Stockpiling Ventilators for Influenza Pandemics

Abstract

In preparing for influenza pandemics, public health agencies stockpile critical medical resources. Determining appropriate quantities and locations for such resources can be challenging, given the considerable uncertainty in the timing and severity of future pandemics. We introduce a method for optimizing stockpiles of mechanical ventilators, which are critical for treating hospitalized influenza patients in respiratory failure. As a case study, we consider the US state of Texas during mild, moderate, and severe pandemics. Optimal allocations prioritize local over central storage, even though the latter can be deployed adaptively, on the basis of real-time needs. This prioritization stems from high geographic correlations and the slightly lower treatment success assumed for centrally stockpiled ventilators. We developed our model and analysis in collaboration with academic researchers and a state public health agency and incorporated it into a Web-based decision-support tool for pandemic preparedness and response.

Keywords: Texas; United States; Ventilators; influenza; optimization; pandemic; viruses.

Figure 1

Overview of methods for projecting the need to stockpile ventilators for an influenza pandemic, Texas, USA. First, a forecasting model was used to estimate weekly hospitalizations at each site on the basis of historical ILI hospitalization data and CDC ILINet reports. Second, 3 additional factors, along with a spatial correlation coefficient, were used to form a probability distribution for peak-week ventilator demand at each site. Third, an optimization model was solved to determine local and central stockpile allocations and generate trade-off curves between the expected unmet demand and total stockpile and between the probability of unmet demand and total stockpile. CDC, Centers for Disease Control and Prevention; HSR, health service region; ICU, intensive care unit; ILI, influenza-like illness.

Figure 2

Optimal ventilator stockpiles for a mild pandemic scenario, Texas, USA. The total size of the optimal stockpile, summed across the central and 8 HSR stockpiles, decreases as risk tolerance increases. Risk for unmet demand for ventilators is quantified as the expected number of hospitalized influenza patients statewide not receiving necessary ventilation (EUD) (A) and the probability of at least 1 hospitalized patient in Texas not receiving necessary ventilation (PUD) (B). We optimized directly for EUD and calculated PUD post hoc. Red circles indicate EUD/PUD of 5 patients. C) Optimal allocation among central and regional sites when EUD is set to 5 patients, equivalent to a stockpile of 272 ventilators. EUD, expected unmet demand; PUD, probability of unmet demand; HSR, health service region.

Figure 3

Optimal ventilator stockpiles for moderate and severe pandemic scenarios, Texas, USA. The total size of the required stockpile, summed across the central and 8 HSR stockpiles, decreases as risk tolerance (EUD) increases, for both moderate (A) and severe (C) pandemic scenarios. For an EUD of 5 patients (red circles), total stockpiles would be 1,172 (A) and 15,697 (C); optimal allocations to central and regional stockpiles are shown for moderate (B) and severe (D) scenarios. EUD, expected unmet demand; PUD, probability of unmet demand; HSR, health service region.