubMed: 8808985]
• Manton KG, Stallard E, Singer B. Projecting the future size and health status of the U.S. elderly population. In: Wise D, editor. Statistics of the Economics of Aging. Chicago: University of Chicago Press; 1994. pp. 41–77.
• Manton KG, Stallard E, Woodbury MA. Home health and skilled nursing facilities use: 1982 to 1990. Health Care Financing Review. 1995;16:155–186. [PMC free article: PMC4193485] [PubMed: 10140152]
• Manton KG, Woodbury MA, Wrigley JM, Cohen HJ. Multivariate procedures to describe clinical staging of melanoma. Methods of Information in Medicine. 1991;30:111–116. [PubMed: 1857245]
• Marcovina SM, Albers JJ, Jacobs DR, Perkins LL, Lewis CE, Howard BV, Savage P. Lipoprotein[a] concentrations and apolipoprotein[a] phenotypes in Caucasians and Africans: The CARDIA study. Arteriosclerosis and Thrombosis. 1993;13:1037–1045. [PubMed: 8318505]
• Marenberg ME, Risch N, Berkman LF, Floderus B, de Faire U. Genetic susceptibility to death from coronary heart disease in a study of twins. New England Journal of Medicine. 1994;330(15):1041–1046. [PubMed: 8127331]
• Mariotti S, Sansoni P, Barbesino G, Caturegli P, Monti D, Cossarizza A, Giacomelli T, Passeri G, Fagiolo U, Pinchera A, Franceschi C. Thyroid and other organ-specific autoantibodies in healthy centenarians. Lancet. 1992;339:1506–1508. [PubMed: 1351187]
• Marshall JA, Hamman RF, Baxter J, Mayer EJ, Fulton DL, Orleans M, Rewers M, Jones RH. Ethnic differences in risk factors associated with the prevalence of non-insulin-dependent diabetes mellitus. American Journal of Epidemiology. 1993;137(7):706–718. [PubMed: 8484362]
• Materson B, Reda D, Cushman W, Massie B, Freis E, Kochar M, Hamburger R, Fye C, Lakshman R, Gottdiener J, Ramirez E, Henderson W. Single-drug therapy for hypertension in men: A comparison of six antihypertensive agents with placebo. New England Journal of Medicine. 1993;328(13):914–921. [PubMed: 8446138]
• Mathiesen E, Hommel E, Giese J, Parving H. Efficacy of captopril in postponing nephropathy in normotensive insulin dependent diabetic patients with microalbuminuria. British Medical Journal. 1991;303:81–87. [PMC free article: PMC1670656] [PubMed: 1860008]
• Matthews KA, Wing RR, Kuller LH, Meilahn EN, Plantinga P. Influence of the perimenopause on cardivascular risk factors and symptoms of middle-aged healthy women. Archives of Internal Medicine. 1994;154:2349–2355. [PubMed: 7944857]
• Mayberry R, Stoddard-Wright C. Breast cancer risk factors among black women and white women: Similarities and differences. American Journal of Epidemiology. 1992;136(12):1445–1456. [PubMed: 1288274]
• McCance D, Pettitt D, Hanson R, Jacobsson L, Knowler W, Bennett P. Birth weight and non-insulin dependent diabetes: Thrifty genotype, thrifty phenotype, or surviving small baby genotype? British Medical Journal. 1994;308:942–945. [PMC free article: PMC2539758] [PubMed: 8173400]
• McKeigue P, Laws A, Chen Y, Marmot M, Reaven G. Relation of plasma triglyceride and Apo B levels to insulin-mediated suppression of nonesterified fatty acids: Possible explanation for sex differences in lipoprotein pattern. Arteriosclerosis and Thrombosis. 1993;13(8):1187–1192. [PubMed: 8343493]
• McKeigue PM, Shah B, Marmot MG. Relation of central obesity and insulin resistance with high diabetes prevalence and cardiovascular risk in South Asians. Lancet. 1991;337:382–386. [PubMed: 1671422]
• Melby CL, Toohey ML, Cebrick J. Blood pressure and blood lipids among vegetarian, semivegetarian, and nonvegetarian African Americans. American Journal of Clinical Nutrition. 1994;59:103–109. [PubMed: 8279389]
• Modan M, Halkin H, Or J, Karaski A, Drory Y, Fuchs Z, Lusky A, Chetrit A. Hyperinsulinemia, gender and risk of atherosclerotic cardiovascular disease. Circulation. 1991;84:1165–1175. [PubMed: 1884447]
• Moon J, Bandy B, Davison A. Hypothesis: Etiology of atherosclerosis and osteoporosis: Are imbalances in the calciferol endocrine system implicated? Journal of the American College of Nutrition. 1992;11(5):567–583. [PubMed: 1452956]
• National Center for Health Statistics. Health, United States, 1994. Hyattsville, MD: Public Health Service; 1994.
• Okamoto Y. Health care for the elderly in Japan: Medicine and welfare in an aging society facing a crisis in long term care. British Medical Journal. 1992;305:403–405. [PMC free article: PMC1883155] [PubMed: 1392924]
• Ornish D, Brown SE, Scherwitz LW, Billings JH, Armstrong WT, Ports TA, McLanahan SM, Kirkeeide RL, Brand RJ, Gould KL. Can lifestyle changes reverse coronary heart disease? The Lifestyle Heart Trial. Lancet. 1990;336:129–133. [PubMed: 1973470]
• Parra HJ, Luyéré I, Bouramoué C, Demarquilly C, Fruchart JC. Black-white differences in serum Lp(a) lipoprotein levels. Clinica Chimica Acta. 1987;168:27–31. [PubMed: 3665103]
• Pathobiological Determinants of Atherosclerosis in Youth Research Group. Natural history of aortic and coronary atherosclerotic lesions in youth: Findings from the PDAY Study. Arteriosclerosis and Thrombosis. 1993;13(9):1291–1298. [PubMed: 8364013]
• Perez-Stable EJ, Otero-Sabogal R, Sabogal F, McPhee SJ, Hiatt RA. Self-reported use of cancer screening tests among Latinos and Anglos in a prepaid health plan. Archives of Internal Medicine. 1994;154:1073–1081. [PubMed: 8185420]
• Perneger T, Klag M, Feldman H, Whelton P. Projections of hypertension-related renal disease in middle-aged residents of the United States. Journal of the American Medical Association. 1993;269(10):1272–1277. [PubMed: 8437305]
• Perry H, Miller D, Morley J, Horowitz M, Kaiser F, Perry H Jr., Jensen J, Bentley J, Boyd S, Kraenzle D. A preliminary report of vitamin D and calcium metabolism in older Africans Americans. Journal of the American Geriatrics Society. 1993;41(6):612–616. [PubMed: 8505457]
• Peterson ED, Wright SM, Daley J, Thibault GE. Racial variation in cardiac procedure use and survival following acute myocardial infarction in the Department of Veterans Affairs. Journal of the American Medical Association. 1994;271:1175–1180. [PubMed: 8151875]
• Pollitzer WS, Anderson JB. Ethnic and genetic differences in bone mass: A review with a hereditary vs. environmental perspective. American Journal of Clinical Nutrition. 1989;50:1244. [PubMed: 2688394]
• Preston SH, Elo IT, Rosenwaike I, Hill M. African-American mortality at older ages: Results of a matching study. Demography. 1996;33:193–209. [PubMed: 8827165]
• Psaty B, Savage P, Tell G, Polak J, Hirsch C, Gardin J, McDonald R. Temporal patterns of antihypertensive medication use among elderly patients: The Cardiovascular Health Study. Journal of the American Medical Association. 1993;270(15):1837–1841. [PubMed: 8105112]
• Qualheim R, Rostand S, Kirk K, Luke R. Changing patterns of end-stage renal disease due to hypertension. American Journal of Kidney Disease. 1991;18:336–343. [PubMed: 1882825]
• Radl J, Sepers JM, Skvaril F, Morell A, Hijmans W. Immunoglobulin patterns in humans over 95 years of age. Clinical Experimental Immunology. 1975;22:84–90. [PMC free article: PMC1538323] [PubMed: 1212818]
• Ragland K, Selvin S, Merrill D. Black-white differences in stage-specific cancer survival: Analysis of seven selected sites. American Journal of Epidemiology. 1991;133:672–682. [PubMed: 2018022]
• Reed T, Quiroga J, Selby JV, Carmelli D, Christian JC, Fabsitz RR, Grim CE. Concordance of ischemic heart disease in the NHLBI twin study after 14-18 years of follow-up. Journal of Clinical Epidemiology. 1991;44:797–805. [PubMed: 1941031]
• Riggs BL, Melton LJ III. Drug therapy: The prevention and treatment of osteoporosis. New England Journal of Medicine. 1992;327(9):620–627. [PubMed: 1640955]
• Ross RK, Henderson BE. Do diet and androgens alter prostate cancer risk via a common etiologic pathway? Journal of the National Cancer Institute. 1994;86(4):252–254. [PubMed: 8158675]
• Rotimi C, Cooper R, Cao G, Sundarum C, McGee D. Familial aggregation of cardiovascular diseases in African-American pedigrees. Genetic Epidemiology. 1994;11:397–407. [PubMed: 7835686]
• Rutan G, Kuller L, Neaton J, Wentworth DN, McDonald RH, McFate-Smith W. Mortality associated with diastolic hypertension and isolated systolic hypertension among men screened for the Multiple Risk Factor Intervention Trial. Circulation. 1988;77(3):504–514. [PubMed: 3277736]
• Salive ME, Cornoni-Huntley J, Philips CL, Wallace RB, Ostfeld AM, Cohen HJ. Anemia and hemoglobin levels in older persons: Relationship with age, gender, and health status. Journal of the American Geriatrics Society. 1992;40:489–496. [PubMed: 1634703]
• Salonen J, Nyyssonen K, Korpela H, Tuomilehto J, Seppanen R, Salonen R. High stored iron levels are associated with excess risk of myocardial infarction in eastern Finnish men. Circulation. 1992;86(3):803–811. [PubMed: 1516192]
• Sandholzer C, Hallman DM, Saha N, Sigurdsson G, Lackner C, Császár A, Boerwinkle E, Utermann G. Effects of apolipoprotein(a) size polymorphism on the lipoprotein(a) concentration in 7 ethnic groups. Human Genetics. 1991;86:607–614. [PubMed: 2026424]
• Schaefer E, Lamon-Fava S, Johnson S, Ordovas J, Schaefer M, Castelli W, Wilson P. Effects of gender and menopausal status on the association of apolipoprotein E phenotype with plasma lipoprotein levels: Results from the Framingham Offspring Study. Arteriosclerosis and Thrombosis. 1994;14:1105–1113. [PubMed: 8018666]
• Schunkert H, Hense HW, Holmer SR, Stender M, Perz S, Keil U, Lovell BH, Riegger GA. Association between a deletion polymorphism of the angiotensin-converting-enzyme gene and left ventricular hypertrophy. New England Journal of Medicine. 1994;330:1634–1638. [PubMed: 8177269]
• Seelig MS. Vitamin D and cardiovascular, renal, and brain damage in infancy and childhood. Annals of the New York Academy of Sciences. 1969;147:537–582. [PubMed: 4902583]
• Selby JV, Austin MA, Sandholzer C, Quesenberry CP, Zhang D, Mayer E, Utermann G. Environmental and behavioral influences on plasma lipoprotein(a) concentration in women twins. Preventive Medicine. 1994;23:345–353. [PubMed: 8078856]
• Sempos CT, Cleeman JI, Carroll MD, Johnson CL, Bachorik PS, Gordon DJ, Burt VL, Briefel RR, Brown CD, Lippel K, Rifkind BM. Prevalence of high blood cholesterol among US adults: An updatebased on guidelines from the second report of the National Cholesterol Education Program adult treatment panel. Journal of the American Medical Association. 1993;269(23):3009–3014. [PubMed: 8501843]
• Sempos CT, Cooper R, Kovar MG, McMillen M. Divergence of the recent trends in coronary mortality for the four major race-sex groups in the United States. American Journal of Public Health. 1988;78:1422–1427. [PMC free article: PMC1350232] [PubMed: 3177717]
• Sheu WHH, Sheigh SM, Fuh MMT, Shen DDC, Jeng CY, Chen YDI, Reaven GM. Insulin resistance, glucose intolerance, and hyperinsulinemia: Hypertriglyceridemia versus hypercholesterolemia. Arteriosclerosis and Thrombosis. 1993;13(3):367–370. [PubMed: 8443140]
• Shintani S, Kikuchi S, Hamaguchi H, Shiigai T. High serum lipoprotein(a) levels are an independent risk factor for cerebral infarction. Stroke. 1993;24(7):965–969. [PubMed: 8322396]
• Singh J, Dudley AW, Kulig KA. Increased incidence of monoclonal gammopathy of undetermined significance in blacks and its age-related differences with whites on the basis of a study of 397 men and 1 woman in a hospital setting. Journal of Laboratory Clinical Medicine. 1990;116:785–789. [PubMed: 2246554]
• Singh GK, Kochanek KD, McDonan MF. Monthly Vital Statistics Report. 3 supp. Vol. 45. Hyattsville, MD: National Center for Health Statistics; 1996. Advance report of final mortality statistics, 1994.
• Skoog I, Nilsson L, Palmertz B, Andreasson L, Svanborg A. A population-based study of dementia in 85-year-olds. New England Journal of Medicine. 1993;328(3):153–158. [PubMed: 8417380]
• Smith S, Julius S, Jamerson K, Amerena J, Schrok N. Hematocrit levels and physiologic factors in relationship to cardiovascular risk in Tecumseh, Michigan. Journal of Hypertension. 1994;12:455–462. [PubMed: 8064170]
• Social Security Administration. Life Tables for the United States Social Security Area 1900-2080. Baltimore, MD: Social Security Administration; Aug, 1992. Actuarial Study 107. SSA publication 11-11536.
• SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. New England Journal of Medicine. 1991;325(5):293–302. [PubMed: 2057034]
• Sorlie P, Rogot E, Anderson R, Johnson NJ, Backlund E. Black-white mortality differences by family income. Lancet. 1992;340:346–350. [PubMed: 1353813]
• Srinivasan SR, Dahlen GH, Jarpa RA, Webber LS, Berenson GS. Racial (black-white) differences in serum lipoprotein(a) distribution and its relation to parental myocardial infarction in children: Bogalusa Heart Study. Circulation. 1991;84:160–167. [PubMed: 1829398]
• Stallard E, Manton KG. The trajectory of mortality from age 80 to 110. Presented at the Southern Demographic Association Annual Meeting; October 19-20, 1995; Richmond, VA. 1995.
• Strong JP. Atherosclerosis in human population. Atherosclerosis. 1972;16:193–201. [PubMed: 4634813]
• Strong JP, McGill HC. The natural history of aortic atherosclerosis: Relationship to race, sex, and coronary lesions in New Orleans. Exploratory Molecular Pathology. 1963;(suppl.):15–27. [PubMed: 14079281]
• Sullivan JL. The iron paradigm of ischemic heart disease. American Heart Journal. 1989;117:1177–1188. [PubMed: 2653014]
• Sutherland S, Gazes P, Keil J, Gilbert G, Knapp R. Electrocardiographic abnormalities and 30-year mortality among white and black men of the Charleston Heart Study. Circulation. 1993;88(6):2685–2692. [PubMed: 8252679]
• Svetkey LP, George LK, Burchett BM, Mortan PA, Blazer DG. Black/white differences in hypertension in the elderly: An epidemiologic analysis in central North Carolina. American Journal of Epidemiology. 1993;137:64–73. [PubMed: 8434574]
• Takata H, Suzuki M, Ishii T, Sekiguchi S, Iri H. Influence of major histocompatability complex region genes on human longevity among Okinawan-Japanese centenarians and nonagenarians. Lancet. 1987;2:824–826. [PubMed: 2889033]
• Taussig HB. Possible injury to the cardiovascular system from vitamin D. Annals of Internal Medicine. 1966;65:1195–1200. [PubMed: 5333232]
• Taylor HA, Chaitman BR, Rogers WJ, Kern MJ, Terrin ML, Aguirre FV, Sopko G, McMachon R, Ross RN, Bovill EC., TIMI investigators. Race and prognosis after myocardial infarction: Results of the Thrombolysis in Myocardial Infarction (TIMI), phase II trial. Circulation. 1993;88:1484–1494. [PubMed: 8403296]
• Tiret L, de Knijff P, Menzel HJ, Ehnholm C, Nicaud V, Havekes LM., for the EARS Group. ApoE polymorphism and predisposition to coronary heart disease in youths of different European populations: The EARS study. Arteriosclerosis and Thrombosis. 1994;14:1617–1624. [PubMed: 7918312]
• Trapido EJ, Chen F, Davis K, Lewis N, MacKinnon JA. Cancer among Hispanic males in south Florida. Archives of Internal Medicine. 1994;154:177–185. [PubMed: 8285813]
• Trapido E, Chen F, Davis K, Lewis N, MacKinnon J, Strait P. Cancer in south Florida Hispanic women: A 9-year assessment. Archives of Internal Medicine. 1994;154:1083–1088. [PubMed: 8185421]
• Trapido E, McCoy C, Stein N, Engel S, Zavertnik J, Comerford M. Epidemiology of cancer among Hispanic males. Cancer. 1990;65:1657–1662. [PubMed: 2311074]
• Valentine R, Grayburn P, Vega G, Grundy S. Lp(a) lipoprotein is an independent, discriminating risk factor for premature peripheral atherosclerosis among white men. Archives of Internal Medicine. 1994;154:801–806. [PubMed: 8147686]
• Wallace DC. Mitochondrial genetics: A paradigm for aging and degenerative diseases. Science. 1992;256:628–632. [PubMed: 1533953]
• White P. Disorders of aldosterone biosynthesis and action. New England Journal of Medicine. 1994;331(4):250–258. [PubMed: 8015573]
• Williams G. Management of non-insulin-dependent diabetes mellitus. Lancet. 1994;343:95–100. [PubMed: 7903785]
• Witteman J, Grobbee D, Valkenburg H, van Hemert A, Stijnen T, Burger H, Hofman A. J-shaped relation between change in diastolic blood pressure and progression of aortic atherosclerosis. Lancet. 1994;343:504–507. [PubMed: 7906758]
• Yantani R, Chigusa I, Akazaki K, Stemmermann GN, Welsh RA, Correa P. Geographic pathology of latent prostatic carcinoma. International Journal of Cancer. 1982;29:611–616. [PubMed: 7107064]
• Yki-Järvinen H. Pathogenesis of non-insulin-dependent diabetes mellitus. Lancet. 1994;343:91–94. [PubMed: 7903784]