COVID-19: Modeling and Moving Forward

May 17, 2020

by Kim Furlow, Institute for Public Health; Elvin Geng, MD, MPH, director, Center for Dissemination and Implementation; and Karen Joynt Maddox, MD, MPH, co-director, Center for Health Economics and Policy

Members of the Institute for Public Health team are using mathematical and epidemiological modeling to help inform local and state government and health authorities’ responses to COVID-19 (Coronavirus). Dr. Elvin Geng, director of the Center for Dissemination and Implementation and Dr. Karen Joynt Maddox, co-director of the Center for Health Economics and Policy, recently highlighted how a modeling tool developed by Dr. Geng, along with collaborators in San Francisco and Massachusetts, can use local data to examine the trajectory of the virus over time under various scenarios that account for local demographics, social distancing policies and other inputs. This critical information can provide authorities with vital information about how state, regional and local areas might most effectively traverse COVID-19 now and in the future. According to Institute leadership, the virus will be here in the future.

We are going to be living with COVID for at least 12-18 months. We need to find a way to create a ‘new normal’ between the extremes of doing nothing about the virus (which will lead to many deaths) and shutting down for over a year (which would be an impossible ask on our economy and our community). We do need to find the middle ground but recognize that it will shift as we learn what is most effective and least dangerous. —Bill Powderly, MD, Larry J. Shapiro Director, Institute for Public Health

The modeling tool Dr. Geng and team developed is called LEMMA, and works like this:

  • A standard compartmental model that includes SEIR (susceptible, exposed, infected, removed and additional compartments representing hospitalization) epidemiologic model is developed.
  • The model accepts inputs on the transmission parameters that users deem most credible (e.g., length of stay, proportion of patients needing intensive care). The users of the model can vary these to make them most closely match local conditions.
  • Local interventions like the date of implementation of stay-at-home orders or other policies, can also be added to the model.
  • Local data, such as population size and the number of patients hospitalized at any given time are entered, and the model tests projections against the actual data.
  • The model can make projections forward in time, to predict what might happen in terms of cases and hospitalizations in the coming months under a number of user-specified scenarios.

“The tool helps us make projections that are transparent and reflect different scenarios about our policies in order to facilitate discussion among stakeholders at the local and state levels,” says Dr. Geng.

Read Dr. Geng's blog on this important modeling tool

Dr. Geng explains that the idea is to reduce the “reproductive number” or the number of secondary infections that can come from one infected person being in close proximity to one or more non-infected individuals. For example, if each person with COVID-19 gives the infection to three other people, on average, the reproductive number is three. Stay-at-home orders and other public health interventions work to reduce the reproductive number by limiting the number of people with whom each infected person might come into contact. These and other social distancing methods have been recommended by local health officials within the past month.

Dr. Geng points out that when an individual is infected, COVID-19 follows a certain timeline (below) with clinical illness lasting between seven to nine days before symptoms resolve:

For the models to impact public health, they have to be used to impact decision-making. Geng is working with physicians at the BJC Center for Clinical Excellence to compare and contrast projected information from his and various models. Collaborative teams from across the health system as well as from other local health systems look collectively at all projections, and work to implement public health solutions in part, based on these predictions. He is also collaborating with IPH Faculty Scholars, Dr. Joynt Maddox co-director and Dr. Abigail Barker, faculty lead for data and methods, both from the Center for Health Economics and Policy and others to help triangulate local and statewide information. Additionally, Dr. Barker and team have created an interactive data map simulating the likely spread of COVID-19 in counties across Missouri over the next six months, in the presence of high, moderate and low levels of social distancing. These simulations can help policymakers decide the best course of action going forward.

The Data Center at the Institute for Public Health and the Institute for Informatics are working to identify data analytics needs across the city, county and state; collate the data; and communicate results.

The fact that multiple centers from the Institute for Public Health are collaborating with each other and with other WashU and community partners, during this ever-changing period in time is a key part of the Institute’s mission.

“The guiding principle of the Institute for Public Health is to harness transdisciplinary research and service interventions that address significant community and population health problems,” says Dr. Powderly. “There is nothing more pressing in the 12-year history of the Institute than the current COVID-19 crisis. I am tremendously proud that we could bring our strengths in data science, implementation science and community partnerships together to provide approaches for our region to adopt in addressing this crisis.”

Although COVID-19 data changes daily, which then informs the data used in models used by Geng’s team, as of this writing, there are outcomes we can see while moving forward in this “new normal.”

“We can put some of the models into a broader public health context,” Geng adds.

What we know about cases in the state and region:  

  • Louis had a steeper climb in COVID-19-related hospitalizations than Kansas City, which is experiencing a decline in positive cases and in hospitalizations while St. Louis continues to slowly climb.
  • While overall, regional trajectories in hospitalizations are flat across the state, the majority of hospitals seeing the highest number of cases of COVID-19 are in St. Louis.
  • There are a disproportionate number of COVID-19 cases in north St. Louis city and county compared with other areas of the St. Louis region.

Even given these trends, there is mounting political pressure across the state to reopen the economy, but as Dr. Joynt Maddox says, “It goes much deeper than just the street protests.” There are health-related downsides to isolation: more people are staying home and not visiting the hospital for fear of getting the virus. Data from across the U.S. and other countries shows that people are avoiding coming to the hospital for serious conditions like heart attacks and strokes.

Beyond its direct effects on health, Joynt Maddox also reaffirms that the economic impact on health from this virus could be staggering. Being poor has a wide range of negative health effects overall and when we add the current epidemic to the picture, there is an even more dramatic impact on the under-served. As more people lose jobs, and as a result, lose health insurance and access to care, the long-term effect on health could be even bigger than its short-term impacts.

Where do we go from here?

With COVID-19 still prevalent, Joynt Maddox says a compromise in how we move forward may be the best case scenario.

As state and local officials think through their next moves, Joynt Maddox offers a few considerations:

One size does not fit all: Many rural areas appear to be spared for now, mainly because of population density, or as Joynt Maddox calls it, “a natural social distancing.” As a result, there is less of a chance that those “reproductive numbers” will rise. However, she says, “If a rural area is hit with a surge in COVID-19 cases, such as recently happened in rural Georgia, it can mean catastrophe.

These areas have a short supply of nephrologists, cardiologists and other physicians needed to take care of critically ill COVID-19 patients, so a local pandemic could be extremely difficult to manage.”

Don’t make any sudden movements: “Re-opening requires a successful drop in cases, capacity for testing and treatment and constant re-evaluation,” says Joynt Maddox. “Humility is required as well. There is no guarantee that what we do will have the effect that we hoped. As changes occur, we need to rethink strategy and rework these models accordingly in both the short and long term.”

Equity Matters: “We need more investment in resources to treat our vulnerable populations,” Joynt Maddox adds. “There are a disproportionate number of people of color and those living in poverty affected by this outbreak. We are seeing a lack of access to testing, the economic burden of hospitalizations and deaths is large, and it is incumbent on us to insure that we focus on equity in our response to this pandemic.”

Other recommendations include:

Keeping specific focus on populations with high infection growth rates such as the institutionalized, incarcerated and other communal dwellings (aging care facilities). High rates of infection have been reported in these facilities across the state.

Testing health care workers, separation of infected and non-infected hospitalized patients and fully using Telehealth (online doctor/patient visits) through the fall months are some additional recommendations, stemming from modeling outcomes.

Overall, Institute Faculty Scholars such as Drs. Geng and Joynt Maddox who are working with and contributing to these models agree: in St. Louis, targeted testing, quarantining and treatment along with “an adequate social safety net” are needed to drive the infection curve downward and to make it feasible for people to seek testing and treatment, who may not otherwise do so during this time.

The bottom line? “It’s not just going to end in May,” Joynt Maddox points out.

Models show that COVID-19 will not be eradicated anytime in the coming months and planning for the longer term is highly recommended. The Institute for Public Health and its centers will continue to work collaboratively with WashU and community partners to adjust models and make recommendations aimed at informing local decision-making toward the ultimate goal of more positive, healthy days ahead.