Body composition and functional limitation in COPD

The FLOW study of COPD (Function, Living, Outcomes, and Work) is an ongoing prospective cohort study of adult members of a closed panel managed care organization with physician's diagnosis of COPD. Its long-term goal is to determine what factors are responsible for the development of disability in COPD. At baseline assessment, we conducted structured telephone interviews that ascertained COPD status, health status, health-related quality of life, self-reported functional limitations, and sociodemographic characteristics. Subjects then underwent a research clinic visit that included spirometry, bioelectrical impedance, and other physical assessments. Using these baseline data, we evaluated the cross-sectional impact of body composition on the risk of functional limitations among adults with COPD. The study was approved by the University of California, San Francisco Committee on Human Research and the Kaiser Foundation Research Institute's institutional review board.

We studied adult members of Kaiser Permanente (KP), the nation's largest non-profit managed care organization. In Northern California, the Kaiser Permanente Medical Care Program (KPMCP) provides the full spectrum of primary-to-tertiary care to approximately 3.1 million members. In Northern California, KP's share of the regional population ranges from 25 to 30% [25]. The demographic characteristics of KP membership are similar to the overall Northern California population, except for the extremes of income distribution [26].

We identified all adult KPMCP members aged 40–65 years who were recently treated for COPD using a previously described approach [27]. Because an overall study outcome is work disability, younger adults with COPD were recruited. Using KPMCP computerized databases, we identified all subjects who had health care utilization for COPD during the most recent 12 month time period, including 1 or more ambulatory visits, emergency department visits, or hospitalizations with a principal International Classification of Disease (ICD-9) diagnosis code for COPD, which included chronic bronchitis (491), emphysema (492), or COPD (496) PLUS two or more prescriptions for a COPD-related medication during a 12 month window beginning 6 months before the index utilization date and ending 6 months after index date (these medications included inhaled anticholinergic medications, inhaled beta agonists, inhaled corticosteroids, and theophylline). Based on medical record review, we demonstrated that this algorithm is a valid method for identifying adults with COPD [27]. To facilitate attendance at the research clinic, we restricted the sample to persons living within a 30 mile radius of the clinic. The primary care physician for each patient was contacted and given the opportunity to decline contact of their patients. Potential subjects were then contacted by a letter describing the study and given the opportunity to decline by mail. Those not declining were then contacted by telephone to arrange an interview. At the end of the interview, subjects were invited to participate in the research clinic visit. Persons who were found to have other severe life-threatening conditions (e.g., cancer), severe communication or language difficulties (e.g, dementia or stroke), or were not proficient in English were excluded.

This analysis was conducted after the first phase of cohort recruitment. 3144 subjects with COPD were identified and the first randomly sampled 1183 subjects who met all study criteria were eligible for the current analysis. Of the 1183 eligible subjects, 710 (60%) subjects completed structured telephone interviews and 355 (50%) completed the research clinic visit.

Each subject underwent a structured telephone interview that was 30–40 minutes in length and conducted using customized computer-assisted telephone interview software. Interviews ascertained age, sex, race-ethnicity, and educational attainment. Cigarette smoking was measured using questions developed for the National Health Interview Survey [28]. As in previous studies, we defined educational attainment as high school or less, some college, or college/graduate degree [4]. Race-ethnicity was categorized as previously described [4].

Self-reported functional limitation was measured using a previously validated approach used by Sternfeld and colleagues, based on questions from the Framingham Disability Study, Established Populations for Epidemiologic Studies of the Elderly, the Nagle scale, and Rosow and Breslau scales [29]. The scale is comprised of 10 questions that assess the degree of difficulty in multiple domains of basic physical functioning such as pushing, stooping, kneeling, getting up from a standing position, lifting lighter or heavier objects, standing, sitting, standing from a seated position, walking up stairs, and walking in the neighborhood. Subjects who indicated "a lot of difficulty" with one or more functions or not doing a function because they were unable or they were told by a doctor not to were defined as having a self-reported functional limitation [29].

Body composition was assessed using bioelectric impedance (BIA).The Quantum II Bioelectrical Body Composition Analyzer (RJL Systems, Clinton Township, MI) was used. While subjects were lying supine, we applied bipolar electrodes on the middle finger of the right hand and the lateral aspect of the right ankle to obtain measures of resistance and reactance. To calculate lean and fat mass, we used established sex-specific regression equations derived from healthy adults living in Northern California who underwent both BIA testing with the Quantum II device and whole-body dual energy x-ray absorptiometry (DEXA) scans [29]. Lean mass and lean-plus-bone mass were derived from these regression equations (in kilograms); fat mass was obtained by subtracting lean-plus-bone mass from weight (because weight = fat mass + lean-plus-bone mass) [29].

A relative measure of body composition, the lean-to-fat ratio, was calculated by dividing lean mass by fat mass. Previous work has established that lean-to-fat ratio is more closely related to functional limitation than lean mass alone. The lean-to-fat ratio was more strongly associated with walking speed and the risk of self-reported functional limitation among elderly adults than were lean or fat mass [29, 30]. In addition, the lean-to-fat ratio appeared to mediate the beneficial effects of leisure time physical activity on physical functioning [31]. Lean-to-fat ratio has substantive analytic advantages, because it is independent of body size and is not collinear with height (whereas lean mass and height are collinear).

To assess central adiposity (i.e., visceral fat), we measured sagittal abdominal diameter (SAD). SAD and waist circumference are both excellent measures of visceral fat as determined by MRI or CT scanning [32‐36]. SAD appears to be more responsive to weight loss [37]. We chose SAD over waist circumference for this analysis because it correlates more strongly with pulmonary function (both forced vital capacity and forced expiratory volume in 1 second [FEV₁]) [38]. Moreover, preliminary analysis indicated that SAD was related to overall fat mass, whereas waist circumference was not.

To measure SAD, we used the Holtain Kahn caliper (Holtain Ltd, U.K.). Subjects were studied in the supine position. The examiner located the iliac crests, visualized a line connecting the crests, and marked the center of the abdomen along this line. The caliper was then slid under the back and the caliper's upper arm was slid down until it was 2 cm above the abdominal mark. The caliper was then leveled using the bubble level. The caliper's upper arm was then slid down so that it was just touching, but not compressing, the abdomen. The level position was re-confirmed and the distance in centimeters was determined.

Body mass index, as a more general measure of adiposity, was also determined from height and weight measured at the research clinic visit (weight in kilograms/height in meters²). Height was measured by a wall stadiometer in subjects without shoes; weight was measured by a digital scale. Body mass index was categorized into 4 groups using the standard National Heart Lung and Blood Institute/World Health Organization criteria: underweight (< 18.5 kg/m²), normal weight 18.5–24.9 kg/m², overweight (25–29.9 kg/m²), and obese (≥ 30 kg/m²) [39]. Because there were only 9 subjects in the underweight category (3%), they were considered with the normal weight group for analytic purposes.

We assessed functional limitations, which are decrements in basic physical actions, using a multifaceted evaluation that combined a survey-based measure (self-reported functional limitation as described above) and physical assessment. Submaximal exercise performance was measured using the Six Minute Walk Test, which was developed by Guyatt and had been widely used in studies of COPD [40, 41]. We measured submaximal rather than maximal exercise performance (cardiopulmonary exercise testing) because most daily activities and work tasks are likely to require sustained, submaximal exertion rather than high peak exercise levels. We used a standardized flat, straight course of 30 meters in accordance with American Thoracic Society (ATS) Guidelines [42]. Subjects who use home oxygen or who have a resting oxygen saturation < 90% wore supplemental oxygen. Every 2 minutes, the technician spoke standardized encouragement phrases, as recommended by the ATS guidelines. The primary outcome was the distance walked in 6 minutes.

Lower extremity function was measured using the validated Short Physical Performance Battery [43‐45]. The battery included 3 performance measures, which were scored from 0 to 4 points. The standing balance test asks subjects to maintain their feet in a side-by-side, semi-tandem stand (heel of one foot next to the big toe of the other foot), or tandem stand (heel of one foot directly in front of the other foot) for 10 seconds. The maximum score of 4 is assigned for maintaining the tandem stand for 10 seconds; a low score of 1 is assigned for side-by-side standing for 10 seconds, with inability to hold a semi-tandem position for 10 seconds. A test of walking speed requires subjects to walk 4 meters at their normal pace. Participants are assigned a score from 1 to 4 based on the quartile of length of time needed to complete the test. The chair stand test, which reflects lower extremity extensor muscle strength, measures the time required for the subject to stand up and sit down from a chair 5 times with arms folded across the chest. The chair height is standardized for all subjects. Scores from 1 to 4 are assigned based on quartile of length of time to complete the task. A summary performance score integrates the 3 performance measures, ranging from 0 to 12. Previous work indicates that the battery has excellent inter-observer reliability, test-retest reliability, and predictive validity [43‐45].

To assess respiratory impairment, we conducted spirometry according to American Thoracic Society (ATS) Guidelines [46, 47]. Briefly, subjects were tested in a seated position with a nose clip in place. After the technician demonstrated the procedure, subjects performed at least 3 maximal expiratory maneuvers. If reproducibility criteria are not met (FVC and/or FEV₁ variability ≤ 0.2 liters), up to 8 maneuvers were obtained. We used the EasyOne™ Frontline spirometer (ndd Medical Technologies, Chelmsford, MA), which meets ATS criteria. To calculate percent predicted pulmonary function values, we used predictive equations derived from NHANES III [48]. Because FEV₁/FVC ratio is more affected by body size and composition than FEV_1, we used FEV₁/FVC in multivariate analysis to control for pulmonary function impairment [38, 49]. Based on FEV_1,FEV₁/FVC ratio, and respiratory symptoms, COPD severity was staged based on NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria (stage 0 to IV) [6, 50].

Statistical analysis was conducted using SAS software, version 9.1 (SAS Institute, Inc, Cary, NC). We used logistic regression analysis to elucidate the impact of body composition on the risk of self-reported physical functional limitation. The lean-to-fat ratio was chosen as the primary body composition variable (as discussed above). We also examined separate regression models for SAD (an estimate of visceral fat) and BMI (a more general indicator of adiposity). These variables were not included in the same models because of their inter-correlation and the concern for collinearity. To examine potential confounding, 3 sets of analyses are presented that control for covariates: age; age, height, and FEV₁/FVC; age, height, FEV₁/FVC, race (white, non-Hispanic vs. other), educational attainment, and smoking history (current smoking and ex-smoking vs. never smoked). To examine the impact of body composition on submaximal exercise performance (Six Minute Walk Test) and lower extremity functioning (Short Physical Performance Battery), multivariate linear regression was used in analogous fashion. Because weight is mostly composed of lean mass and fat mass, it was not included as a covariate in the regression analysis.

A further series of analyses examined the independent impact of lean mass, fat mass, and visceral fat (SAD) on physical functional limitation when considered in the same regression models. Because lean mass was highly correlated with height and fat mass, we used the approach of Sternfeld and colleagues and developed a residual variable for lean mass from its regression on height and fat mass [29]. The residual variable for lean mass (lean mass_resid) represents the part of lean mass not accounted for by height and fat mass (i.e., the correlations between lean mass_resid and height, and lean mass_resid and fat mass are zero). A residual variable was also developed for SAD from its regression on fat mass and lean mass (i.e., SAD_resid = that part of SAD not accounted for by fat and lean mass).

All regression analyses were stratified by sex, because there was evidence that sex modified the impact of body composition on the risk of functional limitation in many of the analyses. This approach is also consistent with the literature [29, 31, 51]. To assess the impact of GOLD stage 0 on the results, sensitivity analyses were performed restricted to subjects with GOLD stages I or greater; the results were highly consistent with the primary analysis and are not reported here.

Fat free mass index is sometimes used in studies of body composition. The fat free mass index strongly correlated with lean mass (r = 0.91; p < 0.0001). When key analyses were repeated substituting fat free mass index for lean mass, the results were highly similar to those based on lean mass (data not shown).

The mean age was 58 (6.2) years and there was a slight predominance of female subjects (60%) (Table 1). The majority of subjects were white (64%), with a substantial proportion of other race-ethnic groups. The majority (82%) indicated smoking during their lifetime. There was a diversity of educational attainment.

Table 2 shows pulmonary function and body composition measurements. The mean FEV₁ was 1.71 liters and the majority of subjects were GOLD stage I or greater. A slight majority of subjects were obese (54%) based on BMI. A substantial proportion were overweight (20%) or normal weight (24%), whereas very few were underweight (3%).

In men, a higher sagittal abdominal diameter was associated with a greater risk of self-reported functional limitation, but the confidence interval was wide and did not exclude no effect (OR 1.09 per 1 cm increment; 95% CI 0.99 to 1.21) (Table 3). There was no apparent relation between lean-fat ratio or BMI and self-reported functional limitation.

Among women, a higher lean-to-fat ratio was associated with a lower risk of self-reported functional limitation in the fully adjusted model (OR 0.45 per 0.50 increment in lean-to-fat ratio; 95% CI 0.28 to 0.74). Higher sagittal abdominal diameter and obese body mass index were also related to a greater risk of functional limitation (OR 1.15 per 1 cm increment; 95% CI 1.07 to 1.23 and OR 3.50 for obese vs. normal BMI; 95% CI 1.53 to 8.01, respectively).

Body composition was associated with exercise performance on the Six Minute Walk Test, although the effects were greater for woman than for men (Table 4). Among men, a higher lean-to-fat ratio was associated with a greater distance walked in 6 minutes in the fully adjusted analysis (mean difference 40 meters per 0.50 ratio increment; 95% CI 9 to 71 meters). In women, the lean-to-fat ratio was associated with an even greater distance walked (mean difference 162 meters per 0.50 increment; 95% CI 97 to 228 meters). Larger sagittal abdominal diameter and obese BMI were also related to less distance walked in 6 minutes in both sexes (Table 4).

Among men, higher sagittal abdominal diameter was associated with worse performance on the walking speed score and summary performance score of the Short Physical Performance Battery (Table 5). In the female stratum, lean-to-fat ratio, sagittal abdominal diameter, and obese BMI were all related to walking speed score, chair stand scores, and summary performance scores in the expected directions.

To further elucidate the impact of body composition on functional limitation, lean mass (residual), fat mass, and visceral fat (estimated by sagittal abdominal diameter residual) were included simultaneously in each fully adjusted multivariate model (see Methods). Among men, higher fat mass was associated with a decrement in the Six Minute Walk Test (-13 meters per 1 kg fat mass increment; 95% CI -21 to -5 meters) and was possibly related to a greater risk of self-reported functional limitations (OR 1.06 per 1 kg increment; 95% CI 0.99 to 1.13) (Table 6). Higher visceral fat, as estimated by sagittal abdominal diameter, was also related to poorer walk performance (-38 meters per 1 cm increment; 95% CI -68 to -7 meters).

Among women, higher fat mass was related to a greater risk of functional limitations (OR 1.04 per 1 kg increment in fat mass; 95% CI 1.017 to 1.067), 11 meter decrement in the distance walked in six minutes (95% CI -15 to -8), and a poorer SPPB summary performance score (-0.037 points per 1 kg increment; 95% CI -0.053 to -0.020).