|
50% reduction from Feb 2021 to Aug 2021
|
80% reduction from Feb 2021 to Aug 2021
|
|
50% reduction from April 2021 to October 2021
|
80% reduction from Feb 2021 to Aug 2021
|
|
50% reduction from March 2021 to Sept 2021
|
80% reduction from March 2021 to Sept 2021
|
|
50% reduction from March 2021 to Sept 2021
|
80% reduction from March 2021 to Sept 20211
|
|
50% reduction from April 2021 to October 2021
|
80% reduction from April 2021 to October 2021
|
|
50% reduction from April 2021 to October 2021
|
80% reduction from April 2021 to October 2021
|
|
20% increased transmissibility as compared by B.1.1.7 for B.1.617+ variant. 5% prevalence of B.1.617+ nationally on May 29.
|
60% increased transmissibility as compared by B.1.1.7 for B.1.617+ variant. 5% prevalence of B.1.617+ nationally on May 29.
|
|
20% increased transmissibility as compared by B.1.1.7 for B.1.617+ variant. 5% prevalence of B.1.617+ nationally on May 29.
|
60% increased transmissibility as compared by B.1.1.7 for B.1.617+ variant. 5% prevalence of B.1.617+ nationally on May 29.
|
|
40% increased transmissibility as compared with Alpha for Delta variant. Initial prevalence estimated at state-level by teams.
|
60% increased transmissibility as compared with Alpha for Delta variant. Initial prevalence estimated at state-level by teams.
|
|
40% increased transmissibility as compared with Alpha for Delta variant. Initial prevalence estimated at state-level by teams.
|
60% increased transmissibility as compared with Alpha for Delta variant. Initial prevalence estimated at state-level by teams.
|
|
|
- Protection from infection is: - 70% ≤ 65 years - 35% > 65 years - Protection from hospitalization and death is 90%
|
|
- Protection from infection is: - 50% ≤ 65 years - 25% > 65 years - Protection from hospitalization and death is 80%
|
- Protection from infection is: - 50% ≤ 65 years - 25% > 65 years - Protection from hospitalization and death is 80%
|
|
- The same mix of variants circulate throughout the projection period. No change in virus transmissibility
|
- The same mix of variants circulate throughout the projection period. No change in virus transmissibility
|
|
- A more transmissible variant emerges, comprising 1% of circulating viruses on Nov 15.
- The new variant is 1.5X as transmissible as viruses circulating at the beginning of the projection period.
|
- A more transmissible variant emerges, comprising 1% of circulating viruses on Nov 15.
- The new variant is 1.5X as transmissible as viruses circulating at the beginning of the projection period.
|
|
- Slow immune waning, average transition time to partially immune state = 1 year.
- In partially immune state,
|
- Slow immune waning, average transition time to partially immune state = 1 year.
- In partially immune state,
|
||||||||||||||||||||||||
|
- Fast immune waning, average transition time to partially immune state = 6 months.
- In partially immune state,
|
- Fast immune waning, average transition time to partially immune state = 6 months.
- In partially immune state,
|
|
- Advantage Omicron over Delta in South Africa, Rt ratio: 2.8 (Rt = 2.5)
- Intrinsic transmissibility Omicron = 1 x seasonally-adjusted R0 of Delta
- Immune escape = 80%
- Among naive individuals, 50% reduction in severity of omicron infection, relative to Delta (all-age risk of hospitalization and death divided by two; age-specific risks at teams discretion).
|
- Advantage Omicron over Delta in South Africa, Rt ratio: 2.8 (Rt = 2.5)
- Intrinsic transmissibility Omicron = 1.66 x seasonally-adjusted R0 of Delta*
- Immune escape = 50%
- Among naive individuals, 50% reduction in severity of omicron infection, relative to Delta (all-age risk of hospitalization and death divided by two; age-specific risks at teams discretion).
|
|
- Advantage Omicron over Delta in South Africa, Rt ratio: 2.8 (Rt = 2.5)
- Intrinsic transmissibility Omicron = 1 x seasonally-adjusted R0 of Delta
- Immune escape = 80%
- Among naive individuals, no change in severity of omicron infection, relative to Delta.
|
- Advantage Omicron over Delta in South Africa, Rt ratio: 2.8 (Rt = 2.5)
- Intrinsic transmissibility Omicron = 1.66 x seasonally-adjusted R0 of Delta*
- Immune escape = 50%
- Among naive individuals, no change in severity of omicron infection, relative to Delta.
|
|
- Immune escape = 80% of previously immune are susceptible to infection
- All other transmission characteristics at discretion of teams
- Lower levels of escape can be assumed for recently boosted individuals at discretion
- 70% reduction in severity of omicron infection, relative to Delta (all-age risk of hospitalization and death times 0.3; age-specific risks at teams discretion) among all immune classes
Example: A vaccinated person infected with Omicron has 30% the probability of death of a vaccinated person with Delta. Similarly, a naive person infected with Omicron has 30% the probability of death of a naive person infected with Delta.
|
- Immune escape = 50% of previously immune are susceptible to infection
- All other transmission characteristics at discretion of teams
- Lower levels of escape can be assumed for recently boosted individuals at discretion
- 70% reduction in severity of omicron infection, relative to Delta (all-age risk of hospitalization and death times 0.3; age-specific risks at teams discretion) among all immune classes
Example: A vaccinated person infected with Omicron has 30% the probability of death of a vaccinated person with Delta. Similarly, a naive person infected with Omicron has 30% the probability of death of a naive person infected with Delta.
|
|
- Immune escape = 80% of previously immune are susceptible to infection
- All other transmission characteristics at discretion of teams
- Lower levels of escape can be assumed for recently boosted individuals at discretion
- 30% reduction in severity of omicron infection, relative to Delta (all-age risk of hospitalization and death times 0.7; age-specific risks at teams discretion) among all immune classes.
Example:
Similar to above, but with 70% of the probability of death as
compared to the Delta infection.
|
- Immune escape = 50% of previously immune are susceptible to infection
- All other transmission characteristics at discretion of teams
- Lower levels of escape can be assumed for recently boosted individuals at discretion
- 30% reduction in severity of omicron infection, relative to Delta (all-age risk of hospitalization and death times 0.7; age-specific risks at teams discretion) among all immune classes.
Example:
Similar to above, but with 70% of the probability of death as
compared to the Delta infection.
|
|
Projections are initialized with the mix of strains circulating at the start of the projection period.
- Slow immune waning, median transition time to partially immune state = 10 months
- In the partially immune state, there is a 40% reduction in protection from baseline levels reported immediately after exposure (vaccination or infection)
|
- There is a continuous influx of 50 weekly infections of varaint X for the following 16 weeks
- Variant X has 30% immune escape, and the same intrinsic transmissibility and severity as Omicron
- Slow immune waning, median transition time to partially immune state = 10 months
- In the partially immune state, there is a 40% reduction in protection from baseline levels reported immediately after exposure (vaccination or infection)
|
|
Projections are initialized with the mix of strains circulating at the start of the projection period.
- Fast immune waning, median transition time to partially immune state = 4 months
- In the partially immune state, there is a 60% reduction in protection from baseline levels reported immediately after exposure (vaccination or infection)
|
- There is a continuous influx of 50 weekly infections of varaint X for the following 16 weeks
- Variant X has 30% immune escape, and the same intrinsic transmissibility and severity as Omicron
- Fast immune waning, median transition time to partially immune state = 4 months
- In the partially immune state, there is a 60% reduction in protection from baseline levels reported immediately after exposure (vaccination or infection)
|
|
- Protection from natural immunity and VE against infection decrease over time due to waning, but not due to variant mix.
- Risk of severe disease conditional on infection remains unchanged.
- A third booster recommendation is issued October 1st, 2022 for adults 50+ and those with chronic conditions, with reformulated vaccines.
- Booster uptake among the previously vaccinated is reduced by 15% compared to response to the first booster recommendation (x0.85 1st booster coverage). The distribution of who gets a booster among those for whom it is the 3rd, 4th or 5th dose of vaccine, and age differences in coverage within the 50+, is at the teams’ discretion.
- Recommended time between booster doses is maintained.
|
- 50 infections with variant X seeded weekly from Sep 4th-Dec 24th (16 weeks).
- Variant X has 40% immune escape against infection (applies to VE and to protection from natural immunity).
- There is a 20% increased risk of hospitalization and death with variant X, relative to Omicron, conditional on infection and immune status.
- A third booster recommendation is issued October 1st, 2022 for adults 50+ and those with chronic conditions, with reformulated vaccines.
- Booster uptake among the previously vaccinated is reduced by 15% compared to response to the first booster recommendation (x0.85 1st booster coverage). The distribution of who gets a booster among those for whom it is the 3rd, 4th or 5th dose of vaccine, and age differences in coverage within the 50+, is at the teams’ discretion.
- Recommended time between booster doses is maintained.
|
|
- Protection from natural immunity and VE against infection decrease over time due to waning, but not due to variant mix.
- Risk of severe disease conditional on infection remains unchanged.
- A COVID-19 booster campaign starts on October 1st, 2022 for all adults 18+, with reformulated vaccines.
- Coverage of boosters progresses throughout fall 2022 in different age groups at a 10% reduced coverage compared to the 2021-2022 flu vaccine coverage (x0.9 flu vaccine coverage) ; whether individuals get a 2nd or 3rd booster is at teams discretion.
- Boosters are recommended regardless of time since previous receipt of a booster.
|
- 50 infections with variant X seeded weekly from Sep 4th-Dec 24th (16 weeks).
- Variant X has 40% immune escape against infection (applies to VE and to protection from natural immunity).
- There is a 20% increased risk of hospitalization and death with variant X, relative to Omicron, conditional on infection and immune status.
- A COVID-19 booster campaign starts on October 1st, 2022 for all adults 18+, with reformulated vaccines.
- Coverage of boosters progresses throughout fall 2022 in different age groups at a 10% reduced coverage compared to the 2021-2022 flu vaccine coverage (x0.9 flu vaccine coverage) ; whether individuals get a 2nd or 3rd booster is at teams discretion.
- Boosters are recommended regardless of time since previous receipt of a booster.
|
|
- No new variant
- Protection from natural immunity and VE against infection decrease over time due to waning, but not due to variant mix.
- Risk of severe disease conditional on infection remains unchanged.
- Coverage of boosters progresses throughout fall 2022 in different age groups at a 10% reduced coverage (x0.9) compared to historical seasonal flu vaccination; whether individuals get a 2nd or 3rd booster is at teams discretion.
- Boosters are recommended regardless of time since previous receipt of a booster.
|
- 50 infections with variant X seeded weekly from Sep 4th-Dec 24th (16 weeks).
- 40% immune escape against infection (applies to VE and to protection from natural immunity).
- There is a 20% increased risk of hospitalization and death with variant X, relative to Omicron, conditional on infection and immune status.
- Coverage of boosters progresses throughout fall 2022 in different age groups at a 10% reduced coverage (x0.9) compared to historical seasonal flu vaccination; whether individuals get a 2nd or 3rd booster is at teams discretion.
- Boosters are recommended regardless of time since previous receipt of a booster.
|
|
- No new variant
- Protection from natural immunity and VE against infection decrease over time due to waning, but not due to variant mix.
- Risk of severe disease conditional on infection remains unchanged.
- Coverage of boosters progresses throughout fall 2022 in different age groups at a 10% reduced coverage (x0.9) compared to historical seasonal flu vaccination; whether individuals get a 2nd or 3rd booster is at teams discretion.
- Boosters are recommended regardless of time since previous receipt of a booster.
|
- 50 infections with variant X seeded weekly from Sep 4th-Dec 24th (16 weeks).
- 40% immune escape against infection (applies to VE and to protection from natural immunity).
- There is a 20% increased risk of hospitalization and death with variant X, relative to Omicron, conditional on infection and immune status.
- Coverage of boosters progresses throughout fall 2022 in different age groups at a 10% reduced coverage (x0.9) compared to historical seasonal flu vaccination; whether individuals get a 2nd or 3rd booster is at teams discretion.
- Boosters are recommended regardless of time since previous receipt of a booster.
|
|
- Variants have a 25% immune escape from BA.5.2
- Seeding based on combined observed prevalence of Level 5 variants at the start of the projection period
- No change in severity given symptomatic infection
- Teams are free to use available data and information from current and previous rollouts as tehy see fit to define rates
- Teams should assume increasing uptake through October and November as necessary to reach the projected February 1st, 2022 plateau
|
- Variants have a 50% immune escape from BA.5.2
- Seeding based on combined observed prevalence of Level 6 and 7 variants at the start of the projection period
- No change in severity given symptomatic infection
- Teams are free to use available data and information from current and previous rollouts as tehy see fit to define rates
- Teams should assume increasing uptake through October and November as necessary to reach the projected February 1st, 2022 plateau
|
|
- Variants have a 25% immune escape from BA.5.2
- Seeding based on combined observed prevalence of Level 5 variants at the start of the projection period
- No change in severity given symptomatic infection
- Teams are free to use available data and information from current and previous rollouts as tehy see fit to define rates
- Based on current rates, plateau date is flexible as long as it occurs before the end of the simulation (Teams can adjust rates up if needed to achieve adequate coverage by target date)
|
- Variants have a 50% immune escape from BA.5.2
- Seeding based on combined observed prevalence of Level 6 and 7 variants at the start of the projection period
- No change in severity given symptomatic infection
- Teams are free to use available data and information from current and previous rollouts as tehy see fit to define rates
- Based on current rates, plateau date is flexible as long as it occurs before the end of the simulation (Teams can adjust rates up if needed to achieve adequate coverage by target date)
|
Scenario Fullname | Scenario Id | Scenario Name | Social Distancing Measures | Testing-Trace-Isolate | Masking | Vaccine Efficacy | Vaccine Availability | Vaccine Hesitancy | Vaccine |
---|---|---|---|---|---|---|---|---|---|
“Optimistic” Scenario | A-2020-12-22 | optimistic | baseline state orders with regards to NPIs continue for six weeks from their start date (i.e., the date each individual state started the policy regime in place at baseline), interventions step down from baseline to the lowest levels seen since September 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | maintained at baseline levels indefinitely | 95% after two doses, 50% after one dose, doses 3.5 weeks apart | Actually distributed doses in December (approx.), 25 million courses (50 million doses) in January, 25 million courses per month thereafter | NA | NA |
Business as Usual + Moderate Vaccine Scenario | B-2020-12-22 | moderate | current elevated state orders with regards to NPIs continue for stated length or three weeks after the NPI is started if length is unstated; thereafter interventions step down from baseline to the lowest levels seen since May 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | maintained at baseline levels indefinitely | 70% after two doses, 20% after one dose, doses 3.5 weeks apart | Actually distributed doses in December (approx.), 12.5 million courses in January, 25 million courses per month thereafter | NA | NA |
Fatigue and Hesitancy Scenario | C-2020-12-22 | fatigue | current elevated state orders with regards to NPIs continue for stated length or three weeks after the NPI is started if length is unstated; thereafter interventions step down from baseline to an additional 5% below the lowest levels seen since May 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | adherence to these measures steps down from baseline to an additional 5% below the lowest levels seen since September 2020 in a particular jurisdiction over two one-month steps | 95% after two doses, 50% after one dose, doses 3.5 weeks apart. | Actually distributed doses in December (approx.), 12.5 million courses in January, 25 million courses per month thereafter | no more than 50% of any priority group accepts the vaccine | NA |
Counterfactual Scenario | D-2020-12-22 | counterfactual | current elevated state orders with regards to NPIs continue for stated length or three weeks after the NPI is started if length is unstated; thereafter interventions step down from baseline to the lowest levels seen since May 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | maintained at baseline levels indefinitely | NA | NA | NA | no vaccine |
- Baseline date: December 15, 2020 - date of baseline intervention levels
- Start date for first-round scenarios: January 3, 2021 (week ending January 9) - first date of simulated outcomes; model should not be fit to data from after this date
- Simulation end date: at least through week ending April 3, 2021 (13-week horizon); preferably July 3, 2021 (26-week horizon)
- Transmission assumptions: models fit to US state-specific dynamic up until time of submission – no proscribed R0, interventions, etc.
- Pathogenicity assumptions: no exogenous fluctuations in pathogenicity/transmissibility beyond seasonality effects
- Vaccine effectiveness: level of effectiveness and available doses are specified for each scenario; assumptions regarding time required to develop immunity, age-related variation in effectiveness, duration of immunity, and additional effects of the vaccine on transmission are left to the discretion of each team
- Vaccine allocation: between-state allocation is based on population per the CDC/NAS guidelines (proportional allocation); within-state allocation and the impact of vaccine hesitancy are left to the discretion of each team
- Vaccine immunity delay: There is approximately a 14 day delay according to the Pfizer data; because we suspect the post first dose and post second dose delays may be of similar length, we do not believe there is any need to explicitly model a delay, instead groups can delay vaccine receipt by 14 days to account for it
- Vaccine uptake: It is unlikely vaccine uptake will be 100% within prioritized groups, however sufficient data are not available to specify this; we will leave this to team discretion, but we ask that they include these assumptions in their meta-data file
- Vaccine rollout: rollout to follow ACIP recommendations unless known to be contradicted by state recommendations
- Phase 1a: health care workers, long-term care facilities
- Phase 1b: frontline essential workers, adults 75+
- Phase 1c: other essential workers, adults with high-risk conditions, adults 65-74
- NPI assumptions: phased reductions of NPIs in 2021 that align with patterns observed at different times in the course of the epidemic in 2020 (see scenario-specific guidance); teams have some liberty on how to implement these reductions within the guidelines
- Database tracking of NPIs: teams may use their own data if desired, otherwise we recommend the following sources as a common starting point:
- Coronavirus Government Response Tracker | Blavatnik School of Government
- Coronavirus State Actions - National Governors Association
- Geographic scope: state-level and national projections
- Results: some subset of the following
- Weekly incident deaths
- Weekly cumulative deaths since start of pandemic (use JHU CSSE for baseline)
- Weekly incident cases
- Weekly cumulative cases since start of pandemic (use JHU CSSE for baseline)
- Weekly incident hospitalizations
- Weekly cumulative hospitalizations since simulation start
- Weeks will follow epi-weeks (Sun-Sat) dated by the last day of the week
- “Ground Truth”: The same data sources as the forecast hub will be used to represent “true” cases, deaths and hospitalizations. Specifically, JHU CSSE data for cases and deaths and HHS data for hospitalization.
- Metadata: We will require a brief meta-data form, TBD, from all teams.
- Uncertainty: aligned with the Forecasting Hub we ask for 0.01, 0.025, 0.05, every 5% to 0.95, 0.975, and 0.99 quantiles
Scenario Fullname | Scenario Id | Scenario Name | Social Distancing Measures | Testing-Trace-Isolate | Masking | Vaccine Efficacy | Vaccine Availability | Variant Strain | Vaccine Hesitancy |
---|---|---|---|---|---|---|---|---|---|
“Optimistic” Scenario, No Variant Strain | A-2021-01-22 | optimistic_no_var | baseline state orders with regards to NPIs continue for six weeks from their start date (i.e., the date each individual state started the policy regime in place at baseline), interventions step down from baseline to the lowest levels seen since September 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | maintained at baseline levels indefinitely | 95% after two doses, 50% after one dose, doses 3.5 weeks apart | Actually administered doses in December and January, vaccine administration rate observed so far in January persists through the end of the month, 25 million courses distibuted per month thereafter (NOTE: administration refers to actual receipt by an individual, distribution to the doses being sent to states) | no variant strain | NA |
“Optimistic” Scenario, Variant Strain | B-2021-01-22 | optimistic_var | baseline state orders with regards to NPIs continue for six weeks from their start date (i.e., the date each individual state started the policy regime in place at baseline), interventions step down from baseline to the lowest levels seen since September 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | maintained at baseline levels indefinitely | 95% after two doses, 50% after one dose, doses 3.5 weeks apart | Actually administered doses in December and January, vaccine administration rate observed so far in January persists through the end of the month, 25 million courses distibuted per month thereafter (NOTE: administration refers to actual receipt by an individual, distribution to the doses being sent to states) | variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of cases infected by a single case over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model) | NA |
Fatigue and Hesitancy Scenario, No Variant Strain | C-2021-01-22 | fatigue_no_var | current elevated state orders with regards to NPIs continue for stated length or three weeks after the NPI is started if length is unstated; thereafter interventions step down from baseline to an additional 5% below the lowest levels seen since May 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | adherence to these measures steps down from baseline to an additional 5% below the lowest levels seen since September 2020 in a particular jurisdiction over two one-month steps | 95% after two doses, 50% after one dose, doses 3.5 weeks apart. | Actually administered doses in December and January, vaccine administration rate observed to date in January persists indefinitely until the proportion vaccinated reaches the hesitancy threshold | no variant strain | no more than 50% of any priority group accepts the vaccine |
Fatigue and Hesitancy Scenario, Variant Strain | D-2021-01-22 | fatigue_var | current elevated state orders with regards to NPIs continue for stated length or three weeks after the NPI is started if length is unstated; thereafter interventions step down from baseline to an additional 5% below the lowest levels seen since May 2020 in a particular jurisdiction over two one-month steps | constant at baseline levels | adherence to these measures steps down from baseline to an additional 5% below the lowest levels seen since September 2020 in a particular jurisdiction over two one-month steps | 95% after two doses, 50% after one dose, doses 3.5 weeks apart. | Actually administered doses in December and January, vaccine administration rate observed to date in January persists indefinitely until the proportion vaccinated reaches the hesitancy threshold | variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of cases infected by a single case over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model) | no more than 50% of any priority group accepts the vaccine |
- Baseline date: January 23, 2021 - date of baseline intervention levels
- End date for fitting data: January 23, 2021 - no fitting should be done to data from after this date
- Start date for first-round scenarios: January 24, 2021 (week ending January 30) - first date of simulated outcomes
- Simulation end date: at least through week ending April 24, 2021 (13-week horizon); preferably July 24, 2021 (26-week horizon)
- Transmission assumptions: models fit to US state-specific dynamic up until End date for fitting data specified above – no proscribed R0, interventions, etc.
- Pathogenicity assumptions: no exogenous fluctuations in pathogenicity/transmissibility beyond seasonality effects
- Vaccine effectiveness: level of effectiveness and available doses are specified for each scenario; assumptions regarding time required to develop immunity, age-related variation in effectiveness, duration of immunity, and additional effects of the vaccine on transmission are left to the discretion of each team
- Vaccine allocation: between-state allocation is based on population per the CDC/NAS guidelines (proportional allocation); within-state allocation and the impact of vaccine hesitancy are left to the discretion of each team
- Vaccine immunity delay: There is approximately a 14 day delay according to the Pfizer data; because we suspect the post first dose and post second dose delays may be of similar length, we do not believe there is any need to explicitly model a delay, instead groups can delay vaccine receipt by 14 days to account for it
- Vaccine uptake: It is unlikely vaccine uptake will be 100% within prioritized groups, however sufficient data are not available to specify this; we will leave this to team discretion, but we ask that they include these assumptions in their meta-data file
- Vaccine rollout: rollout to follow ACIP recommendations unless known to be contradicted by state recommendations
- Phase 1a: health care workers, long-term care facilities
- Phase 1b: frontline essential workers, adults 75+
- Phase 1c: other essential workers, adults with high-risk conditions, adults 65-74
- NPI assumptions: phased reductions of NPIs in 2021 that align with patterns observed at different times in the course of the epidemic in 2020 (see scenario-specific guidance); teams have some liberty on how to implement these reductions within the guidelines
- Database tracking of NPIs: teams may use their own data if desired, otherwise we recommend the following sources as a common starting point:
- Coronavirus Government Response Tracker | Blavatnik School of Government
- Coronavirus State Actions - National Governors Association
- Geographic scope: state-level and national projections
- Results: some subset of the following
- Weekly incident deaths
- Weekly cumulative deaths since start of pandemic (use JHU CSSE for baseline)
- Weekly incident cases
- Weekly cumulative cases since start of pandemic (use JHU CSSE for baseline)
- Weekly incident hospitalizations
- Weekly cumulative hospitalizations since simulation start
- Weeks will follow epi-weeks (Sun-Sat) dated by the last day of the week
- “Ground Truth”: The same data sources as the forecast hub will be used to represent “true” cases, deaths and hospitalizations. Specifically, JHU CSSE data for cases and deaths and HHS data for hospitalization.
- Metadata: We will require a brief meta-data form, TBD, from all teams.
- Uncertainty: aligned with the Forecasting Hub we ask for 0.01, 0.025, 0.05, every 5% to 0.95, 0.975, and 0.99 quantiles
- Ensemble Inclusion: at present time, in order to be included in the ensemble models need to provide a full set of quantiles
Scenario Fullname | Scenario Id | Scenario Name | Social Distancing Measures | Testing-Trace-Isolate | Masking | Vaccine Efficacy (2-Dose Vaccines) | Vaccine Availability | B.1.1.7 Variant Strain | Vaccine Coverage |
---|---|---|---|---|---|---|---|---|---|
High Vaccination, Moderate NPI | A-2021-03-05 | highVac_modNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change. Declines gradually over a period of 5 months starting at the beginning of March and ending in August at 50% of the effectiveness of control observed for February 2021. Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns), but should occur over the full period. The effectiveness of control in February 2021 should be based on the last two weeks of the month. Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Feb 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place, but still including immunity, vaccination, etc. We recognize that there is uncertainty about what the effects would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | First dose: 90% against disease, 14 days after 1st dose Second dose: 95% against disease, 14 days after 2nd dose Transmission impact at teams’ discretion and should be clearly documented in team’s metadata. Doses 3.5 weeks apart | December, January, and February: based on data on administered doses (second doses should be taken into account) March-August: 35 million administered first doses/month, with the intention of protocols being followed (70M doses/mo) If the maximum level of vaccination specified (e.g., 90% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) The specified scenarios do not include the Johnson and Johnson one-dose vaccine, so it should not be modeled. Next round may include the explicit introduction of J&J vaccine. | Teams can model the B.1.1.7 variant as appropriate to their model. Any assumptions should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of cases infected by a single case over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences in severity, mortality, or VE are assumed in default. | No more than 90%_ of any population group receives the vaccine |
High Vaccination, Low NPI | B-2021-03-05 | highVac_lowNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change. Declines gradually over a period of 5 months starting at the beginning of March and ending in August at 20% of the effectiveness of control (i.e., an 80% reduction in effectiveness) observed for February 2021. Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns), but should occur over the full period. The effectiveness of control in February 2021 should be based on the last two weeks of the month. Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Feb 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place but would still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | First dose: 90% against disease, 14 days after 1st dose Second dose: 95% against disease, 14 days after 2nd dose Transmission impact at teams’ discretion and should be clearly documented in team’s metadata. Doses 3.5 weeks apart | December, January, and February: Administered doses (second doses should take into account) March-August: 35 million administered first doses/month, with the intention of protocols being followed (70M doses/mo) If the maximum level of vaccination specified (e.g., 90% in this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) The specified scenarios do not include the Johnson and Johnson one-dose vaccine, so it should not be modeled. Next round may include the explicit introduction of J&J vaccine. | Teams can model the B.1.1.7 variant as appropriate to their model. Any assumptions should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of cases infected by a single case over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences in severity, mortality, or VE are assumed in default. | No more than 90%_ of any population group receives the vaccine |
Low Vaccination, Moderate NPI | C-2021-03-05 | lowVac_modNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change. Declines gradually over a period of 5 months starting at the beginning of March and ending in August at 50% of the effectiveness of control observed for February 2021. Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns), but should occur over the full period. The effectiveness of control in February 2021 should be based on the last two weeks of the month. Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Feb 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place but would still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | First dose: 50% against disease, 14 days after 1st dose Second dose: 75% against disease, 14 days after 2nd dose Transmission impact at teams’ discretion and should be clearly documented in team’s metadata. Doses 3.5 weeks apart | December, January, and February: based on data on administered doses (second doses should take into account) March-August: 20 million administered first doses/month, with the intention of protocols being followed (40M doses/mo) If the maximum level of vaccination specified (e.g., 50% in this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) The specified scenarios do not include the Johnson and Johnson one-dose vaccine, so it should not be modeled. Next round may include the explicit introduction of J&J vaccine. | Teams can model the B.1.1.7 variant as appropriate to their model. Any assumptions should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of cases infected by a single case over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences in severity, mortality, or VE are assumed in default. | No more than 50%_ of any population group receives the vaccine |
Low Vaccination & Low NPI | D-2021-03-05 | lowVac_lowNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change. Declines gradually over a period of 5 months starting at the beginning of March and ending in August at 20% of the effectiveness of control (i.e., an 80% reduction in effectiveness) observed for February 2021. Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns), but should occur over the full period. The effectiveness of control in February 2021 should be based on the last two weeks of the month. Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Feb 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place but would still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | First dose: 50% against disease, 14 days after 1st dose Second dose: 75% against disease, 14 days after 2nd dose Transmission impact at teams’ discretion and should be clearly documented in team’s metadata. Doses 3.5 weeks apart | December, January, and February: based on data on administered doses (second doses should take into account) March-August: 20 million administered first doses/month, with the intention of protocols being followed (40M doses/mo) If the maximum level of vaccination specified (e.g., 50% in this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) The specified scenarios do not include the Johnson and Johnson one-dose vaccine, so it should not be modeled. Next round may include the explicit introduction of J&J vaccine. | Teams can model the B.1.1.7 variant as appropriate to their model. Any assumptions should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of cases infected by a single case over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences in severity, mortality, or VE are assumed in default. | No more than 50%_ of any population group receives the vaccine |
- Baseline date: See specific details below
- End date for fitting data: March 6, 2021 - no fitting should be done to data from after this date
- Start date for third-round scenarios: March 7, 2021 (week ending March 12) - first date of simulated outcomes
- Simulation end date: at least through week ending June 5, 2021 (13-week horizon); preferably Sept 4, 2021 (26-week horizon)
- Transmission assumptions: models fit to US state-specific dynamic up until End date for fitting data specified above – no proscribed R0, interventions, etc.
- Pathogenicity assumptions: no exogenous fluctuations in pathogenicity/transmissibility beyond seasonality effects
- Vaccine effectiveness: level of effectiveness and available doses are specified for each scenario; assumptions regarding time required to develop immunity, age-related variation in effectiveness, duration of immunity, and additional effects of the vaccine on transmission are left to the discretion of each team
- Vaccine allocation: between-state allocation is based on population per the CDC/NAS guidelines (proportional allocation); within-state allocation and the impact of vaccine hesitancy are left to the discretion of each team
- Vaccine immunity delay: There is approximately a 14 day delay according to the Pfizer data; because we suspect the post first dose and post second dose delays may be of similar length, we do not believe there is any need to explicitly model a delay, instead groups can delay vaccine receipt by 14 days to account for it
- Vaccine uptake: See specific details below.
- Vaccine rollout: rollout to follow ACIP recommendations unless known to be contradicted by state recommendations
- Phase 1a: health care workers, long-term care facilities
- Phase 1b: frontline essential workers, adults 75+
- Phase 1c: other essential workers, adults with high-risk conditions, adults 65-74
- NPI assumptions: phased reductions of NPIs in 2021 that align with patterns observed at different times in the course of the epidemic in 2020 (see scenario-specific guidance); teams have some liberty on how to implement these reductions within the guidelines
- Database tracking of NPIs: teams may use their own data if desired, otherwise we recommend the following sources as a common starting point:
- Coronavirus Government Response Tracker | Blavatnik School of Government
- Coronavirus State Actions - National Governors Association
- Geographic scope: state-level and national projections
- Results: some subset of the following
- Weekly incident deaths
- Weekly cumulative deaths since start of pandemic (use JHU CSSE for baseline)
- Weekly incident reported cases
- Weekly cumulative reported cases since start of pandemic (use JHU CSSE for baseline)
- Weekly incident hospitalizations
- Weekly cumulative hospitalizations since simulation start
- Weeks will follow epi-weeks (Sun-Sat) dated by the last day of the week
- “Ground Truth”: The same data sources as the forecast hub will be used to represent “true” cases, deaths and hospitalizations. Specifically, JHU CSSE data for cases and deaths and HHS data for hospitalization.
- Metadata: We will require a brief meta-data form, TBD, from all teams.
- Uncertainty: aligned with the Forecasting Hub we ask for 0.01, 0.025, 0.05, every 5% to 0.95, 0.975, and 0.99 quantiles
- Ensemble Inclusion: at present time, in order to be included in the ensemble models need to provide a full set of quantiles
Scenario Fullname | Scenario Id | Scenario Name | Social Distancing Measures | Testing-Trace-Isolate | Masking | Vaccination - Pfizer / Moderna | Vaccination - Johnson & Johnson | Vaccination Coverage | B.1.1.7 Variant Strain |
---|---|---|---|---|---|---|---|---|---|
High Vaccination, Moderate NPI | A-2021-03-28 | highVac_modNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change.Declines over a period of 6 months starting in April 2021 and ending in September 2021 at 50% of the effectiveness of control observed for March 2021.Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns).Decline can be gradual or sudden, and can differ in speed between states.The effectiveness of control in March 2021 should be based on the last two weeks of the month.Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Mar 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place, but still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | Vaccine efficacy (2-dose vaccines): First dose: 75% against disease, 14 days after 1st dose Second dose: 95% against disease, 14 days after 2nd dose Effectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.Doses 3.5 weeks apart. Vaccine availability: December, January, February, and March: based on data on administered doses (second doses should be taken into account) April-September: 50 million administered first doses/month, with the intention of protocols being followed (70M doses/mo) If the maximum level of vaccination specified (e.g., 90% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) | Vaccine efficacy (1-dose vaccine): Single dose: 70% against symptoms, 14 days after doseEffectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.. Vaccine availability: March: based on data on administered doses, with continuing at rate current on date of projection for remainder of monthApril-September: 10M administered in April, 15M in May, 20M June, 20M July, 20M August, 20M September administered doses/month. | No more than 90% of any population group receives the vaccine. If the maximum level of vaccination specified (e.g., 90% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new dose administration beyond this amount). | Teams should model the B.1.1.7 variant as appropriate to their model. Any assumptions (e.g., differences in severity/mortality, VE, or natural immunity) should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of infections by a single infected individual over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences between B.1.1.7 and current strains in severity, mortality, or VE are assumed in default. |
High Vaccination, Low NPI | B-2021-03-28 | highVac_lowNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change.Declines over a period of 6 months starting in April 2021 and ending in September 2021 at 20% of the effectiveness of control (i.e., an 80% reduction in effectiveness) of control observed for March 2021.Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns).Decline can be gradual or sudden, and can differ in speed between states.The effectiveness of control in March 2021 should be based on the last two weeks of the month.Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Mar 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place, but still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | Vaccine efficacy (2-dose vaccines): First dose: 75% against disease, 14 days after 1st dose Second dose: 95% against disease, 14 days after 2nd dose Effectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.Doses 3.5 weeks apart. Vaccine availability: December, January, February, and March: based on data on administered doses (second doses should be taken into account) April-September: 50 million administered first doses/month, with the intention of protocols being followed (70M doses/mo) If the maximum level of vaccination specified (e.g., 90% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) | Vaccine efficacy (1-dose vaccine): Single dose: 70% against symptoms, 14 days after doseEffectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.. Vaccine availability: March: based on data on administered doses, with continuing at rate current on date of projection for remainder of monthApril-September: 10M administered in April, 15M in May, 20M June, 20M July, 20M August, 20M September administered doses/month. | No more than 90% of any population group receives the vaccine. If the maximum level of vaccination specified (e.g., 90% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new dose administration beyond this amount). | Teams should model the B.1.1.7 variant as appropriate to their model. Any assumptions (e.g., differences in severity/mortality, VE, or natural immunity) should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of infections by a single infected individual over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences between B.1.1.7 and current strains in severity, mortality, or VE are assumed in default. |
Low Vaccination, Moderate NPI | C-2021-03-28 | lowVac_modNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change.Declines over a period of 6 months starting in April 2021 and ending in September 2021 at 50% of the effectiveness of control observed for March 2021.Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns).Decline can be gradual or sudden, and can differ in speed between states.The effectiveness of control in March 2021 should be based on the last two weeks of the month.Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Mar 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place, but still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | Vaccine efficacy (2-dose vaccines): First dose: 50% against disease, 14 days after 1st dose Second dose: 85% against disease, 14 days after 2nd dose Effectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.Doses 3.5 weeks apart. Vaccine availability: December, January, February, and March: based on data on administered doses (second doses should be taken into account) April-September: 45 million administered first doses/month, with the intention of protocols being followed (90M doses/mo) If the maximum level of vaccination specified (e.g., 75% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) | Vaccine efficacy (1-dose vaccine): Single dose: 60% against symptoms, 14 days after doseEffectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.. Vaccine availability: March: based on data on administered doses, with continuing at rate current on date of projection for remainder of monthApril-September: 5M administered doses/month. | No more than 75% of any population group receives the vaccine. If the maximum level of vaccination specified (e.g., 75% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new dose administration beyond this amount). | Teams should model the B.1.1.7 variant as appropriate to their model. Any assumptions (e.g., differences in severity/mortality, VE, or natural immunity) should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of infections by a single infected individual over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences between B.1.1.7 and current strains in severity, mortality, or VE are assumed in default. |
Low Vaccination, Low NPI | D-2021-03-28 | lowVac_lowNPI | Includes combined effectiveness/impact of all non-pharmaceutical interventions and behavior change.Declines over a period of 6 months starting in April 2021 and ending in September 2021 at 20% of the effectiveness of control (i.e., an 80% reduction in effectiveness) observed for March 2021.Decline can be implemented at teams’ discretion (e.g., daily or monthly stepdowns).Decline can be gradual or sudden, and can differ in speed between states.The effectiveness of control in March 2021 should be based on the last two weeks of the month.Reduction should be implemented based on each team’s best judgment, but should be done in such a way that a 100% reduction (0% of Mar 2021 effectiveness) would approximate an epidemic without NPIs (e.g. no masks, no social distancing) in place, but still including immunity, vaccination, etc. We recognize that there is uncertainty about what transmission would be without NPIs; this uncertainty should be incorporated into the scenario projections. | constant at baseline levels | Included as part of “Social Distancing Measures” above. | Vaccine efficacy (2-dose vaccines): First dose: 50% against disease, 14 days after 1st dose Second dose: 85% against disease, 14 days after 2nd dose Effectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.Doses 3.5 weeks apart. Vaccine availability: December, January, February, and March: based on data on administered doses (second doses should be taken into account) April-September: 45 million administered first doses/month, with the intention of protocols being followed (90M doses/mo) If the maximum level of vaccination specified (e.g., 75% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new doses beyond this amount) | Vaccine efficacy (1-dose vaccine): Single dose: 60% against symptoms, 14 days after doseEffectiveness and impact on infection and other outcomes (hospitalizations, deaths) is at team’s discretion and should be clearly documented in team’s metadata.. Vaccine availability: March: based on data on administered doses, with continuing at rate current on date of projection for remainder of monthApril-September: 5M administered doses/month. | No more than 75% of any population group receives the vaccine. If the maximum level of vaccination specified (e.g., 75% for this scenario) is reached in all population groups, assume that no more vaccination occurs (i.e., do not model new dose administration beyond this amount). | Teams should model the B.1.1.7 variant as appropriate to their model. Any assumptions (e.g., differences in severity/mortality, VE, or natural immunity) should be clearly defined in the metadata.The default assumptions are that the variant is 1.5x more transmissible than current strains and reaches 50% dominance by March 15 and 100% dominance by May 1 (see MMWR report); here a 1.5x increase in transmissibility is defined as the increase in the expected number of infections by a single infected individual over their entire course of infection when there are no interventions or immunity in the population (e.g., a 1.5x increase in R0 in a classic epidemic model). No differences between B.1.1.7 and current strains in severity, mortality, or VE are assumed in default. |
- Baseline date: See specific details below
- End date for fitting data: March 27, 2021 - no fitting should be done to data from after this date
- Start date for third-round scenarios: March 28, 2021 (week ending April 3) - first date of simulated outcomes
- Simulation end date: at least through week ending June 26, 2021 (13-week horizon); preferably Sept 25, 2021 (26-week horizon)
- Transmission assumptions: models fit to US state-specific dynamic up until End date for fitting data specified above – no proscribed R0, interventions, etc.
- Pathogenicity assumptions: no exogenous fluctuations in pathogenicity/transmissibility beyond seasonality effects
- Vaccine effectiveness: level of effectiveness and available doses are specified for each scenario; assumptions regarding time required to develop immunity, age-related variation in effectiveness, duration of immunity, and additional effects of the vaccine on transmission are left to the discretion of each team
- Vaccine allocation: between-state allocation is based on population per the CDC/NAS guidelines (proportional allocation); within-state allocation and the impact of vaccine hesitancy are left to the discretion of each team
- Vaccine immunity delay: There is approximately a 14 day delay according to the Pfizer data; because we suspect the post first dose and post second dose delays may be of similar length, we do not believe there is any need to explicitly model a delay, instead groups can delay vaccine receipt by 14 days to account for it
- Vaccine uptake: See specific details below.
- Vaccine rollout: rollout to follow ACIP recommendations unless known to be contradicted by state recommendations
- Phase 1a: health care workers, long-term care facilities
- Phase 1b: frontline essential workers, adults 75+
- Phase 1c: other essential workers, adults with high-risk conditions, adults 65-74
- NPI assumptions: phased reductions of NPIs in 2021 that align with patterns observed at different times in the course of the epidemic in 2020-21 (see scenario-specific guidance); teams have some liberty on how to implement these reductions within the guidelines
- Database tracking of NPIs: teams may use their own data if desired, otherwise we recommend the following sources as a common starting point:
- Coronavirus Government Response Tracker | Blavatnik School of Government
- Coronavirus State Actions - National Governors Association
- Geographic scope: state-level and national projections
- Results: some subset of the following
- Weekly incident deaths
- Weekly cumulative deaths since start of pandemic (use JHU CSSE for baseline)
- Weekly incident reported cases
- Weekly cumulative reported cases since start of pandemic (use JHU CSSE for baseline)
- Weekly incident hospitalizations
- Weekly cumulative hospitalizations since simulation start
- Weeks will follow epi-weeks (Sun-Sat) dated by the last day of the week
- “Ground Truth”: The same data sources as the forecast hub will be used to represent “true” cases, deaths and hospitalizations. Specifically, JHU CSSE data for cases and deaths and HHS data for hospitalization.
- Metadata: We will require a brief meta-data form, TBD, from all teams.
- Uncertainty: aligned with the Forecasting Hub we ask for 0.01, 0.025, 0.05, every 5% to 0.95, 0.975, and 0.99 quantiles
- Ensemble Inclusion: at present time, in order to be included in the ensemble models need to provide a full set of quantiles
* Vaccine-eligible population. The eligible population for vaccination is presumed to be individuals aged 16 years or older until June 1, 2021. On June 1, the eligible population is presumed to extend to individuals aged 12 years and older. * Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The saturation levels provided in these scenarios are National reference points to guide defining hesitancy, though the speed of that saturation and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling team. The high vaccination 83% saturation is defined using the current estimates from the Delphi group (link) from March 13, 2021 data. The low saturation estimate of 68% is the lowest county-level estimate from the U.S. Census Bureau’s Pulse Survey from March 15, 2021 data (link). Both of these saturation levels are assumed to be among the population eligible for vaccination, not the full population.
Vaccination
Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The saturation levels provided in these scenarios are illustrative National reference points to guide defining hesitancy. The high vaccination scenario (low hesitancy) saturates at approximately 83% vaccination coverage nationally among the eligible population, as defined by current estimates from the Delphi group (link) from March 13, 2021 data (red line in figure). The low vaccination scenario (high hesitancy) saturates at approximately 68% vaccination coverage nationally among the eligible population, defined by the lowest county-level estimate from the U.S. Census Bureau’s Pulse Survey (link) from March 15, 2021 data. The speed of vaccination saturation should be defined by the modeling team, and can be defined as a logistic function (red and blue lines in figure below) or at different speeds (green line below). State or smaller geospatial unit and/or age group hesitancy limits should be defined by the modeling team using their best judgement. Overall national hesitancy should be similar to the illustrative levels defined in the scenarios, though is not expected to be exact. The eligible population for vaccination is presumed to be individuals aged 16 years or older until June 1, 2021. On June 1, the eligible population is presumed to extend to individuals aged 12 years and older.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. High Vaccination, Moderate NPI | highVac_modNPI | A-2021-05-02 |
Scenario B. High Vaccination, Low NPI | highVac_lowNPI | B-2021-05-02 |
Scenario C. Low Vaccination, Moderate NPI | lowVac_modNPI | C-2021-05-02 |
Scenario D. Low Vaccination & Low NPI | lowVac_lowNPI | D-2021-05-02 |
* Vaccine-eligible population. The eligible population for vaccination is presumed to be individuals aged 16 years or older until May 12, 2021. On May 12, the eligible population is extended to individuals aged 12 years and older, through the end of the projection period. * Vaccine hesitancy expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The saturation levels provided in these scenarios are National reference points to guide defining hesitancy, though the speed of that saturation and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling team. The high vaccination 86% saturation is defined using the current estimates from the Delphi group (link, updated from Round 5). The low saturation estimate of 75% is the lowest county-level estimate from the U.S. Census Bureau’s Pulse Survey from Apr 14-26, 2021 data (link), which is updated from Round 5.
NPI: In contrast to past scenarios, we do not specify different levels of non-pharmaceutical interventions (NPI) use here. The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata.
Vaccination
B.1.617+ variant strain with increased transmissibility: The scenarios define the B.1.617-like variants as 20% and 60% more transmissible than B.1.1.7 and other strains circulating in the US and is at 5% national prevalence on May 29, 2021. This 5% proportion on May 29th is a national estimate; teams can use state-specific data if they wish to. Timeframe of the increase in variant prevalence is up to each team, but it should be assumed the variant(s) become dominant due to increased transmissibility. The variant is more transmissible but it is not an immune escape variant; further, no decline of immunity from vaccination (other than VE) or natural infection should be modeled for B.1.617+ or other circulating variants. Other assumptions are at the discretion of each team, but should be documented in metadata. More info on next page.
Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The saturation levels provided in these scenarios are illustrative National reference points to guide defining hesitancy. The high vaccination scenario (low hesitancy) saturates at 86% vaccination coverage nationally among the vaccine-eligible population (updated from 83% in Round 5), as defined by current estimates from the Delphi group (link) (red line in figure, borrowed from round 5, but the same spirit applies to round 6). The low vaccination scenario (high hesitancy) saturates at 75% vaccination coverage nationally among the vaccine-eligible population, defined by the lowest county-level estimate from the U.S. Census Bureau’s Pulse Survey (link) from April 24, 2021 data. The speed of vaccination saturation should be defined by the modeling teams, and can be defined as a logistic function (red and blue lines in figure below) or at different speeds (green line below). State or smaller geospatial unit hesitancy limits should be defined by the modeling team using their best judgment. Overall national hesitancy should be similar to the illustrative levels defined in the scenarios, though is not expected to be exact. The eligible population for vaccination is presumed to be individuals aged 12 years and older.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. High Vaccination, Low Variant Transmissibility Increase | highVac_lowVar | A-2021-06-08 |
Scenario B. High Vaccination, High Variant Transmissibility Increase | highVac_highVar | B-2021-06-08 |
Scenario C. Low Vaccination, Low Variant Transmissibility Increase | lowVac_lowVar | C-2021-06-08 |
Scenario D. Low Vaccination, High Variant Transmissibility Increase | lowVac_highVar | D-2021-06-08 |
Round 7 is an update of Round 6 with updated data and understanding of both the Delta variant and Vaccination hesitancy.
* Vaccine-eligible population. The eligible population for vaccination is presumed to be individuals aged 12 years and older through the end of the projection period. ** Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The saturation levels provided in these scenarios are National reference points to guide defining hesitancy, though the speed of that saturation and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling team. The high vaccination 80% saturation is defined crudely as using the current estimates from the Delphi group, adjusted for potential bias in respondents, who tend to be more highly vaccinated that the general US population (link, updated from Round 6). The low saturation estimate of 70% is based on an adjustment of the Pulse Survey overall estimate, adjusted for survey participant vaccination coverage. This number also mirrors the lowest county-level estimate (73.3%) from the U.S. Census Bureau’s Pulse Survey from May 26-June 7, 2021 (link), which is updated from Round 6.
NPI: In contrast to past scenarios, we don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools intend to return to in-person education in the fall. The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata.
Vaccination
Delta variant strain with increased transmissibility: The scenarios define the Delta (B.1.617.2) variant as 40% and 60% more transmissible than Alpha (B.1.1.7.) Initial prevalence should be estimated or defined by the teams based on sequencing and other relevant data, preferably at the state level. Timeframe of the increase in variant prevalence is up to each team, but it should be assumed the variant(s) become dominant due to increased transmissibility. The variant is more transmissible but it is not an immune escape variant; further, no decline of immunity from vaccination (other than VE) or natural infection should be modeled for Delta or other circulating variants. Other assumptions are at the discretion of each team, but should be documented in metadata. More info on next page.
Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The saturation levels provided in these scenarios are illustrative National reference points to guide defining hesitancy. The high vaccination scenario (low hesitancy) saturates at 86% vaccination coverage nationally among the vaccine-eligible population (updated from 83% in Round 5), as defined by current estimates from the Delphi group (link) (red line in figure, borrowed from round 5, but the same spirit applies to round 6). The low vaccination scenario (high hesitancy) saturates at 75% vaccination coverage nationally among the vaccine-eligible population, defined by the lowest county-level estimate from the U.S. Census Bureau’s Pulse Survey (link) from April 24, 2021 data. The speed of vaccination saturation should be defined by the modeling teams, and can be defined as a logistic function (red and blue lines in figure below) or at different speeds (green line below). State or smaller geospatial unit hesitancy limits should be defined by the modeling team using their best judgment. Overall national hesitancy should be similar to the illustrative levels defined in the scenarios, though is not expected to be exact. The eligible population for vaccination is presumed to be individuals aged 12 years and older.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. High Vaccination, Low Variant Transmissibility Increase | highVac_lowVar | A-2021-07-13 |
Scenario B. High Vaccination, High Variant Transmissibility Increase | highVac_highVar | B-2021-07-13 |
Scenario C. Low Vaccination, Low Variant Transmissibility Increase | lowVac_lowVar | C-2021-07-13 |
Scenario D. Low Vaccination, High Variant Transmissibility Increase | lowVac_highVar | D-2021-07-13 |
Round 8 focuses on waning immunity.
Interpretation: These scenarios illustrate a gradual decay of immune protection with time, rather than the impact of an immune escape variant.
Model structure: Teams are encouraged to model waning immunity as a partial loss of immune protection, where individuals go back to a partially immune state after a period prescribed in the scenarios (mean of 1 or 3 yrs depending on the scenario). Individuals who have reached a partially immune state have reduced probabilities of reinfection and severe disease compared to naive individuals. In scenarios B-D, the distribution of immune decay should follow an exponential distribution. Scenario A has no waning. The same parameters should be used for waning immunity from natural infection and vaccination. Teams are encouraged to model these compartments separately however, in preparation for future scenarios focused on vaccine boosters.
Model parameters defined in scenarios: Parameters in these scenarios are based on observational studies of reinfection (natural immunity), vaccine breakthroughs, and models of decay of antibodies and VE over time. To illustrate the meaning of the scenario parameters, in scenario B for example, we have a protection from infection of 70% for individuals <65yrs in the partially immune state. This means that, for older individuals, the transition out of the partially immune state and into infection is 0.3*force of infection applied to naive individuals of the same age. If we apply this waning parameter to vaccinated people, this corresponds to a VE of 70% against infection. Further, in this scenario, protection against hospitalization and death is 90%. This estimate is similar to VE against hospitalization and death, so it is not a conditional probability. This means that if we follow two individuals over time, one with partial immunity and one completely naive, the probability that the partially immune individual will be hospitalized (die) from COVID19 is 0.1 times the probability that a naive individual will be hospitalized (die). Hence this probability combines protection against infection and protection against hospitalization/death given infection. If we apply this parameter to vaccinated individuals for whom immunity has partially waned, their VE against hospitalization and death becomes 90%.
Unconstrained model parameters: Teams should use their own judgments to parametrize protection against symptoms in the partially immune state, and any reduction in transmission that partially immune individuals may have. Teams can choose to treat individuals who have been infected and vaccinated differently from individuals who had a single exposure to the pathogen/antigen. We do not specify any waning for J&J (for which the starting point VE is much lower): teams can choose to ignore J&J, which represents a small fraction of all vaccinated in the US, or apply a different waning for J&J. We do not specify any waning for those who only get a 1st dose of Pfizer or Moderna and hence never acquire full vaccine immunity. We believe this represents a small fraction of all vaccinated. Teams can choose to apply a different waning to these individuals, or ignore them. All of these assumptions should be documented in meta-data.
Vaccination
Variant progression and transmissibility: Teams should use their own judgment to project the continued progress and transmissibility of the Delta variant across US states. In this scenario, no new variant would arrive in the US between now and the end of the projections. Initial prevalence should be estimated or defined by the teams based on sequencing and other relevant data, preferably at the state level. The variant is more transmissible but it is not an immune escape variant. Teams can set an increased severity of the Delta variant, but this should be documented in meta-data.
NPI: In contrast to past scenarios, we don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools intend to return to in-person education in the fall. Teams should also note the change in CDC mask recommendations for vaccinated people in high-transmission areas on 07/27/2021.The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. No Waning | noWan | A-2021-08-17 |
Scenario B. High Protection, Fast Waning | highPro_fastWan | B-2021-08-17 |
Scenario C. Low Protection, Slow Waning | lowPro_slowWan | C-2021-08-17 |
Scenario D. Low Protection, Fast Waning | lowPro_fastWan | D-2021-08-17 |
To assist with upcoming ACIP recommendations on childhood vaccination (ages 5-11), Round 9 of SMH will concentrate on evaluating the impact of childhood vaccination on COVID-19 dynamics. Results are expected to be needed by mid-September 2021.
Interpretation: These scenarios are intended to demonstrate the impact of vaccination among children ages 5 to 11. We additionally include a stress test axis which illustrates the potential impact of the emergence of a new more transmissible variant.
Model parameters defined in scenarios: With regards to the childhood vaccination axis, the data childhood vaccination begins and the state-level uptake trajectory is defined in the scenario. State-level uptake should reflect the percentage coverage increases observed in the 12 to 17-year-old age group observed since distribution to this group began on May 13, 2021. Baseline state-level age-specific vaccination data can be found here. Teams should specify in their metadata file if they use an alternative source for vaccination uptake. All assumptions about saturation over the course of the projection period should be specified in the metadata. Vaccine uptake among individuals age 12 and over should be the same in all four scenarios. Uptake in these age groups can be extrapolated from past vaccine coverage curves and vaccine hesitancy surveys (Pulse, CovidCast) with the methodology specified in the metadata. With regards to the new variant axis, the date of emergence, starting prevalence, and transmissibility increase compared to the Delta variant is specified by the scenarios.
Unconstrained model parameters: The following parameters are left to the disrection of the teams and should be noted in the metadata
Outputs: In addition to the usual outputs, it would be helpful (but not required) for teams to plan to extract incident and cumulative cases, hospitalizations, and deaths for under 12 years AND 12+ years (ideal). Alternative age-specific projections will also be helpful (e.g., 0-17, 5-17). Please plan to submit quantiles for the complement of the younger age group submitted as it is not possible to extract quantiles for the older age-group by subtracting from quantiles submitted for the total population. This will allow us to provide some information on indirect effects of vaccinating children 5 to 11 years of age. Additionally, please provide population data relevant to the age groups used so appropriate rates can be calculated.
Vaccination
Variant progression and transmissibility: Teams should use their own judgment to project the continued progress and transmissibility of the Delta variant across US states. Initial prevalence should be estimated or defined by the teams based on sequencing and other relevant data, preferably at the state level. Teams can set an increased severity of the Delta variant, but this should be documented in metadata.
NPI: In contrast to past scenarios, we don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools intend to return to in-person education in the fall. Teams should also note the change in CDC mask recommendations for vaccinated people in high-transmission areas on 07/27/2021.The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Childhood Vaccination, No Variant | ChildVax_noVar | A-2021-09-14 |
Scenario B. No Childhood Vaccination, No Variant | noChildVax_noVar | B-2021-09-14 |
Scenario C. Childhood Vaccination, New Variant | ChildVax_Var | C-2021-09-14 |
Scenario D. No Childhood Vaccination, New Variant | noChildVax_Var | D-2021-09-14 |
Round 10 of SMH will concentrate on evaluating the impact of boosters and waning immunity on COVID-19 dynamics. We have designed a 2*2 scenario structure where booster coverage is represented in one axis and the characteristics of waning immunity are on the other axis.
Interpretation: These scenarios are intended to illustrate a gradual decay of immune protection with time, rather than the impact of an immune escape variant.
Model structure: Teams are encouraged to model waning immunity as a partial loss of immune protection, where individuals go back to a partially immune state after a period prescribed in the scenarios (mean of 6 month or 1 year depending on the scenario). Individuals who have reached a partially immune state have reduced probabilities of reinfection and severe disease compared to naive individuals.
The same parameters should be used for waning immunity from natural infection and vaccination.
Model parameters defined in scenarios:
Interpretation of waning parameters is similar to that of round 8.
Specifically, in the optimistic waning scenario, protection from infection is 60% for individuals < 65yrs in the partially immune state. This means that, for these individuals, the transition out of the partially immune state and into infection is 0.4*force of infection applied to naive individuals of the same age. If we apply this waning parameter to vaccinated people, this corresponds to a VE of 60% against infection.
Further, in this scenario, protection against hospitalization is 90% for those under 65 yrs. This estimate is similar to VE against hospitalization and death, so it is not a conditional probability. This means that if we follow two individuals over time, one with partial immunity and one completely naive, the probability that the partially immune individual will be hospitalized from COVID19 is 0.1 times the probability that a naive individual will be hospitalized. Hence this probability combines protection against infection and protection against hospitalization given infection. If we apply this parameter to vaccinated individuals for whom immunity has partially waned, their VE against hospitalization becomes 90%.
Unconstrained model parameters:
Teams can choose different distributions of waning immunity (exponential, gamma); we only prescribe the mean.
Teams should use their own judgments to parametrize protection against symptoms in the partially immune state, and any reduction in transmission that partially immune individuals may have.
Teams can choose to treat individuals who have immunity from natural infection and vaccination differently from individuals who had a single exposure to the pathogen/antigen.
We do not specify any waning for J&J (for which the starting point VE is much lower): teams can choose to ignore J&J, which represents a small fraction of all vaccinated in the US, or apply a different waning for J&J.
We do not specify any waning for those who only get a 1st dose of Pfizer or Moderna and hence never acquire full vaccine immunity. We believe this represents a small fraction of all vaccinated. Teams can choose to apply a different waning to these individuals, or ignore them.
All of these assumptions (especially the distribution of waning times) should be documented in meta-data.
Initial VE (before waning): We suggest that teams use a VE of 80% against symptomatic COVID-19 in those over 65 yrs, and VE of 90% in those under 65 years, while protection against infection and severe outcomes remains at teams’ discretion. This is based on data from US, UK, CDC, NY and CDC MMWR.
Impact of boosters on VE: Boosters should be implemented in a way that individuals who have received a booster shot will revert to the same level of protection that they had before any waning occurred.
Booster coverage:
Booster timing:
Vaccination
Coverage of initial vaccine courses (pre-boosters): Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The coverage saturation, the speed of that saturation, and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling teams. We suggest that the teams use estimates from the Delphi group, adjusted for potential bias in respondents (link) and the Pulse Survey overall estimates, adjusted for survey participant vaccination coverage (link).
Vaccine-eligible population. The eligible population for 1st/2nd dose vaccination is presumed to be individuals aged 12 years and older until November 15, and 5 years and older from November 15 through the end of the projection period.
For vaccine coverage in the 5-11 yo, starting on November 15, 2021, we recommend the same strategy as in round 9. Specifically, state-specific vaccine coverage in 12-17 yrs from May 2021 onwards should be applied to the 5-11 yo.
Variant progression and transmissibility:
Teams should use their own judgment to project the continued progress and transmissibility of the Delta variant, and related lineages, across US states. In this round, there is no new variant that arrives in the US between now and the end of the projections. \
Teams can implement increases in transmissibility or severity of the Delta variant, but these should fit within the scenario specifications and should be fully documented in meta-data.
NPI:
We don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools have returned to in-person education in fall 2021 and high level health officials have noted that “people should feel safe to mingle at Thanksgiving and Christmas”. The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata. Teams should also note the change in CDC mask recommendations for vaccinated people in high-transmission areas on 07/27/2021.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Optimistic waning, widespread boosters | optWan_highBoo | A-2021-11-09 |
Scenario B. Optimistic waning, restricted boosters | optWan_lowBoo | B-2021-11-09 |
Scenario C. Pessimistic waning, widespread boosters | pessWan_highBoo | C-2021-11-09 |
Scenario D. Pessimistic waning, restricted boosters | pessWan_lowBoo | D-2021-11-09 |
Round 11 of the COVID-19 Scenario Modeling Hub will concentrate on evaluating the impact of Omicron on COVID-19 dynamics. We have designed a 2*2 scenario structure where Omicron transmissibility and immune escape are represented in one axis and severity of Omicron are on the other axis. We will consider a 3-month horizon.
Immune escape represents an increase in risk of infection among those with immunity from prior exposure to SARS-CoV-2 (of any kind, vaccination or natural infection), due to changes in the genetic makeup of Omicron. As an illustration, an immune escape of 60% indicates that among those with prior immunity to past variants, 60% will be susceptible to Omicron infection, and 40% will be protected against Omicron infection. Among those infected with Omicron who had previous immunity due to vaccination or prior infection, a reduction in the probability of severe disease may occur. This is specified in the severity axis of the scenarios.
We provide both absolute R0 for Omicron and a fold increase over Delta. Assumptions are based on a ratio of Rt_Omicron to Rt_Delta of 2.8. Here Rt=S(t)*R0*alpha(t), where alpha represents the impact of NPI and seasonal forcing on transmission. We can assume that NPI and seasonal forcing is the same for both variants, so the ratio of 2.8 can be explained as differences in S(t) (immune differences, e.g., link) and R0 (intrinsic transmissibility differences). The parameters chosen for these scenarios cover a possible range of immunity and transmissibility differences between variants that would contribute to an observed Rt ratio of 2.8. We have used intermediate estimates based on results from the MOBS and Bedford labs.
The presence, duration, and extent of waning is left to the team’s discretion. For teams including waning explicitly, we recommend the following:
Model structure: For teams explicitly modeling waning, teams are encouraged to consider immunity as a partial loss of immune protection, where individuals go back to a partially immune state after a period of time which is left to the teams’ discretion (suggested 6 months to 1 year). Individuals who have reached a partially immune state have reduced probabilities of reinfection and severe disease compared to naive individuals. The same parameters can be used for waning immunity from natural infection and vaccination.
Suggested waning parameters: Interpretation of waning parameters is similar to that of Round 8. Specifically, protection from infection is 60% for individuals <65yrs in the partially immune state. This means that, for these individuals, the transition out of the partially immune state and into infection is 0.4*force of infection applied to naive individuals of the same age. If we apply this waning parameter to vaccinated people, this corresponds to a VE of 60% against infection. Further, suggested protection against hospitalization is 90% for those under 65 yrs. This estimate is similar to VE against hospitalization and death, so it is not a conditional probability. This means that if we follow two individuals over time, one with partial immunity and one completely naive, the probability that the partially immune individual will be hospitalized from COVID-19 is 0.1 times the probability that a naive individual will be hospitalized. Hence this probability combines protection against infection and protection against hospitalization given infection. If we apply this parameter to vaccinated individuals for whom immunity has partially waned, their VE against hospitalization becomes 90%.
Unconstrained model parameters:
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Optimistic severity, High immune escape/Low transmissibility increase | optSev_highIE | A-2021-12-21 |
Scenario B. Optimistic severity, Low immune escape/High transmissibility increase | optSev_lowIE | B-2021-12-21 |
Scenario C. Pessimistic severity, High immune escape/Low transmissibility increase | pessSev_highIE | C-2021-12-21 |
Scenario D. Pessimistic severity, Low immune escape/High transmissibility increase | pessSev_lowIE | D-2021-12-21 |
Other submission requirements
Vaccine coverage: Coverage of initial vaccine courses (pre-boosters): Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The coverage saturation, the speed of that saturation, and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling teams. We suggest that the teams use estimates from the Delphi group, adjusted for potential bias in respondents (link) and the Pulse Survey overall estimates, adjusted for survey participant vaccination coverage (link).
Vaccine-eligible population: The eligible population for 1st/2nd dose vaccination is presumed to be individuals aged 5 years and older through the end of the projection period.
Vaccine coverage in the 5-11yo: At team’s discretion.
Vaccine effectiveness: We recommend that teams use the following for VE against symptoms: VE=35% (first dose), VE=80% (2nd dose, > 65 yrs), VE= 90% (2nd dose, < 65 yrs) for Moderna/Pfizer, against Delta. This is the initial VE, before any waning or Omicron. VE is defined here as vaccine effectiveness against symptomatic disease. Teams should make their own informed assumptions about effectiveness and impacts on other outcomes (e.g., infection, hospitalization, death)
Impact of boosters on VE against Omicron: Boosters should be implemented in a way that individuals who have received a booster shot will revert to the same level of protection that they had before any waning occurred. Early data suggests that boosters of mRNA vaccine revert neutralization titers to Omicron to their base levels (the expectation would be that protection against all outcomes would revert to the levels seen with Delta, although there is considerable uncertainty) https://www.pfizer.com/news/press-release/press-release-detail/pfizer-and-biontech-provide-update-omicron-variant
Booster doses:
Booster timing: Current booster eligibility is 6 months after an individual’s 2nd dose.
We don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools have returned to in-person education in fall 2021 and high level health officials have noted that “people should feel safe to mingle at Thanksgiving and Christmas”. The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata. Teams should also note the change in CDC mask recommendations for vaccinated people in high-transmission areas on 07/27/2021. Additional scenario and simulation details
Round 12 of the COVID-19 Scenario Modeling Hub (SMH) will serve as an update of Round 11 to evaluate the impact of the Omicron wave, with updated data and epidemiological understanding. We have designed a 2*2 scenario structure where Omicron transmissibility and immune escape are represented in one axis and severity of Omicron are on the other axis. We will consider a 3-month horizon.
Immune escape represents an increase in risk of infection among those with immunity from prior exposure to SARS-CoV-2 (of any kind, vaccination or natural infection), due to changes in the genetic makeup of Omicron. As an illustration, an immune escape of 80% indicates that among those with prior immunity to past variants, 80% will be susceptible to Omicron infection (or 80% more likely to be infected in a leaky immunity model), and 20% will be protected against Omicron infection. Among those infected with Omicron who had previous immunity due to vaccination or prior infection, a reduction in the probability of severe disease may occur. This is specified in the severity axis of the scenarios. Since boosters seem to restore the protection lost by Omicron’s immune escape, teams can choose to reduce the impact of immune escape on boosted individuals. Alternatively, teams can apply the booster effect as a reduced probability of symptoms, hospitalization and death given infection.
We do not provide guidance on transmissibility, only on immune escape. Teams can use the growth curve of Omicron in the US or elsewhere, or other datasets, to set this parameter.
The presence, duration, and extent of waning is left to the team’s discretion.
Model structure: For teams explicitly modeling waning, teams are encouraged to consider immunity as a partial loss of immune protection, where individuals go back to a partially immune state after a period of time which is left to the teams’ discretion (suggested 6 months to 1 year). Individuals who have reached a partially immune state have reduced probabilities of reinfection and severe disease compared to naive individuals. The same parameters can be used for waning immunity from natural infection and vaccination.
Suggested waning parameters: Interpretation of waning parameters is similar to that of Round 8. Specifically, protection from infection is 60% for individuals <65yrs in the partially immune state. This means that, for these individuals, the transition out of the partially immune state and into infection is 0.4*force of infection applied to naive individuals of the same age. If we apply this waning parameter to vaccinated people, this corresponds to a VE of 60% against infection. Further, suggested protection against hospitalization is 90% for those under 65 yrs. This estimate is similar to VE against hospitalization and death, so it is not a conditional probability. This means that if we follow two individuals over time, one with partial immunity and one completely naive, the probability that the partially immune individual will be hospitalized from COVID-19 is 0.1 times the probability that a naive individual will be hospitalized. Hence this probability combines protection against infection and protection against hospitalization given infection. If we apply this parameter to vaccinated individuals for whom immunity has partially waned, their VE against hospitalization becomes 90%.
Unconstrained model parameters:
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Optimistic severity, High immune escape | optSev_highIE | A-2022-01-09 |
Scenario B. Optimistic severity, Low immune escape | optSev_lowIE | B-2022-01-09 |
Scenario C. Pessimistic severity, High immune escape | pessSev_highIE | C-2022-01-09 |
Scenario D. Pessimistic severity, Low immune escape | pessSev_lowIE | D-2022-01-09 |
Other submission requirements
Vaccine coverage: Coverage of initial vaccine courses (pre-boosters): Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The coverage saturation, the speed of that saturation, and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling teams. We suggest that the teams use estimates from the Delphi group, adjusted for potential bias in respondents (link) and the Pulse Survey overall estimates, adjusted for survey participant vaccination coverage (link).
Vaccine-eligible population: The eligible population for 1st/2nd dose vaccination is presumed to be individuals aged 5 years and older through the end of the projection period.
Vaccine coverage in the 5-11yo: At team’s discretion.
Vaccine effectiveness: We recommend that teams use the following for VE against symptoms: VE=35% (first dose), VE=80% (2nd dose, > 65 yrs), VE= 90% (2nd dose, < 65 yrs) for Moderna/Pfizer, against Delta. This is the initial VE, before any waning or Omicron. VE is defined here as vaccine effectiveness against symptomatic disease. Teams should make their own informed assumptions about effectiveness and impacts on other outcomes (e.g., infection, hospitalization, death)
Impact of boosters on VE against Omicron: Boosters should be implemented in a way that individuals who have received a booster shot will revert to the same level of protection that they had before any waning occurred. Early data suggests that boosters of mRNA vaccine revert neutralization titers to Omicron to their base levels (the expectation would be that protection against all outcomes would revert to the levels seen with Delta, although there is considerable uncertainty) https://www.pfizer.com/news/press-release/press-release-detail/pfizer-and-biontech-provide-update-omicron-variant
Booster doses:
Booster timing: Current booster eligibility is 6 months after an individual’s 2nd dose.
We don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools have returned to in-person education in fall 2021 and high level health officials have noted that “people should feel safe to mingle at Thanksgiving and Christmas”. The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata. Teams should also note the change in CDC mask recommendations for vaccinated people in high-transmission areas on 07/27/2021.
Round 13 of the COVID-19 Scenario Modeling Hub (SMH) considers the interaction of waning immunity against infection (first dimension) and the emergence of a new variant (2nd dimension) over a 52-week period. We follow the usual 2x2 table structure.
Risk of severe disease, conditional on infection, does not wane with time and does not change with variant X (see protection against severe disease section below).
Protection against infection: Waning is defined as a transition to a partially immune state, where individuals retain a long-lasting yet partial level of protection against (re)infection. This can be considered an asymptotic plateau for immunity, where the trajectory of antibodies and other immune components stabilizes on a timescale of 4 or 10 months, depending on the scenario.
We prescribe the relative reduction in protection against infection after the waning period, where comparison is to the levels observed immediately after natural infection or vaccination. For example in the optimistic waning scenarios, a 40% reduction from baseline levels corresponds to the case where protection from infection is 60% of the baseline levels reported immediately after exposure (vaccination or infection). In the pessimistic scenarios, 60% reduction corresponds to the case where protection from infection is 40% of the baseline levels reported immediately after exposure.
We leave the baseline levels of protection at teams’ discretion (eg, VE immediately after a 3rd vaccine dose), and only prescribe the relative reduction that applies after the waning period.
We assume that the timescale of waning does not depend on the number and type of exposures. For example, in scenario A, transition to a partially immune state would occur after a median of 10 mo after 2 vaccine doses, and so would the transition after 3 vaccine doses, or the transition after infection or re-infection.
Teams can choose to bump individuals to a higher level of protection after repeat exposures (where exposure is vaccination or infection), but waning would still occur on a 4 to 10 mo timescale after each new exposure. If repeat exposures raise immunity to a high level, then after 4 or 10 mo of waning, an individual could reach an asymptotic plateau that is higher than where the individual would be 4 or 10mo after a single exposure.
Natural immunity can be treated differently from vaccine-induced immunity, although the characteristics of decay of protection against infection should follow the parameters prescribed in the scenarios.
Teams can choose different distributions of waning immunity (exponential, gamma) as long as the median is as specified in the scenarios.
For scenarios B and D that consider new variant X, the risks of infection will need to be increased by the immune escape parameter provided in the second dimension of the table. Examples: For instance, let’s assume that VE against Omicron infection is 50% immediately after a booster shot in an individual <65 yrs. Then, per scenario A, protection should decline to 60% of the initial value after 10 mo of waning (40% reduction, cf table), so that protection should be 0.50*0.60=30% against Omicron infection 10 mo after boosting. This means VE is 30% against Omicron infection after 10 mo of waning for a boosted individual, or equivalently that their infection risk is 0.7*risk of infection of an unvaccinated individual.
The second example illustrates how a repeat exposure could bump individuals to a higher protection level. Let’s consider the same person from before, who was in a plateau of 30% protection against infection after 10 mo, relative to an unvaccinated individual. Let’s assume that this individual gets infected, immunity is boosted, resulting in a protection of 70% immediately after this new infection. After 10 mo, per scenario A, the residual protection against infection would be 0.7*0.6= 42% in this individual, relative to an unvaccinated individual.
References for VE by variant, number of doses, and time since vaccination, can be found here:
Protection against severe disease: We expect that at this stage of the pandemic, close to 100% of the US population has been naturally infected, vaccinated, or both, so that the entire population has long-lasting protection against severe disease upon (re)infection. The probabilities of hospitalization and death given (re)infection are left at teams’ discretion, with the understanding that this parameter can be calibrated against data during the Omicron wave, or defined based on (recent) literature (see below). It is assumed that the probabilities of hospitalization and death given (re)infection do not wane over the timescale of the projections and apply to all circulating variants, including new variant X. The probability of severe disease given (re)infection can vary by age and/or risk factors however. In other words, these conditional probabilities do not vary with time nor variants, but they can vary based on clinical and demographic host factors.
References on probability of hospitalization, conditional on (re)infection:
Immune escape. In scenarios B and D, we model the emergence of a new variant X, with moderate immune escape characteristics, taken to be 30%. Let’s consider an individual who is currently in a state of immunity to infection, gained from past exposure to SARS-CoV-2 antigens circulating before March 2022 (ie, infection with the wild type, Alpha, Delta, Omicron…) or vaccination (any number of doses). This individual, upon exposure to variant X, will have a 30% probability of infection with X, or a 30% increased probability of infection in a leaky model.
Immune escape only applies to risk of infection with X. Risk of severe disease given infection with variant X is a constant and is the same as that observed with Omicron, per the previous section.
Transmissibility, severity. The intrinsic transmissibility of the new variant should be the same as that of Omicron (same R0 as Omicron, with the R0 value of Omicron left at teams discretion). Similarly, the intrinsic severity of X should be the same as Omicron.
Introduction and ramp up. Variant X is to be seeded in the first week of May 2022 (May 1-7, 2022), with 50 active infections of variant X to be introduced during this week in the US (illustrating incoming variants from abroad). There will be a continuous influx of 50 weekly infections of variant X for the next 16 weeks (weeks starting May 1, 2022 and ending August 20, 2022). Geographic dispersion of these infections is left at teams discretion. The ramp up of the new variant due to local transmission is also left at the teams’ discretion.
Immunity after infection with variant X. Infection with variant X provides immunity to previously observed variants (e.g., Delta, Omicron). After infection with variant X, immune waning should progress as specified by the scenarios.
Scenarios in the 2*2 table specify the risks of infection, while the risks of hospitalization and death conditional on (re)infection are left at teams discretion but remain constant. We do not address case projections in the scenarios, and do not make particular assumptions on case ascertainment. At this point of the pandemic, reported cases include a mix of symptomatic infections reported to local authorities via active and passive surveillance testing, along with an unknown amount of asymptomatic infections. At home testing is not captured in case observations. We assume that over the 1 year projection period, testing propensity will remain constant at the level estimated at the start of the projection period. In other words, the infection to case ratio should be calibrated to observations in the weeks leading to the start of the projection period and be kept constant for the following year.
We assume that vaccine policy is set at the start of the projection period in March 2022 and remains constant for the duration of simulations. As of Feb 2022, vaccination is recommended for all individuals over 5 yrs, (one round of) boosters are recommended for all individuals over 12 yrs, there is no vaccine for children under 5yrs, there is no Omicron-specific vaccine, and two rounds of boosters (4 doses of mRNA vaccines) are not recommended for the general population. If new measures were to be announced before the start of round 13, we would include these measures in the scenarios.
Projection time horizon: We consider a 52-week projection period
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Optimistic waning, No immune escape variant | optWan_noVar | A-2022-02-25 |
Scenario B. Optimistic waning, New immune escape variant | optWan_Var | B-2022-02-25 |
Scenario C. Pessimistic waning, No immune escape variant | pessWan_noVar | C-2022-02-25 |
Scenario D. Pessimistic waning, New immune escape variant | pessWan_Var | D-2022-02-25 |
Other submission requirements
Vaccine coverage: Coverage of initial vaccine courses (pre-boosters): Vaccine hesitancy is expected to cause vaccination coverage to slow and eventually saturate at some level below 100%. The coverage saturation, the speed of that saturation, and heterogeneity between states (or other geospatial scales) and/or age groups are at the discretion of the modeling teams. We suggest that the teams use estimates from the Delphi group, adjusted for potential bias in respondents (link) and the Pulse Survey overall estimates, adjusted for survey participant vaccination coverage (link).
Vaccine-eligible population: The eligible population for 1st/2nd dose vaccination is presumed to be individuals aged 5 years and older through the end of the projection period.
Vaccine coverage in the 5-11yo: At team’s discretion.
Vaccine effectiveness: We recommend that teams use the following for VE against symptoms: VE=35% (first dose), VE=80% (2nd dose, > 65 yrs), VE= 90% (2nd dose, < 65 yrs) for Moderna/Pfizer, against Delta. This is the initial VE, before any waning or Omicron. VE is defined here as vaccine effectiveness against symptomatic disease. Teams should make their own informed assumptions about effectiveness and impacts on other outcomes (e.g., infection, hospitalization, death)
Impact of boosters on VE against Omicron: Boosters should be implemented in a way that individuals who have received a booster shot will revert to the same level of protection that they had before any waning occurred. Early data suggests that boosters of mRNA vaccine revert neutralization titers to Omicron to their base levels (the expectation would be that protection against all outcomes would revert to the levels seen with Delta, although there is considerable uncertainty) https://www.pfizer.com/news/press-release/press-release-detail/pfizer-and-biontech-provide-update-omicron-variant
We don’t specify different levels of non-pharmaceutical interventions (NPI) use; however, teams should consider that most schools have returned to in-person education in fall 2021 and high level health officials have noted that “people should feel safe to mingle at Thanksgiving and Christmas”. The future level of NPIs are left at the discretion of the modeling teams and should be specified in the teams’ metadata. Teams should also note the change in CDC mask recommendations for vaccinated people in high-transmission areas on 07/27/2021.
Round 14 of the COVID-19 Scenario Modeling Hub (SMH) considers the interaction of booster strategies (first dimension) with the epidemiology of circulating strains (2nd dimension) over a 52-week period. We follow the usual 2 x 2 table structure.
In all scenarios, the VE of reformulated boosters available starting Oct 1st, 2022 should be set to 80% against symptomatic disease with non-immune escape strains.
We recommend that in the waned classes, teams consider a reduction from baseline levels of protection ranging between 40 and 60%, corresponding to x0.60 and x0.40 of the baseline levels reported immediately after exposure (vaccination or infection). This follows the same scheme as in round 13. Teams can sample the recommended range of protection reductions, which is 40-60%, or use any value within this range as a point estimate.
These guidelines should not preclude teams from considering longer waning times, especially if they would like to integrate detailed waning data. A recent study suggests that vaccine-induced immunity wanes on long time scales and has not stabilized at 9 months. Accordingly, teams can choose to model longer time scales of waning, with a lower set point than prescribed above. If they do so, teams should ensure that in their formulation, 50% of their population has a 40-60% reduced protection at 4-6 months after (re-)exposure, aligned with the above guidelines.
2. Waning of immunity against severe disease
Absent a new variant, the risk of severe disease conditional on infection remains unchanged. This is the same assumption as in rd 13, which considers that while there is fast waning of immunity against infection, there is no waning in the risk of severe disease conditional on infection.
In scenarios B and D, we model the emergence of a new variant X, with moderate immune escape, set at 40%, and moderately increased risk of severe disease given infection, set at 20%.
Infection with variant X (immune escape): Let’s consider an individual who is currently in a state of immunity to infection, gained from past exposure to SARS-CoV-2 antigens (ie, infection with the wild type, Alpha, Delta, Omicron…) or vaccination (any number of doses). This individual, upon exposure to variant X, has a 40% probability of infection with this variant in an all-or-nothing model, or a 40% increased probability of infection in a leaky model.
Infection with variant X provides immunity to previously observed variants (e.g., Delta, Omicron). After infection with variant X, immune waning should progress as specified by the waning guidelines.
Severe disease with new variant X, given infection. In the new variant scenarios B and D, the risk of severe disease, conditional on infection and an individual’s immune class, increases by 20% relative to Omicron. In other words, the risk of severe disease given a new variant infection is x1.20 the risk of severe disease with Omicron infection, for a comparable individual in the same immune class. The factor by which the risk of severe disease, conditional on infection, increases with the new variant is the same for hospitalization and death. As a result, the new variant CFR is x1.20 Omicron CFR, for a given immune class.
By increasing the severity of variant X, conditional on infection, we implicitly consider the impact of two potential mechanisms that are non mutually exclusive: (i) variant X can partially evade a broad range of immune mediators that prevent progression to severe disease, and (ii) variant X is intrinsically more severe than Omicron, irrespective of immune escape and infectivity features (as seen with Delta).
Transmissibility. The intrinsic transmissibility of variant X should be the same as that of the strains circulating at the start of the projection period (same R0 as Omicron variants and subvariants = same effective transmissibility in a fully naive population). The intrinsic transmissibility of strains circulating at the start of the projection period is left at teams discretion).
Variant introduction and ramp up. Variant X is to be seeded in the first full week of September 2022 (Sep 4-10, 2022), with 50 active infections of variant X to be introduced during this week in the US (illustrating incoming variants from abroad). There will be a continuous influx of 50 weekly infections of variant X for a total of 16 weeks, ending the week of December 18-24, 2022, inclusive. Geographic dispersion of these infections is left at teams discretion. The ramp up of the new variant due to local transmission is also left at the teams’ discretion.
To address potential variation in the rise of variant X between models, and how variant X timing may interact with the timing of the vaccination campaign, we encourage teams to report the mean proportion of cases caused by variant X in each week and location, as prop_X=cases_of_X/all_cases
. This should be reported in the same file as the other targets.
There is evidence that vaccines will be reformulated in June 2022 and improved bivalent boosters will become available by October 2022. It is expected that the VE of reformulated bivalent boosters will provide a moderate improvement above existing boosters; yet the exact VE will depend on circulating strains this fall. Teams should set the VE of reformulated vaccines at 80% against symptomatic disease with the variants circulating at the start of the projection period (including all Omicron variants). For scenarios C and D in which variant X is seeded starting the first week of Sep 2022, a reduction of VE against (infection and) symptomatic disease should be implemented over the theoretical 80%, in line with the 40% immune escape parameter.
The VE of primary vaccine courses and non-reformulated boosters are left at teams’ discretion. Non-reformulated boosters will be discontinued after Oct 1st, 2022.
Relevant references include to set VE for primary courses and non-reformulated boosters include:
For scenarios C and D, teams should use the file here to simulate vaccine coverage in fall-winter 2022-2023. The data in this file provides weekly cumulative coverage by state and adult age groups to apply to scenario C and D. Estimates are based on the reported coverage of the flu vaccine in the 2020-2021 season by month, state and age. We applied the rise in coverage reported during Aug-Oct 2020 to Oct 2022, used a Piecewise Cubic Hermite Interpolating Polynomial to generate weekly coverage estimates, and applied a discounting factor of 0.9 to account for fatigue from repeat COVID vaccination. The distribution of who gets a booster among those for whom it is the 1st, 2nd or 3rd booster, and age differences in coverage within the 18+, is at the teams’ discretion.
Note that a 2nd booster recommendation for 50+ is already in place since March 29, 2022 and will proceed from projection start to October 1st, irrespective of the scenarios. Teams should use their best scientific judgment to model the impact of this recent recommendation.
More broadly, the booster coverage data provided for scenarios C, D are intended to represent the additive efforts of a new campaign with a reformulated vaccine starting in Oct 2022, over all of the (primary and booster) vaccination efforts that have already been undertaken up to that point.
Vaccine coverage among children:
Round 14 should not include reactive changes in NPIs imposed by health authorities to curb transmission, e.g., reinstatement of mask mandates, or closure of schools and businesses. However, teams can incorporate inherent changes in population behavior in response to increasing or decreasing incidences (eg, changes in contacts or masking), if these changes were inferred from earlier phases of the pandemic and are already part of the model. Database tracking of NPIs: teams may use their own data if desired, otherwise we recommend the following sources as a common starting point:
Ascertainment of cases, hospitalizations and deaths will proceed at the same level as they were at the start of the projection period. We will continue to collect the same targets (cases, hospitalizations, deaths) as in prior rounds but suggest that teams provide weekly incident infections (number of new infections each week), if they can. This is to facilitate comparisons between models, including outside of SMH, case ascertainment changes over time. A revised submission file template will be provided. If possible, teams should report the mean proportion of cases caused by variant X in each week and location as a new target named prop_X.
All of the teams’ specific assumptions should be documented in meta-data and abstract. In this round, we ask that teams provide the case ascertainment ratio (cases/infections) used in projections, and a description of how this parameter has been estimated.
Projection time horizon: We consider a 52-week projection period.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Age-restricted booster recommendations, No immune escape variant | restBoo_noVar | A-2022-05-09 |
Scenario B. Age-restricted booster recommendations, New immune escape variant | restBoo_Var | B-2022-05-09 |
Scenario C. Broad booster recommendations, No immune escape variant | broadBoo_noVar | C-2022-05-09 |
Scenario D. Broad booster recommendations, New immune escape variant | broadBoo_Var | D-2022-05-09 |
Other submission requirements
Round 15 is an update of Round 14 that considers the timing of the fall booster campaign (first dimension) with the epidemiology of circulating strains (2nd dimension) over a 40-week period. We follow the usual 2 * 2 table structure. We consider that reformulated boosters will be made available for all adults on Sep 11 or Nov 13, and that booster uptake will be indexed on the flu campaign.
The VE of reformulated boosters available in the fall should be set to 80% against symptomatic disease with all Omicron lineages (including BA.4/5) and pre-Omicron variants. However the VE of reformulated boosters should be decreased for variant X. Variant X partially escapes immunity conferred by all available vaccines and prior infection with Omicron and pre-Omicron lineages.
There may be an expansion of the 2nd booster recommendations to all adult age groups, sometime between the start of the projection period and the time when reformulated vaccines become available. Consideration of this expansion is left at teams’ discretion.
We recommend that in the waned classes, teams consider a reduction from baseline levels of protection ranging between 40 and 60%, corresponding to x0.60 and x0.40 of the baseline levels reported immediately after exposure (vaccination or infection). This follows the same scheme as in round 13-14. Teams can sample the recommended range of protection reductions, which is 40-60%, or use any value within this range as a point estimate.
These guidelines should not preclude teams from considering longer waning times, especially if they would like to integrate detailed waning data. A recent study suggests that vaccine-induced immunity wanes00089-7/fulltext) on long time scales and has not stabilized at 9 months. Accordingly, teams can choose to model longer time scales of waning, with a lower set point than prescribed above. If they do so, teams should ensure that in their formulation, 50% of their population has a 40-60% reduced protection 3-8 months after (re-)exposure, aligned with the above guidelines.
2. Waning of immunity against severe disease
Absent a new variant, the risk of severe disease conditional on infection remains unchanged. This is the same assumption as in rd 13-14, which considers that while there is fast waning of immunity against infection, there is no waning in the risk of severe disease conditional on infection.
In scenarios B and D, we model the emergence of a new variant X, with moderate immune escape, set at 40%, and moderately increased risk of severe disease given infection, set at 20%.
Infection with variant X (immune escape). Let’s consider an individual who is currently in a state of immunity to infection, gained from past exposure to SARS-CoV-2 antigens (ie, infection with the wild type, Alpha, Delta, Omicron…) or vaccination (any number of doses). This individual, upon exposure to variant X, will have a 40% probability of infection with X, or a 40% increased probability of infection in a leaky model.
Infection with variant X provides immunity to previously observed variants (e.g., Delta, Omicron). After infection with variant X, immune waning should progress as specified by the scenarios.
Severe disease with new variant X, given infection. In the new variant scenarios B and D, the risk of severe disease, conditional on infection and an individual’s immune class, increases by 20% relative to Omicron. In other words, the risk of severe disease given a new variant infection is x1.20 the risk of severe disease with Omicron infection, for a comparable individual in the same immune class. The factor by which the risk of severe disease, conditional on infection, increases with the new variant is the same for hospitalization and death. As a result, the new variant CFR is x1.20 Omicron CFR, for a given immune class.
By increasing the severity of variant X, conditional on infection, we implicitly consider the impact of two potential mechanisms that are non mutually exclusive: (i) variant X can partially evade a broad range of immune mediators that may have prevented progression to severe disease, and (ii) variant X may be intrinsically more severe, irrespective of immune escape and infectivity features (as seen with Delta).
Transmissibility. The intrinsic transmissibility of the new variant should be the same as that of the strains circulating at the start of the projection period (same R0 as Omicron variants and subvariants = same effective transmissibility in a fully naive population, with the R0 value of Omicron left at teams discretion)
Variant introduction and ramp up. Variant X is to be seeded in the first full week of September 2022 (Sep 4-10, 2022), with 50 active infections of variant X to be introduced during this week in the US (illustrating incoming variants from abroad). There will be a continuous influx of 50 weekly infections of variant X for a total of 16 weeks, until the week of December 18-24, 2022, inclusive. Geographic dispersion of these infections is left at teams discretion. The ramp up of the new variant due to local transmission is also left at the teams’ discretion.
Because there is little data on variant BA.2.75’s prevalence and fitness in the US, BA.2.75 should not be explicitly taken into account in the models.
In June 2022, FDA recommended that vaccines be reformulated and include two components, an original Wuhan-like strain and an Omicron BA4/BA5 strain. Reformulated bivalent boosters are expected to become available in fall 2022, although the exact timing is uncertain. We assume that reformulated bivalent boosters will provide a moderately improved protection above existing boosters; yet the exact VE will depend on circulating strains this fall. Teams should set the VE of reformulated vaccines at 80% against symptomatic disease with the variants circulating at the start of the projection period (including all Omicron variants). For scenarios C and D in which variant X emerges on Sep 1, a reduction of VE against (infection and) symptomatic disease should be implemented over the theoretical 80%, in line with the 40% immune escape parameter. The VE of a primary vaccine course and a non-reformulated booster are left at teams’ discretion. Note that as soon as reformulated boosters become available on Sep 11 or Nov 13, previously available vaccines will no longer be used.
Relevant references for VE of non reformulated vaccines include:
For scenarios C and D, teams should use the file here to simulate the coverage of reformulated boosters in fall-winter 2022-2023, and shift the start of the fall campaign according to rd 15 scenarios (Sep 11, 2022 or Nov 13, 2022). The data in this file provides weekly cumulative coverage by state and adult age groups to apply. Estimates are based on the reported coverage of the flu vaccine in the 2020-2021 season by month, state and age. We applied the rise in coverage reported during Aug-Oct 2020 to Oct 2022, used a Piecewise Cubic Hermite Interpolating Polynomial to generate weekly coverage estimates, and applied a discounting factor of 0.9 to account for fatigue from repeat COVID vaccination. The distribution of who gets a booster among those for whom it is the 1st, 2nd or 3rd booster, and age differences in coverage within the 18+, is at the teams’ discretion. Note that a 2nd booster recommendation for 50+ is already in place since March 29, 2022 and will proceed from projection start t to the start of the fall campaign with reformulated vaccines, irrespective of the scenarios. Teams should use their best scientific judgment to model the impact of this recent recommendation, and possible expansions to all adults. Vaccine coverage among children:
All of the teams’ specific assumptions should be documented in meta-data and abstract.
Projection time horizon: We consider a 40-week projection period.
Scenario | Scenario name for submission file | Scenario ID for submission file |
---|---|---|
Scenario A. Early boosters, No new variant | earlyBoo_noVar | A-2022-07-19 |
Scenario B. Early boosters, New immune escape variant | earlyBoo_Var | B-2022-07-19 |
Scenario C. Late boosters, No new variant | lateBoo_noVar | C-2022-07-19 |
Scenario D. Late boosters, New immune escape variant | lateBoo_Var | D-2022-07-19 |
Other submission requirements
Round 16 focuses on the impact of bivalent booster uptake (first dimension) with the epidemiology of the variant swarms projected to dominate in the coming months (2nd dimension) over a 26-week period. We follow the usual 2X2 table structure.
The VE of reformulated (i.e., bivalent) boosters currently administered to those five and older should be considered to have an effectiveness of
80% against symptomatic disease to BA.5.2 and all other Omicron variants not modeled in the immune escape swarms scenarios. For variants included in levels 5, 6 and 7, VE should be reduced based on the estimated immune escape factor compared to BA.5.2. If updated data on VE becomes available prior to submission, teams are free to use this data, but it should be noted in the abstract. Variant swarms in levels 5 to 7 partially escape immunity against infection, where immunity is conferred by all available vaccines and prior infection with Omicron BA1 through BA5 and pre-Omicron lineages. Variants by level of escape (from presentation by Cornelius Roemer, based on RBD mutations from BA.2):
Waning of immunity against infection
Models should include waning against infection. The median waning time of protection against infection should range between 3-8 months. Teams can sample this range, or use any value within this range as a point estimate. Teams can consider differences in waning of natural and vaccine-induced immunity, or in waning after Omicron infection vs waning from other types of SARS-CoV-2 exposures; however the median waning time should remain within the 3-8 month range.
We recommend that in the waned classes, teams consider a reduction from baseline levels of protection ranging between 40 and 60%, corresponding to x0.60 and x0.40 of the baseline levels reported immediately after exposure (vaccination or infection). This follows the same scheme as in round 13-15. Teams can sample the recommended range of protection reductions, which is 40-60%, or use any value within this range as a point estimate.
These guidelines should not preclude teams from considering longer waning times, especially if they would like to integrate detailed waning data. A recent study suggests that vaccine-induced immunity wanes00089-7/fulltext) on long time scales and has not stabilized at 9 months. Accordingly, teams can choose to model longer time scales of waning, with a lower set point than prescribed above. If they do so, teams should ensure that in their formulation, 50% of their population have a 40-60% reduced protection 3-8 months after (re-)exposure, aligned with the above guidelines.
Waning of immunity against severe disease
The extent and speed of the waning of protection against severity, conditional on infection, are at the discretion of the teams. Our assumptions are that protection against severity, conditional on infection, wanes on a slower time scale than waning against infection, but may wane eventually. Assumptions regarding waning of protection against severity, conditional on infection, should be provided in the abstract. For reference, several publications have estimates: NEJM, MMWR.
We model the emergence of new variants with different immune escape characteristics by level of escape (based on RBD mutations compared to BA.2). With little data on new emerging variants, a specific variant is not explicitly considered in the scenarios. Instead, new variants are grouped into levels based on their immune escape characteristics. Therefore, each group of variants with a particular level of immune escape can be modeled as a single variant with the specified immune escape characteristics. Levels 5, 6 and 7 variants are taken into account in Round 16, and the detailed characteristics of variants by level are defined in the scenarios as follows: Current variants classified into levels 5 , 6, and 7:
For example, in scenarios A and C, individuals who are previously immunized via either infection with BA.5.2 or vaccination with the reformulated vaccines will have a 25% reduction in the assumed level of protection conferred by that infection/vaccination against infection with Level 5 variants. For individuals who were most recently immunized by a less recent variant (i.e., BA.1) or vaccine (booster 1), protection against infection with Level 5 variants will be reduced by 25% on top of additional immune escape from that variant or vaccine by BA.5.2.
The relationship between immune escape against infection and against symptomatic disease is at the discretion of the teams.
Emerging variants not specified in the scenarios should be treated as not having an epidemiologically significant impact. For example, in scenarios A and C (the level 5 variant scenarios) level 6 and 7 variants should be treated the same as level 0-4 variants. In addition, level 0-4 variants should be considered as low or no immune escape compared to BA.5.
Severe disease with new variants, given infection: The risks of severe disease for both Level 5 & 6/7 variants, conditional on infection and an individual's immune class, are identical with Omicron (including BA.5.x). This is also true for other currently circulating variants. Accordingly, the risk for hospitalization and death, conditional on infection, is equivalent to Omicron.
Transmissibility: The intrinsic transmissibility of the new variant should be the same as that of the strains circulating at the start of the projection period (i.e. the same R0 as Omicron variants and sub-variants = same effective transmissibility in a fully naive population, with the R0 value of Omicron left at teams' discretion).
Initial variant prevalence: The initial prevalence of the Level 5 & 6/7 variants should be based on observed combined prevalence of all variants included in the given level at the start of the projection period in the US. Teams are free to use available data to inform the prevalence of new variants. Teams are free to model importations as they see fit based on their analysis of the local and global epidemiological situation. Geographic dispersion of these infections is left at teams' discretion. The ramp up of the new variant due to local transmission is also left at the teams' discretion.
In June 2022, FDA recommended that vaccines be reformulated and include two components, an original Wuhan-like strain and an Omicron BA.4/BA.5 strain. Reformulated bivalent BA4/5 boosters are currently being administered, and are available to people five years and older. We assume that reformulated bivalent boosters will provide a moderately improved protection above existing boosters; yet the exact VE will depend on circulating strains this fall. Teams should set the VE of reformulated vaccines at 80% against symptomatic disease from BA.5 and all variants not captured in the immune escape scenarios. For immune escape variants, a reduction of VE against (infection and) symptomatic disease should be implemented based on the denoted extent of immune escape. Relevant references for VE of non reformulated vaccines include:
The bivalent booster is authorized for use in individuals ages 5+. The rate of booster uptake and final coverage levels are defined in the scenarios as follows. Accelerating uptake (scenarios A & B): Booster uptake rates accelerate in the coming months and saturate by February 1st, 2023 at 90% of the state-specific flu coverage reported for the 2020-2021 fall-winter season among ages 5+ (provided here). Teams are free to use available data and information from previous rollouts as they see fit to define rates. Current uptake (scenarios C & D): Booster uptake rates stay at rates implied by current data and saturate by the end of the projection period at the level of the uptake of the booster 1 coverage. The plateau date should be based on current rates and is flexible as long as it occurs before the end of the projection period. Teams can adjust rates up if needed to achieve adequate coverage (based on booster 1) by target date. Teams are free to use available data and information from current and previous rollouts as they see fit to define rates. The distribution of who gets a booster among those for whom it is the 1st, 2nd or 3rd booster, age differences in coverage, and heterogeneity in coverage between states, is at the teams' discretion. Dose spacing: Accounting for dose spacing is not required.
Teams should include their best estimate of COVID-19 seasonality in their model but we do not prescribe a specific level of seasonal forcing.
Round 16 should NOT include reactive changes in NPIs imposed by health authorities to curb transmission, e.g., reinstatement of mask mandates, or closure of schools and businesses. However, teams can incorporate inherent changes in population behavior in response to increasing or decreasing incidences (eg, changes in contacts or masking), if these changes were inferred from earlier phases of the pandemic and are already part of the model. Database tracking of NPIs: teams may use their own data if desired, otherwise we recommend the following sources as a common starting point:
The mix of circulating strains at the start of the projection period (October 30, 2022) is at the discretion of the teams based on their interpretation/analysis of the available data and estimates of the level 5 and 6/7 variants at the the time of projection. Variation in initial prevalence between states is left at teams' discretion.
Ascertainment of cases, hospitalizations and deaths will proceed at the same level as they were at the start of the projection period. We will continue to collect the same targets (cases, hospitalizations, deaths) but note that VRBPAC and ACIP are talking about a focus on severe disease moving forward.
All of the teams' specific assumptions should be documented in meta-data and abstract.
Projection time horizon: We consider a 26-week projection period.
Scenario | Scenario name for submission file ('scenario_name') | Scenario ID for submission file ('scenario_id') |
---|---|---|
Scenario A. High boosters, Moderate immune escape variant | highBoo_modVar | A-2022-10-29 |
Scenario B. High boosters, High immune escape variant | highBoo_highVar | B-2022-10-29 |
Scenario C. Low boosters, Moderate immune escape variant | lowBoo_modVar | C-2022-10-29 |
Scenario D. Low boosters, High immune escape variant | lowBoo_highVar | D-2022-10-29 |
Other submission requirements