Planning for a Pandemic: One Hospital's Approach

Background:  The emergence of a novel H1N1 influenza A in April 2009 highlighted potential gaps in our hospital's preparedness for a pandemic. Significant problems included a lack of flexibility in our institutional pandemic response plan and a potential supply shortage.

Objective: To develop a flexible and comprehensive pandemic preparedness strategy at our 920-bed tertiary care medical center in metropolitan Los Angeles, with ~10,000 employees.

Methods: We developed and applied a forecasting model to plan for a possible pandemic influenza illness (H1N1).   1. To construct an accurate clinical model, we identified virulence and communicability (of H1N1) as critical variables, allowing us to evaluate four potential scenarios. We concluded that a “low virulence/high communicability” (Cell 2) scenario was the most likely at our institution for Influenza-Like Illness (ILI), based on global epidemiologic and clinical data as of June 2009. 

	Low Communicability	High Communicability
Low Virulence ILI Illness/death Hospitalization	Cell I Very few Very few Very few	Cell II Many Very few Few
High Virulence ILI Illness/death Hospitalization	Cell III Very few Very few Very few	Cell IV Many Many Many

2. We used historical data from three previous flu seasons to determine monthly estimates of Emergency Dept (ED) patient visits and hospitalizations for ILI, including estimates for the “walking/worried well”. These estimates were ranges and were revised as new information became available.

3. We recognized other variables that may influence our model over time, such as vaccine availability, emergence of seasonal flu, virulence changes, new affected populations (children, pregnant women), and uncertain external support.

4. To reduce the likelihood of supply or resource shortfall for the initial three months, we designed our projections to be overestimates.

5. We applied these estimates to predict the resources (personnel, supplies, beds, etc) necessary to support our institutional response for 3 months.

Results: Based on our model, we estimated a 30-50% increase in monthly ED ILI visits (effective Sep 2009), representing a 5% increase over total ED volume. We estimated that 25-30% of ED ILI visits would require hospitalization. Supply goals for 3 months were met (except H1N1 vaccine) by mid-Sept, as shown.

Item	3 Month Goal
Oseltamivir courses	24,000
Seasonal Vaccine	22,000
N95 Respirators	96,900
Surgical Masks	253,800
Gowns	524,670
Gloves	862,560
Goggles	5,000

Conclusions: The uncertainty to which the H1N1 pandemic would impact our community required early pre-emptive planning and a rolling flexible model. We reassessed our strategy on a weekly basis and made real time modifications based on our actual experience. Our initial two month experience proved that our ED projections were met and our hospitalization projections were overestimated (see figure). Our experience validated this model as flexible and potentially applicable to other scenarios.