What to know
Overview
This activity builds upon the previous two sections, in which you've learned about modeling outputs and how to work with modeling and analytics teams to create them. If you need to review the previous section, visit Activity 1: Deciding which analytic approach is best or return to the main training page.
You are given one public health decision-making case study and several modeling outputs to consider informing the decision.
Review the modeling outputs (Figures 1-4) and consider:
- What type of modeling output are you looking at, and is this appropriate for the situation?
- What kinds of input data were used? What could be the limitations of these data?
- What assumptions could have been made to generate this model? Were they reasonable?
Case Study
It's October, and you're the state epidemiologist. Healthcare providers in your state are seeking your advice on whether and when to hire medical surge staff during this example seasonal respiratory viral season.
Background: Your state has had a few hospital closures recently, and you're worried about possible surges in healthcare demand this season. Your state public health lab has not reported any new variants of COVID-19, and your estimates indicate that uptake of immunizations against seasonal respiratory viruses is lagging in your state so far.
Your healthcare coalition partners are considering contracting with staffing companies to hire surge staff for two months during the respiratory disease season. They are seeking your advice on whether to hire the surge staff and which two-month window would be most impactful.
Review the following figures
There are four figures for review in this section. Look through them and their captions and consider the questions posed in the "Overview" section above.
Historic timing of peak hospitalizations for RSV, influenza, and COVID-19
This figure shows an example peak week for hospital admissions for RSV, influenza, and COVID-19 in a state. Some seasons of RSV and influenza (2020/21 through 2022/23) were excluded, as timing and severity were altered by the COVID-19 pandemic. Historical trends indicate that peaks of COVID-19 and RSV are common in late December or early January, often around the same time, while influenza tends to peak later in the respiratory season.
Estimated timing, and magnitude of peak COVID-19 hospitalizations
This figure shows the timing and magnitude of peak hospitalization week for the forthcoming respiratory season. It's a scenario model because the two scenarios considered are (1) assuming no additional new variant (blue points) and (2) assuming a new immune escape variant (red points). The university conducted 100 simulations for each of the two scenarios. Each point represents the magnitude and timing of peak hospitalizations in a single simulation. In simulations from Scenario 1, with no new variant, the model estimated a smaller peak in late December or early January. In simulations from Scenario 2, with a new variant with immune escape, results showed a larger, later peak, but the expected timing is uncertain.
Estimated cumulative influenza hospitalizations by vaccine uptake
In this hypothetical scenario, CDC generated a stochastic compartmental model projecting cumulative hospitalizations for influenza in the upcoming season at the national level. This model provided estimates of cumulative hospitalizations in three different scenarios.
If influenza vaccine uptake is similar to last season (light blue line), if vaccine uptake is 20% higher than last season (dark blue line), or if it is 20% lower than last season (yellow line). For each scenario, 20 simulations are presented (faint lines). The bolded lines indicate the simulation for each scenario that possessed the median cumulative hospitalization rate.
RSV hospital admission forecasts
This last figure is from a local online media outlet. It shows a projection of RSV hospitalizations through February. The model is an exponential growth projection based on extending the current growth rate of reported hospitalizations over the next three or four months; it expects hospitalization levels to be up to three times the level of last season's peak, and extending well into February.