State of IL COVID-19 Epi-

Modeling Task Force Update

Prepared by Civis Analytics

May 9, 2020
1
Table of Contents

● Executive Summary
● Modeling Update
○ Recap from last time
○ Updated state models and updated resource capacity estimates

● Initial Look at the Restore Illinois Health Regions

○ Model Forecasts by Superregion

● Recommendations for Safely Reopening

○ Sentinel Surveillance Testing

● Next Steps

● Appendix
2
Executive Summary (1/3)
Overall, changes since last update

● In addition to Statewide, now a deeper look at the 4 Project Restore regions. Northwestern was able to
produce model results for each of the 11 EMS regions; UIUC produced a model for each of the
four Restore Illinois regions; UChicago produced 3 regional models (North-central, Northeast, and
Central + Southern). All 3 of the groups’ models were combined to provide separate views of the
Restore Illinois regions.

● Additional data. Models are fit on approximately 2.5 additional weeks of data, and are making use
of additional hospital census (ICU bed occupancy, hospital bed occupancy) data from EMResource.
UChicago is also using data from a Chicago-area hospital system to inform some parameter
choices.

CONFIDENTIAL 3
Executive Summary (2/3)
Updated State Models - Summary of latest results and conclusions

● Compared to the previous model forecasts, the updated forecasts are somewhat more pessimistic
(peak is later) and also cover a somewhat wider range of possibilities. Possible explanations for
this include that we’re seeing a higher transmission rate in the last 2.5 weeks of data added to the
models. For one of the groups, it’s also possible that incorporating length-of-stay data provided by
one of the hospital partners is a contributor to this as well.

● Assuming no changes in mitigation, hospital bed and ventilator resources from COVID-19
patients are still sufficient. Of course, this does not hold true if the state begins relaxing stay-at-
home orders and other current restrictions without putting other counter-mitigations in place.

● Post-mitigation reproduction rates (R0) remain near the tipping point of 1, not well below 1. If
mitigations are lifted without simultaneously implementing tight procedures to monitor, detect
and control outbreaks early, we can expect a resulting surge in cases and deaths.

CONFIDENTIAL 4
Executive Summary (3/3) North-Central
Northeast

Model Results for Each of the Restore Illinois Health Regions Central

Southern

● Looking at the current model forecasts, if mitigations were held in place for the 4 regions defined by Project
Restore, the epidemic would be under reasonable control. Lifting shelter in place on May 30th without other
mitigation to reduce transmission could lead to a second wave in each region, even possibly in areas where there
aren’t currently many deaths.

● The approach as described by Project Restore is phased and therefore falls somewhere in between an indefinite
continuation of current restrictions and a complete relaxation of mitigations. More work is needed to understand
the model trajectories as a result of the state’s phased approach to reopening. Our next update will include
insights from simulations of different ways that partial relaxation of mitigations might affect the epidemic."

● Understanding that reopening IL in the next few weeks may be non-negotiable, the public health and
epidemiologic experts who are part of this task force strongly recommend putting into place robust processes for
detecting signs of an outbreak, isolating cases and controlling widespread transmission early. A preliminary
introduction to these concepts is provided on slides 17-21.
CONFIDENTIAL 5
Modeling Update
Recap: Models shared publicly on 4/22/20

II. 2 scenarios
simulated: (1) Stay-at-
Home never
implemented; and (2)
Lifting Stay-at-Home
abruptly on 4/23/20

III. Predicted peak

hospital resource
needs compared
to resource
I. Statewide projections provided by UIUC and U Chicago; their capacity
Chicago/Cook County models were scaled by population size of state to
do this

CONFIDENTIAL 7
Updated forecasts of COVID-19 for the State of IL
Key Takeaways

● Compared to the
previous report, the
range of the peak is
wider (from now to
mid-June)

● This is later than

previously reported,
possibly because in the
additional 2.5 weeks of
data, we haven’t
observed the declines
in deaths and
hospitalizations that
were previously
forecasted.
CONFIDENTIAL 8
Models still predict that in most cases, the state will have adequate
hospital resources to meet need at the projected peak
This is under the assumption that current level of mitigations will remain in place

State Resource Capacity

(from EM Resource)

Utilization by Non-
COVID Patients

CONFIDENTIAL 9
Reopening Illinois
Model projections for each Restore Illinois Health Region

On 5/5/20, it was announced that as part

of the Restore Illinois plan, the following
EMS regions will be grouped into larger
“superregions”:

● North-Central IL: EMS Regions 1, 2

● Northeast IL: EMS Regions 7-11
● Central IL: EMS Regions 3, 6
● Southern IL: EMS Regions 4, 5

In the following pages, we’ve summarized

the available results corresponding to the
baseline scenario of what would occur if
current mitigations (including stay-at-home
orders) were held in place for each of these
superregions.
CONFIDENTIAL 11
Model Projections for North-Central IL
If current mitigations are held in place, UChicago predicts a gradual surge in cases peaking in early fall

At the peak of the

outbreak, UChicago
estimates that ICU bed
capacity will be met or
exceeded in all cases and
that non-ICU bed
capacity will be met in
the worst case scenario.

CONFIDENTIAL 12
Model Projections for Northeast IL (Chicago area)
Because most cases have occurred in Chicago, the state’s curves most resemble this region’s curves

CONFIDENTIAL 13
Model Projections for Central IL
Of the 4 superregions, Central has the most favorable projections for peak and resource capacity

Central region has had

very low activity to-date
and all 3 of the modeling
groups predict few
deaths as long as
transmission remains
low. Central region also
appears well-equipped
with hospital resources.

CONFIDENTIAL 14
Model Projections for Southern IL
Southern IL superregion includes the East St. Louis area, where high activity is being observed

CONFIDENTIAL 15
Sentinel Surveillance
for Safely Reopening
Illinois
Monitoring, Detecting and Controlling Regional Outbreaks
Testing in sentinel surveillance population is absolutely critical to safely reopening the state

● As Illinois reopens, it’s absolutely critical to prevent new outbreaks of COVID-19 in reopened
areas. All areas remain vulnerable, since all populations are far from the herd immunity
threshold. We estimate approximately 7-8% of the population of Illinois has gained immunity
through infection, but 60-80% need immunity to avoid future outbreaks.

● Safely reopening Illinois requires knowing as quickly and accurately as possible if it looks like
current interventions aren’t working and we need to introduce new strategies. The same
detection systems would allow us to know as soon as possible if current interventions are
working well and a region can move onto the next stage of reopening.

● We need a strong, well-designed surveillance system to get us fast and actionable information
on COVID-19 across Illinois.

CONFIDENTIAL 17
The Limitations of the Data We Have Today

● Changes in the total number of test-positive cases or the

fraction testing positive are an unreliable measure of shifts.
These numbers should not be used to determine policy.
It’s unclear if these trends are
due to changes to increased ● Hospitalizations, ICU occupancy, and deaths are all later
testing capacity and/or changes
to eligibility criteria for testing events in the course of disease, so these events are too
over time late to estimate the rate of transmission in the community.
For reference, symptom onset precedes hospitalization by
approximately 4.2 days, ICU admission by 6.7 days, and
death by 11.7 days.

● In the best of all worlds, we would assess changes in

prevalence--which reveal current transmission rates--
Source: IDPH
through random testing of the general population over
time. But this is impractical due to costs.

CONFIDENTIAL 18
Sentinel Surveillance: Using surrogate populations to track
recent infections and transmission in the general population
Not just increased testing, but testing in a predefined population using consistent criteria

● We propose to track infection rates in two types of sentinel populations: (1) pregnant women presenting for
delivery, and (2) a fixed number of symptomatic individuals appearing daily at outpatient testing sites.

● Tracking newly symptomatic infections means we will have more advance notice that trends are moving in
the wrong direction, which will give us more time to prepare to respond than if we simply tracked
hospitalizations or deaths.

● Another surrogate population that should be established are patients admitted to the medical ICU. As long
as ICU capacity is not exceeded, this is a very reliable indicator of epidemic activity, and has been
recommended for tracking influenza prevalence in epidemic settings. This is one of the strategies currently
used in Hong Kong for COVID-19.

● Several of the members of our modeling task force have significant expertise designing such programs for
respiratory viruses such as influenza and have already begun composing detailed plans for how COVID-19
surveillance testing could happen here in Illinois.

CONFIDENTIAL 19
Next Steps
 High-Level Project Plan
Wk 1 Wk 2 Wk 3 Wk 4 Wk 5 Wk 6 ...

Incorporating hospital data from IDPH (“EMResource) and

Models v.1 creating projections through end of July, assuming current
Modeling

interventions remain in place)

Team

Create statewide projections by running models for Create regional models, incorporate
separate areas of IL and run scenarios e.g. lifting SIP detailed data from IDPH, aggregate to 4
Model Refinements sooner vs. later. Incorporate additional data from Restore Illinois Health regions
IDPH and hospitals.

● EMResource data (aggregate count data from hospitals) from IDPH

Identifying & Onboarding New ● Line-Level data from IDPH
Data, e.g. IDPH line-level data ● Line-Level and aggregate data from select health systems
and hospital data ● Mobility data from IDOT, Google
● Case management data on patients recovering from COVID-19
Civis Team

IDPH, CDPH Northshore Argonne National Others

Discovery interviews with (McKinsey), Cook Health Laboratory
additional modeling groups as County, UIUC, System,
identified UChicago, NW, Vizient/SG2
IHME

Status updates, pipeline building and dashboarding

Consolidation of Modelers’
Results and Productionizing into
Reports, Dashboards
21
Appendix
Non-ICU, ICU and Ventilator Capacity for the Super Regions
and State of IL

Geography Northeast North-Central Central Southern All of Illinois

Hospital Bed
10,903 2,328 1,833 1,394 16,458
Capacity

Adult ICU 2,826 409 289 240 3,764

Vent Capacity 2,780 450 360 240 3,830

Date of Extract 5/5/2020 5/5/2020 5/5/2020 5/5/2020 5/5/2020

Sources: EMResource on May 5, 2020

Additional capacity due to buildout of McCormick Place and other ACS needed to further refine these estimates.

23
PRELIMINARY AND CONFIDENTIAL
Modeling Team Experts
COVID-19 Modeling Team Leads

University of Chicago Title Contact

Sarah Cobey, PhD Associate Professor, Ecology and Evolution cobey@uchicago.edu

University of Illinois at Urbana-Champaign

Nigel Goldenfeld, PhD Professor, Biological Physics nigel@illinois.edu

Sergei Maslov, PhD Professor, Biological Physics maslov@illinois.edu

Northwestern University

Jaline Gerardin, PhD Assistant Professor of Preventive Medicine jgerardin@northwestern.edu

(Epidemiology) and McCormick School of
Engineering

Argonne National Laboratory

Charles Macal, PhD Senior Technical Advisor & Social, Behavioral, macal@anl.gov
and Decision Science Group Leader

Jonathan Ozik, PhD Computational Scientist jozik@anl.gov

24
PRELIMINARY AND CONFIDENTIAL