Burden of Disease and Cost of Illness of Overweight and Obesity in Portugal

Overweight and obesity rates are at epidemic proportions. A recent pooled analysis of worldwide trends in body mass index, underweight, overweight and obesity revealed that between 1975 and 2016, adults continued to gain weight in most western countries. The worldwide number of adults with obesity increased from 100 million in 1975 to 671 million in 2016; with an additional 1.3 billion adults in the overweight range in 2016 [1]. This epidemic is taking a notorious proportion in the global south due to population ageing, urbanization, and lifestyle changes [2].

Overweight and obesity are also a risk factor for a plethora of non-communicable diseases resulting from endocrine and metabolic changes, including cardiovascular disease, type II diabetes, and cancer [3]. As a result, persons with overweight and obesity are more likely to experience lower quality of life [4], have reduced life expectancy [5], demand more outpatient visits, be more often admitted to hospital, and to have more medications prescribed than other persons [6‒9].

Several international studies have estimated the disease and economic burden of overweight and obesity [10] and have reported the healthcare costs of obesity to be a significant proportion of total healthcare expenditures (ranging between 2% and 8%) [11]. In Portugal, a few studies have investigated the costs of overweight and obesity, considering both direct and productivity costs [12‒14], and out of pocket expenditures [15]. However, no studies have estimated the disease burden of overweight and obesity in Portugal.

According to results from the most recent Portuguese National Health Survey with Physical Examination (NHSPE) conducted in 2015 [16], the prevalence of overweight among adults aged between 25 and 74 years was 39.1%, and the prevalence of obesity was 28.6% [17]. Given the high prevalence estimates, overweight and obesity are expected to have a large disease and economic burden on the Portuguese population.

Burden of disease and cost of illness studies are useful in supporting policy and decision-making in the allocation of resources. In this study, the objective was to estimate the disease burden and healthcare costs associated with overweight and obesity in the adult population in mainland Portugal.

Using a prevalence-based methodology, this study estimated the disease burden and healthcare costs of overweight and obesity for the year 2018, considering the adult population aged 18–84 years in mainland Portugal. This year was chosen to reflect the year of the most complete healthcare data in Portugal.

Following the Global Burden of Disease (GBD) methodology [18], healthcare costs were valued in monetary units and disease burden in disability-adjusted life years (DALYs), including the impact of morbidity and mortality directly related to overweight and obesity as well as the attributable impact of morbidity and mortality of a set of 25 diseases for which overweight and obesity are a risk factor (henceforth referred to as diseases related to obesity, DrO). These diseases were selected from the latest 2019 GBD study [19], which identified 25 DrO. Of these 25 diseases, the following 24 were included: type II diabetes, cardiovascular and cerebrovascular diseases (ischaemic heart disease, stroke, hypertensive heart disease, and atrial fibrillation), cancer (oesophagus, colon and rectum, liver, gallbladder and biliary tract, pancreas, breast, uterus, ovary, kidney, thyroid, multiple myeloma, and leukaemia), musculoskeletal diseases (osteoarthritis, gout), Alzheimer’s disease, chronic kidney disease, diseases of gallbladder and biliary tract, asthma, and cataracts. Low back pain was excluded due to lack of reliable national data sources. Depression was additionally included, based on expert opinion.

The prevalence of overweight and obesity was sourced from the National Health Survey with Physical Examination (NHSPE) from 2015 [20]. This is a nationwide survey of the Portuguese population aged 25–74 years, which aims to improve knowledge on the health status, health determinants, and use of healthcare services. Data were collected in 2015 between February and December and included anthropometric measurements (blood pressure, height, weight, waist and hip circumference), a blood collection, and a computer assisted personal interview. The estimated prevalence was 38.9% for overweight (45.4% men vs. 33.1% women) and 28.7% for obesity (24.9% men vs. 32.1% women). This source was complemented by data from the National Food, Nutrition and Physical Activity Survey, which included – amongst other variables – measurements of height and weight of adults between 18 and 84 years old.

RR for each DrO was sourced from the GBD 2019, except for depression which was sourced from the NHSPE. PAFs were later applied to the burden and costs of each DrO to estimate the burden and cost attributable to overweight and obesity. Disease burden was estimated by sex and age group, allowing the use of PAF by sex and age group. However, costs were not available by sex and age group, thus average PAFs were considered (Table 1).

YLL are estimated as the product between the number of deaths and the average life expectancy at which death occurs, assuming the standard life expectancy from the life table developed for the GBD 2019 study [22]. Case fatality for DrO was sourced from the Mortality Database from the Portuguese Health Directorate [23]. The International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes were used to identify mortality for DrO. YLL estimates for acute leukaemia were sourced from previously published work by the authors [24, 25]. For comparative purposes, YLL resulting from all-cause mortality were also estimated [26].

YLD are estimated as the product between the number of patients and the relevant disability weight (DW). A DW is a measure of health loss associated with a certain health outcome. It is measured on a scale from 0 to 1, where 0 represents a state of full health and 1 represents death. DW was sourced from the latest GBD 2019 [19]. Most prevalence estimates for the DrO were sourced from official documents published by the Ministry of Health and from published national epidemiological studies. For acute DrO with short duration, incidence estimates were used as proxies, sourced from the Hospital Morbidity Database. This is a national registry of all public hospital-related care, both inpatient and ambulatory care. Approximately 70% of all inpatient hospital admissions take place in public hospitals [27].

Type II Diabetes. The prevalence of type II diabetes for adults aged 20–79 years was sourced from a study by the Portuguese National Diabetes Observatory [28] which estimated the prevalence by sex and age group in the Portuguese population for the year 2015, using microdata from the PREVADIAB study from 2009 [29]. The DW for type II diabetes considered only patients without complications and those with microvascular complications (retinopathy and neuropathy) to avoid double counting of macrovascular complications included separately. Microdata from the primary care database (SIARS) of the Regional Health Administration of Lisbon and Tagus Valley including data from 195,926 diabetic patients was used to get the proportion of patients without complications (8.4%), and the proportion of patients with retinopathy (4.6%). Expert opinion was used to estimate the proportion with moderate (67%) and severe (33%) retinopathy. The proportion of diabetic patients with total blindness was 0.4% [30]. The proportion of patients with foot ulcer (0.2%) was used as proxy for the prevalence of neuropathy and was sourced from the primary care dashboard from the Ministry of Health [31]. As national data on the prevalence of amputations was not available, this was estimated using DISMOD-II model [32], which estimated a prevalence of 0.6% and 0.2% for minor and major amputations, respectively.

Cardiovascular and Cerebrovascular Diseases. The following diseases were considered: ischaemic heart disease (including acute myocardial infarction and angina), stroke (including ischaemic and haemorrhagic), hypertensive heart disease, and atrial fibrillation. The prevalence of stroke and angina was sourced from National Health Survey 2019. The incidence was based on the Hospital Morbidity Database. The distribution of patients across levels of disability was retrieved from the GBD 2016 study [33]. A DW due to acute stroke was considered for new cases and a DW for chronic stroke for patients who developed stroke in previous years. A weighted average DW for ischaemic and haemorrhagic stroke of 0.241 and 0.235, respectively, was estimated.

Due to lack of relevant national data, the distribution of cases of angina according to severity levels was obtained from the AVANCE registry, a Spanish observational study with 2,039 persons with angina [34]. This study was chosen given the similarities of population characteristics and cardiovascular disease between countries. A weighted average DW of 0.074 was estimated.

The prevalence of myocardial infarction (MI) was assumed to correspond to the incidence as the disability due to myocardial infarction is often associated with the acute phase of the disease, corresponding to the first 28 days [35]. Incidence was sourced from the Hospital Morbidity Database. The DW associated with an episode of MI was estimated as the average of the DW in the 2 days after the event (DW = 0.432) and the DW in the following 26 days (DW = 0.074). This was further converted into an annual DW of 0.007.

Hypertensive heart disease was defined in the GBD 2019 study as symptomatic heart failure due to the direct and long-term nature of hypertension [19]. Prevalence estimates for heart failure were sourced from the EPICA study [36], a community-based epidemiological survey on heart failure prevalence including 5,434 adults aged 25+ attending primary care centres. From this study, we considered the number of symptomatic patients with hypertensive aetiology according to the New York Heart Association (NYHA) Classes II–IV. The NYHA classes II–IV were assumed to correspond to light, moderate, and severe disability, allowing to estimate a DW of 0.045 for hypertensive heart disease.

The prevalence of atrial fibrillation was sourced from two atrial fibrillation studies in Portugal (SAFIRA [37] and FAMA [38]). The proportion of symptomatic patients was sourced from the Hospital Morbidity Database. An average DW of 0.061 was estimated.

Cancer. The following cancers were considered: oesophagus, colon and rectum, liver, gallbladder and biliary tract, pancreas, breast, uterus, ovary, kidney, thyroid, multiple myeloma, and acute leukaemia (acute lymphocytic and acute myeloid leukaemia). The prevalence of all cancers, except leukaemia, was sourced from Globocan 2020 [39]. The prevalence of acute leukaemias was sourced from published literature [24, 25].

The GBD 2019 considers DW corresponding to different stages of disease progression for most cancers (diagnostic and treatment, post treatment remission, metastases, and terminal disease). For breast and colorectal cancers, additional DW was considered in the GBD, pertaining to the presence of mastectomy or stoma. The proportion of patients in each disease stage was estimated using data from the Surveillance, Epidemiology, and End Results Program (SEER) [40] and incidence reported in the National Cancer Register 2018 [41]. The following average DW were estimated for the various cancers: 0.149 (colorectal), 0.113 (uterine), 0.220 (ovarian), 0.176 (kidney), 0.100 (thyroid), 0.244 (multiple myeloma), 0.434 (oesophageal), 0.421 (gallbladder and bile duct), 0.454 (pancreatic), and 0.403 (liver). For breast cancer, the distribution of patients across disease stages was based on Sousa et al. [42], allowing to estimate a DW of 0.124.

Musculoskeletal Diseases. The prevalence of hip and knee osteoarthritis and gout was sourced from the EpiReumaPt study [43], a national epidemiological survey conducted between 2011 and 2013 on the prevalence of rheumatic diseases in a sample of 10,661 adults. DW was retrieved from the GBD 2016 study [33] along with the distribution of prevalent cases across the different levels of disability. Weighted average DW was estimated at 0.060 and 0.056 for hip and knee osteoarthritis, respectively, and 0.035 for gout.

Diseases of the Nervous System. Both depression and Alzheimerʼs disease were considered. Prevalence for depression was sourced from the National Health Survey 2014 [44], which estimated a prevalence of 11.9% among persons aged 15 and older. A weighted average DW was estimated at 0.257, according to the DW for different severity levels of depression (mild, moderate, and severe) sourced from the GDB and the distribution of cases of depression by severity level sourced from Statistics Portugal 2019. The prevalence of Alzheimerʼs disease was sourced from a previous study that estimated the disease and economic burden for mainland Portugal among persons aged 65+ [45].

Other Diseases. Chronic kidney disease (CKD), gallbladder and bile duct disease, asthma and cataracts were also included. The prevalence of CKD across different disease stages 1–5 was sourced from the RENA study [46], which included 3,135 adults attending primary care centres. DW was sourced from the GBD for different stages of CKD. These DW were multiplied by the distribution of patients across disease stages [46, 47] to estimate an average DW of 0.014.

The incidence of gallbladder and biliary tract diseases, sourced from the HMD 2017, was used as proxy for the prevalence, as national data were unavailable. A DW of 0.027 was used, which corresponded to the annual disability accrued from a symptomatic episode with a duration of 1 month.

The prevalence of asthma was sourced from the National Asthma Survey [48], which estimated it based on a self-report of 6,003 participants. The GBD considered three DW corresponding to controlled, partially controlled and non-controlled asthma. The distribution of the prevalence of asthma across these different levels was sourced from a national cross sectional study [49]; a weighted average DW of 0.046 was estimated. The prevalence of cataracts was sourced from the HMD 2017, with no DW attributed.

The cost analysis adopted the perspective of the National Health Service (NHS) and included direct medical costs related to inpatient and specialized outpatient care, primary care, and pharmacological treatment. Total costs included medical costs directly related to overweight and obesity, as well as the proportion of medical costs related to DrO attributable to overweight and obesity. PAF previously described were applied to estimate these costs. All unit costs were retrieved from official sources [50‒52].

Inpatient admissions were sourced from HMD 2017. The ICD-10-CM codes used to identify all relevant admissions are outlined in Table 2.

Obesity. Resources related to treatment of obesity were sourced from the Programa de Tratamento Cirúrgico da Obesidade (PTCO) [53]. This programme is funded by the Portuguese National Health Service (NHS) and provides special access to treatment for patients with obesity. Medication costs were sourced from an IQVIA database (from 2020), where medication sales specifically targeting obesity and weight reduction were retrieved. For retails drugs, the fractions paid by the NHS and by the patients were considered.

Diseases Related to Obesity. Resource use for the DrO were quantified using a “bottom-up” approach based on expert opinion, previously published work, and Ministry of Health documents [54‒56]. Medication costs related to the DrO were based on official sources and the IQVIA sales database.

Table 3 shows the total number of deaths and YLL due to DrO in relation to total mortality and YLL due to all other causes. The total number of deaths due to DrO, within the age groups included, was estimated at 20,080, which corresponds to 32% of overall mortality. We estimated a total of 86,694 YLL attributable to overweight and obesity, which derives from the product of the total YLL due to DrO by the PAF for each disease. This corresponds to 20% of total YLL due to DrO (444,183 YLL) and 6% of YLL due to overall mortality (1,395,233 YLL). Type II diabetes (46%), chronic kidney disease (38%), and gout (38%) were the DrO with largest PAF. However, due to being the cause of more deaths, the largest contributor to total YLL was cardiovascular and cerebrovascular diseases (59%), followed by cancer (24%), and type II diabetes (8%). Overall, the impact of premature death was 5,456 YLL per 100,000 inhabitants.

Table 4 shows total YLD for each DrO as well as the proportion attributable to overweight and obesity. A total of 406,499 YLD were due to DrO, 24% of which (96,654 YLD) were attributable to overweight and obesity. Type II diabetes had the largest PAF (57%), followed by chronic kidney disease (41%) and gout (30%). Depression was the DrO with more YLD attributable to overweight and obesity (42.115 YLD; 44%) followed by cardiovascular and cerebrovascular diseases (19,688 YLD; 20%) and osteoarthritis (12.019; 12%). Prevalence and DW for each DrO are available in Table 5.

The overall burden of overweight and obesity was 183,348 DALY. YLL were largest among men, whereas YLD were largest among women, with overall largest burden for women (55% of total). Total YLD represented 53% of total DALY (96,654 YLL) (see Table 6).

Table 7 presents the total costs attributable to overweight and obesity and to the included DrO. The total economic burden of overweight and obesity was estimated at 1,147 million euro. Of these, only 1% were related to the treatment of obesity per se. Outpatient care costs corresponded to 43% of the total cost (491 million euro), followed by costs related to pharmacological treatment (38%; 438 million euro) and inpatient care (19%; 218 million euro). Type II diabetes (79%), cardiovascular and cerebrovascular diseases (43%) and chronic kidney disease (39%) were the DrO with largest PAF, as well as the DrO with largest contribution to costs (cardiovascular and cerebrovascular diseases were responsible for 38% of total cost, type II diabetes for 34%, and chronic kidney disease for 12%). Figure 1 shows the distribution of the total costs attributed to overweight and obesity by DrO.

The present study is the first to estimate the disease burden of overweight and obesity in Portugal. It also presents updated estimates of the economic burden of overweight and obesity considering the NHS perspective. Following the GBD methodology, these estimates consider the morbidity, mortality and costs directly related to overweight and obesity, as well as the morbidity, mortality, and related costs of various related diseases.

The impact of overweight and obesity in Portugal is substantial, with an estimated prevalence of 38.9% for overweight and 28.7% for obesity [20]. In 2018, overall mortality caused 1,395,233 YLL, with 96,654 (6%) attributable to overweight and obesity. Although the DrO with largest PAF were type II diabetes, cardiovascular and cerebrovascular diseases were the largest contributor to YLL. This reflects the large number of deaths related to these diseases in the population.

Overweight and obesity were also responsible for 96,654 YLD. Similarly, despite type II diabetes being the DrO with largest PAF, depression and cardiovascular and cerebrovascular diseases were the DrO with more YLD attributable to overweight and obesity, partly due to their large prevalence in the population.

The total burden attributable to overweight and obesity was estimated at 183,348 DALY, which corresponds to about 13 days per adult with overweight or obesity. YLD due to disability contributed slightly more to the overall burden (53%).

The total economic burden of overweight and obesity totalled 1,147 million euro, with the vast majority of costs related to DrO, and only 1% due to costs related to treatment of obesity per se. Similarly to the disease burden, cardiovascular and cerebrovascular diseases, and type II diabetes were the largest contributors to total costs. Altogether, they represented over 83% of the total economic burden. Total costs represented approximately 0.6% of the gross domestic product for Portugal and 5.8% of the health budget allocated by the government. The proportion of the health budget is larger than that of other European countries, namely the Netherlands (4%), UK (2%), Sweden and France (2%). Previous estimates of the economic burden of obesity to the healthcare sector in Portugal, from Pereira & Mateus, 1999, reported that costs related to pharmacological treatment corresponded the biggest proportion of the total cost (43%), which in the present study corresponded to 47% of total cost. In 1999 values, healthcare costs of overweight and obesity were estimated at approximately 268 million euro compared to the 1,147 million euro from the present study.

The estimates presented in this study highlight the current impact of overweight and obesity on population health and on the healthcare budget. They also highlight the need for governments to take action to reduce overweight and obesity. Although disease and economic burden studies are useful in themselves to quantify the impact of diseases and demonstrate the potential gains of reducing their prevalence, they do not offer guidance as to which strategies may best achieve this goal. Studies investigating the cost-effectiveness of interventions targeting overweight and obesity can better support decision-makers invested in the reduction of prevalence of this public health problem. Obesity is a complex chronic disease process which results from the interaction of multiple factors including nutrition, physical activity, and genetic susceptibility [3], which are also intertwined with other socioeconomic factors.

Efforts to reduce the impacts of overweight and obesity should be a combined endeavour of both individuals and governments. Investment in treatment of obesity as well as interventions targeting the full spectrum of social determinants of obesity is called for. Evidence exists on the cost-effectiveness of interventions targeting obesity and overweight [57]. The World Health Organisation has published an overview of “best-buy” strategies including community public education and awareness for physical activity and taxes on sugar-sweetened beverages and front-of-package labelling [58]. Policies influencing price or availability of fruits and vegetables [59] are also cost-effective. Making these and other strategies available to populations pose challenges, considering the diverse demographic, cultural, and economic characteristics across the globe [60]. Given the large epidemiological burden of overweight and obesity and population ageing [61], governments need to consider sustainable and equitable health financing solutions to addressing the burden of overweight and obesity. Some of these include increasing public health spending, improving efficiency, promoting equity in access to care [2], as well as promoting prevention and intervention across multiple sectors of the economy [62]. A crucial aspect for many countries around the globe and in particular in the Global South, is achieving universal health coverage to ensure people have access to the care they need. There are several challenges to achieving universal health coverage but many countries are working towards this goal [63, 64]. Bringing together the global north and the global south to tackle this epidemic [65] should pave the way forward.

Some limitations should be considered. One limitation pertains to the availability of data to estimate the prevalence of overweight and obesity in the Portuguese population, which was limited to the upper bound of 84 years of age. Additionally, several data sources were used to estimate the prevalence and mortality of DrO. For oncological diseases, assumption had to be made concerning the distribution of prevalence across disease stages to use the DW presented in the GBD. However, cancer-related YLD represent only 1% of the total YLD attributable to overweight and obesity. Furthermore, we were unable to distinguish between different causes of CKD and isolate diabetes-related CKD due to the lack of more specific data. This overlap may have led to potential double counting in disease categories, affecting the precision of our estimates. Future research with more granular data would be necessary to refine these classifications and improve the accuracy of the results.

Another limitation pertains to the lack of data on outpatient care for which no specific database was available. The authors used the best available evidence to tackle this limitation and used several data sources from databases, literature, and expert opinion. This study has also adopted a narrower healthcare payer perspective to estimating the economic burden of overweight and obesity, excluding costs beyond healthcare. Overweight and obesity have impacts largely beyond the health sector, including lost productivity and social care costs, thus the total burden estimates are undoubtedly larger. A study by Okunogbe et al. [66], estimated the economic impacts of overweight and obesity in eight countries and found that, in Spain, healthcare costs represent only 31% of the total costs of obesity to society. Future studies should consider broader perspectives. Finally, longitudinal studies using microdata and with population controls are important to be able to estimate the excess cost of overweight and obesity in relation to the general population.

These findings highlight the large disease and economic burden of overweight and obesity, as they are related to the development of several non-communicable diseases, including type II diabetes, cardiovascular and cerebrovascular diseases, and cancer. Reducing the prevalence of overweight and obesity should be a priority, which may result in better population health, less burden on the public healthcare system and substantial savings to society.

We would like to acknowledge Administração Central do Sistema de Saúde, IP for providing access to the national diagnostic-related-group database for 2017 and 2018, as well as the Instituto Nacional de Saúde Doutor Ricardo Jorge (INSA) for providing access to the microdata from the National Health Examination Survey (INSEF). We would also like to thank Sociedade Portuguesa para o Estudo da Obesidade for scientific sponsorship.

This study protocol was reviewed and approved by the Ethics Committee of National Institute of Health Doutor Ricardo Jorge, Lisbon, Portugal, date of decision May 27, 2021. Patient consent was not required as this study was based on publicly available data.

The authors declare: M.B., D.F., B.P., and L.S.M. are employees of IQVIA. F.S. is employed by Uppsala University. J.C. is a member of Laboratório de Farmacologia Clínica e Terapêutica, Faculdade de Medicina, Universidade de Lisboa. Both J.C. and F.S. are external consultants of IQVIA. V.C. is employed by Novo Nordisk. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

This study was funded by Novo Nordisk Portugal, Lda. Funding was independent of the study outcomes.

J.C., M.B., and L.S.M. conceived the study and supervised all aspects of its implementation. D.F. and B.P. collaborated in the inception of the study and carried out data analysis. F.S., J.C., M.B., L.S.M., and V.C. drafted and revised the article. P.F., C.M.D., and V.G. contributed with clinical and epidemiological expertise. All authors contributed to the interpretation of results and approved the final manuscript.

There is a restriction applied to the data that support the findings of this study; therefore, the data are not publicly available. Data are, however, available from the authors upon contract agreement and with the permission of Novo Nordisk. Contact the corresponding author.