Integration among national health information systems in Brazil: the case of e-SUS Primary Care

ABSTRACT OBJECTIVE: To measure the degree of integration of the Electronic Citizen’s Record (PEC - Prontuário Eletrônico do Cidadão) of the e-SUS Primary Care Strategy (e-SUS AB - Estratégia e-SUS Atenção Básica) in the view of other Brazilian´s National Health Information Systems (SNIS - Sistemas Nacionais de Informação em Saúde), relating to the internal political-organizational structure of the Brazilian Ministry of Health (MH). METHODS: This is a qualitative case study. Data collection was carried out through document analysis and semi-structured interviews. In the first stage, we sought to clarify how many SNIS were in use in the Primary Care (PC) of the Unified Health System between 2013 and 2017. Then, we defined as criterion the maintenance of data collection interfaces by the Ministry of Health even after the implementation of the PEC/e-SUS Primary Care in order to measure the integration. RESULTS: 31 SNIS were identified in Primary Care. We observed that 12 of them were completely integrated and 15 presented no unification of interfaces related to PEC/e-SUS Primary Care. Another 4 have partial integration. By correlating these data with the political-organizational structure of the MS, we observed a greater integration with the systems managed by the Department of Primary Care and a persisted fragmentation in SNIS, especially those under the management of the Health Surveillance Department (Secretaria de Vigilância em Saúde). The disparity between the integration of the PEC/e-SUS Primary Care with the Health Surveillance SNIS is a sign of the persistence of the division and the false dichotomy between Health Care and Health Surveillance practices and processes in the Ministry of Health – even 30 years after the foundation of the SUS and unification of the state structures of social security hospital care and federal public health in MH. CONCLUSION: Although still insufficient, the systems integration carried out by the e-SUS Primary Care Strategy, which focuses on reducing user interfaces, can be considered a new fact on the SUS information and information policy agenda.


INTRODUCTION
In Brazil, health professionals, managers, and researchers live with dozens of health information systems in their work environments, with little or no integration between them. This problem has been reported since the 1980s 1 and is related to the fragmentation of the State's bureaucratic structures 2 , the absence of semantic and technological standardization [3][4] , and the low maturity of Information and Communication Technology (ICT) governance policies in organizational services 5 .
Between 2013 and 2018, 54 National Health Information Systems (SNIS -Sistemas Nacionais de Informação em Saúde) were identified in operation in Brazil, maintained by the Ministry of Health (MH), including systems for registration, notification of diseases and conditions, control and logistics of supplies and medicines, electronic medical records, laboratory management, accounting control of the production of procedures, among others 6 . Furthermore, they are incorporated into these other systems developed or acquired by states and municipalities.
Often, poor integration requires that the same data be filled in different interfaces, generating rework and increased costs, so turns out to be necessary to maintain several technological solutions with redundant functions. This duplication of records has an impact on the quality of the databases, making analysis and cross-referencing difficult and impacting the reliability of the information produced 4,7 .
Since the creation of the Brazilian Unified Health System (SUS), the state has adopted strategies to improve communication and integration between the SNIS. By the initiative of the Pan American Health Organization, the Inter-Management Network for Health Information (Ripsa -Rede Intergerencial de Informações para a Saúde) was created in the late 1990s, intending to produce relevant and integrated information for decision-making in health 8 . In the 2000s, the implementation of the SUS National User Registry (Cadsus -Cadastro Nacional de Usuário do SUS) was intended to unify the basic registration information of every Brazilian citizen who uses the SUS. The first information management model of Cadsus provided for the registration, storage, and protection of data as attributions of the municipalities, in a decentralized model that faced serious problems related to duplicate registrations, besides the existence of multiple numbers of the National Health Card (CNS -Cartão Nacional de Saúde) for the same user is frequent. This happens when the same user goes to a health service without his card, requiring the reception to make a new registration for him. This problem was only resolved with a structural change in data governance by the centralization of databases in MH. Still, only 14 SNIS were connected to the Cadsus national base in 2014 9 .
In 2013, the MH launched the e-SUS Primary Care Strategy (e-SUS AB -Estratégia e-SUS Atenção Básica), One of its objectives has been to promote greater integration between the SNIS operating in the Primary Care Units 10 by the unification of interfaces data collection in the Simplified Data Collection (CDS -Coleta de Dados Simplificada) and in the Electronic Citizens Record (PEC -Prontuário Eletrônico do Cidadão) software, developed in partnership with the Universidade Federal de Santa Catarina (UFSC) and made available free of charge to the Municipal Health Departments.
According to the MH, in 2016, data sent to the national e-SUS Primary Care database was carried out by 97% of Brazilian municipalities 11 and the electronic medical record would already be installed in 9,227 Primary Care Units, corresponding to 21% of the 42.6 thousand existing in 2017 12 . These numbers may vary by region and criteria used. For example, Lima (2018) carried out a diagnostic study on the implementation of the e-SUS Primary Care Strategy in Minas Gerais, concluding that in only 49.1% of the municipalities the strategy had actually been implemented 13 . Moreover, the advanced degree of implementation of the public policy in question seems unequivocal.
When can we say that two or more systems are integrated? In the field of Public Health, this understanding was generally related to the crossing of different bases, increasing data reliability, and producing new health information 14,15 . In turn, the debate on systems integration in the field of health informatics has been around technological solutions aimed at greater automation of communication between systems (mediated by standardized messages and files, syntactic interoperability) or semantic compatibility between the terms used in each of them (semantic interoperability) 16 . However, there is another aspect of systems integration that has been little addressed in academic production and in public health information and informatics policies. We refer to the so-called integration of user interfaces 17 , related to the design and usability of systems by workers and managers in the services responsible for inserting the primary data in the systems.
Focusing on the user-centered, we propose to describe and measure the integration of the PEC with other SUS Primary Care SNIS, relating it to the political-organizational structure of the MS and testing the hypothesis of the systems fragmentation relationship with the bureaucratic structure of the state.

METHODS
This case study assumes as object the relationship among the e-SUS Primary Care Strategy and the SNIS in Primary Care use, as proposed by Yin (2015) 18 . In the first stage, we sought to clarify how many SNIS were in production in Primary Care between the years 2013 to 2018 -necessary due to the existing conflict between official data sources. The following MH documents related to health information and informatics were analyzed: Information The responses of the MS motivated by two requests for access to information based on Law 12,527/2011 (Access to Information Law) were also analyzed, where it was asked about the list of systems in production in MH (nº 25820004082201795) and in Primary Care (nº 25820005771201806). The complete content of the responses can be accessed openly on the page of the Electronic Information System to Citizen (e-SIC -Sistema Eletrônico de Informações ao Cidadão) a .
In the second stage, the objective was to understand the situation of the integration of each of the SNIS in operation in Primary Care with the Electronic Citizen's Record (PEC) of the e-SUS Primary Care, clarifying whether the integration actions allowed workers and local managers to discontinue the use of the respective SNIS. Three key informants were interviewed: two members of the development and management team of the e-SUS Primary Care Strategy in the period from 2013 to 2017 and one member of the technical team of a national entity representing the municipal health secretariats. The analysis also included contributions from the experience of one of the researchers in the daily use of PEC/e-SUS Primary Care and other SNIS from his work as a physician in a basic health unit in Rio de Janeiro. Only those where there was consensus among the above sources were considered as integrated SNIS.
The following typology was created to classify the situation of integration between systems: • Integrated or unified user interfaces (full integration): When two or more systems become invisible to the final user. This means that the user cannot perceive that there is more than one system in operation, because the interfaces are fully integrated.
• Partially integrated or unified user interfaces (incomplete integration): It happens when integration initiatives between the interfaces were identified, but the need for a Controladoria-Geral da União (BR Continue professionals and managers continues using the two systems in their daily lives, given technical, political, or administrative reasons.
• No integration or unification of user interfaces: it happens when no action has been planned or taken for integration between system interfaces.

RESULTS
A total of 31 SNIS were found in operation in Primary Care between 2013 and 2018, most of them under the technical management of the SAS or SVS (Box).
It was found that the e-SUS Primary Care Strategy performed complete unification of user interfaces with 12 of the 31 nationally-based SNIS in use in Primary Care and incomplete integrations with four SNIS. Another 15 SNIS had no degree of integration, as shown in Table 1.
Integration varied according to the internal sector of the MS responsible for the technical management of the system ( Table 2). Although the SAS and SVS have similar numbers of SNIS in use in Primary Care, there was an important disparity between the integration of their respective systems with the e-SUS Primary Care.

DISCUSSION
The SIS fragmentation is a phenomenon reported by several authors when analyzing health information policies in different regions of the world 27-31 , but we did not find studies that dimensioned such fragmentation, based on the counting and description of the SNIS in operation. The record of 31 SNIS in use in Primary Care from 2013 to 2018 shows that the trend of fragmentation persists in the country, documented in Brazil in several technical studies [32][33][34] .
In countries that promoted projects to change this scenario, the focus has been on better integration between national databases [35][36][37] or the establishment of semantic and technological standards for recording and exchanging information between different systems 38,39 . No describing studies were about SNIS integration, focusing on the user interfaces.
When we analyze the result of the integration carried out by e-SUS Primary Care and the political-organizational structure of the MH, we noted that there was greater integration with the SNIS under the direct management of the DAB (Figure) and a low integration with systems from other secretariats beyond of SAS itself.
In the case of integration among SVS SNIS, including the Epidemiological Surveillance, Care for the Population with HIV/AIDS and Hepatitis, birth, and mortality monitoring and immunization policies, we observed only a partial advance in the integration of the e-SUS Primary Care with the SI-PNI.
The most worrisome case seemed to be the SNIS of the Department of HIV/AIDS, not even mentioned in the list of SNIS in operation in Primary Care in the MH bidding for Primary Care Units computerization 25 .
The disparity between the integration of e-SUS Primary Care with the SVS and SAS SNIS can be considered a sign that some degree of historical division and false dichotomy between Health Care and Health Surveillance practices and processes still persists 40 , even after 30 years since the foundation of the SUS and unification of the state structures of social security hospital care and federal public health in MH.
The governability of the managing area of e-SUS Primary Care over other SNIS proved to be a variable to be considered in the integration process, as there was a higher success rate of integration of e-SUS Primary Care with DAB's own SNIS when compared to the SNIS under the management of other MH departments and secretariats.
The frequency and intensity of the use of SNIS in health services was a lesser factor, given that SNIS with intense use at the local level of the SUS had incomplete or not started integration, such as the Sisreg, Horus, and Sinan Net.

CONCLUSION
Considering the complexity of integration between legacy systems in organizations, it is reasonable that after five years of the e-SUS Primary Care Strategy implementation, the operation of SNIS in Primary Care is just partially integrated. Although still insufficient, the integration of systems carried out with about ⅓ of the SNIS in operation in Primary Care, focused on reducing user interfaces, is a new fact on the agenda of the SUS information and information policy.
The relative uniqueness of the Brazilian experience of integration between health information systems is remarkable. Brazil promoted the creation of a state electronic medical record as a kind of "hub" for other SNIS created by the government, more specifically by the federal sphere. Although we believe that the "developmentalist" stance taken by the Brazilian state contributes to the expansion of computerization of primary care in the country, we pay attention to the need for greater integration of the large volume of market software in use by states and municipalities in the national SNIS bases, following the trend of investment in public regulation policies associated with greater semantic and technological interoperability between systems, as observed in other countries.