Relationship between Waist Circumference, Body Mass Index, and Medical Care Costs

Authors

  • Marc-Andre Cornier M.D.,

    Corresponding author
    1. Department of Medicine, Denver Health Medical Center and Division of Endocrinology, Metabolism, and Diabetes, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado
    Search for more papers by this author
  • Charles W. Tate,

    1. Department of Medicine, Denver Health Medical Center and Division of Endocrinology, Metabolism, and Diabetes, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado
    Search for more papers by this author
  • Gary K. Grunwald,

    1. Department of Medicine, Denver Health Medical Center and Division of Endocrinology, Metabolism, and Diabetes, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado
    Search for more papers by this author
  • Daniel H. Bessesen

    1. Department of Medicine, Denver Health Medical Center and Division of Endocrinology, Metabolism, and Diabetes, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado
    Search for more papers by this author

Denver Health Medical Center, 777 Bannock St., Mail Code 4000, Denver, CO 80204-4507. E-mail: mcornier@dhha.org

Abstract

Objective: Excessive visceral adiposity as measured by anthropomorphic measures may be more closely associated with adverse health consequences than body weight or body mass index (BMI), the more commonly obtained clinical measures. Waist circumference (WC) provides information about regional adiposity and may correlate with health care costs better than body weight or BMI.

Research Methods and Procedures: A total of 424 men (37%) and women (63%) were identified as they were seen in outpatient medical clinics at Denver Health, an integrated health care system serving a largely indigent population. Height, weight, and WC were measured by one examiner. Information on outpatient, laboratory, pharmacy, inpatient, and total charges attributable to each subject for the preceding year were obtained from computerized databases. Data on health care charges were divided into quartiles based on WC and BMI.

Results: Total annual health care charges were significantly greater in the highest WC quartile (WC < 83.3 cm: $6062 ± $784; 83.3 to 93.5 cm: $5968 ± $812; 93.7 to 103.5 cm: $6369 ± $1015; >103.5 cm: $8699 ± $1092; p = 0.047). Those with a WC >103.5 cm generated 85% more inpatient charges than the group with a WC <83.3 cm. Although there was a positive trend, BMI was not found to significantly correlate with total health care charges in this population sample.

Discussion: These results suggest that abdominal adiposity as assessed by WC is associated with increased total health care charges and may be a better predictor of health care charges than the more widely used BMI.

Introduction

The prevalence of obesity has risen dramatically in the United States over the last 20 years (1). In some ethnic and age subgroups, the prevalence now exceeds 60%. This increase has occurred despite the efforts of health care providers and consumers alike to improve the health-related behaviors of the population. This epidemic is all the more concerning given the clear association between excess adiposity and adverse health consequences. Obesity has been clearly shown to be associated with an increased risk of type 2 diabetes, hypertension, coronary artery disease, degenerative arthritis, gall bladder disease, a number of different cancers, and mortality (2,3,4). In addition, it has been estimated that more than 50 billion dollars, or 5.7% of total health care expenditure in the United States, is associated with the management of obesity-associated morbidity (5,6).

A challenge in assessing the impact of obesity on health and health care costs is how to measure obesity itself. Body mass index (BMI) has been widely accepted and used in most epidemiological studies as a surrogate measure of total body fat. This formula takes into account not only body weight but also height. BMI, however, does not take into consideration the distribution of body fat, which is believed to have significant health implications. Abdominal adiposity has been independently associated with coronary artery disease, cerebrovascular disease, insulin resistance, and diabetes mellitus (7,8,9,10,11,12,13). Waist circumference (WC) is an easy measurement that is highly correlated with abdominal or visceral adiposity (14) and with markers of cardiovascular disease risk such as blood pressure, insulin resistance/diabetes, low high-density lipoprotein cholesterol, and hypertriglyceridemia (9,15,16,17).

Currently, most health plans do not cover treatment for obesity in part because of a paucity of data directly documenting the excess costs of caring for obese individuals in specific health care settings. A few studies have shown a direct correlation between BMI and annual total health care costs and missed time from work (5,6,18,19,20,21,22,23). Greater WC has been shown to be associated with an excess burden of ill health but has never been reported to be associated with increased total health care costs (24). Because WC may be a more accurate measure of adiposity associated with health risk, we undertook this study to quantify the association between WC and health care costs and to assess the relationship between WC and BMI. We hypothesized that excessive abdominal adiposity as assessed by WC would be associated with increased total health care costs, and that WC would be a better predictor of health care costs than the more widely used BMI.

Research Methods and Procedures

Adults were sequentially recruited from outpatient medical clinics at Denver Health Medical Center, an integrated health care system serving a largely indigent population in Denver, Colorado, during the period from July to September 1998. The sample population was taken directly from general internal medicine (68%), endocrinology (21%), and cardiology (11%) clinics. Demographic and anthropometric measurements were obtained by a single examiner at the time the patients were being checked in to the clinic. Body weight was measured using a standard scale. Height was measured using a stadiometer. WC was measured with a tape measure at the narrowest part of the torso between the lowest rib and the level of the iliac crests at end expiration while standing (14). BMI was calculated as weight in kilograms divided by height in square meters.

Retrospective information on the outpatient, laboratory, pharmacy, and inpatient charges for each patient for the period 7/1/97 to 6/30/98 (the fiscal year before the anthropometric measurements) was obtained from computerized databases using the patients’ medical record number. These were summed to determine the total health care charges for that patient. All charges incurred by patients at this medical center are captured by the computerized databases, and because these patients are medically indigent, they receive most if not all of their medical care at this medical center, insuring close to complete capture of health care charges for each patient.

The primary endpoints were the total annual charges of medical care according to WC and BMI quartile. Secondary endpoints included annual inpatient and outpatient charges of medical care according to WC and BMI quartile. The data are expressed as mean annual health care charges per quartile ± SEM. World Health Organization criteria were used for the BMI quartile cut-offs. These cut-offs closely approximated the true quartile boundaries. Correlational analysis was performed on WC, BMI, and cost data, but it was felt that analyzing and presenting the data in quartiles was more clinically informative. Statistical analysis was performed with SigmaStat statistical software (Jandel Scientific, San Rafael, CA). One-way ANOVA was used to compare charges across quartiles with a pairwise multiple comparison procedure (Dunn's Method) to isolate the group or groups that differ from the others. Linear regression analysis was also performed, adjusting for age, gender, race, and smoking status. χ2 analysis was performed to compare the percent of patients in each quartile incurring inpatient charges. Correlation analyses were done using the Pearson Product Correlation. Because of skewness of the cost distributions, we repeated correlation and ANOVA analyses using logarithm of cost, which was approximately normally distributed. Results did not change substantially, so we have reported results in the original dollar units because these are more easily interpretable and because total charges for groups can be estimated from mean per-person charges. In addition, we reanalyzed the data after removing the significant outliers of total health care charges to see if this might strengthen the associations. This in fact worsened the associations between WC/BMI and total health care charges. Analyses on the original dollar scale remain valid because sample sizes are large enough for the central limit theorem to hold.

Results

The characteristics of the patients examined are depicted in Table 1. Women comprised 64% of the sampled population. The mean age was 52 ± 14 years (range, 18 to 84 years). The ethnic distribution of the sample population was representative of the overall population served in this health care system. Most patients were nonsmokers (64%). The mean WC was 94.5 ± 16.3 cm (range, 60.2 to 162.5 cm). The mean BMI was in the obese range at 30 ± 8 kg/m2 (range, 16 to 65 kg/m2). As seen in Figure 1, BMI and WC were highly correlated (r = 0.86, p < 0.0001) with 74% of the variance in BMI accounted for by the WC.

Table 1.  Subject characteristics and demographics
CharacteristicsValue
Total subjects, n424
 Male, n (%)154 (36)
 Female, n (%)270 (64%)
Nonsmokers, n (%)292 (69%)
Smokers, n (%)131 (31%)
Age, years ± SD (min to max)52 ± 14 (18 to 84)
Ethnicity, % 
 Caucasian30
 African American27
 Hispanic37
 Other6
Waist circumference, cm ± SD (min to max)94.5 ± 16.3 (60.2 to 162.5)
Body mass index (kg/m2)30 ± 8 (16 to 65)
Figure 1.

Correlation between waist circumference (WC; centimeters) and body mass index (BMI; kilograms per square meter) for 424 patients (r = 0.86, p < 0.0001).

Figure 2 shows the total annual health care charges in relation to the WC. WC was an independent predictor of total charges, although the correlation was weak (r = 0.10; p = 0.04). Because of the extreme skewness of the distribution, we took the log of the charges and found similar results, although with a p value that was slightly smaller. Figure 3 shows the mean total annual health care charges according to WC quartiles (<83.3 cm, 83.3 to 93.5 cm, 93.7 to 103.5 cm, and >103.5 cm). The mean WC of each quartile was 75.1 ± 0.6, 89.1 ± 0.3, 98.3 ± 0.3, and 115.8 ± 1.2 cm, respectively. The highest WC quartile had significantly greater mean total charges than the other quartiles ($6062 ± $784, $5968 ± $812, $6369 ± $1015, and $8699 ± $1092, respectively; p = 0.047) and represented 32% of the overall total charges.

Figure 2.

Correlation between total annual health care charges and waist circumference (WC; centimeters) (r = 0.10, p = 0.04).

Figure 3.

Mean total annual health care charges per waist circumference (WC) quartile. Total annual health care charges were significantly greater in the highest WC quartile (*p = 0.047 for WC > 103.5 cm compared with other quartiles).

As seen in Figure 4, mean annual outpatient charges did not differ between WC quartiles, whereas the mean annual inpatient charges were significantly greater in the highest WC quartile (p = 0.038). In fact, 29% of the patients in the highest WC quartile incurred inpatient charges compared with only 15% to 16% of those in each of the lower three quartiles (p = 0.028). Pharmacy and laboratory charges were not greater in either of the higher WC or BMI quartiles.

Figure 4.

Mean outpatient and inpatient annual health care charges per waist circumference (WC) quartile. Inpatient charges were significantly greater in the highest WC quartile (*p = 0.038 for inpatient charges of WC > 103.5 cm compared with inpatient charges of other quartiles). No differences were seen with outpatient charges.

BMI was not an independent predictor of either total annual health care charges or the log of total charges, nor did adding BMI to WC enhance the regression model. Figure 5 shows the mean total annual health care charges per BMI quartile. The mean total charges were greatest in the highest BMI quartile (>35 kg/m2), although this was not statistically significant (p = 0.08). As with WC, the highest BMI quartile was associated with greater inpatient charges but not with outpatient charges, although again this was not statistically significant (p = 0.07).

Figure 5.

Mean total annual health care charges per body mass index (BMI) quartile. No significant differences were seen in total health care charges across BMI quartiles; p = 0.08 for BMI > 35 kg/m2 compared with other quartiles.

Ethnicity, gender, and smoking status did not correlate with health care charges. Age was weakly correlated with total medical charges (r = 0.15, p = 0.003) and WC (r = 0.13, p = 0.009), and therefore may have been a confounder. However, when the data were adjusted for ethnicity, gender, smoking status, and age, no differences were found compared with the unadjusted data. In all cases above, similar results were obtained for analyses of the logarithm of cost.

Discussion

This study was performed to examine the relationship between health care charges and two different measures of obesity—WC and BMI. A number of conclusions can be drawn from this study. First, greater WC is associated with increased total health care charges. Second, greater BMI is also associated with increased total health care charges although not statistically significant. Third, the differences seen in total medical charges seem to be primarily because of greater inpatient charges in the highest WC category, and not from outpatient, laboratory or pharmacy charges.

Obesity as assessed by BMI or WC is clearly associated with significant morbidity and mortality (2,3,4,7,8), and the prevalence of obesity has risen dramatically and continues to rise in the United States and around the world (1). The management of obesity-associated morbidity is also associated with a tremendous cost to society. Unfortunately, most health care plans do not cover the evaluation or treatment of obesity. This is in part because of the lack of data documenting the excess costs of caring for obese individuals in addition to a perception that treatment is expensive and ineffective. A number of studies have shown that high BMI is associated with increased total health care costs, increased health services use, increased days lost from work, and increased disability (5,6,18,19,20,21,22,23). This study confirms these findings. However, in this study, total health care costs were better correlated with WC than with BMI. WC has been found to be associated with excess burden of ill health and impaired quality of life (24), but to the authors’ knowledge, no other studies have examined the association between WC and health care costs. In this study, the patients in the highest WC quartile (WC > 103.5 cm) had significantly greater total health care charges than those in the other quartiles. Total health care charges in the highest WC quartile were 44% greater than the total charges of the patients in the lowest WC quartile. In fact, WC was found to be an independent predictor of total health care charges. Adding BMI to the model did not enhance the prediction, nor was it an independent predictor of charges. The total health care charges were greater in the highest BMI quartile, but this was not statistically significant.

The greater total health care charges associated with increased WC and BMI were found to be primarily because of increased inpatient care charges. In fact, patients in the highest quartile of WC incurred almost twice the inpatient charges as those in the lowest quartile. Outpatient, laboratory, and pharmacy charges were not found to be greater in the highest WC or BMI quartiles. Other studies have found that high BMI is associated with greater outpatient costs (18,20). The present study also found no association between gender, ethnicity, or smoking status and health care charges, although this study was not designed to detect such differences.

Is WC a better surrogate marker for obesity than BMI? The answer is certainly a matter of debate. Increasingly experts in the field of obesity are using WC as the preferred measure or marker of obesity. It has been shown that WC is well-correlated with abdominal and visceral adiposity (14). Abdominal adiposity and increased WC, especially >100 cm, have been shown to be independently and highly correlated with risks for cardiovascular disease (7,8,9,10,11,12,13,15,16,17). In addition, WC is a simple measurement to obtain, is not affected by lean body mass or volume status, and provides the health provider an opportunity to discuss the relation between fat distribution and health. Whereas some patients may feel that having a WC measurement is somewhat embarrassing, most are comfortable with this measurement, especially after they are educated about the importance of this measurement.

A number of limitations to this study need to be discussed. First, the total sample population was small, limiting the power of the study. This may explain why high BMI was not statistically associated with increased health care charges, why no differences were seen in outpatient, laboratory, and pharmacy charges, and why age was only weakly associated with health care charges. Second, the sample population consisted of more women than men, which may have biased the results. Third, the sample population had a greater BMI and WC than the general population. This is likely explained by the high percentage of minorities represented in the sample population. This may have resulted in an underestimation of the impact of obesity on health care charges. Fourth, only direct medical care charges were evaluated in this study. Indirect costs such as missed time from work were not measured. Finally, the sample population was a random sampling of the Denver Health system, an ethnically diverse population of generally low socioeconomic status, and not of the general community. The results may therefore not apply to the general community.

In conclusion, excessive abdominal adiposity as assessed by WC is associated with increased total health care charges, especially in the charges of inpatient care. WC may also be a better predictor of health care costs than the more widely used BMI. These findings offer more support that obesity, specifically abdominal adiposity, is associated with significant costs to the health care system, and therefore more effort should be placed on the evaluation and treatment of this disease.

Acknowledgments

Support for this work was provided by the Clinical Nutrition Research Unit DK48520, the National Institute of Diabetes Digestive and Kidney Diseases (NIDDK) DK47311 and DK02935, and the National Center for Research Resources (NCRR) RR00192.

Ancillary