Synopsis of Projections Workgroup 4.13 · 4/8/2020 · Model Comparison: External models 1 10 100...
Transcript of Synopsis of Projections Workgroup 4.13 · 4/8/2020 · Model Comparison: External models 1 10 100...
Synopsis of Projections for the COVID‐19 Pandemic
COVID‐19 Projections Workgroup Arizona State UniversityThe University of Arizona
Facilitated by the Arizona Department of Health Services
Models as of 4/8/2020
• Arizona‐specific projections were deemed necessary because the Institute for Health Metrics and Evaluation (IHME) University of Washington model in common use was found to be often too optimistic in its projections and does not consider a variety of scenarios.
• Modeling and projections are highly complex, and require focused attention and effort by a team of experts over several weeks and potentially months.
• ADHS called upon partners at ASU and U of A with special previous experience doing projections and modeling, and who are working with CDC and the National Institutes of Health on COVID‐19 response.
Key findings: • Based on the latest data and considering multiple possible scenarios, this Arizona‐
specific model predicts infections will peak around the middle of May.• This model is highly‐sensitive to physical distancing and increased temperature. • A wide range in outcomes is still feasible because of uncertainties in the number of
undetected cases and the effectiveness of physical distancing.
Arizona‐Specific COVID‐19 Epidemic Projection Model
Epidemiology:
• Growth is rapid, but has slowed
• Doubling times– March 17‐24: 1.7 days
– March 25‐April 8: 5.3 days
• Does not include undetected cases
0
500
1000
1500
2000
2500
3000
3/4/
203/
5/20
3/6/
203/
7/20
3/8/
203/
9/20
3/10
/20
3/11
/20
3/12
/20
3/13
/20
3/14
/20
3/15
/20
3/16
/20
3/17
/20
3/18
/20
3/19
/20
3/20
/20
3/21
/20
3/22
/20
3/23
/20
3/24
/20
3/25
/20
3/26
/20
3/27
/20
3/28
/20
3/29
/20
3/30
/20
3/31
/20
4/1/
204/
2/20
4/3/
204/
4/20
4/5/
204/
6/20
4/7/
204/
8/20
Arizona COVID‐19 Testing
Positive Deaths
4/8: Growth is slowing down. 3/16 limited gatherings; 3/20 bars and restraurants close; 3/31 stay‐at‐home.
4/8: ADHS data show positive confirmations are currently 7% of all tests.
Testing:• Information about negative tests results released March 27.– Positive test result drops to <10%.
• Range of 8%‐10% is consistent with other US cities with community spread
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0
1000
2000
3000
4000
5000
6000
7000
8000
3/10
/20
3/12
/20
3/14
/20
3/16
/20
3/18
/20
3/20
/20
3/22
/20
3/24
/20
3/26
/20
3/28
/20
3/30
/20
4/1/
204/
3/20
4/5/
204/
7/20
Rolling weekly positive COVID‐19 test rate
Tests Positive %
4/8: Positive test rate has increased to close to 10% indicating higher burden of COVID‐19 disease in population.
Transmission• Early stochastic
effects• Fast exponential
growth• Slowing growth
0
1
2
3
4
5
6
7
8
9
3/6/20
3/7/20
3/8/20
3/9/20
3/10/2
03/1
1/20
3/12/2
03/1
3/20
3/14/2
03/1
5/20
3/16/2
03/1
7/20
3/18/2
03/1
9/20
3/20/2
03/2
1/20
3/22/2
03/2
3/20
3/24/2
03/2
5/20
3/26/2
03/2
7/20
3/28/2
03/2
9/20
3/30/2
03/3
1/20
4/1/20
4/2/20
4/3/20
4/4/20
4/5/20
4/6/20
4/7/20
4/8/20
AZ confirmed cases ‐ log scale
Early contact based spread. Limited transmission and contact tracing. Beginnings of community infection.
Undetected disease transmission, sustained community spread, and cases increase in severity.
Reduced transmissiondue (hopefully) to social distancing and increased testing.
Assumptions & Parameters
Assumptions & Parameters
Pyramid of Disease Severity • The assumed
parameters in the model are all sourced from recent publications in the literature
• The top of the pyramid implies significant healthcare resource requirements
TOTAL ASYMPTOMATIC CASES: 18.5%
TOTAL SYMPTOMATIC BUT NONHOSPITALIZED CASES: 65.2%
IN ICU: 7.34%(TOTAL HOSPITALIZED CASES: 16.3%)
VENTILATOR: 6.45%
DEATHS:1.61%
S E
IAsymp
Ipre-symp
ISymp
Ihosp D
R
Ihome
2 days 3 days
3 days
6 days
6 days
2 days
18.5%
20%
ICU
ICUREC
8 days
6 days45%
22%
Arizona COVID‐19 Model
Scenarios:Scenario Description
Scenario 1.
Assumes all infections are known based on a reporting rate of 9% and ``moderate” influence of physical distancing. The estimate of unreported cases obtained by an estimate provided by Shaman et. al. 2020. Assumes no additional mitigation.Summer effect is modeled by reducing transmission efficiency by half from May 15.
Scenario 2.
Assumes a reporting rate of 9% and ``maximal” influence of physical distancing. Assumes ongoing mitigation.Summer effect is modeled by reducing transmission efficiency by half from May 15.
Scenario 3.
Assumes that the current reported cases reflect the actual number of infected individuals as of 4/8/20 and moderate influence of physical distancing. Assumes no additional mitigation.Summer effect is modeled by reducing transmission efficiency by half from May 15.
Scenario 4.
Assumes that the current reported cases reflect the actual number of infected individuals as of 4/8/20 and maximal influence of physical distancing. Assumes no ongoing mitigation.Summer effect is modeled by reducing transmission efficiency by half on May 15.
Scenario 5.
Same as Scenario 1 with extreme summer‐time reduction in transmission efficiency (heat or distancing); Assumes no additional mitigation for physical distancing.
Total Infected • Total infected includes
asymptomatic and pre‐symptomatic individuals, who may be transmitting the disease
• The sharp decline in Scenario #5 due to the reduction in transmission rate due to summer effect– Assumes May 15 for
reduction in transmission– Summer effects not yet
known
Symptomatic Infections
• A large number of the symptomatic infections will recover at home– Due to physical
distancing measures, we assumed that these individuals will transmit the disease at a lower rate
Hospitalized Infections
• A portion of the hospitalized infections are in ICU, which we track separately due to the significant resources required to care for ICU patients
• Under our mid‐range scenario (Scenario #2), the number of hospitalized patients hit 13,091 on May 23
• Scenario #4 estimates a max of 1,258 patients on May 23, similar to IHME estimates of 1,203 on April 22
Patients in ICU• ICU resources can be
critical to save lives• In particular, several
sources have pointed to longer ICU stays by patients that eventually recover
• ICU stays can be as long as 14+ days for these patients
Patients on Ventilator
• A significant fraction of patients (~88%) need mechanical ventilators in ICU
• Rate of mortality among patients on mechanical ventilator is higher than other causes of Acute Respiratory Distress Syndrome (ARDS) (~67%)
Maximum Daily Counts: All scenarios
Low High Mid
Peak Infected 6,875 175,695 88,466
Peak Hopitalized (Daily) 1,259 31,670 15,428
Peak ICU (Daily) 591 14,981 7,126
Peak Ventilators (Daily) 520 13,183 6,270
Projected Infections: Low, medium, high
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
4/8/
204/
15/2
04/
22/2
04/
29/2
0
5/6/
205/
13/2
05/
20/2
05/
27/2
0
6/3/
206/
10/2
06/
17/2
06/
24/2
0
7/1/
20
7/8/
207/
15/2
07/
22/2
07/
29/2
0
Projected Infections: High, Med, Low(Estimate Made 4/8)
High Med Low
Projected Hospitalizations: Low, medium, high
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
4/8/
204/
15/20
4/22
/204/
29/20
5/6/
205/
13/20
5/20
/205/
27/20
6/3/
206/
10/20
6/17
/206/
24/20
7/1/
20
7/8/
207/
15/20
7/22
/207/
29/20
Projected Hospitalizations: High, Med, Low(Estimate Made 4/8)
High Med Low
Projected ICU visits: Low, medium, high
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
4/8/2
04/
15/2
04/
22/2
04/
29/2
0
5/6/2
05/
13/2
05/
20/2
05/
27/2
0
6/3/2
06/
10/2
06/
17/2
06/
24/2
0
7/1/2
0
7/8/2
07/
15/2
07/
22/2
07/
29/2
0
Projected ICU visits: High, Med, Low(Estimate Made 4/8)
High Med Low
Projected Ventilator Use: Low, medium, high
Assumes 88% ventilator utilization for ICU patients
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
4/8/2
04/1
5/20
4/22/
204/2
9/20
5/6/2
05/1
3/20
5/20/
205/2
7/20
6/3/2
06/1
0/20
6/17/
206/2
4/20
7/1/2
0
7/8/2
07/1
5/20
7/22/
207/2
9/20
Projected Ventilation Use : High, Med, Low(Estimate Made 4/8)
High Med Low
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
4/8/2
04/1
5/20
4/22/
204/2
9/20
5/6/2
05/1
3/20
5/20/
205/2
7/20
6/3/2
06/1
0/20
6/17/
206/2
4/20
7/1/2
0
7/8/2
07/1
5/20
7/22/
207/2
9/20
Projected Ventilation Use : High, Med, Low(Estimate Made 4/8)
High Med Low
Model Comparison: External models
1
10
100
1,000
10,000
100,000
1,000,000
10,000,000
1‐Apr
3‐Apr
5‐Apr
7‐Apr
9‐Apr
11‐A
pr13
‐Apr
15‐A
pr17
‐Apr
19‐A
pr21
‐Apr
23‐A
pr25
‐Apr
27‐A
pr29
‐Apr
1‐M
ay3‐
May
5‐M
ay7‐
May
9‐M
ay11
‐May
13‐M
ay15
‐May
17‐M
ay19
‐May
21‐M
ay23
‐May
25‐M
ay27
‐May
29‐M
ay
Forecasts of peak week and peak resources ‐ COVID‐19 Arizona
Total Infected Hospital Beds ICU Vents deaths
healthdata.org
Joe Gerard
Act now ‐worst case
Tim LantCurrent Model ‐Middle Scenario
Authors and Contributors:Team 1: The Biodesign Institute, Knowledge EnterpriseArizona State UniversityTim Lant, PhD, MAS; Megan Jehn, PhD; Esma Gel, PhD; Anna Muldoon, MPH; Heather Ross PhD, DNP, ANP‐BC;Neal W. Woodbury, PhD;
Team 2: The Center for Population Health SciencesArizona Health Sciences, the Zuckerman College of Public Health, and the Center forBiomedical Informatics and BiostatisticsUniversity of ArizonaJoe K. Gerald, MD, PhD;Patrick Wightman, PhD, MPP;Kathy Hiller, MD, MPH, FACEP;Vern Pilling;
Authors and Contributors:Team 3: The Center for Health Information and Research (CHIR)Arizona State UniversityGeorge Runger, PhD;Anita Murcko, MD, FACP;Dieter Armbruster, PhD;Gevork Harootunian;Klim Drobnyh;Seho Kee;Maziar Roodsari;Basam Alasaly;Nassim Idouraine;Logan Cameron;
Team 4: Seidman Research Institute, WP Carey School of Business, Arizona State UniversityTim James, PhDTimothy J. Richards, PhDMark Manfredo, PhD