Nurses screening patients for signs of Ebola at Redemption hospital.
Early on the morning of January 13, 2015, I walked with two twenty-something Liberian contact tracers through the chaotic dirt paths of their local community, New Kru Town, on the northwestern edge of Monrovia, Liberia's capital.
It was late in the Ebola outbreak, which began in March of 2014, and cases were dwindling. The contact tracers were tasked with checking temperatures and inquiring after the health of their neighbors who had been exposed to the virus. Increasingly, the daily rounds passed without incident.
On that particular morning, near the end of their rounds, we encountered a heavy-set, middle-aged woman holding back vomit and visibly sweating as she struggled to present herself to us. She feigned good health, though her symptoms met the definition of a suspect Ebola case. Against her wishes, we called the authorities, who brought her to the nearest treatment center.
The woman’s family lived in a communal house that was ravaged by Ebola. The house had more than five prior deaths—sadly, an all-too-common scenario after nine months of outbreak—and the specter of a sixth was probably more than this family could bear.
Thankfully, the woman tested negative. And on May 9, the World Health Organization declared the Ebola outbreak in Liberia over.
The reprieve lasted for a month and 20 days. And then suddenly, Ebola was back.
It’s amazing to look back at that moment. It was impressive that so much progress was made to combat the outbreak in a region ill-equipped to combat it. The health system was struggling to curb the virus’ spread, buckling under its own weight, crippled by the lack of protective equipment for personnel, no infectious disease training, and few, if any, supplies. What’s more amazing is that we were able to build a picture of the outbreak.
The fact that new cases appeared, however, suggests holes in the data. Did the new Ebola patients in Liberia contract the virus in neighboring countries, or were pockets of infections simply overlooked by the census, as tests have suggested?
More than 27,000 people have contracted Ebola and upwards of 11,000 have died between December 2013 and June 2015. Hundreds of millions of dollars have been spent and hundreds of people were sent to fight Ebola by the United States government alone. While the images, reporting, and statistics depict the stark realities on the ground, they paint an incomplete picture of the largest Ebola outbreak Earth has ever seen. What they omit is the underlying data crisis.
The key to stopping Ebola lies in knowing who’s infected and who they’ve been in contact with. In my experience, more than nine out of 10 suspected cases turn out to be negative for Ebola. On the other side, Ebola-positive patients can slip through the cracks, often because people are afraid to seek treatment. There are countless ways contact tracing can fail. The two Liberians who guided us through New Kru Town could have missed a house or a family member, or there could have been an error in the initial mapping of exposure.
In my experience, more than nine out of 10 suspected cases turn out to be negative for Ebola
The next time you fill up your gas tank, remember that the price per gallon is raw data—alone, it doesn’t mean much. However, take into account what you paid last week, and the price of gas at nearby stations versus the money in your wallet, and you’ll start to build knowledge: information you can act on. You’ll begin seeing patterns, such as whether you pay more on certain days and at specific locations, or if you’re burning through more gasoline than usual.
Finding meaning in disparate data points is easy when it comes to your fuel habits, but it’s several orders of magnitude more difficult with emerging global health threats like Ebola.
The basis for preventing the Ebola outbreak from spiraling out of control lies in being constantly aware of the evolving state of public health. Keeping tabs on that information is manageable with just a few cases or patients, but it quickly became a nightmare once Ebola started spreading at an exponential rate. Even for the most organized health systems, it would have posed a problem. Nigeria had only 20 cases of Ebola, but tracked nearly 1,000 people who had come in contact with those infected. Add that to an under-resourced health system’s lack of preparation, supplies, and staff, and you have a recipe for disaster.
Redemption Hospital in New Kru Town is a free national referral hospital in Liberia. The sprawling complex of concrete buildings is hidden down a dirt road, but it serves thousands of poor citizens who live in the slums surrounding it. Redemption was the bellwether for Ebola's entry into Monrovia and its rapid, uncontrollable spread. The virus overwhelmed the hospital and the entire country long before the international response was mounted.
Every day, hundreds of patients visit Redemption to get their children vaccinated, give birth, and get treated for HIV, as well as many other maladies. Each patient visit is a sentence in a personal and community story that’s buried in reams of medical records and ledgers, but these are stories with the potential to save lives.
Esther Kesselee, a nurse at Redemption, fell ill in late June 2014 before swiftly dying of Ebola. On July 2, 2014, Redemption’s head surgeon had succumbed to Ebola. By the time Liberia was declared Ebola-free, more than 12 Redemption hospital workers had died.
According to Eddie Nyankun, Head of Medical Records at the hospital, the first cases to reach the hospital were classified as “other”—not Ebola—on medical record forms and filed away. A single “other” entry tells the story of one person. Multiple “other” cases, however, build a story to be scrutinized and learned from. That is the crux of the challenge.
Paper records at Redemption Hospital.
I went to Liberia with my colleagues at Gobee Group, a global health consulting firm, to find creative ways to enhance the speed and efficiency of managing Ebola response information. We worked with the Ministry of Health, frontline health facilities, and international response partners to map the people, processes, and technology involved.
Without seeing a direct benefit of reporting the data, the staff wasn’t given much, if any, incentive to deliver quality data to the government
During the outbreak, no one at Redemption could use the data to build a picture of what was going on inside the hospital, let alone the surrounding areas it serves. In other words, Ebola’s spread couldn’t be mapped, which is what’s needed to stop it in its tracks. Since health workers couldn’t see the patterns of the outbreak, they were forced to fly blind, reacting to the fires that need to be put out rather than forming a proactive strategy to halt Ebola’s relentless advance. Not even those who collected the data had a good handle on what it meant on a larger scale.
The paper trail starts at Redemption hospital with a visual examination and interview with each patient to identify symptoms that may indicate Ebola infection. Those patients fortunate to pass screening without discernable signs of Ebola move slowly through the hospital from registration to the examination room. Samples may be sent to the laboratory, and X-rays ordered, before the final diagnosis is determined and the patient is referred to the pharmacy or elsewhere for treatment. At each stage, crucial data begins to materialize in the notes on the medical record and ledgers in each ward.
At the end of each month, a 30-page report with aggregate data concerning a hospital's entire situation, including Ebola-related statistics, is sent to Liberia’s Ministry of Health. “It includes everything from the number of people that came with HIV/AIDs, the number of people with HIV/AIDS that didn’t come back at the right time, the number of maternal mortalities, the number of vaccinations given, on and, on and on,” says Dr. Sharon McDonnell, associate professor at Dartmouth’s Geisel School of Medicine. McDonnell worked closely with the International Rescue Committee on frontline Ebola care, and helped rehabilitate Redemption.
Once the ministry received the report, it would plug the data into DHIS 2, a widely used open source national health management information system, which provides a country-wide, monthly look at the state of public health. However, this broad picture doesn’t help clinicians make decisions at the point of patient care.
“[These health management information systems] are great improvements and horrible tragedies at the same time, like most things, in so far as there is an attempt to collect lots of information at the big data level, which is not going to make any difference to anybody’s life in Redemption,” McDonnell explains. “Then the question is, ‘To what extent does that influence patient care?’ or ‘What happens?’ The ministry may know, but nobody below the ministry is really accessing and using that information.”
Tracking cases at the Liberian Ministry of Health and Social Welfare.
Without seeing a direct benefit of reporting the data, the staff wasn’t given much, if any, incentive to deliver quality data to the government. With 500 to 700 patients that went through Redemption’s doors each day, the information collected was nothing short of mountainous. Redemption, like many other hospitals and treatment centers, wound up with piles of paper and little actionable data to improve patient care and respond better to community health threats.
Many are tempted to solve this data crisis by throwing mobile and web technologies into the mix. Between on-the-go banking and high adoption rates, Africa’s affinity for mobile devices has become something of a trope, so it’s enticing to believe the obvious solution is the best solution.
But mobile technology introduces a host of complications and additional points of failure. There’s the need for cellular connectivity, which is often unreliable. Of course, the ongoing cost and effort of technical maintenance and support needs to be accounted for, which these medical facilities aren’t equipped to handle. When it comes to user-friendliness, devices have to be charged and software bugs will be encountered.
And, as I’ve experienced on the ground, handling mobile devices in addition to clipboards and stacks of paper can be cumbersome. In November 2014, the Paul G. Allen Family Foundation provided 10,000 smartphones to help data collection, but as of March 2015, fewer than 1,000 had been deployed according to officials with the United Nations Mission for the Ebola Emergency Response (UNMEER). Even with the right technology it is difficult for international and local organizations to build, test, deploy, and maintain valuable applications. Digitizing data is critical, but how that happens needs to best fit the situation.
“Everybody understands that there would be tremendous value in having the data digitally, and I think they drastically underestimate how jarring a change it would be operationally.”
The human factor also comes into play. People have established workflows and habits, which can be difficult to change. Even with mobile devices, health workers would still fill out paper forms because part of the medical system’s reporting procedures required it. What mobile devices did, in effect, was provide more work and another piece of equipment to juggle. Such devices also introduce the need for training personnel.
At one point, international technology experts deployed a mobile data solution—Magpi—to collect triage data using tablets at Redemption hospital, but the effort was quickly abandoned once the experts returned home.
“Everybody understands that there would be tremendous value in having the data digitally, and I think they drastically underestimate how jarring a change it would be operationally,” says Brian Busch, Head of Business Development at Captricity, a firm that digitizes data from paper. “You’re not just changing the data and the systems on the backend, but you’re changing the way that the people in the field have to do their jobs. Sometimes I think that’s often underestimated.”
The road to Redemption Hospital.
Redemption Hospital is more fortunate than most health facilities in Liberia. The hospital had four computers and software to manage patient registration, diagnosis, and treatment. The reality is these technologies seldom worked as intended. In my time at the hospital over the course of two trips and three months, the computers were either not operating or the software wasn’t functioning. We spent several weeks trying to locate the developer (a Ghanaian) of the system. After a short Skype call with the hospital’s IT team, the developers quickly discovered the problem was minor requiring a simple reinstall. This minor problem caused months of disruption to the medical records department. This, unfortunately, is typical of the challenges faced by even relatively well-equipped health facilities in fragile states all over the world.
What’s more, a totally new system is more likely to face issues when put under stress. Even without a new technology or process to deal with, Redemption’s national reporting efforts crumbled during the outbreak.
Ebola is resurfacing in Liberia, and the outbreak continues in neighboring Sierra Leone and Guinea. In order to end this epidemic, we need good data.
A better solution lies in empowering people with technology that suits them and their workflows. Ironically enough, a 2,000 year old technology—paper—may be the key for Redemption and hospitals like it to tame the data inherent in an outbreak.
With paper, there’s no need for cellular connectivity, charging, technical maintenance, or training. Better yet, it’s possible to convert the data on paper to a digital form. Captricity, for example, is a service that extracts data from scans of paper using machine learning. With such a solution, hospitals can scan paper forms and automatically convert the information within them to digital data, which can be immediately analyzed and shared with local and national organizations.
Using such a digitization system would save valuable staff time and reduce the chance for input errors since there’s no need for data entry by hand. In addition, having all that data on tap means that family and friends could be automatically texted lab results and condition updates about their loved ones—a big change compared to the radio silence that fed rumors of disappearances. Best of all, it would give health workers the ability to analyze how Ebola is spreading in the community, as well as at the national and regional levels.
The Paul G. Allen Family Foundation provided 10,000 smartphones to help data collection, but as of March, fewer than 1,000 had been deployed
“There’s a false dichotomy in thinking that just because people have access to technology now—where there’s a phone, there’s a tablet, there’s something they could get ahold of—then we necessarily should collect data from them using that device,” Busch said. “Really, the question is what’s the right medium to collect data in any given situation. And there are plenty of times where I think we’re going to find in the future where paper remains a great option, arguably the best option.”
Not only is paper a natural fit for the data collection process itself, but it’s often a better fit for people. It’s no secret that some of us still prefer scribbling notes onto pads. There may even be a cognitive benefit to sticking with papyrus’ descendent. Researchers have found that people process information better when they write things down on paper instead of tapping away on devices.
Of course, it’s not enough to bolt-on technology to existing paper forms. We completely rethought the content and structure of the paperwork in use in Liberia, redesigning them to maximize efficiency and efficacy.
The old version of the paper hospital form and the new version. Image: Mahad Ibrahim
Testing this approach yielded convincing results that converting paper to digital data is not only viable, but that it’s a better approach to massive data management challenges. We recruited two untrained transcribers to enter dummy patient data by hand onto the redesigned forms, which were then separately entered by two other untrained data entry subjects into Google Forms and scanned with a multi-feed scanner.
Not only is paper a natural fit for the data collection process itself, but it’s often a better fit for people
A total of 80 forms were completed in the test. Scanning the forms to have their data pulled out automatically was 38 times faster than data entry completed by hand. While it took 153 minutes for the team to manually input the data, scanning the forms took only four minutes. Although accuracy for number-based data was roughly the same for both processes, letter-based data was 21 percent more accurate when scanned. And that’s just the start. The forms could be optimized for even better results.
Combating Ebola is not a simple issue. A lack of resources has a catastrophic effect on a medical system’s ability to fight the virus. However, armed with the ability to quickly analyze data about the spread of the virus, even under-resourced medical operations have a better chance to curtail its transmission. And paper, as strange as it sounds, might just be the technology to empower those on the frontlines to fight back.
Mahad Ibrahim, Ph.D, is a founder and managing partner of Gobee Group, LLC., a global social innovation design consultancy. Additional contributors to this post were Osvaldo Gomez Martinez, Ph.D, Jaspal S. Sandhu, Ph.D, Alexis Santos, and Shannon M. Hamilton. All photos courtesy Mahad Ibrahim.
Modern Medicine is a series on Motherboard about how health care and medical technology can move forward so rapidly while still being stuck in the past. Follow along here.