Open-source EMR: A New Model for Evidence-based Health Care in Haiti

When the first patients trickled into Hôpital Universitaire de Mirebalais (University Hospital) at the end of March, they triggered a cascade of seemingly banal computer tasks that represents one of the largest undertakings in the history of open-source electronic medical records (EMRs).

In the U.S. and other developed countries, hospitals invest millions of dollars on proprietary medical record systems that gather enormous volumes of data. Such hefty investments aren’t an option in the countries where Partners In Health works, and rarely is the necessary infrastructure in place. In Haiti and other PIH sites, clinics and hospitals have often relied on paper records, which can be difficult to manage and easily destroyed in a fire or flood.

In recent years, however, a global community of doctors, software developers, academics, and tech enthusiasts has come together with a single focus: Building and deploying EMRs that are open source, meaning anyone can use the application for free and modify the code to meet their needs. No patents, no licensing fees, just a collaborative effort. PIH was an early adopter. In 2004, we collaborated with the Regenstrief Institute to create OpenMRS, a formal community of individuals and organizations that contribute their coding expertise to a single open-source EMR platform. Nowadays, software developed under the OpenMRS banner is used in more than 40 countries. 

Our goals were simple yet audacious, given the status quo, Evan Waters, director of PIH’s medical informatics team, said.

But building an open-source EMR for University Hospital—a 300-bed teaching facility in central Haiti with seven different points of entry, a digital radiology suite, and the capacity to serve 500 outpatients per day—would be a “feat of epic proportions,” as Renee Orser, business analyst for PIH’s medical informatics team, said. It had to be highly intuitive for a staff with a wide range of computer literacy. Furthermore, as anyone from PIH’s Monitoring, Evaluation and Quality teams will tell you, collecting data is only useful if they can be extracted, analyzed, and put to work. 

Most implementations of EMRs in low-resource countries have focused on retrospective data collection for specific diseases—meaning the information would be collected on paper and then plugged into a computer. While these systems are helpful for evaluating programs, they don’t immediately improve individual patient care. For University Hospital, the team set out to build a true point-of-care EMR that not only improves data quality, but allows health care workers to be more efficient, and delivers real-time feedback. 

“Having this system in place has made a difference—it is showing people we are doing things differently here, in a more efficient way,” said Dr. Gregory Jerome, director of Monitoring, Evaluation and Quality at Zanmi Lasante, PIH’s Haitian sister organization.

Meeting of the Minds

Before a single line of code was written, the PIH informatics team and developers from Thoughtworks—a private software company that provided hundreds of hours of technical support—embarked on an extensive fact-finding mission. By interviewing Haitian nurses, doctors, and archivists who’d be using the EMR, analysts learned what would work and what wouldn’t. For instance, they learned that visual cues and icon-driven navigation worked better than text-heavy instructions. Little revelations such as these guided every step of the development. 

“Our goals were simple yet audacious, given the status quo,” Evan Waters, director of PIH’s medical informatics team, said. “We aimed to deploy a system that would, from day one, be capable of registering every patient in real-time, provide an easy way to look up the patient’s record as they moved about the hospital, and capture vitals, diagnosis, radiology, and surgery information across the entire hospital.”

To ensure the hospital had the necessary hardware for such a system, leading tech firm Hewlett Packard (HP) donated more than $1.5 million of equipment and flexible spending. “Clinical and clerical staff will be able to access this EMR quickly and reliably on 175 computer terminals located throughout the hospital thanks to HP’s donation,” Waters said. “This opens the possibility of collecting cross-sectional data that will allow PIH to improve individual patient care, better evaluate programs at the hospital, and improve reporting to donors and government partners in Haiti. It also allows clinicians to have access to X-rays from every computer terminal.”

Between August 2012 and March 2013, developers worked relentlessly to build an initial system that was robust and resilient. As tens of thousands of lines of code piled up, a sophisticated yet user-friendly EMR emerged. 

Six weeks before University Hospital opened its outpatient clinic, some of the biggest brains of the OpenMRS community hunkered down in PIH’s Boston office for a marathon of meetings. 

At one end of the table sat Darius Jazayeri of PIH and Paul Biondich of the Regenstrief Institute, among the earliest OpenMRS pioneers. At the other end sat Hamish Fraser, assistant professor of medicine at Harvard Medical School and founder of PIH’s informatics team. Laced in between were a dozen PIHers; a gang of Brazilian coders sent by ThoughtWorks; Pete Szolovits, head of the clinical decision-making group within MIT’s Computer Science and Artificial Intelligence Laboratory; Megan McGuire, an epidemiologist and health informatics coordinator at Médecins Sans Frontières; and Jonathan Teich, a physician from Brigham and Women’s Hospital and chief medical informatics officer at Elsevier. If there’s ever an OpenMRS Hall of Fame, it’s a safe bet that half this table will end up in the inaugural class. 

Over two days and gallons of coffee, these experts drilled down into the broader implications of the project and looked for ways to make University Hospital’s EMR a foundation for the future of OpenMRS. Everyone was acutely aware that a successful launch could provide a new framework for hospitals around the world working with constrained resources. 

Patients see the system with the plastic cards and bar codes, and they understand that we are finding a way to get better information that can help us improve the level of care, Jerome, the Haitian MEQ director, said.

At the end of the two days, Waters and his team displayed the latest iteration of the EMR and showed everyone exactly what the archivists, nurses, and doctors in Haiti would see on their computer screens. With each step—from capturing the patient’s vitals to ordering an X-ray to searching diagnostic codes—the intuitiveness of the interface never diminished. A collective head nod from those at the table approved of security features that protect patient privacy and similar safeguards. 

Before the presentation finished, a small machine in the back of the room spit out a plastic card with a unique identification code. Every patient who enters University Hospital gets one of these IDs, a small but durable manifestation of the EMR that links the individual to the hospital. 

Awash in Data 

By the end of June, more than 12,000 unique patients have been logged into the EMR and given an identification card. When a patient returns for care, the barcode on the ID is scanned, giving doctors and nurses near-instant access to the patient’s recent medical history. 

“Patients see the system with the plastic cards and bar codes, and they understand that we are finding a way to get better information that can help us improve the level of care,” Jerome, the Haitian MEQ director, said.

In its infancy, the EMR has helped Jerome’s team understand the ways patients are moving through the hospital, and why. Staff can monitor how crowded a certain department is and track patients as they’re transferred and discharged. “The EMR has given us the information we need to look at patient waiting times so that we can think about how to improve patient flow and keep satisfaction high,” Jerome said.

By utilizing routinely collected data, University Hospital is positioning itself to be a model for evidence-based health care in Haiti.

The more data the EMR collects, the better informed our decisions and programs will be. One long-term project is to map broad health trends in Haiti’s Central Plateau. Because we record each patient’s weight, height, and address, we’ll eventually be able to map body mass indexes on a village-by-village basis. Such maps can help identify nutritional deficits in certain areas, allowing us to go to patients most in need. 

“As we look to the future, we can take University Hospital as an example of an EMR system that is efficient, useful, and usable,” Jerome said. “The EMR gives us the chance to better manage information about our patients and better understand the profile of the diseases we’re treating. In the long run, this will help improve our decision-making across the whole network of Zanmi Lasante health facilities.”

Clinicians, nurses, and informaticians now meet regularly to review metrics gleaned from the EMR. By utilizing routinely collected data, University Hospital is positioning itself to be a model for evidence-based health care in Haiti. 

Generous support of OpenMRS by IDRC and the Rockefeller Foundation made this work at University Hospital possible.