Economic burden of physical inactivity in hospitalizations due to dementia: a Brazilian nationwide study

Our study aimed at determining the economic burden of leisure-time physical inactivity in hospitalizations due to dementia in Brazil. In this national-wide descriptive study, we used secondary data from the Brazilian Health Informatics Department and from the latest Brazilian National Health Survey (2013). Rate of hospitalization and economic cost were extracted from 2013. Leisure-time physical inactivity prevalence was used to estimate its population attributable fraction for dementia. Outcomes were stratified into sex (man/ woman), age groups (< 50, 50-59, 60-69, 70-79, 80+), ethnicity (white, black, mixed race, other), and region (South, Southeast, North, Northeast, and Central). In 2013, 3,724 people were hospitalized due to dementia in Brazil resulting in a total cost of BRL 17,971,833.85 (USD 7,673,973.05). More than half of cost among men was spent with individuals younger than 59 years, white and mixed-race, and from Southeast region. Proportion of cost was higher in women older than 60 years and white. In 2013, 37% of all hospitalization from dementia in Brazil were attributable to physical inactivity, with values reaching 42.9% and 43.8% for men and women aged 80 or older, respectively. Physical inactivity cost BRL 6,994,254.75 (USD 2,986,546.78) in 2013 on hospitalizations due to dementia in Brazil. Higher attributable-cost was found for patients younger than 50 years and women older than 80. Strategies to preserve the quality of life of these patients are needed to reduce the burden for the patients, their family, and on the healthcare system. Demetia; Exercise; Hospitalization Correspondence N. Feter Rua Luís de Camões 625, Pelotas, RS 96055-630, Brasil. natanfeter@hotmail.com 1 Universidade Federal de Pelotas, Pelotas, Brasil. 2 The University of Queensland, St. Lucia, Australia. 3 Colégio Militar de Porto Alegre, Porto Alegre, Brasil. This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited.


Introduction
Dementia is a chronic and degenerative syndrome that affected around 50 million people in 2019 1 . Population growth, ageing, and increased prevalence of dementia-related risk factors lead Brazil to the second highest age-standardized prevalence of dementia in the world, affecting about 1.7 million people 2 . Whereas the number of deaths attributable to dementia decreased in the last 10 years in some Latin American countries 2 , the mortality rate from this disease increased annually about 12% in men and 13% in women in Brazil from 2000 to 2008 3 .
Moreover, patients with dementia have higher prevalence of comorbidities compared to cognitively healthy older people 4 . Besides leading to increased rate of hospitalization 5 , the treatment of those comorbidities is more complex 6 , expensive 7 , and unequal 4 in patients suffering dementia, resulting in longer and more costly hospital admissions 4 . Consequently, longer hospitalizations are linked to higher incidence of cognitive decline that may become irreversible if treated incorrectly 8,9 .
The total individual cost per year of patients with dementia in Brazil is greater than the global average in all disease stages 10 . In 2016, only 34% of total health financing in Brazil was spent by the government. However, more than 65% of hospitalizations due to dementia were covered by the Brazilian public healthcare system (Brazilian Health Informatics Department -DATASUS. http://www2. datasus.gov.br/DATASUS/index.php, accessed on Mar/2020). In addition, around 50% of people living with dementia in Brazil have another chronic condition 10 . Those facts show the recent past, current, and future social and economic burden of dementia in Brazil.
Although ageing is the strongest risk factor for dementia 11 , more than 50% of all its cases in low-and middle-income countries (accounting for about 70% of all dementia cases in the world 12 ) could be attributable to the following modifiable risk factors: smoking, diabetes, obesity, hypertension, depression, low schooling, hearing loss, social isolation and physical inactivity 13 . Indeed, a 20% reduction in the prevalence of those risk factors would lead to a reduction of 18.1% in the prevalence of dementia in Brazil by 2050 14 .
Despite being associated with lower risk of cognitive decline in later life 15,16 , the prevalence of regular leisure-time physical activity did not increase between 2000 and 2015 in any region of the world 17 . Physical inactivity is characterized in adults as do not achieve at least 150 minutes of moderate-intensity aerobic physical activity per week, or 75 minutes of vigorous-intensity aerobic physical activity per week, or an equivalent combination of both 18 . Although physical inactivity among older people was associated with poor cognitive function 15 , high level of dependency 19 , and increased allcause mortality risk 20 , its prevalence is still high among older adults in Brazil 21 .
Bielemann et al. 22 reported that 14% of cost from hospitalizations due to diabetes, some types of cancer, cardiovascular diseases, and osteoporosis in Brazil were linked to physical inactivity; however, there is no data about its impact on dementia-related hospital admission. Then, we aimed to determine the economic burden of leisure-time physical inactivity in hospitalizations due to dementia in Brazil.

Method
To conduct this national-wide descriptive study, we used secondary data from the Hospital Information Systems of the Brazilian Unified National Health System (SIH/SUS, in Portuguese) (DATASUS. http://www2.datasus.gov.br/DATASUS/index.php, accessed on Mar/2020) and from the latest Brazilian National Health Survey (PNS, in Portuguese) 23 . PNS was approved by The National Commission of Ethics in Research (Regulation n. 328,159) 23 . DATASUS is an open-access non-nominal database, so no ethical approval is needed to extract data from this database.
Nationalwide data about hospitalization and total economic cost from these hospitalizations in Brazil during 2013 were extracted from SIH/SUS. We selected 2013 because it was the latest year with population-based data about physical activity in Brazil measured by validated instrument (International Physical Activity Questionnaire -IPAQ) 24 . Number and cost of hospitalizations was extracted from DATASUS (http://www2.datasus.gov.br/DATASUS/index.php, accessed on Mar/2020). Total hospitalizations were considered as all new authorized hospitalization forms, transferences, and re-Cad. Saúde Pública 2020; 36(1):e00046520 hospitalizations related to dementia. Total economic cost was considered as the value referred to all services to all hospitalizations throughout a specific period (2013). When applicable, we converted Brazilian Real (BRL) to US Dollar (USD) based on conversion rate in data registration year (December 31th, 2013): BRL 1.00 = USD 0.427.
The cause of hospitalizations was informed as the main diagnose that motived them. For dementia, we considered all hospitalizations from which the main diagnose was defined as dementia in Alzheimer's disease (ICD-10: F00; G30), vascular dementia (ICD-10: F01), dementia in other disease classified elsewhere (ICD-10: F02), and unspecified dementia (ICD-10: F03), according to International Classification of Diseases, 10th revision (ICD-10) 24 . We also stratified these outcomes based on sex (man/woman), age groups (less than 50 years, 50-59, 60-69, 70-79, 80 or older), ethnicity (white, black, mixed race, other -yellow, indigenous, and unknown), and region (South, Southeast, North, Northeast, and Central). Age was estimated based on date of birth and date of admission.
Rate of hospitalization was determined by dividing the number of hospitalizations by the total population and from each group (sex, ethnicity, and region). Data about total population in 2013 was extracted from Brazilian Institute of Geography and Statistics (IBGE, in Portuguese) 25 .
To determine the economic burden of physical inactivity in hospitalizations due to dementia, we calculated the population attributable fraction (PAF) for this exposure. We used the formula below based on previous publication 13,22,26 . The PAF aims at identifying the percental reduction in the prevalence of certain disease or health condition (e.g. hospitalization due to dementia) if certain risk factor (e.g. physical inactivity) was eliminated.
Where, p: prevalence of physical inactivity and RR: relative risk. We used the RR calculated by Hamer & Chida 27 in a meta-analysis with a sample of 23,168 participants from different countries (Finland, United States, China, Sweden, Canada, and France). The study identified that physical inactivity was associated with a RR of 1.82 (95%CI: 1.19; 2.78) for dementia. All included studies reported adjusted RR by covariates such as age, gender, and education level.
Self-reported physical activity behavior was extracted from PNS. This cross-sectional populationbased survey was to date the most complete health survey already conducted in Brazil. It covered data about sociodemographic, health status, behavior, and healthcare utilization among Brazilian adult population (i.e. aged 18 or older). Due to its three-stage probabilistic sampling and use of weighted estimates, the information from PNS can be extrapolated to all Brazilian adult population 23 . From 81,167 households included in the PNS, 60,202 individuals completed interviews about lifestyle and health status (response rate: 74.2%). As mentioned earlier, physical activity was measured by the IPAQ, which includes questions on frequency and duration of physical activity in leisure-time. A final score was generated from multiplying the weekly frequency of physical activity by its duration in minutes. Participants that did not achieved at least 150 minutes of moderate physical activity or 75 minutes of vigorous physical activity or an equivalent combination of moderate and vigorous activity per week in leisure time were classified as physically inactive 18 .

Results
In 2013, 3,724 people were hospitalized due to dementia in Brazil. Among men, 61.2% of admissions were of adults aged 69 years or less and 61.5% in Southeast region. Most of hospitalizations from dementia in women occurred in patients older than 70 years (55.2%), white (44.2%), and from Southeast region (64%) (data not shown). Table 1 shows that, although the rate of hospitalizations increases with ageing, men had higher values before aged 80 compared to women.
In Brazil, the total cost of hospitalization from dementia in 2013 was BRL 17,971,833.85 (USD 7,673,973.05). More than half of cost among men were spent with individuals younger than 59 years (54.6%), white and mixed (51.5%), and from Southeast region (87.9%). Women and men had similar distribution among regions, but the proportion of cost was higher in patients older than 60 years (57.8%) and white (53.7%).
In 2013, 77.7% of Brazilian population were physically inactive during leisure time with higher prevalence among women ( Table 2). Men aged less than 50 years had lower prevalence of physical inactivity compared to other male and to female age groups. Men and women aged 70 or older were more physically inactive compared to any younger group from the same sex. Women were more inactive than men regardless of ethnicity and region.
About 37% of all hospitalization from dementia in Brazil were attributable to physical inactivity in 2013, with values reaching 42.9% and 43.8% for men and women aged 80 or older, respectively, and 41% and 40.8% for black women and in North region, respectively. Table 3 shows that the economic burden of physical inactivity in hospitalizations due to dementia was estimated in BRL 6,994,254.75 (USD 2,986,546.78). White patients and at Southeast region had higher cost attributable to physical inactivity regardless of sex. While physical inactive accounts for greater cost among men younger than 50 years, we observed that this exposure has higher impact in hospitalizations of women aged less than 50 and older than 80 years.

Table 3
Economic cost of hospitalizations due to dementia attributable to physical inactivity. Brazil, 2013 (n = 3,724).  43.5% of all dementia-related hospitalizations in patients aged 80 years or older. Also, women had more hospitalizations attributable to physical inactivity than men regardless of ethnicity and country's region.

Cost of hospitalizations for dementia attributable to physical inactivity
The number of studies that evaluated the economic impact of physical inactivity in hospital admissions in Brazil is limited. Sichieri et al. 28 stated that physical inactivity was related to nearly 36% of all diabetes-related inpatient cost in adults aged between 20 and 60 years, values that are near what we reported for overall sample and for those aged up to 60 years (37%). Bielemann et al. 22 reported that about 15% of all hospitalizations from diabetes, breast and colon neoplasm, cerebrovascular and ischemic heart disease, hypertension, and osteoporosis in 2013 were attributable to physical inactivity in Brazil, with the lowest patient fractions from South and Central regions. In our study, despite the different population attributable factor (37%) for dementia, we found the lowest values at the same regions. Our results provide a novel perspective about how physical inactivity is associated with hospital admissions due to dementia in Brazil.
Bielemann et al. 22 exposed that all hospitalizations due to non-communicable chronic disease in Brazil accounted for USD 451,987,141.82 in 2013. From this value, USD 67,395,324.61 were attributable to physical inactivity. From 2000 to 2018, deaths from Alzheimer's disease-type dementia increase about 146.2% in United States, a percentage higher than stroke, heart disease, and breast or prostate cancer1. Likewise, data from SIH/SUS showed that the number and economic cost of hospitalizations due to dementia increased more than any of beforementioned chronic diseases from 2010 to 2017 in Brazil. Although this value found by Bielemann et al. 22 was higher than that found in our study, dementia has impacted the Brazilian healthcare system in last years; however, a larger effect is expected in the next years. Therefore, strategies to reduce this disease economic and social burden are need, especially in Brazil.
The prevalence of physical inactivity, which is linked to around 18% of all dementia cases in the world 29 is persistently high in Brazilian population especially in children, adolescentes 30 and older population 21 . Women are more likely to develop dementia than men, especially in older ages 31 , what may explain the higher rate of hospitalization among women in patients aged 80 years or older. Based on our results, about 44% of hospitalizations from dementia in Brazil were linked to physical inactivity among women older than 80 years in 2017. This population stratum is affected by loss of physical and cognitive capacity, which may act as important barriers to physical activity practice 32 . Thus, public policies need to be designed as ageing-friendly interventions to promote physical activity practice and maintenance during aging.
Moreover, men had higher rate of hospitalization than women among individuals aged 69 years or less. Likewise, the proportion of hospitalizations attributable to physical inactivity was higher among men aged between 60 and 69 years than women. Alzheimer's disease is the most prevalent type of dementia1, and vascular dementia appears as the second leading cause of this syndrome 33 . However, vascular dementia has higher incidence rate among men 34 . This difference may be explained by a worse cardiovascular health in men. Hypertensive 35 and cerebrovascular diseases 36 are common causes of vascular dementia and about 57% of all premature deaths that occurred among men between 30 and 69 years in Brazil in 2017 were due to those diseases (Brazilian Ministry of Health. http://svs. aids.gov.br/dantps/centrais-de-conteudos/paineis-de-monitoramento/mortalidade/dcnt/, accessed on Mar/2020). This picture exposes the "survival bias" in dementia prevalence: men that live longer may have better cardiovascular health, what lead to a decreased risk to develop some ageing-related illness such as Alzheimer's disease 37 . Nevertheless, we still observed high rates of hospitalizations due to dementia in the oldest age group, what corroborated previous findings 1, 6 and emphasized the need to promote low-cost strategies such as physical activity to reduce the burden of dementia in Brazilian health system. Furthermore, whereas high hospitalization rates were found among women, we revealed that men reported higher economic costs from hospitalization due to dementia. One third of men in Brazil do not perform annual screenings (i.e. check-up) 38 . Common imaging medical tests such echocardio-Cad. Saúde Pública 2020; 36(1):e00046520 gram and ultrasonography could decrease the risk for cardiovascular events that are associated with dementia. Then, hospital admissions among men are more likely to be related to advanced stages of dementia, which requires more hospital services and then increased economic cost.
Although rates of hospitalizations were the highest among subjects with ignored ethnicity classification in SIH/SUS, we could observe that it was higher among white people than among black and mixed patients in both sexes. Also, though mixed-race patients had a rate of hospital admission 60% lower than black individuals, the inpatient economic cost was 2.9 times higher when compared with black patients. Despite other studies did not find a higher prevalence of dementia in this ethnic group 39,40 , it was observed an increased inpatient burden of dementia among black individuals. Even though more than half of Brazilian population is declared non-white 25 , the country is constantly facing a racial disparity. Whites have higher monthly income and schooling years than mixed subjects 25 . Likewise, non-whites are more likely to smoke, to have heavy episodic drinking, physical inactivity, sedentarism, and poor diet compared to white individuals 41 . Although the higher hospitalizations rates among white patients may be explained by higher life expectancy in this group 41 , our results indicate a racial disparity among economic cost of hospitalizations due to dementia.
Our study has important limitations. First, the SIH/SUS did not accurately provide the ethnicity group for all patients. Nevertheless, we could report some social disparity related to economic burden of dementia among mixed and white patients. Second, we reported that RR was generated using data from six different countries. However, five of them are classified as high-income countries, what could reduce the translation of this measure into the Brazilian healthcare system. However, as already mentioned by Mukadam et al. 13 , there are no meta-analyses that estimated the RR for dementia by physical inactivity using data exclusively from low-and middle-income countries. Moreover, extracting RR from meta-analysis instead of individual studies is a more precise and comparable method 42 .
Third, Levin's formula was designed to provide PAF using unadjusted RR. Here, the RR estimated by Hamer & Chida 27 was adjusted for some important confoundings such age, sex, and educational level. However, the usage of other methods was not possible. For example, the PAF formula suggested by Miettinen et al. 43 requires the prevalence of the exposure (i.e. physical inactivity) among cases (i.e. hospitalized patients with dementia). Nevertheless, no nationalwide data about physical inactivity in dementia patients from Brazil is available.
Fourth, although physical activity could be measured in different domains such as work and during work/back home trip, we decided to determine physical inactivity only by leisure time for some reasons. Firstly, due to the greater potential for future interventions in this population 44 . Secondly, the lower prevalence of physical activity in commuting domain in individuals older than 40 years in Brazilian population 45 could overestimate our results. Thirdly, the articles included in the systematic review that generated the relative risk of physical inactivity for dementia 27 and other that used similar methodology 26,29 primarily used physical inactivity in leisure time to determine the impact of this risk factor in other chronic diseases such as dementia. Lastly, we presume that this classification may prevent overestimation of our findings, enable greater comparison with other populations, and illustrated the possible impact of leisure time physical activity promotion 13,46 .
Finally, we extracted data about hospitalizations due to dementia using the DATASUS. Although it is a federal open-access nationalwide database, there are some limitations that need to be acknowledged regarding its use. First, DATASUS comprehends data from hospitalizations covered by the Brazilian Unified National Health System -SUS (i.e. Brazilian public healthcare system). Consequently, information about hospitalization in private hospitals may be missing. However, besides the fact that two of three older adults in Brazil have no private health insurance 41 , two thirds of hospitalizations due to dementia were covered by the Brazilian public healthcare system in 2013 (DATASUS. http:// www2.datasus.gov.br/DATASUS/index.php, accessed on Mar/2020). Second, DATASUS may have different levels of sub-notification across the country, which could underestimate our findings.
We conclude that 37% of economic cost from hospitalizations in Brazil due to dementia were attributable to physical inactivity in 2013, with the highest patient fractions occurring in older patients. These results enhance the perspective about the role of physical inactivity in burden of dementia and its impact on Brazilian healthcare system. Dementia is a chronic and progressive disorder with elevated economic and social cost. Therefore, strategies to preserve the quality of life of these patients are needed to reduce the burden for the patients, their family, and on healthcare system.