https://orcid.org/0000-0001-7166-5737
ABSTRACT
This paper describes the development of the Brazilian version of the International Patient Summary (IPS). A joint technical team is developing the project from Syrian-Lebanese Hospital, São Paulo, and the Coordination of Innovation and Informatics in Health, Secretariat of Information and Digital Health, Brazilian Ministry of Health. The five-phased approach used in the development of Brazil IPS starts with the identification of the Brazilian Ministry of Health national terminologies used to represent concepts for problems, diagnosis and procedures, their upload in a terminology server and mapping to IPS ValueSets, the Fast Healthcare Interoperability Resources profiles customization and, finally, the complete Brazil IPS Implementation Guide is described.
INTRODUCTION
The International Patient Summary (IPS) is the result of coordinated and harmonized work involving five Standards Developing Organizations (SDOs): the European Committee for Standardization/Technical Committee 251—CEN/TC 251, the International Organization for Standardization/Technical Committee 215—ISO/TC 215, Health Level Seven International (HL7), Integrating the Healthcare Enterprise (IHE) International and Systematized Nomenclature of Medicine (SNOMED) International led by the Joint Initiative Council (JIC) [1, 2]. The JIC has been playing a significant role in harmonizing Health Informatics standards since 2007 when it was created. IPS was published as an ISO standard in 2021 by the ISO 215 Committee as ISO 27269:2021 [3]. In June 2021, the Group of Seven (G7) member countries signed the Oxford Declaration, committing to adopt IPS as a standard for exchanging health information between countries [4]. The G7 initiative was extended to the Group of 20 (G20), and the first IPS global deployment pilot focused on the Coronavirus disease 2019 (COVID-19) immunization block with the issuance of the international COVID-19 immunization certificate.
The IPS is an electronic health record extract containing essential health information for unscheduled and cross-border care settings [1, 2]. The IPS is made up of different main sections: Header—which includes the patient’s data, the health professional who generated the summary, the custodian organization of the information and the digital signature; Medication Summary—medications being used by the patient; Allergies and Intolerances—patient allergies and intolerances; List of Problems—clinical conditions of the patient; Immunizations—list of vaccines administered to the patient; Procedure History—list of relevant clinical/surgical procedures that the patient has undergone; Medical devices—medical devices used by or implanted on the patient; and Diagnostic results—current patient diagnostic results (laboratory results, anatomic pathology results and radiology results). Additional sections may also compose the IPS: vital signs, past history including obstetric and social history, functional status, care plan and advanced living directives [1, 2]. The IPS is currently under development in several countries on different continents: New Zealand, Vietnam, Holland, Sweden, England and Canada, and in Latin America through Argentina and Brazil [5]. Some countries, such as Canada, also use the IPS as a national patient summary for continuity of care [6].
IPS is based on the HL7 FHIR (Fast Healthcare Interoperability Resources), which was created in 2014 by HL7 International [7, 8]. It represents a modern standard for exchanging healthcare information electronically, designed to address the interoperability challenges healthcare systems worldwide face. One of the significant advantages of FHIR is that it uses standard web technologies, such as HTTP, RESTful application programming interface (APIs) and JSON/XML formats, facilitating lightweight and efficient data exchange. The fundamental building block of FHIR is the resources, which represent pieces of healthcare information, such as patients, observations, medications and procedures. Each FHIR resource encapsulates relevant clinical or administrative data, structured according to standardized FHIR data models. FHIR profiles further enhance interoperability by defining additional constraints, extensions and implementation guidance for specific use cases or communities. FHIR extensions accommodate extra data elements when no field exists in the standard definition. By defining standardized profiles, FHIR promotes consistency and semantic interoperability, ensuring that healthcare data are exchanged in a meaningful and interoperable manner, thereby driving improved patient care and healthcare outcomes. The IPS implementation guide (IG) defines profiles for each of the IPS sections [7, 8].
Brazil has a National Digital Health Strategy (ESD20-28) published in January 2021 [9]. The ESD 20-28 has seven strategic axes: (i) Governance and Leadership for ESD20-28; (ii) Digitization of the three Levels of Care; (iii) Support for Improving Healthcare; (iv) The User as Protagonist; (v) Human Resources Education and Training; (vi) Interconnectivity Ecosystem; and (vii) Innovation Ecosystem. The Brazilian IPS addresses axis 3 of ESD 20-28: support for Improving Care because by sharing the patient’s summary, continuity of care is assured with more quality between the different levels of care: primary, secondary and tertiary. Besides axes 3, the Brazilian IPS also attends axis 4: the User as Protagonist, since it is by the citizen that their summary will be generated, and the citizen will be able to share their summary with whom they consider appropriate. Finally, the Brazil IPS is connected to axis 6 of the Brazilian ESD20-28: Interconnectivity Ecosystem, since it promotes the integration of different sources of clinical information stored in National Healthcare Data Network (RNDS), such as outpatient encounters, immunization registries and lab exams, into one single clinical document. Thus, the development of the Brazilian IPS attends three of the seven strategic axes tESD20-28 since these are the axes most related to the IPS initiative [9].
At the core of the ESD20-28 is the National Healthcare Data Network (RNDS)—a platform that connects health information from different health information systems across states and municipalities throughout the country. The ESD20-28 vision statement says: ‘By 2028, RNDS will be established and recognized as the digital platform for innovation, information, and healthcare services for all of Brazil, for the benefit of users, citizens, patients, communities, managers, healthcare professionals, and healthcare organizations’ [9]. In May 2024, there are >1 billion 400 hundred vaccine registries and ~74 million results of COVID-19 and Monkeypox exams on the RNDS repository. Data from outpatient encounters from primary care started to be sent to RNDS at the end of 2023. Currently, 84.4 million primary care encounters are available on RNDS, as the Brazilian Ministry of Health (BMoH) informs. RNDS is a database organized in clinical documents (FHIR compositions), one for each type of information: immunization registries, laboratory results from COVID and MonkeyPox, and primary care encounters. The patient’s unique identifier indexes these databases, either the taxpayer identification number or the national health card identifier.
The IPS can also be used nationally to provide all the information necessary for the continuity of care. This is under consideration by the BMoH. While the RNDS serves as a centralized platform focusing on specific encounters, it does not provide a comprehensive patient summary. In contrast, the IPS focuses on delivering comprehensive patient summaries to support continuity of care and informed decision-making. Thus, adopting IPS nationally would guarantee that the patient could share their summary whenever necessary, avoiding searching in different encounters to find information needed for the continuity of care.
The IPS-Brazil project is a collaboration between Syrian-Lebanese Hospital and the BMoH, supported by the Support Program for Institutional Development of the Unified Health System (PROADI-SUS) [10].
The objective this work is to describe the Brazilian IPS Project development status and its contribution to the National Digital Health Strategy.
METHODS
A five-phased approach is being adopted for the development of the Brazilian IPS as follows:
Phase One consists of the identification of all dictionaries adopted in the Brazilian RNDS for the sections of Immunization, Lab Exams, Allergies-Adverse Reactions, Medications and Conditions, as well as all the CodeSystems and ValueSets indicated in the IPS IG (http://hl7.org/fhir/uv/ips) and upload these dictionaries to a terminology service compliant to the Sharing Valuesets, Codes and Maps (SVCM) IHE Profile [11] and to the Common Terminology Service 2 standard [12] under the different organizations that are expected to maintain each of these dictionaries. IPS has already established the scope, use case and design principles. Those govern the choices on structure, dictionaries and vocabularies that should be used in Brazil for IPS.
Phase Two focused on constructing the ConceptMaps that express the mapping from the Brazilian dictionaries to the IPS preferred and required terminologies. The mapping process followed the ISO/TR 12300:2014 Health Informatics—principles of mapping between terminological systems [13]. According to the standard, the purpose of the mapping process needs to be clearly stated with the definition of the source and target data dictionaries involved in the mapping. Moreover, it is necessary to indicate the cardinality and equivalence degree of the mapping between source and target. The cardinality must be expressed as: ‘one to one - a single source concept is linked with a single target concept or term; one to many - a single source concept is linked with multiple target concepts or terms; many to one - multiple source concepts are linked with a single target concept or term and many to many - multiple source concepts are linked with multiple target concepts or terms’ [9]. The equivalence degree must be expressed as follows: (i) equivalence of meaning; lexical as well as conceptual. For example, hypertension and hypertension; (ii) equivalence of meaning, but with synonymy. For example, hypertension and high blood pressure; (iii) the source concept is broader and has a less specific meaning than the target concept/term. For example, diabetes and diabetes mellitus type II; (iv) the source concept is narrower and has a more specific meaning than the target concept/term; For example, Septic Shock and Shock; (v) no map is possible. No concept was found in the target with some degree of equivalence (as measured by any other four ratings) [13]. The mapping was first depicted in a spreadsheet and later introduced as ConceptMaps with the equivalence degree in the term on a terminology server. The BMoH technical team validated all mappings, confirming that the concept maps created were correct.
Phase Three focused on the harmonization of IPS Brazil. Thus, all documents on RNDS were analyzed and harmonized with the IPS. All mandatory sections on IPS were supported. The IPS IG was strictly followed.
Phase Four consists of assembling IPS Brazil specifications, depicting all profiles, terminologies, and mappings and project design decisions for each profile.
Phase Five should validate the ability to display an IPS from an international patient using examples available at http://hl7.org/fhir/uv/ips and create a Brazilian IPS with data available at RNDS.
RESULTS
Phase one—terminology definition
International and national terminologies that are part of the information model for Immunizations, Lab Exams, Allergies/Adverse Reactions and Medications, as well as international terminologies adopted in the IPS, including SNOMED Clinical Terms (CT) IPS [14], Logical Observation Identifiers Names and Codes (LOINC) [15] and the HL7 FHIR R4 CodeSystems and ValueSets were identified and uploaded to the terminology management system. Currently, there are 113 Dictionaries (CodeSystems) distributed in eight organizations. The BMoH has 58% of the CodeSystems, while HL7 has 65% of the ValueSets.
Phase two—terminology mapping
For immunizations, Lab Exams and Allergies/Adverse Reactions sections were mapped to IPS’ terminologies, mainly SNOMED CT IPS and LOINC. The Brazilian official Immunobiologicals registrer comprises 103 concepts, including vaccines, antivenoms and some diluents, which were mapped to SNOMED IPS vaccine codes. Except for 12 diluents, 91 vaccine codes were mapped. Among those, 40 vaccine codes were mapped in semantic equivalence. Thirty-three concepts were mapped with subsumption, meaning the Brazilian concept was more restricted than the SNOMED IPS concept. For instance, COVID-19 PFIZER—COMINARTY PEDIÁTRICA is subsumed by SNOMED IPS concept—Vaccine product containing only severe acute respiratory syndrome coronavirus two messenger ribonucleic acid (medicinal product). There was no semantic loss in these mappings since the fact that Brazilian concepts were narrower because they included the manufacturer’s name in the vaccine name. This did not impact the clinical information that a given vaccine was administered.
For allergies, the Brazilian Classification of Allergies and Adverse Reactions—CBARA—was mapped to the IPS’ Allergy/Intolerance set. CBARA is an adaptation from the Portuguese classification of Allergies and Adverse Reactions (CPARA) [16]. The list contains 152 concepts, including allergens, substances and adverse reactions. Most mappings between CBARA and IPS’ Allergy/Intolerance set were mapped in semantic equivalence semantically similar: 141 (92.8%). For 11(7.2%) of the remaining concepts, mapping to the IPS’ Allergy Intolerance set was due to these concepts belonging only to the SNOMED CT Core.
For medications, the Brazilian National Ontology of Medications—OBM, an ontology inspired by the NHS Dictionary of Medicine and Devices (dm + d) —was used to describe the medicines in use by the patients (https://portal-obm.saude.gov.br) [7]. All medications described in OBM were mapped to ATC (Anatomic Therapeutic Classifications), one of the bindings recommended by IPS to classify medications [17]. All the mappings from OBM to ATC felt on equivalence level 4, meaning that the Brazilian term was more restricted since it expresses the substance, concentration, unit of measurement and pharmaceutical form. In contrast, the ATC term only expresses the class to which the medication belongs. Example: from OBM—Paracetamol 500 mg tablet to ATC N02BE01—Paracetamol (acetaminophen). While the Identification of Medicinal Products set of standards proposed by ISO/TC 215 is not fully implemented globally [18], ATC is a classification that can at least inform the medication class the patient is taking.
Only COVID-19 and Monkeypox exams are currently available on RNDS. The BMoH uses LOINC codes to describe these concepts, so there was no need for mapping for observation results. All other lab exams will be included in 2024.
Phases three and four—building and implementing the specification
The IPS Brazil IG was built using FHIR ShortHand—FSH and FHIR IG Publisher. It is based on IPS, adding CodeSystems, ValueSets and ConceptMaps from National dictionaries to IPS terminologies. The specification is supposed to be the highest IG in the national hierarchy, meaning it is open, minimal, agnostic and extensible. To accomplish that, every profile is based on its respective IPS profile, defining national identifiers, extensions, semantic bindings and mappings. The IG was specified following HL7’s national implementation guidelines. Thus, it contains minimal cardinality restriction for its elements, with optional extensions, open slices and extensible bindings. Slicing is one of the most useful capabilities of FHIR when building profiles for specific use cases. In the IPS, IG slices are open, meaning that any resource instance can conform with elements that do not match any defined slices if they satisfy the remaining profile constraints. A harmonization with existing RNDS identifiers and extensions was performed to represent information already available, such as the definition of some national identifiers for individuals and extensions for native Brazilian ethnicities, as well as to capture information on patients’ birth sex and gender identity.
To implement the IPS Brazil, a patient summary curator had to be built. The curator is an FHIR Broker Adapter that consumes outpatient encounters, Vaccination registries, and Observation results for COVID-19 and Monkeypox exams from the following clinical documents standardized by the BMoH: Registro Atendimento Clínico -RAC, Registro de Imunobiológico Administrado -RIA, and Registro de Exame Laboratorial—REL. Based on the patient’s data stored in these different clinical documents, within a past recent and clinically relevant time period, the curator composes the patient’s summary using conditions, problems, lab exams and medications information. The data on RNDS conforms to national and local semantic definitions. Thus, the broker must transform syntax and semantics based on the definitions. So, it applies structure maps to data extracted from RNDS related to problems, conditions, procedures, and medication, transforming it into a valid IPS FHIR document. The semantic mapping between national and IPS terminologies is also specified in the IG that governs the FHIR transformation and translation operations. Also, the semantic mapping is performed using a terminology service, which, in this case, is Open Concept Lab (https://openconceptlab.org/), and concepts are mapped by using FHIR translate transactions. The IPS Brazil IG is currently published as a draft (https://ips-brasil.web.app) (Fig. 1).
IPS Brazil’s architecture
Phase five—validating the IPS Brazil
The IPS Brazil starts when citizens request their summary using Meu SUS Digital (BMoH’s patient health record super App). At this moment, the patient also informs in which language the summary should be generated: Brazilian Portuguese or English. A curator is responsible for selecting documents through outpatient encounters from Registro de Atendimento Clínico—RAC that contain active problems. These documents already contain data from Conditions, Medications and Allergies/Intolerances that will be extracted, composed and transformed into a summary. Subsequently, Immunization registries from Registro de Imunobiológico Administrado—RIA and Observation results for COVID-19 and Monkeypox exams from Registro de Exame Laboratorial—REL are added to the summary. So far, RNDS only has outpatient encounters provided by the Primary Care Electronic Health Record System (PEC e-SUS AB), installed in ~85% of the almost 60 thousand primary care units throughout the country. Therefore, the patient summary currently collects data from outpatient encounters, Immunization Registries, COVID-19 and Monkeypox exams.
Once generated, the IPS Brazil summary can be visualized using the Meu SUS Digital super app. Also, the summary can be shared using a SMART Health Link compliant (https://docs.smarthealthit.org/smart-health-links/) QR Code. Thus, the citizen can, configuring a pin- or time-based access, share their summary via QR code with whoever they/intend to. So far, the decision is not to build a national repository of patient summaries but to allow citizens to save it locally on their mobile devices. Security is therefore enforced by using a pin access code associated with time-based access defined by the citizen.
Through a public-private partnership program in Brazil called PROADI-SUS, BMoH participated in the LACPASS Connectathon in two of the event’s three tracks (https://racsel.org/en/conectaton2023/) [19]. Track 1 dealt with IPS—the objective was for each country to generate a valid IPS according to the IPS IG and to interoperate among other countries according to Mobile access to Health Documents transactions, while Track 2a objective was to generate and interoperate a valid vaccine certificate according to the World Health Organization Digital Documentation and Content Center (WHO DDCC) standard. Sixteen countries from Latin America and the Caribbean (LAC) participated in the event: Argentina, Barbados, Belize, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Panama, Paraguay, Peru, Suriname, Uruguay and Brazil received the highlight award for tracks 1 and 2 participation. According to the LACPASS Connectathon Report 2023 summary sent to the BMoH, Brazil successfully composed three different IPS documents according to the examples. These documents were tested by two different countries: El Salvador and Chile. Brazil also tested different IPS documents generated by Chile, Argentina, Costa Rica and El Salvador. On Track 2a, Brazil was able to generate the COVID certificate recognized by the universal validator of WHO and tested by Chile MoH. All documents were generated in English.
DISCUSSION
According to the Handbook—Digital Health Platform—Building a Digital Information Infrastructure (Infostructure) for Health, from International Telecommunication Union and WHO, one key component of a digital health architecture is the Terminology Service, among others, such as national registries to identify patients, healthcare professionals and healthcare providers as well as security and interoperability standards [20]. Brazil has unique identifiers for all citizens with a national database with >220 million records. In addition, there is a national registry of all healthcare providers, private and public, and all healthcare professionals. A General Data Protection Law (LGPD) in the country was issued in 2018 based on the European General Data Protection Regulation Law [21]. RNDS is dictating the national rule on interoperability by mandating FHIR as the national standard to send data to the RNDS repository. As already mentioned above, all COVID-19 and Monkeypox lab results performed anywhere in Brazil must be sent to RNDS. In addition to Lab reports, immunization registries and outpatient encounters provided by the Primary Care Electronic Health Record System (PEC e-SUS AB) are also sent to RNDS. However, until now, no terminology ‘infostructure’ has been established. The IPS project fills this void by offering an open-source terminology service—OCL with the local dictionaries and SNOMED IPS and LOINC with the respective mapping between them, incorporating it into the RNDS infrastructure.
During the COVID-19 pandemic, BMoH had to rapidly develop the FHIR profiles to receive in the RNDS repository all the COVID-19 and, more recently, Monkeypox exams from private and public labs throughout the country, as well as all the vaccination registries from COVID. In addition to that, the national vaccination registry was transferred to the RNDS repository at the same time.
The FHIR profiles developed for RNDS are published on Simplifier at https://simplifier.net/RedeNacionaldeDadosemSaude. There are 40 FHIR profiles, 72 CodeSystems, 93 ValueSets defined and 32 extensions. These profiles describe specific use cases for sending Lab reports or Immunization Registries. The BMoH considers the Brazilian IPS IG to become the Brazilian BR Core IG because it is open, extensible and agnostic to specific use cases. This means that the IPS IG can become the national reference to accommodate all the interoperability needs from different use cases. The current use cases defined by RNDS, such as primary care encounters, hospital discharge summaries, electronic prescriptions, obstetric discharge summaries and private insurance claims, are now being revised to attend to the IPS BR Core IG requirements.
IPS is a significant contribution to patient care in Brazil, considering the challenges of territorial extension, rich cultural diversity, and over 230 million people—therefore, users—of the Unified Health System (SUS). By adopting IPS, it will be possible for healthcare data to overcome healthcare boundaries inside public health and interconnect with private healthcare systems also available in the country. This is important since it is not rare for citizens to rely on public and private healthcare through different stages of life, depending on socioeconomic, demographic and even disease-related factors. This frequently generates different and non-interoperable clinical information records that can directly impact patients’ care.
IPS also shifts the logic on patient data property while empowering citizens throughout the healthcare system: It focuses on the patient’s right to own and discretionarily share their medical records, reinforcing their role and responsibilities as participants in their treatment. Granting patients this autonomy enables them to effectively advocate for their own care. Moreover, IPS streamlines the process by eliminating the need for patients to spend extra time informing every new healthcare professional about their medical history. Instead, it provides a patient healthcare summary at the point of care, reducing care gaps and unnecessary tests or procedures, ultimately leading to improved patient outcomes.
CONCLUSION
This project makes a structuring contribution to the SUS, specifically to the new Digital Health and Information Secretariat of the BMoH, created in January 2023. The infostructure of health terminology services for RNDS developed by this project will allow the BMoH to manage in a single infrastructure and automatically distribute, via APIs, all the terminologies used in the country, with their respective versions and mappings between them. It is expected that in 2024, the Brazilian IPS will be fully operational in the Meu SUS Digital super application with all the mandatory and recommended sections of the international summary, except for the devices section.
Additionally, the IG of the IPS Brazil is an open and extensible canonical guide of HL7 FHIR R4. It may become the highest hierarchy IG in the RNDS so that other use cases in the country can fit a set of open and extensible profiles.
One critical success factor of the project is the weekly meetings with the BMoH technical team, which allows for fast resolutions of any project need, such as revising the mappings from the local dictionaries or accessing data for proof of concept.
Implementing the IPS in Brazil aligns with the other G20 countries and the GDHP initiative, which aims to accelerate the adoption of the Summary internationally.
This project’s most significant contribution is the construction of the digital health infostructure for expanding the RNDS.
ACKNOWLEDGEMENT
We thank Mr Milton Tawamba da Silveira Junior for editing the figure in the manuscript.
STUDY FUNDING
This project is funded by the Support Program for Institutional Development of the Unified Health System (PROADI-SUS).
APC FUNDING
The Support Program for Institutional Development of the Unified Health System (PROADI-SUS) funded the article-processing charges.
CONFLICT OF INTEREST
None declared.
AUTHORS’ CONTRIBUTIONS
B.F.L. and I.M.A.C. conceived the paper’s overall structure, contributing to all its sections. J.M. was responsible for the Allergy and Adverse Reactions sections of Brazil IPS, and for adapting the FHIR Composition of IPS. She contributed to the Methods, Results and Discussion. A.R.Z., M.A.M. and R.W.M.M. were responsible for the Medication section of Brazil IPS. A.Z. and M.A.M. also contributed to the Methods, Results and Discussion. K.L.A.C.C. contributed to the Results and Discussion section of the paper. B.H.M., E.S.S., G.N.N., P.X.S., J.E.B.O., G.G.O. and S.D.G. contributed to the revision of the text as well as the Results, Discussion and Conclusions. F.R.M. contributed to the discussion and made the editorial corrections and quality control of the text, verifying its compliance with the author’s recommendations. All authors reviewed the final version of the manuscript and approved the submission of this manuscript to this journal.
DATA AVAILABILITY
The data to generate the International Patient Summary reside in the RNDS, and according to the Brazilian HealthCare Privacy Act, LGPD can only be shared if the person who owns the information allows it.
