One country's journey to interoperability: Tanzania's experience developing a national health information exchange

Background and purpose: Robust, exible, and integrated health information (HIS) systems are essential to achieving national and international goals in health and development. Such systems are still uncommon in most low and middle income countries. This article describes a rst-phase activity in Tanzania to integrate the country’s vertical health management information system with the help of an interoperability layer that enables cross-program data exchange. Methods: From 2014 to 2019, the Tanzanian government and partners implemented a ve-step procedure based on the “Mind the GAPS” (governance, architecture, program management, and standards) framework and using both proprietary and open-source tools. In collaboration with multiple stakeholders, the team developed the system to address major data challenges via four fully documented “use case scenarios” addressing data exchange among hospitals, between services and the supply chain, across digital data systems, and within the supply chain reporting system. This work included developing the architecture for health system data exchange, putting a middleware interoperability layer in place to facilitate the exchange, and training to support use of the system and the data it generates. Results: Tanzania successfully completed the ve-step procedure for all four use cases. Data exchange is currently enabled among 15 separate information systems, and has resulted in improved data availability and signicant time savings. The government has adopted the health information exchange within the national strategy for health care information, and the system is being operated and managed by Tanzanian ocials. Conclusion: Developing an integrated HIS requires a signicant time investment; but ultimately benet both programs and patients. Tanzania’s experience may interest countries that are developing their HIS programs.

In 2005, the MOHCDGEC began working with the USAID | DELIVER Project and other partners to improve national health information logistics and strengthen human capacity and systems design. This work led to implementation of the Integrated Logistic System Gateway, a mobile reporting system designed to increase the visibility of logistics data and improve product availability, and subsequently the electronic Logistics Management Information System (eLMIS), which increases supply chain visibility at all system levels [31][32][33]. Other routine and non-routine health information were collected through household surveys (such as the Demographic and Health Surveys), health facility and administrative databases, census and vital events registration data, the Integrated Disease Surveillance and Response framework for communicable disease surveillance, and mobile technology for data collection on immunizations and neglected tropical diseases [35][36][37][38].
Over the past decade, the use of mobile technologies (mHealth), electronic medical records (EMR) at hospitals, medical decision-making tools, and other health-related tracking systems for infectious disease assessment, community-based services delivery, outreach services, and need identi cation and prioritization has increased [35][36][37][38].
However, in Tanzania as elsewhere, these tools have not always linked effectively to the overall HIS data, leading to duplication of work, data quality errors due to manual transmission, and inappropriate use of data. In Tanzania, the lack of electronic data exchange hindered service delivery, and weakened linkages between health information system components. [18][19][20][21] Limited interoperability also diminished information accessibility, compatibility, and sharing across data sources housed at universities, professional councils, non-governmental organizations, the MOHCDGEC, [3] and the Ministries of Education and Finance, resulting in chronic data gaps and missed opportunities to use new and promising practices, tools, and approaches.
To address these gaps, the government of Tanzania and its partners examined several global networks that help countries develop common frameworks and design, and implement HIS architecture and interoperability. One such network is AeHIN (Asia eHealth Information Network), which promotes use of ICT in Asia and uses the "Mind the GAPS (governance, architecture, program management, and standards) framework" to support countries. Another community of practice supporting architecture and standards-based interoperability in health sector is OpenHIE (Open Health Information Exchange) which has developed various resources around HIS architecture, system functionality, and standards, and provides access to various open-source digital solutions that are available for countries to adopt and learn.
To move toward interoperability throughout the health care system, the MOHCDGEC employed AeHIN's ve-step approach. These enabled the incorporation of an interoperability layer into the development and implementation of an integrated health information exchange (TZ-HIE). The work in Tanzania also emphasized two more priorities-improving data use, and capacity building for sustainability adopting a GAPS-CU (capacity and use) approach (see in Figure 1 and details in subsequent text): 1. Identifying leadership structures and roles 2. De ning public health information system priorities 3. Designing the HIE architecture 4. Designing, testing, and implementing the system 5. Building capacity and supporting data use As part of Phase I implementation, the MOHCDGEC prioritized four use cases to demonstrate how interoperability would work.(described in Step 2). This article focuses describes the process by focusing on the rst use case-improving access to and use of data across specialized hospitals. However, the Ministry's approach followed the same ve GAPS-CU steps for design and implementation of each case. The Broadband Commission's Working Group on Digital Health states that, "strong leadership, intersectoral collaboration and clear governance are essential for effective implementation of a national digital health strategy." [1] To develop a harmonized overall health information system in Tanzania, it was crucial to identify the leadership and governance structure for the HIE, including coordination, partnerships, and nancing. MCSP's primary activity was to support a Ministry-led process that brought stakeholders together to review key concepts and global models, review national applications, develop a common vision, and determine high-level requirements for a comprehensive HIE framework.
Through these discussions, the MOHCDGEC and its development partners de ned the high-level data exchange framework that leverages digital health technology to improve key aspects of the health system; achieve the strategic objectives of the Fourth Health Sector Strategic Plan (HSSP-IV); and identify the eHealth system's architecture objectives, priorities, and gaps. A governance structure was established ( Figure 2) under the leadership of the Permanent Secretary of the MOHCDGEC as the chair of the National eHealth Steering Committee. The Ministry's ICT unit served as the secretariat, and a Project Management O ce (PMO) was formed to coordinate development of multiple elements of the HIS. Under the PMO, four technical working groups (TWGs) were created: care delivery, health care resources, decision support, and interoperability. More than 25 government subject-matter experts from national hospitals, partners, and local and regional government o ces took part in the development process and participated in a range of steering committees and working groups.

Fostering discussion and education
To facilitate discussion among the partners and stakeholders, the Ministry held a series of meetings and workshops. The process included review of case studies to examine examples used by other projects and countries, and brainstorming sessions to identify issues and understand key concepts (e.g., measuring change). After intense discussions, participants voted on priorities and approaches, and drivers and challenges were addressed at multiple points in the process. Detailed timelines are described in Figure 3. Among the mechanisms used were: A National eHealth Steering Committee to review and give feedback on the Tz-HIE Project charter.
Tz-HIE TWG was created to review and provide feedback on the requirements speci cations.
Various stakeholder workshops were held on use cases, Requirements Speci cations solicitation, governance principles, acquisition strategy, and development.
A high-level stakeholder meeting was held and o ciated by the Permanent Secretary of the MOHCDGEC.

Clarifying stakeholders' roles
Coordinating participating organizations [2] within Tanzania's eHealth system also required an understanding and alignment of their roles, since each organization operated under its own legal, technical, and political parameters. A situational analysis was conducted to determine the legal and technical environment, governance, data standards, and systems within each participating organization.
Decision-makers in some organizations were hesitant about participating, and it was vital to hold meetings with organizational leaders and staff to determine logistics and partner commitmentstimelines, legal implications, implications for information technology, and roles and responsibilities.
The Permanent Secretary for Health led the process, which helped to rea rm the government's commitment to support the Tz-HIE and improve the country's overall HIS development. This proactive approach to engagement enabled the MOHCDGEC and partners to increase their ownership of and capacity to manage the Tz-HIE. Box 1. Business Use Cases to Demonstrate Interoperability A use case is an explicit, fully documented description of an activity or business/task to be performed, including interactions between multiple people, groups or systems. For example, in the case of maternal health, antenatal care is provided by multiple health workers (doctors, nurses, midwife) at different levels of the health system. The use case will de ne these interactions and identify the data exchange requirements across these multiple systems. The description includes the stakeholders, the inputs needed for the task, the purpose of the task, and what the task is expected to produce. The primary actors and their responsibilities are also identi ed. The MOHCDGED employed use cases to demonstrate improved health information transfer between national and sub-national information systems.

Step 2: De ning public HIS priorities
Once the governance structure is in place, the TWGs focused on the second step to identify the biggest data challenges that could be addressed via interoperability-development of a "middleware" function (see further details in Step 3 below). This brought together senior Ministry o cials and partners to identify, prioritize, and develop use cases, including detailed documentation of requirements and speci cations (Box 1); gather input from stakeholders; customize and con gure the system; and test the system based on use case speci cations.
Based on the prevailing challenges and the Ministry's immediate needs, the project team engaged speci c stakeholder groups to develop four major use cases to enable: 1. Client-level data exchange for priority hospitals, especially national and specialized hospitals. It has been di cult obtain data from these facilities, since they do not report through DHIS2. There was a need to track the performance in these hospitals on a regular basis, looking at bed occupancy, services delivered, deaths occurring, and revenue collected. Stakeholders for the development of this use case included hospital technical and administrative staff. Requirements for each of the de ned use cases were grouped into functional and system requirements.
Functional requirements describe what the system should do-such as its ability to exchange client-level data in a single repository, search for records with data quality issues, and so on. System requirements describe how the system should perform: this includes functions such as queuing, translating, error messaging, etc.
The team observed several challenges with improving electronic data exchange: A variety of digital tools to support various domains in the health sector, e.g., DHIS2, eLMIS, HRHIS, HFR, and others.
Use of different Electronic Medical Records (EMRs) by public and private health facilities use based on their need and process.
Non-standardization of data across multiple systems. Few systems have adopted any data standards for recording information.
Most of the systems e.g. EMRs from Hospitals are missing APIs for integration.
Non-existence of standardized unique identi ers of individual clients/patients in the Tanzanian health sector.

2.3
Step 3: Designing Health Information Exchange architecture

The Enterprise Architecture approach
The Tanzanian Public Service Management and e-Government Agency of the President's O ce developed the national Enterprise Architecture (EA) Approach, national EA policy standards, guidelines, and an operations manual in 2013 to ensure seamless exchange of health information [39][40][41][42]. The Tanzania EA focused on putting the user rst, particularly health care workers and data users, to highlight the decisions they make to provide effective care. Consequently, Tanzania's EA approach required all digital health stakeholders to incorporate interoperability, open standards, exibility, collaboration, and technology into their designs.
The EA approach helps identify information needs across multiple domains and health sector building blocks [45]. The Ministry of Health, Community Development, Gender, Elderly and Children chose the EA approach to help de ne and frame data exchange and interoperability across levels and health domains. This approach de nes the health sector as an enterprise, in contrast to the domain-speci c (i.e., service delivery, supply chain, etc.) traditional approaches to HIS development. EA enables a more harmonized development of HIS and facilitates identi cation of data exchange needs between domains.
The major focus in this activity was on the application architecture-the behavior and interactions of the applications used within a business or enterprise. At the higher level, the Tanzania HIE model provides a structural conception that describes how components t to one another to share and exchange information. At lower levels, the HIE requires detailed interactions between various components of service delivery and systems for a speci c service delivery area. For example, unique identi ers are required to track continuity of care, integrated care, insurance coverage, referrals, etc. The system architecture "maps" these processes and guides development of a harmonized system.

Collaborative decision-making on system elements
Developing this system was time-consuming and required continuous collaboration and feedback. Throughout the ve-year process of design and implementation, Tanzania incorporated input from partners and stakeholders, using a collaborative approach to ensure that the system would meet health sector needs, be easy to learn and use, incorporate the principles and products associated with EA and open access, and promote buy-in.
The National eHealth Steering Committee provided overall leadership and governance for the HIE operation; and the Ministry's ICT Unit served as the secretariat and management o ce or PMO. The Tz-HIE blueprint ( Figure 4) represents a dynamic environment that adapts to changing business, information technology, and data requirements.
The activity in general provided an opportunity to look at the broader health sector-wide need for data and the functions of the proposed data exchange. Once this was done, the next step was to develop use casespeci c data exchange architectures and identify components that need to work together. For the purpose of use case 1, the following components of the architecture ( Figure 5) were identi ed as essential: 1. Health Information Mediator: Information sharing and exchange across systems is mediated through a middleware, the Health Information Mediator, or HIM (shown in Figure 4). A mediator is an essential component of integrated system architecture that facilitates data exchange across multiple systems. It manages functions such as authentication, queuing of messages data translation and data quality check. As of early 2019, the HIM began integrating data across ve health domains-Hospital Information Management, mHealth, HMIS, Immunization, and Logistics-each with corresponding sub-domains and their data. HIM implementation addresses the challenges of the point-to-point data exchange by reducing the number of changes that are required to be made to all system connections when one system is modi ed.
2. Health Data Repository or HDR: A database to act as a central repository for all client data collected from multiple hospitals. The repository provides managers access to a real-time from multiple hospitals in a single database.
3. Terminology services: Houses data standards and data quality protocols and ensures that all transactions are meeting the de ned standards and quality protocols.

A dashboard:
To provide visualization and analytical features of performance across multiple health facility activities through the HDR.
The Ministry is now using the HIE conceptual model to align investments and harmonize future development in a national health information system that uses integrated digital technology to provide data for improved decision-making at all levels of the health system.

Standardizing data and codes for interoperability
Information sharing through the HIM, and comparison of information across multiple facilities, required harmonization and standardization of service codes. Once the use cases were de ned and the architecture designed, the next steps entailed identifying and adopting standards to enable data exchange. For use case 1, a list of data standards were identi ed, including ICD10 codes for diseases and mortality data and Current Procedure Terminology (CPT) codes for recording procedures. In relation to data standards for use case 1, the PMO team observed that: 1. All hospitals were using ICD10 codes for recording disease and mortality information.  Table 1 shows the standards used for data exchange; Table 2 shows rules for data processing.
Addressing the challenge of non-standard codes for procedures required bringing together health care professionals-including clinicians, radiologists, lab technicians, anesthesiologists, surgeons, cardiologists, physiotherapists, and others-who developed a list of services to be standardized across the country. A standard list of services was de ned using the CPT 4 codes from the American Medical Association. These codes were customized for Tanzania and integrated in the terminology services to facilitate data exchange. The list was then mapped against the custom codes used by the hospitals. A digital crosswalk was developed to map the custom hospital codes with the central standard codes, which would enable data from the customized system to the central system. If an exact match was not found, a code nearest to the central code was assigned ( Table 3).
The HIM, or mediator, incorporated a degree of exibility to enable alignment with all the systems. For recording dates, if one system recorded client data using the date/month/year system, and another recorded these data using the month/date/year system, the HIM would always translate those systems into the HIE's operating system.
Data standards are critical for seamless interoperability. However, during implementation, the PMO decided that as an immediate step all custom codes from hospitals would be mapped to the existing CPT code (as shown in Table 3). This was an advantage for the HIM; mandating the use of standards for legacy systems would have been much more challenging, as the systems were already operational and health workers were used to using the custom codes. Over the long term, any new systems being developed will use the standards adopted by the MOHCDGEC, and if there is an opportunity, legacy systems will migrate to using the standardized codes. Introducing new standards did not force organizations to switch, because the exible interoperability layer translated data from custom to standard code sets.

Developing guidance on using the interoperable system
To address emerging issues in the future, the MOHCDGEC developed standards, policy guidelines, and a conceptual framework to organize partners and other stakeholders using digital technology in health care. This approach provided the MOHCDGEC with a complete picture of activities to mobilize and commit resources for speci c activities. By late 2017, development of integrated guidelines for facilities using electronic management information systems-guidelines explaining how to use the HIE-allowed the Ministry to assess and strengthen 9 hospital management information systems from 33 national, specialized, and regional hospitals. The conceptual framework, standards, and policy guidelines were each critical to enhancing the system's ability to integrate data.

2.4
Step 4: Designing, testing, and implementing the system The next step was to customize the system and conduct conformance testing. Once each use case was prioritized and architecture and interoperability needs identi ed (steps 1 and 2), the PMO team reviewed and identi ed an interoperability layer that would t the Ministry's needs. Multiple options on the market were already being used to support interoperability in various settings-such as OpenHIM, [3] OpenFN, [4] HEALTHeLINK, [5] mulesoft, [6] and many others. The PMO team reviewed the performance of various interoperability layer tools against a variety of use cases. The vendors of the tools were given the use case scenarios and asked to demonstrate their tool's functionalities and ability to manage the proposed use cases.
This was a crucial stage in the journey toward improving interoperability in the Tanzanian health sector. The PMO, which consisted of representatives from multiple organizations and experts, played the role of an independent advisory group. The group reviewed the available tools and advised the MOHCDGEC team to adopt the HEALTHeLINK tool (version 3), which has been used in the U.S. for the past 13 years. With support from MCSP, the team of software developers from HEALTHeLINK worked with the MOHCDGEC and partner organizations to con gure the system based on the rules and requirements outlined. The system was then taken through conformance testing to ensure that all requirements were being met and that the system could manage all transactions.
The team found that some systems were unable to participate fully on the HIE, as they were underdeveloped or used outdated technologies that do not support interoperability. To address this challenge, the system was con gured to accept data through le uploads (e.g., export/import of XLS les) to the HIE, providing a exibility that is particularly important in low-resource countries to accommodate key functions. This was a key functionality offered by the chosen tool, HEALTHeLINK.

Tools for users
Comprehensive support tools and structures were developed to ensure that the system ran smoothly and supported sustainability. The HIE's sustainability depends on its exibility and scalability, as seen in its ability to easily accommodate new use cases or extend existing use cases to cover more organizations. Tools developed for Tanzania included a systems installation manual, users' operationalization guide, and system administration and implementation guides. The implementation guide was based on rsthand experiences from the Tanzanian context, and provides insight on the HIE processes and best practices to follow when "on-boarding" or connecting new or existing organizations or systems to the HIE.

2.5
Step 5: Building capacity and supporting data use 2.5.1 Capacity building Training and capacity building for both technical staff and users took place throughout the implementation of Tz-HIE. The Ministry used a structured training methodology and standardized training materials for different groups of users. The training methodology was predominantly hands-on, where participants were given access to the system, with facilitator-led demonstrations and presentations, group assignments, pre-tests, quizzes, and post-tests meant to cement users' understanding and assimilation of the key issues. To further enhance use of the system, the Ministry's ICT department provides three tiers/levels of escalated support structure. Level 1 is the technical support team for users at the health facility level. Level 2 is the operations team; and Level 3 targets system administrators.

On-the-job support
MCSP also provided technical support to the ICT department by placing a full-time advisor seconded to the ICT unit. The Advisor's role was to provide technical support the ICT staff and ensure the eHealth Strategy initiatives are implemented as planned. The seconded advisor role included improving coordination across the development of the various eHealth initiatives (including design and implementation of the mediator), identifying the use cases, engaging different departments of the MOHCDGEC, and playing the role of a secretariat for the digital health initiatives in the health sector.

Global conferences and meetings
MOHCDGEC o cials were supported to take part in various international conferences and study tourssuch as the annual Global Digital Health Forum in Washington, DC, the Public Health Informatics (PHI) Conference in Atlanta, GA, and a 2016 HIS study tour in Boston, Massachusetts. The PHI conference provided an opportunity to learn from experiences with interoperability across multiple programmatic area and geographic states in the U.S. The study tour in Boston enabled the MOHCDGEC team to interact with the Massachusetts Department of Health team and learn why and how they are investing in interoperability, and how they use data to make programmatic decisions.

Data visualization
To foster improvements in use of the available data by the MOHCDGEC leadership team, dashboards were developed to summarize the data and create insights. These were then displayed on wall-mounted TV screens. The system was set to update the dashboards as new data were sent from the hospitals. The dashboards are also available via a web-link, giving managers online access and enabling them to explore the features that interest them, such as performance gaps, time trends, or performance variations by region or organization.

System management
An administrator's dashboard was added to the HIM to summarize the total number of systems connected to the HIM and the number of transactions successfully performed. An additional feature is the capacity to examine transactions that had errors or were not successfully performed. System administrators can use this feature to provide feedback to stakeholders on errors and improvement strategies, which helps to improve data quality.

Results
Following the steps described above, Tanzania successfully completed the HIM implementation steps for all four use cases developed, enabling data exchange among 14 separate information management systems in the Tanzania HIS. The leadership team has completed ve implementation steps for three of the business use cases de ned by the leadership team: 1) client-level data exchange for priority hospitals; 2) aggregate data exchange for DHIS2; 3) health facility registry data extraction. As of this writing, the HIM is able to exchange data among the 14 systems.

Community of practice
The process of developing Tanzania's health system operability led to creation of a local community of practice (COP) with almost 25 participants meeting in person and virtually via WhatsApp and Skype.
These participants learn together, work together, and support each other's work. Some participating organizations had limited technical resources, both in number and in terms of their capacity. The COP enabled these organizations and facilities to share experienced technical staff among participating organizations towards a common goal. For instance, Mbeya Hospital's technical resource provided technical support to staff at Kibong'oto Infectious Disease Hospital.

Institutionalization of an HIE strategy
To help stakeholders consistently and systematically adhere to national standards, the HIE serves as the national reference guide to support data exchange among multiple health systems. Even as the MCSP Project began its close-out, the MOHCDGEC adopted the HIE as part of the national strategy for health care information, stipulating that any further investment in the system would follow the ve-step process outlined above.

Improved data quality and use
The HIM interoperability layer, as a core component in health information exchange, plays an important role in improving data quality and encouraging evidence-based planning and problem solving at all levels of the health system. For example, during implementation of the health data repository, the team realized that the data received at the HIE from participating organizations suffered from quality issues, such as data with future and historical dates, persons with discharge date prior to admission date, and reports of the same individuals dying in more than one ward. The interoperability layer's lters captured these quality issues and requested recti cation from the organization. The data quality problems captured by the layer also acted as a trigger for the organizations to further improve their own electronic systems. Table 4 describes how the business cases changed: the challenges in these areas before implementation of the HIM, and system and programmatic bene ts after implementation.

Improved e ciency in data reporting and management
Improved data availability has been a key bene t of using automated data exchange. Before implementation of the mediator, the MOHCDGEC had no insights into hospital key performance indicators, such as bed occupancy, average length of stay, mortality by cause, revenue generated, or exemptions, because hospitals were not reporting them. Today, an average of 50,000 client records from among the ve specialty hospitals are summarized electronically and sent to the HDR and dashboard every month. This has saved hospital staff the time needed to manually summarize these data each month, and also reduces the data quality errors that are bound to happen due to such a large number of transactions. Future implementation and scale-up of the mediator's use will further reduce the time spent on manually summarizing client-level data, freeing more time to spend on further analysis and problem solving.

Local ownership and capacity
MCSP closed its operations in Tanzania in June 2019, at the end of its agreement with USAID. The MOHCDGEC has the source codes of the mediator and trained staff who continue to maintain and manage the system. The ICT department continues to con gure additional hospitals into the system and connect them to the mediator. The full system (the Health Data Mediator, Health Data Repository, Terminology Services, and dashboard with all software components) and all the data are hosted at the National Data Center in Tanzania; and Tanzanian in-country experts are solely responsible for managing, running, and scaling up the system. Limited external expertise will be required to provide handholding support for maintenance of the system and enhancing system performance and upgrades. In future, external support will be required to add additional use cases, such as HIV/AIDS case based surveillance, RMNCH continuum of care and tracking of clients, etc.

Discussion
Improvements in data use should drive investments for building stronger interoperable health information systems. The steps described here enabled the MOHCDGEC and its partners to establish a strong base, including processes, guidance, and tools, for building interoperability across Tanzania's health data systems.
Although there are still clear gaps in key components of Tanzania's eHealth architecture (e.g., electronic medical records at lower levels of the health system, shared health records, master patient registry, etc.), the Ministry's ICT unit is applying a stepwise approach to develop a comprehensive digital information system that aligns with the country's broader e-Government policies and is guided by requirements and standards. Tanzania has planned implementation of multiple elements of the data exchange framework, and activities are already underway to link applications and automate data interchange. Data linkage and sharing will support greater transparency, inform policy-and decision-making, and improve client outcomes. Investment in technology over time allows the system to keep pace with innovative ways to collect, manage, and analyze data. 7,12 Early cost-bene t analysis of some Tanzania upgrades to the HIS have shown a nancial bene t, though additional bene ts may not be seen for several years [33].
Future data exchange through the HIM mediator will further enhance the health system's ability to share information, improve quality, and reduce duplication in three ways. First, it will facilitate access to data across the continuum of care-for instance, through the shared data repository, health workers can access historical data to check a woman's history of high blood pressure during pregnancy. Second, interoperability improves integrated care and facilitates client tracking and referral from community to health facility. Third, data exchange between multiple domains will provide a wide range of data within a common repository-information about the patient and services provided; the provider and medicines/commodities provided; the service level (facility or community); and the payer and costs. This will be very useful source for quality and performance improvement initiatives. Implementation of the HIM middleware was time-consuming, requiring connection of multiple heterogeneous systems and collaboration with multiple actors, all with competing priorities. Nonetheless, the demanding, continuous foundational activities that we completed during this initial phase not only paved the way not only for disparate systems to exchange data, but also provided a platform for building common understanding among all participants. Ongoing activities, since they build on previous accomplishments, promise to be faster and incur fewer delays compared to previous activities.
Working with third-party vendors also posed challenges. For example, it was di cult to convince offshore vendors to prioritize the system enhancements we needed to build integration and interoperability into the HIE. To obtain adequate engagement and support from the vendor's side, it was critical to provide detailed information on the required steps for each use case as it affected each vendor's product.
The rst-phase activities described here represent a foundation, yet the system itself is still being developed. To sustain these gains and further amplify the use of the interoperability layer, the MOHCDGEC and its partners should continually add to the number of facilities and systems that use the interoperability layer to send data to the national health care repository. Also, developing a Shared Health Records le alongside a centralized Master Patient Index, as subcomponents of the Tanzania HIE, would enhance the exchange while maximizing the bene ts of the interoperability layer. This could, for example, enable better visibility of anonymized patient-level data; enhance integrated and longitudinal care; increase citizen access to health information; enhance decision-making at the time of care, and enable clients to provide feedback on the care they received.

Conclusions
High-income countries have made tremendous progress in interoperability, and continue to learn from and improving their systems. The U.S., for example, has seen major advances in the last 15 years, and states are supporting a range of initiatives across programmatic areas. Canada and European nations have also implemented integrated digital health programs. The concept of leveraging interoperability to improve health care has also gained signi cant notice in global and local public health.
A functioning, interoperable HIS is essential-not only to advance the WHO's Universal Health Coverage agenda, but to ensure that every citizen receives quality care at an affordable cost. Advances made in developed countries, and the emerging global awareness of the importance of gathering and using quality data in health care, create a huge opportunity to share lessons globally and locally.
However, building the system described here takes time and requires collaboration of many partners and careful consideration of public health priorities; the architectural principles, hardware, and software options needed for interoperability; decisions about d esign structure; identi cation of critical gaps and priorities; and establishment of long-term partnerships with multiple committed organizations. A clear understanding that the time-consuming, iterative efforts to improve the system are balanced against the need to build capacity, provide health services, and maintain data collection until the new system is in place is also necessary.
The Government of Tanzania's vision of a healthier population, its strong national health sector policy and eHealth strategy, and its long-term partnerships were all key ingredients in the successful development of the Tz-HIE. Countries that want to make their national health information systems more interoperable and sustainable would bene t from learning more about Tanzania Table 4. Business use case, challenges, and outcomes after implementation of Health Information Mediator