Fiocruz Primary Healthcare Dashboard - Painel Saúde Fiocruz

Health Professionals seldom have at hand daily feedback on their performance on providing continuous care to their assigned population. It's hard to keep track of what is happening with all the population they are responsible for. 

Access through Primary Health Care (PHC) is not enough, when several people are left behind of quality essential care. A pregnant woman may miss an appointment, or forget to bring her syphilis or HIV test results. A senior man with hypertension and diabetes may stop returning for follow ups, and thus not have his blood pressure or glucose measured, or even worse, stop taking medications. Managers lack a consolidated vision of gaps in access to essential health services and experience huge obstacles in prioritizing resources logistics  These examples of types of shortcomings undermine integration, coordination, and longitudinal continuity of care, weakening its quality. 

The examples represent daily challenges for both frontline health workers and policy makers in planning individual and community interventions that may improve the health of populations. 

If we put ourselves in the shoes of frontline health workers and policy makers at local, regional and central level, it is excruciating for them to know data is collected and available, but cannot be easily accessed, analyzed and interpreted according to pressing needs for intervention at the individual, family and community levels.

In Brazil and several other low- and middle-income countries, information to warn and advise health professionals about discontinued care frequently lacks standardization or validation. Although the information is available through billing forms, in local electronic health records, or in community health worker (CHW) notes, the lack of standardization leads to loss of information and rework in repeated and  relentless data entries that hampers health professionals efforts to maintain the flow and continuity of care to their assigned population.

The billing forms with data and information about the health actions performed by health workers are usually sent for accountability to the entities responsible for billing and financial transfer. These agencies in the receiving end  may be part of a philanthropic program, health insurance scheme or the national universal health system:  the time consuming data and information sent rarely return as feedback for local action. 

Electronic Health Records (EHR) are saved locally in relational databases: reports for the final users usually focus on only the number of patient visits or invoices issued. However, frontline health workers have a hard time making sense of local databases, as health sciences schools rarely teach SQL ! 

Although the databases contain a wealth of information about patients, another source, which are precious Community Health Workers (CHW) annotations, are seldom taken into account formally by Physicians and Nurse Practitioners. 

This inefficient situation can be transformed if frontline health workers are able to see the data they regularly produce presented in an user-friendly format, to be used as information required for decision-making to improve quality of healthcare. If they have access and can analyze that information, health professionals at the Primary Health Care level will be empowered to better understand which of their patients requires more attention, what populations must be targeted and what they have missed in their work: a visit, a lab test, a medication or just a routine follow-up.

We developed a software that provides a dashboard with the information that PHC professionals need to understand the demographics of the population they serve, and to go deeper on the follow-up of pregnant women and people living with hypertension and diabetes.

It runs locally on the computers of health professionals or local health managers and analyzes local data, so no Internet access is required. The software accesses data on a SQL database on the local network or on data files saved on the computer, such as Excel Spreadsheets or CSV.  

It can run on Windows or Linux computers by creating a local data science environment based on Python, Pandas and Flask, thus can be easily adapted to other scenarios by local data scientists. Dashboards are rendered in React.js to be visually attractive, responsive and render instantly.

The design and validation process was the result of two years of interviews, field visits and usability tests with PHC professionals, managers and local health authorities. We studied and tested it in 4 cities of different population sizes, from different regions of Brazil and with different scenarios of informatization: with paper based records only, running the standard EHR provided by the Ministry of Health (MoH), running proprietary solutions or a mix of these.

It has shown to be better than the existing methods  as it uses information that is frequently discarded away or sent to the MoH to never return to the local level. 

Our solution transforms local data into information about the continuity of care, by linking multiple different kinds of records: family and housing registers, CHW annotations and visit reports, physician consultations, nursing records, dentistries consultations and collective health educational activities. The link is the patient and his family, the focus is the health history: we are searching for the ones who missed important healthcare appointments or are yet to join operational groups organized for prevention and health promotion. The mission is to not leave anyone behind.

Our goal is to go beyond improving general performance of PHC, we aim to enhance health equity, improving access to better quality care and allowing teams to focus on the ones with the most urgent needs and deeper vulnerabilities.

The users of the solution are workers involved in primary care, including: physicians, nurses, nurse practitioners, dentists, community health workers, PHC managers and the local health administration at city or district level. These are the tested users of the platform, who input data and receive feedback in the dashboarding tools.

As the information is intended to support community and individual healthcare decision making,action and follow-up, the ones actually served are the population assisted by the primary healthcare team. 

Today Brazil has 47,786 family health teams, responsible to providing PHC to 148,627,648 people registered in the communities they assist.

As a team we´ve been there, at the frontline. And we kept our bonds. We are a mixed team of health professionals and health information and informatics specialists, and devoted our careers to health education.

The ones of us who are health professionals are physicians and physical therapists who served in the frontline for considerable time, working and teaching in PHC teams and public maternity clinics of impoverished quarters and towns. 

We are part of a bigger institution, the Oswaldo Cruz Foundation (Fiocruz). Fiocruz is a federal public health research and development institution based in Rio de Janeiro, founded in 1900. Fiocruz has scientific institutes and offices in 12 states around Brazil, developing research in public health, tropical medicine and other health related themes. It produces medicines, vaccines, lab tests and more recently digital health solutions. We have delivered many health informatics solutions before.

Our health informatics previous solutions have been successful at national level, delivering distant education to over 2 million brazilians, 1 million of them are registered health workers, in all 5,570 brazilian cities. Exemples of courses are the specialization course on PHC and the online emergency courses in Zika and COVID-19 pandemics. Refer to www.unasus.gov.br and campusvirtual.fiocruz.br for more information.

We devoted the past three years to develop this solution. During the pandemics field visits were limited, thus we interviewed the frontline workers: healthcare professionals, community health workers, information technology staff and PHC managers and local health authorities by phone and web calls. This year (2022), we visited PHC facilities and local health authorities offices. Those feedbacks from the frontline guided the design and technological strategies and resulted in a solution fit for the purpose: supporting data-informed decisions on primary healthcare.

We are true believers and practitioners of PHC as a pathway to universal health care, committed to improving access to quality health services. We started developing this solution in 2019 with as part of a technical cooperation agreement between our institution (Fiocruz) and the Ministry of Health (MoH) Primary Health Secretariat. 

During this period MoH policies for PHC changed to a strategy focused on specific conditions rather than comprehensive PHC. MoH also abolhished our techinical counterpart, the General Coordination of PHC Information. 

The solution was ready to deploy on July 2022 as schedule, but MoH decided to end the cooperation with Fiocruz on PHC. 

This lead to project activities end, not even a no cost extension was possible.

We see this challenge as an oportunity to convey our message: local data is useful to PHC teams, and can have a transformative effect on the quality of care. The prize will be used to deploy our solution as an open source software to every PHC team and municipality that takes interest in looking closer to their own data ir order to improve teams performance.

Adaptability. Our solution is made for and tested for real-world primary healthcare. It is built upon Brazil’s decade long experience in digitizing PHC. Anywhere in the world PHC requires information about how people live, the family, their housing, local environment, exposure to risks, conditions previous to care and the health record.  

In addition, our solution combines health care management with continuous health education for frontline health workers. Professionals have access to the data produced in their daily work in a timely manner, making it possible to identify and propose best practices and actions, campaigns and interventions in territories, communities and practices, according to the best information and evidence available. Policymakers and local managers can improve logistics and potentially save costs with more accurate actions.

We stressed the adaptability to be sure it would run in such a large and diverse country as Brazil. 

Data input is highly flexible. Records are paper based? One computer may be enough to type the PHC records for an entire team or health district, as currently done in several places in Brazil. Are there only old computers available? Let's run the solution whatever operational system is available, no graphical user interface is needed, just publishing the dashboard to localhost or local area network. 

Data is recorded locally in Excel Spreadsheets? Saved in a MySQL or PostgreSQL database? Or maybe in JSON, XML, TXT and CSV files? We can handle it due to using Pandas in the background. 

There is no Internet access? Records can be entered and processed locally. Each team can save files to a pendrive and they can be consolidated in a city report weekly or monthly. Internet access is unstable and low speed? Files can be transferred in chunks, using a delay torrent network strategy.

Feedback dashboards are highly adaptable. Only one computer available? Run on localhost. Is there a local area network available for the health facility or district? Publish the dashboard there. 

Is a whole city perspective needed? Consolidate data from multiple facilities. The city is big, and high availability is needed? We can docker swarm the solution. Are there privacy concerns? Anonymize data at source, transfer only aggregated data needed for analysis.

We designed and tested the solution in all above-mentioned situations, allowing us to ensure its adaptability. 

To do that, we had to design and implement a software developed in multiple layers: (1) the Dashboard Installer and the PHC Dashboard itself, which in turn was divided in two additional layers, the (2) back-end and (3) the front-end. 

The installer had to deal with the following constraints: some environments did not have an internet connection; the Windows version of the machine was variable; computing capacity was reduced.

The main function of the front-end was to create an interactive interface for the reports, while the back-end had the function of delivering the data resulting from data processing to the front-end. This communication resulted in the Panel reports.

In addition, the software should also meet the following requirements: application server should run in the background as a service; should comply with novel Windows security requirements; send telemetry data back to us, and finally should be able process large amounts of data - the local PHC database size could have a few hundred gigabytes.

For environments with little or no internet connection, an execution system was designed that did not use the internet to view information. The data needed to display the web pages were acquired by the system itself during the first load and weekly at times of low machine use. The software does not require an internet connection to generate and view the reports, the web pages are served and rendered locally.

 Computing power is limited in small cities and PHC facilities in impoverished neighborhoods. To avoid competing with other information systems that are used by the teams, data acquisition and preparation was run in the background overnight. The prepared data is then stored temporarily and then displayed in reports, thus not overloading the computer on which the software was installed. 

Processing was designed to run in parallel where possible. As a result, there was a significant improvement in the execution time of large amounts of data, especially in the environments of medium and large cities.

We acknowledge seriously the diversity of available resources, according to different regions and social situations. Being ready to run on such diverse scenarios is where our innovation lies to provide opportunities of improving quality of care in different settings.

In the next year our goal is to deploy the solution to Brazil, and in the next five years we expect to be operating in several countries. 

We finished the pilot testing by July 2022. We had 50 testers and covered 200,000 people on our pilot. Our goal is to grow a hundredfold, reaching 5.000 users and covering 20 million people by the next year. 

The strategy to deploy the solution in Brazil is to make it available as a freeware downloadable software for Windows - the operating system that is used in 99% of the city health departments in Brazil. It is ready to be used in PHC facilities and cities with up to 100,000 inhabitants. 

By the end of the first year our goal is to begin technical cooperation with larger cities to deploy customized versions of the solution. By the end of year two our goal is to cover 40 million people in smaller cities, and 20 million in the bigger ones, with over 100,000 population. From there we will seek to cover additional 20 million people every year, capping at a hundred million, approximately half brazilian population, 

Our vision for the next five years is that after a successful implementation in Brazil we will be able to deploy an international version of the solution, not only for Portuguese speaking countries. The internationalization will require two actions: (1) extraction, transformation and loading guidelines (ETL) and (2) translation of the user interface and manuals. 

ETL guidelines will explain how to map available data to be consumed by the platform. We use available data on assisted population health assistance and on the determinants of health, such as age, gender, clean water access, nutrition, sanitation and solid waste disposal, access to education and job opportunities, housing needs.

We support two kinds of data sources: relational databases and local files, including commonly used formats such as comma separated values (CSV), hypertext (HTML), Json, XML and spreadsheets (ODF and XLS). 

On the translation of the software for other languages, we will begin with a Portuguese interface, and will initially translate to Spanish, French and English in order to support usage in Latin America and Sub Saharan Africa.

Our first goal is to raise funds to launch the software website and backend. 

After the software is launched we will monitor progress with the following performance indicators:

• Number of registered users

• Number of active users on the previous month

• Number of PHC teams with registered and active users

• Number of Cities with registered and active users

• Total population of assisted population analyzed

• Total number of warnings on pending healthcare actions and vulnerable persons in need of assistance visualized by health professionals

On the first year we will try to raise funds for the next steps, adding two new measures to the years that follow: number of active contractors and total funding raised.

Throughout the 32 years since the Unified Health System (SUS) was created in Brazil, Primary Health Care centers and their family health teams have shown to be strong agents of change to improve the health situation in the country. With the right combination of user friendly access to information and continuous, on the job education they can do much more, with better quality of access and care.

When we started working in Primary Health Care by the 2000's, all records were paper based. That was positive at that time, because papers, particularly health records were kept at the Health Facility and if we needed information it was at hand - literally. Our team, like many others, developed tools in order to monitor and improve our performance, like analyzing the most frequent reasons for patient visits and stratifying health vulnerability.

Informatization, paradoxically, made data harder to access and analyze by the frontline professionals. In the past decade we heard frequent complaints from health professionals, our colleagues and students, that they didn´t have access to valuable information on their performance, information that could be extracted from the data they enter into the information system everyday. 

We performed  fieldwork research on several cities (Brasília, Aparecida de Goiânia, Irecê, Esmeraldas and Ouro Preto). The general situation was the same.

The municipal health authority is not even aware that useful information is routinely collected by family health teams, usually he only knows the PHC health budget, how much of its costs are covered by financial incentives by the Ministry of Health (MoH) and how many facilities and employees there are.

Every city has one PHC coordinator, typically a female nurse practitioner with a PHC frontline background. She is usually aware that data is collected by the health professionals and digitally sent to MoH. She hears the complaints from health professionals that there are many forms to be filled daily, and she also complains that the dashboards offered by MoH are not informative - only the general numbers of people registered and assisted are shown. So she runs a few parallel information systems, using spreadsheet software on her personal computer, with information collected by phone calls to facilities, or typed from printed reports or paper based records.

The frontline health workers, all of them: physicians, nurses, dentists and community health workers have the same complaints. They have to fill very often the same information twice or thrice in different information systems: one for the MoH, another for the PHC coordinator or local health surveillance authority, and some keep a separate record for themselves. But they are never able to see the feedback on the information they provide, and they are unable to keep their records updated.

Part of our theory is that this situation can be changed if data is only entered once, and from that record it is transformed into on hand, useful information for the multiple local workers and managers. 

Then the PHC coordinator and the family health teams will be able to identify people with vulnerabilities that can be addressed, as well as track the adherence of pregnant women and people with hypertension and diabetes to their care plan. And that will lead to a performance change, where they will know beforehand if someone in their community is being left behind, so that they can go after the person in need before it is too late.

We developed this solution and showed it to them in three cities (Irecê, Esmeraldas and Ouro Preto). The answer we got were sparkly eyes, as they couldn't believe it was possible, and that they were relieved even to imagine how much it would make their jobs easier and more productive.

A demo of the solution shown in Ouro Preto can be seen here: https://ouropreto.painelsaude.info . User: admin Password: painelfiocruz22

Several technologies were combined in order to build a tool that is easy to install, read and process data from the PHC facility or Municipal Health Department. 

Considering the diverse scenarios of information technology infrastructure available for the PHC teams and coordination, we created a software that installs and runs a data science environment on the local computer. The solution can read the data available on the PHC facility or City´s health department, apply data analysis methods and then present the aggregated data in interactive reports. In order to do so, the solution was developed in three layers: (1) the Dashboard Installer and the PHC Dashboard itself, which in turn was divided in two additional layers, the (2) back-end and (3) the front-end. 

The installer mission is preparing and installing all the minimum technological resources to be able to run a web server locally, in order to view the reports, connect to the local database, read the data and also create an environment for data  processing and analysis. It was built using C++, HTML, CSS and Javascript.

The backend was developed using Python, Pandas and Flask, using SSL and JSON Web Tokens to communicate with the front end. The Fronted, in its turn, uses React JS, Bootstrap, React Query e Apache eCharts.

Every workday hundreds of thousands of health workers are collecting data that is relevant for performance improvement in Primary Healthcare in Brazil. MoH data shows that the 52,000 physicians, 66,000 nurse practitioners, 97,000 nurses working on PHC report over 200 million consultations every year. Together with the 265,000 community health agents, these professionals report 300 million home visits to check living and health conditions of families in the assisted communities.

Our solution give this data back to them, transformed into the information they need to improve quality and performance.

The development model was carried out based on a participatory approach through field research and interviews with different health professionals in the different locations visited, taking into account rural and urban areas. 

Our team gender balance is 40 male / 60 female. Our team  is made up of people residing in various regions of the country, with different educational and professional backgrounds. The field research, information architecture, prototype design, user experience workflow was led by Renata David who played a fundamental role in the creation and performance of our product. She was the Product Owner of the solution. Although there is Team Lead, a main coordinator of the team, work is managed in a horizontal and equitable manner, where the different areas of knowledge and action dialogue with each other and make decisions together. Furthermore, we are a team that believes in diversity, equity and inclusion as a basis for the growth of society. Our plans are to incorporate into our team more people with different profiles, including black people, LGBTQIAP+, people with disabilities and young apprentices to continue the work done so far.

We expect usage by early adopters, estimated in 5% of the PHC Teams. Brazil has 47,786 Family Health Teams (FHT), distributed across the 5,570 cities. Considering one physician and one nurse practitioner per team, and at least one PHC coordinator at city level, on a rough estimate we expect 5,000 active users of the total 100,000 target population. More than 268.000 CHW fully employed by the municipalities are a pillar of 98% of FHTs.

The software will be available at a dedicated Fiocruz website, hosted in our datacenter, with no additional infrastructure cost. Our budget will be invested in hiring the original team that developed the solution, who will be responsible for dissemination of the solution in public health events and social media and for offering support to the end users.

Our fundraising strategy for the next year consists in applying to grants and challenges that can subside the costs of initial deployment, to draw attention of health managers and authorities in city, state and federal level that take interest in the solution. Paid services will include support for deploying the solution at city level for larger cities, analytics services and customization.

The project in the coming years will rely heavily on the adoption of the solution by financing users, whether we get direct support from the Ministry of Health or from specific municipalities in Brazil. 

CONASEMS, the National Council of Municipal Health Departments, is a privileged target for our fundraising, as much of the funding will come directly from municipalities served. Our success with this prestigious prize would send a strong and credible sign for their engagement with the project in a sustainable manner. 

We will opensource the solution and seek funding for offering support and developing novel functions using a mix of strategies not related to selling the code, such as professional services, adding value with bundled online softwares as services, seeking donations, crowdsourcing, grants and stipends. 

The solution was conceived as an open source solution. The reason for that comes from the field experience and field research: health professionals are already overloaded with tasks of inputting information on different systems. Another software, as useful as it may be, will be firstly received as another burden. 

We don't mean to compete with established local or national electronic health records producers, big or small, public or private. We seek cooperation, our solution can be integrated with existing software for a better user experience in the frontline. 

Thus licensing the software under MIT License will allow us to support organizations willing to integrate the solution on their local information systems and even commercial solutions.

We hope that by winning this challenge we will drive enough attention to the solution to raise more funds for the next steps.

We went so far as finishing the pilot testing of the solution. Our next step is to raise funds for the deployment. We seek to raise 200,000 USD by the following year, id est up to July 2023. It will be enough to deploy the solution and support five thousand (5,000) active users for a year. 

Infrastructure costs and part-time dedication of the 2 Fiocruz technology researchers is already covered , the funds raised will be used to pay the original team involved in the development to support the deployment. 

The Brazilian public health IT market is big enough to sustain our acceleration phase after initial deployment. We will offer the following services: (1) professional services, supporting local health authorities usage of the solution with training, technical support (2) development of novel online dashboards with custom and complementary analysis using data collected by the solutions running locally (4) applied research on local primary health situational analysis for cities and state authorities. Additionally we will seek additional funds for development of novel functions and adaptation to different scenarios, such as donations, crowdsourcing, grants and stipends. Acceleration phase fundraising goal is set to 1,000,000 by early 2024. That is enough to support the solution for more than two years and launch the internationalized version. At this stage we expect to have enough paying users to keep the support of the solution as long as it is useful to PHC managers and workers.

The first evidence is the investment we raised for the first phase, a USD 300,000 grant from the Ministry of Health. However, we keep fundraising and have built a reputation of delivering results on both fronts, applied research and deployment of digital health solutions.

Our plan is to build on different competencies of the Labs involved in the project. 

CIDACS has a record of success in raising funds for applied research, leading to public health solutions that are used to guide national public health authorities and published in respected peer reviewed journals. Three examples are the Hundred Million Brazilian Cohort (https://doi-org.ezproxyberklee.flo.org/10.1093/ije/dyab213 ), the Congenital Zika Cohort (https://www.nejm.org/doi/full/10.1056/NEJMoa2101195 ) and the COVID-19 Vaccine effectiveness evaluation (https://vigivac3.fiocruz.br/artigos.php) .

Fiocruz national programmes regularly deliver digital health solutions to Brazil nationwide, and to PHC in particular. One example successful led by our team is the deployment of the Open University of the National Health System (UNA-SUS) portal, health education hub of MoH-endorsed online and distance learning courses that covered all 5,570 brazilian cities, 1,9 million health professionals with more than 400 courses and 22 thousand open educational resources. 

The Fiocruz Virtual Campus is one of the nodes of UNA-SUS. All Fiocruz online courses are offered by the Virtual Campus. So far 1,003 courses have been offered, for 428,768 health professionals. The Zika and COVID-19 courses were able to reach over 4,400 Brazilian cities in less than 2 months.