Scenario Modeling Hub

A Note on Scenario Modeling Hub Round 3 (January 5, 2023)

In Round 3, we produced mid-season projections of influenza in the US, using data up through December 3, 2022. In the four scenarios, we continued to assess the impact of vaccination effectiveness and reduced residual immunity during the 2022-2023 influenza season, a result of reduced transmission of influenza during the COVID-19 pandemic. Ensemble projections are based on contributions from 12 teams (10 nationally) and cover the period from Dec 4, 2022 to June 3, 2023. Scenarios were identical to Round 2, with an additional 3 weeks of calibration data.

Our main findings include:

  • In the most pessimistic scenario, with low vaccine effectiveness and pessimistic pre-existing immunity, weekly hospitalizations are projected to peak at 47,800 nationally (50% PI, 33,300-69,800). In this scenario, peak hospitalizations have a substantial likelihood to exceed the peak of the severe 2017-18 season (n = 34,385 hospitalizations).
  • Irrespective of scenario, ensemble hospitalizations are projected to peak in mid- to late-December, 2022, though there is variability between states. Some states whose epidemics started later are expected to peak in January.
  • Increased vaccine effectiveness is expected to decrease both the peak and cumulative hospitalizations. Higher vaccine effectiveness has a more pronounced effect on the overall burden in the optimistic immunity scenario (likely because the epidemic is projected to occur later, when a larger fraction of the population is effectively immunized). We project that high vaccination effectiveness decreases cumulative hospitalizations by 27.3% in high immunity scenarios, but only by 14.8% in low immunity scenarios.
  • There is substantial agreement in the trajectory of individual models in this round.
  • At the time of this report, the national trajectory of hospitalizations is declining, falling below the most pessimistic scenario, but still within the other scenarios. It is possible the US has already peaked nationally, though there is substantial variability at the state-level. Additionally, a resurgence remains possible.
  • A few caveats are worth noting:
    • The amount of calibration data available for the new HHS influenza dataset remains limited, and testing practices could change between and within seasons. Similarly, the amount of death data available for calibration is limited.

Table 1. Flu Scenario Modeling Hub round 3 scenarios. More detailed scenario definitions and model characteristics can be found at https://github.com/midas-network/flu-scenario-modeling-hub.


Even the best models of emerging infections struggle to give accurate forecasts at time scales greater than 3-4 weeks due to unpredictable drivers such as a changing policy environment, behavior change, the development of new control measures, and stochastic events. However, policy decisions around the course of emerging infections often require projections in the time frame of months. The goal of long-term projections is to compare outbreak trajectories under different scenarios, as opposed to offering a specific, unconditional estimate of what “will” happen.

As such, long-term projections can guide longer-term decision-making while short-term forecasts are more useful for situational awareness and guiding immediate response.