Introduction

Obesity is a major public health problem in the United States, with increasing prevalence during the last four decades (1,2). Recent estimates suggest that the prevalence of obesity among U.S. adults, defined as a BMI 30 kg/m², increased to 31% in 1999 to 2000 compared with 23% in 1988 to 1994 (2). Obesity has been associated with an increased risk for many chronic diseases, including hypertension, dyslipidemia, cardiovascular disease (CVD),¹ type 2 diabetes, gallbladder disease, osteoarthritis, sleep apnea, respiratory problems, and endometrial, breast, and colon cancers (3). In 1995, it was estimated that $99.2 billion dollars in medical costs could be attributed to obesity, of which $51.6 billion was caused by the direct medical costs associated with obesity-related comorbidities (ORCM) (4).

Despite the clear association between obesity and many chronic diseases, it has also long been appreciated that the natural history and consequences of obesity can vary widely. The possibility of a healthy subtype of obese individuals who do not develop ORCMs has been debated for years (5). As early as the 1940s, Vague (6,7) suggested that the "android" type of obesity, in which fat predominated in the upper body, was associated with an increased risk for obesity-related metabolic diseases, whereas "gynecoid" obesity, with lower body predominance, was more of an esthetic problem. Several researchers have documented the existence of a metabolically normal subgroup of obese individuals who remain without any obesity-related metabolic diseases (5,8). Approximately 30% of obese participants in a large Health Maintenance Organization did not report any ORCMs (9). These studies, when taken together, raise the question of whether the clinical or demographic factors that differentiate obese individuals with and without ORCMs can be readily identified.

Identifying variables associated with having ORCMs may provide clues about the pathophysiology of these comorbidities and may also help to tailor obesity treatment to the subgroup likely to receive the greatest overall health benefit from weight reduction. Pertinent here, such characterization of risk subgroups may improve understanding of obesity-related health outcomes and treatment profiles in African Americans. In the 1999–2000 National Health and Nutrition Examination Survey (NHANES) data for adults 20 years of age, the highest prevalence of obesity was observed in non-Hispanic black women (2). However, the weaker link between obesity and mortality in African-American compared with white populations has suggested a relatively lower impact of ORCMs in the African-American population (10,11). The need to tailor treatment programs for greater effectiveness in African Americans has been suggested on the basis of several observations of smaller weight losses among African Americans than whites enrolled in the same weight loss program (12).

This study assessed the prevalence and correlates of ORCMs in a sample of African-American adults with Class I to Class III obesity (e.g., BMI levels of 30 to 50 kg/m²) (13). Study participants were enrollees in Phase 1 of the Healthy Eating and Lifestyle Program (HELP) Study (14), which evaluated a 10-week behavioral weight reduction program offered in an outpatient setting. Subgroups defined by baseline ORCM status were compared on baseline characteristics, attendance, and program completion. Among those who returned for data collection after the 10-week program, ORCM status subgroups were compared on postprogram weight change.

Research Methods and Procedures

Background

The HELP program was offered by the family practice department of a university-based health system. All enrolled participants were expected to attend a series of 10 weekly classes focusing on nutrition, physical activity, and behavior change. Classes were held in the family practice department conference room and were led by nutritionists, exercise specialists, and a behavioral counselor, usually working in teams of two. Particular care was taken to ensure that the classes were interactive and culturally salient to the African-American participants. The Phase 1 evaluation was a one-group pre-post design. The complete study design, rationale, and main results are reported in detail elsewhere (14). Methods relevant to this analysis are summarized below.

Study Population and Eligibility

The study was advertised to African-American men and women between the ages of 25 and 70 years, whose primary care provider was within the university health system. The BMI eligibility range was 30 (29.5) to 50 (50.4) kg/m². Individuals were eligible regardless of their history of CVD, diabetes, or other ORCMs if their disease status was stable enough for their personal physician to grant permission to enroll in the study. Those who were pregnant, nonambulatory, currently using psychotropic or antidepressant drugs, being treated with chemo- or radiation therapy, or who could not otherwise obtain clearance from their personal physician to participate in the program were not eligible. As reported elsewhere (14), 244 of 380 individuals screened were found to be eligible for the program. Of the 136 not eligible, nearly one-half (n = 67) were outside of the BMI range, and another 20 were ineligible because they did not have a physician within the university-health system. Only 11 were excluded for medical reasons.

Study Procedures

Recruitment, data collection, and intervention procedures were approved by the University of Pennsylvania Institutional Review Board. Participants were recruited between April 2000 and May 2001.

Data Collection

All pre- and postprogram measurements were obtained by the same trained individual. Height was measured in inches without shoes using a portable stadiometer (Road Rod; SECA, Hanover, MD). Weight was determined using a portable Electronic Digital BWB–627A Scale (TANITA, Tokyo, Japan), which had a 600-lb capacity, a stainless steel platform, and a leveling gauge for improved accuracy of measurements. Waist circumference (WC) was measured using a 60-in retractable pliable tape according to the NHANES III protocol (15). Seated blood pressures were measured in all participants using a mercury sphygmomanometer (Baumanometer Desk Model; W.A. Baum Co. Inc., Copiague, NY), according to the American Heart Association guidelines (16). As part of the baseline visit, participants were instructed to have their blood samples taken after at least 12 hours of fasting. The fasting blood samples were drawn on a separate occasion and subsequently analyzed at the Core Laboratory of the General Clinical Research Center for determination of total cholesterol, low-density lipo-protein-cholesterol, high-density lipoprotein-cholesterol, triglycerides, and glucose.

Demographic information, medical and weight history, and smoking and alcohol history were obtained by self-administered questionnaire at the baseline visit. Medical history was initially self-reported on baseline forms. This information was subsequently reviewed during the participant's initial baseline consultation with the study physician and reconciled with the medical history information provided by the participant's personal physician on the medical evaluation form required for final eligibility. A behavioral questionnaire included an item about a perceived physical limitation that "affects your ability to participate in physical activity." The baseline dataset included other variables of interest with respect to potential differences in attitudes or motivation by ORCM status, e.g., the amount of weight loss desired or expected in the program, importance of health versus appearance as motivations for weight loss, and beliefs about the value of weight loss in preventing heart disease. However, these variables were dropped after preliminary analyses showed no association with ORCM, program participation, or program outcomes.

Program participation data were obtained from class attendance records. Weight, WC, blood pressure measurements, and fasting blood samples were repeated at or in conjunction with the postprogram follow-up visit. The average waiting time before the start of classes was 2 months. Participants were scheduled for a follow-up visit within 1 to 4 weeks after the end of the 10 weeks of classes. The time between the baseline and follow-up visit, therefore, ranged from 3 to 6 months.

Statistical Analyses

Data were entered from scannable forms and edited using range checks and logic algorithms, with reference to paper forms to clarify questionable values. All data were analyzed using Intercooled STATA, version 8.0 (17).

Medical history data were used to identify ORCMs by selecting conditions for which obesity is an established risk factor (18). ORCM status was based on preexisting diagnoses, which was feasible because each participant had a personal physician in the university health system and was in contact with that physician to obtain the medical evaluation form required for final study eligibility. The information on fasting glucose, lipids, and blood pressure, although subsequently compared for ORCM subgroups, was not used in the ORCM definitions, primarily because the study assessment protocols were not designed to diagnose these conditions independently of the personal physician's assessment. Participants whose medical history information indicated a diagnosis of high blood pressure, gallbladder disease, gout, or elevated uric acid, obstructive sleep apnea, breathing problems, stroke, angina, heart disease, arthritis, joint pain, diabetes, hypercholesterolemia, or high cholesterol were considered to have ORCMs. Participants with a history of high blood pressure, stroke, angina, heart disease, diabetes, hypercholesterolemia or high cholesterol were considered to have an obesity-related metabolic disease. All persons with a history of diabetes were assumed to have type 2 diabetes; neither the medical history form nor the personal physician's medical evaluation form specified type of diabetes.

BMI was calculated as weight in kilograms divided by height in meters squared. Categorical variables for obesity class were created according to current NIH guidelines (13), as follows: Class I obese, BMI between 30.0 and 34.9 kg/m²; Class II obese, BMI between 35.0 and 39.9 kg/m², and Class III obese, BMI 40.0 kg/m². Categorical variables for WC were created using the recommended sex-specific cut-off (13): men and women with WC 102 and 88 cm, respectively, were considered to have normal WC, whereas men and women with WC >102 and >88 cm, respectively, were considered to have high WC.

Descriptive statistics (mean SD and percentages) were used to describe baseline characteristics for the total sample and for subgroups with and without ORCMs. In bivariable analyses, the ² and Student's t test were used for categorical and continuous variables, respectively, to assess baseline differences by ORCM status. Multivariable logistic regression was used to adjust for potential confounders and thereby determine the independent association between baseline characteristics and the presence of ORCMs, i.e., with ORCM as a binary dependent variable. Model covariates included variables with an associated p < 0.25 in the preliminary bivariable analyses along with potential confounders and effect modifiers from the literature (see Tables 1 and 2 for variables tested) (19). Variables were retained in the final model if their associated p value was <0.05. Both age and WC were maintained as continuous variables after graphical assessment suggested that the assumption of linearity in the logit was met. The association between covariates, e.g., BMI and WC, was assessed. The fit of all logistic models was assessed with the Hosmer-Lemeshow test for goodness of fit. To compare the ability of BMI versus WC to predict the presence of ORCMs, the "roccomp" procedure in STATA was used to test the equality of the area under the curve (AUC) of receiver operator characteristic curves of separate logistic regression models based on each of these variables. Variables that are better at discriminating between those with and without ORCMs will have greater sensitivity and specificity over a wider range of cut-off values, and their AUCs will be larger and closer to 1 (the maximum possible AUC) in value.

Table 1. - Baseline characteristics of study participants overall and by the presence or absence of ORCMs.

Full table

Table 2. - Results of logistic regression showing the magnitude of the association between baseline characteristics and the odds of having ORCMs.

Full table

Results

Baseline Characteristics and ORCM Prevalence

Descriptive statistics for the total sample are shown in Table 1. Most participants were women, had education beyond high school, and were not married. On average, participants reported prior participation in a weight loss program at least once (mean, 1.4), but the percentage reporting such participation was only 58% (data not shown). BMI levels were nearly evenly distributed across the three obesity classes, with 34.6% of the participants in Class I, 32.1% in Class II, and 33.3% in Class III. Almost all (98.3%) participants had a high WC; the four participants with normal WC were all in the Class I obese category.

As shown in Table 1, mean baseline blood pressure, cholesterol, and glucose levels were in the desirable range. Figure 1 shows the prevalence of ORCMs, separately for each condition included in the definition and for having any one of the conditions. About three-quarters of participants had at least one of the ORCMs, with hypertension and osteoarthritis being the most common. About one-half (50.6%) of the participants reported at least one of the parameters for obesity-related metabolic disease, as defined in the Research Methods and Procedures section.

Figure 1.

Prevalence of ORCMs in the HELP study participants at enrollment. * Defined by self-report, confirmed by personal physician. OSA, obstructive sleep apnea and breathing problems were combined; CHD, coronary heart disease, which was a combination of those who reported a diagnosis of heart disease or angina.

Full figure and legend (85K)

Bivariable Associations with ORCM status

Table 1 also shows baseline characteristics according to the presence or absence of ORCMs. As shown, participants with ORCMs were significantly older, less likely to have education beyond high school, and more likely to report a physical limitation that would affect their ability to do physical activity compared with those without ORCMs. Blood pressure was higher for those with versus without ORCM. Among participants who provided baseline blood samples (n = 177), mean fasting plasma glucose and total cholesterol were also higher in the group with versus without ORCM. Seven participants with no history of diabetes had elevated fasting plasma glucose (defined as 126 mg/100 mL), of whom four were in the subgroup with no ORCMs (data not shown). There was no significant difference by ORCM status in the proportion of people who provided or did not provide a fasting blood sample at baseline (data not shown).

The average WC was significantly higher for the group with ORCMs than the group without ORCMs. The differences in weight and BMI between those with and without ORCMs were not statistically significant (p = 0.064 and 0.095, respectively). The proportion with ORCMs did not differ by obesity class (68.3%, 73.6%, and 79.6% for Classes I, II, and III, respectively, ²; p = 0.255; data not shown).

Multivariable Analyses on ORCM Status

Table 2 presents results of multivariable logistic regression models, expressed as odds ratios (ORs) and 95% confidence intervals, before and after simultaneous adjustment for the other row variables, for the independent variables of primary interest from the bivariable analyses. Participant age, WC, educational status, and the perception of a physical limitation were independently associated with having ORCMs in this study sample. A 10-year increase in the age of the participant was associated with 2.6 times the odds of ORCMs in the study sample (OR = 1.10 for 1 year of age; p < 0.001). The participants who reported beyond a high school education showed an average 56% decrease in the odds of having ORCMs (OR = 0.44; p = 0.034). A 1-cm increase in WC was associated with a 5% increase in the odds of having ORCMs for this study sample (OR = 1.05; p < 0.001), which suggests that an increase of 10 cm in WC would be associated with a 63% increase in the odds of ORCMs.

BMI and WC were highly correlated (r = 0.778, p < 0.001). The association between WC and ORCMs persisted even in the adjusted model that included BMI (OR = 1.06; p = 0.039). In contrast, although a one-unit increase in BMI was associated with an average 6% increase in the odds of having ORCMs, this association did not reach statistical significance (OR = 1.06; p = 0.069), and this association was not present when WC was adjusted for in the final model (OR = 0.95; p = 0.453). The AUC based on WC alone (AUC = 0.644) was significantly greater than the AUC for the model based on BMI alone (AUC = 0.578; p = 0.006). In all models tested, there was no statistical interaction between WC and BMI; the magnitude of the association between WC and the presence of ORCMs was similar for all three classes of obesity in the final model listed in Table 2 (OR = 1.08, 1.06, and 1.06 for Classes I, II, and III participants, respectively, data not shown). The final model in Table 2 predicts that a 40-year-old participant with no perceived physical limitations and a WC of 103 cm would have a 60.1% probability of having an ORCM. The predicted probability of ORCMs, in the same hypothetical participant, increases to 68.1% and 73.8% if the WC increases to 112 and 120 cm, respectively (data not shown). Participants who perceived a physical limitation that would affect their ability to participate in physical activity were 16 times more likely to also report ORCMs (OR = 16.0; p = 0.007). When these analyses were repeated using the binary variable describing the presence of an obesity-related metabolic disease as the dependent variable (as defined in the Research Methods and Procedures section), the magnitude of the association between the perception of a physical limitation and the presence of an obesity-related metabolic disease was much lower (OR = 2.23; p = 0.036).

Program Participation and Outcomes

Of the 237 participants who enrolled in the program, 77% (N = 183) attended at least one class. Among these 183 participants, the mean attendance was 5.8 of the 10 classes. Fifty-seven percent of those initially enrolled (N = 134) attended the postprogram data collection visit (Phase I completers). Mean SD weight change among Phase I completers was -1.7 3.7 kg, with 40.3% (N = 54) losing at least 2.25 kg (5 lb) and 16.4% (N = 22) losing 4.5 kg (10 lb). The mean SD change in WC among Phase I completers was -2.4 6.4 cm. None of these variables differed for the subgroups with versus without ORCMs: attending any classes (79% vs. 71%, respectively; p = 0.18), average attendance (5.8 vs. 5.7 classes, respectively; p = 0.89), being a Phase I completer (56 vs. 58%, respectively; p = 0.78), mean weight change (-1.6 and -1.7 kg, respectively; p = 0.88), percent losing = 2.25 kg (40.8% vs. 38.9%, respectively; p = 0.84), percent losing 4.5 kg (17.3% vs. 13.9%, respectively; p = 0.632), or change in WC (-2.1 vs. -3.3 cm, respectively; p = 0.38). An association of having perceived physical limitations with weight change was suggested (weight loss of 1.9 vs. 0.4 kg, respectively, in those without and with limitations) but did not attain statistical significance (p = 0.069). Perceived physical limitations were not related to change in WC (p = 0.451).

Discussion

The individuals in this predominantly female sample may be generally similar to obese African Americans likely to enroll in a university health system, outpatient weight loss program in an urban area such as Philadelphia. The objectives of this analysis were to determine the prevalence of ORCMs among this type of clinical population and to identify possible differences by ORCM status. We were interested in differences at baseline as well in program participation or postprogram weight change, thus obtaining insights about possible indications for differential tailoring of treatment approaches.

The high prevalence of ORCMs and the ranking of specific conditions in this sample were consistent with expectation for African Americans in the Class I to Class III obesity range and recruited in a clinical situation in which ORCMs were likely to have been diagnosed. The program did, however, attract a substantial number of individuals (24% of those enrolled) who did not have specific diagnoses associated with their weight levels. Ascertainment of ORCMs—the key variable of interest—was probably relatively complete in this sample. The design of the study from which these data were taken required that all participants have a personal physician in the health system in which the program was offered and also required that the personal physician provide a medical evaluation form that specifically prompted for a number of conditions, including the specific ORCM used in these analyses. This provided an opportunity to reconcile participant and physician reports to identify ORCM history. The significantly higher clinical values in the ORCM groups for blood pressure, fasting glucose, and cholesterol supported the validity of the ORCM classification, nothwithstanding pharmacological management of these conditions.

Although age did not completely explain the presence of ORCMs, the subgroup with ORCMs were somewhat older and were more likely to perceive limitations on physical activity. These findings were expected given the increased occurrence of ORCMs with aging and with increasing duration of obesity (20) and the effects of both obesity and ORCMs on physical functioning (13,21). The bulk of this association with physical limitations was apparently driven by the nonmetabolic diseases included in the definition of ORCMs, e.g., gout and osteoarthritis, that are known to cause physical disability (21). However, the significant residual association when the analysis was limited only to metabolic diseases is consistent with the possibility of common etiologic pathways for some musculoskeletal and metabolic conditions (22,23). The subgroup with ORCMs also had less education, consistent with the well-established relationship between lower socioeconomic position and poorer health status (24).

As noted in the Research Methods and Procedures section, our preliminary analyses did not identify attitudinal and behavioral differences between the ORCM and no ORCM groups. Such differences might exist in the general population of obese African Americans but not among those who are self-selected on motivation for weight loss. This finding is probably not caused by differential eligibility of those with ORCMs based on motivation, e.g., to differential exclusion of a subsample of those with ORCM who were more or less more motivated for weight loss than those included. Only 8% (11 of 136) of the overall eligibility exclusions were based on medical history. Consistent with this lack of apparent differences in motivation, we also did not identify differences in program participation or postprogram weight loss by ORCM status. Analysis of longer-term results of HELP (i.e., after an additional 8 to 18 months of counseling and follow-up) also did not support an association of ORCMs with weight loss outcomes (14). This implies that, in this population, motivations that are unrelated to health may have been as important as those that were health related, or weight loss program retention and outcomes in African Americans may generally reflect feasibility and logistical factors more than motivations (25).

Taken together, these results would seem to preclude any need, in this type of population and under these circumstances, for differential tailoring of treatment approaches for those with versus without ORCMs, even given the identified differences in age and educational attainment. However, the difference in perceived activity limitations does warrant further study, given the suggested association of perceived limitations with weight loss. The HELP program was unable to provide an adequate focus on increasing physical activity, which may have partly explained the relatively modest weight losses observed (14). Physical activity is an important component of lifestyle weight loss treatment (13,26), and tailored approaches may be needed for individuals who have or perceive activity limitations.

Our finding of the association of WC with ORCMs was unexpected. WC correlates highly with abdominal or visceral obesity (27) and has been shown to be an independent marker of CVD risk in both the general population and in the African-American population (28,29,30,31), even though the amount of abdominal fat for a given BMI may be less in African Americans than whites (31). In identifying individuals who are candidates for obesity treatment, assessment of WC in addition to BMI is recommended for individuals with BMI levels 35 kg/m², because WC of those with BMI >35 kg/m² usually exceeds the cut-points used to designate high-risk WC levels (13). Although this applied to our sample, i.e., the four people with WC below the cut-points had BMI levels <35 kg/m², our analyses of the association of BMI and WC with ORCMs across the BMI 30 to 50 kg/m² range suggest that, in this population, WC was a clear marker of increasing ORCM risk above a BMI of 35 kg/m²—more than BMI itself. This finding is particularly noteworthy because the debate about WC versus BMI as predictors of obesity-related risk has not addressed the question to the upper part of the BMI distribution (32,33,34,35). Much of the literature in this area is focused on evaluating which anthropometric marker of obesity better predicts cardiovascular health risk compared with normal weight subjects. For example, a recent analysis of NHANES data on the WC versus BMI question excluded individuals with BMI >35 kg/m² with the rationale that all individuals with Classes II to III obesity already have high WC based on the National Heart, Blood, and Lung Institute clinical cut-points (32)

From the perspective of patients who have already been deemed "high risk" and are interested in obesity treatment, the anthropometric marker of interest is the one that better predicts their risk of developing further complications from their obesity. These data suggest that WC could be such a marker, although we cannot assess the ability to predict risk from these cross-sectional data. Nor can we rule out a diagnostic bias that would render African-American patients with higher WC levels more likely than others in the same BMI range to be screened for ORCMs, although this seems unlikely.

In summary, among African Americans in the Class I to III obese range who enrolled in a clinical weight loss program, only age, education, the perception of a physical limitation, and WC were independently associated with the presence of ORCMs at baseline, and ORCM status was not associated with retention in the program or with weight loss after the program. The only possible lead identified regarding the need to tailor programs to those with ORCMs relates to the association of physical limitations, which were more common among those with ORCMs, with less weight loss. This association did not attain statistical significance but, on a conceptual basis, would seem to warrant further study. The relatively strong and highly significant association between WC and the presence of ORCMs in this sample suggests that WC may be a more important marker of ORCMs at this higher end of the BMI distribution than has been previously recognized, consistent with the general argument that more emphasis should be placed on WC in the identification and evaluation of obesity-related health risks. WC may be useful for identifying a relatively higher risk subset within obese African-American populations from the perspective of concurrent or, perhaps, imminent comorbidities.

Notes

¹ Nonstandard abbreviations: CVD, cardiovascular disease; ORCM, obesity-related comorbidity; NHANES, National Health and Nutrition Examination Survey; HELP, Healthy Eating and Lifestyle Program; WC, waist circumference; AUC, area under the curve; OR, odds ratio.

References

1. Flegal, K. M., Carroll, M. D., Kuczmarski, R. J., et al. (1998) Overweight and obesity in the United States: prevalence and trends, 1960–1994. Int J Obes Relat Metab Disord. 22: 39–47. | Article | PubMed | ChemPort |
2. Flegal, K. M., Carroll, M. D., Ogden, C. L., et al. (2002) Prevalence and trends in obesity among US adults, 1999–2000. JAMA 288: 1723–1727. | Article | PubMed | ISI |
3. Pi-Sunyer, F. X. (2002) The medical risks of obesity. Obes Surg 12: 6S–11S.
4. Wolf, A. M., Colditz, G. A. (1998) Current estimates of the economic cost of obesity in the United States. Obes Res. 6: 97–106. | PubMed | ISI | ChemPort |
5. Sims, E. A. (2001) Are there persons who are obese, but metabolically healthy? Metabolism 50: 1499–1504. | Article | PubMed | ISI |
6. Vague, J. (1996) Sexual differentiation. A determinant factor of the forms of obesity. Obes Res. 4: 201–203.
7. Vague, J. (1996) The degree of masculine differentiation of obesities: a factor determining predisposition to diabetes, atherosclerosis, gout, and uric calculous disease. Obes Res. 4: 204–212. | PubMed | ISI | ChemPort |
8. Brochu, M., Tchernof, A., Dionne, I. J., et al. (2001) What are the physical characteristics associated with a normal metabolic profile despite a high level of obesity in postmenopausal women? J Clin Endocrinol Metab. 86: 1020–1025. | Article | PubMed | ISI | ChemPort |
9. Quesenberry, C. P., Jr, Caan, B., Jacobson A. (1998) Obesity, health services use, and health care costs among members of a health maintenance organization. Arch Intern Med. 158: 466–472. | Article | PubMed |
10. Sanchez, A. M., Reed, D. R., Price, R. A. (2000) Reduced mortality associated with body mass index (BMI) in African Americans relative to Caucasians. Ethn Dis. 10: 24–30.
11. Stevens, J. (2000) Obesity and mortality in African Americans. Nutr Rev. 58: 346–353. | PubMed |
12. Kumanyika, S. K. (2002) Obesity treatment in minorities Wadden, TA Stunkard, AJ eds. Handbook of Obesity Treatment 416–446. 3rd ed., Guilford Publications New York.
13. National Institutes of Health (1998) Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: the evidence report. Obes Res. 6: S51–S210.
14. Kumanyika, S. K., Shults, J., Fassbender, J., et al. Outpatient weight management in African Americans: the Healthy Eating and Lifestyle Program (HELP) Study. Prev Med. (in press).
15. National Center for Health Statistics (1996) NHANES III Anthropometric Procedures Video US Government Printing Office Washington, DC.
16. Perloff, DGC, Flack, J., Frohlich, E. D., et al. (1993) Human blood pressure determination by sphygmomanometry. Circulation 88: 2460–2470. | PubMed | ISI | ChemPort |
17. StataCorp (2003) Stata Statistical Software Release 8.0 Stata Corp College Station, TX.
18. Pi-Sunyer, F. X. (1999) Comorbidities of overweight and obesity: current evidence and research issues. Med Sci Sports Exerc. 31: S602–S608. | Article | PubMed | ChemPort |
19. Hosmer, D. W., Lemeshow, S. (2000) Applied Logistic Regression 2nd ed John Wiley & Sons New York.
20. Janssen, I., Katzmarzyk, P. T., Ross, R. (2004) Duration of overweight and metabolic health risk in American men and women. Ann Epidemiol. 14: 585–591. | Article | PubMed |
21. Larsson, U. E., Mattsson, E. (2001) Perceived disability and observed functional limitations in obese women. Int J Obes Relat Metab Disord. 25: 1705–1712. | Article | PubMed |
22. Anderson, J. J., Felson, D. T. (1988) Factors associated with osteoarthritis of the knee in the first National Health and Nutrition Examination Survey (NHANES I). Am J Epidemiol. 128: 179–189. | PubMed | ISI | ChemPort |
23. Guralnik, J. M., LaCroix, A. Z., Abbott, R. D., et al. (1993) Maintaining mobility in late life. I. Demographic characteristics and chronic conditions. Am J Epidemiol. 137: 845–857.
24. Berkman, L. F., Kawachi, I. (2000) Social Epidemiology Oxford University Press New York.
25. Bronner, Y., Boyington, J. E. (2002) Developing weight loss interventions for African-American women: elements of successful models. J Natl Med Assoc. 94: 224–235. | PubMed |
26. McTigue, K. M., Harris, R., Hemphill, B., et al. (2003) Screening and interventions for obesity in adults: summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med. 139: 933–949. | PubMed | ISI |
27. Pouliot, M. C., Despres, J. P., Lemieux, S., et al. (1994) Waist circumference and abdominal sagittal diameter: best simple anthropometric indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol. 73: 460–468. | Article | PubMed | ISI | ChemPort |
28. Okosun, I. S., Cooper, R. S., Rotimi, C. N., Osotimehin, B., Forrester, T. (1998) Association of waist circumference with risk of hypertension and type 2 diabetes in Nigerians, Jamaicans, and African-Americans. Diabetes Care 21: 1836–1842. | PubMed |
29. Okosun, I. S., Chandra, KMD, Choi, S., et al. (2001) Hypertension and type 2 diabetes comorbidity in adults in the United States: risk of overall and regional adiposity. Obes Res. 9: 1–9. | PubMed | ISI | ChemPort |
30. Okosun, I. S., Prewitt, T. E., Cooper, R. S. (1999) Abdominal obesity in the United States: prevalence and attributable risk of hypertension. J Hum Hypertens 3: 425–430.
31. Okosun, I. S., Rotimi, C. N., Forrester, T. E. (2000) Predictive values of abdominal obesity cut-off points for hypertension in blacks from West Africa and Caribbean island nations. Int J Obes Relat Metab Disord. 24: 180–186. | Article | PubMed | ChemPort |
32. Janssen, I., Katzmarzyk, P. T., Ross, R. (2004) Waist Circumference and not body mass index explains obesity related health risk. Am J Clin Nutr. 79: 379–384. | PubMed | ISI | ChemPort |
33. Zhu, S., Wang, Z., Heshka, S., et al. (2002) Waist circumference and obesity-associated risk factors among whites in the third National Health and Nutrition Examination Survey: clinical action thresholds. Am J Clin Nutr. 76: 743–749. | PubMed | ISI | ChemPort |
34. VanItallie, T. B. (1998) Waist circumference: a useful index in clinical care and health promotion. Nutr Rev. 56: 300–302. | PubMed |
35. Lean, MEJ, Han, T. S., Morrison, C. E. (1995) Waist circumference as a measure for indicating need for weight management. BMJ 311: 158–163. | PubMed | ChemPort |

Acknowledgments

These analyses were supported, in part, by American Heart Association Grant 9970068N and by Project EXPORT Grant P60 MD000209 from the National Center for Minority Health and Health Disparities of the NIH to Dr. Kumanyika as principal investigator. Partial support was also provided by General Clinical Research Center Grant M01 RR00040 from the NIH, National Center for Research Resources. Dr. Hope was supported by the Kynett-FOCUS fellowship at the University of Pennsylvania.