Waist Circumference as a Predictor of Disability among Older Adults

Authors


Departamento de Medicina Preventiva y Salud Pública, Facultad de Medicina, Universidad Autónoma de Madrid, Avda. Arzobispo Morcillo, sn, 28029 Madrid, Spain. E-mail: fernando.artalejo@uam.es

Abstract

Objective: Few studies have addressed the association between abdominal obesity, as measured by waist circumference (WC), and disability in the elderly. Moreover, those studies were cross-sectional and yielded inconsistent results. The objective of this study was to examine longitudinally the association between WC and self-reported disability among older adults.

Research Methods and Procedures: A prospective cohort study was conducted from 2001 to 2003 in 3235 persons (1411 men and 1824 women) representative of the non-institutionalized Spanish population ages 60 years and older. Baseline information was collected by home-based personal interviews and measurement of WC, weight, and height. Two years later, information on disability was obtained by telephone interview. The association of interest was summarized with odds ratios obtained by logistic regression.

Results: Among persons reporting no disability at baseline, WC predicted disability 2 years later. After adjustment for age, education, tobacco use, alcohol consumption, and physical activity, men in the highest WC quintile had 2.17 (95% confidence interval, 1.15 to 4.09) times more risk of mobility disability and 4.77 (95% confidence interval, 2.50 to 9.13) times more risk of agility disability than those in the lowest quintile. Additional adjustment for BMI, chronic diseases, and cognitive function led to only a slight reduction in this association. Results were similar for women. No statistically significant association was observed between WC and restriction of daily activities, limitation in instrumental activities of daily living, and limitation in bathing or dressing, in either men or women.

Discussion: WC predicts mobility and agility disability in old age. Avoidance of the highest values of WC might decrease the risk of disability in older adults.

Introduction

Disability among older persons has become a major public health problem in developed countries, due to the progressive aging of the population and the ensuing rise in related morbidity (1, 2). Among the principal diseases responsible for disability are chronic disorders, such as stroke, osteomuscular diseases, coronary disease, diabetes, and cancer (3). These diseases are, in turn, related to obesity (4).

A number of studies have confirmed the relationship between disability and excess weight, as measured by BMI, among the elderly (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19). Very recently, Houston et al. (20) reported that, among middle-aged men and women followed over a period of 9 years, waist circumference (WC)1 predicted disability in instrumental and basic activities of daily living. There are very few studies, however, that have examined the relationship between WC and disability among the elderly (21, 22), and none of these has been prospective. Furthermore, such studies have yielded inconsistent results. Visser et al. (21) did not find a significant association between WC and disability in elderly participants in the Framingham study, yet Chen et al. (22) observed a higher frequency of limitations in activities of daily living with an increase in WC among elderly Hispanics in Massachusetts.

To address these limitations in the literature, and to test the hypothesis that larger WC is associated with higher risk of disability in older adults, we measured WC in a population-based cohort of subjects ages 60 years and older and followed them prospectively over 2 years to identify new cases of disability.

Research Methods and Procedures

Study Design and Subjects

The study methods have been described elsewhere (23, 24). Briefly, in 2001, information was obtained from 4008 subjects (1739 men and 2269 women), representative of the non-institutionalized Spanish population ages 60 years and older. Study subjects were selected through probabilistic multistage cluster sampling. First, clusters were stratified by region of residence and size of town. Thereafter, census tracts were selected at random in each cluster, followed by individual households, where information was then obtained from subjects. Data were collected on a total of 420 census tracts in Spain, with subjects being selected in sex and age strata. Subjects who could not be interviewed after 10 failed home visits by the interviewer or because of the subject's incapacity, death, institutionalization, or refusal to participate were replaced with other individuals selected with the same sampling procedure. The overall study response rate among originally selected subjects was 71%. Data were collected by home-based personal interview and physical examination by trained and certified personnel.

In 2003, an attempt was made to contact the subjects again; it was successful in 3235 cases (1411 men and 1824 women). Subjects contacted did not differ significantly in any sociodemographic or lifestyle characteristic from those lost to follow-up, except for the number of chronic diseases diagnosed and reported in 2001 (1.4 diseases among people followed vs. 1.2 diseases among those lost). In 2003, data were collected through a telephone interview conducted by trained staff. In Spain, there is evidence that a phone interview through computer-assisted technology is reliable and valid, in relation to a face-to-face interview of the study participant at home, to measure both lifestyle and use of health services (25, 26).

In all cases, informed consent was obtained from study participants or cohabiting next-of-kin. The study was formally approved by the Clinical Research Ethics Committee of the “La Paz” University Hospital in Madrid.

Study Variables

Self-Reported Disability

At baseline, and again at the 2-year follow-up, disability was assessed by asking participants about the difficulty or limitation experienced in performing 14 activities, classified into the following five types: mobility, agility, restriction of daily activities (RDA), instrumental activities of daily living (IADLs), and basic self-care activities.

Mobility was assessed with the following questions: 1) Do you experience any difficulty in picking up or carrying a shopping bag? 2) Do you experience any difficulty in climbing one flight of stairs? 3) Do you experience any difficulty in walking several city blocks (a few hundred meters)? Agility was ascertained with the question: Do you experience any difficulty in bending or kneeling? RDA was obtained with the question: During the past 4 weeks, did you have to refrain from doing any of your daily activities, because of your physical health? IADLs were measured using Lawton and Brody's test (27), with questions on subjects’ ability to prepare meals, do household chores, and care for clothing being excluded in men. Finally, self-care disability was assessed with the following question taken from the test of Katz et al. (28): Do you experience any difficulty in bathing or dressing yourself without assistance? The respective types of disability were deemed to be present when the answer was affirmative for any of the preceding questions, or when Lawton and Brody's test score was ≤4 among men and ≤7 among women (these cut-off points correspond to the presence of disability in at least one IADL in each sex).

Waist Circumference

At baseline, WC was measured using a plastic, inelastic, flexible belt-type measuring tape, with subjects lightly clothed. Waist circumference, which was deemed to be located at the midpoint between the lowest ribs and the iliac crest, was measured using standardized procedures (29). In each sex, subjects were classified in quintiles of WC.

For 100 study participants selected randomly at baseline in one study town, we compared the measure of WC taken by four trained interviewers with the measure obtained by two members of the research team. Each measure was taken only once, either by the interviewers or the research team. The intra-class correlation coefficient between the two measures, assuming a fixed effect model, was 0.89.

Potential Confounders

In most cases, analyses were adjusted for the following variables, ascertained during the baseline interview: age (60 to 64, 65 to 69, 70 to 74, 75 to 79, 80 to 84, and 85 years or over), educational level (subprimary, primary, secondary, and higher), tobacco use (never-smoker, ex-smoker, and smoker), alcohol consumption (abstainer, ex-drinker, drinker), and physical activity during leisure time. To assess physical activity, interviewees were asked, “Which of the following possibilities best describes most of your leisure-time activity?” For retired or unemployed subjects, questions referred to any time. Replies in the following categories were accepted: 1) sedentary: I do no exercise but remain seated for the major part of my leisure time (reading, going to the cinema, watching television, etc.); 2) occasional activity: some sporadic physical activity or sport (walking, leisurely cycling, gardening, light exercise, recreational activities requiring relatively little effort); 3) regular activity: regular physical activity several times a month (tennis, vigorous exercise, running, swimming, sprint cycling, team sports, or physical training several times a week).

At baseline, weight and height were also measured. Body weight was measured to the nearest 0.1 kg using a calibrated precision scale (Seca Model 812; Vogel & Halke, Hamburg, Germany) with subjects lightly clothed and shoeless. Height was measured to the nearest 0.1 cm using a portable wall-mounted stadiometer (KaWe, Asperg, Germany), with subjects standing in stockinged feet against a wall that had no skirting board. As with WC above, weight and height were measured using a standardized procedure (29) and validated by the researchers on a random sample of 100 individuals. The intra-class correlation coefficients were 0.97 for weight and 0.92 for height. BMI was calculated as weight in kilograms divided by the square of the height in meters (kg/m2). Study participants were classified in sex-specific quintiles of BMI.

In addition, information was collected at baseline on self-reported chronic diseases. Specifically, study participants answered the following question: Have you been told by your physician that you have any of the following diseases? The diseases considered were: chronic obstructive pulmonary disease, ischemic heart disease, cerebrovascular disease, arthritis, cataracts without treatment, diabetes mellitus, Parkinson's disease, cancer at any site, arterial hypertension, and depression. Earlier studies have shown that agreement between self-reported diseases and clinical history is high among the elderly population (30, 31). Finally, cognitive function was assessed using the Mini-Mental State Examination (MMSE), adapted and validated for use in the Spanish population, and subjects were classified into two groups according to their score (≥23 and <23 points of a maximum of 30 points on the test) (32, 33).

Statistical Analysis

Cross-Sectional Analysis

Of the 4008 subjects studied at baseline, information was lacking on WC in 300, on some variable of disability in 308, and on BMI in 12 cases. As a result, the analysis was performed on 3388 subjects (1506 men and 1882 women).

The association between WC and disability at baseline was summarized using odds ratios (ORs) obtained from logistic regression. Three models were constructed. Model 1 was adjusted for age, educational level, tobacco use, alcohol consumption, and physical activity during leisure time. Because WC and BMI are correlated (Spearman correlation coefficient: 0.74 in men and 0.71 in women), Model 2 was further adjusted for BMI, to ascertain to what extent the relationship between WC and disability was statistically independent of BMI. This analysis is important from a biological perspective, because BMI assesses mainly general obesity, whereas WC indicates mainly central obesity. Also, as compared with BMI, WC may be a better indicator of the metabolic effects of obesity. Furthermore, with aging there is a greater relative increase in abdominal fat than in total body fat, making WC a measure of obesity potentially more useful than BMI (34). Finally, Model 3 was additionally adjusted for chronic diseases, depression, and cognitive function (MMSE score), to study whether the relationship of interest might, in part, be mediated by such diseases. All variables were modeled using dummies. Interaction terms, constructed as the product of sex by WC expressed as a continuous variable, were used to test whether the association of WC with disability varied between men and women.

Longitudinal Analysis

To investigate the association between WC and risk of disability 2 years later in life, information on the 3235 subjects with follow-up data were used. Of these, the following subjects were excluded: 245 subjects who had died; 207 with missing information on disability at baseline; and 219 for whom we lacked data on WC or BMI in 2001. Furthermore, for analyzing the respective disabilities, subjects reporting a specific disability in 2001 but with missing data on it in 2003 were eliminated. The strategy used in the statistical analysis was the same as that used in the cross-sectional analysis. Moreover, the analyses were repeated with BMI as the independent variable of interest, to compare the relationship between disability and BMI against that displayed by WC.

Statistical significance was set at p < 0.05 (two-tailed). Analyses were performed separately for men and women, using SAS software, version 8.2 (35).

Results

Table 1 shows the total number of subjects included in the analyses, as well as the prevalence of each type of disability in 2001 and the incidence of each type of disability in 2003. A total of 63.0% of men and 80.5% of women reported one or more types of disability in 2001. Functions most frequently affected by disability were, in descending order, agility, mobility, IADL, RDA, and bathing or dressing. The prevalence and 2-year incidence of each type of disability were higher in women than in men.

Table 1.  Prevalence of disability at baseline (2001) and 2-year incidence of disability (2003), according to sex
 MenWomen
DisabilityNn%Nn%
  • *

    Denominator (at-risk subjects) for calculating incidence of disability: subjects free of each type of disability at baseline, with complete information on disability and other study variables at 2003.

Prevalence at baseline (2001)      
 Mobility150657838.41882109058.0
 Agility150670346.71882126767.3
 RDA150626917.9188252828.1
 IADL150642328.1188278741.8
 Bathing or dressing150618212.1188237820.1
 At least one disability150694862.91882151480.4
2-year incidence (2003)*      
 Mobility58814224.154154143.3
 Agility51217935.042521350.1
 RDA76212616.688220923.7
 IADL82619824.085727532.1
 Bathing or dressing827617.4100117711.6
 At least one disability37116344.026715056.4

Table 2 presents the main characteristics of the study participants according to disability at baseline. In each sex, reporting at least one disability was associated with older age, lower education, lower frequency of current smoking and drinking, higher frequency of sedentariness, and higher frequency of each of the chronic health disorders ascertained, of depression, and of MMSE score <23. Moreover, the proportion of subjects in the higher quintiles of WC and BMI was greater among those reporting at least one disability than in subjects with no disability, in men and women.

Table 2.  Characteristics of study subjects at baseline (2001), according to disability and sex
 MenWomen
 No disability (N = 557)At least one disability (N = 948)No disability (N = 367)At least one disability (N = 1514)
  • SD, standard deviation.

  • *

    Cut-off points for WC (cm): Men: Quintile 1, <93.8; Quintile 2, 93.8 to 99.6; Quintile 3, 99.7 to 104.5; Quintile 4, 104.6 to 111.0; Quintile 5, >111.0. Women: Quintile 1, <87.3; Quintile 2, 87.3 to 94.2; Quintile 3, 94.3 to 100.0; Quintile 4, 100.1 to 107; Quintile 5, >107.0.

  • Cut-off points for BMI (kg/m2): Men: Quintile 1, <25.1; Quintile 2, 25.1 to 27.0; Quintile 3, 27.1 to 29.0; Quintile 4, 29.1 to 31.2; Quintile 5, >31.2. Women: Quintile 1, <25.2; Quintile 2, 25.2 to 27.9; Quintile 3, 28.0 to 30.3; Quintile 4, 30.3 to 33.5; Quintile 5, >33.5.

Age (years, mean ± SD)68.7 ± 6.571.8 ± 8.268.3 ± 6.072.9 ± 7.6
 60 to 64 years (%)28.118.835.218.5
 65 to 69 years (%)35.223.126.819.2
 70 to 74 years (%)18.624.420.321.5
 75 to 79 years (%)11.516.510.717.1
 80 to 84 years (%)5.110.05.313.6
 ≥85 years (%)1.26.91.49.8
Educational level    
 Subprimary (%)33.151.736.361.8
 Primary (%)42.833.049.330.9
 Secondary (%)15.211.18.85.43
 Higher (%)8.84.05.41.8
Tobacco use    
 Never-smoker (%)29.528.692.594.6
 Ex-smoker (%)48.651.34.23.5
 Current smoker (%)21.819.93.21.8
Alcohol    
 Abstainer (%)19.527.057.173.0
 Drinker (%)67.450.237.520.9
 Ex-drinker (%)12.922.65.26.0
Physical activity during leisure time    
 Sedentary (%)22.241.330.053.8
 Occasional activity (%)70.156.064.244.6
 Regular activity (%)7.52.65.61.5
Chronic diseases    
 Chronic obstructive pulmonary disease (%)9.721.95.612.6
 Ischemic heart disease (%)4.912.31.45.7
 Cerebrovascular disease (%)1.63.52.14.2
 Arthrosis (%)28.051.645.571.7
 Cataracts without treatment (%)12.022.118.025.5
 Diabetes mellitus (%)11.916.311.218.0
 Parkinson's disease (%)0.41.70.51.5
 Cancer at any site (%)0.82.10.72.4
 Arterial hypertension (%)64.466.660.872.5
Mental health disorders    
 Depression (%)3.05.438.016.1
 30-item MMSE score <23 (%)5.518.48.925.7
WC*    
 Quintile 1 (%)21.717.830.617.7
 Quintile 2 (%)20.320.429.1318.0
 Quintile 3 (%)22.318.116.321.2
 Quintile 4 (%)19.521.313.820.8
 Quintile 5 (%)15.922.29.922.1
BMI (kg/m2)    
 Quintile 1 (%)19.320.123.119.2
 Quintile 2 (%)20.719.625.918.6
 Quintile 3 (%)21.419.223.019.3
 Quintile 4 (%)19.119.617.820.6
 Quintile 5 (%)18.921.410.322.3

Cross-sectional Analysis

Table 3 shows the OR of disability at baseline by quintiles of WC, adjusted for age, educational level, tobacco use, alcohol consumption, and physical activity. In men, the highest quintile of WC registered a significantly greater frequency of limitation in mobility and agility, and of RDA, than did the lowest quintile. The relationship was not linear. Whereas higher frequencies of mobility and RDA disability were observed solely in the highest WC quintile, agility disability was significantly higher also in the second WC quintile. In men, no statistically significant association was observed between WC and bathing or dressing disability. In women, the results displayed the same direction as in men, but the magnitude of the association between WC and disability was greater (interaction p < 0.05 for all types of disability except RDA). Furthermore, compared with women in the lowest WC quintile, those in the highest quintile were significantly more likely to be disabled in IADL and in bathing or dressing (Table 3). However, after further adjustment for BMI (Model 2), the association of WC with impairment in IADL and bathing or dressing among women lost statistical significance. After additional adjustment for chronic diseases and cognitive function (Model 3), only the association of WC with agility disability and RDA among women remained statistically significant (data not shown).

Table 3.  Adjusted ORs of disability at baseline (2001), according to quintiles of WC, by sex
 WC quintiles (cm)
 Q1 (low) ORQ2 OR (95% CI)Q3 OR (95% CI)Q4 OR (95% CI)Q5 (high) OR (95% CI)
  • ORs adjusted for age, tobacco use, alcohol consumption, educational level, and physical activity during leisure time. Q, quintile; CI, confidence interval; Ref, reference.

  • *

    p < 0.01.

  • p < 0.001.

  • p < 0.05.

Men     
 Mobility1 (Ref)1.17 (0.82 to 1.68)1.07 (0.74 to 1.55)1.07 (0.74 to 1.54)1.65 (1.15 to 2.38)*
 Agility1 (Ref)1.66 (1.18 to 2.35)*1.23 (0.86 to 1.74)1.36 (0.96 to 1.92)1.99 (1.40 to 2.83)
 RDA1 (Ref)1.27 (0.82 to 1.99)0.91 (0.56 to 1.47)1.51 (0.98 to 2.34)1.89 (1.22 to 2.94)*
 IADL1 (Ref)0.72 (0.49 to 1.05)0.43 (0.28 to 0.64)0.61 (0.42 to 0.90)0.73 (0.50 to 1.08)
 Bathing or dressing1 (Ref)0.60 (0.35 to 1.02)1.00 (0.59 to 1.69)0.75 (0.45 to 1.26)0.92 (0.54 to 1.55)
Women     
 Mobility1 (Ref)1.07 (0.79 to 1.46)1.40 (1.03 to 1.90)1.42 (1.04 to 1.94)2.30 (1.66 to 3.18)
 Agility1 (Ref)1.36 (1.00 to 1.85)2.06 (1.49 to 2.83)2.14 (1.55 to 2.97)3.18 (2.25 to 4.50)
 RDA1 (Ref)0.97 (0.69 to 1.37)1.21 (0.87 to 1.68)1.17 (0.83 to 1.64)1.90 (1.37 to 2.64)
 IADL1 (Ref)0.85 (0.60 to 1.19)1.06 (0.76 to 1.47)1.00 (0.72 to 1.40)1.62 (1.16 to 2.27)*
 Bathing or dressing1 (Ref)0.93 (0.61 to 1.41)1.03 (0.70 to 1.54)1.37 (0.93 to 2.03)1.61 (1.09 to 2.36)

Longitudinal Analysis

Tables 4 and 5 show, for men and women, respectively, the adjusted OR of disability at 2 years of follow-up for quintiles of WC at baseline among persons reporting no disability at baseline. In men, Model 1 shows that, compared with subjects in the lowest WC quintile, those in the highest quintile had 2.17 (95% confidence interval, 1.15 to 4.09) times more risk of mobility disability and 4.77 (95% confidence interval, 2.50 to 9.13) times more risk of agility disability (Table 4). These results were not substantially changed in Models 2 and 3, although, in the latter model, statistical significance was lost for mobility disability (Table 4). Results were similar for women (in Model 1, p values for interaction terms with sex were not significant in any of the five types of disability), but in Model 2, the OR of mobility disability lost statistical significance (Table 5). It should be noted that we observed no clear association between functional impairment and the intermediate quintiles of WC, except for agility disability in women, which in Model 1 attained statistical significance in all quintiles (Table 5).

Table 4.  Adjusted ORs of disability at 2 years of follow-up (2003), according to quintiles of WC at baseline (2001), in men reporting no disability at baseline (2001)
 Q1 (low) ORQ2 OR (95% CI)Q3 OR (95% CI)Q4 OR (95% CI)Q5 (high) OR (95% CI)
  • Model 1: ORs adjusted for age, tobacco use, alcohol consumption, educational level, and physical activity during leisure time. Model 2: ORs adjusted for Model 1 variables plus BMI. Model 3: ORs adjusted for Model 2 variables plus chronic diseases, depression, and cognitive function. Q, quintile; CI, confidence interval; Ref, reference.

  • *

    p < 0.05.

  • p < 0.001.

  • p < 0.01.

Mobility     
 Model 11 (Ref)0.86 (0.42 to 1.76)1.24 (0.63 to 2.43)1.15 (0.60 to 2.19)2.17 (1.15 to 4.09)*
 Model 21 (Ref)0.86 (0.41 to 1.82)1.18 (0.55 to 2.53)1.02 (0.47 to 2.21)2.07 (1.10 to 3.91)*
 Model 31 (Ref)0.78 (0.36 to 1.71)1.17 (0.53 to 2.59)1.04 (0.46 to 2.35)1.60 (0.64 to 3.99)
Agility     
 Model 11 (Ref)1.03 (0.53 to 2.03)0.92 (0.47 to 1.83)1.62 (0.86 to 3.04)4.77 (2.50 to 9.13)
 Model 21 (Ref)0.95 (0.47 to 1.92)0.82 (0.38 to 1.79)1.45 (0.66 to 3.21)4.20 (1.64 to 10.72)
 Model 31 (Ref)0.96 (0.46 to 2.01)0.83 (0.36 to 1.89)1.73 (0.75 to 3.93)4.84 (1.81 to 12.93)
RDA     
 Model 11 (Ref)0.60 (0.31 to 1.16)0.48 (0.25 to 0.94)*0.93 (0.51 to 1.69)0.92 (0.50 to 1.70)
 Model 21 (Ref)0.59 (0.30 to 1.16)0.42 (0.20 to 0.92)*0.77 (0.36 to 1.62)0.68 (0.28 to 1.65)
 Model 31 (Ref)0.55 (0.27 to 1.10)0.39 (0.17 to 0.86)*0.67 (0.31 to 1.48)0.55 (0.22 to 1.38)
IADL     
 Model 11 (Ref)0.89 (0.51 to 1.56)0.90 (0.52 to 1.57)1.28 (0.75 to 2.18)1.00 (0.58 to 1.73)
 Model 21 (Ref)0.92 (0.51 to 1.65)0.91 (0.49 to 1.69)1.24 (0.65 to 2.35)0.76 (0.36 to 1.62)
 Model 31 (Ref)0.86 (0.47 to 1.57)0.85 (0.45 to 1.64)1.23 (0.63 to 2.42)0.65 (0.30 to 1.42)
Bathing or dressing     
 Model 11 (Ref)0.95 (0.38 to 2.37)0.96 (0.39 to 2.35)0.98 (0.41 to 2.32)1.63 (0.71 to 3.74)
 Model 21 (Ref)0.92 (0.36 to 2.39)1.10 (0.40 to 3.01)1.26 (0.45 to 3.52)2.19 (0.66 to 7.27)
 Model 31 (Ref)0.96 (0.35 to 2.63)1.15 (0.39 to 3.39)1.28 (0.41 to 4.00)2.09 (0.58 to 7.50)
Table 5.  Adjusted ORs of disability at 2 years of follow-up (2003), according to quintiles of WC at baseline (2001), in women reporting no disability at baseline (2001)
 Q1 (low) ORQ2 OR (95% CI)Q3 OR (95% CI)Q4 OR (95% CI)Q5 (high) OR (95% CI)
  • Model 1: ORs adjusted for age, tobacco use, alcohol consumption, educational level, and physical activity during leisure time. Model 2: ORs adjusted for Model 1 variables plus BMI. Model 3: ORs adjusted for Model 2 variables plus chronic diseases, depression, and cognitive function. Q, quintile; CI, confidence interval; Ref, reference.

  • *

    p < 0.05.

  • p < 0.001.

  • p < 0.01.

Mobility     
 Model 11 (Ref)1.11 (0.63 to 1.95)1.18 (0.67 to 2.08)1.53 (0.87 to 2.69)1.81 (1.02 to 3.20)*
 Model 21 (Ref)1.20 (0.66 to 2.18)1.30 (0.70 to 2.44)1.53 (0.80 to 2.92)1.67 (0.81 to 3.41)
 Model 31 (Ref)1.20 (0.65 to 2.18)1.25 (0.66 to 2.37)1.50 (0.78 to 2.90)1.59 (0.77 to 2.90)
Agility     
 Model 11 (Ref)1.99 (1.02 to 3.90)*2.06 (1.06 to 4.00)*2.25 (1.15 to 4.39)*5.21 (2.58 to 10.50)
 Model 21 (Ref)1.96 (0.96 to 3.99)1.87 (0.90 to 3.89)1.79 (0.84 to 3.82)3.30 (1.42 to 7.70)
 Model 31 (Ref)2.21 (1.05 to 4.66)*2.05 (0.95 to 4.42)1.97 (0.89 to 4.36)3.57 (1.48 to 8.59)
RDA     
 Model 11 (Ref)0.74 (0.43 to 1.25)0.88 (0.52 to 1.48)0.83 (0.50 to 1.38)1.36 (0.81 to 2.26)
 Model 21 (Ref)0.74 (0.42 to 1.29)0.85 (0.47 to 1.53)0.79 (0.43 to 1.46)1.48 (0.75 to 2.94)
 Model 31 (Ref)0.74 (0.42 to 1.31)0.87 (0.48 to 1.59)0.74 (0.39 to 1.38)1.44 (0.71 to 2.93)
IADL     
 Model 11 (Ref)1.18 (0.70 to 1.97)1.06 (0.63 to 1.77)1.91 (1.16 to 3.13)*1.36 (0.82 to 2.25)
 Model 21 (Ref)1.14 (0.66 to 1.97)0.97 (0.54 to 1.72)1.69 (0.95 to 3.03)1.25 (0.66 to 2.36)
 Model 31 (Ref)1.13 (0.65 to 1.95)0.94 (0.53 to 1.70)1.65 (0.92 to 2.98)1.16 (0.60 to 2.21)
Bathing or dressing     
 Model 11 (Ref)1.08 (0.54 to 2.17)1.66 (0.86 to 3.20)1.12 (0.57 to 2.21)1.48 (0.76 to 2.90)
 Model 21 (Ref)1.18 (0.57 to 2.49)1.76 (0.83 to 3.74)1.02 (0.45 to 2.29)1.11 (0.47 to 2.66)
 Model 31 (Ref)1.29 (0.61 to 2.74)1.89 (0.87 to 4.10)1.00 (0.43 to 2.33)1.05 (0.42 to 2.61)

Tables 6 and 7 show the results of the longitudinal analysis with BMI as the independent variable of interest. As compared with individuals in the lowest quintile of BMI, those in the highest quintile had a higher 2-year risk of agility disability in each sex (Model 1). After adjustment for WC, the OR for the association of BMI with agility disability was substantially reduced and lost statistical significance among men (Model 2). Among women, however, the increased risk of agility disability for those in the highest quintile of BMI persisted after adjustment for WC, chronic diseases, depression, and cognitive function (Model 3). Lastly, no clear association was observed between BMI and the other types of disability studied.

Table 6.  Adjusted ORs of disability at 2 years of follow-up (2003), according to categories of BMI at baseline (2001) in men reporting no disability at baseline (2001)
 Q1 (low) ORQ2 OR (95% CI)Q3 OR (95% CI)Q4 OR (95% CI)Q5 (high) OR (95% CI)
  • Model 1: ORs adjusted for age, tobacco use, alcohol consumption, educational level, and physical activity during leisure time. Model 2: ORs adjusted for Model 1 variables plus WC. Model 3: ORs adjusted for Model 2 variables plus chronic diseases, depression, and cognitive function. Q, quintile; CI, confidence interval; Ref, reference.

  • *

    p < 0.001.

Mobility     
 Model 11 (Ref)0.95 (0.47 to 1.90)1.03 (0.52 to 2.05)1.53 (0.81 to 2.89)1.82 (0.95 to 3.47)
 Model 21 (Ref)0.92 (0.44 to 1.92)0.95 (0.44 to 2.07)1.29 (0.60 to 2.79)1.14 (0.47 to 2.78)
 Model 31 (Ref)0.84 (0.39 to 1.81)0.93 (0.41 to 2.08)1.17 (0.52 to 2.64)1.15 (0.45 to 2.91)
Agility     
 Model 11 (Ref)1.39 (0.71 to 2.70)1.37 (0.71 to 2.66)1.92 (1.00 to 3.69)3.88 (2.03 to 7.44)*
 Model 21 (Ref)1.34 (0.66 to 2.74)1.30 (0.61 to 2.81)1.22 (0.53 to 2.78)1.29 (0.50 to 3.32)
 Model 31 (Ref)1.27 (0.61 to 2.66)1.24 (0.56 to 2.77)1.13 (0.48 to 2.66)1.19 (0.44 to 3.20)
RDA     
 Model 11 (Ref)0.74 (0.38 to 1.43)0.85 (0.45 to 1.62)1.05 (0.57 to 1.96)1.18 (0.63 to 2.18)
 Model 21 (Ref)0.91 (0.45 to 1.83)1.18 (0.56 to 2.49)1.31 (0.59 to 2.89)1.41 (0.58 to 3.46)
 Model 31 (Ref)0.92 (0.44 to 1.92)1.18 (0.54 to 2.61)1.35 (0.58 to 3.12)1.42 (0.55 to 3.68)
IADL     
 Model 11 (Ref)0.83 (0.48 to 1.42)0.89 (0.51 to 1.54)0.99 (0.58 to 1.69)1.22 (0.72 to 2.05)
 Model 21 (Ref)0.81 (0.46 to 1.45)0.87 (0.47 to 1.61)0.96 (0.50 to 1.85)1.38 (0.67 to 2.88)
 Model 31 (Ref)0.85 (0.47 to 1.55)0.90 (0.48 to 1.72)0.94 (0.47 to 1.87)1.40 (0.65 to 3.03)
Bathing or dressing     
 Model 11 (Ref)1.74 (0.75 to 4.05)1.03 (0.41 to 2.57)0.85 (0.33 to 2.20)1.66 (0.71 to 3.90)
 Model 21 (Ref)1.65 (0.68 to 4.02)0.88 (0.31 to 2.52)0.62 (0.20 to 1.96)0.90 (0.26 to 3.05)
 Model 31 (Ref)1.83 (0.69 to 4.85)1.14 (0.37 to 3.55)0.66 (0.19 to 2.33)0.91 (0.24 to 3.42)
Table 7.  Adjusted ORs of disability at 2 years of follow-up (2003), according to categories of BMI at baseline (2001) in women reporting no disability at baseline (2001)
 Q1 (low) ORQ2 OR (95% CI)Q3 OR (95% CI)Q4 OR (95% CI)Q5 (high) OR (95% CI)
  • Model 1: ORs adjusted for age, tobacco use, alcohol consumption, educational level, and physical activity during leisure time. Model 2: ORs adjusted for Model 1 variables plus WC. Model 3: ORs adjusted for Model 2 variables plus chronic diseases, depression, and cognitive function. Q, quintile; CI, confidence interval; Ref, reference.

  • *

    p < 0.05.

  • p < 0.001.

Mobility     
 Model 11 (Ref)0.60 (0.34 to 1.05)0.76 (0.44 to 1.32)0.85 (0.49 to 1.48)1.31 (0.75 to 2.28)
 Model 21 (Ref)0.58 (0.32 to 1.03)0.66 (0.37 to 1.20)0.71 (0.38 to 1.31)0.97 (0.49 to 1.92)
 Model 31 (Ref)0.58 (0.32 to 1.04)0.64 (0.35 to 1.17)0.68 (0.36 to 1.29)0.96 (0.48 to 1.92)
Agility     
 Model 11 (Ref)1.54 (0.81 to 2.93)0.96 (0.49 to 1.87)2.00 (1.05 to 3.81)*4.07 (2.07 to 7.99)
 Model 21 (Ref)1.27 (0.64 to 2.49)0.71 (0.34 to 1.46)1.40 (0.68 to 2.89)2.24 (0.99 to 5.06)
 Model 31 (Ref)1.23 (0.61 to 2.50)0.68 (0.32 to 1.46)1.28 (0.60 to 2.71)2.49 (1.07 to 5.80)*
RDA     
 Model 11 (Ref)0.76 (0.45 to 1.30)0.85 (0.51 to 1.44)1.26 (0.77 to 2.06)0.97 (0.59 to 1.61)
 Model 21 (Ref)0.79 (0.46 to 1.38)0.92 (0.52 to 1.64)1.27 (0.71 to 2.28)0.75 (0.39 to 1.47)
 Model 31 (Ref)0.75 (0.43 to 1.33)0.82 (0.45 to 1.48)1.22 (0.67 to 2.24)0.70 (0.35 to 1.39)
IADL     
 Model 11 (Ref)0.89 (0.53 to 1.47)1.18 (0.72 to 1.94)1.51 (0.93 to 2.46)1.24 (0.76 to 2.02)
 Model 21 (Ref)0.86 (0.50 to 1.46)1.08 (0.63 to 1.87)1.28 (0.72 to 2.26)1.02 (0.55 to 1.91)
 Model 31 (Ref)0.84 (0.49 to 1.43)1.03 (0.59 to 1.80)1.22 (0.68 to 2.17)0.99 (0.53 to 1.86)
Bathing or dressing     
 Model 11 (Ref)0.58 (0.29 to 1.16)0.86 (0.45 to 1.64)0.96 (0.52 to 1.77)1.32 (0.72 to 2.42)
 Model 21 (Ref)0.52 (0.25 to 1.07)0.74 (0.36 to 1.50)0.87 (0.42 to 1.79)1.30 (0.59 to 2.88)
 Model 31 (Ref)0.49 (0.23 to 1.03)0.66 (0.32 to 1.37)0.80 (0.38 to 1.70)1.29 (0.56 to 2.96)

Discussion

Our results show that WC predicts mobility and agility disability in older adults of both sexes. Nevertheless, no clear relationship between WC and risk of RDA, IADL, or bathing or dressing disabilities was in evidence.

Among the possible mechanisms to explain our findings, three were most important a priori. First, abdominal obesity is associated with physical inactivity and could, thus, lead to disability. It is unlikely, however, that it would be the principal mechanism, because our results were observed after adjustment for physical activity during leisure time. Second, abdominal obesity is associated with diseases that cause disability. Yet, our results were, for the most part, in evidence even after adjusting for numerous serious chronic diseases. Accordingly, one plausible mechanism is that obesity per se represents a “burden” for the osteomuscular system, which limits agility and mobility.

The results of the cross-sectional analysis point to a higher frequency of limitation in IADL and bathing or dressing among women in the highest WC quintile, although this association was not independent of BMI and serious chronic diseases. It is likely that the most severe disabilities, such as in IADL and bathing or dressing, are frequently mediated by obesity-related diseases. Because the longitudinal analysis was performed with subjects reporting no disability at baseline, such persons, compared with those in the cross-sectional analysis, were thinner (mean WC, 98.2 cm vs. 99.6 cm) and younger (mean age, 70.0 years vs. 71.4 years), and they had less comorbidity (mean number of diseases, 1.4 vs. 1.9). This may have hindered observation of the association of WC with IADL and bathing or dressing disabilities in the longitudinal analysis.

As for the results of earlier cross-sectional studies, Visser et al. (21) failed to observe any association between either WC or waist-to-hip ratio and general or mobility disability in subjects ages 72 to 95 years from the Framingham Study. Visser et al. did not report their results in detail, thereby ruling out accurate comparisons with ours. It is noteworthy, however, that they did observe a clear association between disability and body fatness, because a substantial part of such fat accumulates in the abdomen (36). Furthermore, our results are consistent with those of Chen et al. (22), who observed a cross-sectional association between WC and disability in Hispanic men and women with a mean age of 69 years. This association was independent of BMI and chronic diseases classified into only three groups: arthritis, hypertension, and others. These authors measured disability using a 12-item scale in which agility and mobility disability predominated (22). Finally, our results are also in line with those of a cross-sectional study in the Netherlands, in which larger WC was associated with functional limitations in a relatively young population, ages 20 to 59 years (37, 38).

The results for the longitudinal association between BMI and disability were quite similar to those obtained for WC. When WC and BMI were included in the same model, WC maintained its association with agility disability in the two sexes and with mobility disability in men. Yet, after adjustment for WC, BMI showed an association only with agility disability in women. Thus, our results suggest that abdominal obesity is at least as important as general obesity for producing disability in our population. It is possible that WC is a better indicator of adiposity than BMI among the elderly. In fact, with increasing age, muscle mass is progressively replaced by fat mass, with no necessary gain in weight or BMI. Also, the relative contribution of WC to total fatness is likely to increase with age because fat tends to accumulate in the abdomen in the elderly (34).

To interpret our results correctly, some methodologic remarks are called for. First, our study rules out the possibility that the association between abdominal obesity and disability was attributable to reverse causation—less agility and mobility leading to greater obesity—because it was a longitudinal study that exclusively analyzed persons reporting no disability at baseline. Second, it allows for wide-ranging generalization of the results, because it studied a representative sample of the non-institutionalized older population of both sexes, nationwide. Third, the study was conducted with validated measurements of WC, weight, and height, and analyses were adjusted for a good number of confounders, a number that is similar to or higher than those used in previous studies (21, 22). Yet, we cannot exclude the possibility that unmeasured confounders could explain some of these results. For example, the relationship between WC and disability could be mediated, at least in part, by diseases exacerbated during the 2 years of follow-up.

Finally, our questionnaire has shown validity to assess disability in earlier studies. Mobility disability was assessed with three questions. Two of them (difficulty in climbing one flight of stairs; and difficulty in walking several city blocks) are used very frequently in the literature (12, 39, 40), and there is evidence that difficulty in performing these two specific tasks predicts mortality and nursing home admission in older men and women (41). The third question related to the difficulty in picking up or carrying a shopping bag. Although this task requires certain arm strength, usually it is only a minimum because older adults adapt the contents of the shopping bag to the weight they are able to carry comfortably. Thus, this question assesses mainly lower-extremity function. Moreover, other authors have also used this task to ascertain mobility disability (12). As for the question assessing agility disability (difficulty in bending or kneeling), it is similar to one used in the Short Form-36 to ascertain body function, one of the eight dimensions of the quality of life measured by that questionnaire. The validity and reproducibility of each of the eight dimensions of the SF-36 has been demonstrated previously (42), and this question has also been used by other authors to assess agility (40). As regards RDA, it is a global indicator of health status frequently used in this type of research (43, 44). In addition, there is evidence that RDA is associated with higher use of health services and with disability (43, 44). Also as in previous research (43, 44), we used no information on the other components of the Katz index (e.g., limitations in eating and in using the toilet), because the incidence of disability in such tasks is very low in community-dwelling older persons. Furthermore, it is highly uncommon for disability to develop in these activities (eating, using the toilet) without concurrent disability in bathing or dressing. Lastly, the Lawton and Brody test (27) is the standard scale used to assess limitations in instrumental activities of daily living (45). Other authors have also used the same cut-off point as used in our study (46, 47, 48).

It should be noted that in our study there was a binary response to most disability questions, with the category of “any difficulty” falling into the disability group. Thus, the disability group includes subjects with very mild disability or functional limitations. Although it may limit accuracy in the definition of the outcomes, it increases sensitivity to detect them and provides power to show an association between WC and disability over a follow-up period of only 2 years. Also, our definition of disability has resulted in higher disability prevalence than have earlier studies (21, 22).

Our study has practical implications. Our results suggest that avoidance of the highest values of WC might decrease the risk of agility and mobility disability in older adults. This type of disability is relevant, even though it may seem relatively mild, as it predicts mortality and nursing home admission (41). Because of these factors, and the fact that the association between WC and disability was independent of BMI, our results also provide a new reason for measuring WC, in addition to BMI, in older adults as part of routine clinical practice.

Acknowledgments

This work was supported by Grant 02/563 from Spain's Health Research Fund (Fondo de Investigación Sanitaria), an unrestricted contract with AstraZeneca in Spain, and ISCIII (RCESP red C03/09). N.S.F. was supported by a grant from the Coordenaçao de Aperfeiçoamento de Pessoal de Nível Superior (CAPES BEX 1696/04). E.L.G. was supported by a contract “Juan de la Cierva” from the Ministerio de Educación y Ciencia. The funding bodies had no role in data extraction and analysis, writing of the manuscript, or the decision to submit the paper for publication. There are no conflicts of interest.

Footnotes

  • 1

    Nonstandard abbreviations: WC, waist circumference; RDA, restriction of daily activities; IADL, instrumental activity of daily living; MMSE, Mini-Mental State Examination; OR, odds ratio.

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Ancillary