By Eric Nelson, MD, PhD, University of Florida College of Medicine and College of Public Health and Health Professions and Adam C. Levine, MD, MPH, Brown University Alpert Medical School
Dr. Nelson and Dr. Levine are members of the Diarrhea Innovations Group, housed at PATH.
Apple computer company was founded on three principles:
- Empathy: “We will truly understand [the customers’] needs better than any other company”
- Focus: “In order to do a good job of those things that we decide to do, we must eliminate all of the unimportant opportunities”
- Impute: “People judge a book by its cover. People form an opinion about a company or product based on the signals it conveys. It must convey excellence.”
These principles are rarely extended to global health—where “field appropriate” approaches often become what Paul Farmer characterizes as “crap for poor people.”
Our team has embraced Apple’s principles to explore how we might impute excellence when it comes to last-mile delivery of care for the poor in remote regions. Our mission is to combine medical, technological, and scientific skill sets to discover new approaches to combat infectious disease outbreaks. We do this by listening to both patients and their health care providers in the hope that conversation uncovers unanticipated problems and reveals solutions inspired by the end-user.
Inspired solutions through dialogue
A health care provider at icddr,b tests the Rehydration Calculator developed by our team to assess a child’s level of dehydration. Photo: Brown University Alpert Medical School/Adam Levine.
As a model system for intervention, we chose cholera outbreaks in Bangladesh and Haiti for several reasons: (i) proof of principle against a pathogen like Vibrio cholerae that kills swiftly and inflicts disproportionate harm on the extreme poor can have broad impact; (ii) cholera can be clinically diagnosed and prompt administration of low-cost oral rehydration solution or intravenous fluids as appropriate can drop deaths from 20 percent to less than 1 percent; and (iii) despite limited historical success, new tools and opportunities are coming together that inspire optimism for effective and rapid outbreak control in the future.
With these principles and strategy, we began building a dynamic and comprehensive team that included the Bangladesh Ministry of Health, rural government doctors and patients in Bangladesh, Silicon Valley and Hyderabad technology developers, mapping experts, and academic clinicians and scientists.
After a period of conversation with stakeholders in Bangladesh and Haiti, we identified two main problems with outbreak response. The first was that clinicians were not equipped with decision-support tools to make the right diagnosis and, therefore, treatment plans and clinical reports for epidemiologists were unreliable. The second was that paper-based data collection was unstructured and delayed. With so many data points, it was difficult to determine the most important data and make appropriate clinical decisions. In addition, smartphone survey tools required connectivity that did not exist in outbreak areas, and commercial survey products did not readily allow for team-based data entry.
Learning from failure
Over three years, we have systematically been addressing these challenges while having copious, but instructive, failures.
These failures have all stemmed from an under-appreciation of the fact that the ultimate sophistication is simplicity, another Apple priority. We began by making a decision-support tool that integrated with a data-collection tool and worked like an electronic medical record system for both doctors and epidemiologists. Although the tool failed pre-tests and live-tests, it took a year to acknowledge and stop the project. We pivoted to make a very simple two-page, decision-support calculator (known as Rehydration Calculator) for clinicians and then a separate lean data-collection platform for epidemiologists (the Outbreak Responder platform).
Smartphone decision-support tool: (A) Home page with the Rehydration Calculator for decision support (upper) and the data-collection platform (lower). (B) Data input page. (C) WHO-derived recommendations for medication and fluid administration; units can be changed by side-swiping. “i” represents a link to educational summaries and primary publications. (D) Outbreak Responder data-collection and aggregation platform for the research team (also known as ‘Response Team’). ‘QI’ is a portal for customized quality improvement questions intended to be used by researchers only. Source: PLOS Neglected Tropical Diseases.
We tried to incorporate advanced Google maps techniques to map the dyad of treatment and residence locations. But we failed to appreciate that many communities do not conceive of their space in the two dimensions of a flat map. This led to a simpler solution that seems to work broadly; communities know their geographic hierarchy, such as United States/ California/ Santa Clara County/ Palo Alto. This discovery helped us to shift complexity (e.g., Google maps) away from the users on devices to the back end by providing a simple drop-down menu that queries their geographical hierarchy.
We began to learn what was beyond the scope of our core team’s mission. It became clear that we were developing a vehicle for information and that we were not the ones to populate the vehicle. There was a gap in high-quality research into the performance of diagnostic models and management strategies. Even the World Health Organization (WHO) guidelines for assessing dehydration in patients with diarrheal disease lacked sufficient evidence in accuracy and reliability.
While we may be able to improve adherence to guidelines with our calculator, the improvement is useless if the guidelines have not been rigorously evaluated and shown to work effectively in clinical practice. In response, the ‘we’ in the core team was expanded by establishing a partnership with Dr. Levine at the Brown University Alpert Medical School. Partnering with researchers at icddr,b in Bangladesh, Dr. Levine has begun developing new evidence-based assessment guidelines that may have greater accuracy for assessing dehydration in children with acute diarrhea than current standards.
Eye on customer, focus, and excellence
To impute excellence also requires validation. We view health technologies at times like medications. They may have indications, contra-indications, and unintended consequences. Dr. Nelson conducted a pre-intervention and intervention clinical study on the impact of the Rehydration Calculator at two hospitals in northern Bangladesh during the fall cholera outbreak season. The calculator led to the normalization of dehydration assessment, the correct antibiotic usage, and a decrease in IV fluid usage by approximately half.
These results were promising but the study lacked generalizability and independent control. A large cluster randomized control trial (RCT) is being started to address these limitations. Outbreak Responder was deployed in the pilot and will now be used in the cluster RCT to collect and aggregate nearly one million data points. In parallel, Dr. Levine has begun comparing the utility of paper versus smartphone adaptation of WHO and newer methods for assessing dehydration in children and adults; one goal of this project is to populate the Rehydration Calculator with these newer evidence-based methods.
This journey has had unexpected turns, challenges, and successes. We hope it shows that a mindful eye on customer, focus, and imputing excellence yields important gains. We are still mid-journey and plan to deploy both the Rehydration Calculator with updated assessment algorithms and the Outbreak Responder in Haiti. We view these technologies as additional tools in a larger toolkit that we hope will prove useful in the fight to eliminate cholera from Haiti. We also hope that this story has ripple effects beyond the infectious disease space.
Please contact us with ideas and questions.