The new approach to institutionalise health accounts in Brazil focuses on comprehensively identifying spending by financing schemes and healthcare services. This chapter discusses how this scope could eventually be expanded to unlock the full potential of health accounts to inform evidence-based decision-making in health. Areas where such an extension could be possible include an allocation of health spending to health providers, the identification of revenue sources of the various financing schemes and breaking down health spending by regions or states. Moreover, a number of OECD countries also allocate health spending by beneficiary characteristics such as age, sex or diseases.
Institutionalising Health Accounts in Brazil
A Review of Methods, Data and Policy Relevance
Abstract
The two reports produced in the last decade together with this renewed health accounts initiative should serve as a solid basis for the regular production and reporting of health spending data for Brazil in accordance with international standards. As described in Chapter 2, A System of Health Accounts 2011 provides a flexible toolkit to produce a range of health expenditure data to primarily respond to a country’s information needs, in addition to meeting any international reporting requests.
Beyond Brazil’s current level of reporting (as outlined in Chapter 5), an important next step can be to expand the scope of SHA implementation to incorporate additional dimensions of analysis. This would enhance the applicability and policy relevance of the health accounts to best match Brazil’s health policy priorities. While some extensions may be feasible in the short term, others will require a longer-term commitment. This chapter is intended to serve as an input towards developing a roadmap for future developments.
To gain a more comprehensive understanding of how resources are allocated within the health sector, Brazil should investigate how to expand reporting to include health spending by health provider (HP) in the core SHA framework (as set out in Chapter 2). Nearly all OECD countries, including Chile, Costa Rica and Mexico, provide a breakdown of health spending by provider. A breakdown of spending by provider (HP) cross-classified with financing schemes (HF) and healthcare functions (HC) allows for a more nuanced analysis of health expenditure. Analysing health spending by type of service in combination with health providers is particularly valuable to understand differences in the organisation of health services across countries. Figure 6.1 shows that in some countries, such as Bulgaria, Germany and Greece, hospitals tend to be mono-functional focusing almost exclusively on the provision of inpatient care services (accounting for around 90% of total spending in hospitals). In other countries, such as Finland, Portugal or Chile, hospitals can also play a significant role in providing outpatient and day care services. Allocating health spending by type of service to health providers is also a prerequisite for an appropriate estimation of spending on primary healthcare, where OECD methodologies rely on information included in the HCxHP table of the annual data submissions (OECD, 2018[1]).
Incorporating the provider dimension into Brazil’s health accounts should be seen as a natural next step, as much of the existing data infrastructure is already closely linked to health facility types. Brazil has already made significant progress in this area – as part of the second health accounts report, spending by SUS was successfully categorised by provider (Brasil. Ministério da Saúde, 2022[2]). The well-established National Registry of Health Establishments (CNES) and the requirement for key SUS data sources (e.g. SIA and SIH) to include CNES identifiers create a strong foundation for systematically integrating provider-level information. Given this, allocating SUS expenditure by type of provider would appear highly feasible. For other public financing schemes (such as spending by other federal ministries or subnational governments), as well as private insurance and out-of-pocket (OOP) spending, the direct allocation by provider type is less straightforward. Here, initial effort to map and estimate the providers linked to the ANS could be further pursued, based on plausibility assessments where direct linkages are not yet established.
As with other countries, Brazil uses a health provider classification that reflects national specificities and responds to the structure of the Brazilian health system, as well as reflecting regional variations. Brazil’s national classification of health providers, which includes 45 CNES categories, serves as a crucial tool for organising health expenditure data. However, to ensure international comparability, this classification needs to be mapped to the ICHA-HP framework used in the System of Health Accounts 2011. An initial mapping between the 45 CNES categories, a Brazilian health provider classification (see Table 6.1) and the ICHA-HP classification has already been developed (Brasil. Ministério da Saúde, 2022[2]), providing a foundation for integrating Brazil’s data into global health expenditure analyses. Nevertheless, further refinement is necessary to enhance consistency and accuracy, particularly for providers with multiple functions, such as hospitals that deliver both inpatient and outpatient care. A more precise alignment between national and international classifications will not only improve the reliability of cross-country comparisons but also strengthen Brazil’s ability to analyse health spending patterns in a way that aligns with global best practices.
Hospital expenditure by type of service, 2022 (or nearest year)
Note: “Other” includes preventive care activity; pharmaceuticals if dispensed to outpatients; and unknown services. 1. Includes ancillary services.
Source: OECD (2023[3]), Health at a Glance 2023: OECD Indicators, https://doi.org/10.1787/7a7afb35-en.
Source: Brasil. Ministério da Saúde (2022[2]), Contas de saúde na perspectiva da contabilidade internacional: conta SHA para o Brasil, 2015 a 2019, https://doi.org/10.38116/978-65-5635-028-8.
As seen in previous chapters, the health financing architecture in Brazil is particularly complex. In the estimates based on the refined methodology, health financing data is presented exclusively through the lens of the financing schemes (HF) – that is, referring to the arrangements through which people obtain health services. However, to get a fuller understanding of how health financing works in a country, this view should be complemented by identifying the financing agents (FA) that implement the various financing schemes and the revenues of the financing schemes (FS). As indicated in the methodological manual for the Brazil Health Accounts, “the identification of sources of revenue and financing agents is carried out only in a descriptive manner, without discussing the method of calculating the revenues linked to each type of revenue and to each financing agent”. The manual goes on to highlight that this is seen as an area for further research.
A key aspect of health system financing revolves around the roles of the various financial agents in the transfer of funds through the system. In Brazil, the financial agents include the Ministry of Health, the National Health Fund, state and municipal health secretariats, and private health plan operators, among others. Understanding the interactions between these different agents and mapping the flows between them from sources through to uses will ultimately contribute to a better understanding of the financial landscape within the health system.
To achieve this, it is necessary to develop a comprehensive research agenda focused on financing schemes (ICHA-FS) and financing agents (ICHA-FA). Such an agenda should aim to provide more precise and in-depth estimates related to these financing mechanisms.
Across the OECD, nearly three out of four countries can successfully identify the revenues of all financing schemes, a capability that is also shared by all OECD member countries in Latin America. While the allocation of financing agents to the appropriate categories can present challenges, identifying the revenues of these financing schemes should be a relatively straightforward task for Brazil. In this regard, Brazil shares similarities with several OECD countries that have National Health Service (NHS)-type financing systems with a duplicative private insurance market, including the United Kingdom, Australia and New Zealand.
The first step toward effectively identifying revenues (FS) is the accurate categorisation of financing schemes (HF). By convention, certain revenues are linked to specific categories of financing schemes, meaning that a precise identification of these schemes is critical for mapping revenue sources correctly. This foundational step will ensure a robust framework for understanding and reporting Brazil’s health financing system in a comprehensive and internationally comparable way.
Based on the current understanding of the health financing architecture in Brazil, the types of revenues (ICHA-FS) of each financing regime are already described in the Brazilian health accounts manual. From this, SUS (HF111) is predominantly or exclusively financed via internal transfers (FS1). The same should be true for the civil servant scheme that does not require contribution payments from beneficiaries. The scheme for military and law enforcement personnel, considered as a social security scheme (HF121) is financed by social contributions (FS3) – either by employers or employees – and possible via transfers on behalf of specific groups (FS1). Voluntary private health insurance (HF21), which plays a substantial role in Brazil and generally takes the form of duplicate insurance, is generally financed by voluntary contributions (FS5) – either by employers or private households/employees. However, tax rebates for private insurance premiums are non-negligible. They should be subtracted from voluntary contributions and re‑allocated to subsidies (FS13), in case they are measurable transactions and can be found in public budgets.1 By convention, out-of-pocket payments (HF3) should be allocated to other revenues from households (FS61). Any tax deductions from out-of-pocket spending that can be reclaimed form the public purse should be deducted at the level of the Financing Scheme (from HF3 to HF11). Overall, based on this road map, the identification of all revenues of financing schemes should be feasible.
Since the important split between public and private financing of health is generally carried based on the revenue classification and not the financing schemes classification, an implementation of this additional SHA-dimension is all the more important for international comparisons. This is particularly true for Brazil where the substantial governmental subsidies for the uptake of voluntary private health insurance would go unnoticed. If the analysis is restricted to financing schemes, this would underestimate the role of government spending. From a policy perspective, it would not highlight possible equity issues associated with government support for the uptake of private health insurance, which is concentrated among the more affluent population groups in Brazil (Montoya Diaz et al., 2020[4]).
Additionally, for national purposes, Brazil could show how SUS revenues but also expenditure (collectively allocated to HF111) could be further subdivided into the federal government, the state governments and the municipalities. Descriptions of the various data sources – SIAFI/SIOPS in the case of SUS, the transparency portals for state employee health plans, or DIOPS/ANS in the case of private health plans – indicate the availability of information to estimate detailed revenues of the various financing schemes in Brazil. This could be a powerful tool and inform policy discussions about, for example, whether the various levels of government finance the respective shares as set out in the constitution. Other OECD countries that share some similarities with the Brazilian health system are also interested in the role of the different levels of government in both revenue generation and spending. In Canada, for example, public-sector health expenditure can be distinguished between spending from provincial/territorial governments, federal direct health expenditure, municipal government health expenditure and spending by social security funds (CIHI, 2024[5]). This analysis is also possible for each Canadian province, which enables a comparison of federal involvement in public health spending by province.
Monitoring financial investments in the health sector is crucial for policy makers seeking to assess the physical capacity of healthcare systems now and in the future. The allocation of resources toward new health facilities, diagnostic and therapeutic equipment, and information and communications technology (ICT) plays a vital role in strengthening a health system’s ability to meet the growing and evolving healthcare needs of the population, as well as being resilient in the face of sudden shocks.
To facilitate this, the Joint Health Accounts Questionnaire (JHAQ) data collection includes a dedicated table on “Gross Fixed Capital Formation”2 for each category of health providers. This captures investments in infrastructure (e.g. the construction of primary healthcare centres), machinery and equipment (e.g. hospital beds and MRI scanners), and intellectual property (e.g. the development of digital health databases and telemedicine platforms).
Across OECD countries, gross fixed capital formation in the health sector typically accounts for approximately 0.6% of GDP – significantly lower than current health expenditure, which represents around 9% of GDP (OECD, 2023[3]). In some cases, health accountants may have some health sector-specific data sources to draw on and should align to a certain extent with data that can be found on GFCF in the Annual National Accounts, even if there needs to be some adjustment made to fit closer to the SHA boundaries of health providers.
In Brazil, health sector investments by both public and private entities were reported as part of the second pilot implementation. According to the metadata accompanying the Brazilian JHAQ data submission for 2021, the key data sources from IBGE included:
Pesquisa Industrial Anual (Annual Industrial Survey) – Covers investment in industrial sectors, including medical equipment and infrastructure.
Pesquisa Anual da Construção Civil (Annual Construction Survey) – Captures investments in infrastructure projects, including healthcare facilities.
In addition, it is indicated that SIAFI includes information on budget execution for infrastructure, medical equipment, and capital projects in the health sector, while SIOPS provides insights into both recurrent expenditures and capital investments at a more localised level.
Understanding and integrating these various sources could help policy makers develop a clearer picture of capital investment trends in the health sector. Ultimately, this can ensure that resources are strategically allocated to improve healthcare infrastructure and services. As the demand for more resilient healthcare systems grows, strengthening investment tracking mechanisms will be key to enhancing capacity and responsiveness in both routine and crisis scenarios.
Another extension that could be of great interest to Brazil is the generation of a health spending break-down by state, or even municipality. A good number of OECD countries have implemented their health accounts in a way that this information is calculated on a regular basis, for example, Australia, Canada, Spain, Switzerland or the United States. In Canada, for example, provinces submit data to the Canadian Institute for Health Information (CIHI), an independent federal institution, which compiles different data sources and produces SHA-based accounts at the provincial and federal levels for cross-regional and international comparisons.
Regional health spending estimates are also produced in some Latin American countries. In Mexico, for example, the annual health accounts report includes public spending on health broken down at the state level (Gobierno de México, 2025[6]). The latest available data highlights substantial differences across states (Figure 6.2).
Public spending on health per capita by federative entity, in Mexican pesos (MXN), 2023
Note: Does not include spending by SEDENA, SEMAR, or ISSFAM, as these institutions only report data at national level.
Source: Gobierno de México (2025[6]), SICUENTAS – Gasto Público en Salud, http://www.dgis.salud.gob.mx/contenidos/basesdedatos/da_sicuentas_gobmx.html.
Tracking health spending by region can be particularly useful to measure the performance of health system in a country where funding depends on subnational entities. Comparing how much regions spend per capita on healthcare is hence an important indicator to assess whether issues with equitable geographical access to healthcare exist, which can be particularly relevant in large countries with important socio‑economic differences across geographical regions. Analysing variations in private spending across regions, including both private insurance spending and out-of-pocket payments, can contribute to such as holistic assessment. Differences can indicate possible variations in the benefit packages across regions (if these are determined at a subnational level) or the unavailability of publicly financed healthcare services (because of geographical distance or long waiting times), with people resorting to privately financed healthcare.
For such a pilot exercise, Brazil could benefit from the rich data sources available to measure health expenditure on a national level and apply those to a regional analysis. For SUS spending, accounting for 42% of overall health spending, a breakdown into spending by the federation, the states and municipalities is already incorporated implicitly into the current health account approach. To calculate health spending by states, all federal spending would need to be allocated to the state and municipality benefiting from the federal transfers, and municipality spending regrouped into corresponding spending on state level. For the comparably small amount of health spending from schemes for federal civil servants, assumptions would need to be made on how these would be allocated to regions, possibly based on residence.
For private spending, the establishment a regional breakdown is typically more challenging. Yet, for Brazil, it might be feasible. The relevant data source from the ANS includes state level information for private health insurance spending. This is also true for the household budget survey (POF) used to measure out-of-pocket spending. An additional avenue to explore for a regional breakdown is the provider information based on the CNES registry. This registry allows to allocate all providers at a state level.
Overall, it appears that Brazil could benefit from the extensive availability of data at the subnational level to pilot the implementation of a regional breakdown. This would be a very useful extension of the ongoing health accounts work and could inform discussions between decision-makers at federal, state and municipal levels about the use of financial resources in the health system of the country.
A limited number of OECD countries break down health spending according to beneficiary characteristics and diseases to get a better understanding on how resources are used, and on the burden of the various diseases on public and private budgets. Combined with epidemiological information, estimates on health spending by age groups, sex and diseases can help plan future resource allocations in health, for example, to decide whether more funding is needed for mental health or cancer care to meet expected growth in demand for those services.
Generally, generating a full break down of all health spending by beneficiary characteristics and disease requires the use of data with a high level of granularity, such as patient level data. In countries with centralised health information and unique patient identifiers, combining spending data with patient characteristics and diagnosis can help with an appropriate allocation of health accounts data into age groups, sex and diseases. Yet, even without this, an allocation of spending information is still possible, for example, by identifying a variety of health provider-specific data sets that include information on patient characteristics and diagnosis. At any rate, the generation of SHA-compatible “disease accounts” is typically very resource intense.
Yet, a good number of OECD countries regularly engage in this work, including Australia, Germany, Korea, or the Netherlands. Across Europe, a 2016 study tested the feasibility of breaking down health spending by diseases within the SHA 2011 framework (Eurostat, 2016[7]). Results suggested that in many countries sufficient granularity in health data information systems would allow to distribute health spending data to diseases on an ICD‑10 chapter level. However, in some countries, data limitations left proportions of health spending unallocated (Table 6.2)
Health expenditure by disease for ICD‑10 Chapters, 2013, selected OECD countries
Note: Distribution based on allocatable current health expenditure.
Source: Eurostat (2016[7]), HEDIC – Health Expenditures by Diseases and Conditions, Statistical Working Papers, https://ec.europa.eu/eurostat/documents/3888793/7605571/KS-TC-16-008-EN-N.pdf/6cb33aa4-2e65-4df7-9b2b-1ff171eb1fba?t=1473156921000.
Among Latin America countries, Costa Rica has reported health spending by diseases in recent years, relying on the DIS-classification included in WHO’s Health Accounts Production Tool (HAPT). The DIS classification is itself linked to the International Classification of Diseases (ICD). At the moment, data is available for years 2018‑20, with spending from the 2021‑22 period currently being estimated. The aim is to consolidate the production of spending by disease estimates as part of the annual health accounts reporting cycle (Figure 6.3).
Health expenditure on NCDs in Costa Rica, 2020
Note: Disease classification is based on DIS classification (WHO HAPT).
Source: Adapted from Ministry of Health, Costa Rica (2025[8]), Cuentas de salud, https://www.ministeriodesalud.go.cr/index.php/biblioteca/material-educativo/material-publicado/indicadores-en-salud/indicadores-de-servicios-de-salud/indicadores-economicos-en-salud/cuentas-de-salud.
In the case of Brazil, based on an understanding of the structure of the underlying data sources, it would appear that an integrated approach combining SIGTAP with production data, e.g. the SIH (Hospital Information System) and SIA (Outpatient Information System), could provide a solid basis for the disaggregation of health expenditure by disease.
SIGTAP offers detailed information on medical procedures, medications, and orthoses, prostheses, and special materials (OPM) used in SUS. This information can in many cases be linked to specific diseases, offering a granular view of healthcare services and their associated costs. SIH and SIA complement this information by providing extensive activity data on hospital procedures and outpatient services. These systems categorise diagnoses and treatments using ICD (International Classification of Diseases) codes, enabling a detailed understanding of the prevalence and treatment of various diseases. By integrating the data from these different sources, it should be possible to track the costs associated with specific medical procedures and treatments for disease chapters and sub-groups, using the expansion factors to adjust the disaggregated spending to the overall SUS health spending estimates.
From the documented specifications of the various data sources related to private insurance, it would suggest that there is the potential to use data from DIOPS/ANS, SIP/ANS, and TISS to help estimate spending by disease category. For example, the financial and volume information from health plan operators, including expenses related to reimbursement of services, and other operational and administrative expenses can be explored to evaluate the structure and mapping to standard disease categories. A more challenging area is the significant proportion of spending related to out-of-pocket where typically the Household Budget Survey (POF) lacks any detail to allocate to disease. Certain assumptions can be made to distribute the spending by disease, in the absence of any additional data sources or surveys.
More broadly, there are several overall challenges to consider. First, based on country experience this requires a heavy investment in resource capacity. Ensuring compatibility between different data systems can be difficult, as differences in data formats, coding standards, and reporting practices may require significant effort to harmonise. Technical challenges, such as the need for robust infrastructure and additional skills, also play a role. Finally, interpreting the integrated data correctly and making informed decisions based on the analysis requires a deep understanding of both the healthcare system and the data itself.
In the short to medium term the production of health spending estimates based on diseases and other patient characteristics may be too ambitious for Brazil. In the near term, a more immediate priority may be strengthening the reporting of health spending by health providers (HP) and mapping revenue sources (FS) to financing schemes (HF). Among potential extensions of health accounts, generating a geographic breakdown of health spending appears to be a more feasible and immediately actionable option compared to segmenting expenditures by patient characteristics. Additionally, a geographic analysis by state would likely offer greater policy relevance, making it a practical next step in enhancing the comprehensiveness of Brazil’s health accounts.
References
[2] Brasil. Ministério da Saúde (2022), Contas de saúde na perspectiva da contabilidade internacional : conta SHA para o Brasil, 2015 a 2019, Ipea, https://doi.org/10.38116/978-65-5635-028-8.
[5] CIHI (2024), National Health Expenditure Trends, 2024: Data Tables — Series B, Canadian Institute for Health Information, https://www.cihi.ca/en/national-health-expenditure-trends#additional.
[7] Eurostat (2016), HEDIC - Health Expenditures by Diseases and Conditions, Statistical Working Papers, https://ec.europa.eu/eurostat/documents/3888793/7605571/KS-TC-16-008-EN-N.pdf/6cb33aa4-2e65-4df7-9b2b-1ff171eb1fba?t=1473156921000.
[6] Gobierno de México (2025), SICUENTAS - Gasto Público en Salud, http://www.dgis.salud.gob.mx/contenidos/basesdedatos/da_sicuentas_gobmx.html.
[8] Ministry of Health, Costa Rica (2025), Cuentas de Salud, https://www.ministeriodesalud.go.cr/index.php/biblioteca/material-educativo/material-publicado/indicadores-en-salud/indicadores-de-servicios-de-salud/indicadores-economicos-en-salud/cuentas-de-salud.
[3] OECD (2023), Health at a Glance 2023: OECD Indicators, OECD Publishing, Paris, https://doi.org/10.1787/7a7afb35-en.
[1] OECD (2018), Spending on Primay Care: First estimates, OECD Publishing, Paris, https://www.oecd.org/en/publications/spending-on-primary-care_a75a9bcb-en.html.
[4] Thomson, S., A. Sagan and E. Mossialos (eds.) (2020), Private health insurance in Brazil, Egypt and India, Cambridge University Press, https://doi.org/10.1017/9781139026468.
← 1. The SHA 2011 Manual suggests taking tax credits and income tax deductions generated by health spending into account when measuring OOP spending, in case a transaction exists. The same principle should apply in the case of premium payments to VHI.
← 2. Gross Fixed Capital Formation refers to the acquisition (minus disposal) of fixed assets. These assets need to be used in the production for more than one year.