Level of cognitive impairment for male and female centenarians. Blessed Dementia Scale scores higher than 34 indicate normal cognition; 27 to 33, mild impairment; 21 to 26, moderate impairment; and lower than 21, severe impairment. The subjects who were not able to complete the Information-Memory-Concentration test because of dementia were included in the severe impairment category.
Distribution of Barthel Activities of Daily Living Index (Barthel Index) scores by sex and morbidity group. Frequency of Barthel Index scores for male and female centenarians by morbidity group: age at onset younger than 85 years (A, female; B, male) and age at onset 85 years or older (C, female; D, male).
Terry DF, Sebastiani P, Andersen SL, Perls TT. Disentangling the Roles of Disability and Morbidity in Survival to Exceptional Old Age. Arch Intern Med. 2008;168(3):277–283. doi:10.1001/archinternmed.2007.75
Although it is commonly held that survival to age 100 years entails markedly delaying or escaping age-related morbidities, nearly one-third of centenarians have age-related morbidities for 15 or more years. Yet, we have previously observed that many centenarians compress disability toward the end of their lives. Therefore, we hypothesize that for some centenarians, compression of disability rather than morbidity is a key feature for survival to old age.
This cross-sectional, nationwide study included 523 women and 216 men 97 years or older. The participants were stratified by sex and age at onset (age <85 years [termed survivors] and age ≥85 years [termed delayers]) of chronic obstructive pulmonary disease, dementia, diabetes, heart disease, hypertension, osteoporosis, Parkinson disease, and stroke. Dependent variables were the Barthel Activities of Daily Living Index (Barthel Index) and the Information-Memory-Concentration test of the Blessed Dementia Scale.
Thirty-two percent of the participants were survivors. For men with hypertension and/or heart disease for 15 or more years, the median Barthel Index score was 90 (independence range, 80-100). For female survivors with hypertension, heart disease, and/or osteoporosis, the median Barthel Index score was 65 (minimal assistance range, 60-79). Generally, men had better function than women: 60% of male survivors had Barthel Index scores of 90 or higher compared with 18% of female survivors (P < .001) and 50% of male delayers had Barthel Index scores of 90 or higher compared with 27% of females delayers (P < .001).
Whereas the compression of both morbidity and disability are essential features of survival to old age for some centenarians, for others, the compression of disability alone may be the key prerequisite. Though far fewer in number, male centenarians tend to have significantly better cognition and physical function than their female counterparts.
With the age 85-years-and-older group comprising the fastest growing segment of the American population,1 significant concern has been expressed as to what this growth portends for society.2 Some have contended that survival to exceptional old age is accompanied by comorbidities and prolonged disability.3 In contrast, the “compression of morbidity hypothesis” optimistically posits that there will be a delay in the onset of chronic morbidity and that the length of this delay will be greater than that of any increase in life expectancy.4 In other words, individuals will live longer but experience a shorter period of morbidity. Recently, Fries5 demonstrated a decline in disability trends of approximately 2% per year, while the decline of mortality is approximately 1% per year. Other studies have shown that individuals with better health habits, such as not smoking, regular physical activity, and maintaining a normal weight, live longer and compress the number of years living with disability.6,7
The prevalence of centenarians has increased dramatically over the past 2 decades. Cynics question why anyone would want to live to age 100 years if it entails living with numerous age-related morbidities and disabilities for many years prior to death. In fact, the longevity literature suggests that the opposite may be true. Consistent with the compression of morbidity hypothesis, nearly two-thirds of centenarians delay, if not escape, age-related morbidities. A smaller proportion survive for many years with age-related morbidities.8 These results suggest that there are multiple pathways to achieving exceptional old age that need to be better understood.
Since the time that the Compression of Morbidity hypothesis was introduced, the concept of comorbidities, disability, and frailty have evolved, although their exact definitions and how they are related to one another are still a source of debate. Generally speaking, comorbidity represents the simultaneous presence of multiple diseases, disability represents loss of function, and frailty represents a state in which one is at high risk for developing adverse health outcomes.9
The purpose of this study was to determine the absence or presence of disability among centenarian survivors. Survivors were defined as individuals who developed at least 1 of the following conditions before age 85 years: chronic obstructive pulmonary disease, dementia, diabetes, heart disease, hypertension, osteoporosis, Parkinson disease, or stroke. Age 85 years was selected as a cutoff based on the commonly used definition of the “oldest old.”10 We hypothesize that survivors demonstrate unusual adaptive capacity and functional reserve to better cope with their morbidities and to maintain their independence well into their tenth decade of life. In addition, others have demonstrated that male centenarians function better than their female counterparts.11 We hypothesize that male centenarians are less able to cope with morbidity than female centenarians, and thus only the most robust men are able to survive to extreme old age.
The New England Centenarian Study (NECS) is a cross-sectional study of individuals 97 years and older (whom we herein refer to as “centenarians”) and their family members who live in the United States and Canada. Centenarians were recruited by mass mailings, local census records, and media features. Participants are not excluded by sex, race, function, cognitive ability, and living situation. The NECS protocol is approved and monitored by the institutional review board of the Boston University Medical Center, Boston, Massachusetts, and all participants or their proxies provided informed consent.
Data were analyzed from 739 subjects (523 women [71%] aged 100-119 years and 216 men [29%] aged 97-113 years). The minimum age for men and women corresponds to the top 1% survival age in the 1900 birth cohort as defined by the birth cohort life tables.12
All data for this analysis were collected via telephone interview and by mail. If a participant was unable to complete the questionnaires himself or herself, a family member or friend assisted with completion of the questionnaire. Survivors were defined as individuals who developed at least 1 of the following conditions at age 85 years or younger based on the health questionnaire: chronic obstructive pulmonary disease, dementia, diabetes, heart disease (defined as myocardial infarction, arrhythmia, and congestive heart failure), hypertension, osteoporosis, Parkinson disease, or stroke. The health history questionnaire was validated in 25 centenarian subjects (17 women and 8 men) by comparing responses to medical records. Responses from the health history questionnaire demonstrated a 100% concordance with the subjects' medical records for diabetes and stroke; a 96% concordance for cancer, dementia, Parkinson disease, and chronic obstructive pulmonary disease; a 92% concordance for arrhythmia; an 88% concordance for hypertension; an 84% concordance for CHF; and an 81% concordance for osteoporosis.
The Barthel Activities of Daily Living Index (Barthel Index) is a validated instrument that examines functional domains including activities of daily living.13 The Barthel Index scores range from 0 to 100, with 80 to 100 indicating independence in activities of daily living; 60 to 79, requires minimal assistance; 40 to 59, partial dependence; 20 to 39, heavy dependence; and lower than 20, complete dependence.13- 15
The Information-Memory-Concentration (IMC) portion of the Blessed Dementia Scale is an instrument that examines cognition and has been validated for both telephone and in-person administration.16 The instrument was administered to participants only (not proxies) in person or by telephone interview sometimes with the assistance of nursing home staff or a family member, particularly in the cases of subjects who had difficulty hearing. Scores for this instrument range from 0 to 37, with 33 or higher indicating normal cognition; 27 to 32, mild impairment; 21 to 26, moderate impairment; and 21 or lower, severe impairment.16,17 If the participant was unable to complete the test, the reason was recorded. The most frequent reasons for not completing the instrument were, in order of frequency, significant dementia, hearing and/or visual impairment, and anxiety.
Univariate and bivariate descriptive statistics were computed using the statistical package R 2.1 (http://cran.r-project.org). Variables included age, sex, age at onset of morbidity, Barthel Index score, and the IMC test score. Categories of physical and cognitive function were compared by sex and age at onset of morbidity (<85 years vs ≥85 years). Comparisons were performed using unpaired t tests and χ2 tests of independence in contingency tables. Wilcoxon rank sum tests were used for comparing medians. Monte Carlo simulations were used to compute P values to test independence in sparse contingency tables to avoid asymptotic approximations. P<.05 was considered statically significant.
Basic demographic characteristics including education, race, marital status, and living situation are given in Table 1. Both men and women were similar in terms of years of education and race, and 24% of men and 2% of women were married.
The subjects (N = 739) were categorized into 2 morbidity profiles (Table 1). Among centenarians, 32% (n = 234) were survivors or those who developed age-related morbidity before age 85 years, and 68% (n = 505) were delayers, defined as those who developed age-related morbidity at age 85 years or older (P < .001). The median Barthel Index score was 75 for both groups.
Of the male and female centenarians, 27% and 34%, respectively, were survivors. This difference was not statistically significant (P = .09). Whether they were segregated by survivor status, male centenarians had better function than female centenarians. Without segregating by survivor status, proportionately more men than women were living independently (31% vs 14%; P < .001), had higher mean Barthel index scores (80 vs 61; P < .001), and higher mean IMC test scores (28 vs 23; P < .001).
A greater proportion of men (67%) had normal cognitive function or mild impairment in cognition compared with 42% of women with similar levels of function (P < .001; Figure 1). Scores from the IMC test were unavailable for 78 women (15%) and 29 men (13%). Of the 107 subjects who did not complete the IMC test, 32 were unable to complete the test due to dementia; they were included in the severe impairment category in the subsequent analysis examining cognitive function in men vs women. For the remaining 75 individuals, 50 did not complete the test prior to their deaths, 17 were unable to complete the test due to hearing and/or visual impairment, 4 were omitted by family members requests due to concern of causing anxiety, and 4 refused. Among these 75 individuals, there were no reports of dementia.
Proportionately more female centenarians scored in the dependent range for the Barthel Index score: 16% were partially dependent, 16% were heavily dependent, and 11% were completely dependent. Among male centenarians, 11% were partially dependent, 5% were heavily dependent, and 1% were completely dependent. Female centenarian participants had a more uniform distribution of Barthel Index scores in the 2 morbidity groups (Figure 2). In contrast, male centenarian participant scores were consistently shifted to the right (better function).
As shown in Table 2, men had better function than women: among survivors, the mean Barthel Index scores for men women were 82 vs 60 (P < .001), and among the delayers, the mean scores for men vs women were 79 vs 61 (P < .001).
Of male survivors, 60% had Barthel Index scores of 90 or higher compared with 18% of female survivors (P < .001). Of male delayers, 50% had Barthel Index scores of 90 or higher compared with 27% of female survivors (P < .001).
The prevalence of the age-related morbidities included in this analysis according to age at onset and sex are provided in Table 3. Notably, for male and female centenarian survivors, 66% and 60%, respectively, had heart disease for 15 or more years; 55% and 66% respectively, had hypertension for 15 or more years; and 24% and 20%, respectively, had stroke for 15 or more years.
A key question is whether some centenarians compress disability toward the end of their lives in spite of having morbidities present for several years. In Table 4, the median Barthel Index scores are listed for each of the morbidities surveyed according to age at onset (<85 and ≥85 years) and sex. Male survivors who had hypertension and/or heart disease had Barthel Index scores in the “independent” range and the scores were consistent with that of the male delayers. Thus, male centenarians appear to compress their disability toward the end of their lives even with the history of clinically evident heart disease and/or hypertension for 15 or more years.
In the case of women, for these same morbidities plus osteoporosis, their median Barthel Index score was in the “requires minimal assistance” range and remained consistent for survivors and delayers. Women, like men, generally compress their disability even with clinically evident disease, though their level of function tends to be less.
In our sample 32% of the centenarians (N = 739) were survivors or those who developed at least 1 of the age-related morbidities surveyed in this study before age 85 years; thus, morbidity was not compressed toward the end of these exceptionally long life spans.
Yet, centenarians who had developed heart disease and/or hypertension before age 85 years and still survived to 100 years demonstrated similar levels of function (“independent” in the case of men and “requires minimal assistance” in the case of women) as those who delayed morbidity until after age 85 years.
Interestingly, male survivors with diabetes mellitus maintained significantly better function (P = .03) than delayers. One possibility for this observation is that men who develop diabetes mellitus later in life may have greater difficulty adapting because of other age-related comorbidities, thus resulting in a more rapid functional decline. Another possibility is that individuals who were diagnosed later may have had previously unrecognized and therefore poorly controlled diabetes, resulting in functional decline.
The literature focusing on the “compression of morbidity” hypothesis has used declining disability rates18,19 to support this paradigm, in large part because of a lack of available data for examining morbidity rates.20 Our research extends the work in this field by suggesting that the “compression of morbidity” and the “compression of disability” may in fact be different phenomena. The delayers exemplify the “compression of morbidity” hypothesis4 in that they compress the development of major age-related morbidities until the latter years of their very long lives. The survivors in this study, however, are an interesting exception as individuals who have had clinically significant age-related morbidity for many years and have therefore not compressed their morbidity but, in many cases, have compressed their disability.
Rates of disability prevalence have been examined in other cohorts. In a Danish cohort of 2262 nonagenarians, 50% of the men and 41% of the women were categorized as “nondisabled.”21 We demonstrated similar rates of functional independence for female centenarians (38%); however, a larger proportion of the male centenarians (66%) were independent.
Our results suggest that male centenarians maintained better physical and cognitive function than female centenarians. This is consistent with other studies that have demonstrated that male centenarians are physically and cognitively more fit than their female counterparts.11 These observations may indicate a demographic crossover in which women have better function than men in younger old age, and men, although fewer in number, have better function than women in extreme old age. One explanation for this may be that men must be in excellent health and/or functionally independent to achieve such extreme old age. Women on the other hand may be better physically and socially adept at living with chronic and often disabling health conditions.21- 24
Marital status may be an important contributor to the male centenarians' ability to maintain physical and cognitive function; 24% of men were married in contrast to 2% of female centenarians. It has been noted that individuals with more social ties have less functional25 and cognitive decline.26,27 However, 31% of our male centenarian participants lived alone compared with 14% of female centenarians, suggesting that marriage alone cannot explain the functional advantages that the male participants demonstrated.
The preserved physical and cognitive function among male centenarians should not detract from the overall successful survival noted among female centenarians. In fact, women are much more likely to survive to age 100 years. According to US Census Bureau estimates, approximately 79% of centenarians are women.28 Furthermore, by looking at the overall numbers rather than the percentages, there are more than 3 times as many female survivors and twice as many female delayers.
The relationships among morbidity, disability, and frailty are complicated, overlapping, and interrelated.9 Unfortunately, we were not able to examine the role of frailty in this study, although anecdotally we have observed that many of the centenarians enrolled in the NECS exhibit signs of frailty such as weight loss, muscle and strength loss, loss of endurance, and loss of balance. Ideally one would like to have access to longitudinal data to identify at what point individuals begin to clinically express a disease and when they experience a functional decline and frailty. With the accumulation of large enough samples of these individuals, large and long-term longitudinal studies, such as the Baltimore Longitudinal Study on Aging and the Framingham Heart Study, should eventually be able to more accurately assess the relationship between morbidity, disability, frailty, and longevity. If the prevention and delay of disability proves to be a powerful marker of exceptional longevity, it will be important to understand the underlying mechanisms, both genetic and behavioral.
We acknowledge several limitations in our study. The study is cross-sectional and consists of US centenarians (approximately 97% white); therefore, the results may not be generalizable to other centenarian samples. However, to obtain a sample as generalizable as possible, the NECS does not exclude individuals by sex, race, functional ability, cognitive ability, and living situation. It is possible that our study was biased toward healthier centenarians, but the breakdown of the living situations for the participants suggest otherwise. While 14% of the women and 31% of the men lived independently, the remainder required some degree of support, with 42% of women and 20% of men requiring constant care. These proportions are consistent with our previous assessment of living situations for centenarians in a population-based study.29
While we do examine current function, the questions about age-related morbidities and their age at onset rely on the recollection of participants and their family members. Self-report of age-related morbidities and functional abilities can be subject to biases; however, the health questionnaire used in this study was validated (see the “Methods” section). It should also be noted that the reported age at onset of morbidity can be different from the actual onset and that the participants may have had subclinical morbidity for some time prior to their diagnoses.
We acknowledge that our list of age-related morbidities is not exhaustive. In particular, data on important disabling conditions such as osteoarthritis were not available. The conditions assessed in the health questionnaire were selected based on the literature suggesting that older individuals are best at recalling major health conditions as opposed to minor ones.30 We considered including nonskin cancers but opted against doing so because an individual who was cured of cancer 20 years ago may not have any function-limiting sequelae. We were not able to quantify the degree to which each morbidity could contribute to loss in function. Finally, owing to the nationwide nature of our study population, we were unable to perform physical performance assessments and to examine differences in morbidity severity.
When we first embarked on the NECS we hypothesized that to live to extreme old age, one had to delay or escape age-related morbidities usually associated with increased mortality. However, the subsequent discovery that nearly one-third of male and female centenarians live with such morbidities for 15 or more years led us to rethink this hypothesis.8 We report herein that a substantial proportion of survivors markedly delay or escape disability, with 72% of male survivors and 34% of female survivors with scores in the “independent” range of the Barthel Index. Thus, for some centenarians, compression of both morbidity and disability is an essential feature of their survival to such old age. For others, particularly men, compression of disability and not necessarily morbidity, may be the key prerequisite.
The sex differences among centenarians regarding functional status are dramatic. Our findings confirm those of other studies that compared with female centenarians, male centenarians, although generally far fewer in number, tend to have significantly better cognitive and physical functional status.
Untangling the relationship between compression of morbidity and compression of disability, 2 concepts that have previously been used interchangeably, may shed additional light on the various ways in which people can survive to extreme old age. Determining the mechanisms that facilitate the delay or escape of disability in the face of clinically evident age- and mortality-associated morbidities merits further investigation.
Correspondence: Dellara F. Terry, MD, MPH, Geriatrics Section, Boston Medical Center, 88 E Newton St, Robinson 2, Boston, MA 02118 (firstname.lastname@example.org).
Accepted for Publication: March 16, 2007.
Author Contributions: Drs Terry and Perls had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Terry, Sebastiani, Andersen, and Perls. Acquisition of data: Terry, Andersen, and Perls. Analysis and interpretation of data: Terry, Sebastiani, and Perls. Drafting of the manuscript: Terry, Andersen, and Perls. Critical revision of the manuscript for important intellectual content: Terry, Sebastiani, and Perls. Statistical analysis: Sebastiani. Obtained funding: Terry and Perls. Administrative, technical, and material support: Terry and Andersen. Study supervision: Perls.
Financial Disclosure: None reported.
Funding/Support: This work was supported by National Institute on Aging grants K08 AG22785 and K23 AG026754 (Paul Beeson Physician Faculty Scholar in Aging Awards) (Dr Terry) and grant K24 AG025727 (Dr Perls).