Heart Failure (HF) is a leading cause of mortality and morbidity worldwide, and is associated with the high use of resources and healthcare costs. In the United States, the estimated number of HF patients aged ≥ 20 years increased from 5.7 million, in 2009-2012, to 6.2 million in 2013-2016,^{1 , 2} and is estimated to affect 26 million people around the world. Its prevalence has been increasing fast due to the aging of the population and better care and treatment.^{3 , 4}

In Brazil, the prevalence of HF is of around 2 million patients, and its incidence is of approximately 240,000 new cases per year. This country has the largest universal health system in the world; besides, it is characterized by intense racial admixture, social inequalities, and cultural traditions that may affect the natural history of HF.⁵ It was observed that morbidity and mortality rates related to HF in Brazil are much higher than those observed in developed countries, even when adjusting for region, several hospital beds, and type of institution.⁶

Multiple regions of Brazil have shown a high variation in the quality of care addressed to cardiovascular conditions of high economic burden, such as HF. These registries remain suboptimal, and the lack of optimal therapies is more critical in the public non-academic institutions of the poorest regions of Brazil.⁷

Mortality due to HF is constantly associated with individual, social, economic, and health service indicators.⁸ In this regard, it is relevant to have an in-depth understanding of the spatial distribution of mortality caused by HF to increased knowledge about the areas of greatest vulnerability. The spatial analysis of HF mortality cases throughout the national territory is fundamental for the identification of specific demands and priority locations and, thus, for the planning of interventions that can reduce mortality caused by this disease.

There are no national and international studies about the distribution and spatial analysis of HF mortality rates in the Brazilian territory, denoting an existing knowledge gap.

The present investigation aimed to analyze the spatiotemporal trend of mortality from HF in Brazil, from 1996 to 2017.

During the 21 year-period, 1,242,014 deaths by HF were notified. The temporal trend analysis results showed a significant decrease of 2.3% (95%CI: -2.3 – -2.7) in the mortality rate in all Brazilian territories. By considering the continental dimensions of Brazil, it is necessary to investigate such trends in regional areas. Four of the five national Regions showed a significant reduction in mortality rates, highlighting a minimal reduction in mortality rates in the Northeast (AAPC: -0.7; 95%CI: -0.9 – -0.4). On the other hand, the South increased its mortality rate by 2.9% (95%CI: 2.0 – 3.9) ( Table 1 ).

Table 1

Country Regions	*APC1	‡IP	*APC2	‡IP	*APC3	‡IP	†AAPC
	(§ 95% CI)		(§ 95% CI)		(§ 95% CI)		(§ 95% CI)
Brazil	-6.3 // (-7.7; -1.8)	2001	-1.1 // (-1.3; -0.7)	2017			-2.3 // (-2.3; -2.7)
North	-1.3 (-11.3; 2.4)	2017					-1.3 // (-1.7; -0.9)
Northeast	- 0.7 // (-7.1; -1.5)	2017	9.4 // (6.6; 12.3)	2011	-2.8 (-4.2; -1.4)		-0.7 // (-0.9; -0.4)
Southeast	-7.7 (-8.7; -6.7)	2002	-0.8 (-1.2; -0.3)	2014	-2.4 // ( -0.9; -5.8)	2017	-4.0 // (-5.5; -2.4)
Midwest	-6.2 // (-10.2; -2.1)	2000	-2.3 // (-3.0; -1.7)	2015	-3.3 ( -11.5; 3.8)	2017	-4.0 // (-5.5; -2.4)
South	-7.9 // ( -13.0; -2.5)	1998	37 // (28.9; 45.5)	2001	-1.1 // (-1.3; -0.8)	2017	2.9 // (2.0; 3.9)

*APC: Annual Percentage Change; † AAPC: Average Annual Percent Change; ‡ IP: Inflection Point; § 95%CI: 95% Confidence Interval; // p = < 0.05

Regarding the spatial analysis of the 5,570 municipalities, the maps in Figure 2 demonstrate that crude mortality rates varied from 0 to 200 deaths by 100 thousand inhabitants. In the red areas, it is possible to identify, in period A (1996-2001), that high mortality rates were focused on municipalities of the South and the Southeast. Periods B (2002-2007) and C (2008-2012) demonstrate that, throughout the years, rates spread to the Northeast and the Midwest, but kept high concentrations in the South (similar results to the time series analysis). In the last years, as demonstrated in period D (2013-2017), it is possible to highlight that the municipalities in the North did not present high mortality rates.

Figure 2

In Figure 3 , the rates were smoothed by the empirical Bayesian method and demonstrated that, in period A (1996-2001), mortality was mostly concentrated in municipalities in the South; however, they varied throughout the years. Period B (2002-2007) showed that the Southeast presented high mortality rates, particularly in municipalities located in the states of São Paulo and Minas Gerais. In periods C (2008-2012) and D (2013-2017), it was possible to identify a high death concentration in the Northeast, predominately in Piauí, Bahia, Rio Grande do Norte, and Ceará.

Figure 3

Regarding autocorrelation spatial analysis, the GMI of period A (1996-2001) was 0.41 (p=0.01); in period B (2002-2007), it was 0.28 (p=0.01); period C (2008-2012) it was 0.29 (p=0.01); and GMI in period D (2013-2017) was 0. 31 (p=0.01). These results indicate positive spatial autocorrelation.

Since spatial dependency was verified in all study areas by GMI, we used the Local Moran Index (LMI). The spatial clusters corroborate the results in the empirical Bayesian local maps in all periods. They confirmed that high-highs are more concentrated in the South and spread to Northeast municipalities over the years. The low-low was concentrated in cities of the North and Midwest regions ( Figure 4 ).

Figure 4

In Figure 5 , scan spatial statistics confirmed other analyses and identified that the greater risk of CI mortality was in the South in period A (1996-2001). This pattern advanced in other periods, being also possible to find high RR in the Midwest.

Figure 5

Nevertheless, the last period also demonstrated higher risks in the North and the Northeast. In the last period (2013-2017), almost all cities in the South, Southeast, Northeast, and Midwest had some risk of mortality from heart failure. In the North, the cities in the states of Pará and Amazonas had little risk of mortality, with RR<1 (in blue); only a few cities of Roraima and Acre had clusters with RR>15 (in red).

This is the first study to assess the spatial distribution of HF mortality in Brazil. Initially, the municipalities in the South had a higher mortality rate. This notification pattern characterized operational contexts of “epidemiological silence” in some areas until 2001. Since 2002, mortality rates have increased in municipalities in the Southeast, Northeast, and the Midwest. Municipalities in the North, on the other hand, have shown low mortality rates, revealing the underreporting of cases.

It is important to observe that the municipalities in Brazilian regions with the highest mortality rates are the largest in population size and have more elderly people. National and international surveys have shown that HF mainly affects individuals aged more than 65 years.^{11 , 12} The increasing prevalence of HF is directly related to the most common comorbidities in aging.¹³ Besides, advanced age is a risk factor for self-care, poor treatment adherence, and disease management.^{14 - 16}

The disease is prevalent among the elderly, but a high percentage of the disease is also perceived among adults.¹⁷ In an analysis of specific time trends for age, sex, and race in HF-related morbidity and mortality rates between the years of 1991 and 2015, in the United States, Vasan, Zuo & Kalesan¹⁸ observed an increase in hospitalization rates in people with the disease in the age group of 30 to 49 years. This fact may be related to an inadequate lifestyle, exposing individuals to numerous risk factors for the disease, such as smoking, alcohol consumption, physical inactivity, and comorbidities, with emphasis on hypertension and diabetes.

Poffo et al.¹⁹ attributed the number of HF cases in adults to the earlier manifestation of heart disease, less effective treatment of diseases that lead to HF, and non-adherence to the recommended treatment. Thus, prevention stands out as an important weapon to fight against HF.^{17 , 18 , 20} The high mortality rate in municipalities in the southern region may be associated with the race/ethnicity of people affected by HF, with a higher prevalence of white individuals. While the disease prevails in this population, survival is higher among black people (7.5% vs. 2.2%; p = 0.01),¹⁹ Hispanics, and Asians.²¹

The reason for such a health disparity is complex. The effect of race on the prognosis of HF is uncertain because different studies have revealed divergent results. Race showed no association with the occurrence of complications and hospital readmissions among Australians with HF.²² Husaini et al.,²³ in a study comparing race/ethnicity to hospitalization rates, found that hospitalization rates for black individuals remained higher compared to whites and Hispanics.

In a North American study, in a total of 7,878 patients with HF, 35.8% were black. Among them, diseases such as hypertension, diabetes, and cerebrovascular diseases were prevalent; however, the mortality rate was lower when compared to white individuals.²⁴ Goyal et al.²⁵ analyzed 1,889,608 hospitalizations of patients with HF and found out that other comorbidities, such as anemia and chronic kidney disease, were more common among blacks, but this group presented lower in-hospital mortality when compared to whites.

The divergence in the literature reinforces the need for further research on the association between race/ethnicity and mortality rate due to HF, as well as the main causes of readmission. Such findings may assist in the implementation of preventive programs for hypertension and diabetes among minorities, which can reduce subsequent hospitalization for HF in these patients.

The change in the epidemiological patterns of the disease in the country has also been generated as a result of coverage by health services. Health care in Brazil is guided by a clear definition of territories and the people under the responsibility of each health team. The improvement of access, quality, and infrastructure of health care and coverage of Brazilian municipalities is a concern of health managers.

It is necessary to keep in mind that there are regional differences between the conditions of access to health institutions and the care provided by Brazil. Differences in supply can be observed within the same municipality, between municipalities and between states of the federation, which results in the maintenance of an uneven distribution pattern of supply and access to services.²⁶

The complexity of HF requires knowledge from different professionals who work at different levels of health care. Therefore, it is important to articulate these levels to promote health, quality of care, and user satisfaction. Health providers and patients must take on some responsibility that extends beyond mechanistic or traditional care. It is necessary to empower and enable the individual to protect oneself from situations of vulnerability, thus reducing regional disparities.

The increase in mortality rates due to HF in the municipalities of the North and Northeast can be explained by the low quality of life and stressors, which are the main agents for higher increase in hospitalization. Hospitalizations occur not only due to HF, but also other comorbidities.^{19 , 27}

The Brazilian regions with the highest number of hospitalizations in 2017 were the Southeast, followed by the South and Northeast, and, in the third place, the Midwest and the North.²⁸ Isolated initiatives suggest the existence of significant regional differences in several characteristics of patients hospitalized with HF in Brazil,¹⁷ although these comparisons are methodologically limited by often divergent designs and inclusion criteria.

Despite major advances in health care science and technology, a large gap separates technically achievable advances in health from what individuals and populations actually achieve.²⁹ Financial status is a relevant factor in the management of the disease, related to functional health literacy, adherence, and self-care behaviors. The authors emphasize that the lower the socioeconomic level, the lower the rate of treatment adherence (medicated or non-medicated).³⁰

In Europe, it is estimated that the annual cost of HF treatment is 3.7% of the National Health Fund, with the expectation of doubling the amount in the next 20 years. Medication costs make up for less than 5% of the resources spent on treating people with HF, whereas 79% is related to expenses with hospitalization.³¹

In Brazil, HF has a high financial cost. Kaufman et al.³² revealed that, in the period from 2001 to 2012, the average length of stay for patients with HF was 5.8 days, in 2001, to 6.6 days in 2012; the mortality rate was on the rise, starting from 6.6%, in 2001, and reaching 9.5% in 2012 (a 46.1% increase). The average cost of authorization for hospitalization increased from R$ 519.54, in 2001, to R$ 1,209.56 in 2012 (an increase of 132.8%). Research that sought to describe the number of hospitalizations for HF in Brazilian regions in 2017 and the impact of these hospitalizations on hospital costs showed that, in 2017 alone, the amount of R$ 339,719,216.50 was spent.²⁸

In a recent study, Stevens et al.³³ analyzed the costs associated with treatment, lost productivity due to reduced employment, the costs of providing formal and informal care, and loss of well-being related to the conditions of four major heart diseases in Brazil. The authors found that the costs with HF were R$22.1 billion (USD 6.8 billion) and patients had significant loss in productivity.

The high cost associated with care in HF suggests that new studies are developed to better understand the patients’ choices, so that professionals can be more prepared to help them administer their medications, influence daily behaviors, and encourage taking healthy decisions. Knowing the financial situation and understanding the municipality’s economy can provide additional approaches to improve the prognosis of the disease.

It is worth mentioning that cities at low risk of mortality from the disease need to be analyzed with caution. The area demonstrated as a protective factor for mortality from the disease may be the result of underreporting. Moreover, the migration of patients with HF to neighboring regions due to better opportunities for treatment may make it difficult to interpret the study.

Some limitations regarding this study should be reported. First, it is an ecological study, so the analysis of aggregated level patient data cannot establish relationships between individual-level variables. Furthermore, since our data derived from a national database, it is possible that there is underreporting, and the hospital mortality rate may be underestimated. These limitations may better address the results. On the other hand, the curacy of the analysis, the amount of data and the long analyzed time period provide safety to the evidence generated in this study.

HF mortality followed a spatial pattern in the analyzed period. The notification of cases was initiated in the municipalities of the South of Brazil and, over the years, an increasing pattern in mortality was observed in other regions, with emphasis on the Southeast and the Northeast.

Spatial analysis contributed to show the HF mortality scenario in Brazilian municipalities. The highlighted geographic areas are most susceptible to mortality, requiring specific actions to prevent the disease and promote heath. Therefore, this study may guide actions aimed at improving the quality of clinical care provided in the cities most affected by HF.