Introduction

An estimated 97 million adults in the United States are obese or overweight, and the prevalence of this important public health problem has increased dramatically in recent years (1,2,3,4,5). According to the latest National Health and Nutrition Examination Survey by the National Center for Health Statistics, Centers for Disease Control and Prevention, the prevalence of obesity is 30.5% among men and women, compared with 22.9% reported in the previous National Health and Nutrition Examination Survey conducted from 1988 to 1994. The prevalence of overweight also rose during this period, from 55.9% to 64.5% (3).

Excess weight increases the risk for many disorders associated with high morbidity and mortality such as type 2 diabetes, hypertension, coronary heart disease (CHD),¹ high blood cholesterol, gall bladder disease, osteoarthritis, respiratory problems, and certain malignancies (1,2,5,6,7,8). Obesity seems to lessen life expectancy markedly, particularly among individuals in younger age groups; each year, an estimated 300,000 U.S. adults die of causes associated with obesity (9,10).

The relationship between obesity and depression has also been explored (11,12,13,14,15). Although there does not seem to be a simple association between these two increasingly prevalent disorders (12,13), the findings of a number of recent studies have suggested that obesity is associated with an increased risk for depression (11,14,15). There has been sufficient disparity of results, however, to warrant further research in this area.

With obesity being related to a variety of diseases, it follows that there are associated economic outcomes that can be quantified. The costs to society are both direct and indirect (including increased health care expenses and loss of productivity), as well as personal costs (such as job discrimination and higher living expenses) (16,17,18,19,20). The total cost for obesity in 1995 dollars has been estimated to be $99.2 billion. Approximately $51.6 billion of this amount is attributable to direct medical costs, which represents 5.7% of our national health expenditure (18). Most of the studies examining obesity and related health care costs have used group data, applying estimates of population-attributable risks to estimates of U.S. total costs of care for each disease related to obesity. While these data are valuable, it is also important to quantify the association of obesity and health resource use at the patient level, where potential confounding variables, such as sociodemographics and comorbidities, can be controlled. Two recent studies sought to do this using patient data from a large health maintenance organization (Kaiser Permanente) (21,22).

In the first study, Quesenberry et al. (21) found a significant association between self-reported body weight and annual costs of total outpatient and inpatient health services use for a 12-month period of time, approximating the survey period. Obese patients, however, reported a greater number of comorbid conditions, and statistically controlling for diabetes, hypertension, and CHD essentially eliminated the age- and sex-adjusted association between obesity and total annual costs. The authors concluded that the higher incidence of these diseases drives increased costs in the obese group and that adjusting for comorbidities when measuring the effect of obesity on costs leads to an underestimation of the association.

Another study, by Thompson et al. (22), investigated the relationship between obesity and future health care costs. Here, patients completing a health survey had their costs for all outpatient services, inpatient care, and prescription drugs tallied for the subsequent 9 years, irrespective of any weight change during the study period. Obese patients had 36% higher annual health care costs than nonobese individuals. It is important to note that this study excluded patients who smoked cigarettes or had a history of CHD, stroke, human immunodeficiency virus, or cancer. The authors did not control for type 2 diabetes, hypertension, or hypercholesterolemia in their analyses, stating that these diseases are critical links in the pathway by which obesity leads to increased risk for important diseases.

The purpose of this study was to examine differences in the use of health care services and in the associated charges for 1 year of care for obese and nonobese patients. Obesity was determined by measured values (rather than self-reported), because obesity estimates based on self-reported data tend to be lower than those based on measured data (3,5). We sought to contribute to the findings of previous studies investigating the association between obesity and medical costs by controlling for patient health status and depression, in addition to sociodemographic characteristics, because these have been shown to significantly impact health resource use (23,24,25). By statistically adjusting for health status and depression, measured with instruments widely employed in health services research, we attempted to eliminate potential sources of bias and confounding, while avoiding "overadjustment" by controlling for obesity associated comorbidities.

We hypothesized that obesity is associated with decreased physical health status and a higher incidence of depression. We further expected obesity to be related to higher medical resource use. Last, we hypothesized that, even after controlling for sociodemographic patient variables, general health status, and depression, obese individuals use more medical care services and have higher associated charges than patients without obesity.

Research Methods and Procedures

Study Design

This study was part of a larger project examining physician–patient interaction, physician practice styles, and associated patient outcomes. A total of 509 new patients agreeing to participate comprised the original study population. Subjects were randomly assigned to be cared for by 105 physicians (second- and third-year family practice and general internal medicine residents) at the University of California, Davis Medical Center Primary Care Center. Physicians each saw an average of 4.8 4.6 patients (SD). Patients were interviewed before their initial visit by their primary care provider. Data collected included sociodemographic information, self-reported health status using the Medical Outcomes Study Short Form-36 (MOS SF-36), and evaluation for depression using the Beck Depression Index (BDI). Height and weight measurements were also taken to calculate BMI. To avoid influencing physicians' behavior, physicians were not provided with information from the previsit interview. The use of medical services for the study period was determined by medical record and associated billing record review. The study methods were approved by our institutional Human Subjects Committee.

Measures

BMI is calculated as weight in kilograms divided by the square of height in meters. According to the NIH Guidelines, BMI is the recommended method for measuring obesity in clinical settings. Patients who have a BMI of 18.5 to 24.9 kg/m² are considered normal; those with BMI of 25 to 29.9 kg/m² are overweight; and those having a BMI 30 kg/m² are obese. (1)

The MOS SF-36 is a reliable and valid 36-item questionnaire made up of eight scales: general health, physical function, physical role, mental role, social function, pain, energy, and mental health. Scales are scored so that higher scores reflect better health status; common chronic medical conditions have a unique negative impact on scores (26,.27). Some of the chronic illnesses that have been profiled with the MOS SF-36 include comorbid conditions associated with obesity such as diabetes, hypertension, CHD, arthritis, and lung problems (27). Summary measures can describe a physical component score and a mental component score (28,29). The physical component score was used in this study to measure physical health status. Thus, rather than focusing on the specific diseases associated with increased morbidity, their effect on the individual patient's health status was measured and controlled.

The BDI is a reliable and valid instrument used to measure depressive symptoms (30,31). The abbreviated version includes 13 items weighted and summed to produce a total score (31). A score between 9 and 15 indicates moderate depression, and a score of 16 indicates severe depression. The BDI is used widely for screening and for assessing treatment efficacy (32). In this study, a BDI of 16 was used to identify those patients with severe depression.

Medical center resource use and charges for the 1 year of care were determined by review of the comprehensive medical record. Two physician reviewers noted the number of primary care visits, specialty clinic visits, emergency department visits, hospitalizations, and laboratory, diagnostic, and radiological tests (diagnostic services). The medical record identifies and separates patient information according to these classifications. Medical charges for all these services were obtained from the institutional central billing unit. Charges, used as a proxy for medical costs, were assigned to one of five categories: primary care clinics, specialty care clinics, emergency departments, hospitalizations (including outpatient surgery admissions), and diagnostic services. In addition, yearlong totals for each of the five types of charges were calculated for each patient. Patients were asked to submit quarterly reports by means of prepaid postcards for any medical care obtained elsewhere. Out-of-system use by participants was negligible and subsequently excluded from the analyses.

Results

A complete data set was available for 506 study patients. BMI scores ranged from 15.71 to 74.95 kg/m². The mean and median BMI for this group were 29.78 and 27.99 kg/m², respectively, which shows that the study group's average was in the overweight category.

A total of 205 patients (40.32%) were considered to be obese, with BMI scores 30 kg/m². Sociodemographic and health variables for this obese group of patients were compared with those for nonobese study participants with two-tailed Student's t tests for continuous variables and Pearson ² for categorical variables (Table 1). Obese patients were significantly more likely to be women (p = 0.0101) and had a higher mean age (p = 0.0280), lower mean education (p = 0.0278), and lower self-reported physical health status as measured by the MOS SF-36 (p = 0.0230). In addition, the mean BDI score for obese patients was significantly higher compared with nonobese patients (p = 0.0196).

Table 1. - Sociodemographic and health variables for obese (BMI 30 kg/m²) and nonobese (BMI < 30 kg/m²) patients.

Full table

There were 23 obese patients (11.22% of obese patients) who had BDI scores in the severely depressed range compared with 14 nonobese severely depressed patients (4.65% of nonobese patients). This difference points to the significant relationship (p = 0.0053) between obesity and severe depression. In an effort to determine whether there was any association between these findings and patient sex, female and male patients were analyzed separately. Female obese patients were found to be significantly more severely depressed than nonobese female patients (p = 0.0208), but these relationships did not hold true for male obese patients (p = 0.3789).

The comparison of health care resources use (mean and median) by obese and nonobese patients is shown in Table 2. To reduce the influence of outliers, the natural logarithm of these numbers was used in Student's t tests to identify significant differences. Obese patients had a significantly higher mean number of visits to their primary care clinic (p = 0.0005) and to specialty care clinics (p = 0.0006). They also used a significantly higher mean number of diagnostic services compared with nonobese patients (p < 0.0001). There were no statistically significant differences in the mean number of emergency department visits or hospitalizations.

Table 2. - Comparisons of health care resource use patterns for obese (BMI 30 kg/m²) and nonobese (BMI < 30 kg/m²) patients.

Full table

The mean and median annual per capita medical expenditures for the five categories of medical charges (and total charges) for obese and nonobese patients are presented in Table 3. Once again, using the natural logarithm of these charges (plus $10.00) to reduce the influence of outliers, Student's t tests examined differences in medical care charges for the five categories, comparing obese to nonobese patients. Obese patients had significantly higher primary care (p = 0.0058), specialty clinic (p = 0.0062), emergency department (0.0484), hospitalization (p = 0.0485), diagnostic services (p = 0.0021), and total charges (p = 0.0033) than nonobese patients.

Table 3. - Comparison of mean and median annual per capita medical charges for obese (BMI 30 kg/m²) and nonobese (BMI < 30 kg/m²) patients.

Full table

Regression equations were estimated to relate the logarithm of health resource use patterns (number of visits and tests) and medical charges of all categories to obesity, controlling for self-reported health status, depression, age, education, income, and sex (Table 4). Obesity was significantly related to the use of primary care (p = 0.0364) and diagnostic services (p = 0.0075). Approximately 14% of the variation in the number of log-transformed primary care visits was explained by physical health status, depression, age, education, income, sex, and obesity. Similarly, >15% of the variation in log-transformed diagnostic services use was explained by these variables. Because the dependent variable is expressed in logs, exponentiation of the estimated coefficient for any of the indicator variables (such as the coefficient 0.0893 for the variable "obesity" in the primary care equation) provides the percentage by which the average number of visits and tests for those patients having that characteristic exceeds the average for those who do not. Thus, obese patients had a 9.34% higher number of primary care visits and a 12.05% greater number of diagnostic services than nonobese patients.

Table 4. - Standardized estimates from regression equations in which the log of the mean number of visits and tests and medical charges are explained by obesity (BMI 30 kg/m²), controlling for physical health, depression (BDI), age, education, income, and sex (N = 506).

Full table

The final two columns in Table 4 display results for medical charges. Controlling for patient health status and sociodemographics, there were no statistically significant relationships between obesity and medical expenditures in any of the five categories or for total charges.

Discussion

As hypothesized, our findings showed a relationship between obesity and poorer self-reported health status. Obesity was also found to be significantly associated with depression, especially for female patients, which is in agreement with the finding of other researchers (12,13,14). Sociodemographic factors related to obesity in our study included sex, age, and education. It has been proposed that certain important variables may moderate or mediate the linkage between obesity and depression. Sociodemographic, psychosocial, and genetic characteristics may render certain obese individuals more prone to depression (12,13).

There was also evidence supporting our hypothesis regarding the association of obesity and use of health care services. The mean number of primary care and specialty care visits, as well as diagnostic services, was significantly higher for obese patients. Physicians are more likely to make specialty referrals and order laboratory, radiological, and other diagnostic tests for patients who make more frequent visits and have continuing medical complaints. Charges for primary care, specialty care, emergency treatment, hospitalization, diagnostic services, and yearlong total charges were all significantly higher for obese patients. While it is logical that higher mean numbers of outpatient visits (primary and specialty care) and diagnostic tests for obese patients would be associated with higher charges for these services, it is less clear why emergency treatment and hospitalization charges were higher for obese patients, despite use rates similar to those of nonobese patients. The poorer health status of obese patients may have resulted in more complicated and costlier care when they were seen in the emergency department or admitted to the hospital.

Because health status, depression, and sociodemographic variables are all associated with higher health care costs (23,24), these factors were controlled for in subsequent analyses using regression equations in an effort to clarify the impact of obesity on the use of medical services.

Previous studies examining whether obese patients generate higher health care costs have either attempted to control for confounding variables by not including subjects with certain comorbid conditions (22) or have argued that statistically adjusting for comorbidities is inappropriate because they are intermediates along the causal pathway between increased BMI and increased health service use and cost (21). We chose another approach. In our study, patients with diagnoses associated with obesity were not excluded. Instead, MOS SF-36 physical health status was used as a surrogate for comorbid conditions; the impact of these comorbidities on the individual patient was controlled for by including self-reported health as an independent variable in regression equations predicting health care resource use and charges.

As seen in our previous studies (23,24,25), our regression equation results in this study again highlighted the importance of physical health status, age, depression, and sex in the prediction of health care use. However, even after controlling for health status, depression, and key sociodemographic variables, obese patients continued to have both a significantly greater number of visits to their primary care provider and more diagnostic services ordered to evaluate them.

The difference shown here in medical resource use for obese compared with nonobese patients may be associated with divergent patient health beliefs and help-seeking behavior. Obese patients may believe that they are at increased risk for health problems that may be prevented or attenuated by more frequent visits to their primary providers. In addition, physicians cognizant of the increased risks associated with obesity, may request more follow-up appointments with interval diagnostic testing and monitoring for their obese patients. This differential care and testing may take place in the absence of actual health status differences, because it is motivated by perceived risk for potential health problems. The difference in the associated costs for this "extra" care for obese patients may not be of a magnitude high enough to achieve statistical significance. It may, however, be a wise investment if effective in maintaining desirable health outcomes.

An alternative interpretation of our study results may be that, just as in the Quesenberry study, statistically controlling for comorbid conditions (or their health status measure surrogate) associated with obesity eliminates the association between obesity and medical charges because these diseases drive up the cost of care.

The strengths of our study include the fact that BMI was calculated by actual patient weight and height measurements done by study personnel. This is more accurate than BMI values determined by self-reported patient information that have been used in other studies. The latter have been noted to be substantially lower than measured BMI values. We also assessed patient health status and depression with widely used instruments, allowing us to control for these important variables. In addition to data on use patterns, contemporaneous medical charge data were available to compare the costs for obese and nonobese patients.

There were a number of study limitations that should also be noted. Our study was conducted at a university medical center with primary care resident physicians. These physicians-in-training may be more likely to provide differential care for obese patients than community physicians. Moreover, study participants may represent a different patient population than those cared for in the community. Both self-reported physical and mental health status scores for study patients were lower than national means. In addition, we did not measure patient health beliefs and attitudes, which influence the propensity to seek and use health care services.

In conclusion, our hypotheses were confirmed. Obesity is associated with decreased health status and a higher incidence of depression. Obese patients also have more outpatient clinic visits, diagnostic services, and greater expenditures in all categories. Even when controlling for health status and sociodemographic factors, obese patients have more primary care visits and use more diagnostic services than nonobese patients. Obesity is a chronic condition requiring long-term management with an emphasis on prevention. If this critical health issue is not appropriately addressed, the prevalence of obesity and obesity-related diseases will continue to grow, resulting in escalating total health care costs. Future research should focus on the content of obese patient care visits and their outcomes.

Notes

¹ Nonstandard abbreviations: CHD, coronary heart disease; MOS SF-36, Medical Outcomes Study Short Form-36; BDI, Beck Depression Index.

References

1. National Heart Lung, and Blood Institute (1998) Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: The Evidence Report National Institute of Health Bethesda, MD.
2. Pi-Sunyer, F. X. (2002) The obesity epidemic: pathophysiology and consequences of obesity. Obes Res. 10: 97S–104S. | PubMed | ISI |
3. Flegal, K. M., Carroll, M. D., Ogden, C. L., Johnson, C. L. (2002) Prevalence and trends in obesity among US adults, 1999–2000. JAMA. 288: 1723–1727. | Article | PubMed | ISI |
4. Sturm, R. (2003) Increases in clinically severe obesity in the United States, 1986–2000. Arch Intern Med. 163: 2146–2148. | Article | PubMed | ISI |
5. Mokdad, A. H., Ford, E. S., Bowman, B. A., et al (2003) Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001. JAMA 289: 76–79. | Article | PubMed | ISI |
6. Serdula, M. K., Khan, L. K., Dietz, W. H. (2003) Weight loss counseling revisited. JAMA 289: 1747–1750.
7. Must, A., Spadano, J., Coakley, E. H., Field, A. E., Colditz, G., Dietz, W. H. (1999) The disease burden associated with overweight and obesity. JAMA 282: 1523–1529. | Article | PubMed | ISI | ChemPort |
8. Calle, E. E., Rodriguez, C., Walker-Thurmond, K., Thun, M. J. (2003) Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med. 348: 1625–1638. | Article | PubMed | ISI |
9. Fontaine, K. R., Redden, D. T., Wang, C., Westfall, A. D., Allison, D. B. (2003) Years of life lost due to obesity in the United States. JAMA 289: 187–193. | Article | PubMed | ISI |
10. Alison, D. B., Fontaine, D. R., Manson, J. E., Steven, J., VanItallie, T. B. (2003) Annual deaths attributable to obesity in the United States. JAMA 282: 1530–1538.
11. Roberts, R. E., Kaplan, G. A., Shema, S. J., Strawbridge, W. J. (2000) Are the obese at greater risk for depression? Am J Epidemiol. 152: 163–170. | Article | PubMed | ISI | ChemPort |
12. Faith, M. S., Matz, P. E., Jorge, M. A. (2002) Obesity-depression associations in the population. J Psychosom Res. 53: 935–942. | Article | PubMed | ISI |
13. Stunkard, A. J., Faith, M. S., Allison, K. C. (2003) Depression and obesity. Biol Psychiatry 54: 330–337. | Article | PubMed | ISI |
14. Dixon, J. B., Dixon, M. E., O'Brien, P. E. (2003) Depression in association with severe obesity: changes with weight loss. Arch Intern Med. 163: 2058–2065. | Article | PubMed | ISI |
15. Roberts, R. E., Deleger, S., Strawbridge, W. J., Kaplan, G. A. (2003) Prospective association between obesity and depression: evidence from the Alameda County Study. Int J Obes Relat Metab Disord. 27: 514–521. | Article | PubMed | ChemPort |
16. Seidell, J. C. (1995) The impact of obesity on health status: some implications for health care costs. Int J Obes Relat Metab Disord. 19: S13–S16. | PubMed |
17. Seidell, J. C. (1998) Societal and personal costs of obesity. Exp Clin Endocrinol Diabetes 106: S7–S9.
18. 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 |
19. Colditz, G. A. (1999) Economic costs of obesity and inactivity. Med Sci Sports Exerc. 31: S663–S667. | Article | PubMed | ISI | ChemPort |
20. Wolf, A. M. (2002) Economic outcomes of the obese patient. Obes Res. 10: 58S–62S.
21. Quesenberry, C. P., 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 |
22. Thompson, D., Brown, J. B., Nichols, G. A., Elmer, P. J., Oster, G. (2001) Body mass index and future healthcare costs: a retrospective cohort study. Obes Res. 9: 210–218. | PubMed | ChemPort |
23. Bertakis, K. D., Azari, R., Callahan, E. J., Helms, L. J., Robbins, J. A. (1999) The impact of physician practice style on medical charges. J Fam Pract. 48: 31–36.
24. Bertakis, K. D., Azari, R., Helms, L. J., Callahan, E. J., Robbins, J. A. (2000) Gender differences in the utilization of health care services. J Fam Pract. 49: 147–152. | PubMed | ISI | ChemPort |
25. Callahan, E. J., Bertakis, K. D., Azari, R., Robbins, J. A., Helms, L. J., Leigh, J. P. (2002) Association of higher costs with symptoms and diagnosis of depression. J Fam Pract. 51: 540–544.
26. Stewart, A. L., Hays, R. D., Ware, J. E. Jr (1988) The MOS short-form general health survey. Reliability and validity in a patient population. Med Care 26: 724–735. | Article | PubMed | ChemPort |
27. Stewart, A. L., Greenfield, S., Hay, R. D. (1989) Functional status and well-being of patients with chronic disease conditions. Results from the Medical Outcomes Study. JAMA. 262: 907–913. | Article | PubMed | ISI | ChemPort |
28. McHorney, C. A., Ware, J. E., Raczek, A. E. (1993) The MOS 36-item short-form health survey (SF-36): II. Psychometric and clinical tests validity in measuring physical and mental health constructs. Med Care 31: 247–263. | Article | PubMed | ISI | ChemPort |
29. Ware, J. E., Kosinski, M., Keller, S. D. (1994) SF-36 Physical and Mental Health Summary Scales: A User's Manual The Health Institute Boston, MA.
30. Beck, A. T., Ward, C. H., Mendelson, M. (1961) An inventory for measuring depression. Arch Gen Psychiatry 4: 561–571. | PubMed | ISI | ChemPort |
31. Beck, A. T., Beck, R. W. (1972) Screening depressed patients in family practice. A rapid technique. Postgrad Med. 52: 81–85. | PubMed | ISI | ChemPort |
32. Beck, A. T., Steer, R. A., Garbin, M. G. (1988) Psychometric properties of the Beck Depression Inventory: twenty-five years of evaluation. Clin Psychol Rev. 8: 77–100. | Article | ISI |

Acknowledgments

This project was supported by Agency for Healthcare Policy and Research Grant R 18 HSO6167, now known as the Agency for Healthcare Research and Quality.