Comorbidities and Impairments Explaining the Association Between Diabetes and Lower Extremity Disability

The Women’s Health and Aging Study

  1. Stefano Volpato, MD, MPH12,
  2. Caroline Blaum, MD3,
  3. Helaine Resnick, PHD4,
  4. Luigi Ferrucci, MD, PHD15,
  5. Linda P. Fried, MD, MPH3 and
  6. Jack M. Guralnik, MD, PHD1
  1. 1Laboratory of Epidemiology, Demography and Biometry, National Institute on Aging, Bethesda, Maryland
  2. 2Department of Clinical and Experimental Medicine, University of Ferrara, Ferrara, Italy
  3. 3Departments of Medicine and Epidemiology, the Johns Hopkins Medical Institutions, Baltimore, Maryland
  4. 4MedStar Research Institute, Washington, DC
  5. 5Geriatric Department, “I Fraticini,” Italian National Institute for Research and Care on Aging (INRCA), Florence, Italy


    OBJECTIVE—To elucidate the role of diabetes-related impairments and comorbidities in the association between diabetes and physical disability, this study examined the association between diabetes and lower extremity function in a sample of disabled older women.

    RESEARCH DESIGN AND METHODS—Cross-sectional analysis of 1,002 women (aged ≥65 years) enrolled in the Women’s Health and Aging Study (one-third most disabled of the total community-dwelling population). Diabetes and other medical conditions were ascertained by standard criteria that used multiple sources of information. Functional status was assessed using self-reported and objective performance measures.

    RESULTS—Women with diabetes were significantly more likely to have cardiovascular diseases, peripheral nerve dysfunction, visual impairment, obesity, and depression. After adjustment for age, women with diabetes had a greater prevalence of mobility disability (odds ratio [OR] 1.85, 95% CI 1.12–3.06), activities of daily living disability (1.61, 1.06–2.43), and severe walking limitation (2.34, 1.56–3.50), and their summary mobility performance score (0–12 scale based on balance, gait speed, chair stands) was 1.4 points lower than in nondiabetic women (P < 0.001). Peripheral artery disease, peripheral nerve dysfunction, and depression were the main individual contributing factors; however, none of these conditions alone fully explained the association between diabetes and disability. Conversely, only after adjusting for all potential mediators was the relationship between diabetes and disability reduced to a large degree.

    CONCLUSIONS—Even among physically impaired older women, diabetes is associated with a major burden of disability. A wide range of impairments and comorbidities explains the diabetes-disability relationship, suggesting that the mechanism for such an association is multifactorial.

    The ability to remain mobile is an essential aspect of quality of life and is critical for the preservation of independence in old age. Objective measures of lower extremity function are strong predictors of poor health outcomes, including disability, hospitalization, and death (1,2). In recent years, the search for potentially modifiable conditions associated with impaired mobility and lower extremity function identified several sociodemographic and behavioral characteristics, along with acute and chronic medical conditions (3,4). Among the latter, diabetes has been consistently reported as one of the strongest correlates of the presence of poor lower extremity performance (5,6).

    Several diseases and impairments, including cardiovascular diseases, peripheral neuropathy, obesity, and visual deficits, are more prevalent in diabetic patients and may easily explain the greater burden of physical disability associated with diabetes. Nevertheless, the specific mechanisms underlying the disablement process in older diabetic patients are less clear. Indeed, only a few studies have investigated the role of specific diabetes-related conditions in disability (6), and it is not known whether any particular complications of the disease play a predominant role or whether the disablement process is rather a multifactorial phenomenon, sustained by the global burden of comorbidities. Diabetes is highly prevalent in older people, and its prevalence is expected to increase substantially in the next decades (7). Therefore, elucidating the pathway from diabetes to disability is important for planning strategies aimed at preventing or slowing functional decline in older persons and for tertiary prevention in subjects with diabetes.

    This study examined the association between diabetes and both self-report and objective measures of lower extremity function in a sample of older physically impaired women living in the community. We hypothesized that the pathway between diabetes and physical impairments could be explained by the negative impact on physical function of medical conditions that are common complications of diabetes.


    The Women’s Health and Aging Study is an epidemiological study of the causes and course of disability among the one-third most disabled women aged ≥65 years living in the community (8). Briefly, among 5,316 community-dwelling women randomly sampled from Medicare beneficiaries in Baltimore, MD, 1,409 individuals were eligible for the study because of a Mini Mental State Exam score >18 and difficulty in performing one or more tasks in at least two of the following four domains of functioning: mobility/exercise tolerance, upper extremity abilities, basic self-care, and higher functioning tasks of independent living. Overall, 1,002 women (71% of those eligible) agreed to participate in the study. The study was approved by the Johns Hopkins University Institutional Review Board, and all participants gave informed consent.

    Measures of physical function

    Self-reported information included difficulty with two mobility-related tasks (walking 0.25 mile and climbing stairs) and five basic activities of daily living (ADL) bathing, transferring from bed to chair, using the toilet, dressing, and (eating). Validity and reproducibility of these measures have been previously demonstrated (9). Responses were coded as: none, a little, some, a lot, or unable to perform the task. Using this information, two different outcomes were defined: mobility disability (a lot of difficulty or inability to walk 0.25 mile and/or to climb stairs) and ADL disability (a lot of difficulty or inability in at least one ADL).

    Performance-based measures of physical function included usual 4-m walking speed, five chair-stands test, and balance tests. The three tests were administered after a standardized protocol as previously described (10). Using the results of the three performance tests, a summary performance score was calculated. The score, measured on a 0–12 ordinal scale, represents the sum of the scores of the three individual tests (0–4) and predicts disability, hospitalization, and mortality in older individuals (1,2). Severe walking limitation was defined as being unable to walk or walking at a speed of ≤0.4 m/ s.

    Assessment of diabetes

    Women with a history of physician-diagnosed diabetes and those who were taking oral hypoglycemic drugs or insulin were defined as diabetic. For those not taking diabetes medications or without history of diabetes, disease status was further evaluated by checking the value of total HbA1c, measured using low-pressure cation exchange chromatography (reference range 5–8.6%) (765 Glycomat; Ciba Corning, Palo Alto, CA). This test was available for 65% of the subjects; a value >10% was considered diagnostic for diabetes. Because there is not consensus on the use of HbA1c for the diagnosis of diabetes, this cutoff was defined a priori to achieve good diagnostic specificity. For women with missing HbA1c or HbA1c ≤10%, absence of disease was confirmed using the questionnaire sent to the participant’s primary care physician or by direct contact with the physician.

    Comorbidities and impairments

    Peripheral artery disease (PAD) was assessed using the ankle-brachial index (Doppler stethoscope, Model 841-A; Parks, Aloha, OR). An ankle-brachial index value <0.9 was considered diagnostic (11). Values ≥1.5 were considered missing in five subjects (12). Large fiber sensory nerve function was quantified by measuring vibration perception threshold with the Vibratron II (Physitemp Instruments, Clifton, NJ). The Vibratron II measures the sensitivity of the great toe in detecting small vibratory stimuli. Participants were categorized in three groups according to detection of small vibratory stimuli, as follows: 1) normal function (<3.43 vibration units); 2) mild to moderate dysfunction (3.43 to <6.31 units); 3) severe dysfunction (≥6.31 vibration units) (13). Other potential mediators considered were standardized diagnosis of coronary heart disease, congestive heart failure, stroke, and hypertension. Visual impairment was defined as a visual acuity worse than 20/40 (with corrective lenses, if used) (14).

    Potential confounders

    BMI was computed using measured height and weight. BMI values between 25 and 29.9 kg/m2 were considered overweight and values >30 kg/m2 were considered obese (15). Depressive symptoms were assessed by the Geriatric Depression Scale (cut point ≥14) (16), and cognitive impairment was defined as a Mini Mental State Exam score <24 (17). Prevalence of chronic obstructive pulmonary disease, cancer, osteoarthritis of hips and knees, and hip fracture was also considered.

    Statistical analyses

    We used multivariate polytomous logistic regression to study the association between diabetes and difficulty level in mobility and ADL (little or some and a lot or unable, compared with none). Multivariate logistic regression was used to estimate the association between diabetes status and severe walking impairment, whereas linear regression was used to examine differences in mean summary performance score. Comorbidities and impairments hypothesized to be potential confounders or potential mediators of the association between diabetes and physical function were progressively added to the models. To estimate the contribution of each diabetes-related condition to the association between diabetes and physical function, we compared the point estimate measure of the association for diabetes in the basic model with the point estimates for diabetes in models, including each condition individually (18). In multivariate analyses, separate indicator variables for missing data on PAD (6.8%), visual acuity (9.9%), and peripheral nerve dysfunction (PND; 10.4%) were included.


    Of 1,002 participants in the Women’s Health and Aging Study, 160 (15.9%) had an adjudicated diagnosis of diabetes at baseline, with an average self-reported duration of disease of 13.4 years. The mean age of the study population was 78.8 years. A total of 53% of the sample reported mobility disability, 31.6% had ADL disability, and 26.8% had severe walking limitation. General characteristics of the sample are presented in Table 1, according to diabetes status and duration of disease.

    Age-adjusted prevalence rates of health-related characteristics according to diabetes status are shown in Table 2. Participants with diabetes were more likely to have a cardiovascular condition, severe PND, and visual impairment compared with their counterparts without diabetes (all P values < 0.01). After adjustment for age, the prevalence of coronary heart disease, congestive heart failure, PAD, and hypertension were related to the duration of the disease (all P for trend <0.05). A similar relationship was found for prevalence of depression (P for trend <0.05). With regard to other comorbidities, diabetic women were more likely to be overweight and obese, but there was no association with the duration of disease.

    The age-adjusted prevalence of disability was consistently higher in women with diabetes (Table 3). The associations were significant for all study outcomes: mobility disability (a lot of difficulty or inability to walk 0.25 mile and/or to climb stairs), odds ratio (OR) 1.85 (95% CI 1.12–3.06); ADL disability (a lot of difficulty or inability in at least one ADL), 1.61 (1.06–2.43); severe walking limitation, 2.34 (1.56–3.50). Women with diabetes also had a worse lower extremity summary performance score (β −1.4; P < 0.001). These estimates were not substantially modified by adjustment for demographic factors (Table 3, model 1). Conversely, the strength of association between diabetes and each definition of disability was progressively reduced by subsequently adjusting for other confounders and diabetes complications (Table 3, models 2–4). In the final more parsimonious model, diabetic women had an OR of 1.10 (95% CI 0.64–1.90) for mobility disability, 1.06 (0.67–1.68) for ADL disability, and 1.44 (0.92–2.25) for severe walking limitation. Their summary performance score was 0.48 point lower than in nondiabetic subjects (P = 0.054).

    Finally, we estimated the individual contribution of each diabetes-related condition to the association between diabetes and physical function (Fig. 1). Although there was some variability across the four outcomes, PAD, PND, and symptoms of depression were the most important factors in explaining the association between diabetes and disability. Nevertheless, no one single condition reduced the strength of the association by >30%. For example, controlling separately for PAD, PND, and symptoms of depression decreased the excess risk of mobility disability by 30, 18, and 24%, respectively, whereas the same type of adjustment decreased the excess risk of severe walking limitation by 22, 13, and 18%, respectively. Conversely, when all the covariates were simultaneously controlled for, the excess risk reduction was more substantial for all outcomes. The results were substantially unchanged when BMI was added to demographic factors in the basic models.


    The results of this study show that, among moderately to severely disabled older women, subjects with diabetes have a greater burden of mobility-related disability. Several impairments and chronic conditions that are important causes of disability were more prevalent in women with diabetes. However, only after simultaneously adjusting for all of these factors was the relationship between diabetes and disability reduced to a large degree, suggesting that in older persons, the mechanism linking diabetes and the disablement process is multifactorial.

    Our results reinforce the role of diabetes in the disablement process of older individuals. Indeed, even in a sample of disabled subjects, women with diabetes were affected by a greater level of disability, suggesting that the association could be even stronger in a less-selected sample (i.e., the general population). These findings also extend the results of previous studies, providing new insight into the specific mechanisms underlying the association between diabetes and poor physical function. Diabetes has been consistently related to a broad spectrum of health outcomes, including quality of life, ADL and mobility disability (19), and lower extremity function (20), but the mechanisms responsible for such an association are not clear, particularly in older populations.

    PAD, a main cause of mobility-related disability (21), explained ∼20–30% of the association of diabetes with mobility disability and severe walking limitation, whereas the contribution to explaining ADL disability was much lower. An important contribution was also detected for the presence of PND. However, this impairment was found to be more relevant for the association of diabetes with ADL disability, suggesting that different mechanisms may be involved in different aspects of the disablement process. Of note, the prevalence of PAD and PND were strongly associated with the duration of diabetes. These data support the hypothesis that these impairments are, at least partially, a consequence of the metabolic disease and, therefore, that they act as mediators, rather than as confounders, in the disabling effect of diabetes.

    The greater prevalence of severe depressive symptoms among diabetic women accounted for a substantial proportion of the excess risk of disability associated with diabetes. Several studies have reported that depression is common among patients with diabetes (22). Nevertheless, the mechanism for this link has not been clearly established (23). In our study, depression was associated with the duration of diabetes, adjusting for age, suggesting a potential causal relationship, but the direction of the association cannot be determined, considering the cross-sectional design. In older people, however, depression is a powerful predictor of new disability (24); therefore, longitudinal studies are warranted to elucidate the role of depression in the pathway from diabetes to disability.

    We found that several impairments and comorbidities are important cofactors involved in the disablement process associated with diabetes. For example, obesity, visual impairment, and cardiovascular diseases were also found to be substantial mediators of the association between diabetes and different disability definitions. Our results are in agreement with previous studies showing that these conditions are important causes of difficulty in multiple domains of physical function in older individuals (25,26). Of note, however, is the finding that the association between diabetes and disability was reduced by a large percentage only when all the comorbidities and impairments were simultaneously controlled for. Although we demonstrated the strong impact of conditions such as PAD and PND, these findings indicate that the cause of impaired lower extremity function in older diabetic women is multifactorial. Furthermore, presence of PAD and PND was assessed using sensitive instruments, whereas other conditions, including congestive heart failure and coronary heart disease, were ascertained with less accurate measures. This methodological issue could, at least partially, explain the greater effect detected for PAD and PND.

    We found that, even after adjustment for all covariates, there was a remnant of an association between diabetes and objective measures of lower extremity function, suggesting that additional mechanisms may be involved. For example, we did not have information on other types of neuropathies, including proximal motor neuropathy, that may cause weakness of the proximal muscles of the legs, resulting in difficulty or inability to walk and rise from the sitting position.

    A limitation of this study was that the algorithm used for ascertainment of diabetes did not include fasting glucose level; moreover, HbA1c was available for only 65% of the sample. Consequently, according to the American Diabetes Association criteria (27), some women with fasting glucose level ≥126 mg/dl might have been classified as nondiabetic subjects. Older individuals with newly diagnosed diabetes have greater prevalence of cardiovascular disease compared with subjects without diabetes (28); moreover, even short-term glycemic control has been associated with reduction in several symptoms, including pain, dizziness, and fatigue (29). This evidence suggests that we might have underestimated the strength of the association between diabetes and disability. There are several clinical and biological explanations supporting the association between diabetes and impaired physical function. However, the direction of this association cannot be determined from this cross-sectional study, and therefore, a causal relationship cannot be demonstrated. Conversely, we should consider that, at least theoretically, physical disability could be involved in the development of diabetes.

    Our findings may be relevant for secondary and tertiary prevention in older diabetic patients. The accumulated effect of multiple diabetes-related medical conditions and impairments is responsible for lower extremity impairment and disability in older diabetic women. In addition to appropriate glycemic control, early detection and optimal management of diabetes-related comorbidities may prevent or reduce the burden of physical disability associated with diabetes. The presence of a multifactorial mechanism suggests that older patients with diabetes may greatly benefit from interventions targeting basic impairments such as strength and balance, which are common and powerful mediators of the disablement process. Longitudinal studies and clinical trials should be performed to verify these hypotheses.

    Figure 1—

    Percent reduction in the association between diabetes and different physical function outcomes after adjustment for all potential mediators of the association (full model) and for individual covariates. Only variables with effect ≥5% in at least one outcome are shown. For mobility, ADL, and severe walking disability, bars show the reduction in excess risk of diabetes compared with basic model (diabetes, age, race, education, and smoking). [(OR_basic – OR_adjusted)/OR_basic – 1] × 100. For summary performance measures, bars show the reduction in β-coefficients (from linear regression models) for diabetes compared with basic model (diabetes, age, race, education, and smoking). [1 – (β_adjusted/β_basic)].

    Table 1—

    Selected general characteristics by diabetes status and duration of disease

    Table 2—

    Age-adjusted distribution of selected comorbidities and impairments by diabetes status and duration of disease

    Table 3—

    Multivariate regression analyses for the association of diabetes with different indicators of physical function


    This study was supported by contract NO1-AG-1-2112 from the National Institute on Aging.


    • Address correspondence and reprint requests to Stefano Volpato, MD, MPH, Laboratory of Epidemiology, Demography, and Biometry, National Institute on Aging, 7201 Wisconsin Ave., Gateway Building, Suite 3C-309, Bethesda, MD 20892. E-mail: vlt{at}

      Received for publication 31 May 2001 and accepted in revised form 6 December 2001.

      Caroline Blaum has received honoraria for speaking engagements from Pfizer.

      A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.


    | Table of Contents