Original Article
Obesity
EPIDEMIOLOGY/GENETICS
The Role of Obesity Duration on the Association Between Obesity and Risk of Physical Disability Evelyn Wong1,2, Stephanie K. Tanamas1, Rory Wolfe2, Kathryn Backholer1,2, Christopher Stevenson3, Asnawi Abdullah2,4, and Anna Peeters1,2
Objective: To relate measured obesity duration in mid-life with subsequent incidence of physical disability over and above body mass index (BMI) attained. Methods: Framingham Offspring Study is a longitudinal study that began in 1971. Examination 5 (19911995; “baseline”) and disability onset ascertained from examinations 6-8 (2008) were used. About 2,095 disability-free participants aged 45-65 years at baseline were included. Obesity (BMI 30 kg/m2) duration was calculated between examination 1 and examination 5. Cox regression was used to analyze time to disability. Results: 204 participants developed disability (incidence rate 5 7.9 per 1,000 person-years). Obesity duration ranged from 0 to 22 years (mean of 2.0 years overall, 8.3 years for those with baseline obesity). Obesity duration increased risk of new disability (hazard ratio [HR] 1.07 per year of obesity; 95% confidence interval [CI] 1.05-1.09). This association was attenuated on further adjustment for baseline BMI (HR 1.03; 95% CI 1.00-1.06). Conclusions: Being obese for longer during mid-life increases the risk of later-life disability over and above attained BMI. These results support the need for prevention of weight gain in young adults to avoid an increasing burden of physical disability in later life. Obesity (2015) 23, 443–447. doi:10.1002/oby.20936
Introduction It has recently been estimated that obesity in mid-life increases the risk of disability in old age by two- to fourfold (1). Previous studies examining the relationship between obesity and disability have predominantly measured obesity at a single time point (2). Measuring the health risks associated with obesity using obesity measures at a single time point does not take into account the potential role that obesity duration may play in these risk associations. Given recent trends of obesity rates rising rapidly in younger adults (3) against a backdrop of an ageing population, it is important to elucidate whether living longer with obesity leads to an extra risk of future disability above the level of excess body weight attained. Duration and degree of obesity has previously been shown to be associated with increased risk of diabetes and mortality independent of BMI (4-6). Abdullah et al. proposed a new metric of obesity to
measure health risks and developed a construct termed “obeseyears,” combining duration and degree of obesity, similar to the pack-years measurement of cigarette smoking. As with the association between diabetes and disability, it is likely that similar biological pathways between obesity and disability as well as increases of mechanical load over time would lead to an effect of obesity duration on development of disability independent of BMI. If this hypothesis is true, it would lend support to the need for considering a more refined construct or metric of obesity when estimating the impact of obesity on disability and dependent living in our ageing populations. Only one previous study has analyzed the association between obesity duration and physical limitation (measured as walking limitation) (7). The study analyzed a cohort of 2,055 women and 1,377 men aged over 55 (mean age 68 years) with body mass index (BMI) calculated from recalled weights at 20, 30, 40, and 50 years and height at 20 years. The authors reported an increase risk of walking limitation associated with increased duration of obesity
1 Obesity and Population Health, Baker IDI Heart and Diabetes Institute, Melbourne, Australia. Correspondence: Evelyn Wong ([email protected]) 2 Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne, Australia 3 School of Health and Social Development, Deakin University, Burwood, Australia 4 Department of Biostatistics and Population Health, Faculty of Public Health, University Muhammadiyah Aceh, Indonesia.
Funding agencies: This work was supported by Monash University Australian Postgraduate Award and Baker IDI Bright Sparks Foundation Top-up Award to EW; a VicHealth and National Health and Medical Research Council (NHMRC) Fellowship to AP; and a National Heart Foundation fellowship to KB. This work was also supported by funding from the Australian Research Council (Linkage Project Grant 12010041 & Discovery Grant 120103277) and NHMRC Project Grant (APP1044366). Disclosure: The authors declared no conflict of interest. Author contributions: EW, formulation of research question, design of protocol, data analysis, drafted manuscript, and responsible for manuscript submission and responding to reviewer comments; AP, ST, formulation of research question, design of protocol, interpretation, and commented on manuscript drafts; RW, design of protocol, interpretation, and commented on the final manuscript; KB, CS, AA, interpretation and commented on manuscript drafts. Received: 17 June 2014; Accepted: 23 September 2014; Published online 29 November 2014. doi:10.1002/oby.20936
www.obesityjournal.org
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
443
Obesity
Duration of Obesity and the Risk of Physical Disability Wong et al.
after adjusting for education, smoking, alcohol, and chronic diseases but significance was lost when further adjusted for current BMI. Due to the known limitations of recalled weight (8), it remains unclear whether the association between obesity and disability is simply due to BMI attained, independent of obesity duration. We aimed to evaluate whether obesity duration was associated with incident disability in activities of daily living (ADL) in later life, over and above attained BMI. We further aimed to analyze the relationship between a recently developed obesity construct, which combines duration and degree of obesity, and the development of incident disability. We used 45- to 65-year-olds from the Framingham Offspring Study (FOS), a longitudinal cohort spanning over 30 years.
Methods Framingham Offspring Study The FOS began in 1971 and is ongoing, consisting of offspring of participants from the Framingham Heart Study and their spouses. At the first FOS examination (1971-1975), there were 5,124 participants, aged 5-70 with a mean age of 36 years. The second examination was 8 years later with subsequent follow-up examinations approximately every 4 to 8 years through to examination 8 (20052008). As the original intent of FOS was to study cardiovascular risk factors, at all examinations, extensive medical history, objective measurements of height and weight (where BMI is then calculated), and clinical examination were done. Smoking history was also obtained. Information on education level is only available at examination 2 and was dichotomized to above 12 years of education and 12 years and under. From examination 5 (1991-1995), consistent questions pertaining to ADL disability were added to the participants’ histories. FOS study design, consent, and participant characteristics have been previously described (9). This study protocol was approved by the Monash University standing committee on ethics in research involving humans (CF 11/2075-2011001130).
Examination 5 taken as baseline for physical disability analyses Demographics and data on confounders at examination 5 (1991-1995) provided our baseline data. We used examination 5 as it was the first examination in FOS where disability data was available, and therefore we were able to define a disability-free population at baseline. We included only those aged 45-65 years at this baseline (n 5 2,413). Possible confounders included sex, age, and smoking. Smoking was categorized as current or non-smoker. Missing smoking status at examination 5 was imputed from examination 4 (n 5 1).
Measuring obesity duration We estimated obesity duration using BMI measurements over the approximately 20-year period from examination 1 to examination 5. BMI 30 kg/m2 was regarded as obese. If a participant was obese on at least two consecutive examinations, the period of obesity was the time from the first examination where BMI was 30 kg/m2 to the last of the consecutive examinations where BMI was 30 kg/m2. Total duration of obesity was the accumulation of all periods of obesity between examinations 1-5. Degree of obesity was measured in terms of units in kg/m2 above 29 kg/m2. Where a period of obesity, marked by at least two consecutive examinations with obesity, had
444
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
been identified in an individual, degree of obesity was calculated for each of the examination cycle. Years in between the examination cycle was then multiplied by the degree of obesity at the beginning of that examination cycle and obese-years (10) was the sum of all duration multiplied by degree, up to the baseline year. For example, an individual had obesity from examinations 2-4, with BMI measures of 31 kg/m2, 33 kg/m2, and 34 kg/m2, respectively. The examination cycles are 4 years and the individual’s degree of obesity will then be 2, 4, and 5 for examinations 2, 3, 4 respectively. Since the individual was not obese at examination 5, the duration of obesity was 8 years and the degree of obesity at examination 4 was not added to the calculation for obese-years. Therefore the sum of obese-years for this individual is (4 3 (2 1 4). Obesity duration and obese-years at baseline formed our exposure data. Duration of obesity and obese-years were examined as continuous variables. Where BMI was missing at a specific examination but BMI before and after that examination was available, BMI was imputed using the average between the two available measurements (n 5 408). Participants missing BMI on two or more consecutive examinations were excluded from the analysis (n 5 143) (Figure 1).
Disability as outcome measure Questionnaires from examination 5 through to 8 included questions pertaining to participants’ abilities to participate in an activity in a “normal day.” These were basic self-care activities of bathing, dressing, eating, transferring from chair or bed, and toileting. The possible responses to these questions were “0” for “no help needed, independent”; “1” for “uses device, independent”; “2” for “human assistance needed, minimally dependent”; “3” for “dependent”; and “4” for “do not do during a normal day.” We created a composite measure of disability with disability defined as a response of “1” or greater for at least one of the abovementioned activities. As outlined in Figure 1, we excluded participants who did not have disability data recorded at examination 5 (n 5 14); who reported themselves as having disability at examination 5 (baseline) (n 5 61); or who were missing disability from all of examinations 6-8 (n 5 100).
Statistical analysis We analyzed the association between obesity duration and onset of disability with Cox proportional hazards regression in our total study population as well as in those who were obese at baseline only. Models were adjusted initially for baseline age, sex, and smoking status; then additionally adjusted for baseline BMI. We checked the proportional hazards assumption and found no evidence of any violation of the assumption. We tested for interaction between sex and obesity duration as well as smoking and obesity duration. We did not find evidence of either interaction and therefore we did not stratify our regression models by sex or smoking status. We further tested inclusion of age-squared as well as BMI-squared in the models to explore possible non-linearity in the associations of age and BMI with onset of disability. Age-squared was not significantly associated with disability therefore we did not adjust for agesquared in our models. We tested for interaction between obesity duration and baseline BMI and found no evidence of interaction. Onset of disability was assumed to be at the mid-point between the examination at which a participant first reported having disability and the examination prior. Participants who did not develop disability were censored at date of death, date of examination 8, or last
www.obesityjournal.org
Original Article
Obesity
EPIDEMIOLOGY/GENETICS
Figure 1 Derivation of study population. FOS: Framingham Offspring Study.
follow-up if lost to follow-up. Following the same analysis approach to obesity duration, we analyzed the association between obeseyears at baseline and incident disability with a Cox proportional hazards regression adjusting for age, sex, and smoking. We reported associations between obese-years and disability in terms of 10 obese-years.
Sensitivity analysis We ran the same regression models on the study participants with complete baseline data where BMI was not imputed (N 5 1,728). Education attainment as a marker of socioeconomic position is a potential confounder of the obesity-disability association. Information on participants’ education level was only available at examination 2 as number of years of education. We assumed that given our participants at examination 5 were aged 45-65 years, their education attainment from approximately 12 years earlier would be a good proxy for use at the examination 5 baseline. Two hundred one participants were excluded from this sensitivity analysis due to lack of education data. We reran our regression models and compared effect sizes for obesity duration and obese-years with and without adjusting for education.
TABLE 1 Incidence rates for disability
Men
Women
Total
Number of participants 988 1,107 2,095 Number of new cases of disability 82 122 204 Time at risk (years) 12,257.50 13,529.31 25,786.81 Incidence rate (1,000 person-years) 6.69 9.02 7.91
www.obesityjournal.org
Results Of the 2,095 study participants, 204 developed disability (incidence rate 5 7.91 per 1,000 person-years) and 147 died (Table 1) over a mean of 14 years of follow-up. Obesity duration was right-skewed and ranged from 0 to 22 years (median of zero, mean of 2.0 years). For those with obesity at baseline, the median duration was 7 years (interquartile range 3-16 years; mean 8.3 years). Those who developed disability were older on the average, more likely to be female and current smokers, and had higher BMI at examination 5 and longer mean obesity duration (Table 2). From our Cox regression model adjusting for age, sex, and smoking status, every additional year of obesity increased the risk of developing disability by 7% (hazard ratio [HR], 1.07; 95% confidence interval [CI], 1.05-1.09) (Figure 2). When subsequently adjusting for baseline BMI, it was found that a better fitting model took into account BMI as well as BMI-squared (P-value for BMI-squared term 5 0.03). The hazard ratio relating obesity duration to disability risk was attenuated when additionally adjusted for baseline BMI and BMI-squared although it remained statistically significant (HR 1.03; 95% CI 1.00-1.06). The obese-years construct, which takes into account degree of obesity along with duration, was also significantly associated with disability after adjusting for age, sex, and smoking status. For every additional 10 obese-years, the risk of disability increased by 8% (HR 1.08; 95% CI 1.06-1.10) (Figure 2). In those who were obese at baseline (N 5 541), 92 developed disability (incidence rate 14.54 per 1,000 person-years) and an increase of 10 obese-years was associated with an increased risk of disability by 5% (HR 1.05; 95% CI 1.03-1.08).
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
445
Obesity
Duration of Obesity and the Risk of Physical Disability Wong et al.
TABLE 2 Comparing baseline characteristics by disability
outcome status
Age (mean, SD) % Men % Regular smoker in last year BMI (mean, SD) BMI category % Underweight % Normal weight % Overweight % Obese class I % Obese class II % Obese class III Cumulative years of obesity (mean, SD)
No disability
Disability
Total
1,891 53.7 (5.60) 47.9 18.7
204 56.5 (5.4) 40.2 22.6
2,095 54.0 (5.6) 47.2 19.1
27.4 (4.64)
30.1(6.8)
27.6 (4.9)
0.5 32.5 43.3 17.7 4.2 1.8 2.0 (5.0)
1 22.1 31.9 24 13.2 7.8 5.0 (7.3)
0.5 31.5 42.2 18.3 5.1 2.4 2.3 (5.3)
In our analysis of 1,728 participants with complete BMI data from examination 1-5 (baseline) as well as complete baseline smoking data, we found a similar magnitude of association between duration of obesity and disability to the primary model where some participants’ BMI values were imputed (HR 1.03; 95% CI 1.00-1.07) after adjusting for age, sex, smoking, current BMI, and BMI-squared. The risk association between 10 obese-years increase and disability was also similar to our analysis where BMI was imputed (HR 1.08; 95% CI 1.06-1.10). In our analysis of 1,893 participants who had information on education level, further adjusting for education did not alter the magnitude of the risk association between duration or obese-years and disability.
Discussion In this novel analysis to further our understanding of the association between obesity and the risk of incident disability, we have demon-
strated that longer duration of obesity is associated with increased disability risk over and above BMI attained. Each additional year lived with obesity added a further 3% to the risk of disability for a given attained BMI. In a parallel analysis, each additional 10 obeseyears was associated with an increased risk of disability by approximately 8%. It is important to view the additional risks of obesity duration from the perspective of the observed trends of earlier onset of obesity, and the increasing degree of obesity (11) amidst an aging population. These trends translate to increasing number of years spent living with obesity. Prior studies analyzing the risk of disability from obesity compared to normal weight have generally defined obesity without taking into account duration (2). To the best of our knowledge, only one previously published study analyzed duration of obesity and the risk of physical limitation (7). Stenholm et al. have previously reported an increased risk of walking limitation associated with increased duration of self-reported recalled obesity after adjusting for education, smoking, alcohol, and chronic diseases (7). When analysis was restricted to only those with obesity there was significant linear trend between increased categories of duration and walking limitation although the latter association was not statistically significant after adjusting for current BMI. This study was however limited by using recalled weights of over 30 years to ascertain weight status. It has previously been shown that the longer away weight is recalled, the less accurate it is, with an increase in the potential for misclassification error resulting in a tendency toward a null association (8). Our finding that duration and degree of obesity are related to the development of disability is not surprising and can likely be explained by the pathways between obesity and disability. The disease pathways likely involve various obesity-related conditions such as cardiovascular disease, diabetes, and osteoarthritis (12-16). The pathways may be through either systemic inflammation, deposition of excess fat surrounding main organs such as in the pericardium, an increase mechanical load on or pain in weight-bearing joints, or mediated by effects of hyperglycaemia on muscle function (17-21). Increasing duration and degree of inflammation, fat deposits, and mechanical load may relate to increased risk of developing disability. A major strength of our analysis is the availability of repeated objectively measured height and weight over 22 years by trained research staff enabling a good estimate of obesity duration. This lessens the likelihood of reporting bias. Another strength is the inclusion of only participants aged 45-65 at baseline to address the issues of using BMI as a measurement of adiposity in the elderly due to sarcopenia (1,22). An additional strength is in the availability of repeated measures of ADL disability with questions consistent across the three examinations. Having disability questions at baseline enabled us to analyze a population free of disability at baseline, thereby addressing the possibility of reverse causation in the obesity-disability association.
Figure 2 Risk of disability associated with obesity duration and obese-years. HR: hazard ratio; BMI: body mass index.
446
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
Our analysis was limited by lack of information for a precise calculation of time to event. We therefore assumed that the onset of disability occurred mid-way between the last examination where the participant was known to be free of disability and the examination when disability was reported. This assumption is unlikely to lead to bias or alter the direction or magnitude of the obesity
www.obesityjournal.org
Original Article
Obesity
EPIDEMIOLOGY/GENETICS
duration-disability association. Other limitations of the study include a lack of information on education attainment at baseline and some missing data. We addressed these possible limitations through sensitivity analyses and found little difference in the magnitude of the risk of disability associated with obesity duration and obese-years. The use of a different metric of obesity taking into account the number of years lived with obesity has been previously studied and discussed in relation to estimating broad health risks (4-6,10,23). This recently developed metric is particularly relevant given that obesity prevalence has increased in children and young adults in an era of rapidly ageing populations, resulting in increased number of years lived with obesity. Our findings of significantly increased risk of new disability from obesity duration in addition to BMI attained poses new challenges to estimating the overall health burden and economic costs of obesity. ADL disability is associated with loss of independent living with costs to individuals, the wider community and health care resources. Projection studies of the health burden and costs of obesity in the USA and UK estimated medical costs from obesity-related diseases to increase by $48-66 billion/year in the USA, £1.9-2 billion/year in the UK by year 2030 (24). Wang et al. further estimated losses to quality-adjusted life-years (QALYs) due to obesity-related diseases that are disabling such as osteoarthritis and hypertension. These losses of QALYs estimates were between 2.2 and 6.3 million in the UK and 24.5 and 48.2 million in the USA. These estimates of costs and losses of QALYs were based on both historic and recent obesity trends in both countries. However these estimates did not take into account the possible additional effect of obesity duration on the development of obesity-related diseases, subsequent disability, or mortality. Whilst we have demonstrated an additional 3% increase in risk of disability for each year lived with obesity, Abdullah et al. demonstrated a 3% increase risk of all-cause mortality and 3-5.5% increase risk of diabetes for each year of obesity, above BMI attained (4,5). It would be prudent for future studies into the health and economic burden of obesity to take into consideration the number of years spent living with obesity. Our findings highlight firstly the importance of preventing or postponing the onset of obesity and secondly, that a more nuanced metric of obesity is needed if we are to understand or estimate the true cost and burden of obesity.
Conclusion This is the first study to demonstrate that duration of obesity increases the risk of disability over and above BMI attained. This has important implications when estimating the burden of obesity where obesity prevalence in children and young adults has risen against a backdrop of an ageing population. O
Acknowledgments The authors would like to acknowledge the Framingham Heart Study (FHS) for access to the Offspring Datasets. The FHS is con-
www.obesityjournal.org
ducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the FHS Investigators. C 2014 The Obesity Society V
References 1. Backholer K, Pasupathi K, Wong E, Hodge A, Stevenson C, Peeters A. The relationship between body mass index prior to old age and disability in old age. Int J Obes 2012;1-7. 2. Backholer K, Wong E, Freak-Poli R, Walls HL, Peeters A. Increasing body weight and risk of limitations in activities of daily living: a systematic review and metaanalysis. Obes Rev 2012;13:456-468. 3. Tanamas SK, Shaw JE, Backholer K, Magliano DJ, Peeters A. 12-year weight change, waist circumference change and incident obesity: the Australian Diabetes, Obesity and Lifestyle study. Obesity 2014;22:1538-1545. 4. Abdullah A, Stoelwinder J, Shortreed S, et al. The duration of obesity and the risk of type 2 diabetes. Public Health Nutr 2011;14:119-126. 5. Abdullah A, Wolfe R, Stoelwinder JU, et al. The number of years lived with obesity and the risk of all-cause and cause-specific mortality. Int J Epidemiol 2011; 1-12. 6. The NS, Richardson AS, Gordon-Larsen P. Timing and duration of obesity in relation to diabetes: findings from an ethnically diverse, nationally representative sample. Diabetes Care 2013;36:865-872. 7. Stenholm S, Rantanen T, Alanen E, Reunanen A, Sainio P, Koskinen S. Obesity history as a predictor of walking limitation at old age. Obesity 2007;15:929-938. 8. Stevens J, Keil Je, Waid lR, Gazes PC. Accuracy of current, 4-year, and 28-year self-reported body weight in an elderly population. Am J Epidemiol 1990;132:11561163. 9. Kannel WB, Feinleib M, Mcnamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families: the Framingham Offspring Study. Am J Epidemiol 1979;110:281-290. 10. Abdullah A, Wolfe R, Mannan H, Stoelwinder JU, Stevenson C, Peeters A. Epidemiologic merit of obese-years, the combination of degree and duration of obesity. Am J Epidemiol 2012;176:99-107. 11. Walls H, Wolfe R, Haby M, et al. Trends in BMI of urban Australian adults, 19802000. Public Health Nutr 2010;13:631-638. 12. Folsom AR, Stevens J, Schreiner PJ, McGovern PG. Body mass index, waist/hip ratio, and coronary heart disease incidence in African Americans and Whites. Am J Epidemiol 1998;148:1187-1194. 13. Wong E, Backholer K, Gearon E, et al. Diabetes and risk of physical disability in adults: a systematic review and meta-analysis. Lancet Diabetes Endocrinol 2013;1: 106-114. 14. Pinsky JL, Branch LG, Jette AM, et al. Framingham disability study: relationship of disability to cardiovascular risk factors among persons free of diagnosed cardiovascular disease. Am J Epidemiol 1985;122:644-656. 15. Fowler-Brown A, Wee CC, Marcantonio E, Ngo L, Leveille S. The mediating effect of chronic pain on the relationship between obesity and physical function and disability in older adults. J Am Geriatr Soc 2013;61:2079-2086. 16. Heo M, Pietrobelli A, Wang D, Heymsfield SB, Faith MS. Obesity and functional impairment: Influence of comorbidity, joint pain, and mental health. Obesity 2010; 18:2030-2038. 17. Ridker PM, Hennekens CH, Buring JE, Rifai N. C-Reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women. N Eng J Med 2000;342:836-843. 18. Rosito GA, Massaro JM, Hoffmann U, et al. Pericardial fat, visceral abdominal fat, cardiovascular disease risk factors, and vascular calcification in a community-based sample: the Framingham Heart Study. Circulation 2008;117: 605-613. 19. Teichtahl AJ, Wang Y, Wluka AE, Cicuttini FM. Obesity and knee osteoarthritis: new insights provided by body composition studies. Obesity 2008;16:232-240. 20. Park SW, Goodpaster BH, Strotmeyer ES, et al. Accelerated loss of skeletal muscle strength in older adults with Type 2 diabetes: the Health, Aging, and Body Composition Study. Diabetes Care 2007;30:1507-1512. 21. Brinkley TE, Hsu F-C, Beavers KM, et al. Total and abdominal adiposity are associated with inflammation in older adults using a factor analysis approach. J Gerontol Ser A 2012;67:1099-1106. 22. Goodpaster BH, Park SW, Harris TB, et al. The loss of skeletal muscle strength, mass, and quality in older adults: the Health, Aging and Body Composition Study. J Gerontol Ser A 2006;61:1059-1064. 23. De Koning L, Hu FB. Commentary: Obesity-years—a new metric to measure health effects of obesity. Int J Epidemiol 2011;1-2. 24. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. Health and economic burden of the projected obesity trends in the USA and the UK. Lancet 2011;378: 815-825.
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
447
Obesity
EPIDEMIOLOGY/GENETICS
The Role of Obesity Duration on the Association Between Obesity and Risk of Physical Disability Evelyn Wong1,2, Stephanie K. Tanamas1, Rory Wolfe2, Kathryn Backholer1,2, Christopher Stevenson3, Asnawi Abdullah2,4, and Anna Peeters1,2
Objective: To relate measured obesity duration in mid-life with subsequent incidence of physical disability over and above body mass index (BMI) attained. Methods: Framingham Offspring Study is a longitudinal study that began in 1971. Examination 5 (19911995; “baseline”) and disability onset ascertained from examinations 6-8 (2008) were used. About 2,095 disability-free participants aged 45-65 years at baseline were included. Obesity (BMI 30 kg/m2) duration was calculated between examination 1 and examination 5. Cox regression was used to analyze time to disability. Results: 204 participants developed disability (incidence rate 5 7.9 per 1,000 person-years). Obesity duration ranged from 0 to 22 years (mean of 2.0 years overall, 8.3 years for those with baseline obesity). Obesity duration increased risk of new disability (hazard ratio [HR] 1.07 per year of obesity; 95% confidence interval [CI] 1.05-1.09). This association was attenuated on further adjustment for baseline BMI (HR 1.03; 95% CI 1.00-1.06). Conclusions: Being obese for longer during mid-life increases the risk of later-life disability over and above attained BMI. These results support the need for prevention of weight gain in young adults to avoid an increasing burden of physical disability in later life. Obesity (2015) 23, 443–447. doi:10.1002/oby.20936
Introduction It has recently been estimated that obesity in mid-life increases the risk of disability in old age by two- to fourfold (1). Previous studies examining the relationship between obesity and disability have predominantly measured obesity at a single time point (2). Measuring the health risks associated with obesity using obesity measures at a single time point does not take into account the potential role that obesity duration may play in these risk associations. Given recent trends of obesity rates rising rapidly in younger adults (3) against a backdrop of an ageing population, it is important to elucidate whether living longer with obesity leads to an extra risk of future disability above the level of excess body weight attained. Duration and degree of obesity has previously been shown to be associated with increased risk of diabetes and mortality independent of BMI (4-6). Abdullah et al. proposed a new metric of obesity to
measure health risks and developed a construct termed “obeseyears,” combining duration and degree of obesity, similar to the pack-years measurement of cigarette smoking. As with the association between diabetes and disability, it is likely that similar biological pathways between obesity and disability as well as increases of mechanical load over time would lead to an effect of obesity duration on development of disability independent of BMI. If this hypothesis is true, it would lend support to the need for considering a more refined construct or metric of obesity when estimating the impact of obesity on disability and dependent living in our ageing populations. Only one previous study has analyzed the association between obesity duration and physical limitation (measured as walking limitation) (7). The study analyzed a cohort of 2,055 women and 1,377 men aged over 55 (mean age 68 years) with body mass index (BMI) calculated from recalled weights at 20, 30, 40, and 50 years and height at 20 years. The authors reported an increase risk of walking limitation associated with increased duration of obesity
1 Obesity and Population Health, Baker IDI Heart and Diabetes Institute, Melbourne, Australia. Correspondence: Evelyn Wong ([email protected]) 2 Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne, Australia 3 School of Health and Social Development, Deakin University, Burwood, Australia 4 Department of Biostatistics and Population Health, Faculty of Public Health, University Muhammadiyah Aceh, Indonesia.
Funding agencies: This work was supported by Monash University Australian Postgraduate Award and Baker IDI Bright Sparks Foundation Top-up Award to EW; a VicHealth and National Health and Medical Research Council (NHMRC) Fellowship to AP; and a National Heart Foundation fellowship to KB. This work was also supported by funding from the Australian Research Council (Linkage Project Grant 12010041 & Discovery Grant 120103277) and NHMRC Project Grant (APP1044366). Disclosure: The authors declared no conflict of interest. Author contributions: EW, formulation of research question, design of protocol, data analysis, drafted manuscript, and responsible for manuscript submission and responding to reviewer comments; AP, ST, formulation of research question, design of protocol, interpretation, and commented on manuscript drafts; RW, design of protocol, interpretation, and commented on the final manuscript; KB, CS, AA, interpretation and commented on manuscript drafts. Received: 17 June 2014; Accepted: 23 September 2014; Published online 29 November 2014. doi:10.1002/oby.20936
www.obesityjournal.org
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
443
Obesity
Duration of Obesity and the Risk of Physical Disability Wong et al.
after adjusting for education, smoking, alcohol, and chronic diseases but significance was lost when further adjusted for current BMI. Due to the known limitations of recalled weight (8), it remains unclear whether the association between obesity and disability is simply due to BMI attained, independent of obesity duration. We aimed to evaluate whether obesity duration was associated with incident disability in activities of daily living (ADL) in later life, over and above attained BMI. We further aimed to analyze the relationship between a recently developed obesity construct, which combines duration and degree of obesity, and the development of incident disability. We used 45- to 65-year-olds from the Framingham Offspring Study (FOS), a longitudinal cohort spanning over 30 years.
Methods Framingham Offspring Study The FOS began in 1971 and is ongoing, consisting of offspring of participants from the Framingham Heart Study and their spouses. At the first FOS examination (1971-1975), there were 5,124 participants, aged 5-70 with a mean age of 36 years. The second examination was 8 years later with subsequent follow-up examinations approximately every 4 to 8 years through to examination 8 (20052008). As the original intent of FOS was to study cardiovascular risk factors, at all examinations, extensive medical history, objective measurements of height and weight (where BMI is then calculated), and clinical examination were done. Smoking history was also obtained. Information on education level is only available at examination 2 and was dichotomized to above 12 years of education and 12 years and under. From examination 5 (1991-1995), consistent questions pertaining to ADL disability were added to the participants’ histories. FOS study design, consent, and participant characteristics have been previously described (9). This study protocol was approved by the Monash University standing committee on ethics in research involving humans (CF 11/2075-2011001130).
Examination 5 taken as baseline for physical disability analyses Demographics and data on confounders at examination 5 (1991-1995) provided our baseline data. We used examination 5 as it was the first examination in FOS where disability data was available, and therefore we were able to define a disability-free population at baseline. We included only those aged 45-65 years at this baseline (n 5 2,413). Possible confounders included sex, age, and smoking. Smoking was categorized as current or non-smoker. Missing smoking status at examination 5 was imputed from examination 4 (n 5 1).
Measuring obesity duration We estimated obesity duration using BMI measurements over the approximately 20-year period from examination 1 to examination 5. BMI 30 kg/m2 was regarded as obese. If a participant was obese on at least two consecutive examinations, the period of obesity was the time from the first examination where BMI was 30 kg/m2 to the last of the consecutive examinations where BMI was 30 kg/m2. Total duration of obesity was the accumulation of all periods of obesity between examinations 1-5. Degree of obesity was measured in terms of units in kg/m2 above 29 kg/m2. Where a period of obesity, marked by at least two consecutive examinations with obesity, had
444
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
been identified in an individual, degree of obesity was calculated for each of the examination cycle. Years in between the examination cycle was then multiplied by the degree of obesity at the beginning of that examination cycle and obese-years (10) was the sum of all duration multiplied by degree, up to the baseline year. For example, an individual had obesity from examinations 2-4, with BMI measures of 31 kg/m2, 33 kg/m2, and 34 kg/m2, respectively. The examination cycles are 4 years and the individual’s degree of obesity will then be 2, 4, and 5 for examinations 2, 3, 4 respectively. Since the individual was not obese at examination 5, the duration of obesity was 8 years and the degree of obesity at examination 4 was not added to the calculation for obese-years. Therefore the sum of obese-years for this individual is (4 3 (2 1 4). Obesity duration and obese-years at baseline formed our exposure data. Duration of obesity and obese-years were examined as continuous variables. Where BMI was missing at a specific examination but BMI before and after that examination was available, BMI was imputed using the average between the two available measurements (n 5 408). Participants missing BMI on two or more consecutive examinations were excluded from the analysis (n 5 143) (Figure 1).
Disability as outcome measure Questionnaires from examination 5 through to 8 included questions pertaining to participants’ abilities to participate in an activity in a “normal day.” These were basic self-care activities of bathing, dressing, eating, transferring from chair or bed, and toileting. The possible responses to these questions were “0” for “no help needed, independent”; “1” for “uses device, independent”; “2” for “human assistance needed, minimally dependent”; “3” for “dependent”; and “4” for “do not do during a normal day.” We created a composite measure of disability with disability defined as a response of “1” or greater for at least one of the abovementioned activities. As outlined in Figure 1, we excluded participants who did not have disability data recorded at examination 5 (n 5 14); who reported themselves as having disability at examination 5 (baseline) (n 5 61); or who were missing disability from all of examinations 6-8 (n 5 100).
Statistical analysis We analyzed the association between obesity duration and onset of disability with Cox proportional hazards regression in our total study population as well as in those who were obese at baseline only. Models were adjusted initially for baseline age, sex, and smoking status; then additionally adjusted for baseline BMI. We checked the proportional hazards assumption and found no evidence of any violation of the assumption. We tested for interaction between sex and obesity duration as well as smoking and obesity duration. We did not find evidence of either interaction and therefore we did not stratify our regression models by sex or smoking status. We further tested inclusion of age-squared as well as BMI-squared in the models to explore possible non-linearity in the associations of age and BMI with onset of disability. Age-squared was not significantly associated with disability therefore we did not adjust for agesquared in our models. We tested for interaction between obesity duration and baseline BMI and found no evidence of interaction. Onset of disability was assumed to be at the mid-point between the examination at which a participant first reported having disability and the examination prior. Participants who did not develop disability were censored at date of death, date of examination 8, or last
www.obesityjournal.org
Original Article
Obesity
EPIDEMIOLOGY/GENETICS
Figure 1 Derivation of study population. FOS: Framingham Offspring Study.
follow-up if lost to follow-up. Following the same analysis approach to obesity duration, we analyzed the association between obeseyears at baseline and incident disability with a Cox proportional hazards regression adjusting for age, sex, and smoking. We reported associations between obese-years and disability in terms of 10 obese-years.
Sensitivity analysis We ran the same regression models on the study participants with complete baseline data where BMI was not imputed (N 5 1,728). Education attainment as a marker of socioeconomic position is a potential confounder of the obesity-disability association. Information on participants’ education level was only available at examination 2 as number of years of education. We assumed that given our participants at examination 5 were aged 45-65 years, their education attainment from approximately 12 years earlier would be a good proxy for use at the examination 5 baseline. Two hundred one participants were excluded from this sensitivity analysis due to lack of education data. We reran our regression models and compared effect sizes for obesity duration and obese-years with and without adjusting for education.
TABLE 1 Incidence rates for disability
Men
Women
Total
Number of participants 988 1,107 2,095 Number of new cases of disability 82 122 204 Time at risk (years) 12,257.50 13,529.31 25,786.81 Incidence rate (1,000 person-years) 6.69 9.02 7.91
www.obesityjournal.org
Results Of the 2,095 study participants, 204 developed disability (incidence rate 5 7.91 per 1,000 person-years) and 147 died (Table 1) over a mean of 14 years of follow-up. Obesity duration was right-skewed and ranged from 0 to 22 years (median of zero, mean of 2.0 years). For those with obesity at baseline, the median duration was 7 years (interquartile range 3-16 years; mean 8.3 years). Those who developed disability were older on the average, more likely to be female and current smokers, and had higher BMI at examination 5 and longer mean obesity duration (Table 2). From our Cox regression model adjusting for age, sex, and smoking status, every additional year of obesity increased the risk of developing disability by 7% (hazard ratio [HR], 1.07; 95% confidence interval [CI], 1.05-1.09) (Figure 2). When subsequently adjusting for baseline BMI, it was found that a better fitting model took into account BMI as well as BMI-squared (P-value for BMI-squared term 5 0.03). The hazard ratio relating obesity duration to disability risk was attenuated when additionally adjusted for baseline BMI and BMI-squared although it remained statistically significant (HR 1.03; 95% CI 1.00-1.06). The obese-years construct, which takes into account degree of obesity along with duration, was also significantly associated with disability after adjusting for age, sex, and smoking status. For every additional 10 obese-years, the risk of disability increased by 8% (HR 1.08; 95% CI 1.06-1.10) (Figure 2). In those who were obese at baseline (N 5 541), 92 developed disability (incidence rate 14.54 per 1,000 person-years) and an increase of 10 obese-years was associated with an increased risk of disability by 5% (HR 1.05; 95% CI 1.03-1.08).
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
445
Obesity
Duration of Obesity and the Risk of Physical Disability Wong et al.
TABLE 2 Comparing baseline characteristics by disability
outcome status
Age (mean, SD) % Men % Regular smoker in last year BMI (mean, SD) BMI category % Underweight % Normal weight % Overweight % Obese class I % Obese class II % Obese class III Cumulative years of obesity (mean, SD)
No disability
Disability
Total
1,891 53.7 (5.60) 47.9 18.7
204 56.5 (5.4) 40.2 22.6
2,095 54.0 (5.6) 47.2 19.1
27.4 (4.64)
30.1(6.8)
27.6 (4.9)
0.5 32.5 43.3 17.7 4.2 1.8 2.0 (5.0)
1 22.1 31.9 24 13.2 7.8 5.0 (7.3)
0.5 31.5 42.2 18.3 5.1 2.4 2.3 (5.3)
In our analysis of 1,728 participants with complete BMI data from examination 1-5 (baseline) as well as complete baseline smoking data, we found a similar magnitude of association between duration of obesity and disability to the primary model where some participants’ BMI values were imputed (HR 1.03; 95% CI 1.00-1.07) after adjusting for age, sex, smoking, current BMI, and BMI-squared. The risk association between 10 obese-years increase and disability was also similar to our analysis where BMI was imputed (HR 1.08; 95% CI 1.06-1.10). In our analysis of 1,893 participants who had information on education level, further adjusting for education did not alter the magnitude of the risk association between duration or obese-years and disability.
Discussion In this novel analysis to further our understanding of the association between obesity and the risk of incident disability, we have demon-
strated that longer duration of obesity is associated with increased disability risk over and above BMI attained. Each additional year lived with obesity added a further 3% to the risk of disability for a given attained BMI. In a parallel analysis, each additional 10 obeseyears was associated with an increased risk of disability by approximately 8%. It is important to view the additional risks of obesity duration from the perspective of the observed trends of earlier onset of obesity, and the increasing degree of obesity (11) amidst an aging population. These trends translate to increasing number of years spent living with obesity. Prior studies analyzing the risk of disability from obesity compared to normal weight have generally defined obesity without taking into account duration (2). To the best of our knowledge, only one previously published study analyzed duration of obesity and the risk of physical limitation (7). Stenholm et al. have previously reported an increased risk of walking limitation associated with increased duration of self-reported recalled obesity after adjusting for education, smoking, alcohol, and chronic diseases (7). When analysis was restricted to only those with obesity there was significant linear trend between increased categories of duration and walking limitation although the latter association was not statistically significant after adjusting for current BMI. This study was however limited by using recalled weights of over 30 years to ascertain weight status. It has previously been shown that the longer away weight is recalled, the less accurate it is, with an increase in the potential for misclassification error resulting in a tendency toward a null association (8). Our finding that duration and degree of obesity are related to the development of disability is not surprising and can likely be explained by the pathways between obesity and disability. The disease pathways likely involve various obesity-related conditions such as cardiovascular disease, diabetes, and osteoarthritis (12-16). The pathways may be through either systemic inflammation, deposition of excess fat surrounding main organs such as in the pericardium, an increase mechanical load on or pain in weight-bearing joints, or mediated by effects of hyperglycaemia on muscle function (17-21). Increasing duration and degree of inflammation, fat deposits, and mechanical load may relate to increased risk of developing disability. A major strength of our analysis is the availability of repeated objectively measured height and weight over 22 years by trained research staff enabling a good estimate of obesity duration. This lessens the likelihood of reporting bias. Another strength is the inclusion of only participants aged 45-65 at baseline to address the issues of using BMI as a measurement of adiposity in the elderly due to sarcopenia (1,22). An additional strength is in the availability of repeated measures of ADL disability with questions consistent across the three examinations. Having disability questions at baseline enabled us to analyze a population free of disability at baseline, thereby addressing the possibility of reverse causation in the obesity-disability association.
Figure 2 Risk of disability associated with obesity duration and obese-years. HR: hazard ratio; BMI: body mass index.
446
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
Our analysis was limited by lack of information for a precise calculation of time to event. We therefore assumed that the onset of disability occurred mid-way between the last examination where the participant was known to be free of disability and the examination when disability was reported. This assumption is unlikely to lead to bias or alter the direction or magnitude of the obesity
www.obesityjournal.org
Original Article
Obesity
EPIDEMIOLOGY/GENETICS
duration-disability association. Other limitations of the study include a lack of information on education attainment at baseline and some missing data. We addressed these possible limitations through sensitivity analyses and found little difference in the magnitude of the risk of disability associated with obesity duration and obese-years. The use of a different metric of obesity taking into account the number of years lived with obesity has been previously studied and discussed in relation to estimating broad health risks (4-6,10,23). This recently developed metric is particularly relevant given that obesity prevalence has increased in children and young adults in an era of rapidly ageing populations, resulting in increased number of years lived with obesity. Our findings of significantly increased risk of new disability from obesity duration in addition to BMI attained poses new challenges to estimating the overall health burden and economic costs of obesity. ADL disability is associated with loss of independent living with costs to individuals, the wider community and health care resources. Projection studies of the health burden and costs of obesity in the USA and UK estimated medical costs from obesity-related diseases to increase by $48-66 billion/year in the USA, £1.9-2 billion/year in the UK by year 2030 (24). Wang et al. further estimated losses to quality-adjusted life-years (QALYs) due to obesity-related diseases that are disabling such as osteoarthritis and hypertension. These losses of QALYs estimates were between 2.2 and 6.3 million in the UK and 24.5 and 48.2 million in the USA. These estimates of costs and losses of QALYs were based on both historic and recent obesity trends in both countries. However these estimates did not take into account the possible additional effect of obesity duration on the development of obesity-related diseases, subsequent disability, or mortality. Whilst we have demonstrated an additional 3% increase in risk of disability for each year lived with obesity, Abdullah et al. demonstrated a 3% increase risk of all-cause mortality and 3-5.5% increase risk of diabetes for each year of obesity, above BMI attained (4,5). It would be prudent for future studies into the health and economic burden of obesity to take into consideration the number of years spent living with obesity. Our findings highlight firstly the importance of preventing or postponing the onset of obesity and secondly, that a more nuanced metric of obesity is needed if we are to understand or estimate the true cost and burden of obesity.
Conclusion This is the first study to demonstrate that duration of obesity increases the risk of disability over and above BMI attained. This has important implications when estimating the burden of obesity where obesity prevalence in children and young adults has risen against a backdrop of an ageing population. O
Acknowledgments The authors would like to acknowledge the Framingham Heart Study (FHS) for access to the Offspring Datasets. The FHS is con-
www.obesityjournal.org
ducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the FHS Investigators. C 2014 The Obesity Society V
References 1. Backholer K, Pasupathi K, Wong E, Hodge A, Stevenson C, Peeters A. The relationship between body mass index prior to old age and disability in old age. Int J Obes 2012;1-7. 2. Backholer K, Wong E, Freak-Poli R, Walls HL, Peeters A. Increasing body weight and risk of limitations in activities of daily living: a systematic review and metaanalysis. Obes Rev 2012;13:456-468. 3. Tanamas SK, Shaw JE, Backholer K, Magliano DJ, Peeters A. 12-year weight change, waist circumference change and incident obesity: the Australian Diabetes, Obesity and Lifestyle study. Obesity 2014;22:1538-1545. 4. Abdullah A, Stoelwinder J, Shortreed S, et al. The duration of obesity and the risk of type 2 diabetes. Public Health Nutr 2011;14:119-126. 5. Abdullah A, Wolfe R, Stoelwinder JU, et al. The number of years lived with obesity and the risk of all-cause and cause-specific mortality. Int J Epidemiol 2011; 1-12. 6. The NS, Richardson AS, Gordon-Larsen P. Timing and duration of obesity in relation to diabetes: findings from an ethnically diverse, nationally representative sample. Diabetes Care 2013;36:865-872. 7. Stenholm S, Rantanen T, Alanen E, Reunanen A, Sainio P, Koskinen S. Obesity history as a predictor of walking limitation at old age. Obesity 2007;15:929-938. 8. Stevens J, Keil Je, Waid lR, Gazes PC. Accuracy of current, 4-year, and 28-year self-reported body weight in an elderly population. Am J Epidemiol 1990;132:11561163. 9. Kannel WB, Feinleib M, Mcnamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families: the Framingham Offspring Study. Am J Epidemiol 1979;110:281-290. 10. Abdullah A, Wolfe R, Mannan H, Stoelwinder JU, Stevenson C, Peeters A. Epidemiologic merit of obese-years, the combination of degree and duration of obesity. Am J Epidemiol 2012;176:99-107. 11. Walls H, Wolfe R, Haby M, et al. Trends in BMI of urban Australian adults, 19802000. Public Health Nutr 2010;13:631-638. 12. Folsom AR, Stevens J, Schreiner PJ, McGovern PG. Body mass index, waist/hip ratio, and coronary heart disease incidence in African Americans and Whites. Am J Epidemiol 1998;148:1187-1194. 13. Wong E, Backholer K, Gearon E, et al. Diabetes and risk of physical disability in adults: a systematic review and meta-analysis. Lancet Diabetes Endocrinol 2013;1: 106-114. 14. Pinsky JL, Branch LG, Jette AM, et al. Framingham disability study: relationship of disability to cardiovascular risk factors among persons free of diagnosed cardiovascular disease. Am J Epidemiol 1985;122:644-656. 15. Fowler-Brown A, Wee CC, Marcantonio E, Ngo L, Leveille S. The mediating effect of chronic pain on the relationship between obesity and physical function and disability in older adults. J Am Geriatr Soc 2013;61:2079-2086. 16. Heo M, Pietrobelli A, Wang D, Heymsfield SB, Faith MS. Obesity and functional impairment: Influence of comorbidity, joint pain, and mental health. Obesity 2010; 18:2030-2038. 17. Ridker PM, Hennekens CH, Buring JE, Rifai N. C-Reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women. N Eng J Med 2000;342:836-843. 18. Rosito GA, Massaro JM, Hoffmann U, et al. Pericardial fat, visceral abdominal fat, cardiovascular disease risk factors, and vascular calcification in a community-based sample: the Framingham Heart Study. Circulation 2008;117: 605-613. 19. Teichtahl AJ, Wang Y, Wluka AE, Cicuttini FM. Obesity and knee osteoarthritis: new insights provided by body composition studies. Obesity 2008;16:232-240. 20. Park SW, Goodpaster BH, Strotmeyer ES, et al. Accelerated loss of skeletal muscle strength in older adults with Type 2 diabetes: the Health, Aging, and Body Composition Study. Diabetes Care 2007;30:1507-1512. 21. Brinkley TE, Hsu F-C, Beavers KM, et al. Total and abdominal adiposity are associated with inflammation in older adults using a factor analysis approach. J Gerontol Ser A 2012;67:1099-1106. 22. Goodpaster BH, Park SW, Harris TB, et al. The loss of skeletal muscle strength, mass, and quality in older adults: the Health, Aging and Body Composition Study. J Gerontol Ser A 2006;61:1059-1064. 23. De Koning L, Hu FB. Commentary: Obesity-years—a new metric to measure health effects of obesity. Int J Epidemiol 2011;1-2. 24. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. Health and economic burden of the projected obesity trends in the USA and the UK. Lancet 2011;378: 815-825.
Obesity | VOLUME 23 | NUMBER 2 | FEBRUARY 2015
447
