Successful Aging and Well-Being: Self-Rated Compared With Rowe and Kahn

Abstract

Purpose: This research evaluates the utility of two different definitions of successful aging in predicting well-being. Design and Methods: We assessed the definitions of (a) self-rating and (b) Rowe and Kahn's criteria of absence of disease, disability, and risk factors; maintaining physical and mental functioning; and active engagement with life. We made associations with well-being for each definition using data from 867 Alameda County Study participants aged 65–99 years. Results: The percentage of those rating themselves as aging successfully was 50.3% compared with 18.8% classified according to Rowe and Kahn's criteria. Although absence of chronic conditions and maintaining functioning were positively associated with successful aging for both definitions, many participants with chronic conditions and with functional difficulties still rated themselves as aging successfully; none were so classified according to Rowe and Kahn's criteria. On 14 of 15 measures, self-rated successful aging resulted in sharper contrasts for well-being. Implications: Understanding criteria used by older persons to assess their own successful aging should enhance the conceptualization and measurement of this elusive concept.

Decision Editor: Laurence G. Branch, PhD

Although the concept of successful aging goes back over 50 years (Baker 1958; Butler 1974; Pressey and Simcoe 1950), the term received only minimal use until popularized in a 1987 article in Science by John Rowe and Robert Kahn in which they argued that what many viewed as effects of aging were, in fact, effects of disease. They proposed that those aging successfully would show little or no age-related decrements in physiologic function, whereas those aging "usually" would show disease-associated decrements, often interpreted as the effects of age (Rowe and Kahn 1987). According to Edward Masoro 2001, the Rowe and Kahn definition was attractive because of its implication that it was possible to reach advanced age free of age-associated disease and without experiencing significant physiological deterioration, although the number of such persons was likely to be very low.

Researchers analyzing successful aging using Rowe and Kahn's definition usually modified it by defining as successful those who exhibited minimal (rather than no) disease and disability or who exhibited high levels of physical functioning (Guralnik and Kaplan 1989; Roos and Havens 1991; Seeman, Rodin, and Albert 1993). Even with such modified criteria, the proportion of those classified as aging successfully in these referenced studies was relatively low, ranging from 20% to 33%.

Other researchers followed Schmidt 1994 definition of successful aging as minimal interruption of usual function, although minimal signs and symptoms of chronic disease may be present (Manton and Stallard 1991; Strawbridge, Cohen, Shema, and Kaplan 1996). In these studies the proportion of those classified as aging successfully could exceed 50%, which changed the focus from a minority to a majority of older persons. A third approach involved Baltes and Carstensen 1996 definition which describes successful aging as doing the best with what one has. These latter two, broader-based approaches allow for the presence of chronic disease and, thus, represent a more attainable goal for the majority of persons approaching old age than the more restrictive definitions focusing on those who enter old age relatively unscathed. With the exception of Baltes and Carstensen's definition, however, definitions of successful aging still place a strong emphasis on health and physical functioning rather than a more inclusive conceptualization that would include well-being.

In 1998, Rowe and Kahn expanded their definition to include three criteria: (a) absence of disease, disability, and risk factors like high blood pressure, smoking, or obesity; (b) maintaining physical and mental functioning; and (c) active engagement with life. This last criterion included both being connected to other persons and engaging in productive activities. To be aging successfully one had to meet all three criteria, making it likely that successful aging would still describe only a relatively small proportion of older persons and exclude those with chronic conditions.

Both Rowe and Kahn definitions of successful aging had important positive consequences: No longer could all age-related deficits be dismissed as inevitable concomitants of old age, and an examination of environmental and lifestyle factors that could improve well-being in old age was encouraged along with a shift in focus from those doing poorly to those doing well. However, the implication that life-style changes could ward off most chronic conditions had a potential unintended consequence of reducing interest in better managing age-related functional declines through secondary and tertiary prevention (Kaplan and Strawbridge 1994). The adjective successful has itself proven problematic because it implies a contest in which there are winners and losers; most gerontologists are not ready to call someone unsuccessful merely because he or she is disabled or diagnosed with diabetes. Alternative terms used by other researchers include healthy aging, aging well, effective aging, and productive aging (Baltes 1994; Curb et al. 1990; LaCroix, Newton, Leveille, and Wallace 1997; Morrow-Howell, Hinterlong, and Sherraden 2001); however, except for aging well one could find equal fault with the semantics of these terms as with successful aging.

Whatever term is used, it would be helpful to have a tool able to distinguish between those who are experiencing positive outcomes in old age across a variety of dimensions and those who are not. Thus one characteristic of a good definition of successful aging should be its use in differentiating older persons on quality of life outcomes. Given the variations in the definitions of successful aging, it would also be useful to see what could be learned by letting older persons rate their own success at aging and then compare the associations of these ratings and quality of life outcomes with those obtained by using a definition proposed by health professionals.

The research reported here is based upon just such a comparison. We use a large, representative sample of older persons to compare the prevalence and characteristics of those who self-rate as aging successfully versus those classified using the three criteria proposed by Rowe and Kahn 1998. We compared also relationships with 15 measures of well-being between the two definitions.

Methods

Study Population

We drew participants from the Alameda County Study, a longitudinal study of health and functioning begun in 1965 consisting of 6,928 subjects who were selected by using a randomized household sample stratified on the basis of median county household income (Berkman and Breslow 1983). Alameda is a large urban California county and includes the cities of Berkeley and Oakland. In 1965, its population was representative of the larger population of the United States in terms of gender, age, and minority representation. Subjects still enrolled remain representative of the older United States community-dwelling population on a wide range of variables, including age, ethnicity, and prevalence of chronic conditions. Follow-ups have been conducted in 1974, 1983, 1994, 1995, and 1999 with response rates ranging from 85% to 97%. We sent subjects aged 65 years and older who responded to the 1999 general questionnaire an additional questionnaire containing specific questions on age-related items, including successful aging, activities in old age, and general quality of life. A total of 907 men and women returned these questionnaires for a response rate of 89%; we omitted 40 respondents because they were missing data on the items used for the two definitions of successful aging. Of the 867 respondents remaining, 484 were women and 383 were men; age ranged from 65 to 99 years with a mean of 75 years. African Americans constituted 5.5% of the participants, Asians constituted 4.2%, Hispanics constituted 2.0%, and Native Americans constituted 1.4%. For education, 13.8% had fewer than 12 years, 30.9% had 12 years, and 55.3% had more than 12 years. The percentage of those reporting financial problems (any report of not having enough money in the past 12 months to buy clothing, fill a prescription, see a doctor, pay rent or mortgage payments, or buy food [last 30 days]) was 16.2%. For self-rated health, 23.0% said their health was excellent, 57.8% said it was good, 16.1% said it was fair, and 3.1% said it was poor.

Definitions of Successful Aging

Self-rated Successful Aging

We measured self-rated successful aging by asking participants a single question: How strongly did they agree or disagree with the statement "I am aging successfully (or aging well)"? Response categories were agree strongly, agree somewhat, disagree somewhat, or disagree strongly. Sensitivity analyses on a variety of outcome measures indicated that those agreeing only somewhat were more similar to those who disagreed with the statement than to those who agreed strongly, so we classified only those who agreed strongly as aging successfully; we classified all other responses as not aging successfully.

Rowe and Kahn Successful Aging

Rowe and Kahn 1998 do not operationalize their definition; we operationalized the definition according to their three criteria and specific examples as described in their text (1998): (a) absence of disease, disability, and risk factors; (b) maintaining physical and mental functioning; and (c) active engagement with life. Each will be examined in turn.

Absence of disease, disability, and risk factors.

For absence of disease, we included absence of heart disease, stroke, bronchitis, diabetes, cancer, osteoporosis, emphysema, or asthma. Absence of disability included being able to perform all 7 activities of daily living (bathing, dressing, eating, using the toilet, moving from bed to chair, grooming, or walking across a room). Absence of risk factors included absence of cigarette smoking, hypertension, and obesity, defined according to the National Heart, Lung, and Blood Institute's (1998) guidelines as a body mass index of 30 or greater based upon reported height and weight.

Maintaining physical and mental functioning.

Maintaining physical and mental functioning includes the ability to walk 1/4 mile, the ability to climb one flight of stairs without resting, the ability to stand up without fainting or feeling dizzy, and the ability to remember things without difficulty, to remember where one put something, or to find the right word when talking.

Active engagement with life.

Connections with persons included reporting monthly contact with three or more close friends or relatives. Being productive included reporting any of the following: (a) paid employment, (b) caring for a child or grandchild, (c) active volunteering, or (d) cleaning house. (The last item for productivity is specifically included by Rowe and Kahn 1998.)

As specified by Rowe and Kahn 1998, participants had to meet all three criteria in order to be scored as aging successfully.

Dichotomous Measures of Well-Being

We derived the measure the best old age one could expect from a single item modeled on Cantril 1965 Ladder as revised by Andrews and Withey 1976. On the basis of their current quality of life, participants are asked to select a number or rung on the ladder that corresponds with the best to worst old age they could expect to have. Best old age corresponds to a score of 8 or 9, so this variable was divided at a score of 8 or 9 versus all lower scores.

We assessed the measure very happy by one item ("All in all, how happy are you these days?") categorized as very versus pretty or not too happy.

We based the measure pleased with how life turned out upon a single variable asking for level of agreement or disagreement with "When I look at the story of my life I am pleased at how things have turned out." Responses are divided into strongly agree compared with moderately agree or disagree.

To assess the measure much more energy than others, we asked participants "Would you say you have more or less energy than most people your age?" Responses were divided into much more energy compared with a little more or a little less or a lot less.

To assess the measure enjoy free time a lot, we asked the participants "All in all, how much enjoyment do you get out of your free time?" Responses were divided into a lot compared with some or not very much.

To assess the measure not depressed, we used a set of 12 items that operationalize the diagnostic symptom criteria for a major depressive episode outlined in the American Psychiatric Association 1994Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV). Time frame was almost every day during the last two weeks. Scoring followed the DSM-IV algorithm. About 9% of Alameda County Study subjects scored sufficiently high to be considered depressed (Roberts, Kaplan, Shema, and Strawbridge 2000). Scoring was reversed so that a positive score indicated not being depressed.

With the measure excellent mental or emotional health, we mirrored the most commonly employed self-rated physical health question by asking "All in all, would you say that your emotional or mental health is excellent, good, fair, or poor?" Responses ran approximately 40% for excellent, 50% for good, and 10% for a combined fair or poor. As such it can be used in the opposite way of the DSM-IV depression measure to select those with very high levels (excellent) of emotional or mental health, which is how we scored it here.

With the measure very much feel loved and cared about, we asked "Do you feel loved and cared about?" Responses were divided into very much compared with somewhat or little or very little.

To assess the measure very satisfied with relationships, we asked "Overall, how satisfied are you with your friendships and relationships with others?" Responses were divided into very compared with somewhat or not at all.

We asked two questions of married participants to reflect marital satisfaction. To assess the measure marriage has been very happy, we asked "All in all, how happy has your marriage or relationship been for you?" Responses were divided into very happy compared with happy, somewhat happy, somewhat unhappy, or unhappy or very unhappy. To assess the measure rarely feel not a good spouse, we asked participants to indicate how often they feel they are not as good husbands, wives, or partners as they would like to be. Responses were divided into few times or never compared with sometimes or often.

Continuous Measures of Well-Being

The four continuous measures were all based upon established scales. The Life Orientation Test (LOT) included all six items designated by Scheier, Carver, and Bridges 1994 in their revised Life Orientation Test and was scored as they recommend (higher scores reflect an optimistic orientation whereas lower scores reflect a pessimistic orientation). Reliability (standardized Cronbach's α) reported by Scheier and colleagues 1994 was .68 with younger subjects; the α for our sample was .74. We assessed Perceived Control with Wallhagen's (Wallhagen & Kagen, 1993; Wallhagen and Lacson 1999) 15-item Perceived Control scale (derived from her original 30-item scale) that assesses the extent to which subjects feel their current situation is under control. Reliability for the 15-item version on Alameda County Study subjects was .89. The measure Affect Balance contained the same items and scoring (negative scores subtracted from positive) recommended by Bradburn 1969 for his revised scale. Standardized Cronbach's alpha using our participants was .74. The measure Lower Cynical Distrust contained 7 of the 8 items that were derived factor-analytically from the 50-item Cook-Medley Hostility Scale (Cook and Medley 1954; Greenglass and Julkunen 1989). Sample items included "I think most people would lie to get ahead" and "Most people are honest chiefly because of a fear of being caught." Response options ranged from strongly agree to strongly disagree. This scale has been shown by Human Population Laboratory researchers to predict mortality and myocardial infarction for subjects from the Kuopio Ischemic Heart Disease Study (Everson et al. 1997). The standardized Cronbach's alpha of .81 for the Alameda County Study subjects is the same as reported in the Kuopio analysis. To be consistent with the positive direction of the other three continuous scales, we reverse coded scores for this scale so that higher scores reflected lower levels of cynical distrust.

Statistical Analysis

First, we compared the results of the two successful aging definitions in a simple cross-tabulation format. Then we broke out the percentage of those aging successfully for each definition by selected characteristics, including demographic variables, financial problems, self-rated health, number of chronic conditions, and maintaining physical and cognitive functioning.

We fit separate logistic regression models by using each of the eleven dichotomous well-being measures as an outcome and each definition of successful aging as a predictor. For example, we regressed "best old age could expect" on self-rated successful aging in one logistic model and as defined by Rowe and Kahn in a separate logistic model. We adjusted all logistic models for age and gender.

Similarly, we ran separate multiple linear regression models for each of the four continuous well-being outcomes. To better understand why the regression coefficients were larger for the self-rated results, we calculated also the unadjusted mean scores for those aging successfully compared with those not aging successfully for both models. We performed all statistical analyses with the use of SAS software, version 6.12 (SAS Institute, Inc 1996).

Results

Half (50.3%) of the participants rated themselves as aging successfully, whereas we defined only 18.8% as aging successfully by Rowe and Kahn's three criteria. A cross-tabulation of the two definitions is shown in Table 1 . Of the 163 participants classified as aging successfully according to Rowe and Kahn's criteria, 60 (36.8%) did not rate themselves as aging successfully. Similarly, of the 704 participants classified as not aging successfully by the Rowe and Kahn criteria, 333 (47.3%) reported that they rated themselves as aging successfully.

Table 2 shows the percentage of participants aging successfully according to each definition by a series of individual characteristics. Both definitions resulted in higher prevalences of successful aging among women, younger participants, and those with no financial problems. We found differences on ethnicity (lower rates of successful aging among African Americans than Whites for the Rowe and Kahn model) and education (the self-rated model showed no essential differences whereas a gradient favoring those with more education appeared for the Rowe and Kahn model). Results for self-rated model health revealed decreasing proportions of those aging successfully for both definitions as the responses moved from excellent to poor, but the contrasts were sharper for the Rowe and Kahn model where only 2.9% of those rating their health as fair were classified as aging successfully compared with 17.1% for the self-rated model.

The biggest differences between the two definitions occurred in the number of chronic conditions and maintaining physical and mental functioning. Whereas Rowe and Kahn limited successful aging to those with no chronic conditions, 42.7% of those with one chronic condition considered themselves to be aging successfully; for two conditions the figure was 35.0% and for three or more it was 16.7%. We found similar results for maintaining physical and cognitive functioning, where 35.5% of those not physically or cognitively fit still reported themselves to be aging successfully.

The ability of the two definitions of successful aging to predict well-being for the eleven dichotomous measures is shown in Table 3 . Table 4 shows the two definitions' ability to predict well-being for the four continuous measures. In Table 3 , we used odds ratios to show the strength of the associations between each definition of successful aging and well-being. For measures with low prevalence such as depression (9%) and excellent mental or emotional health (15%), the odds ratios approximate relative risk. For 10 of the 11 measures of well-being, the association was stronger for successful aging that was self-rated than for successful aging that was measured by Rowe and Kahn criteria. The exception was not depressed, which had an odds ratio of 3.85 for self-rated successful aging compared with 5.42 for successful aging rated by Rowe and Kahn criteria. All eleven associations were statistically significant for the self-rated model; odds ratios for the two marital satisfaction outcomes were not statistically significant when we used Rowe and Kahn's definition.

Table 4 presents results for the four continuous measures. Because both successful aging classifications are dichotomous, the nonstandardized coefficients in the table represent age-and-sex-adjusted mean differences in the respective scales between those aging successfully and those classified as not aging successfully. Thus the adjusted mean score difference between those aging successfully and those not aging successfully by self-rating was 1.77 on the Life Orientation Test, 6.52 on perceived control, 1.78 on affect balance, and 1.18 on lower cynical distrust; comparable figures devised by using Rowe and Kahn's definition were 0.93, 2.66, 1.19, and 0.75, respectively. All mean differences were statistically significant, but all differences were larger for self-rated successful aging than for Rowe and Kahn's definition.

The reason for the larger mean differences associated with self-rated successful aging became apparent when the unadjusted means of the four measures were compared between definitions (Table 5 ). There was less difference between the means for those aging successfully according to the two definitions than there was for those not aging successfully. Further, for all four outcomes, means were lower for those classified as not aging successfully by self-rating than for those classified as not aging successfully by the Rowe and Kahn definition. Thus, those classified as aging successfully by either definition had similar scores on the four well-being measures. However, for those classified as not aging successfully, the well-being scores were lower for those using self-rating than for those defined as not aging successfully by the Rowe and Kahn definition.

Discussion

To assess the usefulness of successful aging as a concept, we compared the ability of two definitions to predict well-being in older persons on a range of measures.

When asked to classify their own status, 50.3% of the participants rated themselves as aging successfully compared with 18.8% obtained by applying Rowe and Kahn's three rather stringent criteria: (a) absence of disease, disability, and risk factors; (b) maintaining physical and mental functioning; and (c) engagement with life. This large difference is interesting in itself, because it indicates that a much higher proportion of older persons consider themselves to be aging successfully than is indicated by the most popular definition proposed by health professionals.

Other researchers might operationalize Rowe and Kahn's three criteria so as to obtain an even smaller proportion of those aging successfully. For example, if not having the most common chronic condition in old age (arthritis) were included in Rowe and Kahn's first criterion, the proportion of those defined as aging successfully in our study would drop from 18.8% to 13.6%. A further drop would occur if we add a more vigorous measure for physical functioning than the mere absence of disability or frailty. It would be interesting to see how other researchers might measure the three criteria and to what extent the proportion of those defined as aging successfully would change as a result.

The sharper contrasts in well-being resulting from self-rated successful aging than those resulting from Rowe and Kahn's definition demonstrate that a self-assessment such as the one we obtained is not meaningless. As we noted in the introductory paragraph of this article, any successful aging definition must be reflected in well-being to be valid, unless one wants to restrict success merely to longevity or absence of disability.

Depression, measured according to DSM-IV, was the one outcome where the Rowe and Kahn classification system resulted in a sharper contrast. Because the Rowe and Kahn criteria preclude disease and disability, which are both strong predictors of clinical depression in old age (Roberts, Kaplan, Shema, and Strawbridge 1997), it follows that greater numbers of depressed individuals would be classified as not aging successfully. In contrast, on the broader outcome of excellent mental health compared with good, fair, or poor mental health, self-rated successful aging resulted in the sharper distinction.

The reason for the sharper contrast between well-being scores using self-rated successful aging is apparent through an examination of the actual mean scores associated with the continuous measures. The well-being scores of those classified as aging successfully using either definition is similar. However, greater numbers of individuals who have high well-being scores are classified as not aging successfully according to the Rowe and Kahn definition as opposed to the self-rated definition. Thus, the self-rated definition would likely prove more useful in identifying older persons with low levels of well-being who could become then the focus of further assessment or targeted interventions.

The research reported here should not be construed as saying that physical health and functioning are not important components of successful aging. The proportion of those saying they are aging successfully declines as the number of prevalent chronic conditions increases and is lower for those with functional difficulties. But clearly such criteria are not the whole story. Not only do significant numbers of persons living with such conditions still rate themselves as aging successfully, significant numbers of persons lacking such conditions rate themselves as not aging successfully. On the basis of results obtained when the self-rated data are used, we find that successful aging appears to be a complex concept. This complexity is perhaps akin to the complexity inherent in responses to the seemingly simple question of whether one's health is excellent, good, fair, or poor (Strawbridge and Wallhagen 1999).

Those participants who were inconsistent in terms of the two definitions may hold the key to understanding what successful aging is all about. Discovering why some older persons with no chronic conditions rate themselves as not aging successfully whereas others with multiple chronic conditions feel they are aging successfully might provide insights into how to promote quality, as well as quantity of life. Thus, our own next step is to conduct in-depth interviews with a number of our participants to better understand the criteria they are using to classify themselves as aging successfully or not. We urge other researchers to do the same.

This research was supported by a grant from the AARP Andrus Foundation and by Grant 1R37AG11375 from the National Institute on Aging. We acknowledge the contributions of Betsy Sprouse in developing the question used for self-rated successful aging, Dr. Anita Stewart in suggesting the Cantril's Ladder question, and Sarah Shema and Rosemary Plank for assisting with data collection.

References