Older adults with high waist circumference are at risk for reduced self-reported health, physical activity scores, and disability. Our results also suggest that the presence of OA leads to markedly reduced quality of life and physical activity scores than in subjects without OA in those with elevated visceral adiposity.
Elevated waist circumference has been associated with progression of OA in both cross-sectional and longitudinal epidemiological studies [25, 26]. Our study specifically targets the association of WC with geriatric outcomes, including disability, quality of life and physical activity that have not been adequately evaluated longitudinally. Previous findings have demonstrated inconsistent relationships between WC and functional decline in older adults [27, 28]. Yet, our results are consistent with those  who demonstrated that the highest WC quartile was associated with mobility and agility disability in an older cohort at 2-year follow-up. We found robust relationships between visceral obesity with respect to degree of functional impairment elderly persons over time suggesting a compelling rationale for clinician encouragement of weight reduction targeting visceral or central obesity in older adults at risk for osteoarthritis.
We found that after stratification by OA status, physical component scores of the SF-12, PASE score, and LLDI scores were consistently lower in those with OA. Additionally, gait speed was lower in this cohort. These results are not surprising but the differences between high WC quartile and lowest appeared to be present only in those without OA. These findings suggest that the rate of decline between WC quartiles likely has plateaued well before the development of OA. While purely hypothesis generating, this data could suggest the importance of identifying visceral adiposity in a population at risk knee OA population. Providers should consider promote weight loss to delay physical disability in this population.
Our findings underscore the importance of visceral adiposity measured by easily obtained clinical variables as an important and easily obtained metric with significant predictive value in long-term outcomes for older adults at risk for osteoarthritis. Yet caution is warranted in interpreting our results that may underestimate the natural history of decline observed. First, our cohort was relatively young (mean was ~68 years). The onset of disability increases with increasing age  and hence longer follow-up may be needed to delineate the natural course of the obesity in relation to OA. Frailty and disability often are preceded by years of preserved or compensated functional decline which would not be reflected in our findings. Second, the recruitment was based on a community-based cohort who would be able to participate in this clinical trial and may not be fully representative of the general population of older adults in the community. Third, the degree of co-morbidity was quite modest in the study population, implying a healthier population. While these results provide a reasonable basis for future study of important longitudinal geriatric outcomes, the magnitude of our changes are modest and may not necessarily be clinically significant. A longer follow-up with a cohort with advanced medical and sociodemographic characteristics would be required to confirm our results in a more medically and functionally compromised population of older adults. Lastly, WC quartiles as our main predictor was in line with previous author’s recommendations that it facilitates comparison of anthropometric indices between studies . Incorporating WC as a continuous variable would require polynomial regression modeling, which would lead to challenges with interpretability of our results from a clinical standpoint.
We acknowledge a number of other limitations in our study. While the dataset was designed to observe longitudinal outcomes specific to OA, our analysis may not have coincided with the primary scope of the initial study design. Additionally, of those aged 60 or greater, we were missing data on a number of distal (six-year) outcomes. While we attempted to mitigate this issue by comparing included subjects vs. non-included subjects, and used a statistical methodology to account for such issues, our missing cohort appeared to have higher degree of comorbidity and socioeconomic status, suggesting that our results may indeed be an underestimate of the true effect observed. Importantly, our analysis differed from one published by Colbert et al. , who utilized the OAI and its four-year outcomes to assess the impact of obesity and race on gait speed and WOMAC scores . Although our data was consistent with theirs, these authors did not specifically look at older adults, nor did they utilize PASE, SF-12 or LLDI as primary outcomes. Our results are at risk for possible over-adjustment as well. We deliberately presented unadjusted data to demonstrate the similar trends after accounting for a priori variables that we believed could impact our estimates. We agree that certain variables including social support, depression, and physical measures could conceivably have been omitted and/or excluded in our analysis. To account for possible confounding between predictor variables in the Charlson index, we restricted our modified index to those factors likely not influenced by obesity or disability outcomes. Although the estimates were slightly different, the trends were not. Lastly, future studies should relate anthropometric with body composition measures and distal outcomes.
BMI fails to differentiate between central and peripheral fat stores . These results focus solely on the impact of central adiposity on functional outcomes in older adults. We previously demonstrated in a similar population that BMI impacts quality of life and physical function [Batsis et al. Under Review, Public Health Nutrition], and this current study suggests that abdominal obesity could be a separate predictor of poor functional outcomes. Our intent in this study was not to differentiate which anthropometric measures is superior in predicting distal outcomes. We were unable to adjust for BMI in our modeling as the two variables were highly collinear. This was not surprising in this sample as study recruitment focused on those with risk factors for OA (including overweight and obese based on BMI) making the metrics more homogenous. We believe that future studies should consider not only the impact of elevated WC in normal BMI individuals, but the corollary to identify the potential contribution of each metric. Particularly, in older adults, there is a strong consensus that other simple anthropometric measures should be considered for assessment of adiposity, over and above BMI. While the R2 in our modeling did not differ greatly (Table 7), using WC may still be an important anthropometric variable to measure. Not considering WC as an alternative measure may ignore a considerable sample of subjects at otherwise risk of adverse outcomes as we previously have demonstrated  suggesting the need for future analyses to determine the impact of these combined metrics on outcomes.
Lastly, sarcopenia, the loss of muscle mass and quality with aging [33, 34], impacts function and quality of life, and may be particularly present in subjects with OA . WC is strongly associated with sarcopenic obesity and insulin resistance , both of which are related to functional decline, frailty and disability, and thought to be partially mediated by an increase in pro-inflammatory markers, including IL-1, IL-6, and TNF-a. Additionally, leptin levels are also associated both prevalent and incident central obesity and OA . We suspect that there may be a link between the functional decline observed in the highest WC group in those with OA that may potentially be leptin-mediated. We also suspect that higher degrees of pro-inflammatory cytokines viscerally may be implicated in this phenomenon. Whether the alterations in body composition due to progressive replacement of muscle mass by fat mass, in addition to loss of skeletal muscle mass and function in the aging process may be implicated in this functional decline needs further evaluation. Future study should further examine whether leptin and changes in pro-inflammatory cytokine levels predict functional impairment longitudinally.