The Medical Burden of Heart Failure: A Comparative Delineation with Cancer in Brazil

Background: Due to its poor prognosis and mortality rates, heart failure (HF) has been recognized as a malignant condition, comparable to some cancers in developed countries. Objectives: To compare mortality from HF and prevalent cancers using data from a nationwide database in Brazil. Methods: This was a descriptive, cross-sectional study using secondary data obtained from Brazilian administrative databases of death records and hospitalization claims maintained by the Ministry of Health. Data were analyzed according to main diagnosis, year of occurrence (2005-2015), sex and age group. Descriptive analyses of absolute number of events, hospitalization rate, mortality rate, and in-hospital mortality rate were performed. Results: The selected cancers accounted for higher mortality, lower hospitalization and higher in-hospital mortality rates than HF. In a group analysis, HF showed mortality rates of 100-150 per 100,000 inhabitants over the period, lower than the selected cancers. However, HF had a higher mortality rate than each type of cancer, even when compared to the most prevalent and deadly ones. Regarding hospitalization rates, HF was associated with a higher risk of hospitalization when compared to cancer-related conditions as a group. Conclusions: Our findings indicate that HF has an important impact on mortality, hospitalization and in-hospital mortality, comparable to or even worse than some types of cancer, representing a potential burden to the healthcare system.

(IBGE) using methods described in the Brazilian National Population Projections by age and sex: 2000-2060. 17 Estimates are calculated using data from the Brazilian 2010 Demographic Census and information of births and deaths obtained from official records.

Statistical analysis
The data were aggregated for calculations of mortality and hospitalization, in absolute numbers and rates, by disease and year of occurrence. Each event (hospitalization, death and in-hospital death) was coded according to the ICD-10 classification (after accounting for ill-defined or undefined causes of death) and the aggregated groups of causes were analyzed considering the year of occurrence (2005 to 2015).
For death events, a redistribution method of ill-defined causes of death (Chapter XVIII of the ICD-10 -Symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified) was used as described by Soares et al. 15 Undefined causes of death were redistributed using the proportion of each defined cause except for external causes (which were assumed to contribute to a small proportion of in death records). These reclassified death records were summed to the absolute number of deaths initially coded with eligible ICD-10 codes for the study.
Data were then descriptively compared year by year between selected diseases using graphs.
As a descriptive study, all data on death and hospitalization that met the eligibility criteria were organized and stored in a Microsoft Excel spreadsheet. Thus, sample size calculation was not applicable.

Results
Due to the nature of the study -a retrospective database study without patient-level data -information about participants is not disclosed.

Study design
This was a descriptive, cross-sectional study using secondary data obtained from the SIM (Mortality Information System) 13 and the SIH (Hospital Information System) of the Information Technology Department of the Brazilian Ministry of Health. 13 The SIH is an administrative database of data from hospitals of the Brazilian unified Health System (SUS), including admission dataauthorization forms, demographics, hospitalization cause -length of stay and in-hospital mortality, which are used for health service and system planning and knowledge production in the field of public health. 14 The SIM provides nationwide population-based data about mortality -main cause and secondary causes of death, and demographics, obtained from death certificates. As for SIH, these data help in planning of health services and programs. 15 Both SIM and SIH are publicly available databases created and maintained by DATASUS. 16 The analysis comprised a period of eleven years, of registries between 2005 and 2015 of individuals aged ≥ 45 years, age when cardiovascular disease is most diagnosed. Files containing anonymized data were downloaded directly from the DATASUS website in their original format. Data on hospitalization and death were retrieved from the SIH and SIM databases, respectively, of the 26 states and the Federal District in Brazil. Data cleaning and validation was conducted by the investigators to identify completeness and integrity of available data.
We considered the 10th International Classification of Disease (ICD-10) code for Heart Failure I50, and the most prevalent cancers: C16 stomach cancer, C18 colon cancer -grouped with C19 malignant neoplasm of recto sigmoid junction and C20 rectum cancer, C34 trachea, bronchi and lung cancer, C50 breast cancer (except for male cases of breast cancer for both death events and in-patient admissions), C53 cervix cancer and C61 prostate cancer.
Brazilian population projections were obtained from DATASUS website. 16 These projections are obtained from the Brazilian Institute of Geography and Statistics Int J Cardiovasc Sci. 2022; 35(4), 514-520   Table 3 shows the percentage of in-hospital mortality of patients with HF and with different types of cancer. HF mortality has increased progressively in these eleven years. Patients with HF hospitalized for compensation had an average mortality in of 11.08%, higher than breast cancer (9.60%) and prostate cancer (10.32%) and lower than other types of cancer.

Discussion
Several authors have called attention to the fact that the mortality of HF patients is high and more pronounced than of patients with some types of cancer. 13,14 In Brazil, the mortality of patients with HF is also high, particularly when compared with mortality rates described in developed countries, but there are no data comparing mortality from HF with cancer in our country. In this article, we made this comparison using data from DataSUS. [15][16][17] The comparative analysis of hospital admissions for HF with admissions for the most prevalent types of cancer revealed significantly higher numbers of patients hospitalized due to HF than cancer (Table 1). Besides, the number of patients who died from HF was significantly higher than those who died from different types of cancer ( Table 2). In addition, considering in-hospital mortality, Int J Cardiovasc Sci. 2022; 35(4), 514-520 we may say that HF was more malignant than breast cancer and prostate cancers (Table 3), as mean mortality rate of patients hospitalized due to acute HF (11.08%) was higher than breast (9.60%) and prostate cancers (10.32%).
In Latin America, HF is the leading cause of hospitalization, with rehospitalization rates of 33%, 28%, 31%, and 35% at 3, 6, 12, and 24 to 60 months of follow-up, respectively. 18,19 Despite treatment advances, HF still has a poor prognosis, with high mortality rates. Five-year mortality rate for HF was estimated at approximately 50%. 20 In Latin America, it is estimated a one-year mortality rate of 24.5%, and in-hospital mortality rate of 11.7%. 19 Brazilian registry data indicate an in-hospital mortality rate of 12.6%. 21 As shown in previous studies in developed countries, 14,20,22 HF can be associated with worse outcomes than some types of cancer. Askoxylakis et al., 20 conducted a systematic review of the literature and noted a five-year survival of approximately 43% for all cancer types and 26-52% for HF, showing that HF in some settings is as deadly as some cancers, and even worse as compared with cancers like breast cancer (73-89%), prostate cancer (50-99%) and colorectal cancer (43-63%). 20 Using a retrospective approach, Stewart et al., 22 identified that the annual incidence of first-ever hospitalization for HF was higher than for cancer in Sweden: 484 versus 373 (lung, colorectal, prostate, and bladder cancer combined) per 100,000 men and 470 versus 350 (lung, colorectal, bladder, breast, and ovarian cancer combined) per 100,000 among women aged > 20 years. The authors also observed that the 30-day and five-year mortality rates were comparable between HF and cancer, and that during the 10-year follow-up period, HF was associated with more premature life-years lost than all common forms of cancer in men but not in women. 22 Mamas et al., 12 conducted an analysis of survival rate comparing HF with some forms of cancer. The authors' findings indicated that HF had significantly worse five-year survival rate (55.8%) than prostate cancer (68.3%) and bladder cancer (57.3%), but significantly better than lung cancer (8.4%) and colorectal cancer (48.9%). In women, HF mortality outcome was worse (49.5%) than breast cancer (77.7%), but better than colorectal cancer (51.5%), lung cancer (10.4%), and ovarian cancer (38.2%). 14 Our data confirm the described in Latin America and in the world 14,20,22 regarding high rates of mortality from HF as compared with some cancers These data reinforce the need to recognize HF as a priority condition in Brazil, mainly by health system managers and policy makers, but also by the general population. Besides the magnitude of the disease burden in terms of deaths and hospitalizations, the decreasing rates observed in temporal series highlight that HF potentially responds to improvement in care with better outcomes that are relevant for both patients and the health care system, once hospitalization is the main cost driver in HF. 22,23 Since the data used in this analysis were representative of all the death certificates and hospitalization claims from the Brazilian public health care system during the 2005-2015 period, it is possible to assume that the findings are applicable to the national setting for mortality data and for the public health care system for hospital admission data. It is important that physicians become aware of these data, to try to make an earlier diagnosis of HF and provide earlier treatment using the best evidence, and thereby modify the natural history of the disease.
It is worth remembering that the CONSENSUS study showed that, although it was possible to modify the course of HF, the mortality remained high (44%) in the control group in the first six months and in the first year (52%). The prescription of enalapril reduced mortality to 26% in the first six months and to 36% at the end of the first year. 23 With the introduction of beta-blockers, mineralocorticoid receptor antagonists, angiotensin receptor antagonists and, more recently an angiotensin receptor neprilysin inhibitor (ARNI), it is possible to substantially reduce the mortality of patients with insufficiently treated HF. [24][25][26] An interesting point often discussed in Brazilian scientific meetings is the interpretation of the recent reduction in the number of hospitalizations for HF per year, as indicated by the SUS data. Some presentations interpret such decrease as a result of better management of the cases, without taking into account, however, the significant reduction in the number of SUS beds (Table 4) in recent years. With a smaller number of beds, physicians are pressured to admit only the most serious patient who will have the highest mortality, even with the best treatment available.
We can conclude that HF alone promotes more hospitalizations and deaths than some types of cancer. The mortality of patients with HF was higher than the one observed in patients with breast or prostate cancer, a result similar to other studies around the world. This concept of HF malignancy should be better disseminated so that more attention would be paid to patients with the syndrome, as its prognosis varies with treatment (e.g. timing and dosage, use of neurohormonal blockers), according to national guidelines.

Limitations
The results of hospitalization and in-hospital mortality of the present study are probably not applicable to the private health care system, since the access to health care services, treatment patterns and epidemiological profile of patients are markedly different between both settings. The main limitation of this study is its retrospective approach based on administrative databases that were not specifically designed for the purposes of the study. For this reason, detailed clinical data about diagnosis and treatment were not available, limiting our ability to adjust for the comorbidity burden of HF, for example. Also, it was not possible to differentiate between HF with reduced and preserved ejection fraction. Another limitation was that it was not possible to use record linkage to combine HF-and cancer-related hospitalization and mortality data to identify unique patients. Another limitation of this study was the absence of patient-level longitudinal data that could allow further analysis including survival analysis, as previously performed by other researchers. 14,20 Despite these limitations, both SIM and SIH databases have been widely used for epidemiological research in Brazil with valid and well-accepted results. 15,16 These aspects can be further explored in futures studies conducted in Brazil, including HF cost studies, to provide greater knowledge about the clinical and economic burden of HF in the country.