SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Objective: To examine the effect of obesity and cardiometabolic risk factors on medical expenditures and missed work days.

Methods and Procedures: The 2000 and 2002 Medical Expenditure Panel Survey (MEPS), a nationally representative survey of the US population, was used to estimate the marginal effect of obesity (BMI ≥ 30) on annual per-person medical expenditures and missed work days for patients with diabetes, dyslipidemia, or hypertension using multivariate regression methods controlling for age, sex, race, ethnicity, education, income, insurance, and smoking status. Maximum Likelihood Heckman Selection with Smearing retransformation was used to assess medical expenditures, and Negative Binomial regression was used for missed work days.

Results: Normal weight individuals with diabetes, dyslipidemia, or hypertension had significantly greater medical expenditures than those without the respective condition ($6,006 (5,124–6,887), $4,760 (4,102–5,417), $3,911 (3,345–4,476)) and obesity significantly exacerbated this effect ($7,986 (7,397–8,574), $7,636 (7,072–8,200), $6,197 (5,745–6,649); $2007; all P < 0.05). In addition, diabetes, dyslipidemia, and hypertension resulted in greater missed work days (3.1 (0.94–6.21), 3.2 (0.42–7.91), 1.4 (0.0–3.52)) (all P < 0.05 except hypertension), which resulted in greater lost productivity ($433, $451, $199) and obesity significantly exacerbated the deleterious effect on work days (8.7 (4.44–15.2), 5.5 (2.18–10.5), 4.5 (2.92–6.34)) and lost productivity ($1,217, $763, $622) (all P < 0.05). In addition, medical expenditures increased for increasing weight category and increasing number of risk factors.

Discussion: Obesity significantly exacerbates the deleterious effect of diabetes, dyslipidemia, and hypertension on medical expenditures and productivity loss in the United States. Obesity is preventable and public health efforts need to be undertaken to prevent its alarming increase in order to reduce the incidence and effect of cardiometabolic risk factors.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

It is well established that obesity is a current and rapidly mounting public health crisis (1,2,3,4). It is estimated that between 20 and 30% of the adult US population is obese (5,6,7), and the rate of obesity is increasing dramatically in both adults and children (7,8). Indeed, overweight or obesity is one of the top leading health indicators addressed in the United States' Healthy People 2010 goals (9).

Simultaneous with the increased incidence of obesity, the incidence of diabetes, hypertension, and dyslipidemia is increasing dramatically (4) for men and women and for all race, ethnic, age, and education groups in the United States (10,11). There is substantial evidence that cardiometabolic risk factors such as obesity, diabetes, dyslipidemia, and hypertension result in an elevated risk of cardiovascular disease (11,12) and mortality (13).

Obesity has been shown to result in greater health-care utilization, higher medical expenditures, diminished productivity, and employment (14,15,16,17,18,19,20). Diabetes, dyslipidemia, and hypertension individually have also been shown to result in increased medical expenditures and lost productivity (20,21,22,23,24,25,26,27). Previous research has shown that specific clusters of cardiometabolic risk factors have a deleterious effect on medical expenditures, employment, and productivity in the United States (28,29,30). Although there are data on the effect of individual cardiometabolic risk factors on medical expenditures and productivity, limited data exist on the marginal effect of obesity on medical expenditures and productivity associated with common cardiometabolic risk factors such as diabetes, hypertension, and dyslipidemia. Given the increasing prevalence of these risk factors, and the evidence pointing to increased medical expenditures and reduced productivity, it is important to examine how obesity interacts with cardiometabolic risk factors to affect economic outcomes.

This study aimed to estimate the marginal effect of obesity on medical expenditures and productivity associated with diabetes, hypertension, and dyslipidemia in the nationally representative Medical Expenditure Panel Survey (MEPS). A secondary aim of the study was to examine the marginal effect of obesity on medical expenditures and productivity for increasing numbers of cardiometabolic risk factors.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Study design

This study is a retrospective database analysis examining annual medical expenditures and lost productivity. The examination of medical expenditures included all medical expenditures whether paid by the individual patient or the payer. The analysis of lost productivity included missed work days and their respective cost for employed individuals.

Data source

The MEPS is a nationally representative survey of the US civilian noninstitutionalized population that contains detailed information on demographic and socioeconomic characteristics, health conditions, insurance status, smoking status, missed work, and utilization and cost of health-care services (31). The current research pooled the annual 2000 and annual 2002 MEPS public use data (31). All analyses incorporated MEPS weights to ensure nationally representative estimates. Further details on the MEPS are available online (http:www.meps.ahrq.gov).

Study variables

Information from the MEPS Household Component survey was used to ascertain self-reported medical conditions, sociodemographic attributes, and BMI. Reported current body weight and height were used to calculate BMI (31) and then four categories were constructed: underweight (BMI < 18.5), normal (BMI 18.5–24.9), overweight (BMI 25–29.9), and obese (BMI ≥ 30) (National Heart Lung and Blood Institute, 1998 #382). Self-reported medical conditions were mapped to Clinical Classification Category codes based on medical and pharmacy utilization and self-report. For the multivariate analysis, each condition (including diabetes, dyslipidemia, and hypertension) was classified into six mutually exclusive dichotomous (yes/no) variables: normal weight without the condition (reference), normal weight with the condition, overweight without the condition, overweight with the condition, obese without the condition, obese with the condition. A second set of analyses examined the effect of the number of risk factors in combination with weight category. For this analysis, individuals were categorized into 12 mutually exclusive groups: normal weight with no conditions, normal weight with one condition, normal weight with two conditions, normal weight with all three conditions and the same respective four categories for overweight and obese. In order to examine the mediating effects of chronic comorbidity, the total number of reported chronic conditions (minus the condition of interest, i.e., diabetes) were added together to create a count variable called “number of chronic conditions” (NCC). This variable was created by adding all reported chronic Clinical Classification Category codes.

Data analysis

In the first set of analyses, a separate regression was constructed for each condition controlling for sociodemographic characteristics. For example, to examine the marginal effect of obesity on medical expenditures associated with diabetes medical expenditures were regressed on normal weight with diabetes (normal weight without diabetes was the reference), obese without diabetes and obese with diabetes, controlling for overweight without diabetes, overweight with diabetes, underweight, age, sex, race, ethnicity, education, income, smoking, and insurance status. The same approach was then followed for dyslipidemia and hypertension separately. A second set of analyses was conducted as a form of sensitivity analysis to examine the mediating effects of chronic comorbidity. In these regressions, the same analyses were conducted with the inclusion of the NCC variable to control for all other chronic conditions (the total number of reported chronic conditions). An additional set of analyses were conducted to examine the effect of combinations of weight category and conditions using the aforementioned 12 variables. This research used a Maximum Likelihood Heckman Selection model (32) with logarithmic transformation of expenditures and Smearing retransformation using the naïve (normal) assumption for residuals (33). Medical expenditure data were inflated to be expressed in $US 2007. Negative binomial regression was used for missed work days. The Bureau of Labor Statistics 2007 Average National Hourly Wage across US occupations was used to estimate the cost of missed work days and bed days for national indirect cost estimates (34).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Unadjusted descriptive statistics

Unadjusted descriptive statistics are displayed in Table 1. After pooling the 2000 and 2002 files, there were 43,221 unique individuals with complete information on all of the sociodemographic variables of interest. It appears that obesity, diabetes, and hypertension are more prevalent for increasing age, lower levels of educational attainment and income, Hispanic, black and American-Indian populations in the United States MEPS sample.

Table 1.  Unadjusted prevalence of diabetes, hypertension, dyslipidemia, and obesity by demographic and socioeconomic attribute (MEPS 2000–2002)
inline image

Table 2 presents unadjusted mean NCC, utilization, expenditures, and missed work days by condition and weight category. The NCC, utilization, expenditures, and missed work days were higher for obese compared to normal weight individuals. There was also a clear trend of higher comorbidity, utilization, expenditures, and missed work days for obese individuals with the conditions than normal weight individuals with the respective conditions. Individuals with diabetes and obesity appeared to have the highest (worst) values on all outcomes.

Table 2.  Unadjusted descriptive statistics: age, comorbidity, utilization, and expenditures by condition and weight category (MEPS 2000–2002a: per person per year)
inline image

Adjusted results from multivariate regression analyses

Medical expenditures. In the multivariate regressions controlling for sociodemographic characteristics, individuals with diabetes, dyslipidemia, and hypertension had greater medical expenditures compared to normal weight individuals without the respective condition: diabetes $6,006; dyslipidemia $4,760; hypertension $3,911 (Table 3). In addition, there was a clear increasing trend corresponding with increasing weight categories. The presence of obesity significantly increased the medical expenditures associated with each condition ($7,986, $7,636, and $6,197, respectively). Controlling for all other chronic conditions (NCC) in the multivariate regressions appeared to attenuate the deleterious effect of obesity on diabetes, dyslipidemia, and hypertension on medical expenditures (Table 3). In addition, in multivariate analysis controlling for sociodemographic characteristics, medical expenditures were greater for increasing numbers of cardiometabolic risk factors and obesity appeared to exacerbate this effect (Table 4). Similarly, controlling for all other chronic conditions in the multivariate regression attenuated the effect of obesity on medical expenditures associated with increasing numbers of cardiometabolic risk factors (Table 4). Figure 1 shows the sources of medical expenditures for different conditions and weight categories. Individuals with cardiometabolic risk factors such as obesity, diabetes, dyslipidemia, and hypertension have greater overall medical expenditures and this is due primarily to hospital, pharmaceutical, and office-based expenditures.

Table 3.  Multivariate regression results of the effect of obesity on medical expenditures associated with diabetes, dyslipidemia, and hypertension (MEPS 2000–2002)
inline image
Table 4.  Multivariate regression results of the effect of obesity on medical expenditures associated with increasing numbers of RFsa (MEPS 2000–2002)
inline image
image

Figure 1. Unadjusted medical expenditures by source (MEPS 2000–2002). MEPS, Medical Expenditure Panel Survey.

Download figure to PowerPoint

Lost productivity. Diabetes, dyslipidemia, and hypertension also appeared to result in a greater number of missed work days and lost productivity compared to normal weight individuals without the condition in multivariate analyses controlling for sociodemographic characteristics (Table 5). Similar to medical expenditures obesity exacerbated this effect. Again, controlling for all other chronic conditions attenuated the magnitude of missed work days and productivity loss (Table 5). Missed work days and lost productivity also appeared to increase for increasing numbers of risk factors and weight category. However, controlling for all other chronic conditions significantly reduced this correlation with many coefficients not reaching statistical significance (Table 6).

Table 5.  Multivariate regression results of the effect of obesity on lost work days and productivity associated with diabetes, dyslipidemia, and hypertension (MEPS 2000–2002)
inline image
Table 6.  Multivariate regression results of the effect of obesity on lost work days and productivity associated with increasing numbers of RFsa (MEPS 2000–2002)
inline image

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

The results of this nationally representative study show that obesity exacerbates the deleterious effect of diabetes, dyslipidemia, and hypertension on medical expenditures and lost productivity. The results also show that medical expenditures and lost productivity are greater for increasing numbers of cardiometabolic risk factors and, again, obesity exacerbates this relationship. In addition, the results suggest that the development of other chronic conditions (other than diabetes, dyslipidemia, and hypertension) is responsible for a portion of the increased medical expenditures and productivity loss associated with cardiometabolic risk factors. Finally, the results show that the majority of medical expenditures come from hospitalization, pharmaceutical, and office-based sources.

To the authors' knowledge, there are no prior publications examining the marginal effect of obesity on medical expenditures and productivity loss associated with diabetes, dyslipidemia, and hypertension. However, there have been prior studies examining the cost of individual cardiometabolic risk factors and clusters thereof as well as cardiovascular disease.

The direct medical cost of obesity in the United States is estimated to be as high as $92.6 billion (2002), ranging from 5.5 to 9.1% of total medical spending (14,15). Previous research has shown that obesity is associated with an increase in medical utilization including prescription drugs, physician visits, hospitalizations, and emergency room visits (16,17). Sturm estimated that individuals with obesity incur medical expenditures that are 36% higher than normal weight individuals and estimated the direct medical cost of obesity to be greater than that of smoking and problem drinking for individuals 18–65 years of age based on a 1997 national telephone survey (19). Finkelstein et al. estimated that the direct medical cost of obese individuals were 37% higher than normal weight individuals ($732 higher; $1,057 in 2007), resulting in a total of $51.5 billion in the United States using the 1996 and 1997 National Health Interview Survey (NHIS) and the 1998 MEPS ($74.3 billion in 2007) (ref. 15). In an analysis from the employer perspective, obesity has been shown to result in increased medical expenditures for employed individuals that could cost up to $280,000 for a firm of 1,000 employees (20).

Hodgson et al. found that medical expenditures associated with diabetes were significant in the United States, after including unrelated conditions, comorbidities, and/or chronic complications (21). Based on extrapolation methods, Hogan et al. estimated that the per capita medical expenditures of individuals with diabetes were 2.4 times higher than those without (22). Hypertension has been estimated to result in expenditures of $3,787 ($5,456 in 2007) higher than those without hypertension, using a variety of national data sources (23). Hypertension and diabetes were associated with per-person direct medical expenditures for treatment of $4,073 and $5,646 ($5,869 and $8,134 in 2007), respectively, in the 1996 MEPS (24). In addition, combinations of cardiometabolic risk factor clusters have been shown to result in $5,477 ($5,918 in 2007) per person in medical expenditures in the 2000–2002 MEPS after controlling for sociodemographic characteristics and all chronic conditions (29). The methodological approach by Sullivan et al. is very similar to that used in this research in that it controlled for sociodemographic characteristics and chronic disease comorbidity but different in that it focused on specific clusters of cardiometabolic risk factors (defined as overweight (BMI ≥ 25) with at least two of the following: diabetes, dyslipidemia, and hypertension).

Previous research has also examined the effect of individual cardiometabolic risk factors and clusters of risk factors on productivity. After controlling for age, gender, and employment status, Wolf and Colditz estimated that obese individuals missed 39.3 million work days resulting in a cost of $4 billion in the United States based on the 1994 NHIS (16). Finkelstein et al. found that obesity results in increased absenteeism in employed individuals (20). Bungum et al. showed that increasing BMI was associated with work absence among city employees in a large metropolitan area in the South (35). Burton et al. found that obesity was associated with more short-term disability claims and a greater number of sick days in an employed population of First Chicago Bank (36,37). In addition, Robbins et al. found that obesity was associated with lost work days in a population of active duty personnel in the US Air Force in 1997 (38). Several studies have documented the negative effect of diabetes on productivity and employment (25,26,27). In addition, previous research has shown that individuals with cardiometabolic risk factor clusters were 40–45% less likely to be employed and missed 179% more work days due to illness resulting in $17.3 billion ($18.7 in 2007) in lost productivity in the United States MEPS (again defined as overweight (BMI ≥ 25) with at least two of the following: diabetes, dyslipidemia, and hypertension) (28,30).

In summary, although no previous research has examined the marginal effect of obesity on diabetes, hypertension, and dyslipidemia, previous research has consistently shown the negative effect of these conditions alone and in combination on medical expenditures and productivity. Although different in their methodological approach, these important studies demonstrate the significant economic effect in the United States and lend credibility to the findings from the current study. Consistent with the literature discussed above, the results of the current study demonstrate increased medical expenditures and lost productivity associated with diabetes, dyslipidemia, and hypertension. However, this study adds important new knowledge to this literature by quantifying the marginal effect of obesity on medical expenditures and lost productivity associated with these common cardiometabolic risk factors.

There is strong clinical evidence of the link between cardiometabolic risk factor clusters examined in this research and the development of cardiovascular disease as well as the exacerbation of other comorbidities. Comprehensively controlling for all chronic conditions (including cardiovascular disease) thus removes a significant portion of the cost attributable to cardiometabolic risk factors resulting in a downward bias. However, not controlling for chronic conditions may result in including extraneous costs that are not directly attributable to cardiometabolic risk factors, causing an upward bias. For these reasons, both levels of analysis are presented as a form of sensitivity analysis. Nonetheless, estimates from both levels of analysis demonstrate the significant medical expenditures and productivity loss due to the effect of obesity and individual cardiometabolic risk factors and their combinations.

It is clear from the results that obesity significantly exacerbates the deleterious effect of cardiometabolic conditions (and combinations of these conditions). Although it is not entirely understood how obesity interacts with these conditions to result in greater medical expenditures and lost productivity, the results are consistent with previous research showing that obesity itself results in greater medical expenditures and lost productivity. In addition to the deleterious effect of obesity itself, another possible reason may be the development and exacerbation of other conditions: this research appears to support this notion because the results showed that controlling for all other chronic conditions attenuated the deleterious effect of obesity.

There are limitations to our research. Our analysis focuses on a nationally representative sample in the United States. Hence, to the extent that other countries and populations do not resemble those in the United States our findings may not be generalizable. The prevalence of diabetes, obesity, hypertension, and dyslipidemia estimated are consistent with other survey-based national-level estimates in the United States (6,24,39,40,41), but it is possible they may be biased downward. Similar to the Behavioral Risk Factor Surveillance System and the NHIS, MEPS is based on self-report. Self-reported conditions may be underreported (42) and the extent of underreporting may vary by race and ethnicity (43,44,45). Overweight respondents may also underestimate their weight and overestimate their height (46,47). It has been estimated that up to 35% of individuals with diabetes have not been diagnosed (48). Unlike the National Health and Nutrition Examination Survey, MEPS does not contain information on laboratory values and undiagnosed diabetes, hypertension, or dyslipidemia, which likely results in an underestimate of prevalence for these conditions. As a result, the estimates of medical expenditures and productivity loss may be underestimates to a commensurate degree due to the fact that the comparison group was normal weight individuals without the condition of interest. (Thus the comparison group may include individuals who have the condition or are not normal weight). However, other data sources that contain more specific laboratory data (such as undiagnosed diabetes) do not contain the rich array of medical expenditure and productivity information available in MEPS. In addition, there is a risk of omitted variable bias. Although we comprehensively controlled for all important observed variables, there may be unobserved characteristics that affect the outcomes studied. Finally, the types of costs included may affect the results. For example, this study only examined lost productivity for employed individuals, ignoring the lost productivity of those not working; the direct cost analysis did not include direct nonmedical costs; and this analysis also restricted its scope to the current annual burden, rather than the lifetime cost. Inclusion of any of these factors would have resulted in a greater cost burden.

In conclusion, the current study provides important evidence in a nationally representative population that obesity exacerbates the negative effect of diabetes, hypertension, and dyslipidemia on the US economy. Despite the clear clinical and economic toll for individuals as well as employers, governments, and health-care payers, the incidence of these cardiometabolic risk factor clusters is increasing significantly in almost every demographic group (10,11). Obesity is preventable and public health efforts need to be undertaken to prevent its alarming increase in prevalence such as encouraging a healthy diet and physical activity. While this study focused on the adult population, the increasing prevalence of obesity among children is alarming and will result in an increasing economic burden of cardiometabolic risk factors in the future. Unless successful public health efforts are undertaken, the economic toll of obesity and cardiometabolic risk factors will continue to grow unabated and result in lower productivity and higher medical expenditures in the US economy.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

This research was supported by a research grant from Sanofi-aventis.

Disclosure

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

This research was conducted with a research grant from Sanofi-aventis (Sullivan). Dr Ben-Joseph is an employee of Sanofi-aventis.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES