Cost of pandemics: Using epidemiological modeling and insurance claims data to estimate direct healthcare costs related to an influenza pandemic

Introduction: Influenza pandemics can lead to a substantial increase in healthcare utilization which, in turn, results in a wide range of monetary losses driven by healthcare claims. Methods: We estimated hospitalization and physician visit counts as well as paid private health insurance claim amounts for the 2009 H1N1 influenza pandemic in the United States using a stochastic epidemiological and financial modeling approach. Financial model parameters were derived from an analysis of 1.1 million private health insurance claims for influenza-related hospital stays and physician visits in 2009. Regarding hospital-based claims, we estimated mean cost per day and fit univariate distributions for length of hospital stay by age category, gender, U.S. state, and ICU involvement. Paid amounts for physician visits were estimated from claims data using the mean amount paid per patient by age category, gender, and geographic location. Both medical and pharmaceutical costs were incorporated into our financial model. Results: Our modeled simulation of the 2009 H1N1 influenza pandemic included a total of nearly 300,000 excess hospitalizations and over 6.5 million physician visits. The estimated total associated cost to the health insurance industry for medical care and procedures was nearly $3 billion, with additional costs coming from pharmaceutical claims. We found that the duration of hospital stay could best be explained by Weibull distributions with five days being the mean length of stay for all hospitalizations during this event. Roughly 10% of all hospitalizations were estimated to have ICU involvement with significant geographic variation and differences seen by age and gender. Our modeled hospitalizations produced private health insurance claims that ranged from $2,000 to $40,000. Paid claims for physician visits averaged about $300 and varied by age, sex, and geographic location group. Results were validated against available published values and compared well. Discussion: The development of models to estimate the potential costs of healthcare utilization during an influenza pandemic can aid decision makers in preparedness planning. Comparison of our results with historical data from the 2009 H1N1 influenza pandemic suggests these methods could be used to better quantify those potential losses.