Eric Perakslis, PhD is a Rubenstein Fellow at Duke University, Lecturer, Department of Biomedical Informatics at Harvard Medical School and Innovation Advisor to Médecins Sans Frontières. He has significant experience leading technology efforts in infectious disease outbreaks.
As the COVID-19 epidemic scales exponentially across the United States, calls for expended use of telehealth, innovative technology solutions and optimization of life-saving critical care hospital beds clearly highlight unmet needs in the American healthcare system.
Based on lessons from both recent Ebola Virus Disease (EVD) outbreaks in West Africa and the Democratic Republic of Congo (DRC), those of us who are experienced in outbreak response know that the difference between success and failure in responding to the current pandemic will depend equally on what is done and how it is done.
As a nation that prides itself on independence, innovation and ingenuity, the United States must understand that ill-considered heroics can cost lives and that a coordinated response is the best response. That is, if the measure of success is the number of lives saved.
Solutions must be conceived, built and deployed on the ground
One of the first rules of humanitarian and disaster response is that the boots on the ground (BOTG) must be in control. When it comes to technology delivery, this has multiple essential implications. First, the ultimate arbiter of requirements is the field team. The last thing patients or front-line responders need is programmers sitting at home writing code and arguing with health workers in the trenches about functions and features. It never works. Even when agreement is perceived via remote conversations, the reality on the ground may be different or may change instantly, negating previously agreed-upon specifications.
My own personal experience with these hard facts occurred toward the end of the West African EVD outbreak.
In May of 2015, as the case count was trending toward zero and our efforts turned to rebuilding local health systems that had been devastated early in the outbreak, I was writing apps that would enable the proper triage of a possible Ebola patient. These apps were somewhat complex algorithmically but had to be presented graphically to make this process as easy as taking a fast-food order.
This is not difficult—the apps are menu-driven and graphical. Workers simply input symptoms by selecting pictures and the menu walks them through the process. I spent several weeks building and testing the apps based on forms that had been emailed to me directly from the clinic.
When I arrived a week later, however, the people who had emailed me the material I used to develop the apps told me that that the forms were incorrect and they had never seen them before. Having anticipated this possibility, I spent the next 36 hours completely re-writing the apps and the project was highly successful.
My lesson? The time I spent coding apps remotely from emailed specs was wasted; I should have traveled earlier and built the apps on the ground. They would have been correct the first time and the project could have started at least two weeks earlier.
Use privacy-preserving technologies at the outset
The humanitarian response sector operates with a deep understanding that all interventions in crisis settings have corresponding risks to the immediate victims as well as to the responders. Key to mitigating these risks are ethical frameworks that protect all parties from immediate and longer-term consequences. As new procedures and technologies are quickly deployed against COVID-19, there is neither reason nor excuse to jeopardize patient privacy or expose healthcare workers and institutions to additional liability risk.
Because data sharing is essential to combating this pandemic, privacy-preserving technologies should be employed at the outset of implementing any technical solutions. For example, tokenization is a well-understood privacy-preserving technique for facilitating data sharing. A good start would be to automatically tokenize every COVID-19 test result, thereby enabling detailed data sharing across various response capabilities.
Importantly, digital health tools contain the inherent capacity to ensure ethical medical intervention. In light of this, any calls to weaken patient protections for the sake of technological priorities must be viewed both skeptically and critically.
Focus on consistent, automated and standardized data collection
Even in a public health emergency, consistent if not fully standardized data collection is a necessity, not a luxury.
The West African EVD outbreak that struck Guinea, Liberia, and Sierra Leone outpaced the ability of any one government to stop it. This necessitated that the World Health Organization (WHO) play a coordinating role — one that proved highly beneficial. Although the WHO’s response was not perfect, it nevertheless included the publication of a strategic plan that included communications strategies, training on personal protective equipment, case definitions and medical and epidemiological data collection and management standards.
Activities were coordinated across 60 specialized Ebola treatment units that were capable of providing approximately 3,000 beds for Ebola care in the three countries most affected by the outbreak. Further, more than 40 organizations and 58 foreign medical teams deployed an estimated 2,500 international personnel as well as thousands of local staff.
The United States is already at this scale of response for the COVID-19 pandemic, and we anticipate continued exponential growth. Given the magnitude of current and future challenges to healthcare and public health systems and resources, adopting a common approach to data collection and sharing is essential. Such a step need not be difficult: a simple digital questionnaire comprising 5-10 questions and employed during every telehealth session would afford substantial insights into the presentation, triage, treatment and follow-up of the disease.
In Sierra Leone we did this with inexpensive Android apps that ensured high-quality data collection and availability. The key to the success of this effort was that the coordinated response effort provided standard definitions, questionnaires and data management requirements that were employed with surprising efficacy and consistency across a decentralized multinational response.
If we standardize data collection via a simple triage app or case report form, people will use them, regardless of the format—especially if data collection can be done by nonclinical staff, thus allowing doctors and nurses to devote more of their precious time to patient care.
Make use of all “free” metadata and technology capabilities
Another essential lesson from the experience of responding to outbreaks in low-resource settings is to “use all parts of the animal.” For example: when we replaced or supplemented paper contract tracing with digital data collection, accuracy and reliability were improved thanks to the other “free” capabilities already available with the mobile devices. The global positioning system (GPS) capabilities of the cheap Android phones we used provided exact geolocation coordinates.
Video recording captured and documented complex consent discussions in multiple languages with village chieftains. Training videos could be reviewed on-demand and repeatedly by rapidly-trained workers who were rushing into complex and potentially dangerous situations.
As we spin up our response to the COVID-19 pandemic, we need to apply this type of thinking about the exploitation of native technology features and metadata to telehealth capabilities. Starting with the foundation of privacy-preserving tools and techniques, the IP addresses, duration, and timestamps of telehealth sessions could be used to establish a real-time dashboard of medical consults for every state, region, and town.
Overlaying tokenized COVID-19 test results could provide a view of disease incidence at a city-block level of detail that would improve the certainty of risk determination and treatment recommendations. In low-resource settings, which the United States is quickly becoming, taking a “waste not, want not” approach to technologies and metadata is essential.
Use pre-existing, purpose-built toolsets
Among the most painful lessons from the West African Ebola outbreak were the importance of time and the understanding that smaller interventions deployed earlier would have prevented major systemic stresses later. Many efforts to deliver technology solutions started from scratch and took too long to build and deploy. Amid the demands of the current pandemic, we don’t have the luxury of forgetting these lessons.
There are already specialized, fit-for-purpose toolsets available for infectious disease outbreaks. CommCare by Dimagi, for example, is an open-source Android platform that has COVID-19-specific contact tracing applications and other toolsets ready to deploy. All parties seeking to obtain or deliver technology solutions should consult experts and seek off-the-shelf solutions BEFORE anyone writes a single line of code.
Patients are waiting, and the “when” may be more important than the “how.” Or, in other words: smaller solutions delivered when needed beat grand solutions delivered after the need has passed.
The battle with the current pandemic is being fought in clinics, doctor’s offices, hospitals and via telehealth sessions as I write this, and there is no time to waste. The people on the front lines are our “boots on the ground.” Let’s get them every tool they need as quickly and effectively as we can.