Dementias in Brazil: increasing burden in the 2000–2016 period. Estimates from the Global Burden of Disease Study 2016

ABSTRACT Introduction: Dementia is a globally relevant health problem, which places a great burden on patients and their families. This study aimed to estimate the burden associated with Alzheimer's disease (AD) and other dementias in Brazil. Methods: In this descriptive study, we investigated the estimates obtained by the Global Burden of Disease study. We described the prevalence of AD and other dementias, years lived with disability (YLDs), age-standardized mortality, years of life lost (YLLs), and disability-adjusted life years (DALYs) among individuals aged 60 years or older between 2000 and 2016, with their respective 95% uncertainty intervals (95%UI). Results: During this period, the age-standardized prevalence of AD and other dementias per 100,000 people increased by 7.8%, from 961.7 (95%UI 828.3–1,117.5) to 1,036.9 (95%UI 882.0–1,219.5), with approximately 1.5 million people living with dementia in Brazil. The incidence increased by 4.5%. Similarly, all age-standardized rates had an upward trend (mortality: 3.1%; YLLs: 5.8%; YLDs: 7.9%; and DALYs: 6.3%). Mortality profiles increased with age in both years. Dementias were ranked fourth among the leading causes of death in people aged ≥70 years in 2000, rising to second place in 2016. In 2016, it also represented the second and third leading causes of disability among older women and men, respectively. Conclusion: Population growth and aging have resulted in an increased burden of AD and other dementias in Brazil. Preventive and early diagnostic measures are essential to mitigate the burden associated with these diseases.

Dementia is a neurological condition that results in a great burden to patients and their families 1,2 . By 2015, around 40 million people in the world had been living with dementia. This figure is estimated to double by 2040, with lower socioeconomic status (SES) countries as home to most dementia patients. The increasing dementia population is a great concern for health professionals and policymakers worldwide 1,2 .
Although the main causes of dementia -Alzheimer's disease (AD) and vascular dementia -are strongly associated with non-modifiable risk factors such as advancing age and positive family history, they are also related to preventable conditions such as atherosclerotic disease, cardiovascular disease, diabetes, and dyslipidemia 3 . In 2016, Brazil's leading risk factors related to the burden of chronic noncommunicable diseases were obesity, systolic hypertension, and hyperglycemia; all of them are associated with dementia risk and preventable 4 .
The Global Burden of Disease (GBD) study investigates the burden of diseases worldwide, using a standardized methodology that allows spatial and timely comparisons of estimates. Since 1990, this study has annually described the loss of health due to more than 300 diseases across 195 countries, facilitating comparisons between regions, countries, and states. Loss of health is expressed by "disability-adjusted life years" (DALYs) -the potential years of life lost due to death or disability -which combines information on mortality and morbidity 5,6 .
Estimates of dementia in Brazil varied greatly, even when only results of population-based or community-dwelling studies were taken into consideration. In some studies, the prevalence of dementia among Brazilian people aged 65 years or older was similar to that observed in other Latin American countries (around 7%) 7,8 . Most Brazilian studies were conducted in the state of São Paulo. The prevalence rates of dementia among patients aged 60 years or older in the cities of Ribeirão Preto 9 , São Paulo 10 , and Tremembé 11 were 12.5, 12.9, and 17.5%, respectively. In a small city in the state of Minas Gerais, the figure for those aged 75 years and older was 27.5% 12 . Incidence studies are scarce. The incidence of dementia among random-selected older adults (≥65 years) was around 14 per 1,000 person-years, twice as high as the estimate for AD 13,14,15 . A higher incidence of AD (14.8, 95% confidence interval [95%CI] 9.04-22.94) was found among community-dwelling older adults (≥60 years) from Porto Alegre 16 . These differences may reflect the lack of standardization concerning study design, diagnosis of dementia, age composition of the populations, and year of data collection.
Understanding the spatial distribution and temporal trend of the burden of dementia is the first step to implement preventive and therapeutic measures. Using the estimates of the GBD-2016 study, this investigation aimed to determine the burden of AD and other dementias in Brazil and its regions.

METHODS
This study aimed to describe the GBD-2016 estimates of the burden of AD and other dementias. All estimates derived from national and subnational data obtained and analyzed through the collaboration of a Brazilian network of researchers, the Brazilian Ministry of Health, and the Institute of Health Metrics and Evaluation (IHME) of the University of Washington 17 . In our study, we only used demographic and health data from the Brazilian population aged 60 years and older between 2000 and 2016 because they are more reliable than those collected in the earlier period of 1990-2000.
Although GBD estimates are annually provided, we investigated Brazilian estimates from GBD-2016, as a detailed description of methodology and a comprehensive worldwide analysis of AD and other dementias were performed with these estimates 19 . Dementia was defined based on the codes of the Diagnostic and Statistical Manual of Mental Disorders, 4 th and 5 th editions (290, 291.2, 291.8, 294, and 331), and the International Classification of Diseases (F00, F01, F02, F03, G30, and G31), 8-10 th editions 19 . Unlike most disease definitions used in the GBD study, the estimates of AD and other dementias were analyzed and modeled together, without considering specific causes 19 .
Demographic and vital data for mortality analysis were mainly provided by the Brazilian Institute of Geography and Statistics and the Mortality Information System (Brazilian Ministry of Health) 17 . Primary mortality data were adjusted first for underreporting and then for garbage codes. The methods for adjusting death underreporting were described by Foreman et al. 20 . Codes that should not be considered basic causes of death, including ill-defined causes (stated in Chapter XVIII), are classified as garbage codes. They are redistributed to other defined causes called target diseases. All garbage codes are redistributed to target diseases using statistical algorithms, judgments based on literature reviews, or proportional distributions. Details of the modeling and model validation were also published 20,21 . Besides the absolute number of deaths and standardized mortality by year, gender, and age, the metric years of life lost (YLLs) expresses the effect of premature deaths, which is obtained by multiplying the number of deaths by the years lost before the standard life expectancy in each age group. Life expectancy was determined based on the reference life table of the GBD study 21 .
The number of years lost due to disability (YLDs) is calculated by multiplying the prevalence of each disease sequela by its disability weight 6,19 . Three sequelae and their respective disability weights derived from a systematic review of seven studies using the Clinical Dementia Rating (CDR) Scale: mild (CDR1), moderate (CDR2), or severe (CDR3) dementia 19 . All data on the prevalence, incidence, and association with AD and other dementias included Brazilian populationbased studies 7,8,9,10,11,12,13,14,15 to model the estimates. The data are available at http://ghdx.healthdata.org/gbd-2016/ data-input-sources.
Age-standardized rates were directly standardized by the age structure of the world population used by the GBD study. The metrics are presented with 95% uncertainty intervals (95%UIs). Uncertainties may stem from data sources and modeling steps, such as sample size variability, adjustments for mortality sources, parameter uncertainty in model estimation, specification of uncertainty for causes of death models, and differences in data availability by age, gender, year, and location 6 . To determine the UIs, all GBD metric calculations were performed 1,000 times so that the 95%UIs were set at the 25 th and 97.5 th of the estimated values of 1,000 estimates 6,19 . All results can be accessed at http://vizhub.healthdata. org/gbd-compare.
The GBD study was approved by the Institutional Review Board of the University of Washington. It also received approval from the Institutional Review Board of the Universidade Federal de Minas Gerais under the protocol CAAE -62803316.7.0000.5149. We were not required to submit our research proposal to local institutional review boards because we accessed the GBD study data from public domain secondary databases. Individual patient data were not collected in this study; hence, informed consent was waived.

RESULTS
In Brazil, AD and other types of dementia were ranked fourth among the leading causes of death in people aged ≥70 years in 2000, rising to second place in 2016. However, they were not leading causes of death among people aged 50-69 years ( Figure 1). Dementia was an important cause of disability among older adults, representing the second and third leading causes of disability among older women and men, respectively ( Figure 2).

DISCUSSION
This study identified a consistent increasing trend in all estimates, despite the confluence between UIs. This reality should not be neglected by health managers.
In 2016, dementia was the fifth leading cause of death and the 23 rd leading cause of DALYs worldwide 19 . Between 1990 and 2016, the prevalence of AD and other dementias doubled globally, from 20.2 million (95%UI 17.4-23.5 million) to 43.8 million (95%UI 37.8-51.0) 18   of older people from a younger age group (age 65 to 69) in Latin American countries, including Brazil, was higher than in developed countries 4 . Besides, developed countries have more resources and social structures to face the challenges of this demographic shift; therefore, a greater impact of AD and other dementias could be predicted in Latin American countries.
Dementia mortality profiles by age were very similar between 2000 and 2016, but presented an important increasing gradient with age. In addition, we detected a higher increase in the age-standardized prevalence than in the incidence of AD and other dementias. This finding probably results from population growth and aging rather than an increase in the risk of disease in this period. Aging represents a victory for humanity; however, it may also represent a greater risk of chronic diseases and disability. This situation may compromise the autonomy of older people. Brazil has been experiencing a rapid and marked aging process. Its older population in 1940 consisted of 1.7 million people (4%), increasing to 14.5 million (8.6%) in 2000. The aging rate is higher among those aged 80 years or older and shows female predominance (55%) 22 . As a result, in 2016, Brazil had the second highest age-standardized prevalence of AD and other dementias per 100,000 people (1,037 -95%UI 882-1,220), after only Turkey (1,192 -95%UI 1,007-1,405) 19 .
The burden of AD and other dementias increased by 6.3% in Brazil. Chronic degenerative diseases such as dementia are associated with higher morbidity, prevalence, and YLDs than mortality and YLLs. Increased longevity implies that the gain in life expectancy is followed by a greater degree of health loss 22,23 . Prevalence increase will become a reality, as most dementias are progressive in nature and have no treatment to modify their course.
To face the burden profile of dementia, health financing must increase pari passu with population growth and aging. A recent study showed that the overall cost of dementia in Brazil ($1,405.72 per capita)     lower than the global dementia cost per capita (approximately $5,284) 24 . Most of these costs are indirect, as the social costs associated with the loss of caregiver productivity, aside from the time and stress experienced by them, are usually informal 24 . Besides patient suffering, the lack of specific coping strategies, whether in primary or specialized care, leads to increased family demand; family members or relatives become informal and untrained caregivers. These caregivers face many challenges, such as diagnosis acceptance, family conflict management, future reprogramming, and, most importantly, engagement in all activities that involve caring for the patient, from providing medications to securing financial and legal support. The overload experienced by caregivers, largely young women, is undoubtedly an indirect cost that can be prevented if taken over by trained professional caregivers 25 . In this study, although women presented a greater burden of disease, men experienced a greater increase in the burden of AD and other dementias in the period. Most Brazilian studies showed a higher prevalence of dementia in women, even though the difference was not statistically significant 8,9,13 . The current best explanation for this difference is that women outnumbered men; however, future studies are necessary to clarify this scenario 26 .
The analysis of estimates from the GBD study has certain advantages. Its standardized methodology allows comparisons over time and across places. In addition, UIs reflect the amount and quality of health information for each location and period.
However, one important limitation concerns the paucity of population-based studies on dementia, not only in Brazil but worldwide. We underline that São Paulo presented rates two or three times higher than those of other states, clearly affecting the national average. This finding may be responsible for the great homogeneity of metrics between the states, which prevented a more careful analysis by region. The diversity of diagnostic methods used for dementia is the second limitation. Across the 237 data sources used by the estimates of the GBD-2016 study, 230 diagnostic procedures were used 19 .
The third limitation relates to the impossibility of classifying dementia into subtypes, as different causes of dementia have different clinical and epidemiological profiles accompanied by different prevention and treatment strategies 19 . The primary analysis showed sharp discrepancies between the prevalence of dementia and the cause-of-death data on the specific cause of dementia 19 . In the future, further discrimination between AD, vascular dementia, and other dementia types may be advisable. Doctors are advised not to report dementia as an underlying cause of death on death certificates because this practice can lead to underreporting of deaths, which is difficult to correct 27 . We also highlight the challenge of measuring disability weights since brief descriptions of health status may not fully capture the complexity of diseases such as dementia and AD 19 .
In conclusion, this study calls attention to the increase in the burden imposed by AD and other dementias in Brazil, along with the paucity of studies in the country. This reality requires health facilities to invest in prevention and research to ensure a better quality of life not only for those with dementia but also for their families.