Econometric analyses of the Spanish influenza epidemic on United States military bases

This dissertation consists of two essays. The first essay estimates the parameters of a Spanish influenza epidemic model using data from 19 U.S. military bases. The second essay examines mortality of the Spanish influenza epidemic on five U.S. naval bases.;Estimates of epidemic models parameters are important for identifying disease outbreaks, evaluating treatments and estimating the impacts of quarantines and vaccine provision. In the first essay, I describe and use newly developed primary data from 19 U.S. military establishments during the Spanish flu epidemic of 1918 to estimate the parameters of the widely used Susceptible-Infectious-Removed (or SIR) model. Daily data allow us to test common assumptions of the model. Empirical estimates indicate that the infectivity parameter of the SIR model is higher during the initial days of the epidemic (especially up to the peak of the epidemic) at each military base and that the Spanish flu was highly infectious relative to more recent influenza strains. Alternative model specifications suggest that the parameter estimates are sensitive to assumptions about asymptomatic cases. Results of the data analyses suggest that if a vaccine that were 100% effective had been available, approximately 69% of the population would have needed to be vaccinated to have prevented the epidemic.;While there are studies that examine the overall the mortality level of the Spanish flu epidemic of 1918-19 on U.S. military bases, there is little analysis of the time pattern of mortality following infection. Using a newly developed data set from patient registers at five U.S. Navy hospitals, I construct life tables summarizing daily patterns of mortality and examine sources of mortality variation among patients. The results indicate that mortality of Spanish flu patients reached a peak approximately ten days after their admission to a hospital, and nearly 10% of all admitted patients died within a two-week period. The case mortality rate for patients initially diagnosed with pneumonia is exceptionally high, approximately 3.5 times greater than that of non-pneumonia patients. Patients admitted in the second week of the epidemic on a base have lower survival rates compared to patients admitted at other times. The Gulfport base had a significantly lower mortality and a different pattern of mortality than the other four naval bases in our sample. There were no statistically significant differences between the distribution of survival times in our data and the distribution constructed by Mills, et al. (2004) from three sets of autopsy reports.