Copyright Gerontological Society of America, Incorporated Jul 1995CHRONIC (90 days) disability and institutional prevalence in the U.S. elderly population declined from 24.54 percent in 1982 (standardized to the 1989 age distribution in 5-year age groups to age 95 + ) to 22.56 percent in 1989 -- an absolute decline of 1.98 percentage points and a relative decline of 8. 1 percent (i.e., .2256/.2454 = .919). This decline was significant (t = 4.7), occurred for community residents with IADL and ADL disabilities, and for persons in institutions (of whom about 97% had one or more ADLs), was relatively larger at later ages, and was not obviously explicable by survey design or measurement factors (Manton, Corder, and Stallard, 1993a). Such declines may be due to factors with known trends such as socioeconomic status. Disability trends were negatively related to increases in cohort education and income (Maddox and Clark, 1992; Suzman, Manton, and Willis, 1992). The proportion of males aged 85 to 89 with 9 + years of education is projected to increase from 37.9 percent in 1980 to 80.0 percent in 2015 (Preston, 1993). For females the projected increase is from 43.3 to 86.3 percent.
However, even if chronic disability prevalence declines, morbidity prevalence could have been stable, or increased, if medical interventions were successful in extending survival and in slowing a disease's progression so that the age at disability onset for persons with a disease increased. For example, in clinical studies of congestive heart failure (CHF), ACE-II inhibitors (a type of antihypertensive drug) not only improved survival but also function (e.g., SOLVD Investigators, 1991). Thus, both the time lived with CHF, and with less disability, increased. Disease prevention, a delay in age at onset, or a disease cure all can cause morbidity prevalence to decline. With chronic diseases whose incidence increases exponentially with age, a deferral in the age at onset may greatly decrease prevalence. For example, the incidence of cataracts increases exponentially with age, doubling in 10 years. Slowing the development of cataracts (e.g., by use of antioxidants) and increasing the age at which they appear by 10 years would reduce their prevalence by half -- if life expectancy did not change (Taylor, 1993). Since treatment of cataracts costs Medicare about $3.2 billion annually, this could significantly reduce costs.
To determine how morbidity prevalence changed as disability declined from 1982 to 1989, we examined 16 medical conditions reported by respondents to the 1982, 1984, and 1989 National Long Term Care Surveys (NLTCS). In all three surveys, essentially the same list of medical conditions (one subdiagnosis was added in 1989) was referred to in the in-person community interview. We examined how the prevalence of medical conditions changed by age and sex, for ADL disabled, IADL disabled, and nondisabled U.S. elderly persons from 1982 to 1989. We also examined morbidity prevalence after standardizing for age, sex, and disability changes 1982 to 1989. If morbidity prevalence declined by the same proportion, specific to age and gender, as chronic disability, then standardizing on disability should eliminate reported declines in morbidity. Only if morbidity declined more rapidly than disability would standardized morbidity rates decline.
METHODS
Data -- The NLTCS
The data analyzed are from the 1982, 1984, and 1989 NLTCS. NLTCS samples were designed to provide data on the U.S. disabled and institutionalized population age 65 and above. In 1982, detailed interviews were administered to Medicare-eligible community residents aged 65 + identified as chronically disabled on screening interviews (80% delivered by phone; 20% in-person). In addition, on the screening interview, nondisabled and institutionalized persons were identified. In 1984 and 1989, all persons previously identified as chronically disabled, or in an institution, were automatically eligible for detailed interviews to track disability change. In addition, a detailed interview was attempted for all institutional residents. Samples of persons identified as nondisabled at a prior survey were rescreened -- as well as new samples of about 5,000 persons who passed their 65th birthday between surveys -- to ensure that the 1984 and 1989 NLTCS samples represented the entire U.S. elderly population at each date.
The interviews contained questions on the health and function of persons, their use of formal and informal LTC services, their family and socioeconomic situation, and how they used personal assistance or special equipment to cope with disabilities. Details of sampling, field methods, and detailed interview response rates (above 95% in all years) are discussed in Manton, Corder, and Stallard (1993a). Because we used list samples of U.S. elderly Medicare-enrolled persons drawn from administrative records, we could continuously track mortality and Medicare service use, i.e., all persons in the three samples were linked to Medicare service use and mortality records for up to 10 years, 1982-1991. During that time, roughly 11,000 sample members died. Since the tracking of changes in function in individuals is a prime survey goal, the same health (with Alzheimer's disease added to the mental health section) and function questions were used in 1982, 1984, and 1989. The 1982 NLTCS was conducted jointly by the Office of the Assistant Secretary for Policy and Evaluation (ASPE) and the Health Care Financing Administration (HCFA). The 1984 NLTCS was sponsored by HCFA and the National Center for Health Services Research (NCHSR is now AHCPR, the Agency for Health Care and Policy Research). The 1989 NLTCS was conducted under sponsorship of the National Institute on Aging (NIA). A new NLTCS, under sponsorship of NIA, entered the field in June 1994. It includes a nondisabled sample component and questions on nutrition and physical activity.
Identifying the Disabled Population
Two stages of interviewing were used in all NLTCS. First, a list sample of Medicare-eligible persons aged 65 + was screened for chronic (90 + days) impairments of activities of daily living (ADLs; Katz and Akpom, 1976), instrumental activities of daily living (IADL; Lawton and Brody, 1970), and residence. In a second interview, there was a detailed assessment of the health and function of persons identified on the screen as being, or expecting to be, chronically disabled, or persons who had been chronically disabled in a prior NLTCS. Persons previously disabled, or institutionalized (i.e., in 1982 for 1984; in 1982 or 1984 for 1989), automatically qualified for a detailed interview after receiving a shortened screen to confirm residence and vital status.
ADL disability was defined as an inability to perform at least one of six activities (i.e., eating, getting in/out of bed, inside mobility, dressing, bathing, and toileting) without equipment or personal assistance. IADL disability was defined as the inability to perform at least one of eight activities (i.e., light housework, laundry, meal preparation, grocery shopping, outside mobility, travel, money management, and telephoning) due to health problems (Manton, Corder, and Stallard, 1993a, 1993b). The report of an inability to perform an IADL was required to be linked to a health deficit before qualifying as a disability to help reduce gender bias for certain IADLs (e.g., cooking, laundry, grocery shopping; U.S. Bureau of the Census, 1989).
The population receiving detailed interviews was divided into three groups. One reported at least one ADL impairment. The second reported at least one IADL, but no ADL, impairment. The threshold for identifying chronically disabled persons on the screen was set low to provide flexibility in defining disabled groups for different research and policy questions. The third group, although given a detailed interview, reported no chronic disability. That the disability measures detected and meaningfully distinguished between levels of disability was confirmed by determining how well disability level correlated with (a) mortality over two years, (b) Medicare service use over a 12-month calendar period, (c) morbidity, (d) difficulty in performing physical tasks (e.g., climbing stairs), and (e) sensory problems. On all of these measures (see Results) the reported ADL and IADL items showed strong predictive validity (Reuben, Siu, and Kimpau, 1992), i.e., mortality, morbidity, and service use were higher for persons reporting more disabilities. Persons reporting no disability had the lowest service use and mortality.
The screen, by design, identified as disabled a small proportion of respondents (about 8-9%) who reported little or no chronic disability on the in-person detailed interview. This "wide net" approach minimizes false-negative reports. Additionally, persons chronically disabled or institutionalized in a prior survey year who recovered function or were de-institutionalized were automatically given a detailed interview to describe persons who recovered function or who survived a long time (i.e., between surveys) with disability.
Persons who completed a detailed interview, although reporting no chronic disability, reported the presence or absence of the medical conditions listed in the questionnaire. A significant number of persons completing the detailed community interview reported medical conditions but not chronic disability. Condition lists (except for the one addition in 1989) were identical in all three years. In the three surveys, there were 2,484 interviews where no chronic disability was reported on the detailed community interview. Roughly half of these persons reported chronic disability on the screening interview, but not on the detailed interview; half reported disability in 1982 (or 1984) and had fully recovered function in 1984 (or 1989). There were a total of 550 detailed interviews with no disability reported in 1982, 989 in 1984, and 945 in 1989, i.e., sufficiently large numbers of cases to make statistical inferences on how the reports of medical conditions by the nondisabled U.S. elderly population differed from reports of conditions by those chronically disabled.
The sample weights used in Tables 1 and 2 reflect longitudinal sample design effects with conservative adjustments for nonresponse (Manton, Corder, and Stallard, 1993a).(Tables 1 and 2 omitted) In Tables 3-5 the weights are rescaled so that persons reporting no disability on the community survey represent the entire nondisabled U.S. elderly population.(Tables 3-5 omitted) This was done for age and sex groups by: (1) estimating the size of the nondisabled population from the sum of the weights for persons identified as nondisabled from either the screen or detailed interview; (2) setting weights for persons reporting no disability on the screen interview to zero, i.e., treating them as nonrespondents to the detailed interview; (3) multiplying the weights for persons reporting no disability on the detailed interview by the ratio of the size of the total nondisabled population to the sum of the weights for persons reporting no disability on the detailed interview. The rescaled weights yield estimates of characteristics of the total U.S. nondisabled elderly population. To determine that persons receiving the detailed interview and reporting no disability were statistically representative of the total U.S. nondisabled elderly population, we compared their mortality and Medicare use to persons who reported being nondisabled when screened.
Standardization in Table 2 is based on the 1989 sex and age distribution in 5-year age groups to age 95 +. Standardization in Tables 3 and 5 (Table 4 is age-specific) is based on the 1989 sex and age distribution in 10-year age groups to age 85 + . The change from 5- to 10-year age groups was done to ensure that rates were stable even after the additional stratifications (i.e., by disability level and type of medical condition) used in Tables 3 and 5.
RESULTS
Representativeness of Mortality and Medicare Service Use of Nondisabled Respondents
The first test of representativeness was based on mortality. If mortality is measured for two years after April 1 (the target population date) in each survey year, the ratio of the mortality rate for persons taking the detailed interview but reporting no disability, relative to persons reporting no disability on the screen, was 1 .01 for males and 0.99 for females. If the mortality rate is calculated for two years after the date of the last interview in 1982, 1984, and 1989, still none of the t-tests of mortality differences between persons taking the detailed interview and reporting no disability and those reporting no disability on the screen were significant (i.e., t's of 1.3 and 0.6 for males and females in 1982; 0.8 and -0.3 in 1984; and 0.3 and 1.7 in 1989).
A second test examined the use of Medicare skilled nursing facilities (SNF) and home health agency (HHA) services for 12 months. Estimates for calendar years 1982 and 1989 are shown in Table 1.
Persons reporting no disability on the detailed interview in 1982 used no SNF days and an average of 0.41 HHA visits in the 12 months following the end of interviewing; 2.86 percent had at least one HHA visit. A t-test of the difference in the proportion of persons using SNFs who reported no disability on the detailed interview and the proportion of persons using SNFs reporting no disability on the screen was insignificant in 1982 (i.e., t = -4.5; persons reporting no disability on the detailed interview used fewer SNF days) and 1989 (t = 0.78). The t-tests between the proportion using HHA services for persons reporting no disability on the screen vs those reporting no disability on the detailed interview are insignificant; 1.5 (1982) and 1.7 (1989). Thus, there were no statistically significant differences in mortality, Medicare SNF, or HHA use between those reporting no disability on the detailed instrument (for whom medical conditions were reported) and those reporting no disability on the screen (who did not report medical conditions).
Disability Gradients and Their Relation to Service Use
A second way to examine persons who reported no disability on the detailed interview is to compare them to persons who reported chronic disability. In making those comparisons we determined if there were meaningful differences in several types of Medicare service used in a 12-month period about the time of the interview (i.e., calendar years 1982 and 1989) across disability levels. A gradient would be evidence of the biological meaningfulness of the self-reported disability measures.
In Table 1 the IADL-only group in 1982 used an average of 0.08 SNF days (0.45% used SNFs) compared to no use for the nondisabled detailed interview responders. The mean number of HHA visits used by the IADL-disabled group is over twice as high as for persons reporting no disability on the detailed interview. Of IADL disabled persons, 4.78 percent used HHA (vs 2.86% for those who reported no disability on the detailed interview). For those with 1-2 ADLs impaired, the mean number of SNF days used in 1982 is 0.37 (1.24% used SNFs). The number of HHA visits increased for those with 1-2 (to 2.47), 3-4 (to 5.42), and 5-6 (13.80) ADLs impaired. There were also gradients across disability levels for the number of SNF days used and the proportions using SNFs and HHA. Thus, the gradient of 
SNF and HHA service use over disability reported in 1982 was large, consistent, and persisted for 12 months.
In 1989, persons who took the detailed interview but did not report disability used 0. 15 SNF days; the IADL-disabled used 0.64 days, four times more. For persons with 1-2, 3-4, and 5-6 ADLs, 1.22, 1.23, and 2.84 SNF days were used. Of persons reporting no disability on the detailed interview, 0.8 percent used SNFs vs 1.00, 3.30, 4.30, and 6.70 percent for the four disabled groups. There was again a persistent gradient in the proportion of persons using HHA services and in the mean number of HHA visits used across disability levels.
Not only are persons taking the detailed interview and reporting no disability representative of persons who reported no disability on the screen, but also (a) there were significant differences in Medicare service use between those who reported no disability and those who reported disability on the detailed interview, and (b) service use (and mortality; see Manton, Corder, and Stallard, 1993a) increased as disability increased. Both the representativeness of persons taking the detailed interview and reporting no disability and the predictive validity of the self-reported disability items on independent measures were demonstrated. Thus, there were biological correlates of the self-reported disability gradient -- including for persons who reported only IADL disability. There are a large number, almost 4,000 respondents in the three surveys, who reported only IADL disability. These include persons disabled because of the interaction of social or housing factors (e.g., grocery shopping, laundry, cooking) with moderate physical impairment or early dementia (i.e., problems in phoning, managing money, or taking medication). Thus, the samples can be used to estimate the prevalence of medical conditions in 1982 and 1989 for severely disabled, moderately disabled, and nondisabled groups.
Another test of the predictive validity of the disability reports is acute hospital use. This was affected by changes in Medicare hospital reimbursement. Starting in October 1983, payments were made through a prospective payment system (PPS) using 473 diagnosis-related groups (DRGs). Hospital admission rates and length of stay declined from 1982 to 1989 because of PPS. Those reporting no disability on the detailed interview in 1982 used 3. 1 days of care compared to 5.4 for those with IADL disability. In 1989, the difference is 2.25 vs 2.50 days -- the IADL-disabled group length of stay declined more. There was a gradient of hospital use across disability level in both periods. Those with 1-2, 3-4, and 5-6 ADLs in 1982 used 5.8, 9.1, and 14.0 days vs 3.8, 6. 1, and 10.5 days in 1989. The ratio of hospital days used changed proportionately, i.e., care used in the nondisabled group in 1989 is 74.9 percent of that used in 1982, and in the 5-6 ADL group it is 72.1 percent. Thus, PPS reduced the hospital use of all disability groups between 1982 and 1989. Reductions in hospital use were associated with increased HHA and SNF use (Table 1). The Medicare average reimbursement (Parts A and B) for persons reporting no disability on the detailed interview was also below the U.S. average (e.g., in 1982, $1,293 vs $1,742; in 1989, $2,223 vs $2,970).
The tests identify four facts. First, the per capita use of SNF or HHA services was low in either the group reporting no disability on the screen interview or in the group reporting no disability on the detailed interview. Second, differences between the two nondisabled groups for mortality, HHA, or SNF use are nonsignificant despite large samples (i.e., 12,405 persons in the 1982 longitudinal sample reported no disability on the screen and 11,748 in 1989; 550 in 1982 and 945 in 1989 reported no disability on the detailed interview. Third, persons reporting no disability on the detailed interview use significantly less service in the 12 months examined than those reporting IADL disabilities in three of four comparisons (1989 SNF use differences were no significant). Thus, there was no statistically significant evidence of mortality, service use, or reimbursement differences on which to reject the national representativeness of persons taking the detailed interview who reported no disability. This was not surprising because surveys of disability in elderly persons often show both significant rates of reversal of disability and short (e.g., 3 to 12 months) periods of disability (e.g., Manton, Stallard, and Woodbury, 1991). In older (e.g., age 75 +) nondisabled populations, a significant proportion probably had prior episodes of disability (e.g., recuperation after surgery or a fracture; improvement of function after cataract surgery to replace a lens; surgery for hip and knee joint prostheses). If persons reporting no disability in the detailed interview had a higher prevalence of prior disability episodes, their would be expected to report more medical conditions than those who reported no disability on the screen, diluting estimated morbidity differences between disabled and nondisabled populations. Finally, Medicare service use increased with disability intensity for a 12-month period when the disability level was determined -- showing that disability effects were persistent.
Changes in Disability and Mortality 19B2 to 1989
In Table 2 we present disability level specific rates, standardized by age and sex, to adjust for changes in the age and sex composition of the U.S. population 1982 to 1989.
Declines occur both for IADL- and ADL-impaired. Those with 3 to 4 ADLs impaired increased. Declines between 1982 and 1989 are significant if 1982 prevalence rates are unstandardized (-1.12%), or age and sex standardized (-1.92%). For those with 5 to 6 ADLs or institutionalized, groups of interest for health care reform, the decline is large (-1.5%).
We examined life expectancy and the age dependence of disability changes in the sample (Manton and Stallard, 1993) by calculating life tables using the 1982, 1984, and 1989 NLTCS disability data and mortality, 1982 to 1991. Life expectancy at age 65, based on the interaction of age-specific disability dynamics, and age- and disability-specific mortality, was 15.6 years for males and 20.9 years for females (Manton et al., 1994a). The male value is close to the 1989 NCHS life table estimate of 15.4 years and, given declines in mortality above age 85, 1988-1992, may be conservative. Female life expectancy is higher than the 1989 NCHS estimate of 14.0 years. Census Bureau estimates of life expectancy at age 65 (15.7 years) exceed the male estimate and are closer to the female estimate (19.5 years; Day, 1993). Estimated male life expectancy (Social Security Administration, 1992) for the 1925 birth cohort (males age 65 in 1990) was 15.3 years; for females, 19.6 years. Since the 1992 SSA cohort estimates used mortality data only to 1988, they do not reflect the 1988 to 1992 decline in mortality for persons aged 85+. Thus, the NLTCS sample and its disability-dependent mortality experience represent cohort and cross-sectional life expectancy estimates fairly well.
Changes in Reported Morbidity Prevalence 1982 to 1989
We now compare the 1982 and 1989 prevalence of 16 medical conditions, and the mean number of conditions. We also examined the morbidity prevalence calculated by applying the 1982 age-, sex-, and disability-specific morbidity rates to the 1989 age-, sex-, and disability-specific population distributions. In analyzing health changes, prevalence rates are sometimes used with period life tables to calculate the life expectancy in specific health states (e.g., Robine and Ritchie, 1991; Sullivan, 1971; World Health Organization, 1984). Instead, we used standardization to determine if the sum of the 1982 age-, sex-, and disability-specific morbidity prevalence rates decreased after being reweighted to the 1989 age, sex, and disability distribution. Thus, effects of 1982 to 1989 changes in age, sex, and disability distributions are eliminated from standardized morbidity prevalence rate changes without making assumptions about changes in disease duration. Because the three surveys were done at three times, disease incidence is unobserved. To estimate it requires making assumptions about disease transitions between survey dates. However, for prevalence to decline, net of age, sex, and disability change, either the incidence rate, disease duration, or both must decline. Furthermore, the prevalence estimates of disability are based on persons having chronic disability, and we showed in Table 1 that significant disability differences in service use persist for 12 months. Thus, the survey identifies chronic disability episodes and associated morbidity. The reported prevalence of disease at the three times is not altered by length-biased sampling (i.e., the probability of detecting a disease episode of short duration is less than capturing one of long duration), although efforts to attribute prevalence changes to disease duration or incidence change would be affected.
Standard errors and tests of morbidity prevalence differences between 1982 and 1989 were calculated using estimator variances based on longitudinal sample weights adjusted for design effects using procedures in Potthoff, Woodbury, and Manton (1992). These are appropriate for longitudinal surveys because they assume that an infinite "super-population" is sampled (e.g., Deming and Stephan, 1941).
In Table 3 are observed and age, sex, and disability standardized prevalences for 16 medical conditions in 1982 and the 1989 observed prevalence for the U.S. elderly noninstitutionalized population (conditions like multiple sclerosis, epilepsy, and cerebral palsy with little prevalence above age 65 and "symptoms," like headaches, were not used).
T-tests for differences in binomial proportions, with degrees of freedom adjusted for sample design effects, were calculated for the standardized 1982 and observed 1989 prevalences. Most changes are significant at the .05 level ( t >= 1.96). The 1982 standardized prevalence for medical conditions significantly decreased by 1989 for the total U.S. elderly noninstitutionalized population for arthritis, arteriosclerosis, dementia (which, in 1989, included an additional diagnosis, Alzheimer's disease), hypertension, stroke, circulatory disease, and emphysema. Parkinson's disease, other heart disease, pneumonia, bronchitis, and hip fractures increased.
The prevalence rates are point estimates based upon self-reports of the sample person of clinically manifest, symptomatic disease. A disease may start long before clinical manifestation. However, because diagnostic technologies improved between surveys (e.g., CT scans detecting more strokes; McGovern et al., 1992) we would expect lead time (i.e., the time from disease onset to clinical manifestation) to decrease, causing an increase in reported prevalence by increasing the time a disease is clinically manifest. Because assessments were made at points in time, the probability of detecting a disease episode in a sample is proportional to its manifest duration. Prevalence declined for a number of conditions as did the average number of conditions reported overall. Hence, either the incidence of disease episodes of the same duration decreased, there was a higher incidence with a reduced disease duration, or there was a much lower incidence with a longer disease duration. In all of these cases, the population burden of morbidity at the time of the survey would decrease if prevalence decreased. For specific diseases, changes in incidence and duration could differ.
In Table 3 we also calculated the annual percentage relative rate of change, both for 1982 to 1989 and 1984 to 1989. The relative rate of change is a function of the size of the rate. Thus, small rates may change relatively rapidly, and large rates relatively slowly. In general, changes were more rapid from 1984 to 1989 than from 1982 to 1989, i.e., changes appeared to accelerate. We calculated change over the longest available interval, 7 years, because chronic disease changes in a population accumulate slowly and require time to reach significance. In contrast, tracking disease changes for an individual may require frequent assessments.
The summary morbidity index is the estimated average number of conditions reported in the 1982 (standardized) and 1989 elderly U.S. population. The index has the advantage of predictive validity, i.e., it predicts mortality, service use, and health changes in other elderly study populations such as the California Multipurpose Senior Services Program study (Vertrees, Manton, and Adler, 1989). It is statistically more stable than prevalence rates for individual conditions. It has the disadvantage of summing across diseases that are biologically diverse. However, although diseases may have different effects on the individual, it seems reasonable because of their predictive ability that, on average, having more diseases is biologically less advantageous than having fewer diseases.
We present percent changes in the mean number of conditions reported in 1989 relative to the 1982 morbidity standardized to the 1989 age-, sex-, and disability-specific population distribution. In Table 3, the 1989 mean morbidity count declined 11.2 percent from the 1982 age, sex, and disability standardized count(t = -9.3), suggesting morbidity declined between 1982 and 1989.
In Table 3 we also present 1984 prevalence rates. In six cases the 1982 to 1984 trends are consistent with declines observed from 1982 to 1989. In seven cases, there is no change from 1982 to 1984. In three cases the change from 1982 to 1984 was in a different direction but not significant. Thus, while there is a suggestion of the population trends manifest from 1982 to 1989, either (a) it takes the longer 7-year period for population trends to become significant, or (b) prevalence changed faster from 1984 to 1989. More detailed analyses of population subgroups and their health changes will be necessary to identify mechanisms of change. Not presented is the institutionalized population. In Table 2, the age and sex standardized prevalence of this group declined (-0.6%). Thus, the morbidity decline in Table 3 is unlikely to be caused by a greater proportion of the morbid population being institutionalized.
We also calculated morbidity prevalence for persons in 1982 and 1989 reporting (a) no disability, (b) no ADL but at least one IADL disability, and (c) at least one ADL disability. There are differences in the disease-specific patterns of change in these groups, e.g., morbidity changes for persons reporting at least one ADL disability differ from those for the total elderly population. Arthritis declined less (-4.6%; t = -5.2). Decreases were larger for arteriosclerosis (-11.4%; t = -11.2) and dementia (-3.5%; t = -5.2). Circulatory disease declines were large (-7.2%; t = -6.8). Pneumonia increased significantly, but not bronchitis. The mean number of conditions for those with ADL disabilities declined from 3.61 to 3.30 (-8.6%; t = -8.3).
In Table 4 are the 1982 to 1989 changes for males and females aged 85 + with at least one ADL disability. This age group Showed the largest increase in the proportion remaining nondisabled after adjusting for survival changes 1982-1989 (Manton, Corder, and Stallard, 1993a). There were also large declines in mortality for persons 85 + between 1988 and 1992.
For males, other heart disease, pneumonia, and fractures increased. For females, pneumonia and fractures increased. For both genders, most conditions declined (e.g., diabetes, arteriosclerosis, hypertension, and circulatory disease). Bronchitis and emphysema declined only for males, as did stroke. The morbidity index for ADL-impaired males aged 85 + declined 15.4 percent(t = -12.1), about a third more than for females aged 85 + (-10.7%; t = -8.3). These declines are generally relatively larger than at age 65-74. Summary statistics for observed and standardized changes in the average number of conditions between 1982 and 1989 are in Table 5.
Both males and females consistently had declines in overall morbidity, declines which were larger after standardizing the 1982 rates to he 1989 age, sex, and disability distributions. Thus, though its size varied, a decline in the number of medical conditions reported was observed consistently over disability level; age, and sex.
DISCUSSION
We found declines in the age, sex; and disability standardized prevalence for 10 of 16 medical conditions reported in the NLTCS from 1982 to 1989 (8 were significant) and for the average number of conditions reported. In ADL-impaired persons aged 85 + , 8 of 11 significant changes for males were declines compared to 7 of 9 significant changes for females. For males, morbidity declines were smaller than for females, except for ADL-impaired males aged 85 +, which were a third larger than for females.
Several explanations of the trends are possible. In measuring disability and morbidity, survey methods may affect results, as might differences in response rates, which are sensitive to health at late ages (Manton, Stallard, and Woodbury, 1991). Because the same field procedures and instruments were used in all three years, and response rates remained uniformly high, those effects ere minimized. For example, the only change in diagnoses was in 1989, when Alzheimer's disease was added to the dementia group. Despite this addition, dementia prevalence declined. A study comparing self-reports of disability using the NLTCS screening instrument and exert evaluations of nurse assessors also showed good reliability (Durako, 1987). Those items were used in Social/HMOs and comparison Medicare fee-for-service samples and predicted mortality, morbidity, and service use (Manton et al., 1994b; 1994c). Thus, the items showed good predictive validity in the NLTCS (Table 1) and two independent sets of data.
The only substantial NLTCS survey changes were that, in 1984 and 1989, previously disabled persons automatically qualified for a detailed interview. We adjusted the definition of disability to be identical in all three years. In the surveys, proxy rates declined with time. Declines in proxy rates indicate reductions, in highly impaired groups (Dorevitch et al., 1992). In addition, we tested the biological significance of the disability gradient from persons with only IADL disability to persons with 5 to 6 ADL disabilities on multiple measures (i.e., mortality; hospital use, HHA and SNF; the relation to physical impairment; morbidity) and found large and consistent gradients suggesting high predictive validity (Reuben, Siu, and Kimpau, 1992). Since services were measured for 12 months around the survey date, the estimated gradient (including those with IADLs) was persistent and likely conservative (i.e., the most frail are differentially likely to die or improve in 12 months).
Second, diagnostic precision has likely increased because of technical advances (e.g., improved magnetic and radiological imaging; fiber optics used in arthroscopic, laparoscoptic, and other exploratory and therapeutic procedures). As diagnostic procedures improve we would expect "generic" diagnoses for elderly persons to be replaced by larger numbers of more precise diagnoses and disease "lead" time to decrease. This group also manifested large increases in education and the largest increases in the proportion staying nondisabled from 1982 to 1989 (Manton, Corder, and Stallard, 1993a). With increased education, one might expect more precise and detailed self-reporting of conditions in 1989 than in 1982, because the ability to diagnose many conditions improved with time (e.g., Alzheimer's disease). In spite of these trends the mean numbers of medical conditions reported declined.
Thus, there is evidence from the NLTCS of both disability and morbidity prevalence declines from 1982 to 1989. The review of survey methods and the use of conservative assumptions suggest that those declines are not likely due to diagnostic or survey artifact. Both declines were associated with increased life expectancy at late ages. Despite morbidity and disability declines, the rapid growth of the 65 + and the 85+ U.S. population will increase absolute levels of acute and LTC service needs. Nonetheless, morbidity and disability declines will strongly influence the absolute future amounts and types of services needed. Evaluations of future health needs will, therefore, have to balance the growth of the elderly population, disability and morbidity trends, and opportunities to accelerate morbidity and disability declines due to advances in biomedical research.
We examined changes in the prevalence of specific conditions. Respiratory diseases (e.g., pneumonia) increased. Many circulatory diseases decreased. Dementia, because of improved education and possibly reduced stroke and circulatory disease risk, declined significantly in most groups. Skoog et al. (1993) found that a large proportion of dementia cases above age 85 had a significant vascular component possibly amenable to intervention (Hachinski, 1992; Phillips and Whisnant, 1992). Recent research has implicated a blood lipid (apo-lipoprotein E) in Alzheimer's disease (Roses et al., in press; Strittmatter et al., 1993) with risk higher for those with the E4 genotype. Since this apolipoprotein appears to affect transport of lipids across cell membranes, this may partly explain the observed linkage of circulatory disease and dementia (Aronson et al., 1990). Thus, declines in morbidity prevalence are not homogeneous -- they differ by age, sex, type of disease and disability. However, overall there is evidence of morbidity declines which occur on an age-, sex-, and disability-specific basis over a period when mortality declined.
The trends for CVD are products of complex forces with controversy over the relative contributions of prevention, disease treatment, improved survival, and changes in diagnoses. The 1982 to 1989 increase (+2.7% standardized; 3.4% observed) for chronic heart disease is consistent with heart attack victims living longer. The National Hospital Discharge Surveys showed that the U.S. age-standardized hospitalization rate for CHF increased 60 percent from 1973 to 1986 (Ghali, Cooper, and Ford, 1990). The decline in dementia (-1.2%; t = -3.7) is consistent with the increased education in elderly cohorts which, along with increased physical activity and reductions in CVD, are negatively related to dementia (e.g., Evans et al., 1992). The decline in arteriosclerosis (-6.5%; t = -7.7) suggests improvements in diagnostic precision with probably the anatomic site where an atheroma caused a circulatory event being given precedence over a generic diagnosis. This is consistent with the decline (-35.3%) in reporting arteriosclerosis as an underlying cause of death on U.S. death certificates from 1982 to 1989 and improved treatment of CVD at late ages (e.g., Ko et al., 1992; Muhlbaier et al., 1992). However, if more anatomically specific diagnoses are made (e.g., CT scans increasing the number of strokes detected; McGovern et al., 1992); one might expect equal, or increased, numbers of medical conditions to be reported due to better diagnostic specificity and the increased education and medical sophistication of elderly persons. This did not happen in general.
There is evidence that the detection and treatment of hypertension has improved -- at least to 1980 (NCHS, 1992) and possibly, at a slower rate, to 1987 (McGovern et al., 1992). The use of more stringent (i.e., lower blood pressure thresholds) diagnostic criteria, especially for persons with co-morbidity (e.g., diabetes, renal disease), is a recent phenomenon. While hypertension has been better controlled there is controversy over how mush that contributed to, e.g., stroke reduction (Casper et al., 1992; Klag, Whelton, and Seidler, 1989). It may be that reductions in stroke to date are due to smoking and other risk factor reductions(Niessen et al., 1993). In addition, in several studies (e.g., Colantonio, Kasl, and Ostfeld, 1992; Tsuji et al., 1994) decreased function was a stroke risk factor. Thus, improved functioning may reduce stroke which, in time, would further reduce disability in a positive feedback loop. The benefits of hypertension control may be greater for new classes of antihypertensives (e.g., ACE-II inhibitors; the first drug, captopril, was approved for use in severe hypertension in 1981; use in mild to moderate hypertension in July 1985; Materson and Preston, 1994) which produced declines in both mortality and hospitalization rates for CHD (e.g., a reduction of 120 hospitalizations per 1000 person-years in the SOLVD 1991 trials). Despite these factors, the prevalence of reported hypertension declined (-6.5%; t = -6. 1).
An increase in pneumonia and bronchitis is consistent with the emergence of infectious complications possibly associated with nutritional deficiency and declines in immune efficacy with age e.g., Grossman, 1988) an possibly antibiotic efficacy with time. The decline in emphysema may represent cessation of smoking at later ages or reduction of occupationally related pulmonary disease post-WWII.
Morbidity declines can be related to historical trends. Originally, it was theorized that chronic disease would increase as life expectancy and chronic disease risk factors increased in industrial societies (e.g., Dubos, 1965; Omran, 1971). Recent data challenge this notion. Lanska and Mi (1993) analyzed stroke mortality declines and found reporting artifacts insufficient to explain declines in U.S. stroke mortality after 1925. Niessen et al. (1993), in a Dutch study, found that both stroke incidence and mortality declines were necessary to explain annual declines of 3.1 percent for men and 4.0 percent for women in stroke mortality between 1979 and 1989. Those declines were less than U.S. declines of 5.7 and 5.2 percent. Mozar, Bal, and Farag (1990) analyzed the role of viruses in atherogenesis and found that viral infections might explain increased CHD mortality early in the 20th century. Early declines in CHD risk in this model are due to improved commercial food processing. They related temporal and spatial CHD changes to specific food-borne infections identified in 1932, 1952, 1962, and 1980 and subsequent modification of federal and state food handling regulations.
Fogel (1993) compared the prevalence of medical conditions reported for Civil War veterans aged 65 + in 1910 (birth cohorts of 1825 to 1844) with veterans aged 65 + in the 1985-1988 National Health Interview Surveys. Mobidity declined 6 percent per decade over the 75-year period, e.g., heart conditions were 2.9 times more prevalent in Civil War veterans than in more recent veterans. This was attributed to nutritional changes consistent with research on the effects of antenatal nutrition on fetal organ development and subsequent chronic disease risk (Barker et al., 1992a, 1992b; Hales et al., 1991). It was found that the fetus husbanded nutritional resources and gave different priorities to organs. The brain received the highest priority. The liver, pancreas, and lung suffered from antenatal nutritional deficiency elevating circulatory, pulmonary disease, and diabetes risks above age 65. Historical patterns of antenatal nutritional deficiency thus may partly explain long-term morbidity trends.
The relation of dementia to education has been observed in several studies, although the mechanisms by which risk is reduced are unclear. Dementia may be related to circulatory diseases which decline more rapidly in better educated groups who more readily accept life style and other risk factor interventions (Aronson et al., 1990). Data (1960-1990) from National Health and Nutritional Examination Surveys (NHANES) showed that several risk factors declined significantly. Evaluations of fat consumption (1920-1984) first showed increases, then decreases starting in the mid-1960s (Stephen and Wald, 1990). Large declines in cholesterol(e.g., at ages 65 to 74) may partly explain a 54 percent decline in CHD mortality from 1963 to 1990, i.e., the 6 to 8 percent decline in cholesterol observed in NHANES might, over the long run, reduce CHD mortality by 12 to 32 percent (Johnson et al., 1993; Sempos et al., 1993). Other explanations suggest that (a) better educated persons have a longer way to decline before being diagnosed as demented, and (b) education may directly affect neurological development.
Declines in some cancers may be caused by reduction in smoking, improved antioxidant intake (e.g., vitamins C and E), improved physical activity (Blair et al., 1989a, 1989b), and reduced exposure to pathogens (e.g., Helicobacter pylori, a causative agent in 60 to 80% of stomach cancer; ere have been large cohort-related declines in U.S. stomach cancer rates; Forman, 1990, 1991). A reduction in arthritis may be linked to physical activity and education where maintenance of activity may slow disease progression (Callahan, Bloch, and Pincus, 1992). The prevalence of hip fractures among elderly Swedish females declined between 1965 and 1983 (Naesson et al., 1989). In Japan, the cohort occurrence of female vertebral compression declined from 1958 to 1986 (Fujiwara et al., 1991). These declines in female fracture rates were linked to nutritional changes affecting osteoporosis. Clinical trials and other studies suggested exogenous estrogens may reduce fracture rates (by 77% if started before the first fracture; Overgaard et al., 1992), as well as CVD and CVD risk factors (Nabulsi et al., 1993). Nearly three million U.S. women were taking estrogens by 1985 for postmenopausal symptoms. Multiple diseases may be linked to common physiological processes, e.g., the rapid release of calcium in osteoporosis may accelerate the age rate of increase in female CVD rates postmenopausally (Browner et al., 1993; Moon, Bandy, and Davison, 1992). Factors controlling osteoporosis (e.g., exogenous estrogens), and consequently atherogenesis, may also affect lipid profiles, CVD, and possibly dementia. Thus, a number of biomedical and epidemiological findings are consistent with U.S. morbidity prevalence declines from 1982 to 1989, although it seems that many biomedical innovations were emerging in that period so that declines may continue in the future.
ACKNOWLEDGMENTS
This research was supported by NIA grants 5R37 AG-07198, 5R37 AG-07025, and 1R01 AG-07469, and the Health Care Financing Administration.
Address correspondence to Dr. Kenneth G. Manton, Center for Demographic Studies, Duke University, Box 90408, Durham, NC 27708-0408.
REFERENCES
Aronson, M., W. Ooi, H. Morgenstern, A. Hafner, D. Masur, H. Crystal, W. Frishman, D. Fisher, and R. Katzman. 1990. "Women. Myocardial Infarction, and Dementia in the Very Old." Neurology 40:1102-1106.
Barker, D., K. Godfrey, C. Osmond, and A. Bull. 1992a. "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 6:35-44:
Barker, D., T. Meade, C. Fall, A. Lee, C. Osmond, K. Phipps, and Y. Stirling. 1992b. "Relation of Fetal and Infant Growth to Plasma Fibrinogen and Factor VII Concentrations in Adult Life." British Medical Journal 304:148-152.
Blair, Steven N., William B. Kannel, Harold W. Kohl, Nancy Goodyear, and Peter W. F. Wilson. 1989a. "Surrogate Measures of Physical Activity and Physical Fitness." American Journal of Epidemiology 129:1145-1156.
Blair, S., H. Kohl, R. Paffenbarger, D. Clark, K. Cooper, and L. Gibbons. 1989b. "Physical Fitness and All Cause Mortality: A Prospective Study of Healthy Men and Women." Journal of the American Medical Association 262:2395-2401.
Browner, W., A. Pressman, M. Nevitt, J. Cauley, and S. Cummings. 1993. "Association Between Low Bone Density and Stroke in Elderly Women: The Study of Osteoporotic Fractures." Stroke 24:940-946.
Callahan, L. F., D. A. Bloch, and T. Pincus. 1992. "Identification of Work Disability in Rheumatoid Arthritis: Physical, Radiographic and Laboratory Variables Do Not Add Explanatory Power to Demographic and Functional Variables." Journal of Clinical Epidemiology 45: 127-138.
Casper, M., S. Wing, D. Strogatz, C. E. Davis, and H. A. Tyroler. 1992. "Antihypertensive Treatment and U.S. Trends in Stroke Mortality: 1962 to 1980." American Journal of Public Health 82:1600-1606.
Colantonio, A., S. Kasl, and A. Ostfeld. 1992. "Level of Function Predicts First Stroke in the Elderly." Stroke 23:1355-1357.
Day, Jennifer. 1993. Population Projections of the United States, by Age, Sex, Race, and Hispanic Origin: 1993 to 2050 (Series P25-1104, Current Population Reports). Washington, DC: U.S. Government Printing Office.
Deming, W. Edwards and Fredrick F. Stephan. 1941. "On the Interpretation of Censuses as Samples." Journal of the American Statistical Association 36:45-49.
Dorevitch, M., R. Cossar, F. J. Bailey, T. Bisset, S. Lewis, L. Wise, and W. J. MacLennan. 1992. "The Accuracy of Self and Informant Ratings of Physical Functional Capacity in the Elderly." Journal of Clinical Epidemiology 45:791-798.
Dubos, Rene. 1965. Man Adapting. New Haven, CT: Yale University Press.
Durako, S. 1987. "Evaluation of Social/Health Maintenance Organization Demonstration." Final report on survey field methods. Rockville, MD: Westat.
Evans, Denis A., Paul A. Scherr, Nancy R. Cook, Marilyn S. Albert, N. Harris Funkenstein, Laurel A. Beckett, Liesi E. Hebert, Terrie T. Wetle, Laurence G. Branch, Marilyn Chown, Charles H. Hennekens, and Janas O. Taylor. 1992. "The Impact of Alzheimer's Disease in the United States Population." In Richard Suzman, David Willis, and Kenneth G. Manton (Eds.), The Oldest-Old. New York: Oxford University Press.
Fogel, Robert W. 1993. "Economic Growth, Population Theory, and Physiology: The Bearing of Long-Term Processes on the Making of Economic Policy." Presentation to The Nobel Prize Award Committee, December 9.
Forman, D. 1990. "Geographic Association of Helicobacter Pylori Antibody Prevalence and Gastric Cancer Mortality in Rural China." International Journal of Cancer 46:608-611.
Forman, D. 1991. "Helicobacter Pylori Infection: A Novel Risk Factor in the Etiology of Gastric Cancer." Journal of the National Cancer Institute 83:1702-1703.
Fujiwara, S., S. Mizuno, Y. Ochi, H. Sasaki, K. Kodama, W. J. Russell, and Y. Hosoda. 1991. "The Incidence of Thoracic Vertebral Fractures in a Japanese Population, Hiroshima and Nagasaki, 1958-1986." Journal of Clinical Epidemiology 44:1007-1014.
Ghali, J. K., R. Cooper, and E. Ford. 1990. "Trends in Hospitalization Rates for Heart Failure in the United States, 1973-1986." Archives of Internal Medicine 150:769-776.
Grossman, Charles M. 1988. "Serum Cholesterol Variations in Individual Patients." Atherosclerosis 71:193-195.
Hachinski, V. 1992. "Preventable Senility: A Call for Action Against the Vascular Dementias." Lancet 340:645-648.
Hales, C., D. Barker, P. Clark, L. Cox, C. Fall, C. Osmond, and P. Winter. 1991. "Fetal and Infant Growth Impaired Glucose Tolerances at Age 64." British Medical Journal 303:1019-1022.
Johnson, C., B. Rifkind, C. Sempos, M. Carroll, P. Bachorik, R. Briefel, D. Gordon, V. Burt, C. Brown, K. Lippel, and J. Cleeman. 1993. "Declining Serum Total Cholesterol Levels Among U.S. Adults: The National Health and Nutrition Examination Surveys." Journal of the American Medical Association 269:3002-3008.
Kalz, Sidney and C. Amechi Akpom. 1976. "A Measure of Primary Sociobiological Functions." International Journal of Health Services 6:493-508.
Klag, M., P. Whelton, and A. Seidler. 1989. "Decline in US Stroke Mortality: Demographic Trends and Antihypertensive Treatment" Stroke 20:14-21.
Ko, Wilson, Jeffrey P. Gold, Richard Lazzaro, John A. Zelano, Samual Lang, O. Wayne Isom, and Karl H. Krieger. 1992. "Survival Analysis of Octogenarian Patients With Coronary Artery Disease Managed by Elective Coronary Artery Bypass Surgery versus Conventional Medical Treatment." Circulation 86(5):II-101:II-197.
Lanska, D. and X. Mi. 1993. "Decline in US Stroke Mortality in the Era Before Antihypertensive Therapy." Stroke 24:1382-1388.
Lawton, M. Powell and Elaine M. Brody. 1970. "Assessment of Older People: Self-Maintaining and Instrumental Activities of Daily Living." The Gerontologist 9:179-186.
Maddox, George L. and Daniel O. Clark. 1992. "Trajectories of Functional Impairment in Later Life." Journal of Health and Social Behavior 33:114-125.
Manton, Kenneth G., Larry S. Corder, and Eric Stallard. 1993a. "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 48:S153-S166.
Manton, Kenneth G., Larry S. Corder, and Eric Stallard. 1993b. "Changes in the Use of Assistance and Special Equipment 1982 to 1989: Results from the 1982 and 1989 NLTCS." The Gerontologist 33:168-176.
Manton, Kenneth G., Robert Newcomer, James C. Vertrees, Gene L. Lowrimore, and Charlene Harrington. 1994b. "A Method for Adjusting Capitation Payments to Managed Care Plans Using Multivariate Patterns of Health and Functioning: The Experience of Social/Health Maintenance Organizations. Medical Care 32:227-297.
Manton Kenneth G., Robert Newcomer, Gene R. Lowrimore, James C. Vertrees, and Charlene Harrington. 1994c. "Social/Health Maintenance Organization and Fee-for-Service Health Outcomes Over Time." Health Care Financing Review 15:173-202.
Manton, Kenneth G. and Eric Stallard. 1993. "Morbidity and Mortality Issues in Developing Countries in the Post-Reproductive Years." In James Gribble and Samuel Preston (Eds.), The Epidemiological Transition: Policy and Planning Implications for Developing Countries. Washington, DC: National Academy Press.
Manton, Kenneth G., Eric Stallard, and Max A. Woodbury. 1991. "A Multivariate Event History Model Based Upon Fuzzy States: Estimation From Longitudinal Surveys with Informative Nonresponse." Journal of Official Statistics (published by Statistics Sweden, Stockholm) 7:261-293.
Manton, Kenneth G., Eric Stallard, Max A. Woodbury, and J. Ed Dowd. 1994a. "Time-Varying Covariates in Models of Human Mortality and Aging: Multidimensional Generalizations of the Gompertz. " Journal of Gerontology: Biological Sciences 49:B169-B190.
Materson, Barry J. and Richard A. Preston. 1994. "Angiotensin-Converting Enzyme Inhibitors in Hypertension: A Dozen Years of Experience." Archives of Internal Medicine 154:513-523.
McGovern, P., G. Burke, J. Spafka, S. Xue, A. Folsom, and H. Blackburn. 1992. "Trends in Mortality, Morbidity, and Risk Factor Levels for Stroke From 1960 Through 1990: The Minnesota Heart Survey." Journal of the American Medical Association 268:753-759.
Moon, J., B. Bandy, and A. Davison. 1992. Hypothesis: Etiology of Atherosclerosis and Osteoporosis: Are Imbalances in the Calciferol Endocrine System Implicated. Journal of the American College of Nutrition 11:567-583.
Mozar, H. N., D. G. Bal, and S. A. Farag. 1990. "The Natural History of Atherosclerosis: An Ecologic Perspective." Atherosclerosis 82: 157-164.
Muhlbaier, Lawrence H., David B. Pryor, J. Scott Rankin, L. Richard Smith, Daniel B. Mark, Robert H. Jones, Donald D. Glower, Frank E. Harrell, Jr., Kerry L. Lee, Robert M. Califf, and David C. Sabiston. 1992. "Observational Comparison of Event-Free Survival With Medical and Surgical Therapy in Patients With Coronary Artery Disease." Circulation 86:198-204.
Nabulsi, A. A., A. R. Folsom, A. White, W. Patsch, G. Heiss, K. K. Wu, and M. Szklo. 1993. "Association of Hormone-Replacement Therapy with Various Cardiovascular Risk Factors in Postmenopausal Women." New England Journal of Medicine 328: 1069- 1075.
Naessen, T., R. Parker, I. Persson, M. Zack, and H. O. Adami. 1989. "Time Trends in Incidence Rates of First Hip Fracture in the Uppsala Health Care Region, Sweden, 1965-1983." American Journal of Epidemiology 130:289-297.
National Center for Health Statistics. 1992. Health, United States. U.S. DHHS Pub. No. (PHS)92-1232. Washington, DC: U.S. Government Printing Office.
Niessen, Louis W., Jan J. Barendregt, Luc Bonneux, and Peter I. Koudstaal. 1993. "Stroke Trends in an Aging Population." Stroke 24:931-939.
Omran, Abdul. 1971. "The Epidemiologic Transition: A Theory of the Epidemiology of Population Change." Milbank Memorial Quarterly 49:509-538.
Overgaard, K., M. A. Hansen, S. B. Jensen, and C. Christiansen. 1992. "Effect of Salcatonin Given Intranasally on Bone Mass and Fracture Rates in Established Osteoporosis: A Dose-Response Study." British Medical Journal 305:556-561.
Phillips, S. J. and J. P. Whisnant. 1992. "Hypertension and the Brain." Archives of Internal Medicine 152:938-945.
Potthoff, Richard F., Max A. Woodbury, and Kenneth G. Manton. 1992. "Equivalent Sample Size and Equivalent Degrees of Freedom Refinements for Inference Using Survey Weights Under Superpopulation Models." Journal of the American Statistical Association, Theory and Methods 87:383-396.
Preston, Samuel H. 1993. "Demographic Change in the United States, 1970-2050." In Kenneth G. Manton, Burton H. Singer, and Richard M. Suzman (Eds.), Forecasting the Health of Elderly Populations. New York: Springer-Verlag.
Reuben, David B., Albert L. Siu, and Sokkun Kimpau. 1992. "The Predictive Validity of Self-Support and Performance-Based Measures of Function and Health." Journal of Gerontology: Medical Sciences 47:M106-M110.
Robine, J. M. and K. Ritchie. 1991. "Healthy Life Expectancy: Evaluation of Global Indicator of Change in Population Health." British Medical Journal 302:457-460.
Roses, A., W. Strittmatter, D. Schmechel, A. Saunders, and M. Pericak-Vance. In press. "Apo-E Predisposes to Alzheimer's Amyloidosis." Proceedings of the National Academy of Sciences.
Sempos, Christopher, James I. Cleeman, Margaret D. Carroll, Clifford L. Johnson, Paul S. Bachorik, David J. Cordon, Vicki L. Burt, Ronette R. Briefer, Clarice D. Brown, Kenneth Lippel, and Basil Rifkind. 1993. "Prevalence of High Blood Cholesterol Among US Adults: An Update Based on Guidelines From the Second Report of the National Cholesterol Education Program Adult Treatment Panel." Journal of the American Medical Association 269:3009-3014.
Skoog, I., L. Nilsson, B. Palmertz, L. A. Andreasson, and A. Svanborg. 1993. "A Population-Based Study of Dementia in 85-year-olds." New England Journal of Medicine 328: 153-158.
Social Security Administration, Office of the Actuary. 1992. Life Tables for the United States Social Security Area 1900-2080 (Actuarial Study No. 107. U.S. DHHS, SSA Pub. No. 11-11536). Baltimore, MD: Social Security Administration.
The SOLVD Investigators. 1991. "Effect of Enalapril on Survival in Patients With Reduced Left Ventricular Ejection Fractions and Congestive Heart Failure." New England Journal of Medicine 325:293-302.
Stephen, A. M. and N. J. Wald. 1990. "Trends in Individual Consumption of Dietary Fat in the United States, 1920-1984." American Journal of Clinical Nutrition 52:457-469.
Strittmatter, Warren J., Ann M. Saunders, Donald Schmechel, Margaret Pericak-Vance, Jan Enghild, Gay S. Salvesen, and Allen D. Roses. 1993. "Apolipoprotein E: High-Avidity Binding to Beta-Amyloid and Increased Frequency of Type 4 Allele in Late-Onset Familial Alzheimer Disease." Proceedings of the National Academy of Sciences 90: 1977-1981.
Sullivan, D. F. 1971. "A Single Index of Mortality and Morbidity." HSMHA Health Reports 86:347-354.
Suzman, Richard M., Kenneth G. Manton, and David P. Willis. 1992. "Introducing the Oldest Old." In Richard Suzman, David P. Willis, and Kenneth G. Manton (Eds.),The Oldest Old. New York: Oxford University Press.
Taylor, A. 1993. "Cataract: Relationships Between Nutrition and Oxidation." Journal of the American College of Nutrition 12: 138-146.
Tsuji, Ichiro, Yuko Minami, Penelope M. Keyl, Shigeru Hisamichi, Hirotake Asano, Makito Sato, and Kazoo Shinoda. 1994. "The Predictive Power of Self-Rated Health, Activities of Daily Living, and Ambulatory Activity for Cause-Specific Mortality Among the Elderly: A Three-Year Follow-Up in Urban Japan." Journal of the American Geriatrics Society 42:153-156.
U.S. Bureau of the Census. 1989. "Long-Term Care Survey Field Representative's Manual Personal Visit Interview." U.S. Department of Commerce (LTC-S(PV)). Bureau of the Census internal document.
Vertrees, James C., Kenneth G. Manton, and Gerald S. Adler. 1989. "The Cost Effectiveness of Home and Community Based Care." Health Care Finance Review 10:65-78.
World Health Organization. 1984. "The Uses of Epidemiology in the Study." World Health Organization Scientific Group on the Epidemiology of Aging. Geneva: World Health Organization.
My Research