Life Course Social and Health Conditions Linked to Frailty in Latin American Older Men and Women

Abstract

Background. Gender, social conditions, and health throughout the life course affect functional health in later life. This article addresses two specific hypotheses: i) life-course social and health conditions are associated with frailty; and ii) differential exposure and/or vulnerability of women and men to life-course conditions may account for gender differences in frailty.

Methods. Data originated from a cross-national survey of older adults living in five large Latin American cities. Frailty was defined as the presence of three or more of five criteria: unintentional weight loss (10 pounds during the past year), self-reported exhaustion/poor endurance, weakness (grip strength), limitations in lower extremities, and low physical activity; a prefrail state was defined as the presence of one or two of the above criteria. Associations between frailty and social and health indicators were examined using a proportional odds ordinal logistic regression.

Results. Prevalence of frailty varied from 0.30 to 0.48 in women and from 0.21 to 0.35 in men. Childhood (hunger, poor health, and poor socioeconomic conditions), adulthood (little education and non-white-collar occupation), and current social conditions (insufficient income) were associated with higher odds of frailty in both men and women. Comorbidity and body mass index were related to frailty, but their effects differed in women and men. Male/female age-adjusted odds of frailty varied from 1.55 (Bridgetown) to 2.77 (Havana). Differential exposure and vulnerability partially explained differences between women and men.

Conclusion. Theoretical models to explain gender and social differences in frailty should use a life-course perspective.

FRAILTY is defined as a biologic syndrome of decreased reserve and resistance to stressors, resulting from cumulative declines across multiple physiologic systems, and causing vulnerability to adverse outcomes (1,2). Older persons with chronic diseases (3,4) and certain conditions (i.e., obesity or depression) (5) are more likely to be defined as frail. Social conditions present during the life span and related to endocrine and other physiological changes may also explain the frailty syndrome (6). Low birth weight as well as poor childhood cognition and motor development—all of which may result from poor social conditions—are important early-life predictors of muscle strength, sarcopenia, and sensory loss (7–10). Non-white-collar occupation, insufficient income, and poor social networks have been also identified as associated factors (11). Frailty may be associated with gender (12,13). Higher levels of exhaustion and weakness, as well as low levels of physical activity, have been reported among women as compared to men (14–16); the same applies to the prevalence of the frailty syndrome, regardless of how it is defined (16,17). Researchers have shown that life-course social factors are gender-specific, that is, women are more frequently exposed and, in some cases, more vulnerable to them (18–20).

In this article, frailty among Latin American older populations is examined. The specific aging experience of these populations (21), characterized by poverty and poor social conditions along with high comorbidity and disability and a scarcity of health and social services, has only recently been recognized (22). Our own research among Latin American older persons indicates that a poor material environment during childhood is related to poor physical functioning and mental health (23,24) and that men's and women's differential exposure and vulnerability to social conditions and biological factors are associated with gender differences in physical function and mental health in Latin American older persons (24,25). As a result of these previous findings, two hypotheses are examined in this study: i) life-course social and health conditions are associated with frailty; and ii) differential exposure and/or vulnerability to life-course social and health conditions may account for gender differences in frailty. Linking life-course factors to frailty will increase our understanding of the social origins of frailty.

Materials and Methods

Study Population

The SABE project (Salud Bienestar y Envejecimiento; Spanish for Health, Well-being and Aging) is a multicentric cross-sectional study, conducted in 1999–2000, involving 10,661 men and women 60 years old or older, in seven Latin American and Caribbean (LAC) cities: Buenos Aires, Argentina (n = 1043); Bridgetown, Barbados (n = 1508); Sao Paulo, Brazil (n = 2143); Santiago, Chile (n = 1301); Havana, Cuba (n = 1905); Mexico City, Mexico (n = 1311); and Montevideo, Uruguay (n = 1450) (26). The aim of the SABE survey was to study health and well-being in older people in Latin America and the Caribbean. Detailed information on sampling and the SABE questionnaire are reported elsewhere (26). In this study, data from Buenos Aires and Montevideo were not analyzed, as data on physical and/or anthropometric measures were not collected. Response rates varied from 95.3% in Havana to 80.3% in Bridgetown. Assisted interviews were conducted in 4.3% of cases in Bridgetown, 5.9% in Mexico City, 9.0% in Havana, 9.2% in Santiago, and 12.9% in Sao Paulo.

Physical Frailty

Frailty was operationalized using the five components proposed by Fried and colleagues (2). These components are nutrition, strength, endurance and energy, mobility, and physical activity.

Nutrition was defined self-reported unintentional weight loss of >10 pounds (3 kg) during the previous three months (one point). SABE data do not provide objective measurements of weight loss.

Strength was defined as physical performance on the grip strength test. For participants able to take the test, weakness was defined according to gender and body mass index (BMI) (23). For men, BMI was grouped into four categories: ≤24; 24.1–26; 26.1–28; and > 28. For each category, the cutoffs for grip strength were set at ≤29; ≤30; ≤30; and ≤32. For women, BMI was categorized as ≤23; 23.1–26; 26.1–29; and >29. The corresponding grip strength cutoffs were ≤17; ≤17.3; ≤18; and ≤21. Respondents fulfilling the criteria and unable to take the test because of physical limitations were assigned one point.

Endurance and energy were defined based on two questions on the Geriatric Depression Scale (27) (the scale used in SABE to measure depressive symptoms): “Do you have lots of energy (yes/no)”; and “Have you dropped many of your activities or interests (yes/no)”. A negative response to the first question and/or a positive response to the second were considered indications of poor endurance/lack of energy.

Mobility

Walking time was not measured in the SABE survey as in Fried and colleagues (2001). Herein we considered limitations in lower extremity mobility. Participants were considered to have lower body functional limitations if they experienced difficulty walking 100 yards and/or climbing one flight of stairs (28).

Physical activity

Low energy expenditure was assessed via the question: “In the last twelve months, have you exercised regularly or participated in vigorous physical activity such as playing a sport, dancing or doing heavy housework 3 or more times a week?” Respondents answering “no” were assigned one point. The SABE surveys did not measure the number of kilocalories per week expended doing exercise.

Data on certain frailty components were missing on <10% of respondents in all cities, with the exception of 15% with respect to strength (Mexico City and Sao Paulo) and 18% with respect to mobility (in Bridgetown). Respondents for whom data were available on all or all but one of the components of frailty were included in this analysis. An ordinal variable was created following Fried and colleagues (2001), that is, nonfrail if the person did not score positively with respect to any of the five frailty components (0), prefrail if the person scored positively with respect to one or two components (1), and frail if the person was scored positively with respect to three or more (2).

Life-Course Conditions

Childhood health and socioeconomic circumstances were assessed via the questions: “During the first 15 years of your life: 1) What was your family's economic situation? (good/average/poor); 2) Was your health excellent, good or poor? (excellent/good/poor); 3) Were there times when you went hungry? (yes/no).” Adulthood socioeconomic circumstances were determined by the respondent's education and lifelong occupation. Four levels were used to measure education: No schooling (no formal education, participant cannot read or write); primary (between 1 and 6 years of education); some secondary (between 7 and 12 years); and postsecondary (>12 years). Occupation was recorded according to the International Standard Classification of Occupations (ISCO-88) and grouped into five categories as reported previously (24): a) higher level white collar (HWC); b) lower level white collar (LWC); c) skilled and unskilled blue collar workers; d) housewives; and e) farm workers. Housewives were analyzed in a separate category, as 9.7% (n = 1031) of the Latin American women in the sample had never been gainfully employed. Current material and social resources were defined as perceived sufficiency of income and marital status. Perceived income was self-reported as sufficient or insufficient. Marital status was categorized as one of two groups, that is, presence or absence of a partner.

Current Health Conditions

Comorbidity was established as self-report of hypertension, diabetes, cancer, lung disease, heart disease, stroke, and/or arthritis. A summed score of any of the medical conditions reported by respondents was created (range 0–7) and further categorized as 0–1 versus ≥2 reported chronic conditions. Participants' weight and height measurements were taken according to standard protocols. BMI was calculated as kilograms per meter squared.

Statistical Analysis

First, a descriptive analysis was performed on the breakdown of life-course social conditions, comorbidity, and frailty components by sex. Odds ratios (ORs) from proportional odds logistic regressions were obtained to relate the variable frailty (frail, prefrail, frail), to each predictor. First, age-adjusted proportional ORs by gender, life-course conditions, and comorbidity in each of the five Latin American cities were obtained (29). To test the differential exposure hypothesis, the gender effect was estimated in an equation including all life-course exposures, comorbidity, and age for each city. For the differential vulnerability hypothesis, interactions between gender and life-course conditions were tested. Interaction terms were tested by blocks according to life-course conditions, that is, first, all interactions with childhood conditions; second, interactions with education and occupation; third, interactions with marital status and perceived income in the same block; and, fourth, interactions with health conditions. Interactions were included in final models if the deviance test was significant at a p level of.002 (0.05/20 = four blocks by five countries).

Results

Distribution of life-course social conditions, comorbidity, and each of the components of frailty are reported in Table 1. More than 50% of respondents considered that their families lived in poverty when they were children and perceived their health status as low; a larger proportion of respondents living in Santiago, Sao Paulo, and Mexico reported having experienced hunger as compared to their counterparts from Bridgetown and Havana. Women's living conditions during childhood were generally better than men's. The rate of illiteracy and manual work (blue collar, farmer) was higher in three cities (Santiago, Sao Paulo, and Mexico City); in addition, women were more likely to be illiterate and have been manual workers. More than half of SABE respondents perceived their income as insufficient, and more than three quarters of them in Havana were of this opinion. More women than men did not have a life partner and perceived their income as insufficient. In all five cities, women reported more chronic conditions than men and had higher BMI values.

Differences between women and men for each frailty component were observed, with some exceptions, that is, for nutrition (Mexico City and Sao Paulo), endurance (Santiago and Sao Paulo), and strength (Bridgetown, Mexico City, and Sao Paulo) (Table 1). For all cities, the prevalence of frailty was higher among women than men, but overall prevalence varied across cities, being the lowest (26.7%) in Bridgetown and the highest (42.6%) in Santiago de Chile (Table 2). With respect to the frailty score, frailty means were higher among women than men in every city (differences are all significant; p <.05).

Age was related to greater odds of frailty in all cities. In age-adjusted models and before adjustment for social and chronic conditions, the odds of women being frail were higher in all cities (Table 3). Poor health during childhood, perception of insufficient income, absence of a partner, presence of two or more chronic conditions, and higher BMI were also related to greater odds of frailty among respondents in the five cities studied. Furthermore, experience of hunger (except in Havana and Mexico City), less than secondary schooling (except in Bridgetown), and low-skilled occupations (except in Bridgetown and Santiago de Chile) were significantly associated with greater odds of frailty.

Multivariate models indicate that current conditions (perception of insufficient income and comorbidity) remained related to greater odds of frailty in all cities (Table 4). Interactions between life-course conditions and gender did not reach a level of significance. Interactions between health conditions and gender on frailty reached significance (p <.002) in two cities (Bridgetown and Santiago) for BMI, and in Sao Paulo for chronic conditions (Table 4). Male/female differences in the odds of frailty were reduced but remained significant after introducing life-course health and social factors and interactions between gender and health conditions (Table 4).

Separate analyses by gender were carried out to observe the effect of BMI as a categorical variable on frailty in all cities (Table 5): Obesity (BMI > 30) was related to frailty in women but not in men in all cities; additionally, in two cities (Bridgetown and Sao Paulo), undernutrition was also related to frailty in women. In the case of chronic conditions, an interaction was observed only in Sao Paulo, that is, among men, the odds of frailty related to the presence of two or more chronic conditions was higher (OR: 3.59; 95% confidence interval [CI], 2.51–5.13) than among women (OR: 1.71; 95% CI, 1.29–2.28).

Discussion

Principal Findings

We have observed that gender and early-life conditions are significantly and consistently associated with frailty after 60 years of age in all five cities. Across cities, women and those who had experienced impoverished childhoods were more likely to be frail. Lack of schooling, a manual occupation, being a housewife, and perceived economic hardship later in life were related to greater likelihood of frailty. Significant differences between women and men were observed in the relationship between chronic conditions, BMI, and frailty.

Methodological Issues

We have analyzed frailty as an ordinal scale, as proposed by Fried and colleagues. Although the construct of the variable “frailty” was inspired by the Cardiovascular Health Study (CHS) (2), some differences between the SABE and CHS frailty scores were encountered: SABE estimates weight loss during a 3-month period, whereas Fried and colleagues measured it during a 12-month period; two items from the GDS, rather than the CES-D, were used to measure low endurance and exhaustion; and we substituted lack of vigorous physical activity for the 18 items in the activity scale, and lower-extremity limitation for walking time. These differences in the definition of frailty, and possible background risk differences (cultural, and other social and biological factors) may limit the comparison between SABE prevalence estimates and those of other studies. Lack of mortality data constitutes one limitation of this study. Gender differences in frailty may be because of selective survival into old age, and poor life-course social conditions may increase mortality differently among men and women, and across cities. A better understanding of frailty dynamics will require mortality data.

Recall bias of information related to childhood has been recognized. Comparing information obtained from registers and recall measures, an underestimation of the association between childhood circumstances and health later in life has been reported (30). However, some studies indicate that recall information in older persons is a reliable measure (31,32). Methods to reconstruct the early-life experience of aged persons are necessary to estimate the effect of life-course experiences on the health of older persons more accurately (33). Our study used different indicators for exposures associated with low socioeconomic status at different points in the life course, circumventing the frequent problem of highly correlated socioeconomic measurements. This social class distinction of indicators at different points throughout the life course may allow for the identification of childhood conditions not mediated by social class, and of adult social class not mediated by current social class in old age (34).

Comparison with Other Studies

Our results support a high prevalence of frailty, overall and within each of its five components, in all LAC cities; the percentage of LAC-dwelling elders fulfilling the criteria of frailty ranged from 26.7% (Bridgetown) to 42.6% (Santiago). As stated above, it is difficult to compare SABE estimates with those of other populations (35,36), as definitions of some of the frailty components differed, and may, in our case, result in higher estimates of frailty. However, the prevalence of weakness, exhaustion, and weight loss rates were obtained using measurements similar to those of the CHS. Results suggest that the overall prevalence of frailty within SABE populations is higher than that observed in other populations. For instance, poor grip strength/weakness was present in 20.6% of respondents in the Women's Health and Aging Study, whereas in all SABE cities, prevalence was higher than 40%. Moreover, prevalence rank order (ascending) across criteria was quite similar in SABE (weight loss, low endurance, mobility limitations, weakness, and poor physical activity) and in previous studies (37) (weight loss, exhaustion, poor physical activity, weakness, and mobility limitations). However, other factors (besides differences in definition) may account for the high levels of frailty in LAC populations. High levels of obesity and diabetes were observed in LAC populations as compared to other populations (38). These chronic problems may account for the high prevalence of the weakness components of the frailty scale (5). LAC elders have been exposed to worse socioeconomic conditions across the life course than elders within North American and European populations have, resulting in a higher burden of chronic conditions and other exposures related to poor health later in life (22).

Excess physical frailty was found in this study among individuals with chronic diseases, those living under poor social conditions, and those who had enjoyed fewer opportunities during their life course. While associations between frailty and cardiovascular disease, hypertension (3), stroke, diabetes mellitus, hip fracture, chronic obstructive pulmonary disease, arthritis, cancer (4), and obesity (5) have been previously reported, longitudinal studies should provide more insights into the mechanisms relating life-course social conditions to frailty in later life (i.e., programming hypothesis or cumulative biological burden). In addition, Latin American women were frailer than men, and this finding applied to most of the dimensions of frailty measured. Social and chronic conditions did not totally explain the higher odds of frailty in women, suggesting the presence of unmeasured underlying factors. Differences in biological factors, such as muscle strength, may account for the greater rate of frailty in women. Other life-course exposures more frequent among LAC elderly women than men (39) but not studied herein may also contribute to consistently greater odds of frailty. Such exposures include low food intake and poor nutrition because of lack of social support and networks (40), low rates of physical activity related to lack of physical activity at younger ages, poorer perception of health, lack of self-sufficiency, and greater exposure to unsafe neighborhoods (15). Differences in vulnerability between women and men to high BMI and chronic conditions did not explain the gender gap in frailty (see Table 4). However, the fact that obese women were more likely to be frail supports the findings of a previous study showing that risk of mobility disability related to BMI was higher in women than in men (41), and acknowledges the importance of “fat frail” women in future diagnosis and interventions in elderly populations (5).

We have noted differences across cities. Women and men in Havana and Santiago were more likely to report being frail according to all frailty components; associations between social conditions and frailty were not found in some cities (i.e., Bridgetown), whereas in others, the effect persisted in multivariate analysis (i.e., Havana and Santiago, for education); and in some cities women and men differed with respect to health conditions related to frailty. Differences across cities are difficult to explain using available data and may be traced to history (the specific social circumstances during early periods of life), socioeconomic context (how lifetime social conditions are determined), and availability of services (services that may buffer the effects of poor social conditions and health on frailty) within the country in question (42). Further research is required to link these contextual factors to frailty.

Conclusion

We have provided evidence regarding the link between life-course social conditions and the health of elderly women and men in Latin America. This time, our results support the hypothesis that disadvantages existing in early life and reproduced along the life course may account for the physical frailty syndrome. Our results support the hypothesis that women age with higher odds of frailty than men, as has been observed with respect to other health outcomes. Theoretical models to explain gender and social differences in frailty should emphasize the use of a life-course perspective.

This study was supported by a grant from the Institute of Gender and Health, Canadian Institutes for Health Research (CIHR/IRSC).

References