Epidemiology team

Monday, May 1, 2017

Communication is key to informing policy

While the end goal of disease modeling is to influence policy to more effectively control and eliminate disease, there is often a tension between the technical concerns of research and the practical concerns of policy. At the 5th Annual Disease Modeling Symposium, several experts took part in the panel “From data to decisions: The role of modeling in optimizing research allocation in the fight against disease.” They discussed many different strategies to reduce that tension, but the recurring theme throughout was the importance of improving communication between modelers and policymakers.

Policymakers usually have a lot of simple incidence and prevalence data on hand, but they often make big decisions every week using spreadsheets without even simple data analysis, let alone the complex models disease researchers use. Panelists emphasized the importance of first showing policymakers what can be gained by using a model. They must see the value in modeling before they see the model predictions—it can be especially difficult to see the impact of the model if it affirms their initial intuition. “Try to make the black box of modeling a bit gray,” said Dr. Emilie Pothin, statistical modeler in epidemiology at Swiss TPH/CHAI.

However, the goal is not to turn policymakers into modelers or to share every aspect of your model with them. “The policymakers don't actually want your model," said Dr. Rick Steketee Director of the Malaria Control and Elimination Partnership in Africa (MACEPA) at PATH. They, nor their staff, have time to evaluate the model. Instead, first work with them to identify their questions and needs and translate those into goals you can meet with your model.

Panelists agreed that one of the biggest problems when presenting model predictions is how differently each group conveys and understands uncertainty. Scientists are adept at measuring and interpreting uncertainty because the field places a high value on precision. However, policymakers live in a world of quick decisions, often interpreting a researcher’s focus on uncertainty as evidence that their model has no value at all. Many recommended identifying and working closely with an ally in the health ministry or program who has a better understanding of technical details and uncertainty. "It's difficult if we don't have a translation layer," said Dr. Pothin.

When policymakers will not share their data with modelers, this ally can also be instrumental in understanding the data and connecting the dots to set effective policy. "[We must] increase the capacity of people that have access to that data,” said Dr. Paulin Basinga, deputy director of Integrated Delivery’s Country Primary Health Care Initiative.

When presenting model predictions, the key is to convey uncertainty in a format that intuitively conveys that uncertainty, said Dr. Steketee. For example, use a map overlaid with different colors to represent the model output. “People appreciate visual uncertainty more than spreadsheet uncertainty.” In addition, sometimes a simple, anecdotal summary of the output of the model is the most impactful. Dr. Basinga also recommended best case and worst case scenario projections, particularly when they are tied to dollar amounts.

This brings us to the other big gap that must be bridged—policymakers must work within strict time and budgetary constraints to which modelers are less attuned. Ministers of health must constantly justify the cost of their policies, and generally must operate on the same or less money as allocated in the prior year. One successful strategy used in Rwanda, said Dr. Basinga, was to conduct a bottleneck analysis to identify what changes could be made, without increasing budget, to improve health outcomes the most.

Due to these constraints, thinking is shorter-term than in the research community. "There is a scientific calendar and a political calendar. And politics can't wait," said Dr. Bernardo Hernandez Prado, Associate Professor at the Institute for Health Metrics and Evaluation (IHME) at the University of Washington. In contrast, science is slow, more focused on precision than speed. Likening global health to the central nervous system, Dr. Guillaume Chabot-Couture, Senior Research Manager at IDM, says modelers must get better at using their reflexes. You must sometimes drop your long-term research goals to reflexively address policymakers’ short-term needs.