Karine R. Sahakyan, MD, PhD, MPH; Virend K. Somers, MD, PhD; Juan P. Rodriguez-Escudero, MD; David O. Hodge, MS; Rickey E. Carter, PhD; Ondrej Sochor, MD; Thais Coutinho, MD; Michael D. Jensen, MD; Véronique L. Roger, MD, MPH; Prachi Singh, PhD; Francisco Lopez-Jimenez, MD, MS
This article was published online first at www.annals.org on 10 November 2015.
Grant Support: By the National Institutes of Health (grant HL00711-36; Dr. Sahakyan), American Heart Association (grant 11SDG7260046; Dr. Singh), European Regional Development Fund (Project FNUSA-ICRC [grant CZ.1.05/1.1.00/02.0123; Drs. Somers, Sochor, Singh, and Lopez-Jimenez]), and Czech Ministry of Health (grant NT13434-4/2012; Dr. Sochor).
Disclosures: Dr. Somers reports grants from the National Institutes of Health and Philips Respironics (to the Mayo Foundation for the study of sleep and cardiovascular disease); personal fees from Respicardia, ResMed, Sorin Group, Philips Respironics, GlaxoSmithKline, U-Health (China; www.ttdoc.cn), and Ronda Grey; and other compensation for work with Mayo Health Solutions and their industry partners on intellectual property related to sleep and cardiovascular disease, outside the submitted work. Dr. Sochor's institution, St. Anne's Hospital, was supported by the European Regional Development Fund (Project FNUSA-ICRC [grant CZ.1.05/1.1.00/02.0123]); further, Dr. Sochor was awarded grants from the Ministry of Health of the Czech Republic and SCOPES (NT13434-4/2012 and IZ73Z0_152616, respectively). Authors not named here have disclosed no conflicts of interest. Disclosures can also be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M14-2525.
Editors' Disclosures: Christine Laine, MD, MPH, Editor in Chief, reports that she has no financial relationships or interests to disclose. Darren B. Taichman, MD, PhD, Executive Deputy Editor, reports that he has no financial relationships or interests to disclose. Cynthia D. Mulrow, MD, MSc, Senior Deputy Editor, reports that she has no relationships or interests to disclose. Deborah Cotton, MD, MPH, Deputy Editor, reports that she has no financial relationships or interest to disclose. Jaya K. Rao, MD, MHS, Deputy Editor, reports that she has stock holdings/options in Eli Lilly and Pfizer. Sankey V. Williams, MD, Deputy Editor, reports that he has no financial relationships or interests to disclose. Catharine B. Stack, PhD, MS, Deputy Editor for Statistics, reports that she has stock holdings in Pfizer.
Reproducible Research Statement:Study protocol: Available at ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/nhanes/nhanes3/1A/ADULT-acc.pdf. Statistical code and data: Available at www.cdc.gov/nchs/nhanes/nh3data.htm.
Requests for Single Reprints: Francisco Lopez-Jimenez, MD, MS, Division of Cardiovascular Diseases, Mayo Clinic, 200 First Street Southwest, Rochester, MN 55905; e-mail, firstname.lastname@example.org.
Current Author Addresses: Drs. Sahakyan, Somers, Rodriguez-Escudero, Sochor, Roger, Singh, and Lopez-Jimenez: Division of Cardiovascular Diseases, Mayo Clinic, 200 First Street Southwest, Rochester, MN 55905.
Mr. Hodge and Dr. Carter: Division of Biomedical Statistics and Informatics, Mayo Clinic, 200 First Street Southwest, Rochester, MN 55905.
Dr. Coutinho: University of Ottawa Heart Institute, 40 Ruskin Street, Ottawa, Ontario K1Y 4W7, Canada.
Dr. Jensen: Division of Endocrinology, Mayo Clinic, 200 First Street Southwest, Rochester, MN 55905.
Author Contributions: Conception and design: K.R. Sahakyan, J.P. Rodriguez-Escudero, R.E. Carter, T. Coutinho, F. Lopez-Jimenez.
Analysis and interpretation of the data: K.R. Sahakyan, V.K. Somers, D.O. Hodge, R.E. Carter, O. Sochor, T. Coutinho, M.D. Jensen, V.L. Roger, P. Singh, F. Lopez-Jimenez.
Drafting of the article: K.R. Sahakyan, V.K. Somers, J.P. Rodriguez-Escudero, R.E. Carter, P. Singh, F. Lopez-Jimenez.
Critical revision of the article for important intellectual content: K.R. Sahakyan, V.K. Somers, J.P. Rodriguez-Escudero, D.O. Hodge, R.E. Carter, O. Sochor, T. Coutinho, M.D. Jensen, V.L. Roger, P. Singh, F. Lopez-Jimenez.
Final approval of the article: K.R. Sahakyan, V.K. Somers, J.P. Rodriguez-Escudero, R.E. Carter, T. Coutinho, M.D. Jensen, P. Singh, F. Lopez-Jimenez.
Provision of study materials or patients: F. Lopez-Jimenez.
Statistical expertise: D.O. Hodge, R.E. Carter, F. Lopez-Jimenez.
Obtaining of funding: V.L. Roger.
Administrative, technical, or logistic support: V.K. Somers, P. Singh, F. Lopez-Jimenez.
Collection and assembly of data: R.E. Carter.
Sahakyan K., Somers V., Rodriguez-Escudero J., Hodge D., Carter R., Sochor O., Coutinho T., Jensen M., Roger V., Singh P., Lopez-Jimenez F.; Normal-Weight Central Obesity: Implications for Total and Cardiovascular Mortality. Ann Intern Med. 2015;163:827-835. doi: 10.7326/M14-2525
Download citation file:
Published: Ann Intern Med. 2015;163(11):827-835.
Published at www.annals.org on 10 November 2015
The relationship between central obesity and survival in community-dwelling adults with normal body mass index (BMI) is not well-known.
To examine total and cardiovascular mortality risks associated with central obesity and normal BMI.
Stratified multistage probability design.
NHANES III (Third National Health and Nutrition Examination Survey).
15 184 adults (52.3% women) aged 18 to 90 years.
Multivariable Cox proportional hazards models were used to evaluate the relationship of obesity patterns defined by BMI and waist-to-hip ratio (WHR) and total and cardiovascular mortality risk after adjustment for confounding factors.
Persons with normal-weight central obesity had the worst long-term survival. For example, a man with a normal BMI (22 kg/m2) and central obesity had greater total mortality risk than one with similar BMI but no central obesity (hazard ratio [HR], 1.87 [95% CI, 1.53 to 2.29]), and this man had twice the mortality risk of participants who were overweight or obese according to BMI only (HR, 2.24 [CI, 1.52 to 3.32] and 2.42 [CI, 1.30 to 4.53], respectively). Women with normal-weight central obesity also had a higher mortality risk than those with similar BMI but no central obesity (HR, 1.48 [CI, 1.35 to 1.62]) and those who were obese according to BMI only (HR, 1.32 [CI, 1.15 to 1.51]). Expected survival estimates were consistently lower for those with central obesity when age and BMI were controlled for.
Body fat distribution was assessed based on anthropometric indicators alone. Information on comorbidities was collected by self-report.
Normal-weight central obesity defined by WHR is associated with higher mortality than BMI-defined obesity, particularly in the absence of central fat distribution.
National Institutes of Health, American Heart Association, European Regional Development Fund, and Czech Ministry of Health.
Australian National University
November 12, 2015
Normal-Weight Central Obesity and Mortality Risk: How should We Interpret Statistical Models with Interactions?
Sahakyan and colleagues (1) analyzed the 1988-1994 National Health and Nutrition Examination Survey to examine mortality risk associated with body mass index (BMI) and waist-to-hip ratio (WHR). They stated that “WHR, but not BMI, was associated with high mortality risk after including both variables in the model” (Results) and that “[P]ersons with normal-weight central obesity had the worst long-term survival” (Abstract). These two statements are not compatible with each other. If BMI were not associated with mortality risk, then people who are normal-weight (based on BMI) but centrally obese (based on WHR) should not have a worse survival than those who are obese based on both BMI and WHR.Additionally, the authors misinterpreted the model results for males. Under their final model in Appendix 1, for males, the quadratic terms for both BMI and WHR and the interaction between the quadratic terms are statistically significant at α=0.05. It is well recognized in statistics that when there are higher-order interactions in the model, lower-order main effect terms should not be interpreted on their own. Their results indicate that both BMI and WHR are associated with male mortality risk, rather than no association for BMI with high mortality as stated in the Results of the article. Figure 1 in the article shows that at the same WHR value, normal-weight men have a higher mortality than overweight or obese men, based on BMI; the mortality elevation is non-significant when WHR is 0.89 and significant when WHR is 1. However, the authors defined obesity as having a BMI of 33, which is conventionally considered as falling into the category of moderate obesity. It would be useful to provide comparisons against severe obesity (BMI 35 and greater). In 2007-2008, the prevalence of having a BMI 35 and greater is 14.3% for US adults aged 20 and above (10.7% for males and 17.8% for females), and the corresponding figures for having a BMI 40 and greater is 5.7%, 4.2% and 7.2% (2). The significant quadratic terms in the article suggest that even after adjustment for WHR, mortality risk could be higher for the severely obese than normal-weight men. For females, Figure 2 in the article indicate that the centrally obese who are normal-weight have a lower mortality than the centrally obese who are overweight or moderately obese, and the differences are marginally significant. Clearly, the former have no worse survival than the latter.REFERENCES1. Sahakyan KR, Somers VK, Rodriguez-Escudero JP, Hodge DO, Carter RE, Sochor O, et al. Normal-Weight Central Obesity: Implications for Total and Cardiovascular Mortality Risk in Persons With Normal-Weight Central Obesity. Annals of Internal Medicine. 2015;N/A(N/A):N/A-N/A.2. Flegal KM, Carroll MD, Ogden CL, Curtin LR. Prevalence and Trends in Obesity Among US Adults, 1999-2008. JAMA. 2010 January 20, 2010;303(3):235-41.
F Ramirez Lafita MD, FACP, M Castella MD, MA Amigo MD
Dept of Occupational Health, ANAV. L'Hospitalet de l'Infant, Tarragona, Spain
December 24, 2015
Further screening on cardiovascular risk factors
Several studies have found that measures of abdominal obesity (waist circumference, waist-to-hip ratio, and waist-to-height ratio) are more closely related to CVD morbidity and mortality than is BMI (1) Sahakyan et al (2) have performed a thoughtful study where normal weight central obesity defined by waist-to-hip ratio (WHR) was found to be associated with higher mortality rates. WHR and waist circumference measures were superior to their defined BMI values as predictors of total and cardiovascular mortality rates but, authors recognize that other risk’ factors could also contribute to this increase. In fact, visceral adipose tissue deposit plays an important pathogenic role in the metabolic syndrome (3). Visceral fat accumulation is associated with dyslipemia, hypertriglyceridemia, insulin resistance and inflammation. The accumulation of fat in the visceral cavity and particularly in the liver is commonly accompanied by inflammatory processes and contributes to the genesis of insulin resistance. These conditions and the metabolic syndrome are closely associated with nonalcoholic fatty liver disease (NAFLD). Recent studies have found that liver fat content increases the risk of cardiovascular events in long-term follow-up (4) In the last years we have run a cross-sectional study in workers of an Electric Power Industry to investigate the relationship of NAFLD with obesity, central obesity and vascular risk factors. Preliminary results in 480 employees (mean age 46) have shown that NAFLD was correlated with overweight, central obesity, increased blood pressure, dyslipemia, hypertriglyceridemia, type 2 diabetes, hyperuricemia and higher Framingham’s coronary risk index (p<0.001). Our findings were especially interesting in younger than 40. Authors suggest that persons with normal weight and central obesity represent an interesting target population for developing preventive strategies and lifestyle modification in order to prevent CVD. As NAFLD is closely associated with abdominal obesity, dyslipidemia, hypertension, and Type 2 diabetes, which are all features of the metabolic syndrome, should liver ultrasound be considered a useful practice in such preventive strategies? 1. Czernichow S, Kengne AP, Stamatakis E, et al. Body mass index, waist circumference and waist-hip ratio: which is the better discriminator of cardiovascular disease mortality risk? Evidence from an individual-participant meta-analysis of 82 864 participants from nine cohort studies. ObesityReviews 2011; 12(9): 680-7.2. Sahakyan KR, Somers VK, Rodriguez-Escudero JP, Hodge DO, Carter RE, Sochor O, et al. . Normal-Weight Central Obesity: Implications for Total and Cardiovascular Mortality Risk in Persons With Normal-Weight Central Obesity. Ann Intern Med. 2015;163(11):827-835. doi:10.7326/M14-25253. Kim LJ , Nalls MA, Eiriksdottir G, Sigurdsson S, Launer LJ et al. Associations of visceral and liver fat with the metabolic syndrome acrossthe spectrum of obesity: the AGESReykjavik study.4. Pisto P, Santaniemi M, Bloigu R, Ukkola O, Kesäniemi A. Fatty liver predicts the risk for cardiovascular events in middle-aged population: a population-based cohort study. BMJ Open 2014;4: e004973. doi:10.1136/ bmjopen-2014-004973
Rickey E. Carter, David O. Hodge, Francisco Lopez-Jimenez
January 26, 2016
Authors’ Response to “Normal-Weight Central Obesity and Mortality Risk: How should we interpret statistical models with interactions?”
Yu raises important points about the interpretation of our models. Yu’s critique of the statement that BMI did not have an association in men is appropriate given the presence of the statistically significant interaction. However, the interpretation of a single parameter estimate in the model was not the basis for this statement. Figures 1 and 2 were specifically developed to present the combined effects of these variables (i.e., the combined interaction and main effects as Yu suggests as the appropriate course). From a pragmatic perspective, the higher order interaction was estimated with a coefficient of 0.002 (=ln(1.002)), so even large changes away from the mean BMI of 27 had little effect on the estimated hazard. This minimal effect is shown clearly in Table 2. Specifically, holding WHR constant and allowing BMI to change (i.e., compare alternating rows in the tables), one observes much smaller changes in expected mortality than when one holds BMI constant and allows WHR to change (i.e., compare row pairs as presented). For example, the ten-year mortality ratio for a 40 year old male with BMI of 37.0 and a WHR of 0.89 relative to a 40 year old with a BMI of 27.5 and a WHR of 0.89 is ~1.0 (=2.2/2.3). The mortality ratios for the same BMI but different levels of WHR were >1.7. So, while technically we agree with Yu regarding the statistical significance of BMI, the practical implications of the interaction were limited. In women, BMI as an interaction term with WHR or as a main effect was not support by the data.Yu also expresses interest in understanding comparisons of different anthropometric profiles other than what we presented in Figures 1 and 2. The estimates in these figures were based on the parameter estimates in Appendix Table 1. As such, readers are able to calculate the point estimate for hazard ratios for combinations of interest using the normal process of specifying values for variables. The final point Yu makes is not easily addressed using information in Figure 2 along. Table 2 illustrates that the expected mortality for women (within age stratum) who are centrally obese is universally higher than those without central obesity. While statistical comparisons of these mortality estimates were beyond the scope of the paper, one can observe a general rank ordering of the estimates. As expected, the normal weight, no central obesity may be the ideal anthropomorphic state.
to gain full access to the content and tools.
Learn more about subscription options.
Register Now for a free account.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only