Direct Healthcare Cost of Obesity in Brazil: An Application of the Cost-of-Illness Method from the Perspective of the Public Health System in 2011

Background Obesity is a global public health problem and a risk factor for several diseases that financially impact healthcare systems. Objective To estimate the direct costs attributable to obesity (body mass index {BMI} ≥ 30 kg/m2) and morbid obesity (BMI ≥ 40 kg/m2) in adults aged ≥ 20 incurred by the Brazilian public health system in 2011. Settings Public hospitals and outpatient care. Methods A cost-of-illness method was adopted using a top-down approach based on prevalence. The proportion of the cost of each obesity-associated comorbidity was calculated and obesity prevalence was used to calculate attributable risk. Direct healthcare cost data (inpatient care, bariatric surgery, outpatient care, medications and diagnostic procedures) were extracted from the Ministry of Health information systems, available on the web. Results Direct costs attributable to obesity totaled US$ 269.6 million (1.86% of all expenditures on medium- and high-complexity health care). The cost of morbid obesity accounted for 23.8% (US$ 64.2 million) of all obesity-related costs despite being 18 times less prevalent than obesity. Bariatric surgery costs in Brazil totaled US$ 17.4 million in 2011. The cost of morbid obesity in women was five times higher than it was in men. Conclusion The cost of morbid obesity was found to be proportionally higher than the cost of obesity. If the current epidemic were not reversed, the prevalence of obesity in Brazil will increase gradually in the coming years, as well as its costs, having serious implications for the financial sustainability of the Brazilian public health system.


Introduction
The obesity epidemic is recognized by the World Health Organization as a public health problem. Healthcare expenditure on this chronic disease is known to rise in direct proportion to body mass index (BMI) [1]. In 2008, 502 million adults (aged 20+) were obese worldwide [2], with a 10% prevalence in men and 14% in women. The prevalence of obesity is even higher in the Americas -26% in both sexesand is on the rise, having doubled between 1980 and 2008 [3]. According to the Brazilian Consumer Expenditure Survey conducted in 2008-2009, obesity affects 14.8% of the Brazilian population, including 12.5% of adult men and 16.9% of adult women. Obese individuals represent nearly 25% of all overweight men and nearly a third of all overweight women [4]. The prevalence of overweight and obesity has increased over the last four decades. Morbid obesity (body mass index {BMI} ! 40 kg/m 2 ) is rising at a higher rate than obesity in Brazil. Between 1989 and 2003, its prevalence in men quadrupled (from 0.08% to 0.32%), while its prevalence in women increased by 44% (from 0.66% to 0.95%), but it still remained higher in women [5].
2.8 million people die each year around the world as a result of being overweight or obese [6]. Obese people have a higher risk of type 2 diabetes and cardiovascular diseases such as coronary heart disease and strokes, as well as hypertension, osteoarthritis and several types of cancer, including colorectal, breast, uterine and renal cancer [7][8][9][10][11][12][13][14]. When the relative risk of a disease is higher for obese people, a portion of cases can be directly attributed to obesity [15]. In the BMI range of 30-35 kg/m², average survival is reduced by 2 to 4 years; in the 40-45 kg/m² range, it is reduced by 8 to 10 years [16], or by 5 to 20 years depending on sex, age and ethno-racial group [17]. A study conducted in the United States estimated that 14% of all cancer deaths and 20% of women's deaths are related to obesity [18]. The higher the BMI, the greater is the risk of comorbidity [19][20][21][22].
Overweight and obesity contribute to the global burden of disease [23]. In 2002, the proportion of the disease burden attributable to a higher BMI was 58% for type 2 diabetes, 21% for ischemic heart disease, 39% for hypertension, 23% for stroke, 12% for colon cancer, 8% for postmenopausal breast cancer, 32% for endometrial cancer and 13% for osteoarthritis [24]. In this regard, there is undeniably a connection between rising rates of obesity and rising medical spending [25], as well as disability-adjusted life years (DALYs) and years lost due to disability (YLDs).
Studies that emphasize the disproportionality of the cost of treating people with different degrees of obesity can assist health managers in making decisions about how to allocate resources [26,27] dedicated to preventing and treating this disease. Accordingly, this study aimed to estimate the direct costs attributable to obesity (BMI ! 30 kg/m 2 ) and morbid obesity (BMI ! 40 kg/m 2 ) in adults aged 20 and over incurred by the Brazilian public health system in 2011.
A search was conducted of all obesity-related outpatient and inpatient costs. These amounts were added to the costs of other diseases that are related to obesitydiseases and conditions that could have been avoided if obesity had been prevented [28,29]. These costs were obtained by calculating the proportion of total disease cases associated with obesity (Population Attributable Risk -PAR) and multiplying this measure by the cost of treating these comorbidities. The PAR for obesity was calculated for each disease and categorized by BMI range and sex. PAR was calculated using the following formula [30]: in which: • P = prevalence of obese (or morbidly obese) individuals, obtained from the 2008-2009 Consumer Expenditure Survey [4], and • RR = relative risk for a certain disease in obese (or morbidly obese) individuals vs. eutrophic individuals.
To calculate the PAR, relative risks (RR) were used (and their confidence intervals) obtained when the risk of incidence of each disease in obese (or morbidly obese) patients was compared to the risk of incidence in normal weight patients. The latter data were obtained in cohort studies and meta-analyses published in international journals after 2000 and indexed in the following databases: Medline, Evidence-Based Medicine Reviews (EBMR), Cochrane Database of Systematic Reviews, Lilacs, Scielo and Medilene via Pubmed. When the RR was not found, the risk was estimated using the odds ratio (OR). If the outcome is relatively rare in population (10% or less), the relative risk could be estimated by odds ratio [31,32]. The meta-analyses adopted for the RR or OR were identified by Pubmed and selected with the AMSTAR [33] tool, which enables the assessment of systematic reviews. Most of the studies found were conducted in the United States and Europe. One of the main sources of relative risk for this study was Guh et al 2009 [9], who used the following criteria to select studies for inclusion in their metaanalysis: prospective cohort study of the population of a Western country, relevant results, sample size of at least 200 individuals and risk estimate based on disease incidence rather than disease mortality rate.
Obesity prevalence was obtained from the Consumer Expenditure Survey (Pesquisa de Orçamentos Familiares in Portuguese), which employed a nationally representative sample of the Brazilian households carried out by the Brazilian Institute of Geography and Statistics (IBGE, acronym in Portuguese). Data from 55,970 households and 188,461 respondents were collected between 2008 and 2009. Weight and height were measured using a portable electronic scale and a portable stadiometer, respectively [4].
Costs related to hospital (inpatient) and ambulatory (outpatient) care were included in this study. The following direct healthcare costs were obtained: medical consultations, inpatient care, bariatric surgeries, etc.; the cost of medium-and high-complexity procedures, including complementary procedures (clinical pathology, radiology, ultrasound exams, nuclear medicine, interventional radiology and computerized tomography); and the cost of medications, orthotics and prosthetics provided by the Brazilian public health system. This information, which was organized according to the chapters of the International Disease Classification (ICD-10 -WHO, 1995), was obtained from the Brazilian Health Ministry's Hospital Information System and Outpatient Information System. The identification of the procedures related to obesity, morbid obesity and associated diseases was made through the ICD-10, being registered in the Table Management System of Procedures, Medical drugs, Orthotics, Prosthetics and Special Materials of the Unified Health System (SIGTAP in Portuguese), available in the Health Information Systems of the Ministry of Health. These databases correspond to all procedures performed in the public health system and financially supported by the Ministry of Health. Therefore, they included all people living in Brazil that accessed the public health system, all age groups and both sexes. The public system is responsible for approximately 75% of medical care in Brazil [34].
As this study is an analysis of the public health service's costs, we did not include family expenses such as transportation, food (including dietary products), lodging or patient care by family providers.
The study protocol was reviewed and approved by the Ethics in Research Committee of the Faculty of Health Sciences of University of Brasilia (report no. 120-10). All data came from secondary database without identifying people. study, as well as their relative risk or odds ratio related to obesity and morbid obesity. Diabetes is a disease with a higher risk of incidence in people with obesity or morbid obesity. Besides diabetes, the diseases with the greatest risk of incidence in obese men are osteoarthritis and pulmonary embolism, while the greatest risk of incidence in obese women is for pulmonary embolism and endometrial cancer. The pathologies with increased risk of incidence in men with morbid obesity (besides diabetes) are hypertension and esophageal cancer; in women, they are endometrial cancer, gallbladder cancer and hypertension (Fig 1). An association between obesity and skin cancer was only found in men [13], and an association was also found between obesity and three female cancers: ovarian, breast and endometrial. Table 1 reports the population attributable risks (PAR) calculated for some diseases associated with obesity and morbid obesity, according to sex. Table 2 presents the main results of this study: the costs attributable to obesity and morbid obesity, which incurred on the Brazilian public health system budget in 2011, according to associated diseases. These costs totaled US$ 269.6 million, which corresponded to 1.86% of all Ministry of Health's expenditure related to hospital and ambulatory care in Brazil. Comparing the costs by healthcare level, 59.2% of the total cost was hospital admissions (inpatient care) and 40.8% corresponded to ambulatory procedures (outpatient care). When analyzed by associated disease, obesity-attributable costs were highest for ischemic heart disease, followed by breast cancer, congestive heart failure and diabetes ( Table 2).

Costs according to severity of obesity
This study was the first one to estimate morbid obesity costs in Brazil, which corresponded to $64.2 million in 2011. The highest costs attributable to morbid obesity were for ischemic heart disease and diabetes, probably because ischemic heart disease treatment is expensive and diabetes is a disease that has the highest incidence risk in people with obesity or morbid obesity. Despite the fact that morbid obesity was 18 times less prevalent than obesity, morbid obesity costs accounted for 23.8% of all obesity-related costs. Bariatric surgery was the single highest cost, accounting for $17.4 million, or 27.1% of this total. Between 2008 and 2011, the number of bariatric surgeries increased annually, as did the annual cost of such procedures (Table 3). Fig 2 reports the proportion of costs attributable to morbid obesity in relation to total obesity costs by comorbidity. Diabetes, gallbladder cancer, breast cancer and hypertension were found to be the diseases in which morbid obesity accounted for more than 25% of total obesity costs (Fig 2).

Costs according to gender
In Brazil, the costs attributable to obesity in women were double the costs for men in 2011 (Table 4). Of this difference, US$ 28.8 million corresponds to female cancers (endometrial, breast, and ovarian). The cost of morbid obesity in women was five times higher than it was in men.

Discussion
Costs attributable to obesity accounted for 1.86% of medium-and high-complexity healthcare expenses ($22.6 billion) in 2011 [35]. Although this figure does not take into account the total cost of healthcare, it is similar to other studies. A compilation of studies of direct costs worldwide estimated that countries spend between 0.7% and 2.8% of their total health budget on obesity [36]. World Health Organization studies in Europe indicate that direct costs due to obesity generally total 2-4% of national healthcare expenses [37]. The USA is associated to considerably higher shares of the total medical healthcare costs, ranging from 5.5% to 20.1% [38].
Recently Bahia et al (2012) [39] reported on the cost of overweight-and obesity-related diseases in Brazil; however, their study was based on self-reported weights and heights from a non-representative sample (adults living in state capitals). Our analysis was based on surveys that are representative of the Brazilian population, conducted through home visits carried out by the Brazilian Institute of Geography and Statistics (IBGE) and involving more accurate height and weight measurements. If we compare the obesity-attributable costs obtained, we see that our study calculated higher costs, which is probably due to its more accurate estimation of obesity prevalence and the higher number of comorbidities analyzed. Another obesity study conducted by Sichieri et al. [40] estimated the cost to the Brazilian public health system of inpatient care for individuals aged 20 to 60 associated with overweight and obesity and their comorbidities in 2001. The costs totaled $36 million, representing 3.02% of total inpatient costs for men and 5.83% for women. The amounts found in our study were five times higher ($157.6 million) due to the increased prevalence of obesity in this period and the inclusion of the elderly (over 60 years of age).
When comparing the obesity costs of different countries, it is important to consider not only the differences in methodologies used and in population and epidemiological characteristics, but also the fact that the healthcare systems and services provided vary greatly from country to country and even between regions of the same country [26]. It is also important to emphasize that age and sex influence the results of studies into the relationship between obesity and healthcare costs [36].
Turning to comparative costs between the sexes, our finding that the cost of obesity is higher in women has also been identified in other studies of the cost of obesity in Brazil [39,40]. However, in our analysis the cost of obesity in women was found to be proportionally higher. This result can be explained by several factors: a) a higher prevalence of obesity and especially morbid obesity in women; b) women use health services more than men [39]; c) US$ 28.8 million Fig 1. RR or OR for diseases associated with obesity (BMI ! 30 kg/m 2 ) and morbid obesity (BMI ! 40 kg/m 2 ) in adults aged 20 and over, according to sex. Sources: 6 WHO 2012, 9 Guh, 2009, 10 IASO (1) (*), 11 Martinez, 2004, 12 Mokdad et al, 2001, 13 Renehan et al, 2008 (1), 14 Yang, 2009 was attributable to three exclusively female types of cancer (breast, ovarian and endometrial); and d) the cost of morbid obesity in women was five times higher than in men, probably because most of the relative risks (or OR) were higher in women with morbid obesity and most bariatric surgeries conducted by the public health system in 2011 were for women. The costs related to bariatric surgery totaled $17.4 million (27.1% of morbid obesity costs). Compared with the findings of a study [41] conducted between 1999 and 2006, we found that in 2011 more bariatric surgeries were done than in previous years, resulting in higher costs for the Brazilian health system.

Cost of Obesity in Brazil
It is generally accepted that the Brazilian public health system has a large unmet demand for bariatric surgery. It is estimated that 1.5 million adults in the country have morbid obesity and only 26,853 surgeries were performed between 2003 and 2011 in the Brazilian public health system [5,35]; this in a country where about 75% of the population are dependent on public health services.
When we compare the cost of the disease with its prevalence in the population, we find that the cost of morbid obesity (with a prevalence of 0.81% and a cost of $64.2 million) was proportionally 4.3 times higher than the cost of obesity (with a prevalence of 14.8% and a cost of $269.6 million). This is consistent with the findings of other studies [1,27,42,43], which show a direct correlation between rising costs and higher BMI, except that we found the proportion of costs related to morbid obesity to be even higher than in other surveys [44][45][46]. In this study, it was not possible to calculate per capita costs as the Brazilian health information systems use procedures, not individuals, as a unit of analysis, since they were developed for the purposes of reimbursing costs of procedures rather than for epidemiological analyses.
This analysis did not aim to calculate the indirect costs of obesity (sick leave, reduced productivity at work, premature death), but it is important to remember that there are other costs involved with this health problem. Obese people tend to be hired less, earn lower wages and take longer to be promoted than people of normal weight, as well as less likely to receive appropriate preventative healthcare such as cancer screening tests [22]. Obese individuals can suffer from reduced quality of life, depression, anxiety and low self-esteem, characteristics that are more common among extremely obese people [22]. These are just some examples of the indirect and intangible costs attributable to obesity that also need to be considered in a more comprehensive analysis. For example, when compared to normal weight individuals (20 kg/m 2 BMI < 25 kg/m 2 ), obese adults (BMI ! 30 kg/m 2 ) have been found to make 38% more visits to primary care doctors and moderately and severely obese people have been found to have 34% and 74% more days of hospitalization, respectively [49]. Also, individuals with a BMI of greater than 30 kg/m 2 have been found to receive 1.84 more medication dispensations in general and 3.4 times more for cardiovascular diseases [49]. When obesity prevention strategies fail (physical activity, intake of healthier foods, multidisciplinary care, medication), bariatric surgery can still be an effective way of reducing morbimortality and costs. Crémieux et al [50] estimated the return on investment for bariatric surgery from the insurer's perspective in the United States, indicating that costs were fully recovered after 53 months; when the procedure was laparoscopic rather than laparotomic, the cost recovery period dropped to 25 months. In Brazil, Sussenbach et al [51] showed that bariatric surgery reduces associated diseases: 36 months after surgery, patients showed 97% resolution for type 2 diabetes mellitus, 98% for systemic arterial hypertension and 95% for dyslipidemias. In terms of costs, the authors found that average costs (medication, professional Cost of Obesity in Brazil care, examination) after surgery were substantially lower than before it: 36 months after bariatric surgery, costs reduced to 30% for patients without comorbity, 19% for patients with one comorbity, and 32% for patients with two or more comorbities [51].

Study limitations
We understand that it would be ideal to use relative risks (RR) from studies conducted in Brazil. However, while the absolute prevalence of some obesity-related diseases varies from country to country, the relative risk of any particular disease (risk for obese people compared with risk for normal weight people) is quite similar around the world [52]. Despite the fact that the PAR for morbid obesity is probably greater than for obesity for diseases such as ovarian cancer, gastric cancer, congestive heart failure, pulmonary embolism, pancreatitis, chronic back pain, pancreatic cancer, gallbladder cancer, cholecystitis and cholelitiasis, the same PAR value for obesity was adopted to calculate morbid obesity costs since we could not find the RR associated with these pathologies in the BMI ! 40 kg/m 2 range. It is also important to consider that not all the expenses incurred by the public health system are recorded in information systems, such as funds transferred for the distribution of medicines for diabetes and hypertension in the primary healthcare network.

Conclusion
The direct cost of obesity in adults to the public health system in Brazil in 2011 was estimated at US$ 269.6 million. This covers the cost of medium-and high-complexity procedures involved in treating obesity and diseases attributable to obesity. These costs attributable to obesity corresponded to 1.86% of all Ministry of Health's expenditure related to hospital and ambulatory care in Brazil (59.2% of the total was hospital admissions and 40.8% corresponded to ambulatory procedures). The costs attributable to obesity in women were double the costs for men and the cost of morbid obesity in women was five times higher than it was in men. The obesity-attributable costs were highest for ischemic heart disease, followed by breast cancer, congestive heart failure and diabetes. When analyzed by severity, the cost of morbid obesity was found to be proportionally higher than the cost of obesity. Despite its low prevalence in the population (0.81%), morbid obesity now accounts for almost a quarter of the costs attributable to obesity in the public health system, totaled $64.2 million in 2011. Bariatric surgery was the single highest cost, accounting for $17.4 million, or 27.1% of costs attributable to morbid obesity.
If the current epidemic were not reversed, the prevalence of obesity in Brazil will increase gradually in the coming years, as well as its costs, having serious implications for the financial sustainability of the public health system. With regard to public policies, it is essential to strengthen the actions already started in Brazil for the prevention and control of obesity, such as: a) increasing the supply and availability of healthy foods, including the reduction of prices of fruit and vegetables; b) the promotion of healthy environments to facilitate adequate food choices and physical activity; c) adoption of effective interventions at all points of the health care network in order to ensure the completeness of obesity care in the public health system; d) food and nutrition education, as the publication of the Food Guide for the Brazilian population and its dissemination; e) restriction of unhealthy food marketing, especially for children.
Studies of this nature need to be conducted on a regular basis in Brazil in order to monitor the economic impacts of the obesity epidemic in the coming years.